key: cord- -bz g a authors: davis, kailah; staes, catherine; duncan, jeff; igo, sean; facelli, julio c title: identification of pneumonia and influenza deaths using the death certificate pipeline date: - - journal: bmc med inform decis mak doi: . / - - - sha: doc_id: cord_uid: bz g a background: death records are a rich source of data, which can be used to assist with public surveillance and/or decision support. however, to use this type of data for such purposes it has to be transformed into a coded format to make it computable. because the cause of death in the certificates is reported as free text, encoding the data is currently the single largest barrier of using death certificates for surveillance. therefore, the purpose of this study was to demonstrate the feasibility of using a pipeline, composed of a detection rule and a natural language processor, for the real time encoding of death certificates using the identification of pneumonia and influenza cases as an example and demonstrating that its accuracy is comparable to existing methods. results: a death certificates pipeline (dcp) was developed to automatically code death certificates and identify pneumonia and influenza cases. the pipeline used metamap to code death certificates from the utah department of health for the year . the output of metamap was then accessed by detection rules which flagged pneumonia and influenza cases based on the centers of disease and control and prevention (cdc) case definition. the output from the dcp was compared with the current method used by the cdc and with a keyword search. recall, precision, positive predictive value and f-measure with respect to the cdc method were calculated for the two other methods considered here. the two different techniques compared here with the cdc method showed the following recall/ precision results: dcp: . / . and keyword searching: . / . . the f-measure were . and . respectively (dcp and keyword searching). both the keyword and the dcp can run in interactive form with modest computer resources, but dcp showed superior performance. conclusion: the pipeline proposed here for coding death certificates and the detection of cases is feasible and can be extended to other conditions. this method provides an alternative that allows for coding free-text death certificates in real time that may increase its utilization not only in the public health domain but also for biomedical researchers and developers. trial registration: this study did not involved any clinical trials. the ongoing monitoring of mortality is crucial to detect and estimate the magnitude of deaths during epidemics, emergence of new diseases (for example, seasonal or pandemic influenza, aids, sars), and the impact of extreme environmental conditions on a population such as heat waves or other relevant public health events or threats [ , ] . the surveillance of vital statistics is not a novel idea; mortality surveillance has played an integral part in public health since the london bills of mortality were devised in the seventeenth century [ ] . the bills served as an early warning tool against bubonic plague by monitoring deaths from the to the s. today, mortality surveillance continues to be a critical activity for public health agencies throughout the world [ ] [ ] [ ] [ ] . pneumonia and influenza are serious public health threats and are a cause of substantial morbidity and mortality worldwide; for instance, the world health organization (who) estimates seasonal influenza causes between , to , deaths worldwide each year [ ] while pneumonia kills more than million people worldwide every year [ ] . worldwide, the morbidity and mortality of influenza and pneumonia have a considerable economic impact in the form of hospital and other health care costs. each year in the united states approximately million persons acquire pneumonia and, depending on the severity of the influenza season, to million people in the us contract influenza [ ] . these numbers contribute to approximately . million hospitalizations, of which . million are pneumonia cases [ ] and the remainder for influenza [ ] . moreover, pneumonia cases and influenza together cost the american economy . billion dollars in [ ] . in the netherlands it has been estimated that influenza accounts for and days of hospitalization per , highrisk and low-risk elderly, respectively [ ] . due to the public health burden and the unpredictability of an influenza season, strong pneumonia and influenza surveillance systems are a priority for health authorities. mortality monitoring is an important tool for the surveillance of pneumonia and influenza which can aid in the rapid detection and estimates of excess deaths and inform and evaluate the effect of vaccination and control programs. traditionally, influenza mortality surveillance often uses the category of "pneumonia and influenza" (p-i) on death certificates as an indicator of the severity of an influenza season or to identify trends within a season; however, only a small proportion of these deaths are influenza related. it has been reported that only . [ ] [ ] [ ] [ ] [ ] . % of all pneumonia and influenza deaths are influenza related [ , ] . the non-influenza-related pneumonia deaths tend to be stable from year to year and fluctuations in this category are largely driven by the prevalence and severity of seasonal influenza. as a result, the p-i category is an important sentinel indicator. in the us, death certificates are the primary data source for mortality surveillance whose findings are widely used to exemplify epidemics and measure the severity of influenza seasons [ ] . currently, there are three systems to monitor influenza-related mortality; one system in particular, the cities mortality reporting system, provides a rapid assessment of pneumonia and influenza mortality [ ] . each week, this system summarizes the total number of death certificates filed in us cities, as well as the number of deaths due to pneumonia and influenza. however, even these data can be delayed by approximately - weeks from the times of death. this delay can be attributed to one of the following reasons: ) timeliness of death registration and ) reviewing of the death certificates to identify pneumonia and influenza deaths [ , , ] . the registration and reviewing of death certificates varies by states and, as a result, there is variability in length of time to report a death to cdc. for instance, states with paper-based death registration system typically perform manual reviews of the death certificates which can take up to weeks; however states with electronic death registration systems (edrs) may perform automatic reviews which can decrease this time significantly. the current cities mortality reporting system surveillance system also lacks flexibility for expanding the number of conditions and/or the geographic distribution. moreover, the unavailability of coded death records due to the complexity of the national center of health statistics (nchs) coding process results in multiple strategies to identify common outbreaks such as pneumonia and influenza deaths, which greatly vary by jurisdiction. to bypass the lengthy nchs process, a variety of approaches have been attempted that are close to 'realtime' but less than optimal. for instance, in utah keyword searching is used to identify pneumonia and influenza deaths; although this method is fast and easy to implement, it can easily result in the over or under estimation of cases. this can occur by missing cases due to misspelled terms, synonyms, variations, or the selection of strings containing the search term. other research groups [ , ] have demonstrated the feasibility of using mortality data for real time surveillance but all used "free text" search for the string "pneumonia", "flu" or "influenza." as noted earlier, although this method can provide the semi quantitative measurements for disease surveillance purposes, keyword searches can also result in an array of problems that result from complexities of human language such as causal relationships and synonyms [ ] . therefore, the lack of coded death data that may not be available for months [ ] seriously limits the use of death records in automated systems. at this time, there is little published on the automatic assignment of codes to death certificates for automatic case detection. currently the coding of death certificates is a complex process which involves many entities. in the us, where we are focusing this study, the codes on death certificates that are generated by the national center for health statistics (nchs) depend on information reported on the death certificate by the medical examiner, coroner, or another certifier, and there is substantial variation in how certifiers interpret and adhere to causeof-death definitions [ ] . the cause of death literals are coded into international classification of diseases tenth revision (icd- ) [ ] and the underlying and multiplecause-of-death codes are selected based on the world health organization coding rules. these coding rules have been automated by cdc with the development the mortality medical data system (mmds) which consists of four programs: super mortality medical indexing classification and retrieval (supermicar) data entry; mortality medical indexing classification and retrieval (micar); automated classification of medical entities (acme) and transax (translation axes). supermi-car was designed to facilitate the entry of literal text of causes of death in death certificates and convert them into standardized expressions acceptable by micar. it contains a dictionary which assigns an entity reference number (ern) to statements on the death certificate. these erns are fed into micar which transforms the erns into icd- codes by using specific mortality coding rules; the rules require look-up files and a dictionary. acme and transax then selects the underlying and multiple causes of death respectively. icd- codes from micar are fed into acme which assigns the underlying cause of death using decision tables. the decision table contains all possible pairings of diseases for which the first disease can cause the second. in the latest version of the system, acme is comprised of eight decision tables including three tables of valid and invalid codes, causal relationships (general principle and rule ), and direct sequel (rule ), and three other tables needed by modification rules. figure provides the workflow for the mmds system. of the . million deaths that occur each year - percent are automatically coded through super-micar, and the remaining records are then manually coded by nosologists, a medical classification specialist [ ] ; this is a tedious and lengthy process lasting up to months. although the automation process has decreased the time required for coding death data to - weeks, the national vital statistics data is not available for at least two years. therefore, local health department still manually code records or perform basic process techniques to quickly characterize disease patterns [ ] . records that were processed through super-micar or were manually coded are then processed through the remaining components (micar , acme and transax) of mmds. in , micar had a throughput rate of - %, while acme rate was percent. moreover, based on a reliability study, acme error rate for selecting the underlying cause is at onehalf percent, while transax, the multiple cause codes had a one-half percent error rate [ ] . due to the high processing rates and low error rates, mmds is considered by practitioners as the gold standard for the processing and coding of death certificates in the us and other countries (such as canada, the united kingdom (uk) and australia). therefore, we used the codes produced by this system as the "gold standard" when comparing with the methods developed here. in , the us steering committee to reengineer the death registration process (a task force representing federal agencies, the national center for health statistics and the social security administration, and professional organizations representing funeral directors, physicians, medical examiners, coroners, hospitals, medical records professionals, and vital records and statistics officials (naphsis) published the report "toward an electronic death registration system in the united states: report of the steering committee to reengineer the death registration process." this report explained the feasibility of developing electronic death registration in the united states [ ] and argued that these electronic death records have the potential to be an effective source of information for nation-wide tracking and detecting of disease outbreaks. however, little actions have been taken to implement such recommendations in a comprehensive manner. as of july , electronic death registration systems were operating in states, the district of colombia, and in development or planning stage in a dozen others [ ] . information representing the 'cause of death' field on the death certificates is free text. one major goal of natural language processing (nlp) is to extract and encode data from free-texts. there have been many research groups developing nlp systems to aid in clinical research, decision support, quality assurance, the automation of encoding free text data and disease surveillance [ ] [ ] [ ] . although, there have been a few nlp applications to the public health domain [ , ] , little is known about its capability to automatically code death certificates for outbreak and disease surveillance. recently, medical match master (mmm) [ ] , developed by riedl et al at the university of california davis, was used to match unstructured cause of death phrases to concepts and semantic types within the unified medical language system (umls). the system annotates each death phrase input with two types of information, the concept unique identifier, cui, and a semantic type both assigned by the umls. mmm was able to identify an exact concept identifier (cui) from the umls for over % of 'cause of death' phrases. although, the focus of this study was to use nlp techniques to process death certificates, the description of this system reported in the literature did not show how well coded data from an nlp tool along with predefined rules can detect countable cases for a specific disease or condition. the purpose of our project is to create a pipeline which automatically encodes death certificates using a nlp tool and identify deaths related to pneumonia and influenza which provides daily and/or weekly counts. we compared the new technique developed here with keyword searching and mmds as exemplars of the easiest possible approach and the current "gold standard", respectively. the comparison of the techniques was done by calculating recall, precision, f-measure, positive predictive value and agreement (cohen's kappa). we obtained , de-identified electronic death records all with multiple-cause-of-death from the utah department of health (udoh) for the period january to december . the records included a section describing the disease or condition directly leading to death, and any antecedent causes, co-morbid conditions and other significant contributing conditions. an example of a paper and electronic death certificate are shown in figures and respectively. all death certificates used in this study have been processed using the mortality medical data system (mmds) and the record axis codes were received from udoh. for our study we randomly selected , ( %) records. all death records included in the study were previously also coded by nchs into icd- , but this information was not used for our coding, it was only used as posteriori to assess to quality of the automatic coding. we chose to apply the centers of disease control and prevention case definition of pneumonia and influenza deaths defined by cdc's epidemiologist staff through personal communication. therefore, the operational definition for deaths from influenza includes deaths from all types of influenza with the exception of deaths from haemophilus influenzae infection and deaths from parainfluenzae virus infection. pneumonia deaths include deaths from all types of pneumonia including pneumonia due to h. influenza and pneumonia due to parainfluenzae virus. the exceptions include aspiration pneumonia (o . , o , o . , j .-and p .-) , pneumonitis (j . , j -j ), and pneumonia due to pneumococcal meningitis (j , g . ) . pneumonia and influenza related deaths were defined as one of the diagnoses listed in table which were reported in any cause of death field. these codes were selected through manual review of the icd- version manual [ ] . the death certificates pipeline, dcp, was developed to identify pneumonia and influenza cases. the pipeline consisted of two components. the first component of the system was the natural language processor, for which we used metamap [ ] , and the second component was the definitional rules that were applied to the output generated by metamap. the study procedures for this pipeline included: preprocessing, nlp, extraction of coded data and the detection of pneumonia and influenza cases (figure ). spelling errors are common on death certificates; therefore, the death records were first processed through a spell checker to identify misspellings. although the umls sl has a spell suggestion tool called gspell [ ] [ ] [ ] , we decided not to use it and chose to utilize aspell [ ] . our motivation for this decision was based upon an evaluation which showed aspell outperforming gspell; aspell performed better on three areas of performance which were ( ) whether the correct word was ranked in the top ten; and ( ) whether the correct word was found at all [ ] . perl (www.perl.org), a high-level computer programming language that aids in the manipulation and processing of large volume of text data was then used to prepare the cause of death free text for nlp. the preprocessing also involved the removal of non-ascii characters; this was a required technical step for metamap processing. step : natural language processing metamap was used to convert the electronic death records to coded descriptions appropriate for the rule based system. metamap [ ] , developed by the national library of medicine (nlm), is useful in identifying biomedical concepts from free-form textual input and maps them into concepts from the unified medical language system (umls) metathesaurus [ , ] . metamap works by breaking the inputted text into words or phrases, map them to standard terms, and then match the terms to concepts in the unified medical language system (umls) [ ] . for each matched phrase, metamap classifies it into a semantic type then returns the concept unique identifier (cui) and the mapping options which are ranked according to the strength of the mapping. output from metamap. text bolded in the output from nlp represent the code and its corresponding phrase. step : extraction of coded data the data produced by metamap (xml format) was processed through a perl script to extract the inputted text and its corresponding meta-mapped cuis. this extracted data was outputted to a text document. step : identification of p-i deaths the identification of pneumonia and influenza cases involved two steps: ) identifying cuis relating to pneumonia and influenza and ) use of the cuis to create a rules based algorithm to identify cases. details of each step are explained in the following paragraphs. to determine which cui codes were relevant for identifying pneumonia and influenza deaths it was necessary to create a "cui code list" that represents all the icd- codes of interest (see table ). to create this list, we generated a subset of the umls ab database [ ] using the metamorphosys [ ] tool provided by the national library of medicine, nlm. the umls database includes many vocabularies, therefore, to determine which vocabularies are relevant to our aims we used the procedure used by riedl three queries were performed on the subset described above to map pneumonia and influenza icd- codes to cuis and identify related pneumonia and influenza concepts. each query was then placed in a separate database, all duplicates were removed and a sub-query was run to ensure that only the icd- codes in table were included in this list. this produced distinct concept identifiers (cuis) relating to pneumonia or influenza. these codes were used to develop the rules to identify the cases of interest. the coded data produced by metamap was accessed by rules, aimed at identifying the presence of pneumonia and influenza based on the coded data. the rules for identifying these deaths used the cui code list described above. the rule looks at each cause of death field (underlying cause, additional causes, etc.) to flag records with relevant codes. these rules used boolean operators (and, or, not) and if-then statements to create a chain of rules ( figure ). the list of cases identified by our automated detection system was compared with those identified by two other methods: a) keyword searching and b) the reference standard: the icd- codes given by the cdc mmds method. for key-word searching we followed the process to evaluate the performance of both techniques against the reference standard, we needed to specify what constituted a match. each death record is associated to a unique number; therefore, we considered a match if the unique identifier was identified by the comparator and also found by the reference standard. three standard measures were used to evaluate the performance of one method in relation to the reference standard used in this study: precision (equivalent to positive predictive value; recall (equivalent to sensitivity or true positive rate), and f-measure. kappa statistics were used to assess agreement and mcnemar's test was used to analyze the significance between the two methods. all calculations were performed in r [ ] . to calculate these values, pneumonia and influenza related deaths were examined by comparing the reference standard output vs. the two comparators: dcp and keyword search. for both comparators, the deaths were counted and categorized as true posi-tives (cases found by the comparator-pneumonia deaths being correctly classified); false positives (incorrect cases found by the comparator-the number of pneumonia and influenza deaths incorrectly identified by the comparator); false negatives (correct cases not found by the comparator-the number of pneumonia deaths not identified by the comparator). precision, recall and f-score were calculated as follows: precision = true positives/(true positives + false positives) ( ) recall = true positives/(true positives + false negatives) ( ) f-measure = *(p r/ p + r) ( ) mcnemar's test was also calculated to evaluate the significance of the difference between the two comparators. to calculate this value a confusion matrix was created where a is the number of times both methods have correct predictions; b is the number of times method has a correct prediction and method has a wrong prediction; c is the number of times method has a correct prediction and method has a wrong prediction; d is the number of times both methods have incorrect predictions. ethics approval was not required for this study. identifying variables that could be used for re-identifying individuals were excluded from the study data. the records were processed and analyzed on a server with two opteron dual-core . ghz processors and gb ram at the center of high performance computing at the university of utah. using keyword searching the cpu processing time to identify pneumonia and influenza cases was . seconds and the wall time was . seconds. for the dcp, the total cpu processing time was . seconds. the nlp portion of the pipeline attributed to . percent of the processing time (nlp- seconds). while the dcp execution time is much longer, still it is well within the "in real time" realm. for instance, it would take , . seconds cpu time seconds for dcp to code and flag all the weekly death records of the us ( , ). recall and precision were calculated at a . confidence intervals; the f-measure was also calculated. the performance of each method is described below. of the , records analyzed keyword search identified records as pneumonia and influenza deaths, being identified as false positives. precision for keyword searching was calculated at %. of the false positives, records correctly mentioned pneumonia in the cause of death text but their corresponding icd- codes failed to provide any code related to pneumonia, while records were flagged because it included the sub-string "pneumonia" in the additional cause of death field. the death literal for these two records were "bacteremia due to streptococcus pneumonia" and "streptococcal pneumoniae septicemia", the remaining errors were due to the entry of the death literals; in all cases the negation of 'aspiration pneumonia' either due to: ) 'pneumonia' being in a separate cause of death field to 'aspiration' or ) 'pneumonia' not being directly followed by 'aspiration' in the death text (example "pneumonia due to secondary aspiration"). a total of false negatives were recorded, yielding a recall of %. the false negatives could be generalized into two categories: ) misspellings of pneumonia on the death certificated (n = ) and ) appropriate pneumonia or influenza icd- code was coded but the death literals did not mention an appropriate scanned phrase (n = ). f-measure was also calculated at %. a high level of agreement was seen among keyword searching and the reference standard (kappa . ). utilizing the death certificates pipeline (dcp), we identified records as pneumonia and influenza deaths, of which were false positives. the precision for this method was calculated at %. like the keyword searching method, of the false positives, records mentioned pneumonia in the cause of death field but their corresponding icd- codes failed to provide any code related to pneumonia and the remaining errors were due to the reporting of aspiration pneumonia on the death certificate. this method had only false negative for the death literal stating "recurrent aspiration with pneumonia", thus yielding a recall at . %, being less than keyword searching. f-measure was calculated at %. the level of agreement between the pipeline and the gold standard was almost perfect with a cohen's kappa of . . the precision and recall scores that are reported above suggest that the dcp is a better method for identifying pneumonia and influenza deaths than keywordsearching. therefore, we investigated if this observation is supported by statistical analysis. performing a fisher's exact test at α = . , significant difference was seen for both recall (p = . e- ) and precision (p = . ). the mcnemar's test result also showed dcp to be a better method with a p-value = . e- . for the pneumonia and influenza cases found by the reference standard, dcp correctly identified cases, missed one case and incorrectly flagged nine cases. most failures were due to discrepancies between the death literal and its respective icd- code. for the only case which the pipeline did not match, the phrase 'recurrent aspiration with pneumonia' was present in the death literal. metamap coded this literal as aspiration pneumonia which was excluded from the cui code list, but its respective icd- included j . for the additional cases which were not present in the reference standard, we noticed two categories of errors: )cases where the string 'pneumonia' is present in the death literal but not coded into icd- and ) the reporting of aspiration pneumonia on the death certificate. the first category of errors was not due to metamap or the rule algorithm, but perhaps due to the coding process. as described earlier, mmds produces entity axis and record axis codes. the entity axis codes would be a more appropriate reference standard for they provide the icd codes for the conditions or events reported as listed by the death certifier and maintains the order as written on the death certificate [ ] ; but as noted earlier only the record axis codes were made available for this study. the algorithm used to produce record axis codes from the entity axis data removes duplicate codes and contradictory diagnoses within the entity axis data to produce the more standardized record axis [ ] . for example, if a medical examiner reports pneumonia with chronic obstructive pulmonary disease both conditions will be shown in entity axis code data. however, in record axis code data, they will be replaced with a single condition: chronic obstructive pulmonary disease with acute lower respiratory infection (j . ). we were unable to verify that codes related to pneumonia were present in the entity axis codes for the six cases; therefore, we can only speculate the reason for this failure. the second category of errors was due to the reporting of aspiration pneumonia on the death certificate. in cases where the string "aspiration" and pneumonia" were not reported in the same text field metamap processed the string separately thus yielding two codes: one for aspiration and the other pneumonia, instead of one code for "aspiration pneumonia" [c ]. in an initial review of metamap we found metamap had difficulties processing the phrase "pneumonia secondary to acute aspiration", therefore, our rule detection algorithm excluded cases where the code for pneumonia and aspiration were present in the same text field. to our knowledge, this is the first published report on using a natural language processing tool and the umls to identify pneumonia and influenza deaths from death certificates. we found that automated coding and identification of pneumonia and influenza deaths is possible and computationally efficient. the death certificates pipeline developed here was statistically different to keyword searching and has higher recall and precision when compared to the current semi-automatic methods in use by the cdc. a good recall is required to help capture the 'true' p-i deaths and a good precision is needed to avoidoverestimating the number of p-i deaths. this study also indicated that keyword searching underestimated pneumonia and influenza deaths in utah. the simple keyword search method not only decreased recall and precision but also reduced the level of agreement. when reporting counts for surveillance purposes it's best to be as accurate as possible; however, there's a trade-off between recall and precision. for disease surveillance, increased precision enables public health officials to more accurately focus resources for control and prevention, therefore, although both methods had good precision the pipeline developed would be more advantageous to utilize. metamap did an excellent job at extracting cause of deaths from free-form text which is consistent with the results of reid et al [ ] . most of the concepts were present in the umls which attributed good recall. both recall and precision depended on the comprehensiveness of the cui code list. the performance of this system is determined largely by the coverage of terms and sources in the umls. both keyword searching and the system's weakest point is its lack of precision. most of the concepts the system did not identify had either the aspiration text in another field or pneumonia was mentioned in the cause of death text but not coded ( cases fit these criteria). the sample size was sufficient to show difference between the two methods. it is important to note that utilizing trained nosologists, who would manually code the death certificates, would have developed an absolute gold standard which may or may not be a better reference standard than icd- codes. however, our motivation for utilizing icd codes was influenced due to the fact that the use of icd codes to identify all-cause pneumonia has been examined and has showed to be a valid tool for the identification of these cases [ , ] . in terms of timing, while keyword searching is faster than the dcp, our method is also sub / second range, which implies that it is possible to process the daily utah deaths (~ ) in approximately . seconds and all deaths in the us (~ ) in approximately . seconds using current hardware. this timing would be much faster than the minimum of two weeks to receive the coded data from the current cdc process. moreover, these timings make it apparent that this system can be integrated in a real time surveillance system without introducing any additional bottlenecks. there are several potential limitations with this analysis. first, the generalizability of the findings is limited because the death records were only from one institution. although death certificates have a standardized format, the death registration process and the reviewing of death records differ by institutions. udoh utilizes keyword searching to identify pneumonia and influenza cases, other institutions may use more accurate (manual review) or less accurate methods for finding cases. second, a separate evaluation of the nlp component of the dcp was not performed. further research is needed to examine the use of nlp on electronic death records across institutions and countries which may have different documentation procedures. this study shows that it is feasible to achieve high levels of accuracy when using nlp tools to identify cases of pneumonia and influenza cases from electronic death records while still providing a system that can be used for real time coding of death certificates. identification of concept identifiers related to the cdc's case definition of pneumonia and influenza was very important in producing a highly accurate rule for the identification of these cases. future work will aim to improve the preprocessing phase of the pipeline by providing the inclusion of the spellchecker used by the cdc's mortality medical data system. future work will also involve evaluating the flexibility (e.g. identification of different diseases) of the system to deploy the pipeline tool, along with other public health related analytical tools, as a grid service to provide to real time public health surveillance tool that uses data and services under the control of different administrative domains. we have shown that it is feasible to automate the coding of electronic death records for real-time surveillance of deaths of public health concern. the performance of the pipeline outperformed the performance of current methods, keyword searching, in the identification of pneumonia and influenza related deaths from death certificates. therefore, the pipeline has the potential to aid in the encoding of death certificates and is flexible to identify deaths due to other conditions of interest as the need arises. participants of a workshop on mortality monitoring in europe: monitoring excess mortality for public health action: potential for a future european network public health surveillance: historical origins, methods and evaluation communicable disease surveillance the new automated daily mortality surveillance system in portugal description of a new all cause mortality surveillance system in sweden as a warning system using threshold detection algorithms a method for timely assessment of influenza-associated mortality in the united states mortality surveillance - , england and wales. deaths and rates by sex and age group for th revision causes, a-list and chapters. london: great britain office of population census and surveys prevention and control of seasonal influenza with vaccines: recommendations of the advisory committee on immunization practices (acip) national hospital discharge survey: summary influenza-associated hospitalizations in the united states american lung association state of lung disease in diverse communities economic evaluation of influenza vaccination. assessment for the netherlands mortality associated with influenza and respiratory syncytial virus in the united states estimating seasonal influenza-associated deaths in the united states: cdc study confirms variability of flu flu activity & surveillance: reports and surveillance methods in the united states real-time surveillance of pneumonia and influenza mortalities via the national death certificate system prospective surveillance of excess mortality due to influenza in new south wales: feasibility and statistical approach medical language processing: computer management of narrative data death certificate comparability of cause of death between icd- and icd- : preliminary estimates world health organization: international statistical classification of disease and related health problems, tenth revision version for selected data editing procedures in an automated multiple cause of death coding system using the umls and simple statistical methods to semantically categorize causes of death on death certificates description of the national center for health statistics software systems and demonstrations toward an electronic death registration system in the united states: report of the steering committee to reengineer the death registration process national association for public health statistics and information systems classifying free-text triage chief complaints into syndromic categories with natural language processing automatic detection of acute bacterial pneumonia from chest x-ray reports automated encoding of clinical documents based on natural language processing using nlp on va electronic medical records to facilitate epidemiologic case investigations evaluating natural language processing applications applied to outbreak and disease surveillance a: effective mapping of biomedical text to the umls metathesaurus: the metamap program a frequency-based technique to improve the spelling suggestion rank in medical queries lexical systems: a report to the board of scientific counselors umls language and vocabulary tools metamap: mapping text to the umls metathesaurus the unified language system (umls): integrating biomedical terminology umls distribution team: r: a language and environment for statistical computing. vienna: r foundation for statistical computing entity axis codes documentation of the mortality tape file for community-acquired pneumonia: can it be defined with claims data? icd- codes are a valid tool for identification of pneumonia in hospitalized patients aged > or = years identification of pneumonia and influenza deaths using the death certificate pipeline this study has been supported in part by the grants from the national library of medicine (lm ) and from the centers of disease control and prevention center of excellence (ip hk - ). the authors declare that they have no competing interests. all the authors contributed equally to this research. all authors read and approved the final manuscript.submit your next manuscript to biomed central and take full advantage of: key: cord- -q ivezyq authors: saikumar, pothana; kar, rekha title: apoptosis and cell death: relevance to lung date: - - journal: molecular pathology of lung diseases doi: . / - - - - _ sha: doc_id: cord_uid: q ivezyq in multicellular organisms, cell death plays an important role in development, morphogenesis, control of cell numbers, and removal of infected, mutated, or damaged cells. the term apoptosis was first coined in by kerr et al. to describe the morphologic features of a type of cell death that is distinct from necrosis and is today considered to represent programmed cell death. in fact, the evidence that a genetic program existed for physiologic cell death came from the developmental studies of the nematode caenorhabditis elegans. as time has progressed, however, apoptotic cell death has been shown to occur in many cell types under a variety of physiologic and pathologic conditions. cells dying by apoptosis exhibit several characteristic morphologic features that include cell shrinkage, nuclear condensation, membrane blebbing, nuclear and cellular fragmentation into membrane-bound apoptotic bodies, and eventual phagocytosis of the fragmented cell (figure . ). in multicellular organisms, cell death plays an important role in development, morphogenesis, control of cell numbers, and removal of infected, mutated, or damaged cells. the term apoptosis was fi rst coined in by kerr et al. to describe the morphologic features of a type of cell death that is distinct from necrosis and is today considered to represent programmed cell death. in fact, the evidence that a genetic program existed for physiologic cell death came from the developmental studies of the nematode caenorhabditis elegans. as time has progressed, however, apoptotic cell death has been shown to occur in many cell types under a variety of physiologic and pathologic conditions. cells dying by apoptosis exhibit several characteristic morphologic features that include cell shrinkage, nuclear condensation, membrane blebbing, nuclear and cellular fragmentation into membrane-bound apoptotic bodies, and eventual phagocytosis of the fragmented cell (figure . ). cell death is central to the normal development of multicellular organisms during embryogenesis and maintenance of tissue homeostasis in adults. during development, sculpting of body parts is achieved through selective cell death, which imparts appropriate shape and creates required cavities in particular organs. in adults, cell death balances cell division as a homeostatic mechanism regulating constancy of tissue mass. deletion of injured cells because of disease, genetic defects, aging, or exposure to toxins is also achieved by apoptosis. in essence, apoptotic cell death has important biologic roles not only in development and homeostasis but also in the pathogenesis of several disease processes. dysregulation of apoptosis is found in a wide spectrum of human diseases, including cancer, autoimmune diseases, neurodegenerative diseases, ischemic diseases, viral infections, and lung diseases. our knowledge of cell death and the mechanisms of its regulation increased dramatically in the past two decades with the discovery nevertheless, necrosis has been shown to occur in cells having defects in apoptotic machinery or upon inhibition of apoptosis, and this form of cell death is emerging as an important therapeutic tool for cancer treatment. autophagy autophagy, which is also referred to as type ii programmed cell death, is characterized by sequestration of cytoplasm and organelles in double or multimembrane structures called autophagic vesicles, followed by degradation of the contents of these vesicles by the cell's own lysosomal system (see figure . ). the precise role of autophagy in cell death or survival is not clearly understood. autophagy has long been regarded as a cell survival mechanism whereby cells eliminate long-lived proteins and organelles. in this regard, it is argued that autophagy may help cancer cells survive under nutrientlimiting and low-oxygen conditions and against ionizing radiation. , however, recent observations that there is there is early membrane damage with eventual loss of plasma membrane integrity and leakage of cytosol into extracellular space. despite early clumping, the nuclear chromatin undergoes lysis (karyolysis). apoptosis: cells die by type i programmed cell death (also called apoptosis); they are shrunken and develop blebs containing dense cytoplasm. membrane integrity is not lost until after cell death. nuclear chromatin undergoes striking condensation and fragmentation. the cytoplasm becomes divided to form apoptotic bodies containing organelles and/or nuclear debris. terminally, apoptotic cells and fragments are engulfed by phagocytes or surrounding cells. autophagy: cells die by type ii programmed cell death, which is characterized by the accumulation of autophagic vesicles (autophagosomes and autophagolysosomes). one feature that distinguishes apoptosis from autophagic cell death is the source of the lysosomal enzymes used for most of the dying-cell degradation. apoptotic cells use phagocytic cell lysosomes for this process, whereas cells with autophagic morphology use the endogenous lysosomal machinery of dying cells. paraptosis: cells die by type iii programmed cell death, which is characterized by extensive cytoplasmic vacuolization and swelling and clumping of mitochondria, along with absence of nuclear fragmentation, membrane blebbing, or apoptotic body formation. autoschizis: in this form of cell death, the cell membrane forms cuts or schisms that allow the cytoplasm to leak out. the cell shrinks to about one-third of its original size, and the nucleus and organelles remain surrounded by a tiny ribbon of cytoplasm. after further excisions of cytoplasm, the nuclei exhibit nucleolar segregation and chromatin decondensation followed by nuclear karyorrhexis and karyolysis. decreased autophagy during experimental carcinogenesis and heterologous disruption of an autophagy gene, beclin (bcn ), in cancer cells , suggest that breakdown of autophagic machinery may contribute to development of cancer. other interesting studies have shed some light on the relationship between autophagy and apoptosis. these investigations have shown prevention of caspase inhibitor z-vad-induced cell death in mouse l cells by rna interference directed against autophagy genes atg and bcn and protection of bax −/− , bak −/− murine embryonic fi broblasts against staurosporine-or etoposide-induced cell death by rna interference against autophagy genes atg and bcn . however, both of these studies were done in cells whose apoptotic pathways had been compromised. thus, it remains to be seen whether cells with intact apoptotic machinery can also die by autophagy and whether apoptotic-competent cells lacking autophagy genes will be resistant to different death stimuli. paraptosis has recently been described as a form of cell death characterized by extensive cytoplasmic vacuolation (see figure . ) caused by swelling of mitochondria and endoplasmic reticulum. this form of cell death does not involve caspase activation, is not inhibited by caspase inhibitors, but is inhibited by the inhibitors of transcription and translation, actinomycin d, and cycloheximide, respectively, suggesting a requirement for new protein synthesis. the tumor necrosis factor receptor family taj/troy and the insulin-like growth factor i receptor have been shown to trigger paraptosis. paraptosis appears to be mediated by mitogen-activated protein kinases and inhibited by aip /alix, a protein interacting with the calcium-binding death-related protein alg- . autoschizis autoschizis is a recently described type of cell death that differs from apoptosis and necrosis and is induced by oxidative stress. in this type of death, cells lose cytoplasm by self-morsellation or self-excision (see figure . ). autoschizis usually affects contiguous groups of cells both in vitro and in vivo but can also occasionally affect scattered individual cells trapped in subcapsular sinuses of lymph nodes. the nuclear envelope and pores remain intact while the cytoplasm is reduced to a narrow rim surrounding the nucleus. the chromatin marginates along the nuclear membrane, and mitochondria and other organelles around the nucleus aggregate as a result of cytoskeletal damage and condensation of the cytosol. interestingly, the rough endoplasmic reticulum is preserved until the late stages of autoschizis, in which cells fragment and the nucleolus becomes condensed and breaks into smaller fragments. eventually, the nuclear envelope and the remaining organelles dissipate with cell demise. genetic studies in the nematode worm c. elegans led to the characterization of apoptosis. activation of specifi c death genes during the development of this worm results in death of exactly cells, leaving cells intact. further studies revealed that apoptosis can be divided into three successive stages: ( ) commitment phase, in which death is initiated by specifi c extracellular or intracellular signals; ( ) execution phase; and ( ) clean-up phase, in which dead cells are removed by other cells with eventual degradation of the dead cells in the lysosomes of phagocytic cells. the apoptotic machinery is conserved through evolution from worm to human. in c. elegans, execution of apoptosis is mediated by ced- and ced- proteins. commitment to a death signal results in the activation of ced- by ced- binding. the ced- protein prevents activation of ced- by binding to ced- . , mechanisms of apoptosis caspases studies over the past decade have indicated that two distinct apoptotic pathways are followed in mammalian systems: the extrinsic or death receptor pathway and the intrinsic or mitochondrial pathway. the executioners in both intrinsic and extrinsic pathways of cell death are the caspases, which are cysteine proteases with specifi city to cleave their substrates after aspartic acid residues. the central role of caspases in apoptosis is underscored by the observation that apoptosis and all classic changes associated with apoptosis can be blocked by inhibition of caspase activity. to date, mammalian caspases (caspase- to - , caspase- , and mouse caspase- ) have been identifi ed. caspase- was later found to represent a bovine homolog and caspase- appears to be a murine homolog of human caspases- and - , respectively. caspases are normally produced as inactive zymogens containing an n-terminal prodomain followed by a large and a small subunit that constitute the catalytic core of the protease. they have been categorized into two distinct classes: initiator and effector caspases. the upstream initiator caspases contain long n-terminal prodomains and one of the two characteristic protein-protein interaction motifs: the death effector domain (ded; caspase- and - ) and the caspase activation and recruitment domain (caspase- , - , - , - , - , and - ). the downstream effector caspases (caspase- , - , and - ) are characterized by the presence of a short prodomain. apart from the structural differences, a prominent difference between initiator and effector caspases is their basal state. both the zymogen and the activated forms of effector caspases exist as constitutive homodimers, whereas initiator caspase- exists predominantly as a monomer both before and after proteolytic processing. initiator caspase- has been reported to exist in an equilibrium between monomers and homodimers. although the initiator caspases are capable of autocatalytic activation, the activation of effector caspases requires formation of oligomeric complexes with their adapter proteins and often intrachain cleavage within the initiator caspase. caspases have also been divided into three categories based on substrate specifi city. group i members (caspase- , - , and - ) have a substrate specifi city for the wehd sequence with high promiscuity; group ii members (caspase- , - , and - and ced- ) prefer the dexd sequence and have an absolute requirement for aspartate (d) at p ; and members of group iii (caspase- , - , and - and the "aspase" granzyme b) have a preference for (i/ l/v)exd sequences. several reports have suggested a role for group i members in infl ammation and that of group ii and iii members in apoptotic signaling events. the extrinsic pathway involves binding of death ligands such as tumor necrosis factor-α (tnf-α), cd ligand (fas ligand), and tnf-related apoptosis-inducing ligand (trail) to their cognate cell surface receptors tnfr , cd /fas, trail-r , trail-r , and the dr series of receptors, resulting in the activation of initiator caspase- (also known as fadd-homologous ice/ced- -like protease or flice) and subsequent activation of effector caspase- ( figure . ). the cytoplasmic domains of death receptors contain the "death domain," which plays a crucial role in transmitting the signal from the cell's surface to intracellular signaling molecules. binding of the ligands to their cognate receptors results in receptor trimerization and recruitment of adapter proteins to the cell membrane, which involves homophilic interactions between death domains of the receptors and the adapter proteins. the adapter protein for the receptors tnfr and dr is tnfr-associated death domain protein (tradd) and that for fas, trail-r , trail-r , and dr is fas-associated death domain protein (fadd). the receptor/ligand and fadd complex in turn recruits caspase- to the activated receptor, resulting in the formation of death-inducing signaling complex (disc) and subsequent activation of caspase- through oligomerization and self-cleavage. depending on the cell type and/or apoptotic stimulus, caspase- can also be activated by caspase- . activated caspase- then activates effector caspase- . in some cell types, cleavage of caspase- by caspase- also requires a mitochondrial amplifi cation loop involving cleavage of proapoptotic protein bid by caspase- and its translocation to the mitochondrial membrane, triggering the release of apoptogenic proteins from mitochondria into cytosol (see figure . ). in these cell types, overexpression of bcl- and bcl-xl can block cd -induced apoptosis. tumor necrosis factor-α is produced by t cells and activated macrophages in response to infection. although tnf-α-mediated signaling can be propagated through either tnfr or tnfr receptors, the majority of biologic functions are initiated by tnfr . binding of tnf-α to tnfr causes release of inhibitory protein silencer of death domain protein (sodd) from tnfr , which enables recruitment of adapter protein tradd. signaling induced by activation of tnfr or dr diverges at the level of tradd. in one pathway, nuclear translocation of the transcription factor nuclear factor-κb (nf-κb) and activation of c-jun n-terminal kinase (jnk) are initiated, which results in the induction of a number of proinfl ammatory and immunomodulatory genes. in another pathway, tnf-α signaling is coupled to fas signaling events through interaction of tradd with fadd. the tnfr -tradd complex can alternatively engage traf protein, resulting in activation of transcription factor c-jun, which is involved in survival signaling. furthermore, binding of receptor interaction protein to tnfr through tradd results in activation of transcription factor nf-κb, which suppresses apoptosis through transcriptional upregulation of antiapoptotic molecules such as traf , traf , ciap , ciap , and flip. the flice-associated huge protein was identifi ed to be a ced- homolog interacting with the ded of caspase- and was shown to modulate fas-mediated activation of caspase- . another class of protein, flip (flice inhibitory protein), was shown to block fasinduced and tnf-α-induced disc formation and subsequent activation of caspase- . cytotoxic t cells play a major role in vertebrate defense against viral infection. they induce cell death in infected cells to prevent viral multiplication and spread of infection. cytotoxic t cells can kill their targets either by activating the fas ligand/fas pathway or by injecting granzyme b, a serine protease, into target cells. cytotoxic t cells carry fas ligand on their surface but also carry granules containing the channel-forming protein perforin and granzyme b. upon recognizing the infected cells, the lymphocytes bind and secrete granules onto the surface of infected cells. perforin then assembles into transmembrane channels to allow the entry of granzyme b into the target cell. upon entry, granzyme b, which cleaves after aspartate residues in proteins ("aspase"), activates one or more of the apoptotic proteases (caspase- , - , - , - , and - ) to trigger the proteolytic death cascade (see figure . ). fas ligand/fas and perforin/granzyme b systems are the main apoptotic machinery that regulates homeostasis in immune cell populations. cells can respond to various stressful stimuli and metabolic disturbances by triggering apoptosis. drugs, toxins, heat, radiation, hypoxia, and viral infections are some of the tnf-α tnfr complex can also elicit an antiapoptotic response by recruiting traf , which results in nf-κbmediated upregulation of antiapoptotic genes. in cytotoxic t lymphocyte-induced death, granzyme b, which enters the cell through membrane channels formed by the protein perforin, activates caspases by cleaving them directly or indirectly. intracellular pathways: lack of survival stimuli (withdrawal of growth factor, hypoxia, genotoxic substances, etc.) is thought to generate apoptotic signals through ill-defi ned mechanisms, which lead to translocation of proapoptotic proteins such as bax to the outer mitochondrial membrane. in some cases, transcription mediated by p may be required to induce proteins such as bax. translocated bax undergoes conformational changes in the outer membrane to form oligomeric structures (pores) that leak cytochrome c from mitochondria into the cytosol. formation of a ternary complex of cytochrome c, the adapter protein apaf- , and the initiator caspase- results in the activation of caspase- followed by sequential activation of effector caspase(s) such as caspase- and others. the action of caspases, endonucleases, and possibly other enzymes leads to cellular disintegration. for example, the endonuclease cad (caspase activated dnase) becomes activated when it is released from its inhibitor icad upon cleavage of icad by an effector caspase. antiapoptotic proteins such as bcl- and bcl-xl inhibit the membrane-permeabilizing effects of bax and other proapoptotic proteins. cross-talk between extra-and intracellular pathways occurs through caspase- -mediated bid cleavage, which yields a kda protein that migrates to mitochondria and releases cytochrome c, thereby setting in motion events that lead to apoptosis via caspase- . the stimuli known to activate death pathways. cell death, however, is not necessarily inevitable after exposure to these agents, and the mechanisms determining the outcome of the injury are a topic of active interest. the current consensus appears to be that it is the intensity and the duration of the stimulus that determine the outcome. the stimulus must go beyond a threshold to commit cells to apoptosis. although the exact mechanism used by each stimulus may be unique and different, a few broad patterns can be identifi ed. for example, agents that damage dna, such as ionizing radiation and certain xenobiotics, lead to activation of p -mediated mechanisms that commit cells to apoptosis, at least in part through transcriptional upregulation of proapoptotic proteins. other stresses induce increased activity of stress-activated protein kinases, which result ultimately in apoptotic commitment. these different mechanisms converge in the activation of caspases. a cascade of caspases plays the central executioner role by cleaving various mammalian cytosolic and nuclear proteins that play roles in cell division, maintenance of cytoskeletal structure, dna replication and repair, rna splicing, and other cellular processes. this proteolytic carnage produces the characteristic morphologic changes of apoptosis. once the caspase cascade is initiated, the process of cell death has crossed the point of no return. the roles of various caspases in apoptotic pathways and their relative importance for animal development have been examined in genetic studies involving knockout of different caspase genes. a caspase- (interleukin [il]- b converting enzyme [ice]) knockout study suggested that ice plays an important role in infl ammation by activating cytokines such as il- b and il- . however, caspase- was not required to mediate apoptosis under normal circumstances and did not have a major role during development. surprisingly, ischemic brain injury was signifi cantly reduced in caspase- knockout mice compared with wild-type mice, suggesting that infl ammation may contribute to ischemic injury. caspase- deficiency leads to impaired brain development and premature death. also, functional caspase- is required for some typical hallmarks of apoptosis such as formation of apoptotic bodies, chromatin condensation, and dna fragmentation in many cell types. lack of caspase- results in the death of embryos at day with abnormal formation of the heart, suggesting that caspase- is required for cell death during mammalian development. in support of this fi nding, knockout of fadd, which is required for caspase- activation, resulted in fetal death with signs of abdominal hemorrhage and cardiac failure. moreover, caspase- -defi cient cells did not die in response to signals from members of the tnf receptor family. however, cells lacking either fadd or caspase- , which are resistant to tnf-α-mediated or cd -mediated death, are susceptible to chemotherapeutic drugs, serum depriva-tion, ceramide, γ-irradiation, and dexamethasone-induced killing. in contrast, caspase- has a key role in apoptosis induced by intracellular activators, particularly those that cause dna damage. deletion of caspase- resulted in perinatal lethality, apoptotic failure in developing neurons, enlarged brains, and craniofacial abnormalities. in caspase- -defi cient cells, caspase- was not activated, suggesting that caspase- is upstream of caspase- in the apoptotic cascade. as a consequence, caspase- -defi cient cells are resistant to dexamethasone or irradiation, whereas they retain their sensitivity to tnf-α-induced or cd -induced death because of the presence of caspase- , the initiator caspase involved in death receptor signaling that can also activate caspase- . overall, these observations support the idea that different death signaling pathways converge on downstream effector caspases (see figure . ). indeed, caspase- is regarded as one of the key executioner molecules activated by apoptotic stimuli originating either at receptors for exogenous molecules or within cells through the action of drugs, toxins, or radiation. in c. elegans, biochemical and genetic studies have indicated a role for ced- upstream of ced- . upon receiving death commitment signals, ced- binds to pro-ced- and releases active ced- . however, when overexpressed, ced- can inhibit the activation of pro-ced- by binding to ced- and sequestering it away from pro-ced- . therefore, ced- and ced- are involved in activation of apoptosis, and ced- inhibits apoptosis. after the discovery of caspases as ced- homologs, a search for activators and inhibitors analogous to ced- and ced- led to the discovery of diverse mammalian regulators of apoptosis. the plethora of these molecules and their functional diversity allowed them to be classifi ed into four broad categories: ( ) adapter proteins, ( ) the bcl- family of regulators, ( ) inhibitors of apoptosis (iaps), and ( ) other regulators. as stated earlier, two major pathways of apoptosis, involving either the initiator caspase- or the initiator caspase- (see figure . ), have been recognized. signaling by death receptors (cd , tnfri) occurs through a well-defi ned process of recruitment of caspase- to the death receptor by adapter proteins such as fadd. recruitment occurs through interactions between the death domains that are present on both receptor and adapter proteins. receptorbound fadd then recruits caspase- through interactions between deds common to both caspase- and fadd forming a disc. in the disc, caspase- activation occurs through oligomerization and autocatalysis. activated caspase- then activates downstream caspase- , culminating in apoptosis. the inhibitory protein, flip was shown to block fas-induced and tnf-α-induced disc formation and subsequent activation of caspase- . of particular interest is cellular flip, which stimulates caspase- activation at physiologically relevant levels and inhibited apoptosis upon high ectopic expression. cellular flip contains two deds that can compete with caspase- for recruitment to the disc. this limits the degree of association of caspase- with fadd and thus limits activation of the caspase cascade. it also forms a heterodimer with caspase- and caspase- through interactions between both the deds and the caspase-like domains of the proteins, thus activating both caspase- and caspase- . apoptotic protease activating factor- (apaf- ), a ced- homolog in mammalian cells, affects the activation of initiator caspase- . this factor binds to procaspase- in the presence of cytochrome c and ′deoxyadenosine ′-triphosphate (datp) or adenosine triphosphate (atp) and activates this protease, which in turn activates a downstream cascade of proteases (see figure . ). by and large, apaf- defi ciency is embryonically lethal and the embryos exhibit brain abnormalities similar to those seen in caspase- knockout mice. these genetic fi ndings support the idea that apaf- is coupled to caspase- in the death pathway. unlike ced- in nematodes, apaf- requires the binding of atp and cytochrome c to activate procaspase- . the multiple wd repeats in the c-terminal end of apaf- have a regulatory role in the activation of caspase- . the ced- homolog in mammals is the bcl- protein. bcl- was fi rst discovered in b-cell lymphoma as a protooncogene. overexpression of bcl- was shown to offer protection against a variety of death stimuli. the bcl- protein family includes both proapoptotic (bcl- , bcl-xl, bcl-w, mcl- , nr , and a /bfl - ) and antiapoptotic proteins (bax, bak, bok, diva, bcl-xs, bik, bim, hrk, nip , nix, bad, and bid). these proteins are characterized by the presence of bcl- homology (bh) domains: bh , bh , bh , and bh (figure . ) . the proapoptotic members have two subfamilies: a multidomain and a bh -only group (see figure . ). the relative ratio of pro-and antiapoptotic proteins determines the sensitivity of cells to various apoptotic stimuli. the best-studied proapoptotic members are bax and bid. exposure to various apoptotic stimuli leads to translocation of cytosolic bax from the cytosol to the mitochondrial membrane. bax oligomerizes on the mitochondrial membrane along with another proapoptotic protein, bak, leading to the release of cytochrome c from the mitochondrial membrane into the cytosol. other proapoptotic proteins, mainly the bh -only proteins, are thought to aid in bax-bak oligomerization on the mitochondrial membrane. the antiapoptotic bcl- family members are known to block bax-bak oligomerization on the mitochondrial membrane and subsequent release of cytochrome c into the cytosol. , after release from the mitochondria, cytochrome c is known to interact with the wd repeats of the adaptor protein apaf- , resulting in the formation of the apoptosome complex. seven molecules of apaf- , interacting through their n-terminal caspase activation and recruitment domain, form the central hub region of the symmetric wheel-like structure, the apoptosome. binding of atp/datp to apaf- triggers the formation of the apoptosome, which subsequently recruits procaspase- into the apoptosome complex, resulting in its activation . activated caspase- then activates executioner caspases, such as caspase- and caspase- , eventually leading to programmed cell death. the iaps, fi rst discovered in baculoviruses and then in insects and drosophila, inhibit activated caspases by directly binding to the active enzymes. these proteins contain one or more baculovirus inhibitor of apoptosis repeat domains, which are responsible for the caspase inhibitory activity. to date, eight mammalian iaps have been identifi ed. they include x-linked iap (xiap), c-iap , c-iap , melanoma iap (ml-iap)/livin, iaplike protein- (ilp- ), neuronal apoptosis-inhibitory protein (naip), bruce/apollon, and survivin. in mammals, caspase- , - , and - are inhibited by iaps. there are reports suggesting aberrant expression of iaps in many cancer tissues. for example, ciap is overexpressed in esophageal squamous cell sarcoma ; ciap locus is translocated in mucosa-associate lymphoid lymphoma and survivin has been shown to be upregulated in many cancer cells. the caspase inhibitory activity of iaps is inhibited by proteins containing an iap-binding tetrapeptide motif. the founding member of this family is smac/diablo, which is released from the mitochondrial intermembrane space into the cytosol during apoptosis. in the cytosol, it interacts with several iaps and inhibits their function. the other mitochondrial protein, omi/htra , is also known to antagonize xiap-mediated inhibition of caspase- at high concentrations. a serine protease, omi/htra can proteolytically cleave and inactivate iap proteins and thus is considered to be a more potent suppressor of iaps than smac. it has been reported that the heat shock proteins hsp , hsp , and hsp can inhibit caspase activation by cytochrome c either by interacting with apaf- or other players in the pathway. [ ] [ ] [ ] a high-throughput screen identifi ed a compound called petcm (α-[trichloromethyl]- -pyridineethanol) as a caspase- activator. further work with petcm revealed its involvement in apoptosome regulation. this pathway also includes oncoprotein prothymosin-α and tumor suppressor putative hla-dr-associated proteins. these proteins were shown to promote caspase- activation after apoptosome formation, whereas prothymosin-α inhibited caspase- activation by inhibiting apoptosome formation. in an apoptotic cell, the regulatory, structural, and housekeeping proteins are the main targets of the caspases. the regulatory proteins mitogen-activated protein/extracellular signal-regulated kinase kinase- , p -activated kinase- , and mst- are activated upon cleavage by caspases. caspase-mediated protein hydrolysis inactivates other proteins, including focal adhesion kinase, phosphatidylinositol- kinase, akt, raf- , iaps, and inhibitors of caspase-activated dnase (icad). caspases also convert the antiapoptotic protein bcl- into a proapoptotic protein such as bax upon cleavage. there are many structural protein targets of caspases, which include nuclear lamins, actin, and regulatory proteins such as spectrin, gelsolin, and fodrins. degradation of nuclear dna into internucleosomal chromatin fragments is one of the hallmarks of apoptotic cell death that occurs in response to various apoptotic stimuli in a wide variety of cells. a specifi c dnase, cad (caspase-activated dnase), that cleaves chromosomal dna in a caspase-dependent manner, is synthesized with the help of icad. in proliferating cells, cad is always found to be associated with icad in the cytosol. when cells are undergoing apoptosis, caspases (particularly caspase- ) cleave icad to release cad and allow its translocation to the nucleus to cleave chromosomal dna. thus, cells that are icad defi cient or that express caspase-resistant icad mutant do not exhibit dna fragmentation during apoptosis. apoptosis plays a critical role in the postnatal lung. regulated removal of infl ammatory cells by apoptosis helps in the resolution of infl ammation in the lung. recent evidence also supports a role for apoptosis in the remodeling of lung tissue after acute lung injury and in the pathogenesis of chronic pulmonary hypertension, idiopathic pulmonary fi brosis, and chronic obstructive pulmonary disease. , acute lung injury/acute respiratory distress syndrome acute lung injury, which clinically manifests itself as the acute respiratory distress syndrome (ards), involves disruption of the alveolar epithelium and endothelium, increased vascular permeability, and edema. two main hypotheses link the pathogenesis of ards to apoptosis, namely, the "neutrophilic hypothesis" and the "epithelial hypothesis." these two hypotheses are not mutually exclusive, and both could play important roles in the pathogenesis of ards. the neutrophilic hypothesis suggests that neutrophil apoptosis plays an important role in the resolution of infl ammation and that the inhibition of neutrophil apoptosis or the inhibition of clearance of apoptotic neutrophils is deleterious in ards. , studies in humans showed that bronchoalveolar lavage fl uids from patients with early ards inhibit the rate at which neutrophils develop apoptosis in vitro. the inhibitory effect of bronchoalveolar lavage fl uids on neutrophil apoptosis is mediated by granulocyte/macrophage colony-stimulating factor, and possibly by il- and il- . , a membrane surface molecule, cd , has been shown to play an important role in the clearance of apoptotic cells in vivo and in vitro. in a model of bleomycin-induced lung injury, cd -defi cient mice failed to clear apoptotic neutrophils, which was associated with worsened infl ammation and increased mortality. activation of phagocytic cells inhibits production of proinfl ammatory cytokines, including il- β, il- , il- , granulocyte/ macrophage colony-stimulating factor, and tnf-α and increases release of anti-infl ammatory mediators such as transforming growth factor-β, prostaglandin e , and platelet-activating factor. , the net effects of these changes could favor resolution of infl ammation. the epithelial hypothesis suggests that the apoptotic death of alveolar epithelial cells, in response to soluble mediators such as fas ligand, contributes to the prominent alveolar epithelial injury characteristic of ards. several lines of evidence suggest a role for the fas/fas ligand system in epithelial cell apoptosis. fas is expressed on alveolar and airway epithelial cells, , and its expression increases in response to infl ammatory mediators such as lipopolysaccharide. fas-mediated lung cell apoptosis is modulated by surfactant protein a, which inhibits apoptosis in vivo. chronic obstructive pulmonary disease chronic obstructive pulmonary disease, caused primarily by smoking, generally refers to chronic bronchitis and emphysema. several factors, including protease/antiprotease imbalance, oxidative stress, cigarette smokederived toxins, and infl ammation mediated by neutrophils, macrophages, and cd + t cells, have been shown to contribute to the disease process. furthermore, matrix metalloproteinase and vascular endothelial growth factor receptor inhibition, , but not fas/fas ligand, have been shown to play role in the development of emphysema. asthma allergic asthma is characterized by intermittent or persistent bronchoconstriction and has been linked to airway remodeling and chronic infl ammation, with increased numbers of eosinophils, cd + t cells, and mast cells. although at present a role for apoptosis in asthma is not confi rmed, studies ex vivo have shown reduced apoptosis of circulating peripheral cd + t cells and eosinophils in asthma, which might contribute to infl ammation. corticosteroids used to reduce infl ammation in asthma have been shown to induce eosinophil apoptosis. pulmonary fi brosis is characterized by epithelial damage, fi broblast proliferation, and deposition of collagen. although the mechanism of alveolar epithelial cell apoptosis in pulmonary fi brosis is not known, several reports have suggested fas pathway, angiotensin pathway, activated t cell-derived perforin, il- stimulation, and transforming growth factor-β activation to play critical roles. because insuffi cient apoptosis is often associated with tumorigenesis, modulation of apoptotic and antiapoptotic targets seems to be an attractive approach to cancer therapy. lung cancers can be divided into small cell lung cancers (sclcs) and non-small cell lung cancers (nsclcs). the sclcs are relatively more sensitive to anticancer drugs and irradiation than are the nsclcs, but the molecular basis for this difference is not clearly known. evaluation of apoptosis-associated substances has shown that caspase- , fas, and fas ligand are often downregulated in sclcs but not in nsclcs. an investigation of the basis for these differences revealed that there were no differences in the levels of bax and bcl-xl, but the expression of bcl- was found to be signifi cantly higher in sclc than in nsclc cell lines. the observation that in some cases bcl- can be converted into a proapoptotic bax-like death molecule may offer an explanation for the paradoxic expression of bcl- in sclc. the lack of expression of procaspase- , - , - , and - reported in sclc suggests that these caspases probably do not contribute to spontaneous apoptosis in these cells. apoptosis regulators apaf- and procaspase- are overexpressed and are functional in nsclc cell lines. in both types of lung cancer, apoptotic stimuli result in cytochrome c release and activation of caspase- and caspase- , but only sclc cell lines showed a relocalization of caspase- into the nucleus ; this suggests that the resistance of nsclc cell lines is probably due to defective relocalization of caspase- . the expression of caspase- and caspase- in nsclcs was found to be similar to normal lung tissue. however, these cell lines express the apoptosis inhibitor and splice variant of caspase- casp b. in vitro, chemotherapy-resistant nsclc cell lines exhibit decreased caspase- and caspase- expression, which suggests an inhibition of apoptosis induction via apoptosome formation in nsclc. additionally, both nsclc and sclc cells express high and almost equal levels of survivin. the resistant nsclc cells showed higher expression of c-iap , and the radiosensitive sclc cells exhibited increased expression of xiap. these results suggest no correlation between the level of expression of the iaps and the difference in the radiosensitivity between nsclc and sclc cells. cell death has become an area of intense interest and investigation in science and medicine because of the recognition that cell death, in general, and apoptosis, in par-ticular, are important features of many biologic processes. involvement of many genes in the death process suggests that cell death is a complex phenomenon with many redundant mechanisms to ensure defi nitiveness. the realization that defective cell death plays a central role in the pathogenesis of diseases has stimulated work on therapies targeted to these processes, and this work will undoubtedly continue in the future. apoptosis: a basic biological phenomenon with wide-ranging implications in tissue kinetics genetic control of programmed cell death in the nematode c. elegans programmed cell death in animal development apoptosis: defi nition, mechanisms, and relevance to disease apoptosis as a therapeutic target for the treatment of lung disease four deaths and a funeral: from caspases to alternative mechanisms dual signaling of the fas receptor: initiation of both apoptotic and necrotic cell death pathways alkylating dna damage stimulates a regulated form of necrotic cell death a novel response of cancer cells to radiation involves autophagy and formation of acidic vesicles autophagy: in sickness and in health tissue protein turnover during liver carcinogenesis reduced autophagic activity in primary rat hepatocellular carcinoma and ascites hepatoma cells regulation of an atg -beclin program of autophagic cell death by caspase- role of bcl- family proteins in a non-apoptotic programmed cell death dependent on autophagy genes an alternative, nonapoptotic form of programmed cell death paraptosis: mediation by map kinases and inhibition by aip- /alix autoschizis: a novel cell death inhibition of the development of metastases by dietary vitamin c:k combination autoschizis: a new form of cell death for human ovarian carcinoma cells following ascorbate/menadione treatment. nuclear and dna degradation the molecular biology of apoptosis evolutionary conservation of a genetic pathway of programmed cell death interaction between the c. elegans cell-death regulators ced- and ced- interaction and regulation of the caenorhabditis elegans death protease ced- by ced- and ced- caspases: enemies within vital functions for lethal caspases mechanism of xiapmediated inhibition of caspase- insights into the regulatory mechanism for caspase- activation a combinatorial approach defi nes specifi cities of members of the caspase family and granzyme b. functional relationships established for key mediators of apoptosis signalling by cd and tnf receptors: not only life and death apoptosis control by death and decoy receptors the tnf receptor -associated protein tradd signals cell death and nf-kappa b activation fadd, a novel death domain-containing protein, interacts with the death domain of fas and initiates apoptosis caspase- is the direct activator of caspase- in the cytochrome c-induced apoptosis pathway: absolute requirement for removal of caspase- prodomain two cd (apo- /fas) signaling pathways induction of cell death by tumour necrosis factor (tnf) receptor , cd and cd : a role for tnf-r activation by endogenous membrane-anchored tnf tumor necrosis factor (tnf) receptor signaling downstream of tnf receptor-associated factor . nuclear factor kappab (nfkappab)-inducing kinase requirement for activation of activating protein and nfkappab but not of c-jun nterminal kinase/stress-activated protein kinase involvement of mach, a novel mort /fadd-interacting protease, in fas/apo- -and tnf receptor-induced cell death the ced- -homologous protein flash is involved in fas-mediated activation of caspase- during apoptosis viral fliceinhibitory proteins (flips) prevent apoptosis induced by death receptors memory and distribution of virus-specifi c cytotoxic t lymphocytes (ctls) and ctl precursors after rotavirus infection fasdependent cd + cytotoxic t-cell-mediated pathogenesis during virus infection transcriptional regulation during p waf /cip -induced apoptosis in human ovarian cancer cells activation of c-jun nh -terminal kinase/stress-activated protein kinase (jnk/ sapk) is critical for hypoxia-induced apoptosis of human malignant melanoma characterization of mice defi cient in interleukin- beta converting enzyme reduced ischemic brain injury in interleukin- beta converting enzyme-defi cient mice caspase- is required for dna fragmentation and morphological changes associated with apoptosis targeted disruption of the mouse caspase gene ablates cell death induction by the tnf receptors, fas/apo , and dr and is lethal prenatally fadd: essential for embryo development and signaling from some, but not all, inducers of apoptosis reduced apoptosis and cytochrome c-mediated caspase activation in mice lacking caspase the ins and outs of programmed cell death during c. elegans development c-flip(l) is a dual function regulator for caspase- activation and cd -mediated apoptosis the fl ip side of flip apaf- , a human protein homologous to c. elegans ced- , participates in cytochrome c-dependent activation of caspase- an apaf- .cytochrome c multimeric complex is a functional apoptosome that activates procaspase- apaf (ced- homolog) regulates programmed cell death in mammalian development autoactivation of procaspase- by apaf- -mediated oligomerization bcl- inhibits death of central neural cells induced by multiple agents bcl- family proteins role of hypoxia-induced bax translocation and cytochrome c release in reoxygenation injury association of bax and bak homo-oligomers in mitochondria. bax requirement for bak reorganization and cytochrome c release bcl- prevents bax oligomerization in the mitochondrial outer membrane mechanisms of caspase activation and inhibition during apoptosis diablo promotes apoptosis by removing miha/xiap from processed caspase iap family proteins-suppressors of apoptosis identifi cation of ciap as a candidate target gene within an amplicon at q in esophageal squamous cell carcinomas the apoptosis inhibitor gene api and a novel q gene, mlt, are recurrently rearranged in the t( ; )(q ;q ) associated with mucosa-associated lymphoid tissue lymphomas a novel anti-apoptosis gene, survivin, expressed in cancer and lymphoma a serine protease, htra , is released from the mitochondria and interacts with xiap, inducing cell death omi/htra catalytic cleavage of inhibitor of apoptosis (iap) irreversibly inactivates iaps and facilitates caspase activity in apoptosis hsp functions as a negative regulator of cytochrome c-dependent activation of procaspase- heat-shock protein inhibits apoptosis by preventing recruitment of procaspase- to the apaf- apoptosome negative regulation of cytochrome c-mediated oligomerization of apaf- and activation of procaspase- by heat shock protein distinctive roles of phap proteins and prothymosin-alpha in a death regulatory pathway caspase-dependent cleavage of signaling proteins during apoptosis. a turn-off mechanism for anti-apoptotic signals caspasemediated proteolysis during apoptosis: insights from apoptotic neutrophils programmed cell death contributes to postnatal lung development granulocyte apoptosis and its role in the resolution and control of lung infl ammation apoptosis is a major pathway responsible for the resolution of type ii pneumocytes in acute lung injury mechanisms of structural remodeling in chronic pulmonary hypertension induction of apoptosis and pulmonary fi brosis in mice in response to ligation of fas antigen essential roles of the fas-fas ligand pathway in the development of pulmonary fi brosis granulocyte apoptosis and the control of infl ammation macrophage engulfment of apoptotic neutrophils contributes to the resolution of acute pulmonary infl ammation in vivo modulation of neutrophil apoptosis by granulocyte colony-stimulating factor and granulocyte/macrophage colony-stimulating factor during the course of acute respiratory distress syndrome g-csf and il- but not gm-csf correlate with severity of pulmonary neutrophilia in acute respiratory distress syndrome interleukin- involvement in early acute respiratory distress syndrome: relationship with polymorphonuclear neutrophil apoptosis and patient survival resolution of lung infl ammation by cd macrophages that have ingested apoptotic cells in vitro inhibit proinfl ammatory cytokine production through autocrine/paracrine mechanisms involving tgf-beta, pge , and paf phosphatidylserinedependent ingestion of apoptotic cells promotes tgf-beta secretion and the resolution of infl ammation recombinant human fas ligand induces alveolar epithelial cell apoptosis and lung injury in rabbits fas expression in pulmonary alveolar type ii cells expression of fas (cd ) and fasl (cd l) in human airway epithelium natural protection from apoptosis by surfactant protein a in type ii pneumocytes upregulation of gelatinases a and b, collagenases and , and increased parenchymal cell death in copd inhibition of vegf receptors causes lung cell apoptosis and emphysema oxidative stress and apoptosis interact and cause emphysema due to vascular endothelial growth factor receptor blockade glucocorticoid-induced apoptosis in human eosinophils: mechanisms of action increased circulating levels of soluble fas ligand are correlated with disease activity in patients with fi brosing lung diseases bleomycin-induced apoptosis of alveolar epithelial cells requires angiotensin synthesis de novo the perforin mediated apoptotic pathway in lung injury and fi brosis interleukin- induces tissue fi brosis by selectively stimulating and activating transforming growth factor beta( ) early growth response gene -mediated apoptosis is essential for transforming growth factor beta -induced pulmonary fi brosis united states lung carcinoma incidence trends: declining for most histologic types among males, increasing among females progress in understanding the molecular pathogenesis of human lung cancer loss of expression of death-inducing signaling complex (disc) components in lung cancer cell lines and the infl uence of myc amplifi cation conversion of bcl- to a bax-like death effector by caspases differences in expression of pro-caspases in small cell and non-small cell lung carcinoma defective caspase- relocalization in non-small cell lung carcinoma increased expression of apaf- and procaspase- and the functionality of intrinsic apoptosis apparatus in non-small cell lung carcinoma rescue of death receptor and mitochondrial apoptosis signaling in resistant human nsclc in vivo expression of inhibitor of apoptosis proteins in small-and non-small-cell lung carcinoma cells key: cord- - wfyaxcb authors: ubokudom, sunday e. title: physical, social and cultural, and global influences date: - - journal: united states health care policymaking doi: . / - - - - _ sha: doc_id: cord_uid: wfyaxcb in chap. , we examined the technological environment of the health care policy-making system. specifically, we examined the classification, evolution, and diffusion of medical technology; the effects of medical technology on medical training and the practice of medicine; effects on medical costs, quality of care, and quality of life; effects on access to care; the ethical concerns raised by medical technology; and the practice of technology assessment. we concluded the chapter by observing that the growth of technology, as well as other human endeavors, affects other important aspects of our lives, most notably, the air we breathe, the food we eat, the generation of radioactive by-products and toxic chemicals, the manufacture of illicit drugs, and the generation of natural and man-made hazards. in other words, in addition to their effects on the health care system, technology and other human activities affect many other aspects of our lives that are associated with health. the who's defi nition of health as "a complete state of physical, mental, and social well-being, and not merely the absence of disease or infi rmity" (who ) , is primarily based on the wellness model. in this defi nition, emphasis is put on the fact that health is not merely the absence of disease, but also involves a social dimension. therefore, it also emphasizes the social and fi nancial support systems identifi ed in table . of chap. . this defi nition of health, as involving the combination of physical, mental, and social well-being led to the concept of the "health triangle." the health triangle left out the spiritual dimension of health, which has recently gained signifi cant attention in the literature due to a growing interest in the notion of holistic health. holistic health stresses the importance of all the things that make a person whole and complete. in addition to the three dimensions of the health triangle, of his analysis (szreter , p. ) . subsequent studies revealed that the cessation of the large-scale redistribution of income and wealth from the very rich to the poorest in society had adverse effects on the health of the population. for example, when unhealthy behaviors and lifestyles were held as constant as possible, studies showed that people of lower socioeconomic status were more likely to die prematurely than were people of higher socioeconomic status (isaacs and schroeder , p. ; smith et al. , p. ; davey smith et al. , p. ) . the relationship between physical, social and cultural, and global environmental factors and health status is very well documented. in a letter to the editor of the jama , winkelstein ( winkelstein ( , p. argues that curative medical care, or those practices that are used for the care and rehabilitation of the sick, which involve most of the physical and designed social technologies listed in table . of the previous chapter, is not the same as health care. medical care, as he defi nes it, makes only modest contributions to the health status of the population. on the contrary, the health status of the population is largely determined by a different set of factors that involve important physical, social, and economic components. these include preventive medicine, genetic predisposition, social and economic circumstances, environmental conditions, lifestyles and behaviors, and medical care (mckeown ; kannel et al. ; belloc and breslow , p. ; bunker et al. ; bunker et al. , p. ; marmot et al. marmot et al. , p. bell and standish , p. ; mcginnis et al. , p. ; wilkinson wilkinson , p. . we briefl y examine each of the identifi ed determinants of health below. preventive medicine seeks to minimize the occurrence of illness and disease. unlike the medical model that is reactive and seeks to contain disease and ill-health after they have occurred, preventive medicine is proactive and seeks to minimize the likelihood of the occurrence of disease and ill-health. generally, there are three areas or types of preventive measures, namely: primary prevention, secondary prevention, and tertiary prevention. primary prevention seeks to stop or minimize the development of disease or ill-health before it occurs. primary prevention may involve counseling against smoking, in order to prevent the development of chronic emphysema or chronic obstructive pulmonary disease (copd) and lung cancer. other primary interventions may include the promotion of an active lifestyle or exercise program, in order to minimize the likelihood of excess body fat and heart disease; driver education and mandatory seatbelt and motorcycle helmet laws, in order to reduce motor vehicle accidents and accidental head injuries; vaccinations for various forms of diseases and illnesses, such as measles and rubella, which can minimize the occurrence of early childhood diseases and mortality; and water purifi cation and sewage treatment programs that can minimize the occurrence of typhoid, cholera, and other waterborne diseases. secondary prevention involves the early detection and treatment of disease. health screenings and periodic and regular health examinations, such as hypertension screenings, mammograms, and pap smears, serve as examples of secondary prevention measures. these examples fall under the broad category of health promotion discussed in chap. . the benefi ciaries of these programs are currently healthy people who are targeted to improve their health-related behaviors in order to minimize their chances of developing catastrophic and expensive illnesses. as was discussed in chap. , secondary prevention measures are some of the most cost-effective steps employers take to lower their health benefi t costs ( coffi eld et al. , p. ) . tertiary prevention measures involve steps taken to reduce the complications of diseases or illnesses, or to prevent further illnesses. they involve rehabilitative practices and the monitoring of the process of health care delivery. the infection control practices in hospitals and other improvements in the methods of health care delivery discussed in chap. , under the postindustrial period of the evolution of the health care system, which are intended to reduce the occurrences of nosocomial infections and iatrogenic illnesses, are practical examples of tertiary prevention measures. other examples include patient education, nutrition counseling, and behavior modifi cation programs that seek to prevent the recurrence of disease and illness (timmreck , p. ) . since the mid- s in the united states, there have been signifi cant reductions in heart disease, stroke, personal injury, and non-tobacco-related death rates foege , p. ; banta and jonas , p. ) . similarly, the data presented in table . of chap. show signifi cant declines in death rates related to heart disease, cancer, stroke, infl uenza and pneumonia, chronic liver disease or cirrhosis, human immunodefi ciency virus (hiv) disease, suicide and homicides, from to . these particular declines appear to be the result of preventive health measures, such as early screening, detection and treatment of hypertension, the provision and utilization of pneumonia and infl uenza vaccinations, moderate alcohol intake or abstinence, safe sex practices, suicide prevention and anger management programs, increased use of seatbelts and reductions in driving-underthe-infl uence episodes, smoking cessation, and the lowering of dietary fat and cholesterol. if, at least, some of the declines in mortality discussed above are due to preventive measures, the preventive strategy has yielded signifi cant gains in health. perhaps, it is this recognition of the importance of preventive services that led to the establishment of the us preventive services task force (uspstf) in . most likely, it was the recognition of the crucial role that preventive medicine plays in enhancing population health that led to the convening of the uspstf in by the us public health service. the task force is a leading independent panel of nationally recognized nonfederal experts in prevention and evidencebased medicine. programmatic responsibility for the task force was transferred to the agency for health care research and quality (ahrq) in (uspstf procedure manual ). the uspstf is assigned the responsibility of making evidence-based recommendations that address primary and secondary preventive services targeting conditions that represent a substantial burden in the country, and that are provided in primary care delivery settings or made available through primary care referrals. the task force's recommendations are intended to improve clinical practice and promote the public health. tertiary prevention measures are outside the scope of the uspstf. even though the main audience for task force recommendations is the primary care provider, the recommendations are also used to guide programmatic, funding, and reimbursement decisions by policy-makers, managed care organizations, public and private payers, quality improvement organizations, research institutions, and consumers. beginning at the end of may , the uspstf changed the grades it assigns to its recommendations. it assigns one of fi ve possible letter grades, a, b, c, d, or i, to each of its recommendations, including "suggestions for practice" associated with each grade. the agency also defi nes the levels of certainty regarding the net benefi t of each of its recommendations. the task force's reduction of the grade given for evidence quality from "b" to "c" for routine mammograms in women under the age of years generated signifi cant controversy among health professionals and politicians (kinsman ) . in addition to the mammography recommendations stated above, the uspstf has recently recommended against screening for testicular cancer in adolescent or adult males (grade d recommendation) (uspstf , p. ) . it has also concluded that there was insuffi cient evidence to assess the balance of benefi ts and harms of screening for bladder cancer in asymptomatic adults (moyer , p. ) , and that prostate-specifi c antigen (psa) screening was associated with psychological harms, while its potential benefi ts remained uncertain (lin et al. , p. ) . table . shows the approach adopted by the agency in june , to rank its recommendations. health is dependent upon biological factors. our predispositions to health or disease begin to take shape at the moment of conception. these predispositions are embedded in our genetic code. the genetic code guides the development of the proteins that determine our phenotypes (sizes, shapes, personalities, hair color, etc.) and genotypes or those aspects of our genetic codes that we cannot see, such as the biologic limit of our life expectancies (mcginnis et al. , p. ; khoury et al. ; bell and standish , p. ; starfi eld , p. ; blum ; centers for disease control and prevention (cdc) ) . genetic factors predispose individuals to certain diseases. but although an individual may have a strong likelihood of developing a particular disease, this propensity to develop the disease is signifi cantly enhanced by environmental factors. for example, some studies demonstrate that there is a genetic basis for alcoholism (reich ) . but a person who has never taken a drink will not become an alcoholic. some triggers, in this case, the availability and consumption of alcohol, are necessary for the individual to progress from being genetically predisposed to alcoholism to actually (berkman and breslow ; burnett ; banta and jonas , p. ; davis and webster , p. ) . these examples suggest that the interaction between genetic factors and the environment in producing a particular disease is complex. while people have little or no control over their genetic makeups, the lifestyles and behaviors they freely choose and the surroundings where they live can have signifi cant infl uences on the likelihood of developing a particular disease to which they are genetically predisposed. to further the discussion of the infl uence of genetics on health, mcginnis et al. ( , p. ) cite studies which show that although only about % of deaths in the united states may be attributed to purely genetic diseases, about % of late-onset disorders, such as diabetes, cardiovascular disease, and cancer, have some genetic component. for example, the brca gene accounts for only between and % of breast cancers in the united states; only about % of colon cancers may be explained by genes, and only about % of elevated serum cholesterol levels may be explained by familial hyperlipidemia. similarly, studies of identical twins focusing on the occurrence of schizophrenia, and other twin studies examining the occurrence of dementia in older people, have found that about half of each might be explained by genetic factors. further, while about two-thirds of the risk of obesity might be genetic, the risk is expressed only with exposure to controllable lifestyle factors (baird , p. ; muller , p. ; panjukanta et al. , p. ; kendler kendler , p. rowe and kahn ) . the institute of medicine (iom) ( , p. ) reported that americans in , compared with those who lived in , were healthier, lived longer, and enjoyed lives that were less likely to be marked by injuries, ill health, or premature death. but the gains in health reported by the iom were not shared equally among the population of the united states. at the moment, as was also the case in , gains in health status are not shared fairly or equally by all americans. americans with a good education, those who hold high-paying jobs, and those who live in serene and comfortable neighborhoods live longer and healthier lives than those with lower levels of education and income, and those who live in crime infested, overcrowded, and less comfortable and cohesive urban areas (isaacs and schroeder , p. ; bell and standish , p. ; lantz et al. lantz et al. , p. navarro , p. ; satcher , p. ; williams , p. ; metzler , p. ; kilbourne et al. kilbourne et al. , p. berkman and lochner , p. ) . there are several pathways through which social and economic circumstances affect health. those with good educational achievements are more likely to attain higher socioeconomic status than the poorly educated (angel et al. ; barr ; bartley ; mirowsky and ross , p. ) . people of lower socioeconomic status die earlier and are more susceptible to undesirable life events than people on higher socioeconomic levels, a pattern that holds true in a progressive fashion from the poorest to the richest (mcleod and kessler , p. ; adler et al. , p. ; adler and newman , p. ; guralnik et al. , p. ; mcdonough et al. mcdonough et al. , p. . this trend also holds whether one looks at education or occupation (national center for health statistics , p. ; kaplan and keil , p. ). these differences are said to be due to the fact that people of higher socioeconomic status have healthier behaviors and lifestyles than those of lower socioeconomic status. people of higher socioeconomic status are less likely to smoke, and are far more likely to eat healthier foods and to engage in leisure-time physical exercise (national center for health statistics , p. ; pratt et al. , p. s ; giles-corti and donovan , p. ). according to isaacs and schroeder ( , p. ) , as a result of "a sedentary lifestyle and unhealthy eating habits, obesity and the diseases it fosters now characterize lower-class life." poor eating habits and a sedentary lifestyle alone do not explain the differences in health between high and low socioeconomic people. rather, another explanation for the differentials lies in the distribution of income or the income gradient between the low and high socioeconomic groups. in a study of white americans using census data, undertaken by smith et al. ( , p. ) , men earning less than $ , per year were . times as likely to die prematurely as were those earning $ , or more. a similar study of british civil servants conducted about years before the american study showed that when smoking and other risk factors were controlled for, those who were in the lowest employment category were more than twice as likely to die prematurely of cardiovascular disease as were those in the highest employment category (davey smith et al. , p. ) . the fi ndings of these studies have led to the theory that inequitable distribution of income and wealth, or the socalled income and wealth gradient, causes poor health (sen (sen , p. , daniels et al. ; deaton , p. ). as noted above, the relationship between health and income is referred to as a gradient. this terminology emphasizes the gradual relationship between the two variables. health improvements are directly related to improvements in income throughout the income distribution, and poverty has more than a "threshold" effect on health (deaton , p. ) . the us national longitudinal mortality study (nlms) published by the national institutes of health (nih) ( ) showed that the proportional relationship between income and mortality was the same at all income levels, implying that the absolute reduction in mortality for each dollar of income was much larger at the bottom of the income distribution than at the top. apart from income, mortality is also known to decline with wealth, rank, and with social status (marmot et al. (marmot et al. , p. (marmot et al. , (marmot et al. , p. . similarly, studies also show marked differences in life expectancy by race and by geography or people's places of residence. for example, there is a -year gap in life expectancy between white men who live in the healthiest counties or localities and black men who live in the unhealthiest counties (murray et al. , p. ; gittelsohn , p. ; marmot marmot , p. kawachi and berkman ) . the brief discussion in this section points to the effects of numerous, and possibly interrelated, social and economic factors on health. income might affect health just as health might affect income; the distribution of income and wealth might affect health. similarly, education, race, minority status, geography, employment, housing, discrimination and social isolation, nutrition, lifestyle, stress, health practices, and coping skills might affect health. it does not appear to matter very much which of the above factors is stressed, especially since they are more likely to be interdependent than independent. disease risks exist, most often, along a continuum (rose ) . risks are rarely dichotomous. according to lochner ( , p. ) , there is no clear division between risk and no risk with regard to, for example, levels of blood pressure, cholesterol, alcohol or tobacco use, physical activity, diet and weight, etc. this gradient of risk also exists for many social and environmental conditions, such as socioeconomic status, social isolation, occupational and environmental exposure, and air quality. put differently, the numerous studies on the determinants of health that we are unable to fully summarize individually here for lack of space, point to the fact that even though the human and material resources at our disposal, the foods we eat, our levels of education, the houses we live in, the quality of the environments where we live and work, to name but a few, affect every person's health, the effects may vary in direction and scope from person to person, depending on the differences in their unique circumstances. improvement in environmental conditions is an important goal of the us government, as can be inferred from the emphasis on environmental quality outlined in healthy people . that document clearly states that factors in the physical and social environment play major roles in the health of individuals and communities. the physical environment is operationalized to include the air, water, and soil through which exposure to chemical, biological, and physical agents may occur. the physical environment can harm individual and community health, especially when individuals and communities are exposed to toxic substances, irritants, infectious agents, and physical hazards in homes, schools, and work sites. the physical environment can also promote good health, for example, by providing clean and safe places for people to work, exercise, and play ( healthy people , p. ). therefore, the physical environment is perhaps one of the most important factors that should be considered when classifying the health status of an individual (wikipedia ) . environmental factors, such as air and water quality, exposure to pesticides and toxic waste, and housing conditions, have major effects on health and human development. for example, substandard air and water quality have been directly associated with diseases such as cancer, asthma, certain birth defects, and some neurological disorders (grant makers in health , p. ) . similarly, many forms of cancer are associated with dioxin, polychlorinated biphenyls (pcbs), and mercury (friis ) . also, airborne particulate matter, tobacco smoke, and ground-level ozone, have been known to cause asthma attacks in children. exposure to lead, which can be found in peeling paint or in the soil and air in many poor communities, has been associated with impaired cognitive and behavioral development and low birth weight among children born to exposed mothers, and is also known to cause kidney damage (friis ) . in recognition of the danger of environmental contamination, bell and standish ( , p. ) urge communities to act on their behalf to make changes in the policies that affect their physical, social, and economic environments. they state, plausibly, that "policy, place, and community" matter. combined, policy and community can alter or ameliorate the underlying forces that lie at the heart of the determinants of health. for example, they argue that policy determines the behaviors or things that are allowed, encouraged, discouraged, and prohibited. policy also determines whether industrial facilities will be sited near residential neighborhoods, how industrial facilities treat their neighbors; how dense neighborhoods will be; what materials can be used to build houses; who will live in a neighborhood; whether businesses can locate in a neighborhood; and whether there are tax or other incentives available for locating in a neighborhood (bell and standish , p. ). in the developed communities or countries, environmental epidemiologists are concerned about such things as gene-environment interactions, environment-environment interactions, particulate air pollution, nitrogen dioxide, ground-level ozone, environmental tobacco smoke, radiation, lead, video display terminals, cellular telephones, and persistent organic pollutants (pops) that act as endocrine disruptors. exposure to these downstream or proximate environmental vectors (exposures that are closely related in time and space to the ill-effects they cause) affect both health and well-being (encyclopedia of public health ) . in the developing communities, the primary environmental determinants of health are said to involve biological agents in the air, water, and soil that account for most deaths. for example, diarrheal diseases acquired from contaminated food or water, malaria, intestinal parasitic infections, respiratory diseases caused by biological and chemical agents in both indoor and outdoor air, wreak havoc in the developing countries. these environmental hazards take a far greater toll on human life and suffering in absolute terms compared to those environmental vectors of concern in the developed countries (encyclopedia of public health ) . the above environmental vectors that cause havoc in the developing countries also abound in the poor localities of the united states and other developed countries. wealthy people are more likely to live in better homes and locations where they are less exposed to environmental risks than poor people (friis ; mcleod and kessler , p. ; giles-corti and donovan , p. ; shi and singh , p. ; grant makers in health , p. ) . for example, although the rates of asthma have been rising in the country, the disease affects low-income people disproportionately. whereas the national prevalence rate of childhood and adult asthma is put at about %, some african-american communities report about % of children suffering from asthma. also, puerto rican children are reported to have the highest prevalence of active asthma of any us ethnic or racial group. in california, latino children are reported to be hospitalized for asthma at a rate that is % greater than that of white children. obviously, environmental hazards are some of the reasons for these disparities ( healthy people ; joint center for political and economic studies and policylink , p. ; flores et al. , p. ) . despite the gains in environmental quality since the advent of the environmental movement in the s, mainstream environmental policies neglected the problems identifi ed in low-income communities because the inhabitants of those areas lacked the political and economic resources to press for environmental justice. however, since its start around , the environmental justice movement has resulted in the cleanup of hazardous waste sites, the redevelopment of brown-fi elds, the shutdown of incinerators, and the establishment of parks and conservation areas in low-income communities. additionally, in low-income communities, local pollution problems are being addressed, cleaner and more accessible means of public transportation are made available, and wild lands and unique habitats are being protected (faber and mccarthy ) . these changes are due to interest group pressure, the recognition of the externalities associated with environmental degradation, and the value of a clean environment to the health and well-being of all persons, rich and poor. mcginnis et al. ( , p. ) contend that behavior choices constitute the single most important domain of infl uence over health prospects in the united states. lifestyle and behaviors involve many dimensions, including dietary choices, engagement in physical activity, sexual behavior and recreation, including the choice to smoke and to ingest alcohol, the wearing of motor vehicle seatbelts and motorcycle helmets, and other responsible behavior when operating motor vehicles. because lifestyle and behavioral factors are under the control of individuals, the public is very likely to defi ne lifestyle and behavioral health problems as being self-induced. the choices we make with regard to the many dimensions of lifestyle and behavior enumerated above have signifi cant impacts on personal and population health. for example, dietary factors have been associated with coronary heart disease and stroke; colon, breast, and prostate cancers; and diabetes (us department of health and human services ) . similarly, a sedentary lifestyle has been associated with increased risk for heart disease, osteoporosis, dementia, diabetes, and colon cancer (us department of health and human services ) . furthermore, research shows that diets rich in fruits and vegetables, low-fat dairy foods with reduced saturated and total fat, and low sodium diets can lower blood pressure (appel et al. (appel et al. , p. svetkey et al. , p. ; sacks et al. , p. ) . the primary differences between how we perceive behavioral change now from much earlier perceptions is the great awareness that individual behavior occurs in a social context (berkman and lochner , p. ) , be it the place of work or abode, the family, the place of worship, the peer group, the school system, the stage of development, etc. for example, the results from the national youth risk behavior survey (yrbs) demonstrated that numerous high school students engaged in behaviors that increased their chances of dying from motor vehicle crashes, other unintentional injuries, homicide, and suicide. specifi cally, the survey results showed that . % of those surveyed had rarely or never worn a seatbelt during the days preceding the survey; . % had ridden with a driver who had been drinking alcohol; . % had carried a weapon during the days preceding the survey; . % had drunk alcohol during the days preceding the survey; . % had used marijuana during the days preceding the survey; and . % had attempted suicide during the months preceding the survey (grunbaum et al. , p. ) . the authors of the yrbs concluded that "priority health-risk behaviors, which contribute to the leading causes of mortality and morbidity among youths and adults, are often established during youth, extend into adulthood, are interrelated, and are preventable." the examination of the main causes of death in the united states, which we shall shortly discuss in the next section of this chapter, will shed further light on behavioral risk factors. meanwhile, suffi ce it to say that lifestyle and behavioral factors constitute some of the important determinants of health that health policy must seek to address. even though it is agreed that the contribution of medical care to improved health is not as pronounced as the other factors just examined, curative medical care-those practices, technologies, and organizations that society and the medical profession use to cure and rehabilitate the sick-is nonetheless a key determinant of health (blum ; cdc ) . the centers for disease control and prevention (cdc) estimate that only about % of premature deaths in the united states can be attributed to inadequate access to medical care, while the remaining % can be accounted for by individual lifestyle and behaviors ( %), genetic profi les ( %), and social and environmental conditions ( %) (cdc ) . the reason why medical care is the least important determinant of health is because it is reactive, not proactive-it waits for disease and illness to occur before intervening, so to speak. in other words, while individual and population health are somehow associated with having access to curative care, access to preventive services is of greater signifi cance. therefore, health can improve signifi cantly, and the prevalence of disease can decline dramatically, without effective medical care, due to the other determinants of health (sigerist , p. ; mckeown , p. ; banta and jonas , p. ). this knowledge is very likely the reason why williams and jackson ( , p. ) and isaacs and schroeder ( , p. ) advocate the broadening of the concept of health policy to include the other determinants of health that were not usually seriously considered when discussing health policy. this knowledge, too, is the primary reason for this chapter of the book. we can elaborate further on the importance and relevance of the determinants of health by linking them to the ten leading causes of death in the united states. where possible, the analysis will link the incidences of mortality reported in the country that are associated with each, some, or combinations of the determinants of health. table . shows the ten leading causes of death in the united states for and . we present, below, the ten leading causes of death in the country for and in order to attempt to link some of them to treatable or preventable behaviors and exposures. in other words, we shall attempt to show that most of the deaths can be associated with factors that mainly fall under the social, economic, environmental, and lifestyle and behavioral determinants of health that we have just discussed. most of the ten leading causes of death presented above are nongenetic and can be prevented or treated. diseases of the heart, cancers, cerebrovascular diseases or strokes, chronic lower respiratory diseases, unintentional injuries, diabetes, infl uenza and pneumonia, and infection-and high blood pressure-induced nephritis can be curtailed, prevented, or treated. for example, cigarette smoking is linked with an increased risk of heart disease, chronic lower respiratory disease, and cancer; obesity is a major health risk for diabetes, hypertension, coronary heart disease, and some forms of cancer; alcohol causes a wide variety of accidents and injuries, increases the risks for high blood pressure, irregularities of the heart, and stroke; fl u vaccines can minimize infl uenza deaths; and seeking treatment for infections can prevent septicemia. additionally, although there is a genetic basis for nephrosis and nephrotic syndrome, the conditions can occur as a result of infection (such as strep throat, hepatitis, or mononucleosis), use of certain drugs, and diabetes. furthermore, although age and family history are important risk factors for alzheimer's disease, longstanding high blood pressure and a history of head trauma are suspected risk factors for the disease as well mcginnis and foege ( , p. ) identifi ed and quantifi ed the major external or nongenetic factors that contributed to deaths in the united states in . deaths associated with socioeconomic factors and access to medical care, although important contributors to the total deaths recorded in the country, were not included in the study because of the diffi culty quantifying them independent of the other factors reported in the study. about years after the mcginnis and foege study, mokdad et al. ( mokdad et al. ( , p. ) used a similar methodology to quantify the nongenetic factors that contributed to deaths in . the results of the two studies cited above showed that about half of all deaths that occurred in the united states in both and could also be attributed to a small number of largely controllable behaviors and exposures, including tobacco, diet and activity patterns, alcohol, microbial and toxic agents, fi rearms, sexual behavior, motor vehicle accidents, and illicit drug use. the results of the causes of death studies reported by mcginnis and foege and mokdad and his colleagues are consistent with the fi ndings of the national yrbs cited earlier in this chapter. the survey results showed that in the united states, . % of all deaths among youth and young adults aged - years were due only to four causes: motor vehicle crashes, other unintentional injuries, homicide, and suicide. the deaths attributable to these causes among the identifi ed population group were . , , . , and . %, respectively (grunbaum et al. , p. ) . furthermore, substantial morbidity and social problems were said to result from the approximately , pregnancies that occurred each year among women - years (ventura et al. , p. ) , and from the estimated million cases of sexually transmitted diseases (stds) that occurred each year among persons - years (institute of medicine ; eng and butler ) . similar to the studies on the actual causes of death in the united states in and , the yrbs also found that the leading causes of mortality and morbidity among all age groups in the country were related to behaviors that contributed to unintentional injuries and violence, tobacco use, alcohol and other drug use, sexual behaviors that contributed to unintended pregnancies and stds, including hiv infection, unhealthy dietary behaviors, and sedentary lifestyles. in , almost years after the yrbs discussed above, the cdc quantifi ed the death rates among teenagers aged - years between and . not surprisingly, the ten leading causes of death for the teenage population remained constant throughout the period. they were as follows: accidents or unintentional injuries, % of deaths; homicides, % of teenage deaths; suicide, %; cancer, %; and heart disease, %. further analysis showed that motor vehicle accidents accounted for almost three quarters ( %) of all deaths from unintentional injury; and that non-hispanic black males had the highest death rate among all teenagers, with homicide being the leading cause of death for them (minino ) . the determinants of health that have occupied our attention up to this point are not only affected by the broad national and personal factors we have identifi ed but are also affected by broad global or international factors (shi and singh , p. ) . therefore, the rest of this chapter is devoted to examining the infl uences of global factors on the health care system and the health policymaking process. foreign policies involve the political relationships between countries and the outside world. foreign policy development generally concerns the protection of a country's national interests, usually defi ned in terms of security, economic prosperity, and ideological goals (lee et al. , p. ) . increased globalization has led to the broadening of foreign policy concerns to include health. conversely, it is now recognized that international trade and fi nance, migration and population mobility, environmental change or global warming, the emerging and reemerging infectious disease paradigms, natural disasters, and global insecurity or terrorism have clear and observable consequences for human health (kassalow ; mcinnes and lee , p. ; lee et al. , p. ; katz and singer , p. ; campbell-lendrum et al. , p. ; fidler , p. ; macpherson et al. , p. ; labonte et al. ) . we shall briefl y examine how these components of globalizationinternational trade, population mobility, infectious diseases, global warming or climate change, and natural disasters and terrorism-affect countries' health care and policymaking systems generally, and the united states' health care and policymaking systems in particular. we begin with international trade. the principal agents of global international trade and fi nance include such international agencies as the world bank, the international monetary fund (imf), and the world trade organization (wto). it has been reported that the market-biased or effi ciency-oriented austerity policies these organizations promote or sponsor have resulted in reduced expenditures for social programs in developing countries, thereby impairing population health and slowing the advances in literacy, fertility reduction, and improved reproductive health of the women of the developing countries (kinnon , p. ; gray ; watts ) . some specifi c examples of international trade and fi nance policies include the following: trade liberalization or the lowering of tariffs and other barriers to imports that has led to the doubling of the value of world trade from % of world gdp in to % in (world bank ; the reorganization of production and service provision across multiple national borders by multinational or transnational corporations, such as outsourcing or the pursuit of integration into global value chains, resulting in a global labor market (world bank , p. woodall ) ; the conditions attached to world bank and imf loans, and to the rescheduling of loan payments, including structural adjustment programs (saps); fi nancial liberalization, which exposes national economies to the uncertainties created by large and volatile short-term capital fl ows; the signifi cant growth in the world's urban population caused by transnational economic integration; the promotion of export-oriented agricultural development that does not consider the social and environmental consequences of such actions, which result from the pressures on governments around the world to increase export earnings (stonich and bailey , p. ) ; and the promotion and reinforcement of a market-oriented concept of health sector reform that strongly favors private provision and fi nancing (petchesky ; koivusalo and mackintosh , p. ). critics of the above international trade and fi nance policies argue that it is not at all clear that globalization leads to substantial poverty reduction. they point to the large-scale and extreme unequal distribution of wealth and income in the countries that have been identifi ed as "globalizers" witnessing rapidly growing economies. it is argued that even a little redistribution of income through progressive taxation and targeted social programs would go farther in terms of poverty reduction than many years of solid economic growth (jubany and meltzer ; paes de barros et al. ; de ferranti et al. ) . further, it is argued that as countries compete for foreign direct investment and outsourced production, the need to appear business-friendly may limit their ability to adopt and implement labor standards, occupational safety and health regulations, and other redistributive programs (cornia ) ; global integration of production may cause a sharp decline in the wages of, and demand for, low-skilled workers; large amounts of debt limit the ability of many developing and developed countries to meet other human needs related to health, education, water, public safety, sanitation, nutrition, etc.; globalization may lead to an intensifi cation of worldwide social relations which link distant localities in such a way that local happenings are shaped by events occurring many miles away, and vice versa (giddens , p. ) ; much of the urbanization caused by international fi nance and trade policies occurs in countries that have limited resources to provide urban infrastructures; and the emphasis on private fi nancing and provision of health care leads to large-scale underinsurance and uninsurance in both the developed and developing countries (labonte and schrecker , p. ) . globalization and the quest for exports are also blamed for increased smoking and tobacco-related mortality in the developing countries (murray and lopez , p. ) . also noteworthy is the escalation in the sale of weapons, much of it facilitated by western governments. the wars that have raged on and off in sub-saharan africa, latin america, and asia are tragic examples of the ill effects of aggressive weapon sales to these places (mcmichael and beaglehole , p. ) . although the adverse effects of globalization discussed above tend to affect developing countries more than the united states, there are signifi cant adverse consequences of globalization for the united states as well. some of these include the perpetuation and exacerbation of the gap between the rich and the poor, a large public debt profi le that puts signifi cant pressure on social and other safety net policies and programs, the prevalence of uninsurance and underinsurance, job insecurity and reduced wages, the collapse of large manufacturing businesses, increased availability and demand for illicit drugs, and the emergence of new infectious diseases that spread more easily due to increased migration and population mobility (ubokudom and khubchandani , p. ) . for example, american labor unions complain that the north american free trade agreement (nafta) with canada, mexico, and the united states, which came into force on january , , has led to the loss of american jobs. job loss causes stress, loss of income and the fi nancial means to pay for medical care. from the onset, health issues were not at the heart or margins of foreign policy theory or practice for two reasons. first, the protection and promotion of population health did not factor into world leaders' calculations of what "competition in anarchy" (the condition from which foreign policy dynamics fl ow) required of their countries, nor was health for all seriously (as opposed to rhetorically) considered a pathway to a better world. second, those who were engaged in public health did not participate signifi cantly in discussions of foreign policy (fidler , p. ) . therefore, there were only small and nonsubstantial linkages between health and foreign policy (harris , p. ) . actions linking health issues or problems with foreign policy have been strongest when the potential impact on economic prosperity, national security, the environment, and development is severe. this has resulted in attention to health threats that are acute and severe, those that are projected to result in mass casualties, and those that are believed to be geographically widespread. in contrast, long-term health risks, or health risks that cause minor health problems, affect a limited number of people, or are not geographically widespread, attract little attention in relation to foreign policy. in other words, acute epidemic infections and major public health emergencies, such as natural or human-induced disasters, bioterrorism, and chemical and radiation accidents, have received signifi cant attention (fidler , p. ; lee et al. , p. ; katz and singer , p. ) . a few specifi c examples of "attention-receiving" public health problems include the previously unknown human immunodefi ciency virus/acquired immunodeficiency syndrome (hiv/aids) which appeared in the united states in the early s; the hantavirus, believed to have originated in korea; eastern equine encephalitis, which is found in the eastern and north-central united states, canada, parts of central and south america, and the caribbean islands; western equine encephalitis, which occurs primarily in the western and central united states, canada, and parts of south america; the polio virus that is believed to have originated in india in ; the spread of severe acute respiratory syndrome (sars) from china in ; and the outbreak of the deadly h n -swine flu-infl uenza believed to have originated in mexico (cdc ; shi and singh , p. ; friis , p. ) . in summary, many health problems, particularly infectious diseases, are widely recognized as global concerns that cross national and international boundaries. consequently, countries frequently include in their foreign policies strategies on these diseases that have the potential to threaten their domestic interests. this is likely to lead to higher prioritization, more attention, greater political support, and more funding. for example, in the united states, projections of the impact of hiv/ aids on the workforces of many countries, and the prevalence of hiv among military personnel in several regions of the world, contributed to the determination that hiv/aids was a security issue. similarly, awareness of the havocs caused by previous infl uenza pandemics and the economic impact of the small and short outbreak of sars led to serious preparations by the who and its member states for the next infl uenza pandemic (katz and singer , p. ) . this understanding has led to many international agreements covering health and the environment, including the agreement on sanitary and phytosanitary measures, the international standards organization's classifi cation system for food labeling, the un framework convention on climate change, and the kyoto protocol, to name a few. data from the national aeronautics space administration (nasa) show that the earth's surface has warmed by about . °c between january and november . that period was reported to be the warmest january-november in the nasa goddard institute for space studies (giss) analysis, which covers years. the period was only a few hundredths of a degree warmer than , so it is possible that the fi nal giss results for the full year, , would be warmer or in the same range as . further, the available data also show that the earth's surface has warmed by more than . °c over the past century and by about . °c in the past decades (nasa ) . therefore, contrary to frequent assertions that global warming has slowed in the past decade, global warming has proceeded in the decade that ended in just as fast as it did in the prior decades (nasa ) . the health hazards posed by climate change and global warming are inequitable, diverse, global, and probably irreversible over human time scales (patz et al. , p. ; campbell-lendrum et al. , p. ) . they include increased risks of extreme weather, such as fl oods and storms, fatal heat waves, long-term drought conditions in many areas of the world, surface water pollution and groundwater contamination, the melting of glaciers that supply freshwater to large population centers, salination of sources of agricultural and drinking water, increased rates of water extraction that may precipitate declines in supply, and creating a conducive environment for the global killers that are very sensitive to climatic conditions, such as malaria, diarrhea, and protein-energy malnutrition (campbell-lendrum et al. , p. ; friis , p. ) . as we noted under the actual causes of death, these three global killers cause many deaths in the united states; they are also said to account for about three million deaths worldwide each year (who ) . the relationship between migration, population mobility, and health is receiving renewed attention due to the emerging and reemerging infectious diseases that were discussed previously in this section. the health of both legal and illegal migrants to any country are affected by the determinants of health discussed earlier in this chapter, as well as by the risks that are present in their country of origin or that arise from the migration process itself (macpherson and gushulak , p. ) . this is very true of the united states where a signifi cant portion of the annual population growth is due to migration. the effects of population mobility and migration on the country's health care system and the provision of health services are reported daily in the pages of newspapers. first, there is likely to be increased demand for services due to population growth, whether that growth is due to increased fertility rates or migration. for example, the exponential growth in medicaid expenditures in states that border mexico are said to be due to the increased demand for medical services by illegal immigrants as well as by the medical needs of an aging population. second, offi cials of the states that share boundaries with mexico complain about increased violent crimes committed by illegal immigrants, crimes that take a heavy toll on population health and health care expenditures. third, increased migration compels more health services planning, infrastructure maintenance, development and training of a diverse medical workforce to cater for the increasingly diverse population, and the establishment of public health programs for health promotion, health protection, and disease prevention (macpherson et al. , p. ; cohen et al. , p. ) . and, fourth, the opinion pages of newspapers carry citizens' letters that attribute the success of previous terrorist campaigns to the nearly open border policy the united states maintained prior to september , ( / ). since the / attacks, border security and entry visa requirements have been tightened. border control measures are now centered on inspecting and excluding goods, vessels, and people that pose serious health or terrorist threats to the united states. other countries have similar measures. the world has changed. indeed, the world has changed signifi cantly. while most people are actively planning on how to make their lives better, a few others are actively planning on how to destroy lives and settle political and ideological differences through acts of violence. no place and people are immune from the threats of violence, terrorism, and natural disasters. in the past or years, the united states has experienced disasters that have led to a rethinking of how to keep the population safe. the terrorist attacks in the united states on september , , an unsuccessful attempt to initiate an anthrax epidemic in october , and the devastation caused by hurricane katrina of the atlantic hurricane season led to signifi cant loss of lives and property and revealed defi ciencies in the public health and emergency response systems in the country. because of both underfunding and understaffi ng, and perhaps because the changes that have taken place in the world were not anticipated, the public health system was unable to develop or implement a comprehensive program of preparedness, prevention, response, and recovery (us general accounting offi ce ) . following the disasters, state, local, and federal public health agencies began to identify weaknesses in the nation's public health infrastructure and to reevaluate existing disaster response plans (baker and koplan , p. ). the shortcomings revealed in the nation's disaster response plans elevated public health to an important national instrument for anticipating and dealing with terrorism, infectious disease outbreaks, and natural disasters. the guidance on responses to chemical, biological, radiological, nuclear, and explosive threats provided by the cdc, and by other national organizations and universities, helped individual state governments to develop statewide policies that took their unique concerns into account (ziskin and harris , p. ; shah shah , p. gebbie and turnock , p. ) . public health plans to deal with terrorist threats, infectious diseases, and natural disasters now involve public health agencies at the federal, state, and local levels of government; other government and private agencies, such as the departments of justice and defense; the food and drug administration; private, public, and nonprofi t hospitals, clinics, and nursing homes; private and public practitioners, such as nurses and physicians; blood supply organizations, such as the american red cross; police and fi re departments; and individuals and groups throughout the country. as would be expected, expenditures for government public health activities, while still low relative to expenditures for medical care, rose from $ billion in to about $ . billion in , an increase of . % from (centers for medicare and medicaid services (cms) ) . it remains to be seen if this enthusiasm for public health, demonstrated by increased funding since , can be sustained. the law that is used as the basis for most of the new emergency preparedness measures is the homeland security act of . in addition to the strengthening of the public health infrastructure, the law also called for improved inspections of food products entering the united states. it calls for better measures to contain attacks on food and water supplies, to protect vital infrastructures, such as nuclear facilities, and to track biological materials anywhere in the country. further, the provisions of the law have been used to justify tough and controversial interrogation techniques, such as waterboarding. similarly, presidential executive order , signed by george w. bush on april , , authorizes the apprehension, detention, or conditional release of individuals with suspected communicable diseases, such as sars, cholera, diphtheria, infectious tuberculosis, plague, smallpox, yellow fever, and viral hemorrhagic fevers such as ebola (the free dictionary ) . in summation, international trade and fi nance, infectious disease epidemics, global warming and climate change, population mobility, and natural disasters and terrorism signifi cantly affect the united states health care delivery and policymaking systems. in addition, medical technology and us health care professionals and consumers are also affected by global factors. for example, because the united states is widely believed to be the world leader in the development and utilization of high-technology medical protocols, foreign dignitaries come here for specialty care. also, nurses and foreign medical school graduates (fmgs) move to the united states to acquire licenses to practice in the country. this so-called brain drain causes shortages of medical practitioners in the developing countries and alleviates some of the shortages in the health professional shortage areas of the united states. furthermore, telemedicine allows us physicians to transmit radiological images to other countries where they are analyzed at lower costs. on the other hand, us consulting pathologists and radiologists provide their services to other parts of the world. also, advanced medical equipment and supplies that are abandoned here a few years after deployment are shipped to the developing and less technology-intensive developed countries at low costs. the high costs paid by us consumers are used to subsidize the low costs paid by the developing countries (ubokudom and khubchandani , p. ) . this chapter has identifi ed the impacts of physical, social, cultural, and global factors on health and health policymaking. health can be defi ned under the medical or wellness models. the health status of the us population, or the population of any other country for that matter, is largely determined by factors that have important physical, social, and economic dimensions. these include preventive medicine, genetic disposition, social and economic circumstances, environmental conditions, lifestyles and behaviors, and medical care. these determinants of health are associated, in various degrees, with the real or actual causes of death in the country. research demonstrates that most of the deaths in the country are attributable to a small number of largely controllable behaviors and exposures, or due to factors that fall under the preventive, social, economic, environmental, and lifestyle and behavioral determinants of health. these determinants of health are not only affected by the broad national and personal factors identifi ed in the chapter, they are also affected by global or international factors, including trade and fi nance, outbreaks of infectious diseases, climate change, natural disasters, and the threats of terrorism and population mobility. but even though most of the deaths in the country are the result of social, cultural, economic, environmental, and global factors, medical care is also an important determinant of health that cannot be ignored. an insurance card is one of the important factors that infl uence access to medical services. consequently, the next chapter examines demographic factors, most especially americans' ability to access medical services, and the disparities in health among segments of the population. socioeconomic inequalities in health: no easy solution socioeconomic disparities in health: pathways and policies poor families in america's health care crisis a clinical trial of the effects of dietary patterns on blood pressure in why are some people healthy and others not? strengthening the nation's public health infrastructure: historic challenge, unprecedented opportunity in jonas's health care delivery in the united states health disparities in the united states: social class, race, ethnicity and health unemployment and ill health: understanding the relationship communities and health policy: a pathway for change relationship of physical health status and health problems health and ways of living: the alameda county study social determinants of health: meeting at a crossroads planning for health pathways to health: the role of social factors the role of medical care in determining health: creating an inventory of benefi ts genes, dreams, and realities global climate change: implications for international public health policy healthy people: the surgeon general's report on health promotion and disease prevention national health expenditure projections priorities among recommended clinical preventive services the case for diversity in the health care workforce policy reform and income distribution is inequality bad for our health explanations for socioeconomic differentials in mortality: evidence from britain and elsewhere the social context of science: cancer and the environment policy implications of the gradient of health and wealth inequality in latin america & the caribbean: breaking with history? the hidden epidemic: confronting sexually transmitted diseases green of another color: building effective partnerships between foundations and the environmental justice movement the health of latino chindren: urgent priorities, unanswered questions, and a research agenda essentials of environmental health the public health workforce, : new challenges the consequences of modernity socioeconomic status differences in recreational physical activity levels and real and perceived access to a supportive physical environment on the distribution of underlying causes of death social determinants of health false dawn: the delusions of global capitalism youth risk behavior surveillance-united states educational status and active life expectancy among older blacks and whites marrying foreign policy and health: feasible or doomed to fail? united states department of health and human services (usdhhs) deaths: leading causes for institute of medicine (us) committee on health and behavior: research, practice, and policy. health and behavior: the interplay of biological, behavioral, and societal infl uences class-the ignored determinant of the nation's health breathing easier: community-based strategies to prevent asthma the achilles' heel of latin america: the state of the debate on inequality , fpp - . ottawa, canada. canadian foundation for the americas (focal) regional obesity and risk of cardiovascular disease: the framingham study socioeconomic factors and cardiovascular disease: a review of the literature why health is important to u.s. foreign policy health and security in foreign policy neighborhoods and health overview: a current perspective on twin studies of schizophrenia fundamentals of genetic epidemiology advancing health disparities research within the health care system: a conceptual framework world trade: bringing health into the picture statement on the politicization of evidence-based clinical research in commercialization of health care: global and local dynamics and policy responses globalization and the social determinants of health: the role of the global marketplace (part of ) socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of u.s. adults bridging health and foreign policy: the role of health impact assessments religion and health: is there an association, is it valid, and is it causal? benefi ts and harms of prostate-specifi c antigen screening for prostate cancer: an evidence update for the u.s. preventive services task force health and foreign policy: infl uences of migration and population mobility human mobility and population health: new approaches in a globalizing world health inequalities among british civil servants: the whitehall ii study inequalities in death-specifi c explanations of a general pattern the spiritual history religion and depression: a review of the literature religious involvement and mortality: a meta-analytic review income dynamics and adult mortality in the united states the case for more active policy attention to health promotion actual causes of death in the united states health, foreign policy and security the role of medicine: dream socioeconomic status differences in vulnerability to undesirable life events the changing global context of public health septicemia social determinants of health: what, how, why, and now social patterns of distress mortality among teenagers aged - years: united states actual causes of death in the united states screening for bladder cancer: u.s. preventive services task force recommendation statement hereditary colorectal cancer: from bedside to bench and back patterns of mortality by county and race: - . cambridge, ma: harvard center for population and development studies alternative projections of mortality and disability by cause - : global burden of disease study (phs) - ), - . national institutes of health. . a mortality study of . million persons by demographic, social, and economic factors: - follow-up what we mean by social determinants of health meeting the millennium poverty reduction targets in latin america and the caribbean letter: linkage of familial combined hyperlipidaemia to chromosome q -q impact of regional climate change on human health global prescriptions: gendering health and human rights levels of physical activity and inactivity in children and adults in the united states: current evidence and research issues nephrotic syndrome: nephrosis alzheimer's disease; senile dementia-alzheimer's type (sdat) retrieved biologic-marker studies in alcoholism factors infl uencing the view of patients with gynecologic cancer about end-of-life decisions the strategy of preventive medicine successful aging effects on blood pressure of reduced dietary sodium and the dietary approaches to stop hypertension (dash) diet commentary: include a social determinants of health approach to reduce health inequities development as freedom the formation of the emergency medical services delivering health care in america: a systems approach socioeconomic differentials in mortality risk among men screened for the multiple risk factor intervention trial: i. white men resisting the blue revolution: contending coalitions surrounding industrial shrimp farming health services research: a working model effects of dietary patterns on blood pressure: subgroup analysis of the dietary approaches to stop hypertension (dash) clinical trial rethinking mckeown: the relationship between public health and social change severe acute respiratory syndrome (sars) poor diets, little exercise leading cause of preventable illness and deaths world development report : workers in an integrating world an introduction to epidemiology the ecology of health policymaking and reform in the united states of america united states department of health and human services (usdhhs). . the surgeon general's report on nutrition and health physical activity and health: a report of the surgeon general bioterrorism: public health response to anthrax incidents of section : overview of u.s. preventive services task force structure and processes trends in pregnancy rates for the united states, - : an update epidemics in history: disease, power and imperialism unhealthy societies: the affl iction of inequality social sources of racial disparities in health socioeconomic differences in health: a review and redirection men's health: chronic lower respiratory diseases world health organization (who). . preamble to the constitution of the world health organization as adopted by the international health conference state health policy for terrorism preparedness key: cord- -gs i wbe authors: dada, m.a.; lazarus, n.g. title: sudden natural death | infectious diseases date: - - journal: encyclopedia of forensic and legal medicine doi: . /b - - - / - sha: doc_id: cord_uid: gs i wbe nan a wide range of deaths from natural causes is encountered in the field of forensic medicine. despite the advances in the diagnosis and treatment of infectious diseases, a substantial number of sudden and unexpected deaths are caused by infections. in most medicolegal systems these deaths are subject to a forensic investigation. the world health organization defines sudden death as that occurring within h of the onset of symptoms. some authors variably define sudden death as that occurring within , , and h of the onset of symptoms. forensic pathologists should be aware of the importance of infectious causes of sudden death in the present era of bioterrorism and emergent and reemergent diseases. genetic engineering has led to the development of highly infectious and virulent strains of microorganisms (e.g., anthrax). emerging infectious diseases are infections whose incidence has increased in recent years and/or threatens to increase in the near future. reemergence refers to the reappearance of a known infection after a period of disappearance or decline. death from infectious agents may occur as a direct consequence of the infection or from complications such as immunosuppression caused by the infection and adverse reactions to therapeutic drugs. sudden death due to infectious disease may be classified by organ system involvement (e.g., cardiac -myocarditis; nervous system -meningitis and encephalitis) or according to the etiological agent (e.g., viral, chlamydial, bacterial, fungal, protozoal, or helminthic) . the common infectious causes of sudden death by organ system are listed in table . the morphological findings at autopsy will depend on the type of organism, the site involved, and the host's response to the organism. microbiological demonstration of an organism does not equate to disease, as a host may be colonized by bacteria or the patient may have an asymptomatic viral infection. the exquisite sensitivity of molecular tests, e.g., polymerase chain reaction, may exacerbate this problem if the results are not correlated with the pathological findings at autopsy. categories of human pathogens include prions; viruses; chlamydiae, rickettsiae, and mycoplasmas; bacteria; fungi; protozoans; and helminths. infection by prions, rickettsiae, and mycoplasmas is not normally associated with sudden and unexpected death. viruses are ubiquitous and cause a spectrum of disease in humans. these may range from asymptomatic infection, severe debilitating illness, to sudden death. viral infections causing sudden death usually involve the cardiac, respiratory, or the central nervous system. morphologic findings in viral infections may include intranuclear and/or intracytoplasmic inclusions, multinucleate giant cells, and tissue necrosis (cytopathic effect). in many cases the diagnosis can only be made on special investigations, e.g., culture, electron microscopy, serology, or molecular testing. viral hemorrhagic fevers such as marburg, lassa, and ebola virus may cause sudden death in children. if there is any suspicion of a viral hemorrhagic fever, special care must be taken to avoid unwarranted exposure to health workers. the local public health officials must be informed and consideration given to limited autopsy examination in consultation with a virologist (e.g., postmortem blood sampling and liver biopsy). cardiac involvement usually takes the form of myocarditis. although many viruses may cause myocarditis (table ) , coxsackie a and b are responsible for most cases. fulminant coxsackievirus infection may also cause leptomeningitis, florid interstitial pneumonitis, pancreatitis, and focal hepatic necrosis. coxsackie b viruses should also be considered as a cause of sudden infant death. at autopsy, the myocardium is usually mottled and flabby. histology reveals focal infiltrates of inflammatory cells (neutrophils and/or lymphocytes, plasma cells, and macrophages). at least two foci of individual myofiber necrosis associated with - inflammatory cells are required for the histological diagnosis of myocarditis. focal aggregates of lymphocytes not associated with necrosis may be seen in elderly patients and are not diagnostic of myocarditis. myocardial involvement may be patchy. for adequate histological sampling, it is recommended that at least six sections be taken from various areas of the myocardium, including the left ventricle and nodal tissue. indirect damage to the myocardium may occur as an allergic response to a viral infection and eosinophilia, e.g., in eosinophilic myocarditis. this is a rare cause of sudden death in apparently healthy children due to the cardiac toxicity of eosinophils. studies have shown that persons undergoing severe mental or physical stress may have reduced immunity to viral infections. in the investigation of sudden death in athletes, the diagnosis of viral myocarditis must be considered. enteroviral infection may also play an important role in coronary plaque instability and may precipitate coronary thrombosis, leading to ventricular tachyarrhythmias and sudden death. viral infections of the respiratory system sudden death due to viral involvement of the respiratory system may be due to fulminant viral pneumonitis or bacterial pneumonia complicating an initial viral pneumonitis. viruses implicated include respiratory syncytial virus, human herpesvirus- , and parainfluenza virus in children, and adenovirus and influenza a and b in adults. microscopically, the findings of a viral pneumonitis are usually nonspecific and include edema and widening of the interstitial septa with a mononuclear cell infiltrate. in some cases, diagnostic viral inclusions may be demonstrated. emergent diseases such as severe acute respiratory syndrome (sars) have a high mortality and may cause death within hours. sars refers to an acute respiratory illness caused by infection with a novel coronavirus currently known as the sars virus. postmortem histopathological evaluations of lung tissue show diffuse alveolar damage consistent with the pathologic manifestations of acute respiratory distress syndrome. there is usually mild interstitial inflammation with scattered alveolar pneumocytes showing cytomegaly, and enlarged nuclei with prominent nucleoli. when faced with the finding of diffuse alveolar damage at autopsy, the pathologist should consider other infective causes such as influenza, para influenza, respiratory syncytial, and adenoviruses, chlamydia, mycoplasma, pneumococcus, legionella, and pneumocystis. sudden death may occur due to direct infection of the nervous system or a complication of a viral infection such as toxoplasmosis in human immunodeficiency virus/acquired immunodeficiency syndrome (hiv/aids). herpes simplex virus- encephalitis is usually due to reactivation of latent infection. commonly affected sites include the temporal lobe(s) (medial before lateral), the inferior frontal lobe(s), and the sylvian cortex(es). at autopsy there is widespread and asymmetrical necrosis. in fulminant cases there is prominent hemorrhage and swelling with raised intracranial pressure and brain herniation. histological findings include perivascular cuffing by mononuclear cells (figure ) and, in a small number of cases, intranuclear inclusions may be seen in astrocytes and neurons. in adult hiv infections, sudden death from infective causes may be due to opportunistic infections (e.g., toxoplasmosis) or rupture of mycotic aneurysms. in viral central nervous system infections the brain may appear macroscopically normal, especially in very young, elderly, debilitated, and immunocompromised individuals. specimens should be taken for microbiology and histology. serum and cerebrospinal fluid (csf) should be sent for antibody studies. tissue for histological examination should be taken from normal, obviously abnormal, and transition areas. routine sections should be taken from the cerebral cortex (all four lobes), thalamus, basal ganglia, hippocampus, brainstem, and cerebellum. as poliomyelitis has been described as a cause of sudden death in infants, autopsy protocols in sudden death should include histological examination of spinal cord and dorsal root ganglia. chlamydia pneumoniae may be associated with myocarditis and sudden unexpected death. bacterial infections are responsible for sudden unexpected death in adults and children. in the pediatric population bacterial infections of the respiratory, gastrointestinal, and central nervous system account for the majority of cases of sudden death. bacterial infections of the cardiovascular system bacterial causes of myocarditis include corynebacterium diphtheriae, neisseria meningitidis, and borrelia burgdorferi. in b. burgdorferi, cardiac involvement occurs in - % of cases and death may occur as a result of conduction disturbances. in diphtheritic myocarditis myocardial damage is caused by the release of toxins. bartonella-induced silent myocarditis has been described as a cause of sudden unexpected cardiac death in athletes. granulomatous myocarditis may also lead to sudden death ( table ). the mechanism of death includes arrhythmias, cardiac rupture, coronary occlusion, obstruction to pulmonary blood flow leading to fatal hemorrhage, and impaired myocardial contractility. cardiac tuberculosis is usually an autopsy diagnosis. histological examination of the myocardium shows a nodular, miliary, or diffuse infiltrative pattern. the coronary arteries may show narrowing or complete occlusion due to an intimal or diffuse tuberculous arteritis. it is uncommon to demonstrate acid-fast bacilli within the lesions. molecular tests such as the ligase chain reaction (lcr) and polymerase chain reaction (pcr) may be used to demonstrate the organism. sudden death in infective endocarditis occurs as a result of perforation of a free-wall myocardial abscess or rupture of a valve leaflet. staphylococcus aureus is responsible for - % of cases and is the major cause in intravenous drug abusers. other bacterial causes include haemophilus, actinobacillus, cardiobacterium, eikenella, and kingella (hacek group). negative bacterial cultures may be found in % of cases as a result of prior antibiotic therapy. the most common sites of infection are the aortic and mitral valves, except in intravenous drug abusers, where the right-sided valves are primarily affected. tertiary syphilis causing aortitis may cause sudden death from rupture of aortic aneurysms with aortic dissection. the mechanism of death is either blood loss with hypovolemic shock or a fatal cardiac tamponade from intrapericardial rupture. bacterial infections of the respiratory system sudden death from acute epiglottitis occurs from respiratory obstruction caused by swelling of the epiglottic folds, uvula, and vocal cords. the most common cause of acute epiglottitis in developing countries is haemophilus influenzae type b. in countries with established immunization programs, the incidence of h. influenzae epiglottitis has decreased and other bacteria, such as streptococcus, staphylococcus, and pneumococcus, have been implicated as possible causes. postmortem blood cultures are positive in - % of cases. lobar pneumonia ( figure ) and confluent bronchopneumonia are the most frequent cause of sudden death from acute pulmonary disease. some - % of lobar pneumonia is due to streptococcus pneumoniae type . bronchopneumonia is caused by staphylococci, streptococci, h. influenzae, pseudomonas aeruginosa, and coliform bacteria. pulmonary tuberculosis may result in hemoptysis, which can cause hypovolemic shock and sudden death. histologically, caseating granulomas are found. acid-fast bacilli are demonstrated using the ziehl-neelsen stain (figure ) . corynebacterium diphtheriae produces a gray pseudomembrane from the pharynx to the larynx, and this may lead to respiratory obstruction and sudden death. legionnaire's disease is associated with outbreaks of sudden death. the disease is caused by legionella pneumophila, a facultative intracellular organism. it causes severe pneumonia in the elderly, in smokers, and in immunocompromised patients. the organisms may be transmitted via droplet spread from contaminated air-conditioning units and water coolers. the organism may be demonstrated by a modified silver stain (dieterle stain) or by immunofluorescence and culture. pyogenic meningitis may cause sudden death. the causative organism varies according to the age of the patient ( table ) . the location of the exudates depends on the organism. in h. influenzae it is basally located. in pneumococcal meningitis it occurs over the convexities of the brain in the parasagittal region ( figure ) . microscopic examination reveals neutrophils filling the subarachnoid space with extension of the inflammation into the leptomeningeal veins in fulminant cases. blood spread is the most common means of entry; however other routes of infection include local extension of infection, e.g., paranasal sinusitis, osteomyelitis, direct implantation, and via the peripheral nervous system. diffuse bacterial meningitis may follow rupture of a brain abscess, which may lead to sudden death. the organisms may be demonstrated by microbiological culture of the csf and examination of gram stains of the csf and brain tissue. bacterial urogenital tract infections fulminant acute bacterial pyelonephritis may lead to septicemia, causing sudden death. at autopsy, the kidneys show tubular necrosis with interstitial suppurative inflammation. renal papillary necrosis may also be present. severe bacterial enterocolitis may lead to sudden death, especially in the young. the pathogenesis of the diarrhea depends on the cause. vibrio cholerae and clostridium perfringens cause diarrhea by ingestion of a preformed toxin that is present in contaminated foods. enteroinvasive organisms such as salmonella, shigella, and enteroinvasive escherichia coli invade and destroy mucosal epithelial cells. death occurs as a result of dehydration and electrolyte imbalance. bleeding peptic ulcers that are caused by helicobacter pylori may be the first indication of an ulcer and account for % of ulcer deaths, many of which are sudden and unexpected. fulminant bacterial peritonitis secondary to acute appendicitis, acute salpingitis, ruptured peptic ulcer, diverticulitis, strangulated bowel, and cholecystitis may cause sudden death. primary peritonitis may occur postsplenectomy and in patients with splenic hypoplasia. patients with sickle-cell disease may have anatomical or functional asplenia. the former is due to repeated bouts of infarction leading to autosplenectomy. the latter is due to a defect in opsonization of encapsulated bacteria. massive bilateral adrenal hemorrhage with adrenocortical insufficiency may occur as a result of septicemic shock from overwhelming bacterial infection (waterhouse-friderichsen syndrome). the most common association is with neisseria meningitidis septicemia; however, other virulent organisms, e.g., h. influenzae and pseudomonas species, may also lead to this syndrome. sudden death due to fungal infection may occur in an immunocompromised host such as in hiv/aids. organisms include cryptococcus (meningitis or disseminated disease) and pneumocystis carinii (pneumonia). intravenous drug abusers are susceptible to endocarditis due to fungi such as candida. these patients are prone to fungal thromboembolism, leading to sudden death. sudden death may also be due to a complication of fungal diseases such as fatal subarachnoid hemorrhage complicating actinomycotic meningitis or fatal hemoptysis complicating pulmonary mucormycosis. diagnostic modalities include culture of the organism and the histological demonstration of the organisms in tissue. this may be facilitated by special stains such as the periodic acid-schiff (pas) or grocott's methenamine silver stain. fatal cardiac tamponade may occur with intrapericardial rupture of an amebic liver abscess due to entamoeba histolytica. fatal amebic meningoencephalitis may be caused by naegleria fowleri. the organism enters the arachnoid space through the cribriform plate of the nose. there is meningeal hemorrhage with fibrinoid necrosis of blood vessels. cerebral malaria does not usually cause sudden death. however, it may be the primary cause of sudden death in nonimmune persons. susceptible individuals are tourists, business travelers, and sailors. at autopsy, the brain is swollen and may have a ''slate gray'' color due to the brown-black malarial pigment called hemozoin. histology reveals petechial hemorrhages as well as intravascular parasitized red cells. small perivascular inflammatory foci called malarial or dü rck's granulomas may be present. sudden death in malaria may also be due to rupture of an enlarged spleen. an enlarged spleen is fragile and more vulnerable to rupture. other infections that may lead to splenic rupture and sudden death are infectious mononucleosis and typhoid. sudden death due to cardiac involvement in chagas disease (trypanosoma cruzi) occurs in - % of acute cases. the damage to the myocardium causes fatal ventricular tachycardia. histological examination shows myofiber necrosis with an acute inflammatory reaction. clusters of organisms may be found within dilated myofibers, resulting in intracellular pseudocysts. clinically occult helminthic diseases such as hydatid disease (echinococcus granulosus) and neurocysticercosis (taenia solium) may cause sudden death. in neurocysticercosis death may occur due to epilepsy or raised intracranial pressure. parasitic cysts containing scolices are present, especially in the subarachnoid space, cortical sulci, and cortical gray matter. large multilocular cysts (racemose cysts) may be present in the basilar cisterns near the cerebellopontine angle ( figure ) . isolated cardiac hydatid cyst is an uncommon manifestation and accounts for fewer than % of all hydatid disease. sudden death may be the initial manifestation of the disease. death may be due to involvement of the left ventricular myocardium or to massive pulmonary embolism. all autopsies must be approached using universal precautionary principles. in sudden deaths complete autopsy examination is recommended with appropriate tissue and body fluid sampling for special investigations. autopsy sampling for microbiological investigations is indicated in the following circumstances: sudden unexpected deaths in children and adults, deaths in immunocompromised patients, deaths in patients with clinically suspected infections, and deaths with organ changes of infection. the problems encountered with autopsy microbiological testing are contamination during procurement of the sample because of poor technique or due to the postmortem spread of commensals. to prevent false-positive postmortem blood cultures the following should be observed: the body should be refrigerated as soon as possible; and movement of the body should be limited to decrease the possibility of postmortem bacterial spread. an aseptic technique should be used to collect the sample, which should be stored and transported in the correct medium and temperature. close liaison with the microbiology and virology laboratories is important to guide collection, preservation, transport, and evaluation of specimens. this is particularly important in cases where there are positive cultures with negative histological findings. sampling at multiple sites and determining the antibiotic sensitivities may be helpful in determining the significance of positive cultures. the finding of a ''pure'' as opposed to ''mixed'' culture helps to determine the significance of the findings. the type of organism in relation to the site where it was cultured also helps to differentiate contaminants from significant positive cultures. relevant special techniques should be used by the pathologist in order to improve the diagnostic yield in infectious diseases ( table ) . in a small group of cases (so-called negative autopsies) no obvious cause of death is apparent after detailed initial external and internal examination. the incidence of negative autopsies is %- %; this figure improves to about % when special tests such as postmortem chemistry and microbiology are carried out. infectious agents are not a common cause of sudden death. even in cases with little or no morphological changes, investigation of appropriate autopsy samples by recently developed laboratory techniques may prove invaluable and shed light on the cause of death. children: sudden natural infant and childhood death; sudden natural death: cardiovascular; central nervous system and miscellaneous causes in the year an estimated people died from suicide around the world. this represents an annual global mortality rate of . per population. according to the world health organization (who), suicide is the th leading cause of death worldwide. it leads among violent causes of death (e.g., suicide, homicide, traffic deaths). among those aged between and years, suicide is the fourth leading cause of death, and violence against the self is the sixth leading cause of disability. suicidal behavior ranges in degree from merely thinking about ending one's life, through developing a plan to commit suicide and obtaining the means to do so, attempting to kill oneself, to finally carrying out the act of ''completed suicide''. the term ''suicide'' is based on the latin words sui (of oneself) and caedere (to kill). the encyclopaedia britannica defines suicide as: ''the human act of self-inflicting one's own life cessation.'' however, it is often difficult to reconstruct the thoughts of people who commit suicide unless they have made clear statements before their death since all suicidal deaths are not clearly planned. in many legal systems, a death is certified as suicide if murder, accidental death, and natural causes can all be ruled out and if the circumstances are consistent with suicide. this article deals with fatal suicidal behavior. this is the term proposed for suicidal acts that result in death and that directly concern forensic medicine; it does not cover nonfatal suicidal behavior, attempted suicide, or deliberate self-harm, i.e., suicidal actions that do not result in death and which may be referred to psychiatrists. even if it is not always clearly planned, suicide is a result of an act deliberately initiated and performed by a person in expectation of its fatal outcome. suicide is also now a major public health problem, as evidenced by epidemiologic data. according to who, taken as an average for countries for which complete data are available, the age-standardized suicide rate for was . per . the rate for males was . per and for females . per . the rate of suicide is almost universally higher among men compared to women by an aggregate ratio of . to . for some countries the most recent data are shown in table . over nearly years , for countries for which complete data are available, the suicide rates seem to have remained quite stable. geographically, changes in suicide rates vary considerably. according to the french national institute on demographic studies (ined; institut national des etudes démographiques), which provides reliable information on suicide mortality, the rates range from . per in the russian federation to . per in hungary, . per in japan, an introduction to neuropathology forensic pathology introduction to medico-legal practice general pathology of infectious diseases demonstration of infectious agents in tissue key: cord- -hrl ywth authors: sens, mary ann; koponen, mark a.; meyers, sarah title: other pediatric accidental deaths date: - - journal: forensic pathology of infancy and childhood doi: . / - - - - _ sha: doc_id: cord_uid: hrl ywth injury, both intentional and accidental, is the most common cause of death in children throughout the world. many injury patterns that are seen in children are similar to those in adult populations; however, others differ, reflecting the age, stature, and development of children. this chapter focuses on differences in injury patterns of children across the spectrum of childhood development and growth, including injuries and fatalities that primarily occur in infancy and early childhood. risk factors for injury and death are identified within the context of childhood development. topics covered include deaths associated with motorized and nonmotorized vehicles including pedestrian, occupant, and operator fatalities; farming and ranching deaths; drowning, boating, and diving deaths; fires and burns, including electrical deaths; animal-related deaths; falls; and airway-associated deaths. there are significant disparities in childhood injury and death among racial, ethnic, geographic, and socioeconomic groups. a brief consideration of these differences is provided to assist in forensic case examination. accidental injuries cause a high proportion of deaths in children. in some age groups, injury is the leading cause of childhood death. injuries and death differ across the spectrum of childhood development and growth (agran et al. (agran et al. , . an understanding of these differences is critical to the performance of a forensic investigation and autopsy in children and adolescents; conversely, well-performed forensic investigations and autopsies will identify risk factors and opportunities to reduce childhood morbidity and mortality from injuries. pediatric vehicle-related injuries and deaths are commonly seen in forensic pathology. infants and children may be occupants in motor vehicles or may be pedestrians struck by motor vehicles. children may be operators of recreational vehicles, including bicycles, and in adolescence, drivers of motorized vehicles. worldwide, over , children die annually in road-related crashes, and children accounted for % of all road fatalities (world health organization b). childhood deaths from vehicle-related events are highest in africa and are consistently increased in low-and middle-income countries as compared to higher-income countries throughout the world (world health organization ) . each year in the united states (usa), about , children under the age of years are fatally injured while passengers in motor vehicles, and another , sustain injuries, many of which significantly impair the child's quality of life (department of transportation (us), national highway traffic safety administration (nhtsa) ). the centers for disease control and prevention (cdc) in the usa estimates that restraint systems for children (positioning, age, and weight-appropriate car or booster seats, lap and shoulder harnesses, etc.) prevent - fatalities each year (centers for disease control and prevention e). these same initiatives also reduce childhood fatalities in other regions of the world. however, restraint systems may also cause injury and death, particularly when improperly used for the age, size, and development of the child, or when a child is seated in the front seat of a vehicle. the american academy of pediatrics and the cdc have each produced guidelines and best practices for child passenger safety (centers for disease control and prevention e; american academy of pediatrics committee on injury, violence and poison prevention ). children under the age of years should always ride in the rear seats of vehicles. as a general guideline, rear-facing car seats should be used until the age of years. after years of age, forward-facing child seats should be used until the child is over years of age or lb ( kg). belt-positioning booster seats should be used after that point until adult lap-shoulder seat belts fit appropriately at upper thighs and chest when a child is over years of age and ( . m) tall. disabled children have a similar incidence and distribution of injuries related to safety restraint use as compared to children without disabilities (huang et al. ). a review of forensic findings from passenger airbag injuries demonstrates that face, upper extremity, and chest injuries are more common in children than adults, and isolated head injuries are more common in infants as compared to older children (sato et al. ) . in the majority of airbag injuries, a child was unrestrained or improperly restrained in the front seat, and all injured infants were in the front seat in rear-facing child restraint seats ( fig. . a, b) . the smaller stature of children and infants and the position of the head closer to the upward deploying airbag may account for the mechanism of these injuries. the relatively flat and horizontal character of the atlanto-occipital (ao) articulation in children also places them at greater risk of ao dislocation or separation (saveika and thorogood ) . other large studies have confirmed passenger airbag injuries when children under years of age occupy the front passenger seat (newgard and lewis ) . head injuries occur in children restrained with adult seat belts (fig. . a, b ). an adult-designed shoulder harness may cross the child's face and/or neck. children may slip the belt under an arm or behind the back, compromising the restraint of the upper torso making these areas susceptible to injury in a collision with rapid deceleration. in a crash, the unrestrained forward motion of the child's upper body may cause the head to collide with knees or parts of the vehicle. small children may also "submarine" completely out of an adult restraint system on impact, become airborne, collide with the interior of the vehicle and/or other occupants, or may be ejected from the vehicle (tibbs et al. ). children are more vulnerable to blunt abdominal injury from vehicular seat-belt restraints as compared to adults, particularly when child booster seats are not used or the child is inappropriately restrained by the lap-shoulder harness system designed for adults (lapner et al. ) . current recommendations call for use of special child restraint systems until the child is over ( . m) tall, and the lap restraint can be positioned on the upper thighs (not the abdomen) and the shoulder harness crosses the chest while not crossing directly over the neck (american academy of pediatrics committee on injury, violence and poison prevention ). the immature pelvis of a child under the ( . m) height requirement cannot be appropriately restrained with the lap-shoulder harness until the child is tall enough to position the harness in the correct areas (across the thighs and not the abdomen). abdominal-wall contusions are infrequent in both optimally and suboptimally seat belt-restrained children, but when abdominal contusions are present, they often herald intra-abdominal injury (lutz et al. ) . intestinal injury, perforation, and shearing of the fixed ligaments and mesenteric structures are the most common abdominal manifestations of pediatric seat belt-related injuries (davies ) . because the small anterior pelvis of a child is not appropriately stabilized by the lap restraint system designed for adults, deceleration allows the seat belt to override the child's pelvis, producing hyperflexion of the child over the lap belt. this forces the abdominal organs against the vertebral column and momentarily increases the intraluminal pressure of the hollow organs, resulting in sudden compression. in addition to the abdominal injuries, the deceleration-induced compression and hyperflexion of the abdomen may also cause lumbar spine and sacral injuries (hart et al. ; papavasiliou et al. ) . intestinal compromise may be immediate or may occur days to weeks after the injury (hardacre et al. ; lynch et al. ) . pedestrian deaths in children demonstrate age differences in the types and patterns of injury. in children under the age of years, deaths from "backover" accidents in driveways peak in incidence, likely due to the small stature of children, their tendency to "hide" or play behind parked vehicles, and their lack of awareness of the dangers of moving vehicles. crush injuries of young children, usually occurring from low-velocity impacts resulting from vehicles backing out of home driveways, are often fatal due to blunt-force injuries of the head (including closed-head injuries), skeleton, and torso (partrick et al. ) . the most common vehicles involved in these accidents are sport utility vehicles (suv), trucks, and minivans, as opposed to passenger cars (sedans, compact cars, etc.), perhaps due to the increased frequency of larger vehicle ownership by families with small children and the increased "blind spot" of vehicles with higher ground clearance behind the vehicle. in contrast to most other vehicular head injuries, these "backover" injuries are generally static; that is, they do not have significant rotational and accelerationdeceleration components. pedestrian injuries of older children tend to be on roadways rather than home driveways. this reflects cognitive awareness of the age groups: younger children are more at risk of mid-road impacts, associated with impulsive behaviors, such as darting into traffic in areas other than crosswalks; older children and adolescents are more likely to be involved in pedestrian accidents at crosswalks and intersections. as compared to head injuries sustained in "backover" collisions, head injuries in pedestrian roadway accidents in older children and adolescents tend to be complicated by rotational and acceleration-deceleration components as opposed to isolated static forces. the site and character of pediatric pedestrian injuries also reflects the child's age and development (chakravarthy et al. ). the point of vehicular impact (bumper) in a small child is at the head or chest level, both sites that are above the child's center of gravity. because of this injury location, the initial impact usually results in a second impact, with the initial impact projecting the child away from the vehicle and against another fixed object, usually the ground. consequently, the child may be run over by a subsequent vehicle. in contrast, the impact point in older children is below the child's center of gravity, resulting in impact patterns similar to that of adults. in older children, the impact point is at the legs with the body projected over or onto the hood of the car. rotational motion is seen in these impacts with the legs rotating over the head on impact and with increasing speed, and the entire body may somersault over the vehicle following hood/windshield impact. the injury patterns seen in children reflect these different impact patterns, with head, neck, and chest injuries more common in pediatric pedestrian injury populations than in the adult population (chakravarthy et al. ). pedestrian accidental deaths are more common in males and in children from minority groups, particularly in lower socioeconomic areas (american academy of pediatrics committee injury, violence, and poison prevention ). this may relate to socioeconomic considerations including denser traffic patterns, increased vehicular speed, on-street parking, fewer guard-monitored crosswalks near schools or playgrounds, poor sidewalk maintenance, increased need for pedestrian transportation within lower socioeconomic communities, and reduced safety awareness (hotz et al. ). gender differences, as well as the potential influence on parental supervision, have been examined in the context of improving pedestrian safety (barton and schwebel ). geographic information system (gis) studies have assisted in targeting high-risk areas and led to a reduction in both injuries and fatalities (weiner and tepas ; statter et al. ). children may be drivers/operators of a variety of nonmotorized vehicles (wagons, bicycles, gravity go-carts) or "off-road" non licensed motorized vehicles (golf carts, all-terrain vehicles, snowmobiles, motorized go-carts). accidents from these vehicles cause a disproportionate incidence of injury in children when compared to adult populations (siman-tov et al. ; lord et al. ; hamming and henry ; curran and o'leary ; kelleher et al. ; decou et al. ; rice et al. ) . children operating bicycles may be injured from falls, collisions with objects, or collisions with other vehicles, including motorized vehicles. head injuries, including traumatic brain injury, are the most common cause of fatalities involving bicycles; however, blunt-traumatic injury to abdominal organs, abdominal-wall injury, and genital injuries may occur, usually from impact with handlebars or bicycle frame (de jong et al. ; klin et al. ; rowell and chin ; widni et al. ; nellensteijn et al. ). helmet use reduces the occurrence and severity of head injuries in bicycles and other motorized and nonmotorized vehicle accidents. (barnes et al. ; juang et al. ; pardi et al. ). some trauma centers report an increase in severe thoracic and abdominal injuries from bicycle accidents in children as helmet use increases, reducing the incidence of severe head injuries. (klin et al. ) . a review of , bicycle-related injuries in the national pediatric trauma registry (usa) further identified children with mental disorders as a high-risk group for bicycle injury and suggested targeted prevention programs (li et al. ) . training and licensing prior to the operation of off-road and all-terrain vehicles is not as regulated, or as common, as licensed road vehicles (upperman et al. ) . the vehicles are often operated off-road and on irregular terrains, which contribute to vehicle instability and subsequent vehicle rollovers (finn and macdonald ) . crush injuries and asphyxia may result from rollovers of heavy off-road vehicles onto individuals. blunt trauma, fractures, and amputations may occur from impacts with trees, fences, or other obstacles. "clothesline" injuries, including deep lacerations or cervical fracture, may occur to the face and neck when vehicles pass under horizontally strung fence lines. in some instances, multiple children are involved in an accident when an off-road vehicle is being driven with children riding on sleds or skateboards tied behind the vehicle (similar to waterskiing). if the vehicle makes a sudden turn, the increased rotational velocity may throw the child who is on the skateboard/sled away from the vehicle, resulting in blunt-force injuries due to the primary impact and/or a secondary impact with a subsequent structure (tree, vehicle, etc.). the majority of these accidents occur at lower velocities than traditional motor vehicle accidents, although some motorized off-road vehicles can reach speeds of mph ( km/h) or more, increasing the likelihood of fatal injuries and approaching patterns seen in traditional motor vehicle collisions. adolescent drivers of licensed motor vehicles have higher accident and death rates as compared to adults. injury patterns are similar, since adult size and stature are generally reached by the age of legal vehicular licensing. many provinces, states, and countries have limited the adolescent ( - years of age) motor vehicle licensing in several manners, including increasing the minimum driving age, mandating driving courses, restricting hours of driving, limiting the number of adolescents in a vehicle, or requiring a licensed adult driver in the car (russell et al. ). these and other regulations reflect efforts to reduce accidents and fatalities among adolescents until driving skills become more accomplished. some jurisdictions may also limit or have additional requirements before licensing adolescents to operate motorcycles or may restrict operators less than years of age to a smallersized motorcycle. on private property and some indian reservations, state or provincial driving laws may not be applicable or enforceable, and children may drive motorized vehicles at a younger age. there are some unusual vehicle injury mechanisms that forensic pathologists need to consider in pediatric death investigations. infants are more susceptible to carbon monoxide toxicity due to the greater affinity of fetal hemoglobin. individuals with hemoglobinopathies, such as sickle-cell anemia or thalassemia, may also have increased susceptibility to carbon monoxide toxicity (blumenthal ) . rarely, defective motor vehicle exhaust systems may produce high levels of carbon monoxide within the interiors of moving or idling vehicles. "open-air" carbon monoxide poisoning is a well-known hazard around combustion engine exhaust, even from vehicles in outdoor spaces. asphyxia and positional asphyxia may occur with vehicle rollovers, entrapment within vehicle restraint systems, or in automatic window or door closure systems. findings at autopsy are similar to other asphyxia deaths, with petechial hemorrhages often seen in eyes and intrathoracic cavity and hemorrhage within sinuses and sphenoid processes. cyanosis, often intense, is present above the restraint area. often the lividity patterns reflect the entrapment, clothing, and/or vehicular restraints ( fig. . a-e). delayed sequelae of motor vehicle injuries may cause death days or weeks after the motor vehicle accident, following a period of apparent recovery in the intervening time. in these instances, investigative skill is required to correctly ascertain the initiating cause of injury. chest impacts may produce cardiac contusions, including injury to ventricular free walls, papillary muscles, or the ventricular septum. these injured areas may rupture acutely, causing death ( fig. . ) . alternatively, the contusion may result in an aneurysm, infarction, or ventricular rupture days or weeks after the initial injury. cardiac contusion results from blunt-traumatic impact to the chest with compression of the heart against the sternum and/or spine. the momentarily high intrathoracic pressure with deceleration may also contribute to this type of injury. abdominal trauma may produce ileus or injury to hollow organs, particularly the small intestine, resulting in rupture and/or peritonitis several days following the injury. vascular trauma does occur in children, but less commonly than in the adult population (eddy et al. ; riches et al. ; choit et al. ) . these vascular injuries usually present acutely with hemorrhage or infarction, but may be delayed in discovery or presentation. although uncommon, children may also develop pulmonary thromboemboli as a result of local trauma to the leg (fig. . ) or prolonged immobilization and present as sudden death days or weeks later. death from hyperthermia of infants and children left in cars is well known and increasing in incidence since the recommendation of placing infants in carriers in the back seat. usually these tragic deaths result from a caregiver forgetting that the infant is in the vehicle. the temperature in closed vehicles may rapidly rise well over the ambient temperature, and car interiors may reach o f ( o c) or more within minutes (mclaren et al. ) . infants are most at risk for this event since older children may be able to independently leave vehicles and/or alert adults to their presence. autopsy findings are generally nonspecific in car hyperthermia deaths, although intrathoracic petechiae are often seen (fig. . ) . body temperature at time of discovery is important to confirm hyperthermia; however, this is often not available to the forensic investigator. interior vehicle temperature is often not recorded and, if available, may not be reliable, as opening the car door will rapidly disperse heat, resulting in an erroneously low measurement. due to the rapidity of heat exposure in a vehicle, vitreous electrolytes may not demonstrate a dehydration pattern, as is seen in slower exposure to high temperatures. if a victim survives, complications of disseminated intravascular coagulation (dic) and multisystem organ failure may be fatal hours-to-days after the hyperthermic event. there is considerable variation in the certification of the manner of death in these tragic events, with some practitioners certifying as accidents and others as homicides. collisions with trains, either as an occupant in a vehicle or as a pedestrian, have high fatality rates. in contrast with adult pedestrians who are usually intoxicated and trespassers on rail yards, railroad employees, or individuals committing suicide, child pedestrians may be playing on or crossing rail tracks, riding bicycles or walking next to a moving train, or attempting to board, exit, or ride on top of a train. traumatic amputations and head injuries are the most common features of these accidents (thompson et al. ; blazar et al. ). farm-and ranch-related injuries and deaths in children are slowly decreasing in many parts of the world; however, farm and ranch operations remain among the most dangerous occupations, second only to mining, in terms of injuries and deaths (solomon ; lachowski ; cdc b) . unlike mining, children are often involved in farming and ranching activities, exposing them to dangers, both by working and living within farming and ranching environments (pickett et al. ) . most childhood injuries occur in a nonworking capacity rather than being actively engaged in agricultural work at the time of the injury (hendricks and goldcamp ) . estimates of the cost of injuries and deaths of children on us farms alone exceed . billion usd annually (zaloshnja et al. ). there are several dangers in the farm and ranching environment. heavy equipment is used daily, creating risks for children by operating equipment, riding with others on the equipment, or simply being in the vicinity of operation. farm machinery is consistently the most common cause of injury and death in both the north american and the uk farm-related incidents (angoules et al. ). drowning in natural lakes, irrigation ä fig. . (continued) lividity patterns contrast with blanched areas of compression. note the imprint from eye glasses. (b) scleral petechiae from the same child. numerous petechial hemorrhages, some confluent, are present in the sclera, conjunctiva, and the periorbital skin. retraction of the eyelid may be necessary to fully appreciate the presence and extent of petechial hemorrhages. petechial hemorrhages are caused by rupture of small vascular channels, usually venules, and are often present in cases of asphyxia, classically above the point of compression where venous return is compromised. (c) petechiae on the visceral surface of the lungs in the same child. (d) epicardial petechiae from the same child. visceral pleural and epicardial petechiae were first described as "tardieu spots" and initially described as pathognomonic of asphyxial deaths. they are commonly seen in asphyxia, but may be present in a variety of non-asphyxial mechanisms of death. (e) congestion and hemorrhage within the inner ear of the same child. pressure changes from compression may cause sphenoid and inner ear hemorrhages similar to those seen in barotrauma or drowning deaths dams, or streams within farms is another common cause of death in farm children. asphyxial deaths can occur from hypoxic closed spaces or slipping into grain storage bins. lacerations, skeletal fractures, amputations, and blunt-traumatic injuries result from falls occurring in farming activities or entrapment in farm equipment. complex tools, firearms, and sharp implements are necessary for farming and ranching and pose a danger of injury and death in children. falls from, or kicks by, large animals and a wide variety of animal bites also occur more commonly within rural settings. off-road vehicles, such as atvs and snowmobiles, are more common in rural settings, increasing the risk of injury and death from the use of these vehicles by children (lim et al. ). there are minimal gender differences in injury and death in children under the age of years in farm and ranching activities; however, over the age of years, boys have a higher rate of injury than girls . boys are more involved in hazardous tasks involving heavy equipment and tractors, whereas girls more commonly are involved in animal care and agriculture. disparities exist for minority populations with higher incidence of injury and death in farms owned by hispanics, african-americans, or native americans (goldcamp et al. a, b; layne et al. ). finally, the remote location and low population density of many farming communities lead to slower emergency response and increased travel time to trauma centers. all of these factors tend to increase the mortality of injuries. tractors are the leading cause of farm fatalities and injuries in all age groups, but pose several particular hazards to children (schwebel and pickett ) . inspection of the safety status of tractors in rural kentucky (usa) revealed significant safety issues in the majority of examined vehicles (cole et al. ). similar conditions of tractors are likely present in most rural communities and explain the high incidence of injuries and fatalities related to children operating, or present around, this type of . thymic petechiae were seen at autopsy. (b) epicardial petechiae and hemorrhages were present. petechiae were also present on the visceral pleural surface. intrathoracic petechiae are consistently seen in hyperthermic deaths and are attributed to terminal gasping in an auto-resuscitation attempt farm vehicle and the diverse forensic findings that may occur in tractor deaths. nearly half of the tractors examined in the study lacked rollover protection structures (rops). most tractors did not have seat-belt and harness systems, or these systems were difficult to use, putting all operators, particularly young operators or youngsters riding with adults, at risk. many tractors had a narrow front-wheel stance, creating stability issues and increasing the risk of rollover accidents. many tractors had loose or damaged seats. protective shields for starting were present on just over half the trailers examined, and nearly a third had the starting mechanism fully exposed. rearwheel fenders exposed operators to moving tires in nearly % of the tractors. fully functional mounting and dismounting access steps and handrails were present in less than half of the operating tractors and, when present, were designed for an adult body. many tractors had no functional lights, and nearly % had no rear-view mirrors. complicating these design and safety issues is the fact that all of these devices are designed for adult operators, not children. ergonomic studies have demonstrated that even optimally functioning devices may have significant limitations for young operators who have not reached adult stature (chang et al. ). operation of tractors, equipment, and vehicles is common on family farms well before the age of normal driving licensing. in the usa and canada, children from farms and ranches have an average of years' experience in driving tractors and other farm equipment on family farms prior to formal licensing for motor vehicles (marlenga et al. ) , indicating that many farm and ranch children routinely begin driving farm equipment at the age of years or younger. children on farms and ranches also have higher incidence of falls, lacerations, animal bites, and large-animal blunt-trauma injury. when these are fatal, the trauma is similar to that seen in adults. with any perforating injury, secondary microbial infection may occur. secondary infections may be more common with injuries sustained on farms and ranches, reflecting contamination with dirt, animal excrement, and other foreign substances present in these environments. farm and ranch residents may also be exposed to pesticides, herbicides, and a variety of hazardous inorganic chemicals, many of which are not common in urban settings and may cause severe injury and death (neidich ) . the use of alternative or supplemental heating, such as woodstoves, contributes to the higher incidence of rural fires and fire-related deaths. even exposure to hazardous crops, such as tobacco, may be a cause of morbidity in farm children (mcknight and spiller ) . drowning is the second leading cause of childhood death by unintentional injury in the usa, the first leading cause in australia, and the third most common cause in the world (world health organization c). actual deaths may be higher, since world figures do not account for flooding, tsunamis, boating, and water-transport deaths. drowning is more common in children under the age of years, a fact likely associated with minimal swimming skills, lapses in supervision, impulsive child behavior, and attraction to water. drowning is more common in males and often increased in lower socioeconomic classes and minority groups with children of african-american heritage at . times the rate of caucasian agematched children and native american/alaskan native children . times the rate of caucasian age-matched children (centers for disease control and prevention c). infants and nonmobile children are at highest risk of death in home bathing situations or immersion in buckets containing water. older children more commonly drown at swimming locations such as pools and natural water sites. most of the drowning fatalities are related to the child's inability to swim; however, even children with proficient swimming skills acquired in pools or still water may drown with unexpected tidal and current flows present in some natural bodies of water. the contribution of trauma, natural disease, carbon monoxide, drugs, alcohol, hypothermia, or other risks must be investigated in drowning deaths. many drownings occur when individuals have no intention of entering the water, that is, falling off docks or boats, water-transport accidents, or breaking through ice in a snowmobile or car. according to statistics of the international life saving federation (ilsf), approximately % of drownings occurs in individuals with swimming skills, or in water less than a meter deep, and % within m of the pool edge or shore (international life saving federation ). drowning is an asphyxial death resulting from immersion in a liquid, usually water, with resulting anoxic changes. "near drowning" is defined as a recovery or resuscitation following submersion/immersion. following a "near-drowning incident," individuals may recover completely or succumb at a later time from complications of anoxic injury and secondary medical complications. rarer occurrences of drowning deaths include "hyperventilation drowning," seen in swimmers, usually youths, who hyperventilate to increase underwater swimming distance. the resultant hypocarbia may decrease the physiological stimulus for air, resulting in anoxia, unconsciousness, and subsequent drowning. "secondary drowning" and "delayed drowning" are terms used when an individual, often a young child, aspirates water and appears to recover but develops a sudden deterioration of pulmonary function several hours after the event. some use the term "immersion syndrome" when an individual has a catastrophic event, usually cardiovascular, while in the water and dies from the precipitating event or loses consciousness, resulting in drowning. certification of these immersion-syndrome deaths may be problematic and is often classified as accidental drowning when the precipitating event cannot be recognized or proven. a consideration of potential components of drowning assists in the forensic interpretation of findings (burford et al. ) . the pathophysiology of drownings involves lung and airway effects with resulting cerebral anoxia. in most drownings, the initial insult results from water within the airways and lungs causing inadequate oxygenation. aspiration of mud, debris, sand, or vomit into the bronchial tree and alveoli may also occur and complicates the physiological response. this also increases the risk of acute respiratory distress syndrome (ards), chemical pneumonitis, and pneumonia if the individual is successfully resuscitated. freshwater inactivates pulmonary surfactant on contact, resulting in atelectasis, further complicating oxygenation and increasing the pulmonary blood shunting physiological response. osmotic forces may draw more fluid into alveoli from pulmonary intravascular spaces. the resulting anoxia and acidosis have the most immediate and consequential impact on central-nervous-system (cns) function, although other organ systems may be involved in pathophysiological changes. these changes include cardiac dysrhythmias, blood-pressure changes from catecholamine release, carbon dioxide retention, and hypothermia. metabolic acidosis is significant in "near-drowning" victims. although initial experiments suggested that fluid and electrolyte imbalance and hemoglobin/hematocrit changes from osmolality differences played a major role in drowning, these changes may not be as common or as important as first thought. deaths are reported in "secondary" or "delayed drowning" scenarios where acute lung injury and rapid pulmonary failure from inhaled water, debris, or chemicals within water produce death several hours after the event. forensic findings in all drowning deaths are nonspecific; thus, drowning is a diagnosis of exclusion. numerous tests have been proposed for establishing the diagnosis of drowning, including diatom presence and electrolytes or other substance differences in right versus left cardiac blood; however, none have proven sufficiently valid for forensic use. tenacious froth, often with slight blood tingeing, may be present in the mouth and nares and is thought to result from an admixture of water, pulmonary surfactant, mucus, and blood plasma protein (fig. . ) . this finding is supportive of drowning but is not present in all cases, and the absence should not be used as criteria for excluding drowning. internally, froth and water are often present in the airway; the lungs are often heavy and edematous. during the process of drowning, individuals may swallow large quantities of water, distending the stomach. this constellation of heavy, waterlogged lungs often with water in the stomach is termed "wet drowning" and is present in the vast majority of drowning deaths; however, in a minority (about %) of drownings, there is no water in lungs, airways, or stomach. laryngospasm is postulated to occur in these cases closing the airway and resulting in "dry drowning." in all drowning cases, the struggle to breathe or submersion below m of water may also result in pressure changes reflected in hemorrhage in the inner ear and sinuses that may occasionally be seen at autopsy (fig. . ) . hemolysis of red blood cells may be present if copious quantities of freshwater are present. often lividity patterns have a bright red tone when drowning has occurred in cold water (fig. . ). all of these findings however are nonspecific and may occur in other deaths. conversely, drowning may occur with none of these signs. finally, in deaths from "near drowning" or "secondary drowning," the findings at autopsy will more closely reflect the anoxic and secondary complications rather than the original drowning event. forensic investigations of all drowning deaths should rule out the presence of trauma or other incapacitating events eg. (insect/aquatic stings, traumatic falls, entanglement in docks/ropes/flora, fatigue) as a component to the drowning. the presence of natural disease, including cardiac arrhythmias, epilepsy, metabolic derangements (eg. diabetes), and other potentially incapacitating conditions, should also be investigated in drowning fatalities. carbon monoxide is present, often in fig. . foam cone in drowning. this -year-old child drowned in freshwater. a tenacious froth, often blood tinged, exudes from the nares and/or mouth in many drowning deaths. it is caused by agonal breathing and the mixing of air, water, and respiratory mucous. it may extend to the trachea and more distal bronchi. although commonly seen in drowning, it is not specific; drownings may occur without this finding, and similar froth may be seen in other non-drowning deaths, such as acute drug overdoses/toxicities fig. . hemorrhage in the mastoid sinuses in drowning. this is a common, but nonspecific, finding in drowning and is thought to be related to pressure changes lethal levels, in boats and around other internal combustion engines, and carbon monoxide quantitation should be performed in any water-or near-water-related death in which gas-powered engines were possibly present in the vicinity at the time of the incident (jumbelic ) . routine drug and alcohol screening should be performed to rule out incapacitating toxicities such as alcohol or opiate intoxication. depending on potential medical and other case findings, additional toxicology or clinical testing, including vitreous electrolyte, ketone, and glucose levels, may be helpful (byard and summersides ) . forensic investigations of drowning deaths should document water temperature to assess the potential role of hypothermia in the context of the case. if there is a possibility of diving contributing to the death, careful neck examination, including a posterior neck dissection and cervical spinal cord examination, should be performed to rule out contributions to neck and cervical injury. fires and burns are a common cause of injury and death in children throughout the world (international association for the study of insurance economics ). burns are usually unintentional, but may be intentionally inflicted. in house fires, children under years are at the highest risk of death, followed by the elderly (centers for disease control and prevention ). fire deaths have striking disparities in occurrence, with victims more likely to be from a lower socioeconomic group and of native american or african descent as compared to caucasians (istre et al. ; flynn ) . the incidences of both fires and fatalities are higher in rural areas contrasted with urban populations. most fire deaths occur in the home, particularly in homes without working smoke detectors. infants and very young children are at greater risk of succumbing fig. . skin coloration in cold-water drowning. lividity patterns may have brighter red tones in cold-water drowning, hypothermia, or refrigeration. carbon monoxide can also produce a cherry red coloration and should be investigated in any drowning where the potential exists for exposure to internal combustion engine exhaust to fire-and combustion-related injury due to the presence of fetal hemoglobinrelated carbon monoxide toxicity and the inability to recognize and appropriately react to the fire. children may tend to hide within the house or seek out sentimental items, such as personal belongings or pets, delaying their escape from the burning house. young children may be the cause of the fire, resulting from inappropriate use of matches, lighters, and candles, and are thus at the fire's origin and are therefore exposed to the first effects of the fire. a brief analysis of the components of fires is helpful in understanding the forensic findings (peck ) . smoke is readily produced in fires, particularly home fires, and comprises the airborne solid and liquid particulates of combustion. smoke is a direct eye and pulmonary irritant and is invariably accompanied by the toxic gas carbon monoxide (co) with environmental levels of % or more. the carboxyhemoglobin (cohb) saturation level in a human exposed to this level of co gas will rise to % or more within seconds (peck ) . other toxic gases such as hydrogen cyanide (from synthetic fibers or materials) or the highly irritating acrolein (from wood and natural products) may also be present (einhorn ) . concomitantly, environmental oxygen levels in the fire environment diminish from the consumption of oxygen by combustion. humans become significantly physically impaired with oxygen levels below %, and mental judgment is compromised at % (peck ) . house fires may smolder with oxygen levels at %, and all fires have an oxygen-deficient environment. the hypoxic environment likely contributes to fatalities in fires, but this impact cannot be directly measured by current technology used at autopsy. thermal conditions of a fire are also important to some forensic findings (peck ) . temperatures in most house fires reach f ( c) within minutes. in some structures, such as an aircraft, much higher temperatures (approximately , o c) may be present within minutes. a fire often reaches a "flashover" point, where the thermal radiation causes surfaces to reach their ignition temperature. at this point, near-simultaneous ignition of materials occurs, and the fire spreads to involve the complete room or structure within seconds. temperatures may soar at this point to over , f ( , c) within the structure. similar rapid acceleration in combustion and temperature may occur when partial collapse of a structure or opening a door creates a rush of oxygen causing rapid fire growth. all fire deaths must be autopsied to confirm the identity of the victim, establish the cause of death, and eliminate non-fire causes of death. the autopsy should include full-body radiographic studies to detect debris, projectiles, or skeletal fractures especially if the external examination is compromised by extensive fire damage. in many children, particularly infants and small children, dental identification is precluded due to the immature developmental stage of dentition and the lack of dental records. this may require investigators to seek dna confirmation of identity when visual identification is not straightforward. although rare, temperatures and duration of house fires may reach cremation thresholds, resulting in limited remains of small individuals, including children, necessitating thorough scene examination. in these instances, assistance from a forensic anthropologist may be helpful in increasing the percentage of remains recovered. the abundant smoke of most house fires is demonstrated at autopsy by readily apparent soot present in nares (fig. . ) and airways (figs. . and . ) . the presence of soot below the true vocal cords is indicative of breathing soot-filled air and is not a fire artifact. in conjunction with the thermal injury, carbon monoxide saturation may be within the lethal range and may play a more significant role in causing the death as compared to the thermal injuries. pulmonary edema and froth in the nares (fig. . ) and upper airway are often present. skin blistering may be present (fig. . ) , and geographic desquamation of the hands, also this may be from direct thermal injury or may occur postmortem as an artifact of the fire/heat referred to as degloving, may occur ( fig. . ). more severe heat injury results in extensive charring and reduction in body mass, sometimes with marked reduction in the size and bulk of the extremities and head. bodies may assume the so-called pugilistic posture, a postmortem artifact of heat-related, differential muscle contraction. victims of "flashover" fires or explosions, such as from natural gas leaks or incendiary devices, often have minimally elevated or normal carbon monoxide saturation levels and an absence of soot in the airways, which may be intensely hyperemic (fig. . ). in this scenario, death is usually related to heat-induced laryngospasm or vagal-reflex cardiac arrhythmia. pulmonary contusion or hemorrhage may be present from blast injuries if an explosion has occurred; debris or bomb material may be embedded in the victims. hemorrhage in the inner ears and sphenoid processes may be apparent at autopsy in such explosions and subsequent fires. . "degloving" in thermal injury. the epidermal layer may blister and "deglove" in fire deaths. this is usually a postmortem artifact. fingerprint impressions may be made from the "glove" to aid in identification if suitable preinjury prints are available for comparison fig. . tracheal hyperemia in thermal injury. in flash fires or steam burns, there may be minimal soot deposition within the airway. the trachea may be intensely hyperemic from burns of superheated air. edema is often present traumatic injuries, falls, and blunt trauma from structural collapse may be present in fire fatalities. these injuries may be the primary cause of death, contribute to the cause of death by restricting egress from the fire, or occur postmortem. heat-related fractures of bones, including the skull, occur in fires and should not be mistaken for premortem injury. a common fire-related artifact is an epidural accumulation of blood and marrow; however, this is readily distinguished from a premortem epidural hematoma by the granular, foamy, or flaky texture and brownish coloration. the victim's skin may be dry with large splitting defects exposing the underlying fascia or abdominal cavities (fig. . ) . clothing or objects near the victim may burn or melt and be deposited on the skin. care must be taken to distinguish these postmortem artifacts from actual premortem injury. assistance from anthropologists is helpful in many fires to accurately recover and identify bone fragments which may be difficult to recognize among fire debris at the scene (figs. . and . ). radiological examination is extremely helpful, but suspicious findings must be confirmed by direct examination (fig. . ) . toxicology testing for alcohol and drugs is important in fire fatalities, since the use of these substances may contribute to the start of the fire or failure to successfully leave the burning structure. testing for carbon monoxide is needed in all fire deaths, and testing for additional toxicants (such as cyanide) may be necessary for a complete forensic investigation. select cases may benefit from co saturation levels at different anatomic sites; for example, co saturation of a fresh subdural hematoma or an area of trauma may reveal different levels compared with to cardiac blood, indicative of prefire trauma contributing to death. in cases where there are no/minimal airway-soot fig. . skin splitting artifact in thermal injury. skin splitting in fires is a known postmortem artifact and should not be mistaken for sharp force injury. note the yellow base of the surface and lack of vital reaction. the slight reddish appearance is from drying and nonlethal-co saturation levels, death prior to fire must be seriously considered in the investigation. if an incendiary device is suspected, consultation with experts at the time of the initial investigation and autopsy will assist in the specialized collection of evidence from victims which is required for case investigation. contact with or immersion in hot liquids or gases is more common in infants and small children than adults. scalding is the most common burn injury in the pediatric age group (shah et al. ) ; however, most accidental scalding injury is not fatal, so this is less commonly encountered at autopsy. the skin in a scalding injury is markedly erythematous, blistered, or, with severe injury, completely denuded with intense underlying erythema. sharp borders are invariably present between injured and uninjured areas. a characteristic of scalding injury is the distinctive injured/uninjured pattern created by different modalities of injury, including pouring, splattering, or immersion. in an accidental scalding injury, irregular patterns of splattering are often present on the face, hands, or upper body often from children pulling or tipping pans of hot liquids from stoves. in immersion injury, larger, more confluent areas of burning are seen which correspond to the child being dipped or held in hot water. the buttocks, back, and feet are small-bone fragments such as this may not be recognized by responders. consultation with forensic anthropologists is extremely helpful in the recovery of human material in a fire scene the most commonly burned areas. characteristic of the immersion burn is the sparing of injury in skin folds, such as inguinal regions or behind the knees, resulting from the child withdrawing the lower extremities to avoid the immersion. large-immersion pattern scalding injury is indicative of inflicted trauma. children under years of age are most at risk of immersion injury, and this form of inflicted trauma is often precipitated by adult frustration with toilet training of the young. deaths from scalding injuries result from direct thermal injury or complications of these injuries. vitreous chemistry and microbiological studies may be of value in investigating these deaths. water is the most common medium for scalding injuries, likely due to its universal availability. infants' and young children's skin is more susceptible to thermal and scalding injury (diller ) , and burns can occur within seconds with water temperatures f ( c) or higher (feldman et al. (feldman et al. , . many residential water heaters are set to this temperature or higher, increasing the risk and ease of injury to children. fatal chemical burns are uncommon in pediatrics but may occur in young children from exposure to caustic household chemicals, such as lye (elshabrawi and a-kader ) . most of the fatal injuries involve ingestion, and deaths may occur following prolonged medical intervention. fig. . misleading radiologic image in a fire death. these fragments were recovered from a house fire. routine radiologic imaging showed a radio-opaque object suggestive of a projectile (a: circled). retrieval and subsequent cleaning of this object (b, c) demonstrated it was a metal button from jeans contact with hot objects is a cause of childhood burns and is often seen in child abuse (toon et al. ). in the forensic setting, these are usually a component of an inflicted-injury pattern and not the primary cause of death. in many instances, patterned impression of the causative object, such as a hot plate, cigarette, poker, or other object, is present. it is critical to document these pattern injuries with extensive photography and detailed measurements. as with other patterned injuries, body curvature and positioning may cause interrupted patterns or may assist in interpretation of body posture or defensive positioning when inflicted. deaths from fireworks are rare, but a large percentage of these fatalities are in children. deaths usually result from blunt-traumatic injury and/or flash fire resulting from fireworks (smith et al. ) . rarely blast and explosive injuries are a component of these deaths. in many burn injuries, death occurs after considerable time has passed from the time of the injury with numerous intervening medical and surgical procedures. the original pattern of burn injury may not be present, and there are usually a number of medical complications leading to the demise. burn victims are susceptible to a wide variety of infections, electrolyte and metabolic derangements, and medical complications including ards, dic, and multisystem organ failure. careful documentation of medical interventions and events is part of the forensic autopsy of these patients. death certification must accurately relate medical complications back to the initiating injury. most childhood electrical fatalities result from low-voltage electrocution in the home (rabban et al. ). household appliance cords or extension cords are the most common causes of electrocution in children, followed by wall outlets. younger children may contact cords orally or via a conductive foreign object such as a key or pin. less commonly, contact with poorly grounded and electrified appliances or heaters may cause electrocution. high-voltage electrocution is rare in young children but may be seen in adolescents, particularly among males. an even rarer event is death due to lightning injury in the pediatric population; however, fetal death due to maternal electrocution may be seen even when the mother survives the lightning strike. the findings of electrical injury in children are similar to those seen in adults, and the elements of forensic investigation and autopsy are identical. electrical injury is challenging since often there are no distinguishing features. a high index of suspicion, through scene investigation and consultation with electricians, is critical to identifying and appropriately categorizing these deaths. low-voltage (< , v) electrocution is the most common type of electrocution in both children and adults. the classic "joule burn" or electrical mark is present in approximately half of the cases. when present, it identifies the point of electrical current entering or exiting the body. the skin may be whitish (fig. . ) surrounding a raised, oval, or round crater with hyperemic borders and a pale center (fig. . ) . the shape of the burn may relate to the conductor shape in contact with the skin. charring or edema is often present, and prolonged contact may result in deep burning of tissue. microscopically, a "streaming" appearance of cells and nuclei is often present accompanied by thermal homogenization of the dermis, deep fascia, and underlying tissue fig. . ). the epithelial basal layer of the skin is often detached. these changes are supportive, but not diagnostic of electrical injury. occasionally, trace evidence of metal deposition from wires may be present at the site of electrical contact. the point where electric current exits the body is usually more difficult to identify and may be entirely absent. when present, it may have many of the characteristics of the site of current entry. deaths from low-voltage electrocution are caused from electrical disturbances, primarily ventricular fibrillation of the cardiac conduction system or respiratory paralysis from interference with brainstem function, both autonomic disturbances that leave no physical trace to be identified at autopsy. high-voltage electrocution (over , v) is uncommon in the pediatric population and, when present, has the same characteristics of blast injuries and flash burns as with adult deaths due to high-voltage electrocution. death is usually related to blast effects, thermal injury, and secondary trauma rather than to an electrical interruption of cardiac or brainstem function. with extremely high voltage, charring and bone fusion may be present, and the muscles will appear gray-white from the effects of temperature. skin may be denuded, and the hair is often scorched or a lighter color following high-voltage electrocution. lightning strikes are uncommon events, and, because of the extremely short time interval of electrical exposure, over two-thirds of victims survive a lightning strike. in the event of a death, the majority of findings may be nonspecific. a faint, erythematous, arborizing pattern on the skin, the so-called lichtenberg figure, may be present initially, but will subsequently fade in a relatively short time frame. rupture of eardrums and opacification of corneas may occur in lightning injury. falls are a part of normal childhood, an expected result of the need for exploration, as the young push their limits in developing coordination and locomotion skills while navigating their environment. most falls are inconsequential; however, each year in the usa, . million children seek emergency department care for falls (centers for disease control and prevention d). analysis of fall data in children can identify intervention strategies to reduce this common cause of injury (khambalia et al. ) . falls are responsible for death in over , children annually (world health organization a) and may be a component of inflicted injury of children. delineation of these injuries is one of the most challenging tasks in forensic pathology. fatal injury from falls usually involves trauma to the head, although extremity fractures and internal injuries can occur and may cause death. the assessment of the degree of injury relative to the presenting story is a key component in distinguishing accidental and inflicted injury in the evaluation of a fall. in general, short-distance falls (less than ft or m) and falls down stairs rarely produce fatal injury to children (chadwick and salerno ; chadwick et al. ; chiaviello et al. a) . the presence of an adult may complicate injury to the child as falls occurring with children being carried down stairs by an adult have a higher injury index and more frequent skull fractures than falls down stairs by children themselves (joffe and ludwig ) . the added mass of an adult and secondary impact(s) may be responsible for the increased severity of injury. numerous studies of falls in specific situations exist in the literature including falls from buildings (vish et al. ) , playgrounds (centers for disease control and prevention a; petridou et al. ) , beds/couches (belechri et al. ) , high chairs (schalamon et al. ) , infant walkers (shields and smith ; chiaviello et al. b ), child-restraint seats (desapriya et al. ) , infant slings/ carriers worn on adult bodies (frisbee and hennes ) , heights (thompson et al. ; murray et al. ) , and in hospital settings (jamerson et al. ). case reports of severe injury from accidental stairway and short-distance falls exist (lantz and couture ) , although the presence of severe head injury is more indicative of inflicted trauma. disparities exist in pediatric falls. falls with injuries are more common in children of nondominant populations and lower socioeconomic classes (faelker et al. ; shenassa et al. ) in most but not all countries (engström et al. ) . this is correlated to older and poorer conditions of environmental factors, such as reduced repair of playground equipment, older model cribs and beds, and poor maintenance of stairwells, balconies, and windows in lower socioeconomic populations. forensic investigation and autopsy in childhood deaths from falls is challenging since inflicted versus accidental trauma is often the question. careful and extensive photography of injuries, full radiological examination, fixation of the brain, and extensive histology are considered a baseline in the performance of the autopsy. specialized dissections, such as removal of optic nerves, eyes, and posterior neck dissections including examination of cervical spinal roots (matshes et al. ) may assist in interpretation. when there is a question of inflicted skeletal trauma, removal of the affected bone and the contralateral control bone may be important for evaluation. careful review of all clinical records and police investigation is required for optimal interpretation. correlation or identifying points of discordance with the given clinical history of how the event occurred is critical for case assessment. it may be helpful to have at least one meeting with all interested professionals, including clinicians providing care, radiologists (if appropriate), police and investigating agencies, child-protective services, and legal prosecutorial representation if jurisdictionally allowed. the adage "we speak for the deceased as discoverers of truth, not makers of cases" rings true in evaluation of the difficult area of pediatric deaths due to falls. the rapid advances in imaging and pediatric head-trauma interpretation will likely assist in defining appropriate studies and interpretations as more evidencebased practices in forensic medicine evolve. most children have many positive exposures to animals -from appreciation of wild animals, visits to zoos, and caring for animals as pets, companions, or livestock. care and responsibility for animals is a major step in normal childhood development; abuse of animals is a critical warning sign of behavioral problems. with this overarching exposure to animals, childhood injuries and fatalities occur in a variety of animal-related scenarios (bury et al. a, b) . the majority of these fatalities occur either with large animals or with commonly encountered animals such as dogs. injuries and deaths associated with large animals are increasingly seen both in and outside rural environments. the non rural increase is primarily due to the popularity of horseback riding among urban and suburban youth. a horse can weigh over , lb ( kg) and reach speeds approaching mph ( km/h), both accounting for the inherent dangers of this sport. this danger is amplified for young children where the injury rate for children in horseback riding is nearly twice that of adults from both falls and kicks from horses (jagodzinski and demuri ) . a mexican proverb states "it is not enough for a man to learn to ride; he also must learn how to fall," alluding to the danger admixed with the joy of this sport. most injury in horseback riding occurs when the rider falls off the horse; occasionally, this is complicated by a secondary kick from the horse, being stepped on by the horse, or through entrapment in a stirrup and subsequent dragging. the use of "break-free" stirrups can decrease the risk of entrapment and dragging following a dismounting fall. in nearly a third of horse-related injuries, the child is not riding the horse, but is kicked while in the vicinity of the horse. head injuries are the most common and the most serious type of injury, accounting for most fatalities in horserelated accidents (ghosh et al. ) . in children, upper-extremity injury is more common than lower-extremity injury. injuries to the back, pelvis, chest, and abdomen do occur, but with much less frequency in children as compared to adult horse enthusiasts. horse-related injury is second in severity only to pedestrian versus motor vehicle injury, and horse-related injury has greater severity than injury from atvs, bicycles, and passenger-related motor vehicle crashes (bond et al. ) . helmet use reduces the severity of injury and the likelihood of death. nearly five million dog bites occur every year in the usa, with over % occurring in children under years of age (centers for disease control and prevention (cdc) ). children, particularly those under the age of years, are more likely to have provoked dogs, often unintentionally through play, and are often incapable of escape or defending themselves from a dog attack. most dog bites are generally single bites to the extremities and are usually not fatal; however, most dog attacks and fatalities involve multiple bites to the head and neck region where extensive mutilating and defleshing injuries may occur. exsanguination and air embolism via neck veins are common causes of death in dog attacks. most dog-attack fatalities have multiple puncture wounds from the teeth and extensive defleshing from the tearing and hunting behavior of the dog(s). severity of injury and increase in fatal attacks may be increased when more than one dog is involved due to "pack" behavior (tsokos et al. ). abrasions and shallow incised wounds from paws or friction contact abrasions from surfaces may also be present. large breed dogs may cause crushing injuries to the very young, including infants and toddlers, resulting in severe skull fractures and cerebral injuries. several studies have examined the incidence and breed characteristics of fatal dog attacks, but may be confounded by the lack of accurate data into breed type actually present in a community as well as the effects of legislation restricting ownership of certain breeds (raghavan ; sacks et al. ) . bite-prevention programs show initial promise in educating young children about safe behavior with dogs (meints and de keuster ) . special considerations at autopsy of a fatality involving a dog bite include documentation of any canine odontological trauma for comparison (de munnynck and van de voorde ) . when an animal is captured, examination of material between the teeth or within the gastrointestinal tract at necropsy may confirm human tissue. consultation with veterinarians is very useful for testing of an animal for rabies, tumors, or other conditions that may have led to the attack, including examination for prior animal maltreatment, starvation, or training for fighting activity. investigation of the events leading to the attack, including a history of dog behavior, breed characteristics, and any precipitation or provocation by the victim, is helpful in reconstructing the events and interpreting injury patterns. bites and other trauma from animals may also be a cause of delayed death from microbial infections contracted from oral animal flora, human skin flora, or subsequent contamination of the original bite (dendle and looke ) . a wide variety of wild animals, on both land and sea, may cause injury and death in both provoked and unprovoked attacks. children are often more susceptible due to their decreased ability to flee or defend themselves, lower cognitive awareness of danger, and provoking animal attack behavior either intentionally or through play. the smaller mass and stature of young children also increases risk of fatal outcome of a wild-animal attack. many animals may attack causing injury and death including bears, feral dogs, coyotes, cougars, and wolves on land and sharks or other predatory fish in the sea. predatory land animals usually maul and bite victims, and if hunger precipitated the attack, the predators may devour most of the victim. large non predatory animals, such as cattle or moose, may cause injury and death from a stampede or trampling of victims. in water, sharks are the most common human predator. sharks may swim very near land and docks and attack surfers and swimmers. shark attacks are characterized by large bites and tearing of limbs and viscera. examination of the body for bite marks or fragments of teeth may assist in species identification. crocodiles and alligators are territorial animals and may attack individuals within their home wetlands. alligators are generally timid but may lose their fear of humans when fed or with repeated contact. in contrast, large crocodiles are more predatory by nature. both species can move very fast on both land and in water, increasing the danger of attacks to small children. these species will occasionally drag victims underwater, preserving a portion of the body for later consumption. their jaws exert the most pressure of any predator (erickson et al. ) , with up to , lb per square inch pressure ( . mpa, bar, atm), causing instantaneous massive trauma to victims. all these predators produce findings in child victims similar to those seen in adults, although children may be more susceptible due to their small size and inability to defend or flee from an attack. rarely, smaller sea life may sting or envenomate victims, causing injury or death, by either venom or secondary drowning. in all these animal attacks, the findings at autopsy are similar to those in adults. with the wide variety of animals and predators throughout the world, forensic pathologists should be aware of regional species that may cause injury and death. fatal envenomation may occur from several land-based (snakes, scorpions, spiders) or marine-based creatures (jellyfish, stonefish, octopus, portuguese man-o'-war, cone snails). autopsy findings vary with specific venomous species and may be subtle, requiring detailed historical investigation and/or careful cutaneous evaluation for envenomation site (williams and milroy ) . venomous-snake fatalities are uncommon, but young children may be particularly susceptible due to their small body mass ). insect bites and stings may also produce death from anaphylaxis, zoonotic transmission of disease, or rarely overwhelming envenomation. worldwide, the deadliest "animal" is the mosquito, causing malaria across wide areas of the tropics as fig. . cutaneous site of a snake bite. this -year-old child collapsed suddenly while playing in the tall grass at a campsite. her family reported a single scream before the collapse. she was unresponsive and could not be resuscitated. later, this site of likely envenomation was found on her thigh fig. . dissection at site of snake envenomation. hemorrhage is clearly present in the subcutaneous tissue tracts, and direct envenomation of an artery is identified a variety of animals and insects are natural scavengers in the postmortem period (byard ) . the disfigurement caused by postmortem predation is considerable, and care must therefore be taken to avoid confusion with premortem injury fig. . postmortem ant bites. this -month-old infant was found deceased after a sleep period. scene investigators were concerned with the lesions identified on her arm, raising the possibility of child abuse. the irregular and shallow erosions are typical of postmortem ant bites. these normally have a yellow base, but drying artifact can darken the lesions, suggesting a vital reaction. it is critical to correctly assess postmortem artifact from abuse and traumatic injury in any forensic investigation fig. . postmortem rodent activity. lesions from rodent postmortem predation have shallow and irregular contours and are often centered on natural orifices, such as eyes or sites of premortem injury. note the uniform shallow contours and yellow base of this lesion, both indicative of postmortem rodent predation. similar patterns may be seen with some aquatic predators, such as crabs in cases of drowning in crab-infested waters (figs. . and . ). there are striking regional differences in postmortem scavenging activity, and forensic pathologists should be aware of local scavenging patterns. forensic pathologists and investigators should also be aware that household pets often are involved in postmortem predation (buschmann et al. ). choking hazards are greatest in very young children due to the proclivity of infants and toddlers to place objects in their mouths, the lack of cognitive awareness of dangers in this age group, the smaller upper airway, and poorer cough reflex in very young children. in contrast with adults, where alcohol, drugs, and/or neurological impairment may be predisposing factors in airway-associated deaths, these factors are usually not a component of pediatric airway-associated fatalities. food, coins, and balloons are the most common objects causing obstruction in children (altkorn et al. ; rimell et al. ) . total obstruction of the upper airway produces rapid loss of consciousness and death. occasionally, a child is found dead without a history or suspicion of airway compromise, even in a supervised environment. physical findings in airway-associated ("choking") deaths may or may not include external petechiae or head and neck cyanosis; however, internal intrathoracic petechiae are usually present. the object is usually readily found within the larynx or trachea (figs. . , . , and . ), or at the carina (fig. . ). occasionally resuscitative efforts may force the object into the proximal major bronchi. objects smaller than the expected caliber of a child's airway may cause death either through laryngospasm or direct obstruction of a scarred and narrowed airway. children most at risk for airway scarring are those who had been previously intubated, particularly as a preterm neonate. game-playing asphyxial deaths occur primarily in older children and young adolescents (le and macnab ; andrew et al. ). widely termed the "choking fig. . airway obstruction by a balloon. this was the autopsy finding in a -month-old infant who was found deceased by his caregiver. the father had lain down with the infant on a quilt and both fell asleep. on awakening, the father found the infant unresponsive a few feet off the quilt. autopsy revealed an object in the upper trachea game" within the literature and community, this activity usually involves a form of strangulation to achieve a brief period of euphoria. strangulation may be achieved by use of hands, ligature, or noose and may be self-administered or done by others. rarely, a variant of chest compression produces the asphyxial state. the resulting cerebral hypoxia is reported as a "high" with light-headedness or a few seconds of unconsciousness. with release of external neck compression, a "rush" or "tingling" is described among children experiencing this activity. this activity is commonly known among preadolescents and goes by over local variations in both name and activity (centers for disease control and prevention (cdc) ; macnab et al. ) and is seen throughout the world (noirhomme-renard and gosset ). the cases presenting to a forensic service usually involve solo-game playing and are often initially thought to be suicidal hanging deaths (andrew and fallon ) . the physical findings at autopsy are similar to other strangulation or hanging deaths (figs. . and . ) , and the key to recognition is within the scene and death bean obstruction at the carina. this -month-old child was visiting his grandparents, crawling on the kitchen floor during a family gathering. he stopped moving and adults thought he was napping. a few minutes later, they noticed a darkening blue tone to his lips and tried to arouse him. he was unresponsive and ems was called. he presented to the medical examiner's office as a sudden death in infancy during sleep. a sack of pinto beans had spilled in the kitchen the day before; adults thought all were cleaned up. the child had found a single bean in a floor crevice investigation ( fig. . ). the activity is usually hidden from adults, but is widely acknowledged among peers, who usually regard the activity as a safe, drug-free "high," a part of thrill-seeking activity and risk-taking behavior of early adolescence. there are no indications of suicide at the scene or by history, and there is no evidence of autoerotic activity. the incident usually occurs in a private location that the child regards as safe, often within a home or school. the children are often "good kids," without behavioral, mental health, school, or drug/alcohol use problems. there are minimal gender differences in most studies, although some studies show a male predominance. in the usa, it is more common in hispanic and native american/alaskan native populations as compared to caucasian populations (toblin et al. ) . key to recognition at the scene is the "private yet safe" location, the simplicity of the noose, absence of suspension, and facile ability to self-extricate from the compressing object (fig. . ) , which usually is a simple loop against which youngsters lean. there are often wear marks on hooks or furniture from previous episodes or historical incidences of syncopal episodes, voice changes, new onset headaches, abrasions on the neck, disorientation after time alone, or accessing various electronic activities (websites, blogs, chat rooms) fig. . layered neck dissection in ligature hanging. careful neck dissection after removal of the cns and body organs may reveal hemorrhage within strap muscles of the neck, corresponding to the site of constriction for asphyxial games. the absence of suicidal ideation, depression, or farewell messages is also important to ascertain. invariably, peers will either acknowledge similar past group or individual "game playing" or confirm that the victim sought information about game-playing asphyxia. although some describe this as a new risk, it may be a continuum of asphyxial games of past preadolescent generations, such as "breath holding" or hyperventilation to achieve momentary alterations of consciousness. the recent deadly inclusion of ligatures and nooses greatly increases the potential for death, especially when children engage in this activity alone. autoerotic or sexual asphyxia deaths in children are rare and when they occur tend to be in older adolescent males (sauvageau and racette ) . key findings of autoerotic deaths are generally scene-related with sexually explicate material and/or sexual devices present, privacy of the scene, evidence of masturbatory activity, sexual fetishes, bondage or masochistic activity, and often elaborate asphyxial producing and escape mechanisms (shields et al. a, b) . a review of cases from australia and sweden revealed no autoerotic deaths below the age of and confirmed the male predominance seen in previous studies (byard and winskog ) . fig. . scene findings in game-playing asphyxia. this simple loop was present suspended from slats of an upper bunk. the child would lean against the loop as part of game playing; however, in this instance, unconsciousness prevented the simple action of lifting the head to remove neck constriction accidental deaths and injuries occur in all socioeconomic, ethnic, and racial groups. however, disparities exist in the occurrence, severity, and numbers of fatalities from pediatric injuries caused by both accidental and non accidental modalities. in the usa, african-american (brown ) , hispanic (mallonee ) , and native american children (goldcamp et al. a; berger et al. ) all have higher rates and severity of injury and death over caucasian children, even when adjusted for socioeconomic class. rural and frontier children are also at higher risk of injury and death during childhood as compared to their urban and suburban counterparts (cherry et al. ) . forensic pathologists play an important role in correctly identifying traumatic injury and risk factors within larger communities, as well as in groups within the population they serve through a complete medical legal death investigation. accidental injury in infancy and childhood is a common, and often preventable, cause of childhood morbidity and mortality. many of the types of injury are similar to adults; however, others have differences in incidence and character of injury that are unique to the stature and development of a child. it is imperative that forensic investigations of these deaths include information salient both to the preventability of future deaths and addressing the disparities that exist within gender, socioeconomic, and other class characteristics. injury remains the most common cause of childhood death in most parts of the world. prevention strategies rely in no small part on accurate assessment of the causes of death and injury in children. forensic pathology and forensic medicine can contribute significantly in efforts to improve child safety, health, and life. asphyxial games in children and adolescents update on "the choking game prevalence and current concepts of management of farmyard injuries prevention of traumatic brain injury in youth and adolescents the roles of age, gender, inhibitory control and parental supervision in children's pedestrian safety bunk versus conventional beds: a comparative assessment of fall injury risk injury and injury prevention among indigenous children and young people train injuries in children carbon monoxide poisoning pediatric equestrian injuries epidemiology of injury and the impact of health disparities drowning and near-drowning in children and adolescents: a succinct review for emergency physicians and nurses animal-related fatalities -part i: characteristic autopsy findings and variable causes of death associated with blunt and sharp trauma animal-related fatalities -part ii: characteristic autopsy findings and variable causes of death associated with envenomation, poisoning, anaphylaxis, asphyxiation, and sepsis post-mortem decapitation by domestic dogs: three case reports and review of the literature animals, autopsies and artifacts vitreous humor sodium levels in immersion deaths autoerotic death: incidence and age of victims -a population-based study nonfatal dog bite-related injuries treated in hospital emergency departments -united states centers for disease control and prevention. web-based injury statistics query and reporting system (wisqars) [online]. national center for injury prevention and control home and recreational safety; playground injuries: fact sheet workplace safety and health topics: agricultural safety injury prevention and control: motor vehicle safety. child passenger fact sheet likelihood of death of an infant or young child in a short fall of less than vertical feet deaths from falls in children: how far is fatal? pediatric pedestrian injuries: emergency care considerations limitations in fields of vision for simulated young farm tractor operators children's health in the rural environment stairway-related injuries in children infant walker-related injuries: a prospective study of severity and incidence abdominal aortic injuries associated with chance fractures in pediatric patients safety status of farm tractors that operate on public highways in four rural kentucky counties paediatric trauma associated with all-terrain vehicles buckled-up children: understanding the mechanism, injuries, management, and prevention of seat belt related injuries blunt splenic trauma in children: are we too careful? forensic approach of fatal dog attacks: a case report and literature review snowmobile injuries and fatalities in children review article: animal bites: an update for management with a focus on infections national highway traffic safety administration (nhtsa) infant injuries from child restraint safety seat misuse at british columbia children's hospital adapting adult scald safety standards to children the epidemiology of traumatic rupture of the thoracic aorta in children: a -year review physiological and toxicological aspects of smoke produced during the combustion of polymeric materials caustic ingestion in children socioeconomic differences in injury risks in childhood and adolescence: a nation-wide study of intentional and unintentional injuries in sweden insights into the ecology and evolutionary success of crocodilians revealed through bite-force and tooth-pressure experimentation socioeconomic differences in childhood injury: a population based epidemiologic study in ontario tap water scald burns in children tap water scald burns in children: a population-based study of all-terrain vehicle-related head and spinal injuries characteristics of home fire victims. quincy: national fire protection association adult-worn child carriers: a potential risk for injury horse-related injuries in pediatric patients nonfatal injuries to household youth on native american operated farms nonfatal injuries to household youth on racial minority-operated farms in the us pediatric injuries associated with the use of children's wagons in the united states delayed onset of intestinal obstruction in children after unrecognized seat belt injury pediatric sacral fractures injury surveillance for youth on farms in the changing farm injury trends by sex for youth living on us farms preventing pediatric pedestrian injuries children with special health care needs: patterns of safety restraint use, seating position, and risk of injury in motor vehicle crashes international life saving federation: drowning prevention. drowning facts and figures deaths and injuries from house fires horse-related injuries in children: a review child health corporation of america nursing falls study task force. pediatric falls: state of the science stairway injuries in children sledding injuries: a rationale for helmet usage open air carbon monoxide poisoning unsafe at any speed-kids riding all-terrain vehicles risk factors for unintentional injuries due to falls in children aged - years: a systematic review bicycle-related injuries in children: disturbing profile of a growing problem blunt pancreatic trauma in children engagement of children in agricultural work activities-scale and consequences of the phenomenon fatal acute intracranial injury, subdural hematoma, and retinal hemorrhages caused by stairway fall children in crashes: mechanisms of injury and restraint systems youth living on hispanic-operated farms: injuries and population estimates in the u self strangulation by hanging from cloth towel dispensers in canadian schools factors related to the presence of head injury in bicyclerelated pediatric trauma patients fatal and non-fatal machinerelated injuries suffered by children in alberta, canada, - examining ontario deaths due to all-terrain vehicles, and targets for prevention incidence and clinical significance of abdominal wall bruising in restrained children involved in motor vehicle crashes intestinal stricture following seat belt injury in children asphyxial games or "the choking game": a potentially fatal risk behaviour injuries among hispanics in the united states: implications for research agricultural work activities reported for children and youth on north american farms shaken infants die of neck trauma not of brain trauma green tobacco sickness in children and adolescents heat stress from enclosed vehicles: moderate ambient temperatures cause significant temperature rise in enclosed vehicles don't kiss a sleeping dog: the first assessment of "the blue dog" bite prevention program pediatric falls: is height a predictor of injury and outcome? ingestion of caustic alkali farm products paediatric blunt liver trauma in a dutch level trauma center effects of child age and body size on serious injury from passenger airbag presence in motor vehicle crashes the "choking game" and asphyxial games: epidemiological and clinical data the complexity of seat belt injuries including spinal injury in the pediatric population: a case report of a -year-old boy and the literature review the effect of bicycle helmet legislation on pediatric injury driveway crush injuries in young children: a highly lethal, devastating, and potentially preventable event structure fires, smoke production, and smoke alarms injuries in public and private playgrounds: the relative contribution of structural, equipment and human factors pediatric farm injuries involving non-working children injured by a farm work hazard: five priorities for primary prevention mechanisms of pediatric electrical injury. new implications for product safety and injury prevention fatal dog attacks in canada snowmobile injuries and deaths in children: a review of national injury data and state legislation fatal childhood vascular injuries associated with seat belt use characteristics of objects that cause choking in children laparoscopic repair of traumatic abdominal wall hernia from handlebar injury graduated driver licensing for reducing motor vehicle crashes among young drivers breeds of dogs involved in fatal human attacks in the united states between and air bag injuries -a literature review in consideration of demands in forensic autopsies autoerotic deaths in the literature from to : a review airbag-mediated pediatric atlanto-occipital dislocation falls from highchairs the role of child and adolescent development in the occurrence of agricultural injuries: an illustration using tractor-related injuries epidemiology and profile of pediatric burns in a large referral center social disparities in housing and related pediatric injury: a multilevel study success in the prevention of infant walker-related injuries: an analysis of national data autoerotic asphyxia: part i atypical autoerotic death: part ii bicycle injuries: a matter of mechanism and age the rockets' red glare, the bombs bursting in air: fireworks-related injuries to children accidental injuries in agriculture in the uk targeting pediatric pedestrian injury prevention efforts: teasing the information through spatial analysis railyard amputations in children pediatric short-distance household falls: biomechanics and associated injury severity the child as a projectile unintentional strangulation deaths from the "choking game" among youths aged - years -united states children with burn injuries -assessment of trauma, neglect, violence and abuse extensive and mutilating craniofacial trauma involving defleshing and decapitation: unusual features of fatal dog attacks in the young all-terrain vehicle rules and regulations: impact on pediatric mortality pediatric window falls: not just a problem for children in high rises application of electronic surveillance and global information system mapping to track the epidemiology of pediatric pedestrian injury analysis of nonsexual injuries of the male genitals in children and adolescents fatal envenomations -snakes and other creatures update. world health organization, health statistics and health information systems world health organization. children and road traffic injury. world health organization/unicef. department of violence and injury prevention and disability media center fact sheets: drowning incidence and cost of injury among youth in agricultural settings key: cord- - zstcag authors: kim, t.; lieberman, b.; luta, g.; pena, e. title: prediction regions for poisson and over-dispersed poisson regression models with applications to forecasting number of deaths during the covid- pandemic date: - - journal: nan doi: nan sha: doc_id: cord_uid: zstcag motivated by the current coronavirus disease (covid- ) pandemic, which is due to the sars-cov- virus, and the important problem of forecasting daily deaths and cumulative deaths, this paper examines the construction of prediction regions or intervals under the poisson regression model and for an over-dispersed poisson regression model. for the poisson regression model, several prediction regions are developed and their performance are compared through simulation studies. the methods are applied to the problem of forecasting daily and cumulative deaths in the united states (us) due to covid- . to examine their performance relative to what actually happened, daily deaths data until may th were used to forecast cumulative deaths by june st. it was observed that there is over-dispersion in the observed data relative to the poisson regression model. an over-dispersed poisson regression model is therefore proposed. this new model builds on frailty ideas in survival analysis and over-dispersion is quantified through an additional parameter. the poisson regression model is a hidden model in this over-dispersed poisson regression model and obtains as a limiting case when the over-dispersion parameter increases to infinity. a prediction region for the cumulative number of us deaths due to covid- by july th, given the data until july nd, is presented. finally, the paper discusses limitations of proposed procedures and mentions open research problems, as well as the dangers and pitfalls when forecasting on a long horizon, with focus on this pandemic where events, both foreseen and unforeseen, could have huge impacts on point predictions and prediction regions. the current coronavirus disease (covid- ) pandemic [ ] , caused by the sars-cov- virus, is providing statisticians, data scientists, machine learners, and other modelers a real-time laboratory to test and demonstrate their forecasting skills and abilities, with the quality of their forecasts assessable in a matter of days, weeks, or months. see, for instance, https://covid -projections.com from the masachussetts institute of technology (mit) and the institute of health metrics (ihme)'s https://covid .healthdata.org/united-states-of-america based at the university of washington in seattle, as well as [ ] discussing the complexities of modeling pandemics. of particular interests are the forecasting of the numbers of daily cases , deaths, and hospitalizations, or the cumulative cases, deaths, and hospitalizations attributable to covid- at a future date in a specified country or a locality (e.g., a county, state, or province) on the basis of currently observed cases, deaths, and hospitalizations data. such forecasts are of critical importance since they are major components in the decision-making process by government officials, business leaders, and educational and university administrators regarding the termination of lockdowns, lessening of social distancing and other mitigation regulations, opening of businesses, or continuing with online class formats in k- schools, colleges, and universities. the left panel of figure provides the daily number of reported deaths due to covid- for the united states (us) with respect to the number of days since december , until may , , which is day in the figures, as reported by the european center for disease control (ecdc) [ ] [see section a. ]. for a given date/day, including weekends, in the data set, the numbers reported are from the preceding day, which is due to a processing lag in reporting. the right panel of figure is a scatterplot of the cumulative number of deaths in the us due to covid- . given these daily and cumulative deaths data set, it is of interest to forecast the number of cumulative deaths in the us by, say, may , (corresponding to day ), which is memorial day, and to ask whether by that day the cumulative number of deaths in the us due to covid- will have surpassed the ominously depressing and grim milestone of , cumulative deaths. later, for our illustration, we will consider the problem of forecasting the cumulative number of deaths in the us due to covid- at the end of may , and compare our forecast with what eventually occurred. finally, we attempt to forecast the cumulative number of deaths by july th based on the data on july nd. such forecasting problems are clearly non-trivial since there is the distinct possibility that whatever model we had fitted in the observed time-frame may not apply to the time period under forecast, the ever-present danger and risk of extrapolation. aside from the fitted model most likely not being the true data generating model -recalling the aphorism attributed to george e. p. box [ ] that all models are wrong, but some are useful -there are other factors, some beyond our control, that could impact the number of reported deaths at a future time, such as premature easing of social distancing and re-opening of business establishments, virus mutations, better diagnostic tools, changing hotspots, overburdened health care facilities, introduction of effective treatments, beneficial or detrimental actions by local, state, and/or federal entities, changing definition deaths cumulative deaths due to covid- , under-or over-reporting of deaths, timely development of a vaccine, protests and riots arising from social unrests, and others. but high-level decision-makers such as government officials, business leaders, educational administrators, and society itself, demand some beacon, however dim such beacons may be, to guide them in their decision-making. statisticians, data scientists, machine learners, and other modelers are always ready and willing to provide such beacons. this paper is in this spirit. we will examine existing methods and develop new methods for constructing prediction regions for random variables that pertain to the number of occurrences of an event of interest. a prediction region contains more information compared to just a point prediction since it provides information about the uncertainty inherent in the prediction. note that with a prediction region we are interested in the would-be realized value of a random variable, not the value of a parameter, hence instead of referring to it as a confidence region, it is instead called a prediction region. the events of particular interest are those that are 'rare' in the sense that, informally, the probability of an event occurring in an infinitesimal interval is also infinitesimal. consequently, our starting point will be the poisson distribution which is a model for the number of occurrences of a rare event, and transition to the more general poisson regression model, and eventually to an over-dispersed poisson regression model which turns out to be a better model in the covid- application. the real-life and practical application for which our methods will be applied is the construction of prediction regions for the daily and cumulative number of deaths due to covid- in the us for a future date given only the daily deaths data until a current date. note that such predictions or forecasts could probably be improved by utilizing other information (such as the capacities of health care facilities; movements of people in a region; information about sensitivity and specificity of diagnostic tests; transmission rates (r ) of the virus; and others), or via stratification by states, cities, or counties and then combining the results from these strata to obtain a point prediction and a prediction region for the whole us. however, we approach the construction of the prediction regions for the daily and cumulative deaths at a future date by just utilizing the observed reported daily deaths data for the whole us, which in a sense is the most reliable available data regarding this covid- pandemic. it might be possible to utilize information about the number of cases or infected individuals, which is also reported daily, but we feel that this is not a reliable information since it is highly dependent on the number of tests that are performed and on the sensitivity and specificity of the diagnostic tests used. in addition, if such information is to be used in the prediction model, then we may not have their realized values at the future date on which the prediction region is desired. we point out that even though we are employing probabilistic models in the form of the poisson or an over-dispersed poisson model, which are derivable from intuitive conditions when dealing with rare events (cf., [ , ] ), our prediction method is still purely data-driven being only reliant on the observed data. the occurrence of a death due to covid- could still be considered as a rare event when viewed in the context of the whole population, though even if it is rare, deaths are still significant and dire events. this is because to die of covid- , generally one first needs to get infected, which at this point is still a rare event, and then having been infected, to die from it. the rate of dying when infected with covid- , if not age-adjusted, is still rather low, less than % (see, for instance, coronavirus (covid- ) mortality rate). because of its rarity, a plausible probability model for the number of deaths due to covid- is therefore the poisson model whose probability mass function (pmf) is given by with z ,+ = { , , , . . .}, and λ > is the rate parameter, which is also the mean and variance of the distribution, and i{·} is the indicator function. for a variable y with this poisson distribution, we write y ∼ p oi(λ). the cumulative distribution function of a p oi(λ) is we start our investigations with this no-covariate poisson model, equivalently, a model with intercept only, since results for the poisson regression model build on this no-covariate model. suppose now that y ∼ p oi(λ), where for the moment we assume that we know the rate parameter note that this region will not be an interval being a subset of z ,+ , though if this region is formed as the intersection between z ,+ and an interval in , then we may call it imprecisely as an interval. subject to this condition, a desirable property of such a region is that its cardinality is as small as possible. if we allow for Γ(λ, α) to depend on a randomizer u , a standard uniform random variable independent of y , the smallest cardinality ( − α)% prediction region is, using a neyman-pearson lemma type argument, given by where, for d ∈ , we define the subsets of z ,+ given by and c(α) and γ(α) determined via with / = . observe that by allowing randomized prediction regions, we have if we do not admit randomized prediction regions, which is achieved by always taking u = in Γ (u ; λ, α), then unless − α is a 'natural' prediction coefficient, we will not achieve equality in the preceding probability statement. the use of the adjective 'natural' is analogous to its use in constructing nonparametric confidence intervals, cf., [ ] . see [ ] on the application of neyman-pearson-type arguments to construct optimal confidence regions, which could be adapted to the construction of prediction regions. there are two other ways of constructing prediction intervals for y when λ is large using normal approximations. to obtain the prediction regions, these intervals are then intersected with z ,+ . letting n (µ, σ ) denote a normal distribution with mean µ and variance σ , we recall that when λ is large owing to the central limit theorem and the delta method (cf., [ ] ), then we have the normal approximations let φ(·) and Φ(·) be the probability density and cumulative distribution functions of a standard normal random variable so that let z α = Φ − ( − α) be its ( − α)th quantile. two approximate prediction regions for y when λ is large, which are based on the above normal approximations, are given by when λ is large, as noted in the construction of Γ , we may approximate the poisson probabilities by normal probabilities, via as such, we obtain the approximation consequently, when λ is large, the regions Γ (u ; λ, α) and Γ (λ, α) should be close to each other. for these prediction regions Γ r (randomized), Γ n (nonrandomized), Γ , and Γ , the exact coverage probabilities and their exact lengths (mean length for Γ r ) could be computed under the p oi(λ) distribution, since λ is known. note that the lengths, which are the differences between the upper and lower integer limits of the prediction regions, are equivalent surrogates of the cardinalities of the regions. figure depicts the exact coverage probabilities (cp), expressed in percentages, and their lengths (expected length for Γ r ), for different values of λ. except when λ takes small values where the coverage probabilities of Γ and Γ are degraded, especially for the latter, the performance of these prediction regions are quite similar. the coverage probability of Γ r is exactly equal to − α, whereas that for Γ n is always at least equal to − α. both Γ and Γ could have coverage probabilities that could be below the nominal coverage level, though as λ increases, these differences become negligible. by construction, Γ r has a shorter interval than Γ n ; for some values of λ, the length of Γ r exceeds that of Γ and Γ , but this is because the coverage probabilities of Γ and Γ are lower than the nominal coverage level. but, in the preceding developments, we have assumed that the rate parameter λ is known, an unrealistic assumption. how do we deal with the situation when λ is unknown? suppose that we had observed a realization y = (y , y , . . . , y n ) of a random sample y = (y , y , . . . , y n ) from p oi(λ), so the components of y are independent and identically distributed (iid) from p oi(λ). our goal is to utilize y to construct a ( − α)% prediction region for an unobserved y , which is independent of y and whose distribution is also p oi(λ). how will we achieve our goal? note that through the sufficiency principle, we may reduce the problem by simply assuming that we had observed t = n i= y i , the realization of the sufficient statistic for λ given by t = n i= y i , which has a p oi(nλ) distribution. the reduced problem therefore is that we have (t, y ) which are independent random variables with t ∼ p oi(nλ) and y ∼ p oi(λ) and our goal is to construct a ( − α)% prediction regionΓ(t, u ; α) for y , which utilizes t , and possibly a randomizer u which is independent of (t, y ). given t = t, the maximum likelihood estimate (mle) of λ isλ(t) = t/n. by virtue of the consistency ofλ(t ) for λ as n → ∞, a seemingly straight-forward approach to constructing a prediction region for y is to replace λ in Γ (u ; λ, α), Γ (λ, α), and Γ (λ, α) in ( ), ( ) , and ( ), respectively, byλ(t) to obtainΓ (t, u ; α) = Γ (u ;λ(t ), α); how do these prediction regions compare with each other in terms of performance, both in the context of their coverage probabilities and also their cardinalities, whose surrogate are lengths? in particular, by substitutingλ(t ) = t /n for λ, how does this impact the coverage probabilities of these prediction regions and are they still valid, even in an asymptotic sense? it is not clear how the substitution of λ byλ(t ) = t /n will impact the exact performance of the first prediction regionΓ . however, for the second and third prediction regionsΓ andΓ , we could alter them to take into account the substitutions, provided that λ is large. as noted earlier, when λ is large,Γ ≈Γ , so the alteration ofΓ should also apply, approximately, toΓ . the change in distributions of the pivotal quantities arising from these substitutions are reflected below, a consequence of the delta-method. from these normal approximations, we could improve the prediction intervalsΓ andΓ into the following prediction intervals, which take into account the impact of these substitutions where, for notational economy, we writeλ forλ(t ) and '∨' for max; for the ceiling function; and for the floor function: note that by intersecting the intervals with z ,+ , the floor and ceiling functions are actually not needed, but we retain them in the formula since when we consider the 'length', this pertains to the length of the interval. observe that if the lower limits of these intervals are not zeros, which will usually be the case for large λ, then it is a simple exercise to show that these two prediction intervals have the same lengths, but they are not identical regions. in trying to adapt the prediction region Γ (u ; λ, α) in ( ) to the situation where λ is unknown, the main idea is to replace λ by an estimate obtained from the observed data. doing so leads to estimates of p(k|λ), k ∈ z ,+ which are then used in determining c(α) and γ(α) in ( ). thus,Γ in ( ) is obtained by using the ml estimates of {p(k|λ), k = , , . . .} given by {p(k|λ(n, t)), k = , , . . .} withλ(n, t) = t/n. this begs the question on whether other possible estimates of {p(k|λ), k = , , . . .} could be utilized which may have better performances than the use of the ml estimates. an approach based on a second-order taylor expansion adjusts p(k|λ(n, t )) and leads to the approximation p(k|λ) ≈p (k; (n, t)) ≡ p(k|λ(n, t)) by usingp (k; (n, t)) in place of p(k|λ) in ( ) results in the prediction region denoted byΓ (n, t; α). another intriguing possibility is to utilize the uniformly minimum variance unbiased estimator (umvue) (see [ ] ) of p(k|λ), given the data (n, t ) with t ∼ p oi(nλ), or equivalently, y , y , . . . , y n which are iid p oi(λ). the umvue of p(k|λ), usually obtained through the rao-blackwell theorem and lehmann-scheffe theorem [ ] , is given bŷ the binomial probability at k with parameters (t, /n). observe, however, that this approximation will lead to zero probabilities for k outside of the set { , , . . . , t}. usingp (k; (n, t)) in lieu of p(k|λ) in ( ) leads to the prediction region denoted byΓ (n, t; α). as yet another idea is to develop a procedure by borrowing from the bayesian playbook [ ] . we suppose that our prior knowledge of the value of the poisson rate λ is represented by a distribution function g. having observed y = y = (y , y , . . . , y n ), the posterior distribution of λ is given by with t = t(y) = n i= y i . the conditional probability mass function of y , given y = y, also called the posterior predictive pmf, is if we are pure bayesians, then we will completely know, or trust, our g, so we could use the predictive pmf p(·|y; g) in lieu of the poisson pmf in ( ) to form a bayesian prediction region for y . usually, however, we may try to estimate g by aĜ(·; y) based on y. this brings us to the realm of the empirical bayes (eb) approach, pioneered by herbert robbins; see [ , , ]. an extreme case is to 'estimate' g by a degenerate distribution at the ml estimateλ = t/n, which leads to just substitutingλ in the poisson pmf, hence results in the prediction regionΓ in ( ) . another possibility is to try to estimate g non-parametrically. however, here we implement this bayesian and eb approaches using a family of conjugate priors, so we assume g is a gamma distribution with mean κ/β and variance κ/β , denoted by g κ,β , whose density function is where κ > and β > . note that κ is the shape parameter and β is the scale parameter. under g = g κ,β , simplifying ( ) we obtain, for y ∈ z ,+ , when κ is a positive integer, the pmf in ( ) corresponds to a negative binomial distribution with parameters κ + t and (β + n)/(β + n + ). the pmf in ( ) could be used in place of the poisson pmf in ( ) to form a bayesian prediction region for y , given (κ, β), denoted byΓ (u, y; (κ, β)). an approach to specifying (κ, β) is to specify a prior mean and prior standard deviation for λ, say m and s, respectively, which yield κ = m /s and β = m/s . the eb approach estimates κ and β from the data y = (y , y , . . . , y n ). such estimation could be done via maximum likelihood using the likelihood function obtained from the joint marginal distribution of (y , y , . . . , y n ) based on the model y i |λ i ∼ p oi(λ i ) and λ i ∼ g κ,β . this likelihood function is given by a method-of-moments approach to estimating (κ, β) based on y fails, however, because negative estimates of κ and β are obtained when the sample variance of y is smaller than its sample mean. at this point we mention previous works dealing with prediction intervals under the poisson model. prediction interval methods for the poisson model have been incorporated in the r package envstats [ ] . an object function in this package is predintpois dealing with construction of prediction intervals under the poisson model. it provides four options for the type of prediction interval to construct. the methods are based on procedures presented in [ , , ] . the option normal.approx in predintpois coincides with the prediction regionΓ based on the normal approximation. in these earlier procedures, randomization was not utilized, hence generally conservative prediction intervals are obtained. other approaches for prediction interval construction under the poisson model, including poisson regression models, are based on bootstrapping and simulation techniques, hence are computationally-intensive [ ] . to compare performance of the prediction regionsΓ (randomized version),Γ ,Γ ,Γ ,Γ , anď Γ with m = , s = , and for n ∈ { , , , , , , , } and λ ∈ { , , , , , , }, we performed simulation studies, with program codes in the r [ ] environment, to determine the coverage probabilities and the lengths of the regions (recall that length is an equivalent surrogate for the cardinality of the regions since we took the ceiling and the floor of the lower and upper limits, respectively, for the intervals that leads toΓ andΓ ). for each combination of n and λ, simulation replications of the basic simulation experiment were performed. table in the appendix presents the results on the coverage percentages, mean lengths of the prediction intervals, and standard deviations of the lengths for different values of λ. the basic simulation experiment is, for a fixed n and λ, to generate t ∼ p oi(nλ) and y ∼ p oi(λ). the t variable could be viewed as . . , y n are iid from a p oi(λ), the prediction regions are then constructed based on the observed (n, t ), with prediction coefficient of %. note that sincě Γ is the bayes prediction region instead of the eb, we only needed the value of t = n i= y i , but if we also used the eb approach, then we would have needed the values of (y , y , . . . , y n ) to estimate (κ, β). after constructing the prediction regions, it is then determined if y is contained in these regions. coverage percentage is the percentage out of the prediction regions that contain the y ; mean (standard deviation) length is the average (standard deviation) of the lengths of the prediction intervals. figure presents plots with respect to n of the coverage probabilities (cp) and mean lengths (ml) for λ ∈ { , , , }. examining table and figure we observe that when λ = , the cp ofΓ is very poor and even deteriorates as n increases. the reason for this is that the realized y tends to equal , but the square-root transformation has a tendency to shift to the right the prediction interval, hence the interval tends to miss y . this result forΓ is consistent with the result when the rate λ is known. when n = ,Γ andΓ have unacceptably lower cps compared to the nominal level, whileΓ also has cps which are below the nominal level, as well asΓ andΓ , though the last two regions have cps closer to the desired level. the length ofΓ tends to be shorter thanΓ andΓ . as n increases, the cps ofΓ andΓ get closer to the desired level, and their lengths tend to be a bit shorter thanΓ andΓ . when λ = , the cps ofΓ ,Γ ,Γ , andΓ are all below the nominal level, whereas forΓ andΓ , their cps exceed or are quite close to the nominal level, except when n = . as a consequence, they ended up having longer mean lengths. these behaviors continue to hold as λ was increased, but with the cps getting closer to the nominal level, especially as n increases. when n is small, the cps ofΓ andΓ are still appreciably lower than the nominal level. when λ is large,Γ ,Γ , andΓ almost have the same performance. summing up our observations from these simulation studies for this no-covariate or intercept only poisson model, in terms of adapting to the estimation of the unknown rate λ,Γ andΓ possess the best performance among these six prediction regions in terms of achieving the nominal level, but they also tend to be longer than the others. negative continuously-differentiable function ρ(·), called the inverse link function, such that this is the so-called poisson regression model and belongs to the class of generalized linear models or the class of non-linear models [ , ] . if y , given x = x , has a poisson distribution with rate λ(x ; θ), and θ is known, then we could construct prediction regions for y according to the methods described in the first part of section when λ was assumed known. when θ is not known, then there is a need to estimate it. let us therefore assume that we are able to observe the sample ) and with the y i s independent and the x i s fixed. we seek to construct a prediction region for y associated with the covariate vector x . first, we introduce the following functions: the log-likelihood function for θ, given {(y i , x i ), i = , , . . . , n}, is given by the associated score vector function is whereas, the observed fisher information matrix is, with x ⊗ = x t x, thus, the expected fisher information matrix is the mle of θ based on {(y i , x i ), i = , , . . . , n}, denoted byθ, solves the equation this will usually be obtained through iterative procedures, such as the iterative newton-raphson method, with the iteration given by by the large-sample theory of ml estimation (cf., [ , ] ), as n → ∞ and under regularity conditions on the sequence of covariate vectors x i , i = , , . . . , n, we have that a consistent estimator of i(θ) is i(θ). by the delta-method, it then follows that the ml estimator of λ(x ; θ) satisfies, as n → ∞, where 'tr' means trace of a matrix. using this result and when λ = ρ(x θ) is large, we obtain the approximate distributions of relevant pivotal quantities for constructing prediction regions. we writeλ for λ(x ;θ) andψ for ψ(x θ ). these pivotal quantities are: . . , n}, from these pivotal quantities, we are then able to obtain approximate prediction regions for y given by: we could also have the prediction region based on Γ from section given bỹ where we note that the dependence on x and (y, x) is throughλ . note, however, that we are simply plugging in the estimate of λ(x ; θ), but without taking into consideration the variability inherent in the estimator λ(x ;θ). a specific inverse link function ρ(·), which we will consider in the application to forecasting deaths in the us due to covid- , is the exponential function ρ(w) = exp(w), so that for this special inverse link function, we obtain the simplifications for the score vector and information matrices functions given by we also mention the extension of the bayesian/eb approaches to constructing prediction regions in the regression setting. we suppose that the parameter θ in λ(x; θ) = ρ(xθ) takes values in a parameter space Θ. the approach then proceeds by starting with a prior distribution Π(·) on Θ which quantifies our prior knowledge about θ. the posterior predictive distribution of y , the response at x , given the data (y, x) = {(y i , x i ), i = , , . . . , n}, is given by where where the product and the sum are taken over the index set associated with (y, x), so will be over { , , . . . , n} for (y, x), and { , , , . . . , n} for (y , x ). generally, there will be no family of conjugate prior distributions on Θ with respect to the poisson regression model, so the function h will not be in a closed analytical form, so that it has to be computed numerically, for instance, using markov chain monte carlo (mcmc) algorithms. nevertheless, upon obtaining the posterior predictive distribution of y given in ( ), a prediction region is then obtained by using this pmf p(y |x , (y, x)) in lieu of the poisson pmf in ( ), analogously to the development of the prediction regionΓ in the intercept only model. the prior distribution Π will involve hyper-parameters, for example, if Θ = p+ , Π could be specified to be a multivariate normal distribution with mean vector µ and covariance matrix Σ, so (µ, Σ) will be the hyper-parameters. for the bayesian, these hyper-parameters will be assigned values, unless an improper prior distribution (e.g., lebesgue measure), which need not involve unknown hyper-parameters, is adopted; whereas, for the empirical bayesian, these hyper-parameters will be estimated using the data (y, x). because of the need to approximate the posterior predictive pmf through numerical methods, these bayesian and eb approaches to constructing a prediction region for y are clearly computationally-intensive, especially if used in a simulation study to investigate their properties, such as their coverage probabilities and their lengths. because of the need to specify a non-conjugate prior and its hyper-parameters and the need for intensive computations, these bayesian and eb procedures are not included in the illustrations, simulations, and applications. it is clear, though, that they are highly viable alternative procedures and should be further explored. we demonstrate these prediction regions, depicted as intervals in the plots, via the following experiment. we specify a sample size n and an order p. we then generate iid realizations w i , i = , , . . . , n, n + , from either a n (µ, σ ) distribution or a standard uniform distribution, and form the covariate vectors x i = ( , w i , w i , . . . , w p i ), i = , , . . . , n, n + . for a specified θ = (θ , θ , . . . , θ p ) t , the poisson rates are computed. the ith response y i is a realization of a random draw from a p oi(λ i ). the response vector is y = (y , y , . . . , y n , y n+ ) t . the goal is to construct a prediction region for ( ), ( ) , and ( ), respectively. the procedures were coded into r functions and these will be made available publicly in due time. we present the results pictorially via a scatterplot of {(w i , y i ), i = , , . . . , n, n + }. the realized value y ≡ y n+ of y is highlighted and the three prediction intervals for y are also plotted. included in the plot is the theoretical curve for the λ(x; θ) as a function of w and we also super-impose the fitted curve. prediction regions for y i at w i , i = , , . . . , n, are also depicted in the plot. , which all contained y . two things to observe from this plot are ( ) the prediction regionΓ was shifted to the right relative toΓ , and ( ) the prediction regionsΓ with respect to the w-values are scissorlike or jagged. the latter is a consequence of the randomization approach in the construction of the prediction regions and this non-smooth behavior becomes more apparent since the realized values of the y i s are small. the third realization, depicted in figure , is from a model with n = , p = , with w i ∼ n (µ = , σ = ), and with θ = ( , . , −. , −. ). the rate curve λ(w) as a function of w goes to zero as w increases, but goes to ∞ as w decreases, with a local minimum and maximum close to w = − and w = , respectively. the target of the prediction regions was in each of these illustrative realizations, the three prediction regions did not vary much from each other in terms of their sizes, except in the first case whereΓ was shorter but barely covered gamma target the value being predicted. a question that now arises is how their coverage probabilities and their mean lengths compare with each other? to gain some insights into these comparisons, we performed simulation studies under the four different models described above, with each simulation run having replications. the sample sizes considered were n ∈ { , , , }. in the appendix, tables , , and summarize the results of these simulations where we report the coverage probabilities (cp), mean lengths (ml), and standard deviation of lengths (sl). examining these tables, it appears thatΓ has coverage probabilities that are below the nominal level (between % and % below in table when n = ), with the discrepancy being more pronounced when the sample size is small. as the sample size is increased, these observed coverage probabilities get closer to the nominal level. this deficiency is due to the estimation of the θ parameter and, as previously noted, theΓ does not take into consideration the variability in the resulting estimator of λ(x ; θ). on the other hand,Γ andΓ both achieve coverage probabilities that are quite close to the nominal level, especiallyΓ , when n is large.Γ , on the other hand, tends to have a lower mean length compared to the mean lengths ofΓ andΓ , with the differences in mean lengths becoming alarmingly large for the model in table . recall that for this model, the rate curve increases to ∞ as w decreases to −∞, and since the w i 's are generated from a normal distribution, on some occasions, w n+ falls outside the range of {w i , i = , , . . . , n}. depending on how different w n+ is from the mean of w , w , . . . , w n , this could lead to a large estimate of the standard error ofλ n+ , gamma target thus leading to very wide prediction regions forΓ andΓ . sinceΓ simply utilized the estimate ofλ n+ , but was totally oblivious to its variability, it was not much affected in such a situation. however, because of its rigidness with respect to this added variability, it could dramatically suffer. we demonstrate this situation by plotting an extreme realization in figure this particular demonstration warns us of the danger and pitfalls of making a prediction for a response variable that is associated with a covariate vector outside the convex hull of the covariate vectors used in the construction of the prediction regions and when the poisson rate hyper-surface generated by the map x → ρ(xθ) is complex. as a word of caution, when performing extrapolation to do predictions, be forewarned of sinkholes littering the forecasting road -and, if it could be avoided, make no forecasts on long, especially very long, horizons. but, alas, this is the type of forecasting problem that is actually realistic and of most interest, such as that of predicting the number of cases or deaths due to covid- in a future date, given the observed data up to a gamma target certain date. based on the results of these simulation studies, the prediction regionΓ appears to be the most preferable among the three prediction regions. in our illustration using the covid- data set in section , we will therefore just present the prediction region provided byΓ . we now present in this section an illustration of the potential application of the procedures discussed in the preceding sections. one of the interesting questions during this covid- pandemic is the forecasting of the number of cumulative deaths in the us at a given date, for example, at the end of may , , given information up to a certain date, say may , . such forecasts are of critical importance since they could partly be the basis of highly consequential and possibly controversial decisions by federal, state, and local governments officials, school administrators, executives of big corporations and small businesses, religious leaders, and many others. such decisions could pertain to when to institute stay-in-place directives, when to issue social distancing or social easing guidelines, when to open business establishments, when to open public places such as shopping malls and ocean beaches, when to allow religious gatherings, etc. data for daily deaths and cumulative [ ] . clearly, the sequence of cumulative deaths does not satisfy the independence assumption, so a non-homogeneous poisson process model [ ] is not an appropriate model for cumulative deaths when viewed as a continuous-time stochastic process. however, a non-homogeneous poisson process could plausibly model the occurrences of deaths in continuous-time, from which it follows that the sequence of daily deaths will be independently poisson distributed with possibly different rates depending on the number of days from the time origin and the specific day of the week, as well as other features such as, for example, the quality of the health care facilities, which is hard to quantify and not available in the european cdc data set. our novel idea therefore is to utilize poisson regression to predict the number of daily deaths according to the methods developed earlier, and then to aggregate these daily forecasts to obtain forecasts of the cumulative deaths. we will use the data set for the us provided by the european cdc ( [ ] ) plotted in figure which are the observed numbers of daily deaths attributed to covid- starting on march , , the day after the first reported death due to covid- , until may , . note that, technically, this will be the deaths data at the end of may , . using this data set gamma target on may th, and given the cumulative number of deaths until then, the goal is to forecast the cumulative number of deaths in the us due to covid- by the end of may , , that is, june , . we limit our illustration to simply utilizing the variable daynum, which is the number of days starting from december , , day which the day of the week, and deaths, the variable representing the daily number of deaths. we surmise that the deaths data set is the most reliable among the data sets that were compiled, compared, for instance, to the data set pertaining to the number of cases or infected people. however, the deaths data set need not also be totally reliable and could be subject to misclassification error and competing causes of deaths., for example, a patient who contracted covid- who dies primarily because of pneumonia may be classified as having died of covid- , but could also be classified as having died, not of covid- , but of pneumonia. see also, for instance, the wsj article [ ] , [ ] , and the bbc news article https://www.bbc.com/news/world- [ ] , the last two discussing the notion of "excess deaths," which are deaths that may have been due to the pandemic, but which are not included in the reported covid- deaths data set. certainly, we could have used other information such as the number of reported cases; by performing separate forecasts in each of the states and the district of columbia, then aggregating; or even by utilizing counties or metropolitan cities as strata, and then combining forecasts from these strata to obtain an overall forecast for the whole us. however, for illustrative purposes, we decided to keep things simple. = ) , consisting of days, we have the daily number of deaths, hence also the cumulative number of deaths. the other variable used in the modeling is day (e.g., sunday, monday, etc.) associated with each value of daynum, which is a categorical or factor variable. we fitted a poisson regression model using the glm function in r with a log-link, with response being y = deaths and covariate vector x = ( , w, w , w , w , w , day), where w = daynum, and day is considered as a factor variable hence is converted into six, instead of seven since we already have an intercept term, dummy variables in the design matrix. we chose this th-order model with respect to daynum since the akaike information criterion (aic) values, computed under the poisson regression model, appear to stabilize starting at this model and adhering to the law of parsimony (occam's razor). the aic values associated with the thand th-order models were actually smaller than for the th-order model; however, these models possess highly unstable predicted values. table summarizes the aic values for the different models, both without and with day in the model. it also contains the estimates of ξ, the overdispersion parameter in a model that will be introduced shortly, and as we will then see, larger values of ξ are indicative of the poisson regression model becoming a more adequate model. the histogram and time plot of the residuals with respect to daynum are provided in figure . recall that the ith residual in poisson regression is defined as whereλ i is the fitted value associated with x i . as such, if the poisson regression model is adequate, one should see a histogram similar to that associated with a centered (at zero) poisson distribution with unit rate, but this will just be an approximation since the rates are estimated, hence that affects the distribution of the residual. similarly, the time plot of the residuals should be randomly distributed on the zero horizontal line. the histogram and time plot in figure , with the time thus, the reported number on april th of , the highest number of daily deaths reported, is an outlier explained by the adjustment made. however, we still included these perceived outliers in the fitting of the fifth-order, with respect to daynum, poisson regression model. later, when we consider forecasting for july th and august st, and since another significant adjustment was made on june th, we will re-allocate each of the adjustments proportionately to the observed deaths on the days on or before the adjustment day. based on the fitted model's residuals, we further assessed the independence assumption of the daily deaths. we do this by creating a contingency table for daynum with six intervals and with residuals being either negative or positive and then performing a test for independence. the observed contingency table is presented in table . a chi-square test for independence based on this table yielded χ c = . on degrees-of-freedom, with associated p-value of . , hence the null hypothesis of independence cannot be rejected. observe, however, that the fit of the model in the early days is not satisfactory, and between daynum to , there was a preponderance of negative residuals, possibly owing to the influence of the adjusted reported daily deaths on daynum . it may appear surprising and counter intuitive to include a day effect in the model, since one recall that our main objective is to obtain a prediction region for the cumulative number of deaths at a specified date, in our case june , (daynum = ). this means we were predicting the cumulative deaths at the end of may , . from daynum = to daynum = , there were a total of days. if we denote by y j , j = , , . . . , , the random variable denoting the daily number of deaths for daynum = j, the random variable denoting the cumulative number of deaths until daynum = k is s k = k j= y k . thus, we are seeking a prediction region for s , given that s = . under the fitted poisson regression model, y j , given with w j = daynum j and d kj , k = , , . . . , , the dummy variables representing whether day is a tuesday, a wednesday, a thursday, a friday, a saturday, or a sunday, respectively, has poisson distribution with rate λ(x j ; θ) = exp {x j θ} . we mention that in our r code for fitting this model, for computational stability, we first centered and standardized the non-constant columns of the x j 's. let α * ∈ ( , ), and the y j s being independent. from the preceding section we know how to construct a ( − α * )% prediction interval [a j , b j ] for y j , where, as mentioned earlier, we will simply utilize the prediction regionΓ . by the independence, we will have thus, if we wanted s +[a • , b • ] to be a ( −α)% prediction interval for s , given s , we could choose α * = − ( − α) / . this procedure will guarantee a conservative ( − α)% prediction interval for s , given s . this is the approach we followed in constructing a (conservative) % prediction interval for s , the cumulative number of deaths due to covid- in the us by the end of may , . we implemented the above procedure and also constructed the prediction intervals at each of the observed daynum which are depicted in figure . examining this figure note that there are more observed deaths outside the prediction curves than what is expected nominally. this indicates that either there is more variability inherent in the stochastic mechanism generating the observed number of daily deaths relative to a purely poisson regression model, or the fifth-order poisson rate model is still inadequate, or both. we propose an approach that introduces over-dispersion with the poisson regression model serving as a hidden model. we mention that our occam's razor-type solution is motivated by frailty modeling in survival analysis (see, for instance, [ ] ). our model assumes the existence of an unobserved positive latent variable z j of mean at w j = daynum j , and the reported number of deaths y j is the integer part of z j y * j , with {y * j } arising from a poisson regression model and with z j and y * j independent. recall that frailty models in survival analysis are used specifically to model correlations among observations; whereas, in our model it serves as an unobserved random contamination component in the observed number of daily deaths. in our implementation, we shall take z j to have a gamma distribution with mean one and variance /ξ (see [ ] ). this ξ is then an additional parameter in the regression model aside from the parameter vector θ. such a model leads to an over-dispersed poisson regression model, with the purely poisson regression model embedded in this model and obtainable as a limiting case when ξ → ∞. inference for such a model requires further study, with possible use of an em-type algorithm, though this could be difficult to implement since the distributions of the y i s are not in closed forms. however, we may implement a z-estimation approach (see, for instance, [ ] ). we first note that the approximate higher-order moments of the y i s are also obtainable. based on these moments, we could form the set of estimating equations, where we recall that ρ(w) = exp(w): by z-estimation theory ( [ ] ) and under regularity conditions, it will follow that, for some (p + ) × (p + ) matrix in fact, let us introduce the (p + ) × vector functions: denote by h the (p + ) × (p + ) matrix function consisting of the derivatives of u with respect to (θ, ξ). the components of this matrix function are: x ⊗ i ρ(x i θ) and h ((y, x); θ, ξ) = ; we could then obtain the estimates via the newton-raphson (nr) method with iteration step it turns out that a simpler way to obtain the estimates of θ and ξ is to first obtain the estimatẽ θ of θ from the first estimating equation. this could be done by using the glm object function in r [ ] with the poisson family and logarithm link. the estimateξ of ξ is then obtained from the second estimating equation using a one-variable nr iteration with θ replaced byθ. define the (p + ) × (p + ) matrices Σ and Ω according to with plim denoting "in-probability limit" as n → ∞. then, the asymptotic covariance matrix of (θ t ,ξ) t is we note that another way of obtaining estimates of Σ and Ω is to obtain their theoretical expressions using higher-order moments of the y j s. these expressions depend on θ and ξ, so estimates could then be obtained by replacing θ and ξ byθ andξ. we also point out that Ξ is generally not equal to Σ − when ξ is finite. in fact, it is imperative that Ξ should be used instead of Σ − since it takes into account the impact of the estimation of ξ. applying the delta-method [ ] , we then have the pivotal quantity result, as n → ∞, given by where quantities with 'ˆ' are estimates obtained by plugging inθ andξ for θ and ξ in their respective expressions. from this pivotal quantity, it follows that an approximate ( − α)% prediction interval for y is given by because of the term /ξ, when ξ is small, this prediction interval will be wider than the prediction intervals under the purely poisson regression model. as ξ → ∞, which makes the model approach the poisson regression model, then this prediction interval will approachΓ . implementing this procedure based on this over-dispersed poisson regression model, we first examine the prediction intervals at each of the observed daynum values, which are shown in figure . we now see that these approximate % prediction intervals cover most of the observed daily deaths. this indicates that the over-dispersed poisson regression model provides a better fit to the observed daily deaths data than the purely poisson regression model whose approximate % prediction intervals are shown in figure . the estimate of ξ turned out to beξ = . . the daily deaths left-panel of figure is the scatterplot of daily deaths, but now including the actual observed values after daynum and until (these are the red dots) and the prediction intervals for the daily deaths past . observe that the prediction intervals for daynum between and are wider than those in figure and the reason for this is the prediction coefficient used is now adjusted for the goal of constructing a prediction interval for s . in this case, α * = . . the right-panel of figure displays the prediction interval for s , given s ; in fact, this also displays the prediction intervals for s k , k = , , . . . , , given s . the red dots are the actual observed values past daynum equal to . the predicted cumulative deaths on daynum = wasŜ = and the conservative prediction interval for s was [ , ] . in both of these plots, notice that the th-order prediction model did not perform well past daynum = , though the prediction interval for s did cover what was actually observed, which was . from figure , an elevated number of daily deaths occurred starting at daynum = (may ). the grim milestone of cumulative deaths due to covid- in the us was also surpassed on this day. to partly assess the sensitivity of the procedure, if we had used the data until daynum = (may rd), the predicted value for s isŜ = and the conservative % prediction interval, given s , is [ , ]. the associated plots in this case are provided in figure . on the other hand, if we had used the data until june , (daynum = ), the fitted value is , which coincided with the observed cumulative number of deaths. that the fitted cumulative number of deaths and the observed cumulative number of deaths on the last day were equal is actually a consequence of the estimating equation, hence in hindsight is not a surprising result. the associated plots in this case are provided in figure . observe in the right-panel of figure that the model for the cumulative deaths based on this th-order model is quite excellent, lending support to our novel approach of modeling the daily deaths data, instead of the cumulative deaths data, for the purpose of making predictions for the cumulative deaths. we summarize the results of this sensitivity analysis in table and figure , where we report the predicted values and the prediction intervals for s under scenarios where data used in the model fitting is up to the different days from daynum equal to up to . based on this analysis, the fifth-order model appears to possess stability since the predictions and the prediction intervals for s remain somewhat consistent as the amount of data being used in the model fitting varies. as to be expected, note that as the forecasting horizon shortens, then the prediction interval also narrows. this is the case, even though we still considered the outlier on april th, which was the adjusted deaths data point, as a legitimate observation. however, any model, especially higher-order models, will have a breakdown point in the sense of yielding seemingly unreasonable predictions, perhaps due to a long forecasting horizon, insufficient amount of data, or wildly changing data points drastically altering estimates which highly impact forecasts. for this th-order model, it appears to break down when the data used is on or before daynum . one possible cause appears to be sharp increases and decreases in the observed daily deaths. for instance, on daynum = the reported daily deaths was , but on the next two days they were and , and these led to a huge jump in the predicted value for s . also, from daynum ranging from to , the reported daily deaths were , , , , , , , , and , and the predictions were highly unstable and only started to stabilize after daynum = . this is a clear warning on the danger of fitting higher-order models or extrapolating with a long forecasting horizon. as the adage goes, attributed to niels bohr [ ] , with similar versions attributed to mark twain, yogi berra, and others: it is difficult to make predictions, especially about the future. with dire thoughts of the inherent dangers of forecasting over a long horizon, and fully cognizant of the many eventualities (e.g., re-opening of economy; nationwide protests and riots due to police brutality; changing hotspots; adjustments on counts; etc.), which we are not taking into account in order to be purely data-driven, but which could drastically alter trajectories of daily and cumulative deaths, on the basis of the observed data up to july , , in which the cumulative deaths , we seek to forecast the cumulative deaths fifteen days forward, which will be july , . we should mention that on june th, there was an adjustment of which occurred "following a state review of death certificates and prior outbreaks" in the state of new jersey [ ] . together with the adjustment made by the state of new york on april th [ ] , this is the second documented non-trivial adjustment made on the daily deaths counts. we provide two point predictions and prediction regions for the target date: -the first one based on considering the observed deaths values on april th and june th as legitimate values, and the second one based on re-allocating the adjustment values on those days proportional to the observed deaths on the days on or before the day of adjustment. without additional information, such a proportional re-allocation of the adjustment values appears to be most sensible, though this approach is not immune to criticism. figure presents the point prediction and the prediction interval for july , based on these two analyses on with a th-order model. when the adjustment values are not re-allocated, the predictions and prediction intervals are depicted in the two top plots, whereas when they are re-allocated, they are in the two bottom plots. with no re-allocation, the point prediction is together with an associated prediction interval of [ , ] for the cumulative deaths. with re-allocation, the point prediction is with an associated prediction interval of [ , ] . observe that without re-allocation, the observed deaths of on june th fell outside of the prediction interval on that date, while the observed deaths of on april th barely fell inside the prediction interval on that date. notice the wider prediction intervals for deaths when no re-allocations were performed compared to those with re-allocations for the observed daynum values. observe also the widening prediction intervals as we go farther away from july nd, indicating high uncertainty on what may happen moving forward. interestingly, the prediction interval for the cumulative deaths on july th when re-allocations were performed is wider than that without re-allocations, and the point prediction is also tad higher. ominously, notice that the prediction curve for deaths appears to be acquiring an increasing trend past july nd, in contrast to the decreasing trend from daynum (april th) to (june th). it remains to be seen if this is the effect of the lessening of social distancing guidelines, re-opening of business establishments and beaches, people gathering because of the current social unrest, or changing hotspots in the country. of course, in forecasting settings with new data points accruing frequently -daily in this covid- pandemic -forecasts should be updated as each new data point accrues. we intend to provide a publicly-accessible software applet to enable interested users to update forecasts with the latest updated data. motivated by the covid- pandemic, we examined the problem of constructing prediction regions for a poisson distributed random variable, both under the no-covariate (that is, intercept only) figure : based on the available data until july , , point predictions and prediction intervals of the deaths and cumulative deaths by july , (daynum = ) based on an analyses where adjustment values were not re-allocated (top plots) and re-allocated (bottom plots). and with-covariate settings. we compared the performances of the different prediction regions through simulation studies. in the regression setting, we also introduced an over-dispersed poisson regression model upon observing over-dispersion in the covid- reported deaths data relative to a purely poisson regression model. with the ultimate goal of predicting cumulative deaths due to covid- at a future date, we first studied how to construct prediction intervals for the daily deaths data, and then utilized these prediction intervals to construct the prediction interval for the cumulative deaths. the final fitted models involved a th-order model in the variable daynum, and also included the factor variable day. based on data until july , , prediction and prediction interval for the july , cumulative deaths were obtained. the methodologies developed have the potential to be used in the monitoring of daily and cumulative deaths during epidemics or pandemics through the construction of prediction regions, which could then be used by decision-makers regarding implementation of social distancing/easing guidelines and deciding on the closure/opening of business, educational, government, and other establishments. however, further studies are needed to compare our methodologies to other methods that have been proposed during this pandemic. the prediction and prediction region procedures we developed also possess limitations. first, in contrast to the susceptible-exposed-infected-recovered (seir) compartment model cf., [ ] , based on a continuous-time markov chain, our model does not posit an upper bound to the number of people that could die, which clearly is not the case. second, especially since it involves higherorder terms in daynum, they are highly sensitive to outliers, such as when huge adjustments are made as in the cases for the states of new york and new jersey [ , ] . it would be desirable to develop procedures that are robust to such non-trivial adjustments, or to procedures that impose constraints on the rate of increase or decrease of the prediction curve via regularization to curb the influential impact of such adjustments, though this will entail developing new theory for the construction of prediction regions. due to these first two limitations, the proposed methods are not suitable for use in long-horizon forecasting, hence our decision to simply forecast days forward. third, the procedures are not adaptive in its choice of the prediction model. possible improvements may occur by choosing the prediction model in a data-dependent manner, but then model choice uncertainty needs to be accounted for in constructing prediction regions. fourth, there could be an advantage in utilizing other bases functions to transform the variable daynum, such as by using laguerre polynomials, legendre polynomials, trigonometric functions, or even splines or wavelets. these limitations of the proposed methods generate several potential research avenues for further studies. table : simulated coverage probabilities, mean of the lengths, and standard deviation of the lengths of the prediction intervalsΓ andΓ j , j = , , , , , for different λ's and n's. for each combination of (n, λ), replications were performed. for λ = . table : table continued. for λ = . n gam cp gam cp gam cp gam cp gam cp gam cp why is nyc reporting surge in virus deaths? tracking covid- excess deaths across countries. the economist prediction intervals for poisson regression a primer on stochastic epidemic models: formulation, numerical simulation, and analysis statistical models based on counting processes nj says , additional residents likely died of coronavirus after review of death records statistical inference. the wadsworth & brooks/cole statistics/probability series theoretical statistics coronavirus: what is the true death toll of the pandemic? european centre for disease prevention and control: an agency of the european union coronavirus disease (covid- ) pandemic: increased transmission in the eu/eea and the uk -seventh update statistical methods for groundwater monitoring why new york's coronavirus death count jumped: the stories of patients who died at home covid- pandemic modeling fraught with uncertainties sojourning with the homogeneous poisson process envstats: an r package for environmental statistics applied life data analysis median confidence regions in a nonparametric model r: a language and environment for statistical computing. r foundation for statistical computing introduction to the theory of nonparametric statistics linear models in statistics adventures in stochastic processes an empirical bayes approach to statistics an empirical bayes estimation problem some thoughts on empirical bayes estimation standardized surveillance case definition and national notification for novel coronavirus disease (covid- ) adventures of a mathematician of cambridge series in statistical and probabilistic mathematics a. tables of simulation results key: cord- -lamfknpt authors: cina, stephen j.; trelka, darin title: sports-related injuries and deaths date: - - journal: forensic pathology of infancy and childhood doi: . / - - - - _ sha: doc_id: cord_uid: lamfknpt physical activity in children and adolescents should be strongly encouraged. while there is a very low risk of death associated with participation in athletics within this age group, the epidemic of childhood obesity and sedentary lifestyle must be combated to ensure the long-term health and quality of life of today’s youth. sports-related deaths due to trauma are usually readily identified; others require careful examination, adjunctive testing, and/or the expertise of consultants. a thorough investigation of circumstances surrounding the death, review of the medical records, and autopsy is mandated in these cases. the loss of a child is always tragic. when death comes to a young person in peak physical condition engaged in athletics, the fatality strikes a blow to an entire community and often attracts the attention of the national media. in some cases, screening studies or training modifications could have prevented the end result. in many others, however, these sudden, unexpected deaths are the result of conditions that cannot reasonably be anticipated or avoided. the benefits of physical activity in young people are incontrovertible. in fact, a presidential initiative has focused on the necessity for solid nutrition and exercise among american youth. physical activity is essential to well-being, and it must be encouraged in children. childhood obesity has dire health consequences and creates a pattern that can result in significant morbidity and mortality in later life (see ▶ chap. , "childhood obesity"). that being said, engagement in athletics can result in serious injury or death, albeit very infrequently. deaths secondary to trauma are fairly self-explanatory so only a brief overview is in order. much of this chapter will focus on natural disease processes and pathologic conditions that can present as sudden death while a child or adolescent is involved in physical activity. a sports-related fatality is one in which death occurs while the participant is engaged in athletics. within this broad category, death can be directly attributed to an injury received during the activity in which case the manner of death is best certified as an "accident" provided that the trauma was received in accordance with the rules of the sport. in the case of death due to trauma inflicted flagrantly outside of the rules of the sport or if the lethal injury was intentionally inflicted, a manner certification of "homicide" may be more appropriate. in many cases, sudden death may be the result of physical stress superimposed upon a natural disease process or pathologic condition, often involving the heart. in this setting, the manner of death should be certified "natural," analogous to myocardial infarction occurring in an older person engaged in physical exertion (froede ) . the sequelae of repetitive blows to the head have attracted much attention in recent years (omalu et al. (omalu et al. , . while the manifestation of repeated concussions usually appears in middle age or later, it is probable that the damage begins when the brain is first jarred with resultant alteration in mental status that is characteristic of a concussion. what is certain is that concussions must be treated as a serious medical condition, and vigilance is necessary to ensure the safety of participants who are in a post-concussive state. the clinical entity known as the "second impact syndrome" (sis) can cause morbidity and mortality following head injury (bey and ostick ) . sis is comprised of two events: ( ) a concussive head injury and ( ) a second head injury within several weeks followed by cerebral edema, herniation, and death. although the incidence is arguable and is yet to be firmly established, it is thought to be a rare outcome of head injury (bey and ostick ) . in any event, any athlete who manifests concussive symptoms following a head injury (e.g., fatigue, confusion, headache, nausea, vomiting) should be closely observed and not be permitted to return to play for - days (bey and ostick ) (table . ). most serious head injuries occur in the traditional contact sports. they run a spectrum that includes superficial lacerations, contusions, and abrasions to skull fractures, cerebral contusions, deep axonal injury (in the delicate white matter of the brain), and intracranial bleeding. when a human head collides with another object, skull fractures may occur. protective gear, such as helmets, minimizes this risk in many sports. in an unprotected head, skull fractures may be associated with tearing of arteries within the bones of the calvarium, including the middle meningeal artery, resulting in epidural hematomas (edh) that may evolve rapidly and compress the underlying brain. cerebral contusions may also occur at the fracture site. these types of injuries are surgical emergencies. with a more significant direct impact to the head, an open fracture may occur with resultant direct injury and extrusion of the brain. impacts to the face can result in fractures with resultant compromise of the upper airways. deceleration injuries to the head can also be devastating. when a moving or falling head makes contact with a firm surface, injuries to the brain and intracranial bleeding may occur. in contradistinction to the cerebral contusions and/or fractures directly subjacent to the site of impact of a moving object with a stationary head, deceleration injuries may be associated with contrecoup cerebral contusions. these lesions are located opposite to the site of impact of a moving head with a stationary surface. common locations for contrecoup cerebral contusions are the inferior aspects of the frontal lobes and the anterior temporal lobes. contrecoup contusions may or may not be accompanied by basilar skull fractures. rapid deceleration of the cranial contents can also result in diffuse axonal injury (dai) in the white matter and intracranial hemorrhage. subdural hematomas (sdh) secondary to venous bleeding following rapid deceleration of the head can result in a potentially lethal increase in intracranial pressure that must be aggressively managed. with a whiplash type of motion or significant hyperextension of the neck, severe injuries to the cervical spine and underlying spinal cord can occur (watanabe et al. ) resulting in paralysis, respiratory arrest, hemodynamic instability, or death. this type of injury can be seen in violent collisions between bodies, after being ejected from a moving vehicle or animal, or upon impact with the ground while the body is tumbling or rolling. violent impacts to the face, as seen in boxing, can also cause the head to snap back or rapidly rotated with laceration or dissection of the vertebral arteries and subsequent subarachnoid hemorrhage (nedeltchev and baumgartner ) . at autopsy, the pathologist should remove the brain, cerebellum, pons, and medulla and may choose to preserve the block in formalin to permit careful sectioning after weeks. if cervical injuries are anticipated, the pathologist should employ anterior and posterior neck dissections and/or vertebral artery dissection for accurate evaluation. injuries to the ribs and internal organs with subsequent internal bleeding can occur with significant impact to the chest, back, or abdomen. while rib fractures can be debilitating and painful, they are not usually lethal unless there are associated vascular or visceral lacerations or collapse of the lung and pneumothorax. internal bleeding is most often associated with lacerations of the spleen or liver following an impact. the bleeding may occur over a matter of hours or days, so a careful history may be required to establish that the injury occurred during participation in a sport. the spleen is particularly prone to injury if enlarged due to infectious mononucleosis, so infection with epstein-barr virus should be considered if rupture occurs after relatively trivial impact. direct impacts to the abdomen can also injure the mesentery, pancreas, or gastrointestinal tract. the kidneys, being relatively protected by their retroperitoneal position and the presence of a thick fat pad, are injured less frequently. a well-established cause of death in athletes receiving a precordial impact is commotio cordis (westrol et al. ; geddes and roeder ) . death is the result of a lethal dysrhythmia related to a blunt force impact to the chest occurring at a vulnerable phase in the cardiac cycle. classically, the athlete is struck in the midchest by a projectile (e.g., a baseball) and collapses within a matter of seconds. reconstruction of the events leading up to death is required to establish this diagnosis as there may be minimal or no anatomic signs to establish chest trauma and the mechanism of death is transient disruption of impulses within the cardiac conduction system. injuries to the pelvis are unusual in contact sports. however, a significant impact to the perineum, a situation that may occur during riding events (e.g., riding cycles, motorbikes, horses), may cause pelvic fractures, injuries to the genitourinary tract, and internal bleeding. impacts to the scrotum and penis can also result in significant pain and morbidity, but they are rarely life threatening. fractures, sprains, strains, and dislocations are commonly encountered in sports, but these injuries are also rarely life threatening. while in a prolonged debilitated state during rehabilitation, however, thrombi may develop in the deep veins of the lower extremities which may embolize to the lungs resulting in sudden death. if a preexisting coagulation disorder is present, the risk for deep venous thrombosis is increased. smokers and female athletes on birth control pills may be at greater risk for this complication. fractures can lead to embolization of marrow elements, predominantly fat, throughout the body (fig. . ). in addition to the problems associated with physical obstruction of vessels by large emboli, disseminated intravascular coagulation (dic) and activation of chemical mediators can result in death (hofmann et al. ) . the fat embolism syndrome typically occurs - h following a long bone fracture or crush injury presenting as adult respiratory distress syndrome (ards). deep trauma to adipose tissues can also cause fat embolization. of note, fat and marrow emboli are often the result of cardiopulmonary resuscitation with associated rib and/ or sternal fractures, so clinical correlation is required. soft tissue injuries have also been associated with the development of "flesh eating" (group a streptococcus) bacterial infections, even with no breach of the integument (chang et al. ). it should be remembered that even if death is due to a natural disease process, such as sepsis or pneumonia, the manner of death should be certified as "accident" if trauma initiated the chain of events that culminated in death. sport participants are often under intense pressure to perform at a high level, and they may be encouraged to maximize performance through the use of supplements and dietary modification. this is not only true of professionals but also of amateurs and individuals as young as preadolescents. when investigating the death of an athlete, a complete dietary history, including inquiry into the use of chemicals, vitamins, herbal supplements, and performance-enhancing drugs, should be obtained and any such substances procured. anabolic steroids have been historically used to facilitate strength and speed increases, muscle hypertrophy, and decreased recovery time and to generally improve athletic performance (hartgens and kuipers ) . pathologic changes manifested following anabolic steroid use may be appreciated on external examination and can include testicular atrophy, male pattern alopecia, male gynecomastia, masculinization, breast size and body fat decreases, clitoral enlargement, acne, and hirsutism in females (us department of health and human services national institute on drug abuse ). anabolic steroid use can result in peliosis of the liver, psychiatric instability, cardiomyopathy, and death fig. . fat and bone marrow elements may embolize from fracture sites to the lungs. this may be a cause of death, an artifact of trauma, or secondary to cardiopulmonary resuscitation (hematoxylin and eosin, h&e  ) (hartgens and kuipers ) . recently, anabolic steroid use has been implicated in suicides and homicides (so-called "roid rage"). the long-term effects of other performance-enhancing substances, such as human growth hormone (hgh) and creatine, have not been well established and are not recommended for children and adolescents. testing for these "performance-enhancing substances" are commonly performed on urine and hair samples through the use of reference laboratories. stimulants and diet aids can also cause or contribute to sports-related deaths. substances containing ephedrine and ephedra alkaloids have been linked to sudden death (haller and benowitz ) . "energy drinks" often contain agents that can contribute to cardiac deaths and hyperthermia as can certain antihistamines (clauson et al. ; lópez-barbeito et al. ) . it goes without saying that illicit drugs, including cocaine and methamphetamine, can contribute to or directly cause the death of athletes. participants in sports requiring lean body mass are at risk for death related to dehydration or metabolic abnormalities associated with anorexia nervosa and bulimia (warren ; misra and klibanski ) . it should be stressed that these disorders afflict adolescents who are concerned about their body image as well as those participating in competitive sports. investigative history consistent with these disorders are drastic weight loss, a history of using either prescription or over-the-counter medications to facilitate urination and defecation, use of appetite suppressants, a history of vomiting after eating, and exercising incessantly. physical manifestations appreciable on external examination may include cachexia; calluses, scars, or abrasions on the hands if fingers are used to induce vomiting; dental caries or loss of tooth enamel from chronic exposure to gastric acid; and periorbital, conjunctival, or scleral petechiae from induced vomiting (department of health and human services office on women's health ). vitreous electrolyte analysis may shed light on deaths due to dehydration or self-imposed starvation or malnutrition, but it may not establish the cause of death in all such cases. once again, a careful investigation may be required to establish this risk factor for sudden death (see ▶ chap. , "starvation, malnutrition, dehydration, and fatal neglect"). cardiac disease is the leading cause of sudden death in athletes engaged in sports and strenuous activities. until proven otherwise, a cardiovascular source of death should be sought when an athlete unexpectedly collapses and dies. this category of death can be broadly divided into infection, congenital conditions (molecular and structural), coronary artery anomalies, neoplasms, and progressive organic diseases. myocarditis is an inflammatory process involving the heart characterized microscopically by an inflammatory infiltrate in the myocardial interstitium accompanied by myocyte necrosis (fig. . ) . in the majority of cases, the inflammation is due to a viral infection (e.g., coxsackie virus and adenovirus), and a lymphocytic infiltrate will predominate. clues to the diagnosis include a recent viral illness and a "floppy" heart upon gross examination. although a viral etiology can be demonstrated in some cases through laboratory studies, in other cases the infectious agent will not be isolated. other myocarditides are caused by bacteria, fungi, parasites, or autoimmune processes. depending on the etiology of the process, the inflammatory infiltrate may consist of giant cells, eosinophils, histiocytes, or neutrophils. histologic sections may require special stains (e.g., brown and hopps, silver, or periodic acid-schiff stains) in order to better delineate microorganisms. sarcoidosis, a granulomatous inflammation of the heart, may be the result of a postinfectious inflammatory response or of an autoimmune process, the etiology of which remains unclear. special stains to rule out tuberculosis and fungi should be employed to support this diagnosis. congenital conditions may manifest themselves at a structural, cellular, or molecular level. there are a litany of metabolic diseases that may infect the heart, including pompe disease and other storage disorders. these are beyond the scope of this chapter and will not be discussed in further detail, other than to say that they may be a cause of sudden death in childhood. many of these diseases are symptomatic early in life (see ▶ chap. , "cardiac channelopathies and the molecular autopsy," ▶ chap. , "other pediatric cardiac conditions," and ▶ chap. , "pediatric metabolic diseases"). at the molecular level, two major considerations are long qt syndrome and brugada syndrome (goldenberg et al. ; escárcega et al. ). these cardiac ion channelopathies may result in sudden, unexpected death in apparently healthy individuals. there is an association with death during swimming with long qt syndrome (choi et al. ) , which may be diagnosed by evaluation at reference laboratories if it is suspected and appropriate samples are obtained. these diagnoses can be made by retrospective analysis of electrocardiograms in some cases; however, in many young people, this antemortem study has never been performed. these conditions cannot be diagnosed at the gross or microscopic level as they are rhythm disturbances. hypertrophic cardiomyopathy (formerly asymmetric septal hypertrophy, idiopathic hypertrophic subaortic stenosis) can be diagnosed grossly and microscopically. in classic cases, the interventricular septum will be markedly thickened when compared to the left ventricular free wall. in other cases, the left ventricle may show concentric hypertrophy; the right ventricle may also be thickened. often, fibroelastosis of the endocardium below the aortic valve is seen as a "jet lesion" (fig. . ) . microscopically, myocyte disarray with intervening fibrosis is the characteristic histologic finding (fig. . ) . this finding may be focal, and multiple microscopic sections of the heart with trichrome staining may assist in the diagnosis. this disease is caused by a protein abnormality in the heart resulting from a mutation in the genes encoding for the sarcomeric proteins (e.g., myosin heavy and light chains, myosin-binding protein c, troponins i and t, and tropomyosin) (harris et al. ) . as hypertrophic cardiomyopathy is an autosomal-dominant inheritable condition in approximately half of the victims, this diagnosis has implications for surviving family members (cirino and ho ) . marfan syndrome affects multiple sites in the body. the cardiovascular manifestation of this condition is cystic medial necrosis of the aorta. this may result in aortic dissection with rupture into the pleural spaces or pericardial sac with cardiac tamponade or dissection of the coronary arteries. marfan syndrome should be suspected in the sudden collapse and death of tall athletes with long hands and feet (arachnodactyly), a desirable physique for basketball and volleyball players. this disease is caused by a mutation in the fibrillin- gene, the product of which is an extracellular matrix glycoprotein that maintains the structural integrity of connective tissues (robinson et al. ) . arrhythmogenic right ventricular cardiomyopathy (arvc) may be inherited in an autosomal-dominant pattern (azaouagh et al. ) . it is characterized by progressive replacement of the myocardium of the right ventricle by adipose tissue fig. . in hypertrophic cardiomyopathy, a fibroelastotic "jet lesion" is often found on the endocardium subjacent to the aortic valve and fibrosis. occasionally, there are a few scattered inflammatory cells. in advanced cases, the left ventricle may also be involved. this condition may be undiagnosed as the findings are subtle in the early stages, and the right ventricle is often undersampled for histologic analysis. this diagnosis can, at times, be difficult to make, and cardiovascular pathology consultation may prove beneficial. structural defects resulting in sudden death may or may not be grossly apparent. valvular anomalies, septal defects, and transposition of the great vessels can be readily identified at autopsy. deaths due to structural anomalies of the coronary arteries may be more subtle. consultation with a cardiovascular pathologist may be helpful in identifying coronary arterial atresia, intramyocardial tunneling, or acute origin from the sinus of valsalva. these experts may also assist in identifying problems with the cardiac conduction system. these may be either aberrant neural pathways or stenoses of the arteries supplying the atrioventricular (av) or sinoatrial (sa) nodes. a microscopic tumor of the av node can also result in sudden death (fig. . ). structural anomalies of other blood vessels may also lead to sudden death. arteriovenous malformations, particularly within the central nervous system, may rupture with catastrophic results. "berry" aneurysms of the cerebral vasculature may enlarge over time, and intense physical exertion with associated elevation of blood pressure (e.g., weightlifting) may precipitate bleeding. aneurysms and pseudoaneurysms of large arteries cause massive internal hemorrhage. primary cardiac neoplasms are rare but they can lead to death. tumors that affect the heart include atrial myxoma, fibroma, and rhabdomyoma, the latter associated with tuberous sclerosis. the heart may also be affected by lymphomas, angiosarcomas, and metastatic disease. the most common cancers metastatic to the heart are lung, breast, melanoma, and leukemia/lymphoma. adolescents are not immune to cardiovascular diseases that kill older individuals. especially in the setting of familial hypercholesterolemia, atherosclerotic coronary artery disease may develop in the mid-teen years. hypertension may also result in myocardial hypertrophy and lethal dysrhythmia; however, this must fig. . myocyte disarray with intervening fibrosis is characteristic of hypertrophic cardiomyopathy. it may be focal, and multiple heart sections should be examined if there is no apparent cause of death in an athlete following autopsy (hematoxylin and eosin, h&e  ) be distinguished from hypertrophic cardiomyopathy, discussed above. whereas hypertrophic cardiomyopathy commonly affects the septum on gross inspection and is associated with myocyte disarray, hypertension generally results in concentric thickening of the left ventricular chamber and enlarged, hypertrophic myocytes with hyperchromatic "box car" nuclei at the microscopic level. lastly, morbid obesity has been associated with sudden death (see ▶ chap. , "childhood obesity"). when faced with an unanticipated subarachnoid hemorrhage, the pathologist should remove the brain themselves with frequent photographic documentation of the process in order to capture occult lesions prior to onset of any removal artifact(s). once removed, the brain, cerebellum, pons, and medulla should be copiously rinsed with water to remove adherent blood and clot. in lieu of water, hydrogen peroxide may be used to facilitate the lysis of adherent blood from the delicate vasculature so that it can be better examined. care must be taken to avoid destruction of subtle vascular malformations and aneurysmal sacs. sickle-cell disease may be diagnosed in childhood, and it can afflict participants in athletics and other strenuous activities. if the diagnosis of sickle-cell disease is known, recognition and treatment of an impending crisis can avert death. many people with sickle-cell trait, however, are unaware of their condition. when subjected to intense physical exertion, high temperatures, and a component of dehydration, a crisis may ensue and death may rapidly follow (scheinin and wetli ; manci et al. ) . a recent viral illness could also be an exacerbating factor. this condition should be considered when an athlete of african or mediterranean descent complains of joint pain, chest pain, fatigue, and weakness prior to collapse. it is often misdiagnosed as a heat-related illness. the diagnosis is made at the microscopic level wherein virtually all organs will be congested by sickled erythrocytes (figs. . and . ) . at the gross level, persons with sickle-cell disease may have fibrotic, atrophic spleens, whereas those with sickle-cell trait may have enlarged spleens, congested with sickled erythrocytes. the diagnosis can be confirmed with hemoglobin electrophoresis (blood best procured in a tube with anticoagulant/edta) and correlated with information obtained regarding prevalence and distribution of this disease within the family. asthma is a common disease among children and teens, and acute attacks may be precipitated by physical activity. in order to certify death due to asthma, the circumstances of death need to reflect a respiratory crisis. in many cases, an inhaler and/or a nebulizer will be found near the victim or with their personal belongings. grossly, the lungs will be hyperinflated, often touching over the heart in the midline, with prominent mucus plugging of the airways. the microscopic findings of chronic asthma in the bronchioles (thickening of the basement membranes, smooth muscle hypertrophy, and mucus gland hyperplasia) will be accompanied by an eosinophil-rich inflammatory infiltrate that extends into the luminal mucus plugs (figs. . and . ) . a history of asthma should not be a default cause of death in these cases without the circumstances and scene findings supporting a respiratory catastrophe. epilepsy, like asthma, may be a cause of death, however, the circumstances should support the diagnosis. in the absence of a witnessed seizure, other causes of death must be excluded prior to attributing death to epilepsy. further, intracranial trauma must be excluded as a cause of the seizure. some epileptics will die suddenly and unexpectedly in the absence of a seizure (sudden unexpected death in an epileptic person or sudep), but this does not typically occur while the victim is engaged in sports. the autopsy may or, more commonly, may not identify the anatomic correlate of the seizure focus in the brain. spontaneous pneumothorax may occur during sports presumably due to increased shear forces at the apex of the lung (abolnik et al. ) . it may cause death if it progresses to a tension pneumothorax, a condition that results in both hypoxia and mechanical alterations of the cardiovascular system. this is a diagnosis that may be missed if it is not suspected. diabetes mellitus can kill children and adolescents engaged in sports. activities which entail dietary restrictions may predispose those with the disease to ketoacidosis. young people more often have type i diabetes and may be insulin dependent. as teens have a tendency toward denial and risk-taking behavior, they may not be fully compliant with their treatment regimens and therefore be prone to significant blood glucose fluxes. the gross findings at autopsy will be minimal in these cases. microscopically, the islets of langerhans in the pancreas may be infiltrated by lymphocytes ("insulitis"), or they may be diminished in number. urine screens for glucose and ketones may be useful, but postmortem blood analysis is unreliable. the best sample for diagnosing diabetes mellitus and ketoacidosis postmortem is vitreous humor. the presence of ketones and significantly elevated glucose (> mg/dl) in the vitreous humor is diagnostic of this condition (chansky et al. ). vitreous glucose levels drop significantly after death, however, so a lower ocular glucose level does not exclude hyperglycemia. further, hypoglycemia cannot be diagnosed postmortem due to the aforementioned postmortem change. in the evaluation of a nontraumatic death occurring during sports, analysis of vitreous glucose, ketones, and electrolyte levels is recommended in all cases. prior to the examination, chest radiographs including lateral and seated views may illustrate free pleural air and displacement of the heart. at autopsy, care should be taken to reflect the skin and soft tissues of the chest without breaching the intercostal tissues or entering the chest cavities. a pocket which should be filled with water can be created using the reflecting chest tissues. the intercostal tissues can then be pierced below the water level to examine whether air bubbles emerge. the findings of chronic asthma (basement membrane thickening, smooth muscle hypertrophy, mucus gland hyperplasia) with an intense eosinophilic inflammatory response in a person who died during an asthma attack during exercise (hematoxylin and eosin, h&e  ) second impact syndrome children with lethal streptococcal fasciitis after a minor contusion injury hyperglycemic emergencies in athletes spectrum and frequency of cardiac channel defects in swimming-triggered arrhythmia syndromes familial hypertrophic cardiomyopathy overview safety issues associated with commercially available energy drinks department of health and human services office on women's health ( ) fact sheet on bulimia nervosa the brugada syndrome evolution of our knowledge of sudden death due to commotio cordis long qt syndrome effects of androgenic-anabolic steroids in athletes adverse cardiovascular and central nervous system events associated with dietary supplements containing ephedra alkaloids in the thick of it: hcm-causing mutations in myosin binding proteins of the thick filament pathophysiology of fat embolisms in orthopedics and traumatology diphenhydramine overdose and brugada sign causes of death in sickle cell disease: an autopsy study bone metabolism in adolescents with anorexia nervosa traumatic cervical artery dissection chronic traumatic encephalopathy in a national football league player chronic traumatic encephalopathy (cte) in a national football league player: case report and emerging medicolegal practice questions the molecular genetics of marfan syndrome and related disorders sudden death and sickle cell trait: medicolegal considerations and implications anabolic steroid abuse. nih publication number endocrine manifestations of eating disorders upper cervical spine injuries: agespecific clinical features causes of sudden cardiac arrest in young athletes key: cord- -yhi hgq authors: kovács, katalin title: social disparities in the evolution of an epidemiological profile: transition processes in mortality between and in an industrialized middle income country: the case of hungary date: - - journal: mortality in an international perspective doi: . / - - - - _ sha: doc_id: cord_uid: yhi hgq the present paper seeks to understand the transformation of mortality patterns in hungary, by which mortality inequalities by education began to appear in the early s, continued to grow in the following years, and now seem to be stabilising. the first part of this paper overviews the theoretical innovations of the last decades regarding the interpretation of cause-specific mortality dynamics, often referred to as epidemiological transition theories, and their relevance for the analysis of mortality inequalities. the paper then analyses the cause-specific trends of mortality for two educational classes between and . the trends were corrected for changes in the coding system and divided into linear (stagnating, increasing or decreasing) periods. causes of death were grouped according to the relationship between the sequences of these periods for the two educational classes. the causes of death were finally clustered into six groups. one group, which is dominated by nutrition-related and cardiovascular diseases, is largely responsible for the onset of mortality inequalities in . the results imply that the quality of nutrition has diverged for the educational classes since , and this fact has left its footprint on the pattern of mortality. the history of food production and availability seems to be in line with nutrition-related mortality, and it is argued that nutrition transition theory provides a very plausible explanatory framework for the growth of mortality inequalities. countries in central and eastern europe (cee) experienced rapid industrialisation under equalising state-controlled regimes, and entered into the globalising international economy two decades ago. their transformation into service economies is still an on-going process. over the past years social inequalities have increased sharply, reaching medium-level income inequalities in an eu context, which is considered high by the citizens of these countries. in the present paper i look at the implications these changes have had on the level and distribution of mortality in hungary, as an example of this group of countries. all-cause mortality is considerably higher in cee countries than in the rest of the european union, but it was recently shown to correspond to the income level of these countries (spijker and von wissen ) . on the other hand, inequalities in mortality by education have been found to be extremely high in all of these countries (mackenbach et al. ) . so far the explanation for these developments has only been provided within a larger context that applies to the whole of the eastern european region, including not only cee and baltic countries but also countries like russia, belarus and ukraine. one of main conclusions has been that they have not so far undergone the healthier life style changes that have occurred in western europe, and this has resulted in a "reversed epidemiological transition", in which an elevated burden of cardiovascular diseases dominates the pattern of mortality (vallin and meslé ) . is this framework applicable to central and eastern europe and does it explain the evolution of their cause of death pattern and high level of inequalities? if so, what role did income play in these processes and what are the specific social processes that triggered these developments? in hungary, inequalities in all-cause mortality were negligible during the 's and widened during the 's. the next one and a half decades brought a further, dramatic, increase in inequalities, which appear to have stabilised at this very high level for the past half a decade ( fig. . ) . as regards broad groups of causes of death, the data suggest that the apparent similarities in all-cause mortality during the s might be attributable to causes other than the lack of inequalities in living conditions between people with different education. this period was characterized by the overmortality of the less educated from cardiovascular diseases and the over-mortality of the more educated from malignant tumours (fig. . and . ). an explanation is called for which will look at the historical development of cause-specific mortality within the framework of the epidemiological transition. in this paper i shall review recent developments in epidemiological transition theory, and test the applicability of some of these theories to the evolution of cause-and education-specific mortality inequalities in hungary between and . smr, / smr, / long-term mortality trends are commonly interpreted within the framework of epidemiological transition theory, outlined years ago by omran ( ) . the original statements of the theory on mortality, fertility and population growth have already been tested, analysed, criticised and modified. by now, epidemiological transition theory and demographic transition theory have split: the first one has gradually shifted towards a focus on cause-and age-specific mortality patterns, while the second is now far more concerned with patterns of fertility and family formation. the original postulates of omran are, without doubt, of a heuristic nature: based on limited empirical basis (in its original form it was based on the long-term causespecific mortality trends of just six countries), it provided a comprehensive picture of the evolution of cause-of-death patterns throughout the history of mankind. in a rather vague division of human history, three transitional phases were distinguished: the 'age of pestilence and famine', the 'age of receding pandemics', and the age of 'degenerative and man-made diseases' (omran ) . stages were differentiated on the basis of average life expectancy, and age-and cause-specific mortality. during the first phase, which encompassed most of human history (the "pre-industrial period", omran ), mortality due to chronic malnutrition, endemic infectious diseases, and high prenatal and maternal mortality shaped the overall high level of mortality, which was further increased by epidemics, famine and wars in the "peak years". in the second stage, which started in the eighteenth or nineteenth century in western societies, mortality declined considerably, mainly due to factors other than medical interventions: improved nutrition, improvement of personal cleanliness, ecological recession of certain diseases, better housing conditions and the start of using contraceptive methods. the cause-of-death pattern was less and less characterized by diseases caused by pandemics but communicable diseases-tuberculosis in particular-were still dominant. the third stage is characterised by the dominance of non-communicable diseases, such as diseases of the circulatory system and different types of cancer. from the perspective of the future development of the theory, the additional characteristics of the stages are less important, though omran's approach, which has been modified several times by himself and others over the past years, remained complex and aimed at explaining the whole of population dynamics. the evidence which accumulated subsequent to omran's original article, coming mainly from countries of the americas, shows little correspondence with this original framework and offers an amazing variety of cause-specific mortality patterns and their changes over time (albala and vio ; castillo-salgado et al. ; costello and osrin ; hill et al. ; huicho et al. ; marshall ; vigneron ; vigneron ) . this evidence was incorporated into the original model as three models, the classic, the delayed and the accelerated models (omran ) . later on this was expanded to six models, the classical western model, the semi western model and four non-western models: the rapid, the upper intermediate, the lower intermediate and the slow (omran ) . other authors suggested a different classification of countries (frenk et al. ) in order to incorporate new evidence that did not fit into the original sorting in the first form of the epidemiological theory. from the perspective of countries in latin america, the concept of an epidemiological transition was in sharp contrast to the mortality experience of many countries of the region, which were characterized by a sharp divide between the mortality patterns of different population groups within one country. this experience questioned the choice of countries as the units of analysis, and even raised doubts about the usefulness of the whole concept of an epidemiological transition. apart from total refutation, the experience of latin american countries are best summarized as a "patchwork pattern" in which different social groups are often segregated geographically, and display diverse mortality patterns corresponding to different stages of the epidemiological transition. in other words, "different epidemiological worlds" live next to each other. for industrialised countries, on the other hand, a large collaborative study of who did confirm the previously proposed trend of age-specific death rates for two broad groups of causes of diseases in the last half of the twentieth century (salomon and murray ) . the more and more sophisticated classification, however, did not help to overcome one of the major theoretical drawbacks of the original theory. despite the very complex, and somewhat apocalyptic, view of the future presented by omran in his last article ( ), epidemiological transition theory presents a linear view of changes in mortality patterns, according to which more developed stages follow less developed ones, alongside with the course of 'modernisation'. this process may take place slowly or quickly, and with some variations, but it also follows a linear route. in this respect, epidemiological transition theory does not differ from theories of modernisation propounded in the 's (carolina and gustavo ) and is very similar to the dominant view of demographic transition theories (melegh andŐri ) . the theory of an epidemiological transition was attractive not only for public health researchers, it can also be viewed as a major contribution to the on-going debate of historians and historical demographers centred around the nature of mortality changes in the last centuries. in countries with a long history of collecting detailed mortality data, the distinctive phases of receding epidemics and the death toll of infectious diseases in general could be identified. due to the great variability within the regions of one single country (for sweden: rogers and nelson ) , this further classification, unlike the contribution coming from the discipline of gerontology, has not become a commonly accepted modification of the original epidemiological transition theory. olshansky and ault ( ) carried out a detailed examination of the age-specific death rates of the us and pointed to the onset of an epidemiological phase that differs from the one specified as the third stage of the epidemiological transition in the original form of the theory. this fourth stage, they suggested, is characterized by the dominance of the same major causes of death as the third stage but with a continuing delay in mortality from some of these causes, leading to a further significant improvement in life expectancy. the new stage, called the 'age of delayed degenerative and man-made diseases' has become a standard part of the most commonly accepted form of the epidemiological transition theory. anthropology or 'evolutionary biology' has also made its contribution to refining epidemiological transition theory by adding a new transition stage, thus refining how the original theory divided up the other end of the historical time-scale (armelagos et al. ). the addition of the "baseline" mortality pattern, called the 'palaeolithic stage', however, is less relevant from the perspective of the current research. the approach of evolutionary medicine, which emphasises the links between the specific nature of human production, diet and other aspects of living conditions, and causespecific mortality, can, however, be beneficial in interpreting contemporary mortality trends as well. another major discipline contributing to the further refinement of epidemiological transition theory was epidemiology itself. recent trends in epidemiological research clearly exhibit some fragmentation. epidemiology was first concerned with certain diseases in detail but recently more comprehensive approaches have emerged. alongside the continuing research of the risk factors associated with specific diseases, trends in mortality due to major groups of diseases have also been studied and the results and hypotheses presented in the framework of "sub-transition models" such as cancer transition and cardiovascular transition (also known as the cardiovascular revolution). the cancer transition is an extension of the classic transition theory that takes into account new discoveries on the role of infections in the development of certain types of cancers. the discovery of the presence of bacteria in the majority of stomach cancer cases promoted the recognition of how important infections are in cancer in general, though the infectious origin of certain other cancers (such as cancer of the cervix, testicular cancer and certain lymphomas) was already well-known. new discoveries triggered the formulation of "cancer transition theory" (gerstein and wilmoth ) , according to which there is a definitive restructuring process in operation within cancer mortality: those with an infectious origin lose their importance and other non-infection-related cancers emerge. most cancers, however, are known to be influenced by some major risk factors such as non-appropriate diet, smoking and excessive alcohol consumption. these well-known risk factors are now more and more closely linked to societal transformation processes, mostly of a global nature. from among these theoretical frameworks we note in particular the theory of a nutritional transition (popkin , popkin and mendez ) . in this framework, major features of food production, distribution and several other characteristics of living conditions are connected to mortality patterns. nutritional transition theory, just like epidemiological transition theory, divides human history into five distinctive patterns, out of which the fourth corresponds to the living conditions of contemporary cee countries. the fifth profile describes the living conditions and dietary habits of the most health-conscious members of the most affluent countries. labelling the phases not as stages but as profiles obscures the fact that these patterns are arranged in historical order so that they also represent some "developmental route". the patterns, however, are connected with a large number of dimensions of actual living conditions. as regards the transition to the fourth ("obesity characterized") profile, several processes, such as "supermarketisation", are connected to several social processes like the demand for safer food, the changing opportunity cost of females' time, technological changes, and changes in logistics and production systems. altogether this transition is technically characterized by the growing importance of edible oil and animal products in human diet. additionally sugar consumption is on the rise, often in the form of consuming sweetened beverages. the shift from high fibre intake to refined grains and additionally declining fruit and vegetable intake is also documented in many countries (popkin ) . transition theories regarding other risk factors are less developed at the moment, but the term "smoking epidemic" is also in use and the influence of strong economic forces has already been recognised (yach et al. ) . regarding the other dominant groups of diseases, cardiovascular mortality has always been regarded as being strongly related to the epidemiological transition. ischemic heart disease in particular often serves as a "marker disease" that indicates a country's position in the phases of the epidemiological transition (heuveline et al. ) . based on the observations of the contemporary occurrence and frequency of different cardiovascular diseases in different regions of the world, a complete framework for "cardiovascular transition" has gained popularity in the past few years. this framework provides a correspondence between particular cardiovascular diseases and stages of the epidemiological transition (califf et al. ) . the linearity of the occurrence of the stages is not stated but it is inherent in the logic of this scheme. the 'pestilence and famine' stage, with life expectancy around years, is characterized by a modest share of cvds in total mortality ( - %) and the dominant forms of cardiovascular mortality are rheumatic heart disease and other infection-related diseases of the circulatory system, cardiomyopathy in particular. the latter disease may also be connected to malnutrition. in the second stage the proportion of deaths caused by cvd grows to - % and cardiovascular mortality is dominated by rheumatic valve disease, ischemic heart disease and haemorrhagic stroke. in this stage life expectancy reaches about years. in the third stage, in which life expectancy reaches years, the proportion of deaths due to cardiovascular diseases is greater than %. the dominant causes of death within cvds are ischemic heart disease, and ischemic and haemorrhagic stroke. in the stage of 'delayed degenerative diseases' the proportion of cvds among all deaths falls below % and life expectancy exceeds years. major cardiovascular causes of death are the same as in the previous stage, with the addition of congestive heart failure. another important observation not exactly linked to stages is a major shift between stroke types: haemorrhagic stroke declines while ischemic stroke emerges (lawlor et al. ) . a fourth stage also appears in some variations of the "cardiovascular transition" schemes. in some cases (yusuf et al. a ) a stage of 'health regression and social upheaval' is visualised, characterised by the re-emergence of rheumatic heart disease and a new increase in ischemic heart disease due to increasing alcoholism. in the increasingly unregulated social environment, violence also becomes more common and hypertensive disease-which is otherwise characteristic of stage according to these authors-also re-emerges. this visualisation, of course, relies heavily on recent russian mortality trends. other authors have predicted the emergence of heart failure as the main characteristic of a future scenario for cvd mortality (bonnux et al. ; gaziano et al. ) . risk factors for cardiovascular diseases were found similar to the ones identified for cancers but the linkage between the single diseases and the exact role of single risk factors is less clear, with some exceptions. for two major different stroke types, for instance, different set of risk factors had already been identified (o'donnall et al. ) , though inappropriate diet, smoking and excessive intake of alcohol play an important role in the development of all cardiovascular diseases. predictions on the future trends of mortality and cause-specific mortality are not restricted to the field of cardiovascular diseases. in his last publication omran ( ) also outlined a fifth stage, the 'age of aspired quality of life with paradoxical longevity and persistent inequalities'. in this he expressed his hope for a future decrease of inequalities in survival, together with an expectation that there was a high probability of the re-emergence of infectious diseases. nevertheless, the 'invisible perils' in the future of mankind are considered by omran as well, such as the possibility of the evolution of an (airborne) virus with abilities similar to those of hiv; the potential misuse of atomic bombs; and high, uncontrolled population growth. omran's view on the unpredictable role of infectious diseases is not unique. several other authors foresaw the future emergence of new diseases and the re-emergence of "old" infectious diseases that were previously believed to have been controlled by medical interventions. notable examples are the emergence of multi-drug resistant tuberculosis and avian flu. following the emergence of the hiv/aids pandemic, the fear of new infectious diseases is spreading. scientific examination reveals, however, that the majority of the emerging and newly recognised diseases are in fact not new but were known only in some peripheral regions of the world and have reached the consciousness of the wealthy only recently (farmer the circumstances of the onset of "newly identified" diseases during the s pointed out that human activity played a triggering role in the majority of the cases. in the integrated view of evolutionary medicine that divides human history only into three epidemiological transitions, the "third transition" is the new era of emerging and re-emerging infectious diseases (harper and armelagos ) . the "end of the antibiotic era", as this approach calls it, results mostly from the intensification of the globalisation process, especially that of the transportation system, which serves as a 'virtual superhighway' for pathogens. figure . outlines the theories providing a comprehensive explanation for changes in patterns of mortality and their phasing. while many epidemiological transition theories cover the whole of the history of mankind, others refer only to developments in the latest centuries, or even just decades. most of them inherently treat the process of change in mortality patterns as "development", i.e. as a linear, and in some respect hierarchical process. possible reverses and uncertainties mostly appear regarding the latest stages-which is probably due to empirical observations being more numerous and diverse regarding the near past. the issue of social disparities is present in nearly all approaches to the epidemiological transition. in most cases social inequalities in mortality or diverse mortality patterns that are characteristic of social classes, strata or groups are discussed in relation to major drivers (or causes) of the epidemiological transition. in some cases drivers or causes are stated only in general, like modernisation, industrialisation and urbanisation. in other cases propositions are well-formulated and corroborated by some empirical evidence. omran, starting from his very first publication, continuously mentioned social disparities in mortality as well as the driving forces listed above but he did not provide a theoretical framework for the application of these in connection with particular mortality or disease patterns specific for single countries or population sub-groups. mckeown ( a mckeown ( , b, mckeown and record ) studied the disappearance of infectious diseases in industrialising england and formulated his famous nutrition hypothesis. detecting a time-lag between the almost complete disappearance of numerous infectious diseases, and a very notable drop in tuberculosis mortality, before the discovery of the appropriate treatment methods (mostly antibiotic drugs), he concluded that the major cause of decreasing mortality was the improvement in the living conditions and, in particular, the nutritional status of the population during the nineteenth century. the nutritional thesis provides an obvious explanation for social disparities in mortality, whose modified versions later appear in recently developed explanatory approaches. evolutionary medicine, with its anthropological orientation, considers the "palaeolithic" baseline to have been free from social hierarchies in early human communities (harper and armelagos ) . notably they also focus on dietary habits. they suggested that there was a low mortality period before human communities settled down, as a result of their varied diet as well as small population size. mortality started to grow when diet became heavily reliant on crops, which were unequally distributed across population strata. in parallel, the growth of average community size led to new, infectious, diseases becoming the leading causes of death. based on this approach one can conclude that unequal access to food results in unequal resistance to diseases, thus inevitably leading to inequalities in mortality. historians and historical demographers, however, present a rather different picture of those centuries of human history which can be characterised by the dominance of infectious diseases. they suggest that some of the infectious diseases exhibit a "discriminative" nature: there is a long incubation period between the moment of infection and the development of the disease and the resistance of the host matters during the process of battling with these diseases. other infections, by contrast, are "quick" enough not to allow time for the host (the human body) to develop resistance and they kill in a short time; consequently, they can be considered "nondiscriminative". several infectious diseases, such as smallpox and mumps, have been observed to change over the centuries, as their originally "non-discriminative" nature turned into "discriminative". it is still unclear if the changing nature of some formerly fatal diseases is due to increased community-level resistance to those diseases or merely to the changing nature of the disease-scape. however, the disappearance of some infectious diseases, notably the plague, is still explained in several alternative ways (slack ) . during early modern times, when infectious diseases dominated mortality, the excess mortality of those in disadvantaged social positions was likely to be more pronounced in those causes of death that were connected with epidemics and pandemics. according to historical demographers, excess deaths were indeed connected to the availability of food, though this relationship was largely influenced by the effectiveness of supportive networks (bengtsson et al. ) , which helped to mitigate the effect of economic hardships (e.g. famine). regarding the plague outbreaks in london, it was observed that the locus of epidemics moved from the central, relatively wealthy parishes to the poorer suburban ones during the seventeenth century. overall, it is likely that social disparities existed in the era of infectious diseases, though their importance might have changed over time, partly due to factors that operated independently of social organisation and human activity and partly due to greater awareness and ability to cope among the wealthy (slack ; hall ) . the early industrial era inevitably brought large mortality inequalities, which are well documented for some countries. time series of mortality data by social groups, however, are not available for many countries. studies using time series on income inequality for industrialised countries have suggested that mortality inequalities were narrowing from the first years of the twentieth century until about , and widening afterwards. detailed british data (pamuk ) has reinforced this view. this process probably took place in varying ways in different regions: in sweden no sign of the emergence of social inequalities in mortality was found till the s (bengtsson and dribe ) . theoretical explanations for modern inequalities have emerged in the fields of both epidemiology and sociology. in the epidemiology of cardiovascular diseases a particular "disease mobility" was observed first: in the beginning of the twentieth century myocardial infarction used to be the disease of the affluent in developed north american and european countries, but in the s and the s mortality rates due to infarction started to decline earlier and more rapidly among the better educated and the better off (marmot et al. ; kaplan and keil ) . these experiences led to the formulation of the social "following pattern" of diseases. based on the concept of the diffusion of innovation, pearson ( ) suggested an "adoption theory": population groups with higher education and/or better income adopt new ideas, products and behavioural patterns more readily. once a risk factor is recognised, it first becomes public knowledge among those with higher social status, mediated by health education or mass media. the messages reach the poorer and/or less educated groups of societies later. the life course perspective for understanding the occurrence of chronic diseases also originates from cardiovascular epidemiology (forsdahl ; kuh and ben-shlomo ) . its scope, however, is much wider, identifying risk factors that act during the in utero period and early childhood, risk factors which are associated with the social position of the parents (davey-smith and hart ). from the point of view of social sciences, these findings call for integrating intergenerational mobility into epidemiological transition models that are used to understand the particular mortality patterns of single countries. in the social sciences the "fundamental causes" concept was introduced in order to understand the relationship between socio-economic status and disease. these fundamental causes do not refer to causes of death but to dimensions of social position which are causally linked to resources that can be used to avoid risk or to minimize the consequences of diseases once they occur. resources include money, knowledge, power, prestige and interpersonal relationships. fundamental causes act, according to the proponents of this idea, when new diseases, new risk factors or new knowledge on risk factors emerge or new medical technologies are introduced (link and phelan ) . in these cases living conditions and access to resources act directly to grant or restrict different groups' access to, and application of, the new technology. social position, therefore, is the fundamental cause of a disease (or death) and not a "proxy", as it was previously treated in epidemiological research. the concept of fundamental causes has only recently been applied to the analysis of cause-of-death patterns (miech et al. ) . the examination of education-specific mortality inequalities and their dynamics over the last decades of us history aimed at testing the fundamental cause hypothesis. a large number of causes of death ( ) were included in this examination. in accordance with the concept of fundamental causes, the analysis found increasing inequalities for most "emerging" causes, e.g. those whose overall rate was in an increase. omran's classic paper on the epidemiological transition ( ) positioned hungary together with the rest of "eastern europe", in the same model as japan. mortality developments have diverged significantly since then. the latest additions to the concept of epidemiological transition provide no direct guidance for understanding overall mortality trends and educational inequalities in mortality. detailed knowledge has accumulated on the changes in mortality profiles in developed high-income countries. mortality trends, especially the burden of infectious and non-communicable diseases, are widely discussed with regard to low income countries. industrialized middle-income countries seem to be neglected in the discussion of the epidemiological transition. in order to fill this gap, first we examine the applicability of one of the previously outlined theories that focus primarily on other regions of the world: the plausible "following" hypothesis. the higher overall level of mortality as well as the cause of death patterns in hungary (and other cee counties), often referred to as "lagging behind" those of western europe, might be interpreted as the mortality pattern of a society in which large population segments who are "lagging behind" produce an overall "delayed" cause-of-death pattern and large mortality inequalities at the same time. if this proposal were true and meaningful, one would observe the same mortality dynamics for the more and the less advantaged segments of the population but with some time lag regarding the latter group. existing data allow us to examine these processes by education only: i shall therefore compare the mortality development of the less and the more educated hungarian adults. as a contrast i also examine the applicability of another popular branch of theories known as risk behavioural factor approaches, in particular, the possible role of nutrition in shaping cause-of-death pattern differences. mortality data for hungary, by education, are available from . for the period between and we calculated age-standardized cause-specific mortality rates by education for the population aged and above. cause-specific death rates were also calculated for the whole population and by education groups. data on the number of deaths by education is provided by the mortality register of the hungarian central statistical office. corresponding population estimates and forecasts were prepared by lászló hablicsek, based on census data from , and (hablicsek and kovács . underlying causes of death were included in the analysis. education level was dichotomized: high (completed years and passed the matura exam) and low. these two groups will be referred to as the less and the more educated. selecting the relevant causes of death was a multi-stage process. first we selected causes cited in discussions of the epidemiological transition theory that linked their theoretical considerations to empirical analysis. the starting point, however, was the broad categorization into the two distinctive groups of causes of death which came out of the who global burden of diseases study (salomon and murray ) . group included the infectious diseases; diseases of the pulmonary system and several diseases connected to malnutrition and maternal mortality. group encompassed all other diseases, except the external causes: injuries, homicide and suicide. looking at a large number of countries over shorter or longer observational periods (from to ) and taking into account total mortality and wealth (as measured by gdp), salomon and murray ( ) found no consistent relationship between external causes and total mortality or wealth, and we decided therefore to leave them out of the present analysis. the next step in selecting the causes of death was based on those considerations which have been summarized in the introduction. additional results from studies that analysed time trends for a number of diseases in specific countries with regard to the epidemiological transition were also included, particularly studies on the epidemiological transition in the netherlands (wolleswinkel-van den bosch ; wolleswinkel-van den bosch et al. ) and in canada (lussier et al. ) . for group causes, the identification of nutrition-related, pulmonary or maternal causes of death is not problematic. the large group of 'infectious and parasitic diseases', as the international classification of diseases calls it, was much more difficult to break down into smaller and meaningful causes of death, because if anything is clear from the literature, it is that infectious diseases are generally declining but they still vary significantly by country. therefore we decided to select all those causes for which more than cases were found for each year during the period between and . this procedure resulted in a list of one disease: tuberculosis. we also added the "new diseases" such as hiv/aids and newly recognised and antibiotic-resistant infectious diseases. these categories turned out to be almost empty. in practice, the study also includes a number of infectious diseases which are traditionally classified under pulmonary diseases (such as influenza, pneumonia) or other major disease groups (peptic ulcer, appendicitis), or whose coding in some periods overlaps other broad cause-of-death groups (meningitis, enteritis). group included different types of cancers and cardiovascular diseases, divided up according to those "sub-theories" of the epidemiological transition which we briefly introduced earlier. for cardiovascular diseases, the categorisation was based on the list of diseases that appear in different versions of the "cardiovascular transition". apart from these, some other distinctions were also made according to major coding categories such as chronic and acute ischemic heart diseases. among cancers, we distinguished in particular all those cancer types with are connected with infections. a further distinction was made by major risk factors, including not only smoking, excessive drinking and obesity, but also environmental and occupational exposures (for a short summary see table . ). this categorisation, however, does not lead to easy interpretation, due to the pervasive and complex nature of the everyday operation of risk factors. some other diseases, specifically discussed by certain authors with respect to the epidemiological transition, such as alzheimer and parkinson's disease, were also added. the list of the causes of death that we selected for analysis is included in the appendix, together with the coding used. age-specific death rates by the selected causes of death (where possible) were used to create standardized mortality rates using the european standard population. mortality trends, resulting from the standardisation process, did not form continuous time series in most cases, as illustrated in fig. . . there were three different icd coding versions in operation during the observed period, and in addition, "automatic coding" was introduced in , which again affected the structure of the (underlying) causes of death, as if another new icd version had been introduced. icd- was introduced in and icd- in . first we fitted the different versions of icd codes, often with the help of literature, in order to achieve the same content for each disease over time. when code-fitting was not obvious, we relied on code-fitting used by others wolleswinkel-van den bosch et al. ; hashibe et al. ; lawlor et al. ) . the resulting time series called "original values" still did not construct continuous curves in this study. there are three known methods to deal with the changes of icd coding system. the first one, the "double or bridge coding" would require coding death in a certain period according to both the outgoing and the new coding systems. this task was carried out only in for the hungarian mortality data. the second method follows the exact matching of the disease categories by four-digit coding (meslé and vallin ) . this method was partly used in this study but only for some specific causes of death. after establishing the coding we followed a third method of fitting the curves (janssen and kunst ) but applying a simpler method than they did. fitting coefficients were calculated by causes of death, but always for the entire population. the same coefficients were used to fit curves for those with lower and higher educational background. the values of the coefficients, listed in the appendix, provide an overview about the reliability of the estimated time series : the closer the coefficients are to , the higher the reliability. no fitting was applied in the case of those causes which were too small to calculate standardized rates or for those which showed outliers "too often", such as influenza. overall mortality trends were similarly not fitted ( fig. . ) . we now turn to consider the relation between the two estimated mortality time series for groups with lower and higher education. for certain causes of death, almost exclusively in those years when the annual number of deaths is very low, it was not possible to determine definitive relations since the low number of deaths did not allow for standardisation, so fitting was also not applied. therefore the general trends of overall mortality due to these causes are difficult to establish. this fact is well illustrated in the case of obesity. from this cause less than deaths were reported annually between and , but about in the following years. as for inequalities, a clearer picture emerges from the distribution of the number of deaths: most of them appeared among the less educated. deaths due to nutritional anaemia, malnutrition and obesity, as well as maternal death almost exclusively happened among those with lower education. for other rare causes of death such as hiv/aids and "newly emerging infectious diseases", however, no such pattern evolves. hiv/aids mortality was the highest in , when deaths were attributed to this disease. the number of cases declined afterwards and people with lower and higher educational attainment seem to be equally affected. among the newly emerging diseases only deaths were reported from , again distributed proportionally between the educational classes. after disregarding the above-mentioned causes of death, we categorized the remaining causes by the relationship between the two mortality time series displayed by the groups with lower and higher education. the classification of the relations rested on a simplified view of the time series. given that we worked with estimated values in the classification, the dynamics represented by the time series were the focus. the time series were broken down to linear (growing, stagnating or declining) phases and the classification was based on the relationship between the sequences of these phases by causes of death, presented by the two mortality time series. time series were broken down into phases using join-point regression analysis, with software provided by the national cancer institute of the united states. this regression is for analysing trends and the software fits data in the simplest possible sequence of linear trends which are connected by the join-points. first a linear trend for the overall period is fitted, then trends with a growing number of joint-points are also fitted and their significances are tested against the null-hypothesis (e.g. having join-points). the tests of significance are based on a monte carlo permutation test. the breakdown of the time series was successful in most of the cases, though the method applied involves some uncertainties. the location of the join-points is provided together with confidence intervals, which were often very wide, covering even - years. in the following classification only those periodicities were considered when confidence intervals for the joint points were shorter than years. uncertainties were taken into account in all those cases when confidence intervals were wider than - years. the sequences of linear trends and the corresponding set of join-points by cause of death are not given here but are available from the author. to examine the "follow-up" hypothesis, first one has to give a clear definition of a follow-up pattern of two curves. the method chosen for this analysis was not to construct a general definition but first to regard the estimated mortality time series for the two educational groups, then to classify them by their type of relation and then to examine the possible interpretations of their being "follow-up" by type. the application of this method resulted in six different groups of diseases, according to the relationship between the mortality trends estimated for the more and the less educated. this classification allows us to investigate the possibility of providing a proper definition of follow-up. in the case of diseases with strongly declining mortality (type i) the definition of follow-up is not obvious at all. the dynamics of decline did not provide any meaningful definition of follow-up, since for the major diseases of this category (pulmonary tuberculosis, haemorrhagic stroke and cancer of the stomach) the timing of strongly declining and the less strongly declining periods, represented by the mortality of the less and more educated, mostly coincide ( fig. . ) . the existence of sequences of declines with a different pace also means that a definition based simply on when mortality of the less educated reached the mortality level of the better educated also gives no clear-cut answer: for instance, the value of tuberculosis mortality of the more educated in was reached years later by the less educated, but the values for the more educated in or in were reached, by the less educated, only or years later, respectively. in the case of influenza, the level of fluctuation highly exceeds the level of inequalities. for rarer diseases that also belong to this class of causes of death, temporal but irregular high peaks of mortality among the better educated would make it difficult to define a follow-up pattern (table . ). in the case of some other diseases, mortality of the less and the more educated also shows similar sequences of periods of linear trends, but the overall trends are not declining (type ii, fig. . ). inequalities change little or not at all over time and the mortality of the more educated never (in "regular" cases such as the hypertensive diseases of the circulatory system or cervical cancer) or just in exceptional years (in the case of mesothelioma and epilepsy) reaches the level of the less educated. providing any follow-up definition seems meaningless in these cases (table . ). in a number of diseases, however, the sequences of the linear periods of different types are also similar for the less and the more educated, but the overall dynamics of the curves turn to be very different. for these causes of death mortality levels are quite similar at the beginning of the period considered here, but at a certain point of time mortality of the two groups starts to diverge quite distinctly (type iii, fig. . ) . regarding most diseases in the class of type iii mortality, negligible differences in mortality characterise the beginning of the observed period and then the same types of linear trends apply to both educational groups, but the levels of mortality iv. the two educational groups start with the same type (inclining, declining or stagnating) of trend and mortality from the disease is higher for the better educated. mortality of the better educated changes its trend in a certain year for the better and finally huge over-mortality of the less educated is present only mortality of the better educated shows major trend change during the period resulting in higher mortality among the less educated by the end of the period , similar to the one we give for type i diseases. in other cases, however, when the mortality of both educational classes increases, there is no sign that mortality of the less educated would follow that of the more educated by any means. it is more plausible that "the same story is played out" for both of the educational groups concerning risk factors or general conditions of life but with very different risk levels. for type iv causes of death ( fig. . ) , the less educated population is characterised by growing mortality, while the mortality of the better educated changed from a growing to a declining trend. similar trend changes can be expected in the future for the mortality of the less educated, but this change will appear later than the end of our observation period. approximate minimum time-lags for the onset of this change are given in table . . in practice these time lags can also be a bit longer, since we cannot be sure if the last couple of years of the observation period represented the beginning of a new type of trend or not. altogether, a clear follow-up pattern was detected only for three-though very important-causes of death (fig. . ) . as regards acute ischemic heart disease, ischemic stroke and breast cancer, the sequences of the rising and declining periods are similar for the less and the more educated with a time-lag, so the mortality of the less educated seems to follow the mortality of the more educated. though it is impressive that the estimated follow-up time is the same for ischemic heart disease and ischemic stroke, it is important to mention that these "scenarios" are also "played out" at different mortality levels. at their maximum values, breast cancer and ischemic stroke mortality of the better educated is % higher and that of acute ischemic heart disease is % higher than those of the less educated, suggesting that follow-up type explanations need to be supplemented for a full understanding. type v causes of death are characterized by different trends for the two educational groups for the whole of the period (see fig. . ). for some, the mortality of the less educated increases and that of the more educated decreases (aneurysm); for others, both are in increasing but with different intensity (cancer of the oral cavity, non-hodgkin disease) or the mortality of the better educated is declining while the mortality of the less educated is stagnating (cancer of thyroid, prostate cancer, peptic ulcer). the possibility that these diseases start to decline or strongly decline among the less educated can certainly be hoped for, but the follow-up time would be longer in these cases than our observation period. there is thus no point in laying down a definition for the purpose of this study. some diseases, typically rare causes of death, could not be classified into the previous types and they are placed into type vi, represented by fig. . . though a clear follow-up pattern was identified for only three causes of death, there is evidence of some kind of follow-up for a large number of diseases but it is not easily identified. time lags are usually long, exceeding more than one or two decades, so that while follow-up may provide a vague and partial explanation for mortality developments and the development of inequalities for the chosen relatively short time period, it certainly does not a provide a full picture. taking a closer look at the onsets of trend changes, it is quite obvious that they cluster in time. most of the changes occurred in the very first years of the s and around . both these periods were important turning points, and the two clusters can thus be interpreted as indicators of two diverging trends in living conditions, in the widest possible sense of the term. in the first cluster we find, surprisingly, a number of nutrition-related causes of death: diabetes, other endocrine diseases, and two strongly nutrition-related cancers (cancer of the uterus and gallbladder cancer). trends of mortality by education diverge from about the same point of time for a number of causes of death related to the circulatory system: chronic ischemic heart disease, arrhythmias, heart failure, atherosclerosis, other diseases of the veins and arteries and pulmonary heart disease. causes of death which are possibly nutrition-related, such as cancer of the rectum, acute ischemic heart disease and ischemic stroke, also show signs of changing mortality relations by education between and . some other causes of death, which are clearly not nutrition-related, such as meningitis, cancer of the liver, cancer of the brain, melanoma, other skin cancer and leukaemia, join this cluster. the most likely interpretation of the existence of this cluster is that these changes reflect the widening inequalities in the quality of diet for the two social groups distinguished by educational level. alternatively, within the risk-factor oriented explanatory framework, one can argue that all these changes are attributable to diverging trends of excessive alcohol consumption, noting that cirrhosis of the liver, the only cause of death which is clearly related to alcohol consumption, started to emerge a couple of years earlier. alcohol-related changes are known to have an immediate mortality impact but some possible effects of the divergence in alcohol consumption cannot be ruled out. altogether, divergence in nutrition seems to provide a more suitable explanatory framework. the second cluster includes causes of death with important trend changes between and . smoking-related causes, such as cancer of the larynx and cancer of the trachea, bronchus and lung, clearly dominate this cluster, joined by some other diseases such as colon, pancreas, kidney and bladder cancer and valve diseases with other than rheumatic origin. attributing the evolution of this cluster to the appearance of the divergence in smoking habits in the two educational classes, it is to be noted that this divergence point seems to be more diffused in time than the one related to the divergence in nutrition: trends of important smoking-related causes of death (cancer of the oesophagus) started to diverge a year earlier than , though this cancer type is also influenced by nutrition. to explain the rise of mortality inequalities between the less and the more educated from the very beginning of the s in hungary, one might turn towards basic sociological approaches which would focus on the changing relations of education and income, assuming that the relationship between the two was non-existent in the s and became gradually stronger over the period between and . from a simplified point of view on the former state socialist states that assumes that these countries had no income inequalities at all, the onset of mortality inequalities during the s must be a mystery. in fact, income inequalities were already present and connected to educational levels during the s in hungary. even after taking compensation in-kind into account-since a large share of incomes was undoubtedly distributed in this formthe income of those with higher education can be estimated as being twice as high as that of people without this qualification (pető and szakács ) . during the s the maturation of the "second economy" partly confused this relationship. in this period the state made some form of economic activities free from its direct control; therefore, in this sector (especially in agriculture) a limited market economy developed. social status was distributed along two axes: in the formal economy, in which income and education were correlated, creating very mild income inequalities, and in the informal economy, in which education and income did not correlate strongly (kolosi ) . since the emergence of the free market economy following , the correlation of income and education has become stronger and stronger, just as in most european countries (tóth ) . this, coupled with the lack of a significant improvement in gdp, led to widening social inequalities and the extension of poverty. the changing relation between income and education therefore plays a certain role in explaining widening mortality inequalities, but it cannot explain the negligible mortality inequalities which existed during the s nor their revival during the s. we should look, therefore, at nutrition-related risk factors. the food supply in hungary was mostly based on domestic production during the s and s. limited exchange with other state socialist countries existed but imports were mainly limited to a small amount of tropical fruits. domestic products, however, were satisfactory for domestic demand. agriculture had developed into one of the leading ones in europe and from the 's there was no food shortage in hungary. the distribution of food was rather even and quality differences by education hardly existed. during the s, with the growth of the "second economy", food provision varied and prices were already partly market-driven. the better-off could use their resources to purchase better quality food and these provisions were available to a large share of the population, but obviously not for everyone. low food prices, together with energy prices which were still subsidized, made it possible for a larger proportion of the population to buy food of satisfactory quality. in the countryside, "around-the-house" agricultural activity was widespread, producing mostly for the household (occasionally producing for the market, too). during the 's the proportion of food grown "around-the-house" was estimated at % of the overall food consumed (ksh ) . from the s the food supply and the price system of the country were placed into a global context. open trade relations provided a great variety of available food, while domestic production, including around-the house output, started to decline. food prices relative to income represented a greater and greater share of household expenditure and competed with rising energy costs. around-the-house agricultural production, which had been characteristic for many households for decades, halved in less than a decade: its share in overall food consumption of % in had shrunk to % by (ksh ). as a result of these processes, the availability of quality food has been shrinking for an ever growing proportion of the population. domestic agricultural production, however, started to recover in the last years of our observation period, as the states that joined the european union in came to benefit from the unified european agricultural policy. the history of food production and food availability seems to run in parallel to the inequalities in nutrition-related mortality, so this narrative provides a very plausible explanatory framework for our findings. if this framework is supported by similar findings from other countries, then we can conclude that the mortality of the middle income industrialised countries, with moderate income inequalities, is still strongly determined by nutritional differences and by the lack of availability of quality food for large proportions of their populations. social differences in food intake have been described both in wealthy and poorer countries and are usually discussed in connection with obesity. major changes in human nutrition have also been described, characterized by a growth in sugar and animal source food intake (popkin ) . in the context of wealthier countries, the poorer nutritional habits of the less educated is usually understood in the context of lack of knowledge, forced habits by tradition or lack of awareness due to putative or real economic interests. in the case of poor countries the phenomena is understood in the context of absolute deprivation and poverty. several facts indicate that none of these scenarios are appropriate for middle income countries. hungarian household surveys, for instance, indicate that the amount of sugar and sweetened beverages consumed is much lower in low income households than in households with higher income. the difference in this respect between the lowest and the highest income quintile households was fivefold in . some features of the differences in food consumption, however, run parallel with the pattern of the western countries, such as the similar levels of pork consumption of households with different income and the large gaps in poultry, fruit and vegetable consumption. relatively high pork intake is the only fact which would suggest that tradition also plays some role in forming nutrition patterns. differences in fruit and vegetable consumption fluctuate and depend on yearly prices (polgár ; ksh ) so there is good reason to attribute these differences to the decline of around the house production and the lack of financial resources. food intake differences by education can largely be explained by rising poverty among the less educated and the changes in the system of food production and pricing. as a generalisation of our findings, we note that the nutritional elements of living conditions are rarely measured in europe and they are usually restricted to the poorest countries. in the first relevant eurobarometer survey, however, less than % of the west european population answered "yes" to the question if paying for food causes any (some or serious) problem, and the corresponding proportion was between and % for central, eastern and baltic countries (not including the czech republic and slovenia). these data refer to the years around . publicly available raw data of the second european quality of life survey ( ) indicate that the question of food quality is still relevant in cee and baltic countries. for the only directly foodrelated question ("can you afford a meal with meat, chicken or fish every second day if you want it?") no more than % of the population gave a negative answer in west european countries, whereas this proportion was around % in most cee and baltic countries and in greece, and even higher in some countries such as slovakia, bulgaria and hungary ( , and %, respectively) . the same question was included in the same year in the european statistics on income and living condition survey and released results (ward et al. ) suggest that that survey yielded a similar picture: no more than % of the population in western europe was affected and %- % in central and eastern europe and in the baltic countries (except for estonia, romania and bulgaria). data indicate that even if starvation-related mortality is negligible in lower-middle income european countries, there are good reasons to assume that the quality of nutrition is still not satisfactory for large proportions of the populations in these countries, and leaves its footprint on their mortality pattern. as far as the history of the hungarian food provision regime is concerned, some of its elements can be regarded as similar to other countries of the region, while some other elements are certainly different. the above-mentioned developments in income inequalities and food provision in the s are probably similar in all cee countries, while the introduction of the second economy was unique to hungary. the development of free market conditions from the s and the degree of exposure to the global competition varied over time and between the countries, as did the role of around-the-house agricultural production. rising income inequalities and the application of a global pricing system, however, seem to lead to similar levels of mortality inequalities in these countries, though the composition of over-mortality by cause differs (leinsalu et al. ). cee and baltic countries, therefore, probably share more common features than differences in this respect. the generalization of the findings for the whole region of "eastern europe", however, seems less fruitful, allowing for the fact that the cee and baltic countries have had consistently lower income inequalities than countries of the former soviet union other than the baltic countries. several other aspects of household economy, such as the overwhelming role of energy expenditure in cee countries, are not present in the same way. the analysis of cause-specific mortality is a challenging task. these studies typically go beyond the time periods of consistent registration systems of causes of death and creating credible time series is demanding. the solution chosen in this paper can be criticized and other alternatives of code bridging should be considered in further research. the classification of causes of death by their relation to mortality developments between the more and the less educated can also be questioned and other alternatives should also be regarded. the method followed by this paper was to decompose the overall time series to sequences of linear trends and there is no doubt that other than linear approximate trends could also have been considered. moreover, the linear approximation itself was carried out with a high level of uncertainty: the exact point of time when trends changed was hard to establish, which introduces some uncertainties about the findings. the changing composition of the population over time is an inherent problem of studies examining long term developments. in our case the share of the population aged years or more with less than secondary school graduation was % in and % in . a more detailed educational classification of the population would have been desirable but was impossible to carry out with consistency due to major changes in the schooling system during the observed period. in our discussion we deliberately avoided some important issues which might naturally be regarded as good candidates for explaining mortality inequalities, such as health care provision and differences in health care utilization. the reason for this neglect was the lack of space to cover all elements of cause-specific mortality inequalities in one paper. instead, we aimed at identifying some general driving forces contributing to widening inequalities. setting up an accurate statistical record of the different health services, which would have been necessary to evaluate their role, was beyond the possibilities of this study. similarly, we had to disregard other, similarly important elements of welfare policy, except for some aspects of income distribution. our discussion addresses only some of all the arguments raised in different theoretical approaches to the epidemiological transitions theory. we limited the scope of the paper to looking at the role of nutrition in the long term development of mortality and mortality inequalities. the intention of providing an explanation for the observed mortality trends in connection with the social processes of hungary in the last four decades has left little space for discussing the applicability of other, similarly attractive, explanatory frameworks that undoubtedly have high potential. epidemiological transition in latin america the case of chile cancer is preventable disease that requires major lifestyle changes evolutionary, historical and political economic perspectives on health and disease the late emergence of socioeconomic mortality differentials: a micro-level study of adult mortality in southern sweden - life under pressure. mortality and living standards in europe and asia contribution of environmental factors to cancer risk estimating clinical morbidity due to ischemic heart failure: the future rise of heart failure overweight, obesity and cancer: epidemiological evidence and proposed mechanism for the virtual coordinating centre for global collaborative cardiovascular research (vigour) organization epidemiological transition: model or illusion? a look at the problem of health in mexico a subregional assessment of demographic and health trends on the americas: - epidemiological transition, medicalisation of childbirth, and neonatal mortality: three brazilian birth-cohorts infectious causes of cancer and their detection, mini review life-course socioeconomic and behavioural influences on cardiovascular disease mortality: the collaborative study social inequalities and emerging infectious diseases living conditions in childhood and subsequent development of risk factors for arteriosclerotic heart disease. the cardiovascular survey in finnmark - health transition in middle income countries. new challenges for health care cardiovascular disease the cancer transition in japan since plague in london: a case study of the biological and social pressures exerted by years of yersinia pestis az életkilátások differenciálódása iskolai végzettség szerint interaction between tobacco and alcohol use and the risk of head and neck cancer: pooled analysis in the international head and neck cancer epidemiology consortium the changing disease-scape in the third epidemiological transition social determinants and the decline of cardiovascular diseases: understanding the links the uneven tides of the health transition epidemiologic transition interrupted: a reassessment of mortality trends in thailand mortality profiles in a country facing epidemiological transition: an analysis of registered data icd coding changes and discontinuities in trends in causespecific mortality in six european countries socioeconomic factors and cardiovascular disease: a review of the literature zöldség és gyümölcsfogyasztás, statisztikai tükör, ii. évfolyam . szám a life course approach to chronic disease epidemiology secular trends in mortality by stroke subtype in the th century: a retrospective analysis educational inequalities in mortality in four eastern european countries: divergence in trends during the post-communist transition from social conditions as fundamental causes of disease does the recent evolution of canadian mortality agree with the epidemiologic transition theory? demographic research the second fatal impact: cigarette smoking, choric disease and the epidemiological transition in oceania for the european union working group on socioeconomic inequalities in health health inequalities among british civil servants: the whitehall ii study the modern rise of population the role of medicine-dream, mirage or nemesis? london: nuffield provincial hospital trust the epidemiologic transition: changing patterns of mortality and population dynamics reasons for the decline of mortality in england and wales during the nineteenth century a második demográfiai átmenet elmélete reconstructing long-tem seria of causes of death. the case of france the enduring association between education and mortality. the role of widening and narrowing disparities risk factors for ischemic and intracerebral haemorrhagic stroke study in countries the epidemiologic transition, a theory of the epidemiology of population change the epidemiologic transition theory: a preliminary update the epidemiologic transition theory revisited thirty years later the fourth stage of the epidemiological transition: the age of delayed degenerative diseases social class inequality in mortality from to in england and wales the global health burden of infection-associated cancers in the year education and income: double-edged swords in the epidemiological transition a hazai gazdaság négy évtizedének története Élelmiszermérlegek és tápanyagfogyasztás global nutrition dynamics: the world is shifting rapidly toward a diet linked with non-communicable diseases the rapid shift of the nutrition. the global obesity epidemic the epidemiologic transition revisited: or what happens if we look beneath the surface? the epidemiological transition revisited: compositional models for causes of death by age and sex the disappearance of plague: an alternative view. the economic history review the response to plague in early modern england: public policies and their consequences socioeconomic determinants of male mortality in europe: the absolute and relative income hypotheses revisited income composition and inequalities in hungary convergences and divergences in mortality. a new approach to health transition the epidemiological transition in an overseas territory: disease mapping in french polynesia epidemiological transition and geographical discontinuities: the case of cardiovascular mortality in french polynesia european inequalities. social inclusion and income distribution in the european union the epidemiological transition in the netherlands reclassifying causes of death to study the epidemiological transition in the netherlands mortality decline in the netherlands in the period - , a turning point analysis globalization and tobacco global burden of cardiovascular diseases part i: general considerations, the epidemiologic transition, risk factors, and impact of urbanization global burden of cardiovascular diseases part ii: variations in cardiovascular disease by specific ethnic groups and geographic regions and prevention strategies key: cord- -f mzwhrt authors: aggrawal, anil title: agrochemical poisoning date: journal: forensic pathology reviews doi: . / - - - - _ sha: doc_id: cord_uid: f mzwhrt a general increase in the use of chemicals in agriculture has brought about a concomitant increase in the incidence of agrochemical poisoning. organophosphates are the most common agrochemical poisons followed closely by herbicides. many agricultural poisons, such as parathion and paraquat are now mixed with a coloring agent such as indigocarmine to prevent their use criminally. in addition, paraquat is fortified with a “stenching” agent. organo-chlorines have an entirely different mechanism of action. whereas organophosphates have an anticholinesterase activity, organochlorines act on nerve cells interfering with the transmission of impulses through them. a kerosene-like smell also emanates from death due to organochlorines. the diagnosis lies in the chemical identification of organochlorines in the stomach contents or viscera. organochlorines also resist putrefaction and can be detected long after death. paraquat has been involved in suicidal, accidental, and homicidal poisonings. it is mildly corrosive and ulceration around lips and mouth is common in this poisoning. however, the hallmark of paraquat poisoning, especially when the victim has survived a few days, are the profound changes in lungs. other agrochemicals such as algicides, aphicides, herbicide safeneres, fertilizers, and so on, are less commonly encountered. governments in most countries have passed legislations to prevent accidental poisonings with these agents. the us government passed the federal insecticide, fungicide and rodenticide act (fifra) in and the indian government passed the insecticides act in . among other things, these acts require manufacturers to use signal words on the labels of insecticides, so the public is warned of their toxicity and accompanying danger. a general increase in the use of chemicals in agriculture has brought about a concomitant increase in the incidence of agrochemical poisoning. organophosphates are the most common agrochemical poisons followed closely by herbicides. many agricultural poisons, such as parathion and paraquat are now mixed with a coloring agent such as indigocarmine to prevent their use criminally. in addition, paraquat is fortified with a "stenching" agent. organochlorines have an entirely different mechanism of action. whereas organophosphates have an anticholinesterase activity, organochlorines act on nerve cells interfering with the transmission of impulses through them. a early humans are believed to have started agriculture around bce. as the knowledge of chemistry grew, so did the use of chemicals in agriculture. today, chemicals are used in agriculture for three main purposes: to increase farm production (fertilizers and related chemicals), to kill pests (pesticides), and to preserve farm products (preservatives). unfortunately, all three classes of chemicals can cause serious poisoning in humans, mainly through improper labeling, storage, or use. most poisonings with agrochemicals occur in predominantly agricultural economies where a lack of hygiene, information, or adequate control creates unsafe and dangerous working conditions. cases of such poisonings also occur in small factories where pesticides are manufactured or formulated with little respect for safety requirements. accidental poisonings may also take place at home when pesticides are mistaken for soft drinks or food products, and often the victims are curious children who can easily reach pesticides if they are not kept safely away from them. then, there are the intentional poisonings, where compounds, such as phosphorus, arsenic, paraquat, organophosphates, and strychnine, are used as agents for suicidal or even homicidal purposes. this may happen because these chemicals are easily available, relatively cheap, and almost certainly cause death. poisoning occurring as a result of improper use of chemicals used in agriculture has been termed "agrochemical poisoning." agrochemical poisoning can be classified as shown in table . agrochemical poisoning remains one of the major causes of morbidity and mortality around the world today ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) , and a review of this relatively untouched subject seems to be justified. experience has shown that above the wide range of chemicals a vast majority of poisonings occur because of pesticides only. the annual report of the american association of poison control center's (aapcc) toxic exposure surveillance system listed a total of , , human exposures to poisons occurring in the united states during the year alone ( ) . out of these, there were , exposures to pesticides ( % of all exposures) and , exposures to fertilizers ( . % of all exposures); a total of fatalities caused by pesticides and one caused by fertilizers were reported. the break-up for pesticide exposure is shown in table , and the fatalities caused by pesticides are given in table . two categories in which deaths were not reported at all were fungicides and repellants. most deaths (n = ) were to the result of insecticides. herbicides and rodenticides accounted for five deaths each, and one death was caused by fumigants. a comparison of poisoning data for the years to ( ) ( ) ( ) ( ) indicates that, although the absolute number of pesticide exposure has been increasing, it is more or less stable at around % of all exposures to poisons; fatalities owing to pesticide poisoning amount to . to % of all fatalities resulting from poisons (table ). in the following sections, those agrochemical poisons that are important from a medicolegal and pathological point of view will be discussed. organophosphorus insecticides are derivatives of phosphoric acid (h po ) or phosphonic acid (h po ) in which all h atoms have been replaced by organic moieties (figs. - ) . l represents the so-called "leaving moiety" and is the most reactive and most variable substituent. it is called so because this moiety "leaves" the organophosphate molecule after it is attached to the esteratic site of the acetylcholinesterase (ache, also known as true cholinesterase type che). r and r are less reactive moieties. most commonly they are poisonous plants (used as green manure, e.g., ricinus communis). . chemicals used to kill pests (pesticides) (i) acaricides (used to kills mites and ticks, also known as miticides, e.g., avermectins, azobenzene, benzoximate, bromopropylate, dofenapyn, nikkomycins, tetranactin). (ii) algicides used to control growth of algae in lakes, canals, and water stored for agricultural purposes (e.g., cybutryne, hydrated lime [component of bordeaux mixture]). (iii) aphicides (used to kill aphids, e.g., triazamate, dimethoate, and mevinphos). (iv) avicides (used to kill birds harmful to agriculture, e.g., -aminopyridine, -chloro-p-toluidine hydrochloride). bactericides (e.g., bronopol, nitrapyrin, oxolinic acid, oxytetracycline). (vi) fumigants (gas or vapor intended to destroy insects, fungi, bacteria, or rodents, used to disinfect interiors of buildings, as well as soil, before planting, e.g., carbon disulfide, sulfuryl fluoride, methyl bromide). (vii) fungicides (e.g., sodium azide, various compounds of copper and mercury, thiocarbamates, captan, captafol). (viii) herbicide safeners (e.g., benoxacor, cloquintocet, cyometrinil, dichlormid, dicyclonon). these compounds basically protect crops from herbicide injury by increasing the activity of herbicide detoxification enzymes, such as glutathione-s-transferases and cytochrome p- . (ix) herbicides/weed killers (e.g, paraquat, diquat, - dichlorophenoxyacetic acid, mecoprop). (x) insecticides (e.g., organophosphorus compounds, organochlorine compounds, carbamates). (xi) microbial pesticides (those pesticides whose active ingredient is a bacterium, virus, fungus, or some other microorganism or product of such an organism, e.g., bti which is made from the bacterium bacillus thuringiensis var. israelensis and used to control mosquito and black fly larvae, bacillus sphaericus and laegenidium giganteum, a fungal parasite of mosquitoes). (xii) molluscicides (used to kill molluscs, such as snails and slugs, e.g., metaldehyde). (xiii) nematicides (used to kill nematodes that feed on plant roots, e.g., , dichloropropene, , -dibromoethane, ethylene dibromide, diamidafos, fosthiazate, isamidofos). (xiv) ovicides (used to kill eggs of insects and mites). (xv) pesticide synergists (e.g., piperonyl butoxide, n-octyl bicycloheptene dicarbozimide, piprotal, propyl isome, sesamex, sesamolin). (xvi) rodenticides (used to kill rodent pests, e.g., strychnine, vacor, antu, cholecalciferol, anticoagulants and red squill). (xvii) virucides (e.g., ribavirin, imanin). (xviii) miscellaneous chemical classes including contaminants and adjuvants of some pesticides which are toxic on their own (e.g., dioxins, present as contaminants of some herbicides produce toxicity of their own). . chemicals used to disturb the feeding/growth/mating behavior etc. of pests, or used for other miscellaneous agricultural purposes (i) bird repellents (e.g., anthraquinone, chloralose, copper oxychloride). (ii) chemosterilants (e.g., , -dibromo- -chloropropane, apholate, bisazir, busulfan, dimatif, tepa). (iii) desiccants (chemicals which promote drying of living tissues such as unwanted plant tops or insects). (iv) defoliants (chemicals which cause leaves or foliage to drop from a plant, usually to facilitate harvest). feeding deterrents or antifeedants (chemicals having tastes and odors that inhibit feeding behavior, e.g., pymetrozine, azadirachtin a). (vi) insect attractants (substances that attract or lure an insect to a trap, e.g. brevicomin, codlelure, cue-lure, dominicalure, siglure). (vii) insect growth regulators (chemicals which disrupt the action of insect hormones controlling molting, maturity from pupal stage to adult, or other life processes, e.g., hexaflumuron, teflubenzuron and pyriproxyfen). (viii) insect repellents (e.g., butopyronoxyl, dibutyl phthalate, diethyltoluamide). (ix) mammal repellents (e.g., copper naphthenate, trimethacarb, zinc naphthenate, ziram). mating disrupters (e.g., disparlure, gossyplure, grandlure). (xi) plant activators (a new class of compounds that protect plants by activating their defense mechanisms, e.g., acibenzolar, probenazole). (xii) plant growth regulators (substances [excluding fertilizers or other plant nutrients] that alter the expected growth, flowering, or reproduction rate of plants through hormonal rather than physical action). . chemicals used for preservation of grains (i) aluminum phosphide. (ii) nitric oxide. available as dusts, granules, or liquids, organophosphorus insecticides are among the most popular and widely used insecticides throughout the world. they began to be synthesized first around with the esterification of alcohols to phosphoric acid. the earliest synthesis of an organophosphate, tetraethyl pyrophosphate, was reported by phillipe de clermont at a meeting of the french academy of sciences in ( ) . many different organophosphorus compounds were synthesized in the early s, but their toxicity was first recognized by lange in . lange stated that inhalation of the vapor of dimethyl or diethyl phosphofluoridate produced a choking sensation and dimness of vision. as nations started looking for lethal gases with the start of world war ii in , interest in these compounds was rekindled. by , schrader in germany and saunders in england and their study groups had synthesized a number of highly toxic organophosphates for possible use in warfare. most notable among these were soman, sarin, and tabun. currently, about organophosphorus compounds are in use as insecticides worldwide. of these, parathion is the most effective for insecticidal use. tetraethyl pyrophosphate enjoys two distinctions among organophosphates: it was the first organophosphate to be synthesized in and is the organophosphorus insecticides are basically ache inhibitors allowing the accumulation of excess acetylcholine at various nicotinic and muscarinic receptors throughout the body including the central nervous system (cns). this essentially results in acetylcholine toxicity. the main symptoms can be remembered by either of the two acronyms sludge (salivation, lacrimation, urination, defecation, gastrointestinal distress, emesis) or dumbels (diarrhea, urination, miosis, bronchospasm and bradycardia, emesis, lacrimation, salivation). rarely, there is chromolachryorrhoea (shedding of red or bloody tears) ( ) because of a disturbance in porphyrin metabolism and its accumulation in lacrimal glands. ld (lethal dose; the amount of a material, given all at once, which causes the death of % of a group of test animals) of these compounds varies from to mg/kg (extreme toxicity) to more than mg/kg (slight toxicity). compounds that are extremely toxic are chlorfenvinphos, diazinon, and methyl parathion, whereas those that are slightly toxic are malathion, acephate, and trichlorphon ( ) . most patients who have ingested a fatal dose will die within hours of ingestion. organophosphorus toxicity has recently been reviewed extensively by rousseau and co-workers ( ). signs of asphyxia are commonly found in fatal intoxications with organophosphorus insecticides. there is congestion of the face and cyanosis of the lips, nose, fingers, and acral parts of the extremities. one of the most remarkable findings is the characteristic odor emanating from the corpse: it has been described as garlic-or kerosene-like and is due to the fact that organophosphates are dissolved on a kerosene base. there is often frothy, bloody staining at the mouth and nostrils, and the pupils may be constricted. a coloring agent, indigocarmine, is added to parathion (e ® ) to prevent its accidental ingestion or criminal use as a poison. this gives rise to a bluishgreenish discoloration of the lips and oral mucosa. the addition of indigocarmine, however, is not a general practice worldwide. for instance, in india and several other asian countries, this practice is not followed. an interesting sign to be observed (albeit only in somewhat less modern mortuaries) is the death of bluebottles and others insects and flies dying immediately after they alight on an opened cadaver at autopsy ( ). the gastric mucosa is congested and may appear hemorrhagic (fig. ) and the stomach contents often contain an oily, greenish scum. the mucosa of the respiratory tract is congested and the airway passages contain frothy hemorrhagic exudate. the lungs show congestion, hemorrhagic pulmonary edema, and subpleural petechiae. the brain is swollen and there is generalized visceral congestion. parathion (e ) has been studied most extensively for histopathological lesions and these are considered to be representative of other organophosphorus insecticides, too ( ) . in the kidneys, there is epithelial necrosis in the straight sections of the renal tubules. in the epithelia of the remaining renal cortical sections, there is pronounced plasma granulation, nuclear wall hyperchromatosis, and clumping and reduction in the chromatin and marginal nucleoli. epithelia in loops of henle and collecting tubules appear swollen. the liver is more resistant to the effects of organophosphates, partly because of its ability to manufacture serum cholinesterase on its own. hepa- tocytes show opaque swelling and glycogen depletion; there are destructive changes in the liver cell strands, detached hepatocytes, and perivascular edema. myocardium, medulla oblongata, and vagal nuclei of the brain show fine, maculate perivascular hemorrhages. limaye has described a type of toxic myocarditis that he had observed in autopsy cases ( ) . kiss and fazekas described focal myocardial damage with pericapillary hemorrhage, micronecrosis, and patchy fibrosis in victims of organophophorus poisoning ( ) . pimentel and da costa ( ) have described the following myocardial ultrastructural changes in fatal poisonings with organophosphorus: multiple circumscribed necroses are found in the skeletal musculature. the oolemma is damaged and sometimes even necrotic. the glomus caroticum shows an increase in the number of dark-cell nuclei, perhaps as a consequence of increased nuclear metabolism owing to augmented demand. ache and butyrylcholinesterase (bche, also known as pseudocholinesterase or type che) levels are depressed in deaths owing to organophosphorus insecticides. the measurement of their levels can assist in the determination of the cause of death ( ) . ache is found mostly in red blood cells, motor endplates, and gray matter, whereas bche is found mostly in plasma, white matter, liver, heart, and pancreas. the physiological function of bche is unknown ( ) , but it is established that bche hydrolyzes suxamethonium (succinylcholine), and for this reason it is of interest to anesthesiologists as well. postulated functions of bche include its role in transmission of slow nerve impulses, lipid metabolism, choline homeostasis, permeability of membranes, protection of the fetus from toxic compounds, and degradation of acetylcholine and in tumorneogenesis ( ) . the plasma cholinesterase (pseudocholinesterase) is more sensitive and levels fall more rapidly than those of the red blood-cell cholinesterase. red blood-cell cholinesterase levels are more satisfactory for the diagnosis of organophosphorus poisoning because they represent the true cholinesterase levels. sample collection and storage (time and temperature) are critical to the catalytic stability of che and thus influence the quality and interpretation of results of the toxicological analysis. fluids and tissues that should be collected at autopsy are blood, cerebrospinal fluid (csf), semen, muscle, brain, liver, heart, and pancreas. the recommended procedures for collection and storage of biological fluids are as follows: . blood must be collected in heparinized tubes. . the samples must be collected and stored in glass rather than plastic containers to avoid contamination by leachates from plastic. . sample contamination with acid or alkali must be avoided. . samples must be immediately refrigerated because che catalytic activity is temperature dependent. . fluid and cellular components of blood, csf, and semen have to be separated. . determine enzyme activity as soon as possible. if enzyme activity is not determined immediately, samples can be stored for several days at °c. if tissues are intended to be stored for longer periods, the storage temperature should be - °c or below. . tissue should be homogenized at ph . to . using a sonicator or nonmetallic homogenizer and then should be stored as indicated above. che activity in blood, serum, and tissues can be measured by a number of methods. one of the most popular is the ph method by michael ( ), whereby a change in ph is measured when che acts on acetylcholine. the principle is that cholinesterase hydrolyzes acetylcholine, thus producing acetic acid, which in turn decreases the ph of the reaction mixture. electrometric determination of the change in ph from . for a definite period of time (e.g., hour) at a specific temperature (e.g., °c) represents the enzyme activity. normal values of che activity as measured by this method (in Δph/hour/ . ml red blood cells or plasma at °c, mean ± standard deviation) are given in table ( ) . in deaths owing to organophosphorus insecticides, the values will be much lower. a % or greater depression of the red blood-cell che level is a true indicator of poisoning. death occurs when levels have decreased by more than %. blood and urine should be preserved for toxicological analysis of che levels. samples from lung, liver, kidney, skeletal muscle, brain, and spinal cord, as well as gastric contents, must similarly be preserved for toxicological analysis of cholinesterase levels ( ) according to the precautions detailed in steps - in section . . . . paranitrophenol is a metabolite of many organophosphates. it is excreted in urine and its presence in urine is characteristic of organophosphorus poisoning. organophosphates usually resist putrefaction and can be detected in the viscera for quite some time after death. wehr ( ) studied five exhumations where the decedents were suspected having been poisoned with parathion. he could detect the degradation products of parathion (aminoparathion and p-nitrophenol) up to years after burial, but after years, neither parathion nor any of its degradation products were detectable. pohlmann and schwerd found evidence of parathion in a corpse exhumed after months ( ) . more recently, karger and co-workers ( ) described a case where they detected paraoxon, the main conversion product of parathion, from the abdominal cavity of a -month-old boy, months after his death. his mother had poisoned him with parathion; her deed was detected when, several months later, her second child-a -year-old girl-also suffered the same fate and parathion was detected in her blood. carbamates (fig. ) are derivatives of carbamic acid. their structure is similar to that of organophosphates (fig. ) . the first recognized anti-che was in fact a carbamate, physostigmine (also called eserine), obtained in pure form in by jobst and hesse from the calabar bean ( ) . some common carbamates used as insecticides today are aldicarb, carbaryl, γ-benzene hexachloride, triallate, propoxur, methomyl, carbofuran, and carbendazim. like organophosphates, carbamates are inhibitors of ache, but instead of phosphorylating, they carbamoylate the serine moiety at the active site. this is a reversible type of binding, and therefore, their toxicity is less severe and of lesser duration ( ) . because they do not penetrate the cns to any great extent, the cns toxicity of carbamates is relatively low. signs and symptoms are the same as those seen in poisoning with organophosphates/organophosphorus insecticides but they are milder in nature. convulsions are not seen in carbamate poisoning. postmortem findings in carbamate poisonings are mostly similar to those found in organophosphates. a bluish discoloration of the mucosa of the mouth and stomach is not seen because the blue green dye indigocarmine is usually not mixed with carbamates. determination of cholinesterase levels is not of much help because these are restored very rapidly in carbamate poisoning. organochlorine pesticides are nonselective insecticides. they are cyclic in nature, have molecular weights between and d, are cns stimulants, and have limited volatility. they are poorly soluble in water but readily soluble in organic solvents and fats, which is the way how they accumulate in the human body. they are very stable, both in the environment and in the body tissues, and can be demonstrated in the bodies of most people born since . based on their chemical structures, organochlorines can be divided into four categories ( fig. ) ( ): (a) dichlorodiphenyltrichloroethane (ddt) and related analogs, such as methoxychlor, (b) hexachlorocyclohexane or lindane, (c) cyclodienes and related compounds (e.g., aldrin, dieldrin, endrin, endosulfan, chlordane, chlordecone, heptachlor, mirex, isobenzan), and (d) toxaphene and related compounds. the best known organochlorine, ddt, was synthesized by the german chemist othmar zeidler in , but he failed to realize its value as an insecticide. it was the swiss paul hermann müller ( - ) who recognized its potential as an effective insecticide. in , ddt was tested successfully against the colorado potato-beetle by the swiss government. the united states department of agriculture used it successfully in . in january , ddt was used to quash an outbreak of typhus carried by lice in naples, italy; this was the first time a winter typhus epidemic could be stopped. so revolutionary was his work that müller was awarded with the nobel prize in medicine in . it is ironic that just years later, in , ddt was banned in the united states. it is perhaps a unique example in the history of science that a nobel prize-winning work was banned within such a short period of time. the main driving force behind this ban was the ecologists' concerns about the persistence of ddt in the environment and its resulting harm to the habitat-humans are equally affected by persistent ddt in the environment. it was rachel carson's book silent spring, published in , which brought the problem to everyone's notice. endrin, one of the cyclodienes, is chiefly used against insect pests of cotton, paddy, sugarcane, and tobacco. it is active against a wide variety of insect pests, and hence is commonly known as plant penicillin. it has been banned in most western countries, but unfortunately continues to be used in several agrarian economies. the mechanism of action of organochlorines is entirely different from that of organophosphates and carbamates. organochlorines act on axonal membranes affecting the sodium channels and sodium conductance across the neuronal membranes. organochlorines also alter the metabolism of acetylcholine, noradrenaline, and serotonin. lindane and cyclodienes appear to inhibit the γaminobutyric acid-mediated chloride channels in the cns. therefore, not very surprisingly, the main symptoms induced by poisoning with organochlorines are cns-related and include vertigo, confusion, weakness, agitation, hyperesthesia or paresthesia of the mouth and face, myoclonus, rapid and dysrhythmic eye movements, and mydriasis (in contrast to organophosphates and carbamates, where miosis is found). other symptoms include nausea, vomiting, fever, aspiration pneumonitis, and renal failure. the fatal dose of ddt and lindane is to g, whereas that of aldrin, dieldrin, and endrin is to g ( ). the conjunctivae are congested and the pupils are dilated. there may be a kerosene-like smell emanating from the mouth and nostrils. this is because most organochlorines are poorly soluble in water and are dispensed as solutions in organic solvents that may have a kerosene-like smell. fine white froth, which may or may not appear hemorrhagic, can be seen around the mouth and nostrils; this is a general effect of pulmonary edema coupled with respiratory distress and therefore, signs of cyanosis are seen on the face, ears, nail beds, etc. the mucosa of the respiratory tract appears congested and the respiratory passages contain frothy mucus which may or may not be tinged with blood. subpleural and subpericardial petechial hemorrhages are common. the lungs appear large and bulky, showing pulmonary edema. the mucosa of the esophagus, stomach, and bowel is congested owing to the irritating effect of organochlorines on the gastrointestinal tract. the stomach contents smell kerosene-like. the visceral organs are congested. hepatic necrosis may be found on cut sections of the liver. in animals killed by ddt, vacuolization around large nerve cells of the cns, fatty change of the myocardium, and renal tubular degeneration can be detected histologically ( ). feces, urine, and subcuatenous adipose tissue (placed in a glass-stoppered vial or a vial with a teflon-lined cap [ ] ) should be collected for toxicological analysis. samples must be frozen before onward transmission to the toxicology laboratoy. nicotine salts, such as nicotine sulfate, were very popular pesticides in the s and s. these compounds generally contained % nicotine (fig. ). now, because most countries have banned nicotine-based insecticides, less than % of home garden insecticides are nicotine-based. these are usually available in powder form. main among these is black leaf- (manufactured by black leaf products company, elgin, il). when nicotine-based insecticides come in contact with moist skin, fatal doses of nicotine may be absorbed through the skin ( ) . apart from occupational exposure to nicotine spray, other methods of fatal exposure include careless storage and inadvertent mixing with foodstuffs, fruits, and vegetables. these insecticides have also been used successfully with suicidal or homicidal intention. brownish froth around the mouth and nostrils is a frequent finding in nicotine poisoning. there is a characteristic odor of stale tobacco emanating from the gastric contents. the esophageal and gastric mucosa is intensely congested, showing a brownish discoloration. liver and kidneys show considerable acute congestion ( ). the liver shows plaque-like granulations in the cytoplasm of centrilobular and intermediary hepatocytes. intrapulmonary hemorrhages and pulmonary edema are typical and there often is detachment of the alveolar epithelium. in the kidneys, there is necrosis and detachment of the epithelia in the straight and convoluted renal tubules. a variety of arterial wall lesions, including lacerations of the elastic interna, are seen that have been connected with extreme fluctuations in blood pressure from the effects of nicotine ( ). an estimated % of all plant species are weeds, with a total of some , species. chemicals, such as common salt, have been used for centuries for weed control. the era of chemical weed control is generally recognized as starting in . bonnet in france found that the bordeaux mixture, already being used on vines to control powdery mildew, also provided control of specific weeds. by the s, farmers were still using simple chemicals for this purpose; for example, copper sulfate (blue vitriol), which was first used for weed control in , was still in use at this time. in the early th century, scientists in europe started using the salts of heavy metals to control weeds but when this was attempted in the united states, the low humidity in the western states prevented these chemicals from being absorbed by the weeds. other chemicals were tried, but most of them had drawbacks. for instance, carbon bisulfide used to control thistles and bindweeds smelled like rotten eggs and was, therefore, quite understandably unpopular. most chemical weed killers of those times (such as sodium arsenate, arsenic trioxide, and sulfuric acid) were highly toxic to humans and had to be used in large quantities (several kilograms per hectare), which was another serious drawback. the first synthetic organic chemical for selective weed control was introduced in . its chemical name was -methyl- , -dinitrophenol, and it could control some broadleaf weeds and grasses in large seeded crops, such as beans. more modern herbicides are now available. these have to be sprinkled in very low doses (grams per hectare) in order to kill weeds and the crop is spared. herbicides are categorized as selective when they are used to kill weeds without harming the crop and as nonselective when the purpose is to kill all vegetation. killing of all vegetation is generally not intended in an agricultural setting. it is required more often in places such as recreational areas, railroad embankments, irrigation canals, fence lines, industrial sites, roadsides, and ditches. both selective and nonselective herbicides can be applied to weed foliage or to soil containing weed seeds and seedlings depending on the mode of action. the term true selectivity refers to the capacity of an herbicide, when applied at the proper dosage and time, to be active only against certain species of plants but not against others. selectivity can also be achieved by placement, such as when a nonselective herbicide is applied in such a way that it reaches only the weeds but not the crop. herbicides can also be classified as contact or translocated. contact herbicides kill the plant parts to which the chemical is applied. translocated herbicides are absorbed either by the roots or the above-ground parts of plants and are then circulated within the plant system to distant parts. timing of herbicide application regarding the stage of crop or weed development forms another basis of classification. a preplanting herbicide is sprinkled on the farm before the planting of the crop. a preemergence herbicide is sprinkled after planting but before emergence of the crop or weeds. finally, a postemergence herbicide is used after the emergence of the crop or weed. herbicides can be applied to weeds in a number of ways. a band application treats a continuous strip, such as along or in a crop row. broadcast application covers the entire area, including the crop. spot treatments are confined to small areas of weeds. directed sprays are applied to selected weeds or to the soil to avoid contact with the crop. in the more recent overthe-top-application, herbicides are applied "over the top" of the crop and weeds shortly after germination. the crops in these instances are naturally tolerant to the specific herbicide or have been genetically engineered to be tolerant to the herbicide used. from a toxicological point of view, the following herbicides are the most important. dipyridyl weed killers include paraquat, piquat, and morfamquat ( fig. ). paraquat is the most important of these three. paraquat ( , ′dimethyl- - ′bipyridylium dichloride) is an important agricultural chemical from a toxicological viewpoint. out of the deaths caused by pesticides reported by the aapcc annual report ( ) , two were the result of paraquat poisoning. paraquat was first synthesized in , but its herbicide activity was discovered very late. its use as an herbicide was first reported in , and paraquat was introduced commercially as a nonselective herbicide in . the introduction of paraquat caused an agricultural revolution because it has some unique properties. it can be sprayed from the ground level or the air and is totally denatured when it comes in contact with the earth. thus, it cannot harm the seeds or young plants that will be placed in the same ground a short time later. indeed, the crop can be planted within days, if not hours, after herbicidal treatment with paraquat. an additional advantage is that plowing is unnecessary aggrawal in many cases with much less soil erosion. paraquat is therefore of immense value in an economic sense ( ) . in countries like sri lanka, its use has resulted in three crops, instead of two, per year being taken off the same field ( ) . paraquat is highly soluble in water and is marketed most commonly as a concentrate containing g paraquat dichloride per liter ( % wt/vol); this is an odorless brown liquid. a "stenching" agent (a pyridine derivative) is added to prevent accidental or criminal poisoning; a bluish-greenish dye is also added for the same reason, and an emetic may be added as well. paraquat is sometimes sold in combination as a mixture with diquat and other herbicides. the liquid concentrate is known as gramoxone (not to be confused with gammexane, which is the trade name for lindane); a weaker, granulated preparation for horticultural use, known as weedol, is also available ( % wt/vol). the solution may be decanted in soda bottles and left unlabelled. because it looks like a cola drink, accidental ingestion may occur. it may be mistaken for vinegar as well; one patient is reported to have sprinkled it on his french fries. wesseling and co-workers ( ) reported that paraquat is the pesticide most frequently associated with injuries among banana workers in costa rica; the injuries involve mostly the skin and eyes. although most fatalities caused by paraquat occur from ingestion, absorption through the skin can also cause fatalities. wohlfahrt ( ) reviewed paraquat poisoning in papua new guinea from to and found that out of fatalities caused by paraquat, six were the result of transdermal absorption. diquat ( , ′-ethylene- , ′-dipyridylium dibromide) is less commonly used than paraquat. it has the same indications and mode of action as paraquat. diquat is, however, used additionally for the control of aquatic weeds. jones and vale ( ) compiled all cases of diquat poisoning published between the years and and found that only cases were reported in detail in the literature, of which ( %) were fatal. conning et al. showed that out of the three dipyridyl weed killers, it was only diquat that produced bilateral cataracts ( ) . diquat was introduced in as a fast-knockdown, contact herbicide and plant desiccant. diquat-only formulations manufactured by syngenta (formerly imperial chemical industries) or its subsidiaries do not contain the dye, "stenching" agent, or emetic added to paraquat ( ). the symptoms include intense pain in the mouth and pharynx, with inflammation and even ulceration of the oral mucosa. esophageal ulceration may lead to perforation with all its attendant risks. renal and hepatic failure develop within to days. the most important effect is on the lungs (pneumotropism), where massive, irreversible pulmonary fibrosis is seen. pulmonary fibrosis is thought to be the result of an increase in the pulmonary concentrations of prolyl hydroxylase, an enzyme which promotes collagen formation. paraquat is one of the few poisons that may produce necrosis of the adrenal glands, possibly leading to hypotension. the fatal dose is to g (about a mouthful of gramoxone). subcutaneous injection of just ml of gramoxone has shown to be fatal ( ) , with death occuring after to weeks as a result of respiratory failure caused by pulmonary fibrosis; greater doses can kill a human within hours. why does paraquat show such remarkable pneumotropism? it has been postulated that inside the pneumocytes, the paraquat dication pq + accepts one electron from reduced nicotinamide adenine dinucleotide and becomes the monocation pq + . (pyridinyl-free radical) (fig. ). the monocation pq + . is unable to cause any injury on its own, but in the presence of molecular oxygen (o ) in the lungs, it is oxidized once again to its dication form (pq + ). in this process, it passes on its electron to the molecular oxygen (o ), which, in turn, becomes the superoxide anion radical (o -. ). this process, known as redox cycling, is sustained by oxygen in the lungs. the superoxide anion radical o -. (reactive oxygen species) generated as a result of this cycle is responsible for cell death. this also explains why oxygen enhances the toxicity of paraquat and should never be administered during paraquat intoxication; by administering oxygen, one is supplying the "raw material" for the formation of the damaging superoxide radical. formation of free radicals is implicated in injuries caused by at least two other poisons-myocardial injury caused by doxorubicin and liver injury by carbon tetrachloride. the related bipyridylium compounds, such as diquat and morfamquat, do not affect the lung as seriously, but rather cause liver damage ( ). there is ulceration around lips and mouth, although it is not as bad as is seen after ingestion of inorganic acids, such as nitric or sulfuric acids. the oral and esophageal mucosa is reddened and desquamated. a unique feature of paraquat ingestion is the formation of pseudomembranes in the pharynx resembling to that seen in diphtheria ( ). patchy hemorrhages in the stomach mucosa are a frequent finding. the liver is pale, showing fatty changes. the kidneys may exhibit pallor of the cortex. the most striking findings are found in the lungs. both type and type alveolar epithelial cells accumulate paraquat and are thereby destroyed. this destruction is followed by inflammatory cell infiltration and hemorrhages; fibroblast proliferation then leads to fibrosis and impaired gas exchange. the lungs are congested, appear stiffened, and retain their shape during evisceration. each lung is typically approx g or more in weight. teare ( ) reported a case of paraquat poisoning (a -year-old man dying of suicidal ingestion of paraquat after days of illness), with the left lung weighing g and the right lung weighing g. blood-stained pleural effusions and fibrinous pleurisy are other typical autopsy findings. cut surfaces of the lungs reveal edema and fibrosis. subendocardial hemorrhages may accompany the aforementioned pathological findings. the pathological features of paraquat poisoning have been reviewed in detail by vadnay and haraszti ( ) . at the beginning of the toxic process, severe degenerative changes appear in the pneumonocytes with fatty infiltration, desquamation, necrosis, and detachment ( ) . later, there is splintering of the basement membranes, fragmentation, aneurysma formation, and multiple ruptures. fibrinous edematous fluid is seen in the interstitium and within alveoli and hyaline membranes can be observed. there is a large-scale dissolution of the pulmonary structure. there may be active proliferation of the bronchial epithelium, forming small adenomata within the pulmonary parenchyma. marked proliferation of fibroblasts with an increase in macrophages in the alveoli (these two mechanisms obliterate the alveolar spaces) can be seen. acute tubular necrosis is a frequent finding in the kidneys. extensive renal cortical necrosis is also seen at times. in the liver, centrilobular hepatic necrosis, cholestasis, and giant mitochondria with paracrystalline inclusion bodies can be detected ( ) . in the myocardium, there is edematous disaggregation of the sarcoplasm and sporadic fragmentation of the myofibrils. paraquat-type herbicides in aqueous solutions have traditionally been determined by colorimetric methods. these involve measurement of the complex formed with some chemical (α-dipicrylamine hexanitrodiphenylmethane). plasma paraquat levels can be assayed by spectroscopy, high-performance liquid chromatography ( ) or radioimmunoassays; levels greater than . μg/ml confirm death by paraquat intoxication. urine paraquat levels can be deter-mined using spectrophotometry, too; levels greater than μg/ml confirm death by paraquat intoxication ( ) . berry and grove introduced an ion exchange and colorimetric method in for the determination of paraquat in urine ( ) . diquat (reglone) is selectively concentrated in the kidneys and causes marked renal tubular damage. in a case of fatal diquat poisoning, mccarthy et al. found esophagitis, tracheitis, gastritis, and ileitis ( ) . autopsy findings and toxicokinetic data in diquat poisoning have been described in detail by hantson et al. ( ) . morfamquat is used far less commonly than the other two bipyridyls, paraquat and diquat. conning et al. have shown that rats that fed on morfamquat developed renal damage ( ). chlorophenoxy herbicides (fig. ) are growth regulators or auxins. they cause abnormal plant growth, thereby ultimately destroying the plant. chlorophenoxy herbicides are commonly used for control of broadleaf weeds in cereal crops and pastures ( ). - dichlorophenoxyacetic acid ( , -d; trimec) has been and continues to be one of the most useful herbicides developed; it is frequently applied to lawns to control broadleaf weeds and is often found in fertilizer products along with other phenoxy herbicides, such as dicamba, mecoprop, and ( -chloro- -methylphenoxy)acetic acid. , -d is easily absorbed through the skin and lungs ( ). on ingestion, , -d causes peripheral neuropathy, muscle weakness, cheyne-stokes respirations, hyperthermia, acidemia, and coma ( ) . the patient is hypotonic, hyporeflexive, hypotensive, and comatose ( ) , and nasogastric aspirate may be guaiac-positive ( ). , -d earned a notorious reputation during the vietnam war as an ingredient of agent orange sprinkled by united states troops over vietnam (see subheading . ). suicidal ingestions of , -d are occasionally reported ( , ) . postmortem findings in deaths caused by chlorophenoxy herbicides are nonspecific. the gastrointestinal mucosa may be intensely congested and/or hemorrhagic. all internal organs are usually congested. confirmatory tests of suspected poisonings with chlorophenoxy herbicides are the demonstration of these herbicides in plasma and urine,which can be detected by radioimmunoassay ( ) and gas liquid chromatography ( ). this category comprises mainly dinitrophenol (dnp), dinitro-orthocresol (dnoc), and pentachlorophenol ( ) . these substances are used in agriculture mainly as selective weed killers for cereal crops. the effects of dnp in stimulating metabolism have been known since , and dnp was used at one time for "slimming." dnp (fig. ) is a potent "uncoupler" of oxidative phosphorylation, causing the energy obtained from the oxidation of nicotinamide adenine dinucleotide and reduction of o to be released as heat. it has been demonstrated that these compounds are dangerous to humans and thus, they are no longer used for medicinal purposes. the principal risk of poisoning is in the agricultural use of concentrated solutions for spraying crops aggrawal (as weed killers). dinitrophenol (dnp) is also used in agriculture for the control of mites and aphids ( ) . absorption occurs by inhalation and thus, breathing apparatus are a must for those who are exposed to this poison. absorption also occurs by ingestion and through the skin. excretion of dnp is extremely slow, so the poison accumulates in the body gradually. the symptoms are fatigue, insomnia, restlessness, excessive sweating, weight loss, and thirst. clinical signs include tachycardia, increase in the rate and depth of respiration, rise in temperature (up to °c and higher) and some yellow discoloration of the sclera. in severe cases, body temperature may keep rising and just before death, it may reach °c. when death occurs, the onset of rigor mortis is rapid. sodium chlorate is a nonselective herbicide. it acts as a soil sterilant at rates of lbs/acre. it is also used as a foliar spray at lbs/acre as a cotton defoliant. it was once avidly advocated as a weed killer, not only because it is effective, but also because it was considered safe. this fallacy was so prevalent that containers of sodium chlorate used to be marked as "nonpoisonous." however, chlorates cause methemoglobinemia. severe hemolysis is a constant clinical feature in sodium chlorate poisoning, with presence of heinz bodies in the red blood cells. acute renal failure and anuria sets in later. anuria occurs because of (a) a direct damaging action of chlorates on the renal tubular epithelium, and (b) mechanical obstruction of the renal tubules by the hemoglobin set free by hemolysis. the fatal dose of sodium chlorate is to g with death occuring within to days. poisoning with sodium chlorate can occur accidentally, suicidally, or even homicidally. accidental poisoning is probably the most common. a -year-old gardner was severely poisoned in a curious way. he was using a concentrated solution of sodium chlorate in an atomizer while a strong wind was blowing. consequently, spray was blown onto his face and he inhaled and ingested some of the solution. symptoms of poisoning started the same evening. he was saved with some heroic effort on the part of the doctors, yet he could only return to full-time work after about year ( ). the skin has a distinctive chocolate-brown color. blood smears may show evidence of hemolysis and heinz bodies. the kidneys are enlarged and their principal change is a brown streaking of the cortex; microscopical examination reveals acute renal tubular degeneration with blockage of tubules by broken red blood cells and brown pigment granules (released hemoglobin owing to hemolysis). glyphosate is an important agricultural chemical from the toxicological viewpoint. out of the deaths caused by pesticides reported by the aapcc annual report ( ) , one was caused by glyphosate. glyphosate is a broad-spectrum, nonselective, systemic herbicide used for control of annual and perennial plants including grasses, sedges, broad-leaved weeds, and woody plants. it can be used on non-cropland as well as on a great variety of crops. although glyphosate itself is relatively harmless, its chemical formulations (e.g., roundup ® , rodeo ® , touchdown ® , gallup ® , landmaster ® , pondmaster ® , ranger ® ) have been used successfully for committing suicide. this is because glyphosate invariably is formulated in a surfactant (polyethoxylated tallow amine), which is quite toxic ( , ) . glyphosate is generally distributed as water-soluble concentrates and powders. mild poisoning results only in gastrointestinal symptoms, such as vomiting, abdominal pain, diarrhea, and nausea, which usually resolve within a day or two. severe poisoning results in intestinal hemorrhage and ulceration, acid base disturbances, renal failure, hypotension, cardiac arrest, pulmonary dysfunction, convulsions, coma, and death. postmortem findings are nonspecific. glyphosate and the concomitant surfactant are demonstrated by toxicological analysis in the gastric contents and other visceral organs. glyphosate levels of mg/ml or more can be detected postmortem in blood, liver, and urine in less than a minute by using p nuclear magnetic resonance ( ). among the several arsenical herbicides available are cacodylic acid, calcium hydrogen methylarsonate, disodium methylarsonate, hexaflurate (asf k), methylarsonic acid, monoammonium methylarsonate, monosodium methylarsonate, potassium arsenite, and sodium arsenite. cacodylic acid (fig. ) is also known as dimethylarsinic acid. cacodylic acid is a white crystalline substance, readily soluble in water and alcohol, and is still used as an herbicide. when it unites with metals and organic substances, it forms salts known as cacodylates. cacodylic acid contains . % of arsenic. fungicides, or antimycotics, are toxic substances used to kill or inhibit the growth of fungi that cause economic damage to crop or ornamental plants. most fungicides are applied as sprays or dusts. seed fungicides are applied as a protective covering before germination. systemic fungicides, or chemotherapeutants, are applied to plants, where they become distributed throughout the tissue and act to eradicate existing disease or to protect against possible disease. bordeaux mixture (cuso cu[oh] caso ) was one of the earliest fungicides to be used ( ) . bordeaux mixture is a liquid composed of hydrated (slaked) lime, copper sulfate, and water. it was accidentally discovered in in the modoc region of france, where farmers, tired of schoolboys pilfering their grapes, sprayed their grapevines with a poisonous-looking mixture of lime and copper sulphate; it was a desperate idea meant just to deter schoolboys from stealing their grapes. however, in , pma millardet from the university of bordeaux observed that the very same mixture effectively controlled the downy mildew of grapes as well. burgundy mixture is a mixture of copper sulfate and disodium carbonate. both bordeaux mixture and burgundy mixture are still widely used to treat orchard trees. copper compounds and sulfur have been used on plants separately and together. synthetic organic compounds are now more widely used because they give protection and control over many types of fungi. cadmium chloride and cadmium succinate are used to control turfgrass diseases. mercury(ii)chloride, or corrosive sublimate, is used as a dip to treat bulbs and tubers. mercury salts used as fungicides include mercurous chloride, mercuric chloride, mercuric oxide, phenylmercury nitrate (fig. ) , tolylmercury acetate, and ethylmercury bromide. organophosphorus fungicides include ampropylfos, ditalimfos, edifenphos, and fosetyl (fig. ) . carbamate fungicides include benthiavalicarb, furophanate, iprovalicarb, and propamocarb (fig. ) ; the toxicity of organophosphates and carbamates has been dealt with earlier. among the most important inorganic fungicides are potassium azide, potassium thiocyanate, sodium azide, and sulfur. other substances occasionally used to kill fungi include chloropicrin, methyl bromide, and formaldehyde. many antifungal substances occur naturally in plant tissues. creosote, obtained from wood tar or coal tar, is used to prevent dry rot in wood. the most important fungicides-from the toxicological viewpoint-aside from organophosphorus and carbamates, are sodium azide and compounds of copper and mercury. copper compounds are also especially important because they are used in agriculture as insecticides and algicides. somerville discussed the metabolism of several fungicides including maneb, mancozeb, zineb, captan, chlorothalonil, benomyl, triadimefon, triadimenol, and cymoxanil ( ). sodium azide is important because it is a potential intentional or accidental poison. aside from being used in agriculture, sodium azide is also used widely in hospitals where it is used as a component chemical in the fluid used to dilute blood samples. sodium azide, like dnp, is an "uncoupler" of oxidative phosphorylation; it also inhibits the enzymes catalase and cytochrome oxidase. ingestion of sodium azide results in nausea, vomiting, diarrhoea, hypotension, and cns symptoms, such as headache, hyporeflexia, seizures, and coma. postmortem findings include edema of the brain and lungs. edema of the myocardium with myocardial necrosis has also been reported ( ) . fig. . fosetyl, an organophosphate fungicide. salts of copper, although mostly used as fungicides, are used for a large number of other purposes in agriculture as well. copper acetate, copper carbonate, cupric -quinolinoxide, copper silicate, and copper zinc chromate are used as fungicidal agents only; copper arsenate is used as insecticide and copper sulfate as algicide, fungicide, herbicide, and molluscicide; copper acetoarsenite is employed as insecticide and molluscicide; copper hydroxide is used as bactericide and fungicide; copper naphthenate is used as fungicide and mammal repellent; copper oleate as fungicide and insecticide; and copper oxychloride as bird repellent and fungicide. chronic exposure to bordeaux mixture in vineyard sprayers causes the socalled "vineyard sprayer's lung." observed mainly in portugal, the disorder includes pulmonary fibrosis ( ) and may lead to lung cancer ( , ) . bordeaux mixture is the only other significant pesticide aside from paraquat that induces significant pulmonary fibrosis with organophosphates coming in a distant third ( ) . the radiological picture in vineyard sprayer's lung resembles that of silicosis with micronodular features in the early stages of the disease ( ) . only in later stages does a picture of massive fibrosis emerge with continuing development of respiratory insufficiency. plamenac et al. ( ) examined the sputum of rural workers engaged for years in spraying of vines. sputum specimens were tested for copper by rubeanic acid. macrophages containing copper granules in their cytoplasm were found in % of the workers engaged in vine spraying compared with none in a control group. other abnormalities, such as eosinophils, respiratory spirals, respiratory cell atypia, and squamous metaplasia, were also found in the sputum. atypical squamous metaplasia was observed in % of vineyard workers who were also smokers ( ). eckert et al. ( ) exposed mice to copper sulfate aerosol for a longer period of time and were able to replicate these changes in the animals' lungs. the authors concluded that the changes seen in vineyard sprayer's lung are a result of copper sulfate toxicity. pimentel and menezes studied the liver of vineyard sprayers by percutaneous biopsy and also at autopsy ( ) . they found histiocytic and noncaseating granulomas containing inclusions of copper as identified by histochemical techniques. they also found that the affected individuals were prone to liver fibrosis, cirrhosis, angiosarcoma, and portal hypertension ( ) . copper sulfate is a popular suicidal poison in india ( ) and copper sulfate was once a very popular homicidal poison ( ) . although no reports of suicide and homicide with bordeaux mixture exist, this is certainly possible. quite possibly such cases did, and still do, occur but have never been reported. mercury is widely used as a fungicide in agriculture. both inorganic and organic salts are used. inorganic mercury fungicides being used as fungicides include mercuric chloride, mercuric oxide, and mercurous chloride. organomercury fungicides include ( -ethoxypropyl)mercury bromide, ethylmercury acetate, ethylmercury bromide, ethylmercury chloride, ethylmercury , -dihydroxypropyl mercaptide, ethylmercury phosphate, n-(ethylmercury)-ptoluenesulphonanilide (fig. ) , hydrargaphen, -methoxyethylmercury chloride, methylmercury benzoate, methylmercury dicyandiamide, methylmercury pentachlorophenoxide, -phenylmercurioxyquinoline, phenylmercuriurea, phenylmercury acetate, phenylmercury chloride, phenylmercury derivative of pyrocatechol (fig. ) , phenylmercury nitrate, phenylmercury salicylate, thiomersal (fig. ) , and tolylmercury acetate. the ingestion of wheat and barley seed treated with methyl mercury fungicides for sowing by a largely illiterate population in iraq led to a major poisoning with mercury in to with a high fatality rate ( ) . the seed-about , tons of it-was intended for spring planting; there had been ample warning that the seed was unfit for consumption, but this warning was disregarded. there was a latent period of several weeks after which pares- thesias began to appear in several victims. paresthesias involved lips, nose, and distal extremities. more serious cases progressed to ataxia, hyperreflexia, hearing disturbances, movement disorders, salivation, dementia, dysarthria, visual field constriction, and blindness. in the most severe cases, individuals remained in a mute rigid posture altered only by spontaneous crying, primitive reflexive movements, or feeding efforts. there were victims with deaths ( ) ( ) ( ) ( ) . seven children remained permanently incapacitated both physically and mentally. this was the second major mercury disaster after the minamata bay disaster in japan occurring between and , when about people were poisoned and died ( ) . phenylmercury acetate has been found to be embryotoxic and teratogenic ( ). in deaths caused by acute mercury poisoning, the mucosa of the mouth, throat, esophagus and stomach is greyish in color showing superficial hemorrhagic erosions; a softened appearance of the stomach wall is characteristic. in cases where the patient survived a few days, the large bowel may show ulcerations. the kidneys appear pale and swollen owing to edema of the renal cortex. microscopically, the kidneys usually demonstrate necrosis of the renal tubules ( ). sperhake et al. ( ) reported the case of a -year-old chemist who died of mercury poisoning. an autopsy carried out hours postmortem revealed unspecific signs of intoxication including severe edema of the lungs and brain, dilatation of the bowel, and marked congestion of the parenchymatous organs. the stomach contained ml of a reddish fluid. between the gastric folds, the mucosa appeared highly preserved with a brownish discoloration, but streaklike erosions in the exposed parts. the mucosal surface of the oral cavity and esophagus also appeared brownish and discolored. histologically, the pre-served areas of the gastric mucosa were totally unaffected by autolysis with an intact epithelial layer, whereas the eroded areas showed loss of mucosal lining with infiltrates of polymorphonuclear granulocytes and lymphocytes. mercury was detected in the epithelial layer of the gastric mucosa in situ using , diphenylcarbazone staining ( . % in % ethanol). tubular necrosis was present in the kidneys. a case of chronic arsenic poisoning in a -year-old man has been described; the man used a sodium arsenite-based fungicide for cultivating his vine yard ( ). methyl bromide (ch br), also known as bromomethane, monobromomethane, embafume, or iscobrome, is mainly used as a gas soil fumigant against insects, termites, rodents, weeds, nematodes, and soil-borne diseases ( , ) . it has been used to fumigate agricultural commodities, mills, grain elevators, ships, furniture, clothes, and greenhouses. its main advantages are its effective penetrating power and absence of danger of fire or explosion hazards. methyl bromide acts rapidly, controlling insects in less than hours in space fumigations, and it has a wide spectrum of activity, controlling not only insects but also nematodes and plant-pathogenic microbes ( ) . about % of methyl bromide produced in the united states goes into pesticidal formulations. pure methyl bromide is a colorless gas that is heavier than air. odorless and tasteless in low concentrations, it has a musty, acrid smell in high concentrations. occupational exposure to methyl bromide also occurs frequently. it is estimated that about , american workers are occupationally exposed to this gas annually. its toxicity is severe and, despite safeguards, cases of acute and chronic intoxication occur, mainly in the fruit and tobacco industries. the maximum allowable concentration of methyl bromide is ppm. concentrations of ppm or less are considered safe. death has been reported to occur at ppm ( ) . methyl bromide can enter homes through open sewage connections, thus causing fatalities. lagard et al. ( ) reported an interesting case of methyl bromide poisoning where methyl bromide caused toxicity in this manner. the sewage pipes serving two houses (one house was fumigated and in the other the poisoning occurred) had been sucked empty only to hours prior to the start of fumigation. because it depletes ozone into the atmosphere ( ) , methyl bromide has been banned in several industrialized countries, except for exceptional quarantine purposes. phosphine, sulfuryl fluoride (see subheading . .) , and carbonyl sulfide are considered viable alternatives. the mucosa of trachea and bronchi is congested and shows petechial hemorrhages. the lungs show subpleural hemorrhages and pulmonary edema. bilateral bronchopneumonia may also be present. the brain is edematous with necrosis of cortical cells, especially in the frontal and parietal lobes. multiple perivascular hemorrhages may be detected throughout the brain and small subarachnoid hemorrhages may be seen in some cases. circumscribed hemorrhages may also be present in stomach, duodenum, myocardium, spleen, and retina. the kidneys are acutely congested and show tubular necrosis on the micromorphological level; the proximal tubules are most commonly affected. in severe cases, the loops of henle and the distal tubules are also affected. the liver is also congested, but liver cell necrosis is not a common feature ( ) . methyl bromide can be detected and quantitatively determined in various biological samples by headspace gas chromatography ( ). sulfuryl fluoride (f o s) is an important agricultural fumigant. according to the annual report of the aapcc ( ), the only death that occurred as a result of fumigants was caused by sulfuryl fluoride (fig. ) . it is an inorganic gas fumigant used in structures, vehicles, and wood products for control of drywood termites, wood-infesting beetles, and certain other insects and rodents. it is also used as a gas fumigant for postharvest use in dry fruits, tree nuts, and cereal grains. it is available under the trade name vikane™ gas fumigant. because methyl bromide has now been graded as an ozone-depleting substance and is being gradually phased out, sulfuryl fluoride is taking its place. because sulfuryl fluoride is an inorganic material, as opposed to the organic methyl bromide, it does not bind onto items being protected and therefore, less quantities of gas are required for the same insecticidal effect. sulfuryl fluoride is a colorless and odorless gas. it does not cause tears or immediately noticeable eye irritation and lacks any other warning property. chloropicrin is added to products containing sulfuryl fluoride to serve as a warning indicator; chloropicrin is a gas that causes eye and respiratory irritation and vomiting. sulfuryl fluoride acts as a cns depressant. symptoms of poisoning include itching, numbness, depression, slowed gait, slurred speech, nausea, vomiting, stomach pain, drunkenness, twitching, and seizures. inhalation of high concentrations may cause respiratory tract irritation and respiratory failure. skin contact with sulfuryl fluoride normally poses no hazard, but contact with liquid sulfuryl fluoride can cause pain and frostbite-like lesions owing to rapid vaporization. occupational sulfuryl fluoride exposure may be associated with subclinical effects on the cns, including effects on olfactory and some cognitive functions ( ) . the oral ld for sulfuryl fluoride in rats and guinea pigs is mg/kg. scheuerman has reported two cases of suicide by sulfuryl fluoride ( ). according to scheuerman, toxicological analysis should include a plasma and urine fluoride level because the toxic effects of sulfuryl fluoride are probably related to this ion. concentrations of fluoride in his cases were and . mg/l, respectively. however, all values have to be interpreted in the light of all information available (kind and length of exposure, symptoms, autopsy findings, etc.) in a given case. aluminum phosphide (alp) is an ideal grain preservative for a number of reasons. it is highly toxic to almost all stages of insects with remarkable penetration power. alp dissolves well in water, oil, and fat. it is considered an ideal seed fumigant since the seeds' viability is not affected and is practically free from residual toxic hazards-provided the seeds have less than % water content. alp is minimally absorbed and easily desorbed from the treated commod- ity, such as wheat grains. it is inflammable at the prescribed dosage and devoid of tainting on fumigated stock. it has a distinct odor, which has been described as a fishy odor. because of this and also because of delays in evolving, phoshine provides considerable safety in handling this fumigant. safety in handling is due to both these reasons. because it has an odor, it is difficult for handlers to accidently ingest it. because the tablet generates the predetermined weight of gas, it is very convenient to administer the exact dose. cost of fumigation is low and its effects on the fumigated stock last longer. alp is easy to transport and handle. unfortunately, no specific antidote to alp is known. alp is used very extensively throughout agrarian economies like india. on exposure to moisture it releases the poisonous phosphine, which percolates through the grain: alp+h _ al(oh) +ph . as long as the grain is stored in airtight godowns, the liberated phosphine remains in the environment, repelling all pests. when the grain is to be used, it is brought out and aerated. this releases phosphine, leaving behind virtually no or only nontoxic residues. alp is generally available as tablets (alphos ® , celphos ® , fumigran ® ), which are dark brown or grayish in color, g in weight, and measuring mm in diameter and mm in thickness. they come in an aluminum container containing ten tablets. alp is also available as . -g pellets. the tablets are composed of pure alp (the active ingredient) and ammonium carbamate/carbonate (the inert ingredient). the ratio of the active and inert ingredient is generally about : . on contact with moisture, each -g tablet evolves about g of phosphine along with carbon dioxide and ammonia, which prevents self-ignition of phosphine gas. this is why it is also called a "protective gas." carbon dioxide and ammonia are liberated by combination of water with other inert ingredients in the tablets. the main function of the inert ingredients is to produce these gases, so phosphine may not ignite easily. the phosphine gas, once liberated, spreads quickly and kills insects and rodents almost in all stages of their development. after complete decomposition of the tablet, alp is left behind as a harmless and nontoxic grayish white residue, which is less than % of the original tablet weight. alp is the leading cause of accidental and suicidal deaths in india ( ) ( ) ( ) ( ) ( ) . it has been implicated in several homicides including dowry deaths (deaths of newlywed brides occurring in relation to dowry and covered under section b of the indian penal code). the mortality rate for poisoning with alp is almost % ( ) . there is an intense garlic-like odor emanating from the mouth and after opening of the stomach at autopsy. all internal organs are congested and show petechial hem-orrhages. pericarditis may be present ( ) . the stomach contents are hemorrhagic and the mucosa shows detachment. residues of alp may be demonstrable in the stomach contents, but rarely can alp itself be detected because it readily reacts with acid and water within the stomach. misra et al. ( ) described eight cases of alp poisoning after ingestion of alp tablets for attempting suicide; the mean age of the patients was years (age range - years). six of the patients died; the mean hospital stay was hours (range - hours). an autopsy was carried out in two patients, revealing pulmonary edema, congestion of the gastrointestinal mucosa, and petechial hemorrhages on the surface of liver and brain. anger and co-workers ( ) reported the case of a -year-old man who committed suicide by ingestion of alp. autopsy revealed signs of asphyxia with marked visceral congestion. the authors also toxicologically analyzed peripheral blood, urine, liver, kidney, adrenal, brain, and cardiac blood. phosphine gas was absent in peripheral blood and urine but present in the brain ( ml/g), the liver ( ml/g), and the kidneys ( ml/g). high levels of phosphorus were found in the blood ( . mg/l) and liver ( . mg/g). aluminum concentrations were highly elevated in peripheral blood ( . mg/l), brain ( μg/g), and liver ( μg/g) compared with the reference values. histopathological findings in alp poisoning have been described in detail by chugh et al. ( ) . various viscera show congestion, edema, and inflammatory cell infiltration. in the myocardium, there are patchy areas of necrosis, whereas the liver shows fatty changes and the lung parenchyma displays gray/red hepatization. the adrenal cortex shows complete lipid depletion, hemorrhage, and necrosis. chugh et al. assumed that the changes in the adrenal cortex could be both a sequel of shock and/or a cellular toxic effect of phosphine. in out of the patients studied by chugh and associates, there was a significant rise in the plasma cortisol level (> nmol/l). in the remaining patients, the adrenal cortex was critically involved and the cortisol level failed to rise beyond normal levels (< nmol/l). pillay ( ) noted that in alp poisoning the heart shows features of toxic myocarditis, necrosis may be seen histologically in both liver and kidneys, and the lungs may demonstrate evidence of adult respiratory distress syndrome (ards). ards has also been reported by chugh et al. ( ) . the dose of the intoxicant in chugh's cases varied from two g) to three tablets (corresponding to and g, respectively). all patients were in shock at admission and developed ards within hours after ingestion of alp. according to these authors, the exhalation of phosphine (which they detected by a positive silver nitrate paper test) was the possible noxious triggering factor in developing ards. in misra at al.'s series ( ) , histopathological changes included pulmonary edema, desquamation of the lining epithelium of the bronchioles, vacuolar degeneration of hepatocytes, dilatation and engorgement of hepatic central veins and sinusoids, as well as hepatocytes showing nuclear fragmentation. in anger's single case ( ) , microscopic examination revealed congestion of inner organs and pulmonary lesions that were attributed to asphyxia. silo filler's disease is another disorder associated with agrochemical poisoning during preservation. corn used for silage is usually grown under conditions of heavy sunlight and drought and its nitrate content is usually very high. when this silage is stored in a silo, the nitrates are fermented into nitrites, which in turn combine with organic acids to form nitrous acid. nitrous acid decomposes into water and a mixture of nitrogen oxides. these are nitric oxide (no), nitrogen dioxide, and dinitrogen tetroxide. the decomposition starts within approx hours of putting the crops into the silo and continues for about days. when entering these silos (which virtually turn into a kind of gas chamber), farm workers may suffer acute poisoning from these gases, and many such deaths have occurred. this type of death in a silo was first described in , but at that time it was wrongly attributed to asphyxia ( ). nos, being relatively poor soluble in water, can reach the terminal bronchioles and even alveoli. within the lungs, the nos react with water to form nitrous and nitric acids, which cause extensive lung damage, resulting in chemical pneumonitis and profuse pulmonary edema. nos trigger histamine release, which causes bronchoconstriction resulting in increased airway resistance. douglas and colleagues ( ) examined patients of silo filler's disease between and . all exposures had occurred in conventional top-unloading silos. acute lung injury occurred in patients, one of whom died. in the fatal case, autopsy findings included early diffuse alveolar damage with hyaline membranes, hemorrhagic pulmonary edema, and acute edema of the airway walls. poisoning with and fatalities owing to fertilizers are rarely encountered but do occur. the annual report of the aapcc toxic exposure surveillance system reported one death caused by fertilizers ( ) ( table ). used as a fertilizer, anhydrous ammonia is a respiratory irritant, which, in high doses, causes pulmonary edema ( ) . exposure most often occurs during transfer operations. ammonia reacts with water to form the strong alkali ammonium hydroxide, which causes severe tracheobronchial and pulmonary inflammation with bronchiolitis obliterans. normally, the peculiar odor of ammonia warns the potential victim. during world war ii, in london, a brewery cellar having ammonia-carrying condenser pipes was temporarily converted into a bomb shelter. during a bombing, a bomb fragment pierced one such pipe resulting in a mortality rate of the affected individuals as high as % ( ) . saito et al. ( ) described the case of a -year-old male who presumably consumed water contaminated with a nitrate fertilizer. on admission to hospital, the man showed drowsiness, deep cyanosis, and dyspnea; the patient died hours later. at autopsy, no particular morphological changes were noted except for the blood being a chocolate-brown color. postmortem toxicology of the blood revealed a methemoglobin concentration of % and the concentrations of nitrate and nitrite were . and . μg/ml, respectively. in deaths caused by nitrate fertilizers, methemoglobinemia and the presence of appreciable quantities of nitrites and nitrates may be demonstrated in cardiac blood and gastric contents (stored at - °c until toxicological analysis) ( ) . capillary gas chromatography-mass spectrometry and capillary gas chromatography with a nitrogen-phosphorus detector can be used to detect nitrates and nitrites in blood. sato and colleagues ( ) described the case of an -year-old woman who supposedly consumed agricultural fertilizer containing ammonium sulfate. she was found lying dead on the ground outside her house. a thorough autopsy could not determine the cause of her death. a beer can was found next to her, and when it was examined, it was found to contain ammonium sulfate. subsequently, ammonium and sulfate ions were detected in her serum samples and gastric contents. the cause of her death was determined as poisoning by ammonium sulfate. in order to further confirm that this death was indeed a result of an ammonium sulfate fertilizer, the authors administered a total dose of mg/kg of ammonium sulfate to three rabbits. the animals developed mydriasis, irregular respiratory rhythms, and local and general convulsions until they came into respiratory failure with cardiac arrest. electroencephalogram showed slow, suppressive waves and a high-amplitude with a slow wave pattern that is generally observed clinically in hyperammonemia in humans and animals. there was a remarkable increase in the concentration of ammonium ions and inorganic sulfate ions in the animals' serum and blood gas analysis showed severe metabolic acidosis. the authors suggested that when the cause of death can not be clearly determined and the previous history is suggestive of ammonium sulfate intake, measurement of ammonium ions, inorganic ions, and electrolytes in blood, as well as in stomach contents, are a prerequisite for the diagnosis. villar and co-workers reported poisoning and death in animals who drank fertilizer-contaminated water ( ) . the water had been hauled in tanks previously contaminated with a nitrogen-based fertilizer. in udaipur, india, chronic fluorotic lesions in cattle and buffalo have been described following consumption of fodder and water contaminated by the fumes and dusts emitting from superphosphate fertilizer plants ( ) . similar lesions have been reported from australia where the main source of fluoride appeared to have been gypsum that was included in a feed supplement and also ingested from fertilizer dumps on paddocks ( ) . gypsum fertilizers have caused several deaths in animals ( ) . similar morbidity and mortality may be seen in humans who drink contaminated water either intentionally or out of ignorance as well. the latter situation is quite possible among the uneducated farmers of agrarian economies. adrian ( ) drew attention to a very unique situation of poisoning related to fertilizers. in several countries, sewage sludges are used on farms as fertilizers because they do contain these materials. however the sewage-not surprisingly-also contains industrial wastes, such as chromium, lead, zinc, cadmium, and mercury. when this sewage is used as fertilizing material, plants tend to concentrate these heavy metals, especially chromium. ingestion of such farm produce may lead to heavy metal poisoning. several other cases of fertilizer poisoning, especially among animals, have been reported, too ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) . in several countries, poisonous plants, such as castor, are used as green manure which can cause poisoning of both humans and animals. soto-blanco and colleagues from the university of sao paulo, brazil, described a case of canine poisoning where castor bean (ricinus communis) cake was used as a fertilizer ( ) . the authors stressed that these cakes may be accidentally ingested by humans as well, and recommended that cake production should include heat treatment to denature the poisonous proteins. nematicides can cause poisoning in banana plantations. wesseling and co-workers, studying pesticide-related illness and injuries among banana workers in costa rica, reported that workers at highest risk per time unit of exposure were nematicide applicators ( ) . slugs are major pests of oilseed rape that are poorly controlled by conventional bait pellets. therefore, compounds, such as metaldehyde and methiocarb, are used as seed dressings to control slugs ( ) . metaldehyde is a popular molluscicide that can cause fatal poisoning; the aapcc annual report ( ) mentions as many as cases of exposure to this agent. kiyota ( ) reported the case of a -year-old mentally retarded man suffering from pica, who ingested about . g of metaldehyde. despite medical treatment, he developed acute lung injury and died after days; he was found to have ascites and splenomegaly. high-performance liquid chromatography revealed . μg/ml metaldehyde in the serum. jones et al. ( ) developed a method to detect metaldehyde in samples of stomach contents by gas chromatography-ion trap mass spectrometry for forensic toxicology investigations. a suicide attempt using metaldehyde was reported by hancock and co-workers ( ) . a case of homicide using metaldehyde has been described by ludin ( ) . detailed overviews of metaldehyde toxicity have been provided earlier by booze and oehme ( ) and longstreth and pierson ( ) . avermectins used as acaricides (avermectin acaricides), insecticides (avermectin insecticides), and nematicides have been used for suicidal poisoning. chung and co-workers ( ) from taiwan studied the clinical spectrum of avermectin poisoning reported to a poison center from september to december . eighteen patients with abamectin (agri-mek; % wt/wt abamectin) exposure and one with ivermectin (ivomec; % wt/vol ivermectin) ingestion were identified ( males, females; age range - years). fourteen out of the patients had been exposed as a result of attempted suicide; one patient died days later as a result of multiple organ failure. algicides have not been reported to cause fatal poisoning in humans; minor ailments owing to algicide exposure include, e.g., contact dermatitis ( ) . aphicides are known to persist in crops ( ) ; their toxicity in house sparrows has been described in detail by tarrant and co-workers ( ) . bird repellants are trigeminally mediated avian irritants ( ) . toxic effects to humans have apparently not been reported so far. chemosterilants are chemicals that aim at destroying the fertility of pests. , -dibromo- -chloropropane is used to induce infertility in rats ( ) . the chemosterilant bisazir is extremely hazardous. ciereszko and co-workers ( ) have recommended that special safety measures are necessary when handling this chemical. however, toxic effects to humans have not been reported in the medical literature so far. antifeedants are chemicals having tastes and odors that inhibit feeding behavior. several chemicals, such as silphinene sesquiterpenes ( ), , , oxadiazoles ( ) , and ryanoid diterpenes ( ) , are used as antifeedants; again, toxic effects to humans have not been reported so far. herbicide safeners are compounds protecting crops from herbicide injury by increasing the activity of herbicide detoxification enzymes such as glutathione-s-transferases ( ) ( ) ( ) and cytochrome p- s. several herbicide safeners are used in agriculture such as benoxacor ( ) and dichloroacetamide ( , ) ; there toxicity in humans has not been reported so far. insect attractants attract or lure an insect to a trap. several of them are available, such as boll weevil attract and control tubes ® (plato industries, houston, tx) ( ), imidacloprid ( ) , and gf- fruit fly bait ( ) . their toxicity has been studied in detail by beroza et al. ( ) . the secondary effects of conventional insecticides on the environment, vertebrates, and beneficial organisms have caused a move to the use of more target-specific chemicals, such as insect growth regulators (igrs) ( ) . igrs are chemicals disrupting the action of insect hormones controlling molting, maturity from pupal stage to adult, or other insect life processes. several igrs are known, such as halofenozide ( ), s-methoprene ( , ) , buprofezin ( ) , tebufenozide ( ) , the chitin synthesis inhibitors teflubenzuron, diflubenzuron ( ) , and hexaflumuron, as well as the juvenile hormone mimic pyriproxyfen ( ) . halofenozide (rh- ) is a novel nonsteroidal ecdysteroid agonist that induces a precocious and incomplete molt in several insect orders ( ) . the antifeedant , , -oxadiazoles also show a considerable amount of igr activity ( ) . the toxicity of these antifeedants to animals has been studied by wright ( ) . pesticide synergists are chemicals that, although they do not possess inherent pesticidal activity, they nonetheless promote or enhance the effectiveness of other pesticides when used combined (synergism). synergists usually increase the toxicity of a pesticide so that a smaller amount is needed to bring about the desired effect. this may reduce the cost of application. an example of a synergist is piperonyl butoxide, often used with pyrethrin, pyrethroid insecticides, rotenone, and carbamate-containing pesticides. piperonyl butoxide is a liver toxicant and a possible human carcinogen ( , ) ; it also inhibits t-cell activation and function ( ) . -chloro-p-toluidine hydrochloride (cpth) is an aniline derivative registered as a selective, low-volume-use (< kg/yr) avicide. rice baits are treated with cpth to cause poisoning in birds harmful to crops ( ) . cpth may be mutagenic. stankowski et al. ( ) conducted three in vitro mutagenicity tests of cpth according to methods recommended by the united states environmental protection agency, e.g., the ames/salmonella assay, the chinese hamster ovary (cho)/hypoxanthine-guanine phosphoribosyl-transferase mammalian cell forward gene mutation assay, and the cho chromosome aberration assay. they found that cpth did not display mutagenic activity using the ames/salmonella or cho/hypoxanthine-guanine phosphoribosyl-transferase assays. however, cpth induced statistically significant, concentration-dependent, metabolically activated increases in the proportion of aberrant cells. the authors concluded that the results were suggestive of minimal mutagenicity effects associated with exposure to cpth ( ) . stahl and co-workers draw attention to the consumption of cpth treated rice baits by nontargeted bird species, such as pigeon (columbia livia) and house sparrow (passer domesticus). cpth can persist in the breast muscle tissues of both targeted and nontargeted birds which may be a potential secondary hazard to scavengers and predators ( ) . toxicity of cpth both in humans and animals has been discussed by several other authors as well ( ) ( ) ( ) ( ) ( ) . if a particular agrochemical poison has been banned in a country, it is not necessarily that poisoning with this agent will not be seen in that particular country. for example, in japan, production of azomite emulsion (an acaricide) has been stopped since . however, moriya et al. in ( ) described a recent azomite-related fatality. poisoning with azomite was confirmed when aramite and azoxybenzene, two effective components of azomite emulsion, were detected in the patient's serum when qualitatively analyzed with gas chromatography-mass spectrometry. the authors concluded that even if an agrochemical poison is banned, the pathologist must still keep the possibility of its ingestion in mind. many times, it is not the active agricultural chemical that is responsible for poisoning but impurities (such as dioxin), surfactants (e.g., polyethoxylated tallow amine used with glyphosate) and adjuvants used along with the chemical. these adjuvants, or "inert" ingredients, could be solvents, stabilizers, preservatives, sticking or spreading agents, or defoamers ( ) and may constitute petrochemical solvents, such as acetone, fuel oil, toluene, and other benzene-like chemicals. these could sometimes be more toxic than the active ingredient. rubbiani drew attention to several of these adjuvants and clinical syndromes produced by them ( ) . according to harry ( ) , toxicity is often due to solvents or surfactants included in the composition of a formula used as an agricultural chemical. when the obligatory declaration on the label about identity and concentration of some of these substances is not provided by the actual legislation in a particular country, the problem becomes more acute. it is also often difficult to determine if the cause of the poisoning is the actual agricultural chemical itself or its adjuvants. metabolites are breakdown products that form when a pesticide is exposed to air, water, soil, sunlight, or living organisms and often the metabolite is more hazardous than the parent compound. an estimated three million cases of agrochemical poisoning are reported from around the world every year, making it one of most serious toxicological problems of the present times. an overwhelming majority of these-more than %-are reported from developing countries, such as india, presumably because these are predominantly agrarian economies. in the united kingdom, pesticides are responsible for only about % of deaths ( ) , whereas in united states, as seen in table , the figure varies between and %. the equivalent figures in india have been reported to be as high as % ( ) . figure shows some common pesticides used in india. accidental poisoning may occur in a number of ways. accidental poisoning can occur if the insecticide is stored inadvertently with foodstuffs ( ) . one of the most shocking cases of mass agrochemical poisonings occurred in the indian state of kerala in (known popularly as the "kerala food poisoning case of ") when bags of foodstuffs, such as wheat and sugar, were inadvertently stored together with those of folidol (parathion) in the same cabin on a ship ( ) . the insecticide leaked and contaminated the foodstuffs; more than people were accidentally poisoned when they consumed these contaminated foodstuffs. out of these, more than people died. mixing of pesticides with foodstuffs may be intentional, albeit entirely because of ignorance and without any criminal intent. such a case came to notice in the late s in lakhmipur in kheri district, in the indian state of uttar pradesh. farmers in this state were found to be preserving food grains with benzene hexachloride. a severe convulsive epidemic broke out among several hundred people because of this ignorance and more than people died. in , improper use and application of benzene hexachloride in the town sunser in the indian state of madhya pradesh resulted in many people falling ill. fortunately, no human died, but there were reports of several bird casualties. in march , a case of agricultural poisoning from india was reported where an entire family was poisoned owing to leakage of pesticides into cereal (sorghum/jowar) stored in the same room ( ) . the indian state of kerala is a major cashew growing region. there have been attempts at aerial spraying of this cash crop with endosulphan. because these areas are close to local residential areas, deleterious effects occurring in humans have caused a major controversy in recent times ( ) . pillay ( ) suggests that accidental poisoning due to pesticides can occur in four different scenarios: (a) occupational exposure among agriculturists and those engaged in the task of pesticide spraying, (b) contamination of foodstuffs on account of negligence, (c) inadvertent ingestion by children, and (d) reusing pesticide containers for storing food or drink (the latter is very common among third-world countries). instances of fatalities among agricultural workers due to accidental exposures have been reported from time to time ( ) . accidental poisoning owing to some pesticides, such as paraquat, occurs in a number of scenarios, e.g., when the mouthpiece of fumigation equipment is sucked by the operator while cleaning and it is suddenly cleared of obstruction, confusion under the influence of alcohol, consumption of contaminated water or foods, accidental ingestion by children, and accidental cutaneous exposure or oral topical application for toothaches by ignorant persons ( ) . robert g. book of bloemfontein, south africa, reported a unique case of accidental poisoning with paraquat: a young woman tried to "achieve a high" by spiking her coca-cola with paraquat. she died after a few days of hospitalization. at the time of her admission she had told the doctor that her husband had maliciously put paraquat in her drink a few days before; however, only days later she changed her version as just mentioned ( ) . it is noteworthy that in india it is very common for married women at the time of their death to shield their murderous husbands by making such statements. whether the woman's first or second statement was correct is anybody's guess. according to harry ( ) , accidental pesticide intoxications are mainly caused by ingestions of diluted fertilizers, low-concentration antivitamin k rodenticides, ant-killing products, or granules of molluscicides containing % metaldehyde, whereas voluntary intoxications are mostly by chloralose, strychnine, organophosphorus or organochlorine insecticides, concentrated antivitamin k products, and herbicides, such as paraquat, chlorophenoxy compounds, glyphosate, and chlorates. suicidal poisoning with agrochemicals, especially organophosphates and alp, is very common in countries like india. one of the main reasons is the easy availability of these agrochemicals. many companies now add an emetic to dangerous agrochemicals, such as paraquat and alp. addition of a "stenching" agent to paraquat has apparently not deterred suicidals from consuming this poison. homicidal poisoning with organophosphorus compounds is possible and from time to time, one gets to hear or read about cases of a homicide commit-ted with these substances. svraka and colleagues have described four cases of homicide with organophosphorus compounds ( ) . however, homicidal poisonings with organophosphorus compounds are rare because of the unpleasant taste of most agrochemicals, especially of organochlorines, such as endrin, but they have been mixed with alcohol, especially toddy (a strong liquor that is very popular in india), which masks its smell and has been used with organophosphorus compounds for homicidal purposes in this way. homicidal poisoning with parathion is much easier ( ) ( ) ( ) ( ) . to prevent this, a coloring agent, such as indigocarmine, is added to parathion. this is, however, not a universal practice. in india for instance, addition of indigocarmine to parathion is not practiced. the commonly used herbicide paraquat is odorless and gives rise to symptoms mimicking viral pneumonitis. these two properties-classically hailed as the properties of an ideal homicidal poison-make it very attractive as a homicidal poison. paraquat is supposed to have a burning taste, but this can be masked in hot liquids or spicy foods ( ) . several homicide cases with paraquat undoubtedly must have gone unnoticed. teare and teare and brown ( , ) described five cases of paraquat poisoning, of which, two were homicidal in nature. the first is a well-documented case (reg vs kenyon and roberts) in which a -year-old man, keith william kenyon, was killed by his wife jennifer kenyon and her friend, david roberts, a consultant on the effects of agricultural chemicals. she purchased gramoxone along with her friend olive hemming (who turned out to be the chief prosecution witness) from a farm shop, and most likely administered it to her husband in repeated small doses. kenyon was taken ill on november , and died days later, on december . during his illness, he displayed all the classical symptoms and signs of paraquat poisoning. postmortem examination confirmed death by paraquat intoxication. mrs. kenyon was convicted of murder, whereas david roberts was acquitted because of lack of evidence against him ( ) . the second case occurred only month later. after christmas , on the falkland islands, four local agricultural workers had been having a boxing day party when some gramoxone was slipped for some unknown reason into one of their beers. the man died after displaying typical symptoms of paraquat poisoning. autopsy confirmed poisoning by paraquat. criminal charges against the other three laborers were contemplated, but eventually it was decided to drop them. paul ( ) described the case of a -year-old woman who killed her husband by mixing paraquat in his steak-and-kidney pie twice. when he developed a sore throat and was prescribed medicine for treatment, she mixed paraquat in the medicine as well. the husband died on june , after suffering a day illness. the cause of death was attributed to cardiac arrest in combination with renal failure and bilateral pneumonia and it was only by a curious chain of circumstances that paraquat was detected in the young man's tissues preserved in the mortuary in a bucket, months after the man's death. his wife and her paramour were found guilty and sentenced. stephens and moormeister from the medical examiner's office of san francisco, ca, reported four cases of homicidal poisoning by paraquat ( ) . of these, the first three murders were perpetrated by one man against members of his immediate family, and the fourth case was equivocal-it could either have been suicide or homicide. the first three murders were committed by a man who had been married five times. his first three wives were alive and healthy. when the fourth wife threatened to divorce him, she found herself ill and died days after the onset of her illness ( days after hospitalization). eight years later, when his fifth wife threatened divorce, she suffered the same fate, and a few months later, his -year-old mother also died. all three showed typical symptoms of paraquat poisoning. the postmortem findings seemed to suggest natural disease of the lungs. although a suggestion of paraquat poisoning was made in all three cases, the concerned pathologist was reluctant to sign death certificates as paraquat poisoning. toxicological analysis in the second and third cases revealed the presence of paraquat in the victims' tissues and this resulted in conviction of the murderer. it was found that the defendant worked as a mechanic on a large agricultural ranch and had easy access to paraquat; his thumb print was found on one of the opened paraquat containers, although he had earlier denied having to do anything with those containers. the fourth case involved a -year-old man, a registered herbicide and pesticide user, who had marital difficulties with his aggressive, "shrew-like" wife who also stood to benefit from a large insurance policy upon his death. while in hospital, the victim denied suicidal ingestion; he died days after the start of his illness. no testing of toxic effects from the compounds he worked with was ever performed, nor was any consideration given to this possibility. the case did not result in court charges for anyone. stephens and moormeister concluded that the reason why such cases will often go unnoticed is because of the reluctance on the part of both clinicians and forensic pathologists to even think in the direction of paraquat poisoning when they see such a clear and typical picture of "viral pneumonia." in their opinion, the clinician should suspect paraquat ingestion in all cases in which there is progressive pulmonary involvement with no features of viral infection ( ) . the pathologist conducting the postmortem would do well to go through the clinical history, if available, in detail to rule out the possibility of paraquat poisoning. in all doubtful cases, a full toxicological analysis should be done and the tissues should be particularly analyzed for paraquat. daisley and simmons from the university of the west indies in trinidad reported two cases of homicide by paraquat poisoning ( ) . both cases occurred in children and the common clinical presentations were gastrointestinal ulceration and acute respiratory distress with pneumomediastinitis. at autopsy, the most prominent finding was bullous lung emphysema. the authors stress that pathologists should be aware of this finding because they feel that if this autopsy finding is seen combined with the typical clinical presentation mentioned in sections . . . and . . ., it is almost diagnostic of acute paraquat poisoning. da costa et al. have dealt with the medicolegal aspects related to paraquat poisoning in detail ( ) . another weed killer that has been used commonly for homicidal purposes is sodium chlorate. in reg vs hargreaves, hampshire (winchester) assizes, april , a -year-old woman was charged with the murder of a -year-old man whom she had known for the last years as an uncle. in august , he made his last will, written out by the accused in her favor. on january , the accused bought the weed killer sodium chlorate apparently for a friend who was a gardener. on january , , the old man died and the postmortem examination showed signs of death from sodium chlorate poisoning. the victim had consumed beer and the remaining beer in the mug contained some mg of sodium chlorate. the jury found the woman guilty of manslaughter and sentenced her to months of imprisonment ( ) . one of the biggest and most well-known medicolegal controversies in connection with herbicides has been that of agent orange. agent orange is the name given to a mixture of herbicides that united states military forces sprayed in vietnam from to during the vietnam war for the dual purpose of destroying crops that might feed the enemy and defoliating forest areas that might conceal viet cong and north vietnamese forces. the defoliant consisted of approximately equal amounts of the unpurified butyl esters of , -d and , , -trichlorophenoxyacetic acid ( , , -t). agent orange also contained small, variable proportions of , , , -tetrachlorodibenzo-p-dioxin-commonly known as dioxin-which is a byproduct of the manufacture of , , -t and is toxic even in minute quantities; dioxin is considered one of the most toxic compounds synthesized by humans. agent orange was delivered in -gallon drums with an orange stripe to distinguish the drums visually from those containing other chemical agents (hence the name). about million liters of agent orange were sprayed over vietnam from low-flying aircrafts. among the vietnamese, it is considered to be the cause of an abnormally high incidence of miscarriages, skin diseases, cancers, birth defects, and congenital malformations (often extreme and grotesque). alterations in manufacturing procedures had reduced the dioxin content in agent orange later to minimal levels. today, , , -t registrations have been cancelled and agent orange was voluntarily removed by the manufacturers in . many united states, australian, and new zealand servicemen who suffered long exposure to agent orange in vietnam later developed cancer and other health disorders. a class-action lawsuit was brought against seven herbicide makers that produced agent orange for the united states military. the suit was settled out of court with the establishment of a $ , , fund to compensate some , claimants and their families. separately, the united states department of veterans affairs awarded compensation to about veterans. agent orange has now been replaced by agent white, a mixture of , -d and picloram, which is longer lasting and more effective. in the united states, the federal insecticide, fungicide and rodenticide act (fifra) was passed in (amended in , , and [ ] ). this act divides all pesticides in four broad classes depending on their toxicity. the label of each pesticide has to contain a signal word depending on its toxicity. the criteria established by the fifra are given in table . according to the fifra, toxic category i pesticides must have the signal words danger and poison (in red letters) and a skull and crossbones prominently displayed on the package label. the spanish equivalent for danger, peligro, must also appear on the labels of highly toxic chemicals. toxic category ii pesticides must have the signal word warning (aviso in spanish) displayed on the product label. toxic category iii pesticides are required to have the signal word caution on the pesticide label. toxic category iv pesticide products shall bear on the front panel the signal word caution on the pesticide label. pesticides formulated in petroleum solvents or other combustible liquids must also include the precautionary word flammable on the product label. this was obviously done to prevent cases of accidental poisoning, and similar acts exist in almost all countries. in india, a predominantly agricultural country, handling of insecticides is governed by the insecticides act and the insecticide rules, (amended in ) ( ) . section of the insecticide rules, classifies insecticides on a similar basis. section also insists on affixing a label to the insecticide container in such a manner that it cannot be ordinarily removed. among other things, it must contain a square, occupying not less than onesixteenth of the total area of the face of the label, set at an angle of °(diamond shape). this square is to be divided into two equal triangles, the upper portion of which shall contain the signal word, and the lower portion the specified color. the classification of insecticides, signal words to be used, and the color of the identification band on the label according to the insecticide rules, of india are given in table . if a pesticide is misused in any way, the person who bought and stored the pesticide may be legally responsible. in the united states, the food quality protection act was passed in as a complementary set of regulations, which, among other important features, specifically recognizes the special situations and usages of pesticides for public health. these laws regulate the registration, manufacture, transportation, distribution, and use of pesticides. the regulations are administered by the environmental protection agency. more than , bright green ld : lethal dose in % of the exposed subjects intoxications caused by plant protection chemicals in forensic toxicology in urban south africa patterns and problems of deliberate self-poisoning in the developing world pesticide poisoning agricultural and horticultural chemical poisonings: mortality and morbidity in the united states effects of chemical factors on health of the population in various regions of russia green revolution agriculture and chemical hazards human poisoning caused by agents for plant protection in human poisoning caused by chemicals for plant protection in an epidemic of pesticide poisoning in nicaragua: implications for prevention in developing countries agrichemical hazards in the south african farming sector chemical hazards to agricultural workers acute pesticide poisoning: a major global health problem agrochemical poisoning in sri lanka annual report of the american association of poison control centers toxic exposure surveillance system annual report of the american association of poison control centers toxic exposure surveillance system annual report of the american association of poison control centers toxic exposure surveillance system annual report of the american association of poison control centers toxic exposure surveillance system annual report of the american association of poison control centers toxic exposure surveillance system goodman & gilman's pharmacological basis of therapeutics clinical and experimental toxicology of organophosphates and carbamates carbamate insecticides acute neurotoxic organophosphate poisoning: insecticides and chemical weapons knight's forensic pathology forensic histopathology acute organophosphorus compound poisoning organophosphate poisoning and complete heart block clinical and experimental toxicology of organophosphates and carbamates biochemical determination of cholinesterase activity in biological fluids and tissues pseudocholinesterase deficiency and anticholinesterase toxicity biochemical and toxicological investigations related to op compounds an electrometric method for the determination of red blood cell and plasma cholinesterase activity handbook of pesticide toxicology ( vols.) handbook of autopsy practice detection of e several years after burial concerning evidence of parathion (e ) in an exhumed corpse after months analysis of consecutive forensic exhumations with emphasis on undetected homicides clinical and experimental toxicology of organophosphates and carbamates pocketbook of pesticide poisoning for physicians. cbs publishers and distributors goldfrank's toxicologic emergencies modern medical toxicology dreisbach's handbook of poisoning-prevention acute nicotine poisoning the bipyridylium herbicides poisoning by paraquat pesticide-related illness and injuries among banana workers in costa rica: a comparison between and paraquat poisoning in papua new guinea mechanisms of toxicity, clinical features, and management of diquat poisoning: a review paraquat and related bipyridyls death from paraquat after subcutaneous injection reduction of paraquat and related bipyridylium compounds to free radical metabolites by rat hepatocytes morphologic lesions in paraquat poisoning simultaneous determination of paraquat and diquat in human tissues by high-performance liquid chromatography the determination of paraquat ( , ′dimethyl- - ′bipyridylium cation) in urine diquat intoxication a case of fatal diquat poisoning: toxicokinetic data and autopsy findings pesticides studied in man extoxnet pesticide information profile: , -d fatal suicidal poisoning due to , -d fatal human ingestion of , -d, a common herbicide radioimmunoassay for , -dichlorophenoxyacetic acid simultaneous gas chromatographic determination of , -d and dicamba in human blood and urine classes of toxic chemicals clinical management of poisoning and drug overdose sodium chlorate poisoning complicated by acute renal failure acute poisoning with a glyphosatesurfactant herbicide safety evaluation and risk assessment of the herbicide roundup and its active ingredient, glyphosate, for humans rapid determination of glyphosate in postmortem specimens using p nmr bordeaux mixture the metabolism of fungicides three fatal sodium azide poisonings vineyard sprayer's lung": a new occupational disease vineyard sprayer's lung vineyard sprayer's lung-a rare occupational disease pesticide-induced pulmonary fibrosis cytologic changes of the respiratory tract in vineyard spraying workers lung changes induced by copper sulfate. an experimental contribution to the so-called "vineyard sprayer's lung liver granulomas containing copper in vineyard sprayer's lung. a new etiology of hepatic granulomatosis liver disease in vineyard sprayers acute copper sulfate poisoning history of toxicology mercury exposure and early effects: an overview methylmercury poisoning in iraq methylmercury poisoning in iraqi children: clinical observations over two years an outbreak of methylmercury poisoning due to consumption of contaminated grain methyl mercury poisoning in iraq. a neurological study minamata disease embryotoxic and teratogenic action of the organomercury fungicide falizan on chick embryos perimortem fixation of the gastric and duodenal mucosa: a diagnostic indication for oral poisoning chronic arsenicism death and injury caused by methyl bromide, an insecticide fumigant a fatal case of acute methyl bromide poisoning alternatives to methyl bromide treatments for storedproduct and quarantine insects halogenated hydrocarbons fatal accident resulting from methyl bromide poisoning after fumigation of a neighbouring house; leakage through sewage pipes death following intentional methyl bromide poisoning: toxicological data and literature review health effects associated with sulfuryl fluoride and methyl bromide exposure among structural fumigation workers suicide by exposure to sulfuryl fluoride cardiovascular manifestations of aluminium phosphide intoxication aluminium phosphide: worse than bhopal epidemic aluminium phosphide poisoning in northern india lal-mm ( ) occupational phosphine exposure in indian workers aluminium phosphide poisoning adrenocortical involvement in aluminium phosphide poisoning acute pericarditis in aluminium phosphide poisoning acute phosphine poisoning following ingestion of aluminium phosphide fatal aluminum phosphide poisoning adult respiratory distress syndrome following aluminium phosphide ingestion. report of cases four cases of sudden death in a silo silo filler's disease chemical hazards to agricultural workers ammonia gas poisoning. forty-seven cases in a london shelter a case of fatal methemoglobinemia of unknown origin but presumably due to ingestion of nitrate a case of fatal intoxication with ammonium sulfate and a toxicological study using rabbits acute poisoning of cattle by fertilizer-contaminated water industrial fluorosis in cattle and buffalo around udaipur chronic gypsum fertiliser ingestion as a significant contributor to a multifactorial cattle mortality sheep deaths after accidental ingestion of gypsum fertiliser contribution of the incidence of urban sewage spreading to dietary chromium hyperkalemia after house plant fertilizer poisoning accidental superphosphate fertilizer poisoning in pregnant ewes urea poisoning in suckler cows acute, fatal illness in cattle exposed to boron fertilizer suspected superphosphate poisoning in calves forensic pathological studies on autopsy of a patient who died after ingesting the liquid fertilizer hyponex nitrite poisoning in herring gulls (larus argentatus) and ringbilled gulls (larus delawarensis) accidental ammonia exposure to county fair show livestock due to contaminated drinking water ricinus communis cake poisoning in a dog seed dressings to control slug damage in oilseed rape case report-fatal snail bait (metaldehyde) overdose presenting aspiration pneumonia determination of metaldehyde in suspected cases of animal poisoning using gas chromatography-ion trap mass spectrometry attempted suicide with a pesticide mixture murder by metaldehyde poisoning metaldehyde toxicity: a review metaldehyde poisoning from slug bait ingestion agricultural avermectins: an uncommon but potentially fatal cause of pesticide poisoning -chloro- , -dinitrobenzene as an algicide. report of four cases of contact dermatitis aphicide persistence on spinach and mustard greens biochemical and histological effects of the aphicide demeton-s-methyl on house sparrows (passer domesticus) under field conditions bird repellents: interaction of chemical agents in mixtures (dbcp)-induced infertility in male rats mediated by a post-testicular effect efficacy of animal anti-fertility compounds against sea lamprey (petromyzon marinus) spermatozoa silphinene sesquiterpenes as model insect antifeedants the toxic and anti-feedant activity of h-pyridazin- -one-substituted , , -oxadiazoles against the armyworm pseudaletia separata (walker) and other insects and mites selective insect antifeedant and toxic action of ryanoid diterpenes tissue-specific expression and localization of safener-induced glutathione s-transferase proteins in triticum tauschii structure of a tau class glutathione s-transferase from wheat active in herbicide detoxification partial characterization of glutathione s-transferases from wheat (triticum spp.) and purification of a safenerinduced glutathione s-transferase from triticum tauschii partial characterization of glutathione stransferase isozymes induced by the herbicide safener benoxacor in maize herbicide safener-binding protein of maize. purification, cloning, and expression of an encoding cdna specific binding of a dichloroacetamide herbicide safener in maize at a site that also binds thiocarbamate and chloroacetanilide herbicides boll weevil (coleoptera: curculionidae) bait sticks: toxicity and malathion content laboratory and field performance of an imidacloprid gel bait against german cockroaches (dictyoptera: blattellidae) effectiveness of gf- fruit fly bait spray applied to border area plants for control of melon flies (diptera: tephritidae) acute toxicity studies with insect attractants insecticidal activity of a nonsteroidal moulting hormone agonist on mosquito larvae and effects on ecdysteroid amounts laboratory evaluation of selective pesticides against the storage mite tyrophagus putrescentiae (acari: acaridae) pulse-exposure effects of selected insecticides to juvenile australian crimson-spotted rainbowfish (melanotaenia duboulayi) laboratory evaluation of selected larvicides and insect growth regulators against field-collected culex quinquefasciatus larvae from urban dhaka inter-regional differences in baseline toxicity of bemisia argentifolii (homoptera: aleyrodidae) to the two insect growth regulators, buprofezin and pyriproxyfen significance of penetration, excretion, and transovarial uptake to toxicity of three insect growth regulators in predatory lacewing adults cytotoxic effects of two antimolting insecticides in mammalian cho-k cells persistence and residual activity of an organophosphate, pirimiphosmethyl, and three igrs, hexaflumuron, teflubenzuron and pyriproxyfen, against the cowpea weevil, callosobruchus maculatus (coleoptera: bruchidae) environmental and toxicological aspects of insect growth regulators mechanistic study on liver tumor promoting effects of piperonyl butoxide in rats piperonyl butoxide technical fact sheet cuprizone and piperonyl butoxide, proposed inhibitors of t-cell function, attenuate experimental allergic encephalomyelitis in sjl mice poisoning with -aminopyridine: report of three cases -aminopyridine poisoning of crows in the chicago area stabilization of the avicide -chloro-ptoluidine as the beta-cyclodextrin adduct -chloro-p-toluidine hydrochloride: in vitro mutagenicity studies for human health hazards determinations improved method for quantifying the avicide -chloro-p-toluidine hydrochloride in bird tissues using a deuterated surrogate/gc/ms method toxicity of the avicide -chloro- -acetotoluidide in rats: a comparison with its nonacetylated form -chloro-p-toluidine the role of renal aromatic n-deacetylase in selective toxicity of avicide -chloro-p-toluidine in birds effects of an acutely toxic dose of the avicide -chloro-p-toluidine in chickens toxicologic studies on the avicide -chloro-p-toluidine tissue distribution and binding of radioactivity in mouse after intravenous administration of [ c] -chloro-p-toluidine a case of suicide suspected of poisoning from taking some agricultural chemicals label review manual chapter : ingredient statement. office of pesticide programs the problem of the presence of dangerous adjuvants in pesticide preparations used in agriculture or households textbook of forensic medicine and toxicology foodborne outbreak of organophosphorus compound poisoning food poisoning due to organophosphorus compounds investigation of a fatality among parathion applicators in california paraquat poisoning in children: survival of three cases homicidal poisoning by paraquat four cases of murder with organophosphorous compound a case of homicidal parathion poisoning six homicides by means of e- committed by a minor criminal poisoning by means of administration of parathion mixed with a medicinal substance a case of murder by parathion (e ) which nearly escaped detection homicidal poisoning by paraquat poisoning by paraquat murder under the microscope-the story of scotland yard's forensic science laboratory homicide by paraquat poisoning the forensic medical aspects of paraquat poisonings sodium and potassium compounds fifra- , glp, and qa: pesticide registration the insecticides act, with the insecticide rules, , as amended by the insecticide (amendment) rules, . delhi law house acknowledgment i wish to thank my wife marygold gupta, a chemist, and my son tarun aggrawal for their whole-hearted support during the writing of this chapter. marygold was especially helpful in making me comprehend the chemical structures of several pesticides. tarun drew several chemical structures and figures on his computer. key: cord- -yapmcvps authors: menzies, rachel e.; menzies, ross g. title: death anxiety in the time of covid- : theoretical explanations and clinical implications date: - - journal: cogn behav therap doi: . /s x sha: doc_id: cord_uid: yapmcvps the recent covid- pandemic has triggered a surge in anxiety across the globe. much of the public’s behavioural and emotional response to the virus can be understood through the framework of terror management theory, which proposes that fear of death drives much of human behaviour. in the context of the current pandemic, death anxiety, a recently proposed transdiagnostic construct, appears especially relevant. fear of death has recently been shown to predict not only anxiety related to covid- , but also to play a causal role in various mental health conditions. given this, it is argued that treatment programmes in mental health may need to broaden their focus to directly target the dread of death. notably, cognitive behavioural therapy (cbt) has been shown to produce significant reductions in death anxiety. as such, it is possible that complementing current treatments with specific cbt techniques addressing fears of death may ensure enhanced long-term symptom reduction. further research is essential in order to examine whether treating death anxiety will indeed improve long-term outcomes, and prevent the emergence of future disorders in vulnerable populations. key learning aims: ( ).. to understand terror management theory and its theoretical explanation of death anxiety in the context of covid- . ( ).. to understand the transdiagnostic role of death anxiety in mental health disorders. ( ).. to understand current treatment approaches for directly targeting death anxiety, and the importance of doing so to improve long-term treatment outcomes. in december , a novel coronavirus was first detected in the city of wuhan, china. within five weeks, the virus, now named covid- , began to dominate global headlines. by mid-may , covid- had resulted in the deaths of more than , people worldwide, with nearly . million cases confirmed (world health organization, ). as cases increased, governments around the world began closing borders, and introducing social distancing restrictions and lockdown orders, in an effort to slow the rapid acceleration of the virus. prior to many of these government responses, reports emerged of individuals choosing to self-isolate, as mass panic swept through communities in waves. anecdotal reports of verbal and physical aggression in grocery stores, hoarding of antibacterial products and other supplies, and racist abuse of individuals with asian appearance increased as fear took over across the world (devakumar et al., ; garfin et al., ) . as individuals scrambled to prevent the threat of covid- in any way they could, online sales of 'immune boosters' and untrialled medicines increased. analyses of google data across just days in march revealed a total of , searches for where to purchase chloroquine and hydroxychloroquine, two drugs which were touted by the media as potentially effective, despite the existing clinical evidence for the efficacy of these drugs being inconclusive (liu et al., ) . emerging research data are already revealing high levels of anxiety concerning the virus, with findings from nearly participants suggesting that greater perceived severity of the virus is associated with poorer mental health outcomes (li et al., ) . arguably, this response from the public should not come as a surprise. fears of death have been proposed to be a central and universal part of the experience of being human (becker, ) . in fact, there is evidence of humans grappling with death anxiety for as long as our species has been recording its history (menzies, b) . we are the only species that we know of that has the cognitive capacity to contemplate and anticipate our own death, yet this impressive ability comes with a downside; we are destined to live our lives 'forever shadowed by the knowledge that we will grow, blossom, and inevitably, diminish and die' (yalom, , p. ). on the one hand, people may develop adaptive ways of coping with their fear of death, such as building meaningful relationships and leaving a positive legacy (yalom, ) . on the other hand, awareness of death may also produce a powerful sense of fear or meaninglessness, and may drive a number of maladaptive coping behaviours (menzies, ) . some of these behaviours (e.g. avoidance) may underlie numerous mental health conditions, while other behaviours may appear, on the surface, not directly linked to death at all. how might our fears of death be shaping our everyday behaviour in ways that we are not even aware of? terror management theory terror management theory (tmt), a social psychological theory based on the work of cultural anthropologist ernest becker, is the leading psychological framework for explaining this effect of death fears on human behaviour (greenberg et al., ) . tmt posits that our awareness of our own death produces a crippling terror, and that humans have developed two distinct buffers in order to allay this fear: cultural worldviews, and self-esteem. cultural worldviews involve shared symbolic concepts of the world, including identifying with cultural values or endorsing belief systems, such as the belief in an afterlife. sharing these cultural worldviews is thought to offer a sense of 'symbolic immortality', by giving an individual a sense of permanence and meaning in the face of death. secondly, self-esteem, gained through fulfilling the expectations of our cultural worldview, is also said to buffer death anxiety, by making one feel like a valuable member of their culture, who will be remembered after death (greenberg, ) . tmt also proposes that humans use different defence mechanisms depending on whether thoughts of death are within or outside of conscious awareness. according to this 'dual process model', when thoughts of death are conscious, we engage in 'proximal defences', which include suppressing these thoughts (e.g. turning off a news report about covid- death tolls), denying one's vulnerability (e.g. 'i'm not in a high risk group, so i'll probably be fine'), or trying to prevent death (e.g. cleaning down all home surfaces with antibacterial wipes) . on the other hand, when thoughts of death leave conscious awareness, we instead engage in 'distal defences', which involve bolstering our two buffers (e.g. by endorsing our cultural worldviews, or enhancing our self-esteem). findings from hundreds of studies have demonstrated support for tmt (burke et al., ) . primarily, these studies have involved a 'mortality salience' design, in which participants in one condition are reminded of their mortality, while participants in the control condition are reminded of an aversive topic that is unrelated to death. these studies have shown that reminders of death drive a vast array of human behaviours, including intention to purchase products (dar-nimrod, ) , driving behaviour (taubman-ben-ari et al., ) , and even suntanning (routledge et al., ) . despite some recent studies questioning the replicability of tmt results (e.g. klein et al., ) , follow-up studies have demonstrated that classic tmt findings do replicate when sufficiently powered (chatard et al., ) . in addition, burke et al.'s ( ) review of tmt experiments found that death reminders yielded moderate effects on a range of behavioural variables, with little evidence of publication bias, further highlighting the strength of mortality salience effects. given this, what role might death anxiety be playing in the current pandemic? death anxiety and the covid- pandemic with the exception of a handful of studies, the majority of tmt research has been conducted under laboratory conditions; i.e. for those in the mortality salience condition, death is usually primed in the form of two short questions about one's death, which participants are asked to respond to. covid- offers an unusual scenario, in which mortality is made salient nearly constantly, given the daily updates on death tolls from the news and social media, and ubiquitous visible death cues in the form of face masks, anti-bacterial sprays and wipes, social distancing and public health campaigns. supporting this idea, laboratory findings have demonstrated that reflecting on current epidemics or virus outbreaks (e.g. ebola, swine flu) produces comparable findings to standard mortality salience primes, increasing the accessibility of death-related thoughts, and increasing defensive behaviour (e.g. arrowood et al., ; bélanger et al., ; van tongeren et al., ) . although it is currently unknown what the long-term effects of mortality salience primes are, the consequence on human behaviour of even minor, subtle reminders of death under laboratory conditions have much to tell us about the behaviours observed during the current pandemic. first, from this perspective, the observed reports of both covert and overt racism towards asian individuals are unsurprising. these observations are supported by a recent study that found a positive relationship between coronavirus-related anxiety and avoidance of chinese food and products (lee, ) , echoing similar observations of avoidance of chinese people following the sars outbreak (keil and ali, ) . these experiences offer a real-world confirmation of the tmt laboratory findings that reminders of death lead people to feel more hostile towards those of different cultural backgrounds to their own, as they are seen as a threat to one's own worldviews. findings across a number of studies reveal that reminders of death increase stereotypical thinking about people of other races (schimel et al., ) , increase aggression against those who criticise one's nation (mcgregor et al., ) , and lead white participants to hold more favourable reactions to white pride advocates (greenberg et al., ) . one study even found that germans interviewed in front of a cemetery reported strongly preferring german products over foreign products, whereas germans interviewed in front of a shop did not show this preference (jonas et al., ) . similar effects have been observed in more than countries worldwide (greenberg and kosloff, ) . so, much of the recent upsurge in xenophobia, or even hostility towards those with different political views, can be explained by the tmt notion that bolstering our cultural worldviews, and aggressing against those that threaten them, are one means of gaining a sense of symbolic immortality. this idea is further supported by the recent observation of mutual discrimination between east asian societies in the midst of the pandemic (e.g. individuals in taiwan avoiding contact with koreans and japanese individuals; lin, ) . whilst the bolstering of one's cultural worldviews is an example of distal defences being engaged during the pandemic, proximal defences, in the form of attempts to ward off death (e.g. spikes in purchases of hydroxychloroquine, a drug falsely touted as a cure to the virus) or denial have also been observed (jong-fast, ). furthermore, despite the unsurprising recency of much of the research, some preliminary data support the idea that death anxiety may be driving a significant amount of psychological distress during this pandemic. evaluation of the psychometric properties of the fear of covid- scale revealed that the item 'i am afraid of losing my life because of coronavirus- ' had the highest factor loading, suggesting that one's worry about one's own fatality risk is highly predictive of broad fears of the virus (ahorsu et al., ) . data from residents of china revealed that estimates of fatality also appear to specifically predict their psychological distress, with low estimates of one's own survival from covid- predicting greater levels of stress and depression on the depression, anxiety and stress scale dass- (wang et al., ) . one large study of australians specifically explored fears of death in the context of the pandemic (newton-john et al., ). the findings revealed a significant positive correlation between death anxiety and anxious beliefs and behaviours related to covid- (e.g. estimated likelihood of contracting the virus, estimated likelihood of wearing a mask in public, etc.), in addition to self-reported health anxiety, and overall psychological distress. furthermore, participant responses to items assessing beliefs surrounding the virus indicated a heightened perception of threat. for example, when participants were asked how likely they would be to die if they contracted covid- in the next months, the mean likelihood estimate was %, a figure more than times the actual australian case fatality rate of < %. so, while death anxiety may indeed be a driving factor in everyday human behaviour, it appears more relevant than ever in the context of the current pandemic. covid- may be understood as a real-life and ever-present mortality salience prime, influencing people's behaviour in ways they may not even be consciously aware of. early findings suggest that fears of death predict anxiety about the virus, which in turn is shown to predict broader psychological distress. these findings may suggest a causal relationship between death anxiety and psychological distress, and this relationship may be exacerbated in the current pandemic. death anxiety has been proposed to be a transdiagnostic construct, underpinning a range of different mental health conditions (iverach et al., ) . for instance, fears of death may manifest in the frequent reassurance seeking from doctors, checking of one's body, and requests for medical testing seen in the somatic symptom-related disorders (furer et al., ) . in a similar vein, panic disorder often features worries about heart attacks during panic attacks, in addition to repeated appointments with cardiac specialists to allay these concerns (starcevic, ) . specific phobias have been argued to have death anxiety at their core for over a century (kingman, ) , with all of the common phobic objects having the potential to directly result in death (e.g. fears of spiders, snakes, flying and heights). fears of death have also been argued to play a central role in various presentations of obsessive compulsive disorder, as clients attempt to prevent death by illness (in the contamination subtype), household fire or electrocution (in compulsive checking), and death to oneself or another due to acting on intrusive thoughts (as seen in aggressive obsessions) (menzies and dar-nimrod, ; menzies et al., ) . existential concerns have also been argued to play a role in the depressive disorders, with concerns surrounding death and meaninglessness being a common theme (ghaemi, ; simon et al., ) . a number of studies have demonstrated significant relationships between self-reported death anxiety and symptomology of various disorders, including separation anxiety (caras, ) , hypochondriasis (noyes et al., ) , post-traumatic stress disorder (martz, ) , depression (ongider and eyuboglu, ) and eating disorders (le marne and harris, ). results from one large clinical sample found significant and positive correlations between death anxiety and number of lifetime mental health diagnoses, number of medications for mental health, dass- depression, anxiety and stress scores, as well as the symptom severity of different disorders . notably, these relationships remained significant after controlling for neuroticism, suggesting the unique role of death anxiety in psychopathology. while limited conclusions regarding causality can be drawn from such correlational designs, a handful of studies have explored the causal role of death anxiety in mental illnesses using a mortality salience design. these have revealed that reminders of death increase avoidance of spider-related stimuli among spider phobics (strachan et al., ) , social avoidance (strachan et al., ) and attentional biases towards threat among the socially anxious (finch et al., ) , and even restricted consumption of high caloric foods amongst women, suggesting the relevance of death anxiety in eating disorders (goldenberg et al., ) . while few studies have used clinical samples, one study investigated the effect of mortality salience on compulsive handwashing, utilising a large sample of treatment-seeking individuals diagnosed with ocd (menzies and dar-nimrod, ) . participants were first primed with either death or a control topic. following a short delay to allow the effects of the prime to become unconscious, they were asked to wash their hands. the findings revealed that reminders of death doubled the time spent handwashing. notably, this increase in handwashing occurred despite no difference in reported anxiety or perceptions of cleanliness. results from another recent mortality salience design appear particularly relevant to the current pandemic. across a sample of participants with panic disorder or a somatic symptom-related disorder, reminders of death were shown to increase time spent checking one's body for physical symptoms, increase perceived threat of one's symptoms, and also increase intention to visit a medical specialist in the near future . these findings suggest that death anxiety drives relevant anxious behaviour for those vulnerable to health-related worries. results from numerous studies appear to suggest that fear of death is indeed a transdiagnostic construct driving a number of mental health conditions, although further research using treatment-seeking and clinical samples is clearly warranted. if death anxiety does underlie numerous disorders, this may explain the 'revolving door' phenomenon often observed in clinical practice, in which an individual receives apparently successful treatment for one disorder, only to present with a distinctly different disorder at a later time point (iverach et al., , p. ) . if death anxiety is indeed 'the worm at the core' (james, , p. ) of the human psyche, then failing to treat it may result in individuals continuing to present with different mental health conditions at various points across their lifespan. fear of death may need to be assessed and explicitly targeted in treatment in order to achieve long-term amelioration in symptoms and foster ongoing client wellbeing. as with any target of clinical treatments, a thorough assessment paves the way to the most effective treatments of death anxiety, tailored to the individual's unique needs. the clinical interview in early sessions should focus on exploring the topic of death, including assessing for any early losses, memories, or experiences associated with death (menzies and veale, ) . it is also essential to assess the individual's specific worries or thoughts about death, as these can vary largely between individuals. for example, worries may revolve around the dying process itself (e.g. pain or loss of cognitive capacities), the feared death of a loved one, fears concerning eternal punishment in the afterlife, uncertainty surrounding life after death, or non-existence itself, and each theme may need to be addressed using distinctly different lines of cognitive challenging. maladaptive behaviours the individual engages in should also be identified during the assessment stage, including any avoidance behaviours (e.g. avoiding the news, hospitals, flying or driving, or suppressing thoughts around death), reassurance seeking (e.g. from family or one's doctor), self-medicating, or compensatory behaviours (e.g. excessive exercise) (menzies and veale, ) . in the context of the current pandemic, it would be important to distinguish behaviours which are adaptive (i.e. behaviours generally recommended by health professionals and public officials, such as wearing a face mask when leaving the house, self-isolating when symptomatic, and regularly washing one's hands for recommended durations) compared with those that are maladaptive (i.e. behaviours that are not in line with standard recommendations and disrupt the individual's life, such as washing one's hands for hours each day, or requesting repeated medical tests for the virus despite lack of symptoms). alongside a standard clinical interview, questionnaires can prove useful in measuring severity of death fears, as well as tracking change following treatment. one recent systematic review of death anxiety measures revealed that there is a strong need for rigorous measures which have been validated in clinical samples, and that many measures in this field lack adequate psychometric properties (zuccala et al., ) . despite this, a number of measures may prove particularly useful in assessing death anxiety. these include the collett-lester fear of death scale-revised (lester, ) , which has been demonstrated to be responsive to treatment effects, and thus appears to be the best choice for exploring clinical change, and the multidimensional fear of death scale (hoelter, ) , for which means for various clinical groups have been reported . the death attitude profile-revised (wong et al., ) may also offer clinical utility, due to its unique assessment of adaptive attitudes, such as three distinct types of death acceptance, which have been shown to predict more positive outcomes (tomer and eliason, ) . despite being understood to be an 'existential given' (yalom, ) , empirical findings fortunately indicate that death anxiety can indeed be ameliorated. one recent meta-analysis examined the effects of randomised controlled trials on death anxiety (menzies et al., ) . this revealed that psychosocial interventions produced significant reductions in death anxiety relative to control conditions. notably, this effect was found to be driven by cognitive behaviour therapy (cbt) interventions, which produced significantly greater improvements in death fears compared with other treatment modalities. in particular, cbt treatments centring on graded exposure therapy were found to be most effective. in fact, alternative treatment options examined by the meta-analysis failed to produce any significant change in death anxiety scores (menzies et al., ) . given these meta-analytic findings, cbt appears to be the most appropriate treatment for addressing death anxiety, and various techniques for doing so have been proposed (see further, menzies, a; menzies and veale, ) . a number of exposure therapy tasks have been recommended in order to ameliorate death anxiety. of course, as with any exposure tasks, these should be specifically tailored to the individual's own unique pattern of avoidance, and situations or themes that the individual has systematically avoided should be prioritised. one exposure task that can be tailored to the individual's specific concerns is that of an 'illness story', recommended by furer et al. ( ) . this involves writing a vivid description of the death of oneself or a loved one, starting with the events leading up to the death (e.g. the initial diagnosis of a terminal illness), progressing to the death itself, followed by the imagined funeral and aftermath. a similar task is popularised by acceptance and commitment therapy, which involves vividly imagining one's own funeral, and writing one's own eulogy and a tombstone inscription (hayes and smith, ) . other exposure tasks may involve visiting places associated with death that the client has avoided, such as hospitals, nursing homes, cemeteries or funeral homes. reading obituaries online or in the newspaper may also offer valuable exposure opportunities, and clients should be encouraged to deliberately seek out those who have died around their own age (furer et al., ) . preparing one's will, or having discussions regarding end-of-life preferences, may also be considered as exposure tasks, and may serve the additional benefit of increasing the individual's sense of control over their death (furer et al., ; henderson, ) . books (e.g. when breath becomes air by paul kalanithi), films (e.g. blade runner, up), television shows (e.g. after life) and music (e.g. all things must pass by george harrison) related to death may all offer valuable and powerful opportunities for exposure, in addition to helping to normalise death. two thousand years ago, the stoic philosophers of ancient greece observed that 'it is not things themselves that trouble people, but their opinions about things' (epictetus, , p. ). this principle lies at the heart of both stoic philosophy (which emphasised the need to accept death as a universal event outside of our control) and cbt. all of us hold an array of beliefs surrounding death, which may fluctuate between being adaptive (e.g. the belief that we would ultimately cope with the death of a loved one) or maladaptive (e.g. the belief that dying will inherently involve pain and suffering). beliefs of this latter type will understandably cause distress for many individuals, and should be explicitly identified and challenged in therapy. for example, in the context of covid- , the distress of some individuals will be grounded on over-estimating the probability of death from the virus; over-estimates of the fatality risk are commonplace (newton-john et al., ) . however, while standard treatments for anxiety may often involve disproving the client's probability estimates (kirk and rouf, ) , this is not recommended in treatments targeting death anxiety. disproving the individual's estimate of dying from any one particular cause (e.g. falling to one's death, dying in a plane crash, or succumbing to covid- ) only serves to address the proximal threat, and will probably do little to address their fear of their own inevitable death, from one cause or another. as such, it is central instead to focus on addressing the cost of death, rather than merely the probability. clients should be guided to cultivating an attitude of 'neutral acceptance' towards death; that is, an acceptance of death as a universal fact outside of one's control, and therefore neither good nor bad (wong et al., ) . standard cognitive challenging techniques can also be used to challenge unrealistic beliefs surrounding death. for example, for individuals fearing pain associated with dying, corrective information may be provided in the form of information from palliative care, and research indicating that dying is less unpleasant than people typically imagine. notably, theoretical orientations outside of standard cbt may also prove valuable in shifting clients' attitudes towards death. approaches from existential psychotherapy may be particularly relevant, and yalom ( ) outlines many relevant treatment recommendations from an existential lens. for example, for clients who express anxiety surrounding the concept of nonexistence, yalom ( ) recommends the use of the stoic 'symmetry' argument, which proposes that humans have already experienced non-existence, that is, prior to their birth. that is, death 'returns us to that peace in which we reposed before we were born. if someone pities the dead, let him also pity those not yet born' (seneca, ) . these clients may also be encouraged to foster gratitude for ever coming into existence at all, an idea persuasively expressed by richard dawkins, who notes that 'we are going to die, and that makes us the lucky ones', as we have 'won the lottery of birth against all odds' (dawkins, , p. ) . in order to help build identification with this idea, one exercise may involve estimating the likelihood of one's existence, by calculating the probability of one's parents ever meeting, followed by grandparents, and so forth (menzies, ) , in order to help the client focus on the incredible unlikelihood of their own dna sequence ever existing at all, rather than focusing on the tragedy of their own impermanence. the recent covid- pandemic has caused an understandable surge in anxiety across the globe. much of the behavioural response to covid- can be understood through the lens of terror management theory, which argues that death anxiety drives much of human behaviour (greenberg, ) . from this perspective, reminders of death (of which there are many in the current pandemic), produce increases in attempts to avoid a physical death (such as by wearing protective gear or self-isolating) or ensure a symbolic immortality (such as by bolstering one's cultural worldviews, and aggressing against those that threaten them). death anxiety, which has recently been proposed to be a transdiagnostic construct (iverach et al., ) , appears to be more relevant now than ever before. in addition to predicting anxiety related to covid- (newton-john et al., ) , fear of death has also been shown to play a causal role across a number of mental health conditions (menzies and dar-nimrod, ; strachan et al., ) . given this, current standard treatments for mental health conditions may benefit from addressing death anxiety directly, in order to prevent the 'revolving door' often seen in mental health services (iverach et al., , p. ) . fortunately, cbt has been demonstrated to produce significant reductions in death anxiety, with exposure appearing to be particularly effective (menzies et al., ) . complementing current treatments with specific cbt techniques addressing fears of death may help to ensure the best long-term outcomes for clients, and protect the individual from future disorders. however, further research is needed to examine whether treating death anxiety will in fact reduce the likelihood of future mental health problems. acknowledgements. none. financial support. none. conflicts of interest. the authors declare no conflicts of interest. ( ) increasing evidence suggests that death anxiety is a key transdiagnostic construct, and may contribute to various mental health conditions. ( ) standard treatments for a variety of disorders may need to be supplemented with specific treatment targeting death anxiety. ( ) recent evidence demonstrates that death anxiety can be effectively reduced using cbt with a focus on exposure therapy. ( ) we suggest a number of cbt treatment strategies, including cognitive reframing of unhelpful thoughts, and exposure tasks tailored to the feared situations, themes or images the individual avoids. ( ) future research is needed to examine whether directly addressing death anxiety does indeed produce long-term improvement in symptoms, and prevent future disorders. brisbane, australia: australian academic press. the fear of covid- scale: development and initial validation ebola salience, deaththought accessibility, and worldview defense: a terror management theory perspective the denial of death supersize my identity: when thoughts of contracting swine flu boost one's patriotic identity two decades of terror manegement research: a meta-analysis of mortality salience research the relationships among psychological separation, the quality of attachment, separation anxiety and death anxiety a word of caution about many labs : if you fail to follow your preregistered plan, you may fail to find a real effect viewing death on television increases the appeal of advertsied products unweaving the rainbow racism and discrimination in covid- responses how to be free: an ancient guide to the stoic life (a. a. long, trans.) terror mismanagement: evidence that mortality salience exacerbates attentional bias in social anxiety treating health anxiety and fear of death: a practitioner's guide the novel coronavirus (covid- ) outbreak: amplification of public health consequences by media exposure feeling and time: the phenomenology of mood disorders, depressive realism, and existential psychotherapy dying to be thin: the effects of mortality salience and body mass index on restricted eating among women terror management theory: from genesis to revelations terror management theory: implications for understanding prejudice, stereotyping, intergroup conflict, and political attitudes. social and personality psychology compass sympathy for the devil: evidence that reminding whites of their mortality promotes more favorable reactions to white racists assessing the terror management analysis of self-esteem: converging evidence of an anxiety-buffering function get out of your mind and into your life: the new acceptance and commitment therapy beyond the living will multidimensional treatment of fear of death death anxiety and its role in psychopathology: reviewing the status of a transdiagnostic construct the varieties of religious experience currencies as cultural symbols -an existential psychological perspective on reactions of germans toward the euro why are so many baby boomers in denial over the coronavirus? vogue multiculturalism, racism and infectious disease in the global city: the experience of the sars outbreak in toronto fears and phobias: part ii. welfare magazine specific phobias many labs : failure to replicate mortality salience effect with and without original author involvement death anxiety, perfectionism and disordered eating coronavirus anxiety scale: a brief mental health screener for covid- related anxiety the collett-lester fear of death scale: the original version and a revision self-control moderates the association between perceived severity of the coronavirus disease (covid- ) and mental health problems among the chinese public social reaction toward the novel coronavirus (covid- ). social health and behavior internet searches for unproven covid- therapies in the united states death anxiety as a predictor of posttraumatic stress levels among individuals with spinal cord injuries terror management and aggression: evidence that mortality salience motivates aggression against worldview-threatening others cognitive and behavioural procedures for the treatment of death anxiety impermanence and the human dilemma: observations across the ages death anxiety and its relationship with obsessive-compulsive disorder the relationship between death anxiety and severity of mental illnesses the effect of mortality salience on body scanning behaviours in mental illnesses creative approaches to treating the dread of death the effects of psychosocial interventions on death anxiety: a meta-analysis and systematic review of randomised controlled trials the dread of death and its role in psychopathology the role of death fears in obsessive-compulsive disorder psychological distress and covid- : estimations of threat and the relationship with death anxiety hypochondriasis and fear of death investigation of death anxiety among depressive patients a dual-process model of defense against conscious and unconscious death-related thoughts: an extension of terror management theory a time to tan: proximal and distal effects of mortality salience on sun exposure intents stereotypes and terror management: evidence that mortality salience enhances stereotypic thinking and preferences how to die: an ancient guide to the end of life terror management and meaning: evidence that the opportunity to defend the worldview in response to mortality salience increases the meaningfulness of life in the mildly depressed body as the source of threat and fear of death in hypochondriasis and panic disorder terror mismanagement: evidence that mortality salience exacerbates phobic and compulsive behaviours the impact of mortality salience on reckless driving: a test of terror management mechanisms beliefs about self, life, and death: testing aspects of a comprehensive model of death anxiety and death attitudes ebola as an existential threat? experimentally-primed ebola reminders intensify national-security concerns among extrinsically religious individuals immediate psychological responses and associated factors during the initial stage of the coronavirus disease (covid- ) epidemic among the general population in china death attitude profile-revised: a multidimensional measure of attitudes toward death coronavirus disease (covid- ): situation report - existential psychotherapy staring at the sun: overcoming the terror of death a systematic review of the psychometric properties of death anxiety self-report measures death anxiety in the time of covid- : theoretical explanations and clinical implications. the cognitive behaviour therapist key: cord- - j r dd authors: hult, henrik; favero, martina title: estimates of the proportion of sars-cov- infected individuals in sweden date: - - journal: nan doi: nan sha: doc_id: cord_uid: j r dd in this paper a bayesian seir model is studied to estimate the proportion of the population infected with sars-cov- , the virus responsible for covid- . to capture heterogeneity in the population and the effect of interventions to reduce the rate of epidemic spread, the model uses a time-varying contact rate, whose logarithm has a gaussian process prior. a poisson point process is used to model the occurrence of deaths due to covid- and the model is calibrated using data of daily death counts in combination with a snapshot of the the proportion of individuals with an active infection, performed in stockholm in late march. the methodology is applied to regions in sweden. the results show that the estimated proportion of the population who has been infected is around . % in stockholm, by - - , and ranges between . % - . % in the other investigated regions. in stockholm where the peak of daily death counts is likely behind us, parameter uncertainty does not heavily influence the expected daily number of deaths, nor the expected cumulative number of deaths. it does, however, impact the estimated cumulative number of infected individuals. in the other regions, where random sampling of the number of active infections is not available, parameter sharing is used to improve estimates, but the parameter uncertainty remains substantial. to understand the spread of the novel coronavirus, sars-cov- , at an aggregate level it is possible to model the dynamic evolution of the epidemic using standard epidemic models. such models include the (stochastic) reed-frost model and more general markov chain models, or the corresponding (deterministic) law of large numbers limits such as the general epidemic model, see [ ] . there is an extensive literature on extensions of the standard epidemic models incorporating various degrees of heterogeneity in the population, e.g. age groups, demographic information, spatial dependence, etc. these additional characteristics make the models more realistic. for instance, it is possible to evaluate the effect of various intervention strategies. more complex models also involve additional parameters that need to be estimated, contributing to a higher degree of parameter uncertainty. a problem when calibrating, even the standard epidemic models, to covid- data is that there are few reliable sources on the number of infected individuals. publicly available sources provide data on the number of positive tests, the number of hospitalizations, the number of icu admission and the number of deaths due to covid- . in some cases, small random samples of an active infection may be available. for example, the swedish folkhälsomyndigheten performed such a test in stockholm with about subjects in early april . moreover, there is still no consensus in the literature on the value of important parameters such as the basic reproduction number r and the infection fatality rate. a useful approach to incorporate the parameter uncertainty in the models is to consider a bayesian framework. in the bayesian approach parameter uncertainty is quantified by prior distributions over the unknown parameters. the impact of observed data, in the form of a likelihood, yields, via bayes' theorem, the posterior distribution, which quantifies the effects of parameter uncertainty. the posterior can be used to construct estimates on the number of infected individuals, predictions on the future occurrence of infections and deaths, as well as uncertainties in such estimates. in this paper an seir epidemic model with time-varying contact rate will be used to model the evolution of the number of susceptible (s), exposed (e), infected (i), and recovered (r) individuals. a time varying contact rate is used to capture heterogeneity in the population, which causes the rate of the spread of the epidemic to vary as the virus spreads through the population. moreover, the time varying contact rate allows modeling the effect of interventions aimed at reducing the rate of epidemic spread. a poisson point process is introduced to model the occurrence and time of deaths. random samples of tests for active infections are treated as binomial trials where the success probability is the proportion of the population in the infectious state. the methods are illustrated on regional data of daily covid- deaths in sweden. it is demonstrated that, by combining the information in the observed number of deaths and random samples of active infections, fairly precise estimates on the number of infected individuals can be given. by assuming that some parameters are identical in several regions, estimates for regions outside stockholm can also be provided, albeit with greater uncertainty. our approach is inspired by [ ] where the authors considers a bayesian approach to model an influenza outbreak. the main extensions include the introduction of the poisson point process to model the occurrence of deaths, the addition of random sampling to test for infection, and an extension to multiple regions. to evaluate the posterior distribution we employ markov chain monte carlo (mcmc) sampling. samples from the posterior are obtained using the hamiltonian monte carlo algorithm, nuts, by [ ] , implemented in the software stan, which is an open source software for mcmc. to model the spread of the epidemic we consider the deterministic seir model [ , ] , which is a simple deterministic model describing the evolution of the number of susceptible, exposed, infected, and recovered individuals in a large homogeneous population with n individuals. the epidemic is modeled by {(s t , e t , i t ), t ≥ }, where s t , e t and i t represent the number of susceptible, exposed and infected individuals at time t, respectively. the total number of recovered and deceased individuals at time t ≥ is always given by n − s t − e t − i t . the epidemic starts from a state s , e , i with s + e + i = n , and proceeds by updating, the parameters are the contact rate β > , the rate ν of transition from the exposed to the infected state and the recovery rate γ > . note that i t represents the number of individuals with an active infection at time t, whereas n − s t is the cumulative number of individuals who have been exposed, and possibly infected, recovered or deceased, up until time t. in the context of the covid- epidemic the contact rate cannot be assumed to be constant, primarily due to interventions implemented in the early stage of the epidemic. moreover, as the seir model describes the evolution at an aggregate level, a time varying contact rate may be used to capture inhomogeneities in the population. if, for example, the epidemic is initiated in a rural area the contact rate may be rather low, but as the epidemic reaches major cities the contact rate will be higher. the resulting seir model with time varying contact rate is given by clearly, one needs to put some restriction on the amount of variation of the contact rate. in this paper a gaussian process prior will be used on the log contact rate, which restricts the amount of variation in time, but is sufficiently flexible to capture the reduction in contact rate after the interventions. when observations on the number of infected and recovered individuals are available, the model ( ) can be fitted to these observations. in the context of covid- , observations on the number of infected and recovered individuals are unavailable. there are many symptomatic individuals who are not tested and potentially a large pool of asymptomatic individuals. in this paper we will rely on the number of registered deaths due to covid- to calibrate the model. in addition we will incorporate the test results from a random sample that provides a snapshot on the number of individuals with an active infection. to model the occurrence of deaths due to covid- we consider the following poisson point process representation. we refer to [ ] for details on poisson point processes. let f denote the infection fatality rate, that is, the probability that an infected individual eventually dies from the infection. consider the number of individuals that enters the infected state on day t, that is, νe t . each such infected individual has probability f to eventually die from the infection. conditional on death due to the infection, the time from infection until death is assumed independent of everything else and follows a probability distribution with probability mass function p s d . each individual that dies may be represented as a point (t, τ ) in e := {(t, τ ) ∈ n : τ ≥ t}, where t denotes the time of entry to the infected state and τ the time of death of the individual. the number of deaths at time τ can then be computed by counting the number of points on the line ∪ τ t= (t, τ ). the number of deaths, and the corresponding time of infection and time of death is conveniently modelled by a poisson point process on e. let ξ be a poisson point process on e with intensity we may interpret a point at (t, τ ) of the poisson point process as the time of infection, t, and the time of death, τ , of an individual who dies from the infection. the number of deaths d τ that occurs at time τ is then given by summing up all the points of the point process on the row corresponding to τ , ξ(∪ τ t= (t, τ )). since the rows are disjoint this implies that d , d , . . . are independent with each d τ having a poisson distribution with parameter throughout this paper p s d is the probability mass function of a negative binomial distribution with mean s d . more precisely, a parametrization of the negative binomial distribution with parameters r, s d will be used, where the value r = will be used throughout as this fits well with the distribution of observed duration from symptoms to death in the study by [ ] . in this section we provide the assumptions on the prior distributions and derive the expression of the likelihood of the model. note that λ τ is a function of all the parameters of the model, θ = ({β t }, ν, γ, s , f, s d ). the parameters, their interpretation and prior distribution are summarized in table . actually, since the contact rate is positive, a gaussian process (gp) prior will be used for the natural logarithm of the contact rate, denoted log-gp in the sequel. the gaussian process has a constant mean µ and a squared-exponential covariance kernel k with parameters α, ρ, δ such that to compute the likelihood the observed number of daily deaths, d , d , . . . , d t , table . specification of the parameters and prior distributions. will be used, in combination with a random sample of n tests for active infection, performed at a time t , when such test result is available. the number of individuals z with positive test result has a bin(n , i t /n ) distribution. the full likelihood is given by: the joint prior is the product of the marginal priors and leads, by bayes's theorem, to the posterior, the expected number of daily deaths λ τ , the cumulative number of deaths and the cumulative number of infected individuals n −s t are all functions of θ and their distribution can therefore be inferred from the posterior p Θ|d,z . by sampling from p Θ|d and iterating the dynamics ( ) estimates of these quantities may be obtained along with the effects of parameter uncertainty. moreover, predictions on the future development of the above mentioned quantities can be obtained by extrapolating the contact rate into the future. as the posterior distribution is unavailable in explicit form it is necessary to employ monte carlo methods. in the next section markov chain monte carlo methods are briefly described to sample from the posterior. . . multiple regions. the seir model ( ) and the derivation of the likelihood ( ) considers a single region. in the context of multiple regions it may be reasonable to assume that some parameters are identical. for example, when considering multiple regions of sweden below it will be assumed that the rate, ν, from exposed to infected, the recovery rate, γ, the infection fatality rate, f , and the duration, s d are identical in all regions. it is tempting to include interaction terms between the regions as infected individuals from one region may travel to another region and cause new infections. in this paper, it will be assumed that each region has its own time varying contact rate that incorporates fluctuations in new infections due to import cases from other regions. the likelihood from multiple regions is simply the product of the marginal likelihood for the individual regions and the prior is the product of the marginal priors for each parameter. thus, for two regions the prior will be the product of two gaussian process priors for the respective log contact rates for the two regions and the product of the marginal priors for the remaining parameters. markov chain monte carlo (mcmc) methods in bayesian analysis aims at sampling from the posterior distribution. this is non-trivial because the marginal distribution of the data, which acts as normalizing constant of the posterior is practically impossible to compute. in mcmc algorithms the posterior is represented as a target distribution. the algorithms rely on the construction of a markov chain whose invariant distribution is the target distribution. standard mcmc methods are based on acceptance-rejection steps, where random proposals are accepted or rejected with a probability that does not require knowledge of the normalizing constant, e.g., metropolis-hasting and gibbs sampling [ , , ] . when the target distribution is complex and multi-modal, standard methods may lead to poor mixing of the markov chain and slow convergence to the target distribution. to overcome slow mixing of the markov chain gradient-based sampling can be applied, which adapt the proposal distribution based on gradients of the target, see e.g. [ ] . in this paper we will employ a hamiltonian monte carlo sampler, the no-u-turn sampler (nuts) by [ ] in combination with automatic differentiation to numerically approximate the gradients [ ] , which is implemented in the open source software stan. in this section the estimates of the number of infected individuals and predictions on the evolution of the number of deaths and number of infected individuals are provided for ten regions of sweden. the epidemic is considered to start on - - and interventions in sweden began on - - . the joint prior distribution is the product of the marginal priors, and the hyper-parameters are specified in table . the choices of hyper-parameter values are made in line with existing literature on the covid- epidemic. as a general principle we have used informative priors on the parameters ν, γ, and s d , whereas the priors on the time-varying contact rate {β t } and the fatality rate f are uninformative. folkhälsomyndigheten reports that the incubation period is usually around days, which corresponds to /ν ≈ . similarly the expected time to recovery is around days, /γ ≈ . the overall infection fatality rate f is estimated to be in the range . − . , see [ , ] . however, since the infection fatality rate is a very important parameter we have used an uninformative prior and simply use a uniform prior, beta( , ). the expected duration from symptoms to death is around days, see [ ] . samples from the posterior are obtained using the nuts-sampler with a burn-in period of samples and samples after burn-in. mean prior %-c.i. individuals performed between - - and - - . it showed that individuals carried the sars-cov- virus. these results are included in the analysis as a binomial sample of size n = and success probability i t /n where the test date, t , is assumed to be - - . a summary of the marginal posterior distributions is provided in table . the posterior distribution of the time varying contact rate is illustrated in figure . note that although there is great uncertainty about the initial contact rate, the model clearly picks up the reduction in contact rate after the interventions began on - - . the contact rate is gradually reduced around the time of intervention and then remains at a low level. this slow reduction of the contact is, however, not due to stiffness of the gaussian process kernel. we have experimented with a sharp break-point in the contact rate at the time of intervention, but it did not provide more accurate results. on the contrary, the data suggests that the reduction of the contact rate is slow. the contact rate is estimated until - - . after this date the posterior is unreliable. this is because many of the deaths of individuals who are infected after - - have not yet been observed. for this reason, the contact rate is only estimated until - - . to perform estimates and predictions on the future number of daily and cumulative infections and deaths, the contact rate has been extrapolated from its value on - - . the posterior distribution suggests that the contact rate is constant, at a low rate, since roughly - - , which motivates extrapolation into the future, assuming that the interventions remains at the present level. after - - the contact rate is extrapolated, by assuming it will remain constant. figure (top left) shows the observed daily number of deaths (black dots) along with the posterior median (dark red) and % credibility interval (red) for the expected number of daily deaths. figure (top right) shows the observed cumulative number of deaths (black dots) along with the posterior median (dark red) and % credibility interval (red) for the expected cumulative number of deaths. we observe that the parameter uncertainty does not substantially impact the expected number of daily deaths and the peak of the daily number of deaths appears to have occurred by mid april. similarly, the expected cumulative number of deaths in stockholm is likely to terminate slightly above . we emphasize that this is the expected number of deaths, λ τ . since we are considering a poisson distribution for the number of daily deaths an approximate %-prediction interval would be λ τ ± √ λ τ , where λ τ is the poisson parameter on day τ . note from the observed number of daily deaths that the empirical distribution of daily deaths appear to be overdispersed, the variance is substantially larger than the mean. this is likely due to reporting of the data. the data presented at https://c .se/ does not correct the reporting of death dates in hindsight. a comparison at the national level with data provided by folkhälsomyndigheten shows that the official records of the daily number of deaths for sweden does not appear to be overdispersed. nevertheless, even after smoothing the data from https://c .se/ by a moving average over a few days, the results of the simulations remain essentially the same. figure (bottom left/right) shows the posterior median (dark red) and % credibility interval (red) for the daily/cumulative number of infected individuals. although the parameter uncertainty has significant impact on the cumulative number of infected individuals, some conclusions are still possible. as of mid may, the cumulative number of infected individuals has almost reached its terminal value and the spread of the epidemic has slowed down significantly. the estimated cumulative number of infected individuals is . % of the population in stockholm. the estimated number of infected individuals by - - is . %, showing that these results are well in line with the reports of the anti-body test performed at kth , which indicated that % of the population in stockholm had developed anti-bodies against the sars-cov- virus by the first weeks of april. we emphasize that the estimate of the cumulative number of infected individuals in stockholm relies heavily on the inclusion of results from the random sampling performed by folkhälsomyndigheten in late march, early april. without this crucial piece of information similar models to the one analyzed here may provide a significantly higher estimate on the cumulative number of infected. table . marginal posterior median and credibility intervals for region stockholm. . . summary of the results for ten regions of sweden. in this section estimates of the cumulative number of infected individuals are provided for the following regions of sweden: ( ) stockholm (population: . · ) ( ) västra götaland (population: . · ) ( )Östergötland (population: . · ) ( )Örebro (population: . · ) ( ) skåne (population: . · ) ( ) jönköping (population: . · ) ( ) sörmland (population: . · ) ( ) västmanland (population: . · ) ( ) uppsala (population: . · ) ( ) dalarna (population: . · ) the daily death counts for the regions of sweden until - - are obtained from the webpage: https://c .se/. there is no random testing providing information on the proportion of infected individuals outside region stockholm. to estimate the contact rate and the cumulative number of infected individuals in regions outside stockholm, we have implemented the multi-region model pairwise, with two regions in each mcmc simulation, where one region is stockholm and the other region is from the list above. it is assumed that the parameters ν, γ, f, and s d are identical in both regions, but the time varying contact rate and the initial proportion of susceptible individuals are different between the regions. the posterior of the contact rates for the different regions are provided in figure table along with % credibility intervals. overall the proportions are low, far from herd immunity. prop observed daily number of deaths (black dots), the posterior median (dark red) and % credibility interval for the expected daily number of deaths. top right: observed cumulative number of deaths (black dots), the posterior median (dark red) and % credibility interval for the expected cumulative number of deaths. bottom left: the posterior median (dark red) and % credibility interval for the daily number of infected individuals. bottom right: the posterior median (dark red) and % credibility interval for the cumulative number of infected individuals. infectious diseases of humans: dynamics and control the geometric foundations of hamiltonian monte carlo basic estimation-prediction techniques for covid- , and a prediction for stockholm contemporary statistical inference for infectious disease models using stan mathematical tools for understanding infectious disease dynamics stochastic relaxation, gibbs distributions, and the bayesian restoration of images evaluating derivatives: principles and techniques of algorithmic differentiation monte carlo sampling methods using markov chains and their applications the no-u-turn sampler: adaptively setting path lengths in hamiltonian monte carlo random measures early transmission dynamics in wuhan, china, of novel coronavirus infected pneumonia substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (sars-cov ) equation of state calculations by fast computing machines estimates of the severity of coronavirus disease : a model-based analysis the lancet infections diseases estimating clinical severity of covid- from the transmission dynamics in wuhan, china key: cord- -d g pbj authors: chen, yifei; zhao, meizhen; wu, yifan; zang, shuang title: epidemiological analysis of the early fatalities in hubei, china, of the coronavirus disease date: - - journal: journal of global health doi: . /jogh- - sha: doc_id: cord_uid: d g pbj background: since the emergence of coronavirus disease (covid- ) in hubei province of china by the end of , it has burned its way across the globe, resulting in a still fast-growing death toll that far exceeded those from severe acute respiratory syndrome (sars) in less than two months. as there is a paucity of evidence on which population is more likely to progress into severe conditions among cases, we looked into the first cluster of death cases, aiming to add to current evidence and reduce panic among the population. methods: we prospectively collected the demographic and clinical data of the first fatalities whose information was made public by the health commission of hubei province and the official weibo account of china central television news center, starting from january through january . the death cases were described from four aspects (gender and age characteristics, underlying diseases, the time course of death, symptoms at the incipience of illness and hospital admission). results: among the fatalities, . % were male, and . % were female, with the median age of years (interquartile range (iqr) = - ). persons aged - made up the largest share. twenty-five cases had a history of chronic diseases. the median time between the first symptoms and death was . days (iqr = . - . ), while the median time between the admission and death was . (iqr = . - . ) days. in persons aged over years, the time between the first symptoms and death decreased with age, and so did the time between the admission and death, though the latter increased again in persons aged over years. the major first symptoms included fever ( . %), cough ( . %), dyspnea ( . %), myalgia and fatigue ( . %). conclusions: among the death cases, persons with underlying diseases and aged over made up the majority. the time between the first symptoms and death decreased with the advanced age. in all the age groups, males dominated the fatalities. viewpoints research theme : covid- pandemic rocketing up, the death toll of the coronavirus disease (covid- ) outbreak has overtaken that of the severe acute respiratory syndrome (sars) during the - epidemic. the covid- outbreak has wreaked havoc on all sectors in china, resulting in city lockdown, traffic restrictions, work shutdown, and school cancellation, etc., first in wuhan, then later in many other cities. many countries have imposed travel restriction, suspended flights, and barred entry of chinese nationals. the sudden shock of the covid- has had a significant impact on the chinese economy [ ] . containment of virus transmission has become a top priority to global public health security. researchers have been racing against time since the outbreak of the covid- , as little was known regarding covid- virus initially. two viral genome studies had indicated that the novel virus is closely related to sars-cov (one research revealing . % and the other . % nucleotide similarity, respectively) [ , ] , which is reminiscent of the calamitous sars outbreak years back. aside from viral genome studies, researchers also looked into the clinical features and epidemiologic characteristics of covid- cases. clinical manifestation of covid- ranges from mild symptoms (low-grade fever, fatigue, sore throat, etc.) that resemble a common cold [ ] , to severe and even fatal respiratory diseases such as acute respiratory distress syndrome [ ] . a study that collected more than cases across china also reported asymptomatic cases of covid- virus infection, accounting for . % of total confirmed cases [ ] . after emergence, the virus spread rapidly through human-to-human transmission [ ] , which was substantiated by a modeling study from los alamos national laboratory indicating the median basic reproductive number (r ) for covid- virus was . ( % confidence interval (ci) = . - . ) [ ] . the mechanism behind the high infectivity of the covid- virus could be explained by a study revealing that covid- virus spike glycoprotein had around to -fold higher affinity with angiotensin converting enzyme ii (ace ) receptor that is widely distributed in human organs, than sars-cov spike glycoprotein [ ] . in addition, the covid- virus spreads mainly from person-to-person contacts via respiratory droplets or contact with infected surfaces or objects [ ] . with a combination of high infectivity and easy transmission, the covid- virus poses a great threat to anyone who has close contacts with an infected person, especially within families and to frontline medical staffers [ ] . to make things worse, transmission from asymptomatic patients was confirmed by a case report covering a german businessman being infected by his asymptomatic chinese business partner from shanghai [ ] . who issued a warning against possible transmission of covid- virus from infected people before they developed symptoms. these findings have raised concerns across the globe, sounded the alarm of a dire situation, and prompted authorities to ramp up quarantine measures. to what extent covid- kills remains vague, as literature in terms of case fatality rate is in scarcity. based on the data compiled by who, the overall case fatality rate of covid- globally was initially estimated at around % [ ] , similar with an overall case fatality rate of . % from a study which collected more than cases in mainland china as of february [ ] . however, the two case fatality rate figures were much lower than the result yielded from the study of wang et al. on a case series of consecutive hospitalized covid- patients (mortality: . %) in a hospital in wuhan, china [ ] . the reason that the study of wang et al. had higher case fatality rate can be attributed to the large scale of the infection in the epicenter of the outbreak (more than cases in early february ), the heavily strained medical system, the lack of protective suits and medical equipment (such as masks, goggles, gloves, and disinfectants). based on the fact that the death toll of covid- topped that of the sars outbreak during - in less than two months, covid- virus infection will deal a more substantial blow to the globe and can be more fear-mongering, than sars. a look into deaths cases may provide more information to the public and sooth panic, as studies suggested that misinformation and inadequate information contribute to unnecessary public panic and subsequent undesirable responses [ , ] . as there is a paucity of evidence on which population is more likely to progress into severe conditions among covid- cases, here, we poured over the first batch of death cases whose information were made public by health commission of hubei province as of january , one day into city lockdown in wuhan, with the purpose to add a new facet to current evidence. data of covid- death cases in hubei were extracted prospectively from the website of health commission of hubei province [ ] and the official weibo (china' s equivalent of twitter) account of china central television news center [ ] , starting from january , when the first deceased patient was reported, through january , when the th was registered. since january the number of death cases has been surging, and the health commission of hubei province has stopped making public the information of death cases. therefore, data collection was terminated at that point. microsoft excel (microsoft, redmond, wa, usa) and spss . software (ibm corp., armonk, il, usa) was used for data analysis. the death cases were described from four aspects (gender and age characteristics, underlying diseases, death time distribution, and symptoms at the incipience of illness and hospital admission). frequencies (%) and median (interquartile ranges [iqr]) were used to describe the data. as of january , the overall case fatality rate for covid- was . % in hubei. among the fatalities, there were males, and females, with a male to female ratio of . : . the youngest age was years, and the oldest age was years, with the median age being years (iqr = - ). the median age for females and males both stood at , though iqr ranged from to for the former, and from to for the latter, respectively. distribution of fatalities by genders and age groups was shown in figure . there were cases aged - years, making up the largest share of . %. coming next was cases aged - , accounting for . %. the same pattern was found for genders, with - years forming the largest share of . % in males, vs . % in females alone, and - years the second largest ( . % in males, vs . % in females alone). among the death cases, had underlying diseases, including males and nine females, accounting for . % of the total. there were cases of hypertension, cases of diabetes, four cases of coronary heart disease, three cases of chronic bronchitis, two cases of cerebral infarction, and two cases of parkinson disease. other diseases included chronic obstructive pulmonary disease, tuberculosis, frequent ventricular premature beats, colon cancer, gallstone, cirrhosis, chronic renal insufficiency, fracture, hip replacement, etc., all keeping a tally of one case, respectively ( table ) . among all the death cases, had one or two underlying diseases, and eight had more than three underlying diseases. among the fatalities, the first case died on january, and the last in the batch died on january , with the period stretching days, during which, the death toll didn't show apparent regularity. to understand the evolution of death, we defined the first symptom day fell on the date on which the patients started to feel their symptoms. we described the period between the first symptom day and the death date as days from the first symptom to death and the period between the date of admission and death as days from admission to death. the median time from the first symptom to death was . days (iqr = . - . ). as for male dead patients, the median time from the first symptom to death was . days (iqr = . - . ), and for the females, the median time from the first symptom to death was . days (iqr = . - . ). the median time from admission to death was . days (iqr = . - . ). as for male dead patients, the median time from admission to death was . days (iqr = . - . ), and for the females, the median time from admission to death was . days (iqr = . - . ). days from the first symptom to death tailed off over the age groups of - , - , - , and > , while the days from admission to death had a similar pattern over the age groups of - , - , - , but rebounded in persons aged over (figure ) . % of the cases died within nine to days since they felt the first symptoms. fever and cough were the main reported symptoms at the onset of illness among the early death cases. twenty patients first complained of a fever, of coughs, of dyspnea, of chest tightness, of myalgia and fatigue, accounting for . %, . %, . %, . %, and . %, respectively. other symptoms included headache, dizziness, chills, and intermittent diarrhea, each keeping a tally of one case. fever and dyspnea were the main reported symptoms at hospital admission among the early death cases. twenty-five patients complained of a fever, of dyspnea, of coughs, of chest tightness, and had the complaints of myalgia and fatigue, accounting for . %, . %, . %, . %, and . %, respectively. other symptoms are shown in table . as of january , the initial overall case fatality rate in hubei province reached . %. later on, newly reported cases in china saw a sharp rise, but the overall case fatality rate has dwindled. as of february, the overall fatality rate in hubei province was . % [ ] , which was far lower than the results of our study. the later declining overall case fatality rate was on one part attributed to the effective treatment of covid- as thousands of medical workers from other parts of china poured into hubei province to aid their fellow workers battling the coronavirus. on the other part, there was viewpoints research theme : covid- pandemic a substantial shortage of test kits at the early stage of the covid- outbreak, making it challenging to identify the infected cases [ ] . afterward, the test kits were supplied in a large amount, making the number of confirmed patients grow significantly. besides, with a continuous flow of medical resources and personnel into the epicenter and the sweeping screening of infected persons in the communities, the infections were identified and admitted to the hospitals (including fangcang shelter hospitals) speedily, reducing the possibility of becoming severe and preventing the widespread of the coronavirus in communities. we discussed the epidemiological characteristics of cases in the early stage of the disease from the following four parts. gender and age characteristics . % of the deaths were male, considerably more than female, which is consistent with the findings of wang w et al. [ ] . single-cell sequencing of covid- virus receptors at tongji university found that asian men were more likely to be infected with covid- virus [ ] , and a study of cases nationwide also found that the death rate of men was more than three times that of women [ ] . the reason that male dominated the fatalities could be explained by the fact that percentage of ace level in men is higher than in women [ ] , rendering men more susceptible to covid- virus. in addition, covid- virus-infected people tend to be older ones [ ] . in a recent lancet article ( february ) [ ] , % of the confirmed cases had chronic underlying diseases, and the median age was . years, indicating that the middle and old aged patients with chronic underlying diseases were more likely to contract the covid- virus. from experience, we can see that patients with chronic underlying diseases are indeed more likely to have disease deterioration or even death. among the death cases, persons with underlying diseases and aged over made up of the majority. hence, we developed a speculation that covid- could worsen in elderly persons with underlying diseases and even more easily progress to death. this is mainly due to the dwindling immunity in the elderly, especially in those with underlying diseases, which directly renders senior people more likely to be in a state of frailty and more vulnerable to infections [ ] , and subsequently leads to worsening of the disease [ ] . among the death cases, persons with hypertension and /or diabetes made up the largest share, which could be explained by the fact that hypertension and diabetes top the chronic disease chart in china [ ] . ace is a crucial regulator of the renin-angiotensin system, and plays a regulatory role in the central regulation of blood pressure and cardiovascular function and could become an attractive target for the treatment of hypertension [ , ] . covid- virus uses the receptor ace to enter into target cells [ ] , precisely the same as sars-cov. turner et al. [ ] found that sars-cov infection affects the function of ace , so we speculate that the covid- virus will also impair the function of ace , and then manipulate the regulation of blood pressure, and have a negative impact on patients with hypertension. on the other hand, hypertension can cause vascular damage. in patients with hypertension, increased vascular stiffness and decreased elasticity are common, followed by vascular remodeling and stenosis [ ] . the pathological results of patients with covid- showed that pulmonary vessels endothelial swelling, luminal stenosis and occlusion, leading to acute lung dysfunction [ ] . the coexistence of hypertension and covid- is a very unfavorable factor to induce lung dysfunction, which is prone to aggravate the condition and even result in death. ace gene can also be expressed in the pancreatic islets. a study showed that the binding of sars-cov to ace damages islets and causes acute diabetes [ ] . covid- virus may also exert such a negative effect on islets through the same mechanism. in persons with preexisting diabetes, the damage of islets by covid- virus could be more severe, and even fatal [ , , ] . in addition, ace is rich in the lungs, heart, kidney, intestine, and testicles, etc. once covid- virus gains entry into the human body, more organs could be attacked by the virus through blood circulation over time [ ] . therefore, early diagnosis of covid- before it progresses into severe conditions is an important measure for older people who have developed a fever and respiratory symptoms [ ] . other measures including reducing chances of exposure to infected cases (eg, banning visits to nursing home residents, avoiding gatherings), early isolation and treatment of symptomatic confirmed cases can be beneficial to the elderly population, especially those with preexisting underlying diseases. viewpoints research theme : in this study, the number of deaths did not show obvious regularity with time within two weeks. in addition, the covid- virus infection rate has been spiking up since january. according to the data of the national health and health commission [ ] and the results from our study, we make the following speculation: the cases gradually became infected around the end of december according to a median -day incubation period and a median . -day period from the first symptom to death [ ] . if we take the later reported maximum incubation period of days into account [ ] , a considerable part of them may have been infected in november. besides, the difference in immune resistance between different genders and ages is also an important reason for the irregular distribution of the time of death [ ] . however, from figure , it can be concluded that with the increase of age, the days from hospital admission to death and the days from the first symptom to death gradually reduced indeed, which shows that covid- poses a great threat to elderly patients [ ] . our study also shows that the days from hospital admission to death rebounded in persons aged over years. since there were only cases aged over years, the finding needs further validation from long-term, large-scale cohort studies. studies have indicated that viral infection in the early stage mainly shows upper respiratory tract infection, manifested as fever, headache, and cough [ ] . huang et al. found a similar result that % of the patients with covid- experienced fever, % had a cough, and % had dyspnea as the first symptom, respectively [ ] . however, among the deaths up to january, . % had a fever, . % had a cough, and . % had dyspnea, as the first symptom. it can be seen that not all infected cases have high body temperature as the first symptom, and the temperature change of old people is not very significant compared with young ones even when they have infectious diseases [ ] . among the severely infected elderly, % ~ % of them have no fever or slow response to fever, which is often a sign of poor prognosis [ ] , and hinders early detection of infection and brings more potential risks to the elderly. therefore, the early repeated examination is a valid response [ ] . however, symptoms changed at the time of admission, with . % of patients showing fever, . % dyspnea, . % cough, indicating dyspnea became the second major symptom. as for the covid- virus infection, severe patients will have chest discomfort, progressive dyspnea, or acute respiratory distress syndrome symptoms [ ] , which indicates the aggravation of the disease. as a result, the proportion of dyspnea symptoms was slightly higher than other symptoms in our deaths. also, although there were cases with limb myalgia and fatigue, headache, and other initial symptoms came in a small quantity, it does not mean that covid- cases presenting the symptoms are in mild condition. there is still the possibility of progression to death, which should arouse the vigilance of medical staff [ ] . the emergence of a new infectious disease poses a particular challenge to epidemiologic research, as identifying the characteristics of the disease and infection prevention and control of an epidemic is a step-by-step process. during the period from to january , the number of confirmed cases of covid- increased -fold [ ] , indicating high infectivity of the novel coronavirus [ ] . such a disease needs to be contained, or at least the spread of it needs to be reined in time. otherwise, the medical system will face enormous pressure, and a large number of infected patients will inevitably die due to the lack of timely treatment. during the covid- outbreak, it is necessary to strengthen the training of medical personnel from all levels of medical institutions, especially those serving at hospitals designated as the treatment center for the disease. at the same time, it is necessary to invest a multitude of resources in outpatient and emergency departments to detect patients to improve the treatment conditions and the capacity to house severe cases. since the elderly and people with underlying diseases are most vulnerable to the attack of coronavirus, and often have serious consequences [ ] , it is urgent to ramp up protection and prevention measures for the elderly, especially those with chronic underlying diseases. it also warns us that in the face of an unknown disease, the protection of vulnerable people is essential. this study has some limitations. first, the data of this study came from the panel data of the official website of the health commission of hubei province, so the clinical information of the cases collected is limited. second, as our study focused on the deaths in the early stage of the outbreak, and the fatalities constitute only a tiny fraction of the overall still-hiking death toll, the specific relationship between male and female, and the variations in the time window from onset to death, and from admission to death among different age groups needs more large-scale studies . the -ncov outbreak joint field epidemiology investigation team, li q. notes from the field: an outbreak of ncip ( -ncov) infection in china -wuhan economic turning point to appear in march despite coronavirus a pneumonia outbreak associated with a new coronavirus of probable bat origin a new coronavirus associated with human respiratory disease in china coronavirus disease (covid- ): symptoms and treatment. clinical manifestation of covid- epidemiological and clinical characteristics of cases of novel coronavirus pneumonia in wuhan, china: a descriptive study chinese center for disease control and prevention. the epidemiologic characteristics of an outbreak of novel coronavirus diseases (covid- ) in china clinical characteristics of coronavirus disease in china high contagiousness and rapid spread of severe acute respiratory syndrome coronavirus . emerg infect dis. . epub ahead of print cryo-em structure of the -ncov spike in the prefusion conformation diagnosis and management of covid- ( th version-revised importation of rare but life-threatening and highly contagious diseases. current situation and outlook transmission of -ncov infection from an asymptomatic contact in germany novel coronavirus ( -ncov) situation report clinical characteristics of hospitalized patients with novel coronavirus-infected pneumonia in wuhan, china. jama. . epub ahead of print developing pandemic communication strategies: preparation without panic what have we learned about communication inequalities during the h n pandemic: a systematic review of the literature briefing on covid- by hubei health commission the official microblog of cctv news center early transmission dynamics in wuhan, china, of novel coronavirus-infected pneumonia updated understanding of the outbreak of novel coronavirus ( -ncov) in wuhan, china single-cell rna expression profiling of ace , the putative receptor of wuhan -ncov. biorxiv. . epub ahead of print epidemiological and clinical features of the novel coronavirus outbreak in china. mdrxiv. . epub ahead of print early epidemiological analysis of the coronavirus disease outbreak based on crowdsourced data: a population-level observational study. lancet. . epub ahead of print the burden of frailty among u.s. veterans and its association with mortality evaluation of mhla-dr expression rate on immune function and prognosis in elderly patients with pneumonia study of the prevalence and disease burden of chronic disease in the elderly in china study on the ace pathway in the mechanism of exercise reduction in hypertension mechanisms of mas receptor-mediated signaling in the vascular endothelium structure of dimeric full-length human ace in complex with b at . biorxiv. . epub ahead of print vascular fibrosis in aging and hypertension: molecular mechanisms and clinical implications clinical pathology of critical patient with novel coronavirus pneumonia (covid- ) ace : from vasopeptidase to sars virus receptor binding of sars coronavirus to its receptor damages islets and causes acute diabetes management suggestions for patients with diabetes and novel coronavirus pneumonia the single-cell rna-seq data analysis on the receptor ace expression reveals the potential risk of different human organs vulnerable to wuhan -ncov infection. front med. . epub ahead of print china news; national wei jian committee. novel coronavirus pneumonia incubation period averages about days be vigilant! days! asymptomatic! enshi diagnosed cases of extra long incubation period expert group on novel coronavirus pneumonia prevention and control of china preventive medicine association. the latest understanding of novel coronavirus pneumonia epidemiology a familial cluster of pneumonia associated with the novel coronavirus indicating person-to-person transmission: a study of a family cluster coronavirus infections and immune responses clinical features of patients infected with novel coronavirus in wuhan watch out for fever-free older adults most of the old people have no fever clinical course and outcomes of critically ill patients with sars-cov- pneu-references monia in wuhan, china: a single-centered, retrospective, observational study single-cell analysis of ace expression in human kidneys and bladders reveals a potential route of -ncov infection. biorxiv. . epub ahead of print novel coronavirus outbreak in wuhan, china, : intense surveillance is vital for preventing sustained transmission in new locations preliminary prediction of the basic reproduction number of the wuhan novel coronavirus -ncov coronaviruses: an overview of their replication and pathogenesis key: cord- -bu vi xz authors: bayes, cristian; rosas, victor sal y; valdivieso, luis title: modelling death rates due to covid- : a bayesian approach date: - - journal: nan doi: nan sha: doc_id: cord_uid: bu vi xz objective: to estimate the number of deaths in peru due to covid- . design: with a priori information obtained from the daily number of deaths due to codiv- in china and data from the peruvian authorities, we constructed a predictive bayesian non-linear model for the number of deaths in peru. exposure: covid- . outcome: number of deaths. results: assuming an intervention level similar to the one implemented in china, the total number of deaths in peru is expected to be ( %ci: . - . ) persons. sixty four days after the first reported death, the % of expected deaths will be observed. the inflexion point in the number of deaths is estimated to be around day ( %ci: . - . ) after the first reported death. conclusion: these estimates can help authorities to monitor the epidemic and implement strategies in order to manage the covid- pandemic. there is a trend to forecast covid- using mathematical models for the probability of moving between states from susceptible to infected, and then to a recovered state or death (sir models) . this approach is however very sensitive to starting assumptions and tend to overestimated the virus reproductive rate. one key point that these models miss is the individual behavioral responses and government-mandated policies that can dramatically influence the course of the epidemic. in wuhan, for instance, strict social distancing was instituted on january rd, , and by march th new infections were close to zero. taking into account this observation, covid et al. ( ) have proposed a statistical approach to model a empirical cumulative population death rate. however, modelling observed cumulative death numbers has the inherent problem of yielding a highly correlated data which, if it is not taken into account, could cause misleading inference results. instead of considering a mathematical model such as sir, that rely heavily on parameters assumptions, we will directly work as in covid et al. ( ) or (zhou et al., ) with empirical data. to this end, we propose to model the daily number of deaths using a poisson distribution with a rate parameter that is proportional to a skew normal density. using a bayesian approach and a prior epidemic china covid- history, we forecast the total number of deaths in peru for the next seventy days. let y t , y t , . . . , y tn be the number of covid- deaths at times t , t , . . . , t n , where time is measured from the first reported death due to covid- , and let us suppose that the death rate will hit a platoon due to the government intervention. we propose then the model y (t i ) ∼ p oisson(λ(t i )), i = , , . . . , n ( ) with death rate where g(t i | α, β, η) denotes the density function of a skew normal distribution with location, scale, and shape parameters, α, β, and η, respectively; p is a maximum asymptotic level parameter and k is the population size. the choice of the skew normal distribution is motivated by its flexibility on the tails and their asymmetry, which can force, as one could expect, a rapidly increase rate at the first stages of the pandemic and a slower decrease of this rate at the last stages. a negative binomial distribution can also be considered for the deaths numbers, but we found empirically a better fit with the poisson model. our model differs of the approach taken by covid et al. ( ) , who directly models the cumulative death rate Λ(t) = t λ(s)ds with a term proportional to a symmetric cumulative normal distribution. this difference is not only found in the formulation, but also in the estimation procedure. while these authors incorporated first the chinese data-more concretely the time from when the initial death rate exceeds e- to the implementation of social distancing-into their model throughout a location-specific inflection point parameter or a maximum death rate and then used a sort of credibility model between short-range and long-rate variants, we followed a bayesian approach that incorporates the chinese data as a prior distribution. the posterior predictive distribution, which is the goal of this model, is then a dynamic object that can be updated with new data about the number of deaths or any other reliable information that may be incorporated as covariates in the model. apart from the total number of deaths, three main quantities of interest can be easily derived from our model. first, a time to threshold death rate, which provides the time after which only . % of deaths will be observed on the population. this will be defined as the . quantile of the g distribution. another characteristic of interest is the inflection point, defined as the time at which the death rate reaches its maximum level. since china was the first country to have experienced a drastic drop in infections and deaths, we are proposing to incorporate this data, into our peruvian death rate predictions, through a prior distribution. figure shows the empirical distribution of number of reported deaths in china. one can notice that the reported numbers on february and (red points) were and deaths, respectively. there is some controversy about the reported numbers from china on these days, the reason why we are considering the average number for these days. figure shows the observed and predicted death rates (a) and the daily number of deaths (b) in china. the predicted rates and their associated % prediction credible intervals were obtained, under a bayesian approach, by considering a non-informative prior and the chinese official death reports. table summarizes the information given in figure . this information will be considered as a prior for modelling the number of deaths in perú with the exception of the p parameter, where a weakly informative prior, n ( , ), will be considered for log (p). initial values for the estimation process will be sampled from the prior distribution that was constructed with the chinese data. the chinese data to be used in this work was obtained from the european centre for disease prevention and control, institution that daily publishes statistics on the covid- pandemic. the daily death reports in perú, on the other hand, were obtained from the local authorities. taking into account the observed likelihood function, easily derived from ( ) and ( ) we now briefly describe our predictions for the spread of covid- in perú. the time here will be understood to be measured in days after the first reported death in the country. in addition, figure .b shows the expected number of deaths per day and its associated % predictive credibility interval. in particular, the expected number of deaths is . ( %ci: - ) and . ( % ci: - ) at and days, respectively. overall, the total number of deaths is expected to be . ( %ci: . - . ) persons. the model estimates that the number of days, since the first death, to achieve the threshold death rate will be . ( %ci: . - . ) days. at this time, the % of expected deaths will have been observed. the estimated inflection point is . ( %ci: . - . ), which means that we expect to spend approximately days before the covid- death rate starts to decline after the first reported death case. three different scenarios were considered, each keeping the weakly informative prior lognormal distribution for p. in all cases, the posterior mean for the fitted chinese model was taken as the mean prior for the modelling of data from perú. the first scenario (i) takes also the posterior chinese variances as the prior peruvian variances. the second scenario (ii) induces flexibility in the prior distribution by increasing their corresponding standard deviations by a factor of five. finally, the third scenario (iii) increases the standard deviations by a factor of ten. table and figure shows that the point estimates are not heavily affected, but the precision pays the price of not having yet enough data from perú. considering the information on daily number of deaths from china and from perú, our estimation of total number of deaths will be . ( % ci: . - . ) and % of those deaths will occur . ( % ci: . - . ) days after the first death reported case. the present study has some limitations. although perú has not followed all the measures taken in china, we assumed that peruvian interventions will have similar effects as the ones observed in that country. however, we expect our model is flexible enough to be driven by the peruvian data. furthermore, the model will increase its precision as more data from perú becomes available. several extensions are possible for the model. for instance, covariates as daily social mobility indicators or country age distributions can be included in the model, in particular on the p parameter that measures the total number of deaths. we expect our model can be useful to guide some policies that need to be taken by the peruvian government in order to overcome the covid- pandemic. for example, the proposed model can be useful to measure the impact of the covid- pandemic on the we would like to thanks rodrigo carrillo larco for providing us the detailed information of stan: a probabilistic program- ming language forecasting covid- impact on hospital bed-days, icu-days, ventilator-days and deaths by us state in the next months rstan: the r interface to stan forecasting the worldwide spread of covid- based on logistic model and seir model key: cord- -zjnr vwm authors: altmejd, adam; rocklov, joacim; wallin, jonas title: nowcasting covid- statistics reported withdelay: a case-study of sweden date: - - journal: nan doi: nan sha: doc_id: cord_uid: zjnr vwm the new corona virus disease -- covid- -- is rapidly spreading through the world. the availability of unbiased timely statistics of trends in disease events are a key to effective responses. but due to reporting delays, the most recently reported numbers are frequently underestimating of the total number of infections, hospitalizations and deaths creating an illusion of a downward trend. here we describe a statistical methodology for predicting true daily quantities and their uncertainty, estimated using historical reporting delays. the methodology takes into account the observed distribution pattern of the lag. it is derived from the removal method, a well-established estimation framework in the field of ecology. the new corona virus pandemic is affecting societies all around the world. as countries are challenged to control and fight back, they are in need of timely, unbiased, data for monitoring trends and making fast and well-informed decisions (nature, ) . official statistics are usually reported with long delay after thorough verification, but in the midst of a deadly pandemic, real time data is of critical importance for policymakers (jajosky and groseclose, ) . the latest data are often not finalized, but change as new information is reported. in fact, reporting delays make the most recent days have the least cases accounted for, producing a dangerous illusion of an always improving outlook. still, these unfinished statistics offer crucial information. if the pandemic is indeed slowing, we should not wait for the data to be finalized before using it. rather, we argue that actual case counts and deaths should be nowcasted to account for reporting delay, thus allowing policymakers to use the latest numbers availiable without beinig misled by reporting bias. such predictions provide an additional feature that is perhaps even more important. they explicitly model the uncertainty about these unknown quantities, ensuring that all users of these data have the same view of the current state of the epidemic. in this paper we describe a statistical methodology for nowcasting the epidemic statistics, such as hospitalizations or deaths, and their degrees of uncertainty, based on the daily reported event frequency and the observed distribution pattern of reporting delays. the prediction model is building on methods developed in ecology, referred to as the "removal method" (pollock, ) . to help motivate why such forecasting is needed, we now turn to the case of sweden. the model is flexible by design, however, and could easily be applied to other countries as well. the swedish public health agency updates the covid- statistics daily . during a press conference, they present updates on the number of deaths, admissions to hospitals and intensive care, as well as case counts. one of the reasons for following these indicators is to enable public health professionals and the public to observe the evolving patterns of the epidemic (anderson et al., ) . in relation to policy, it is of specific interest to understand if the growth rates changes, which could indicate the need for a policy response. however, in each daily report only a proportion of the number of recent deaths is yet known, and this bias produces the illusion of a downward trend. the death counts suffer from the longest reporting delay. in their daily press conference, the swedish public health agency warns for this by stopping the reported -day moving average trend line days before the latest date. but not only are deaths often reported far further back than days, a bar plot still shows the latest information, creating a sense of a downward trend. in fact, this might be the reason why the number of daily deaths have been underestimated repeatedly. at the peak, deaths were initially believed to level out at around per day, but after all cases had been reported more than two weeks later, the actual number was close to (Öhman and gagliano, ) . we propose to use the removal method, developed in animal management (pollock, ) , to present an estimate of the actual frequencies at a given day and their uncertainty. the method has a long history dating back at least to the s (leslie and davis, ) . however, the first refined mathematical treatment of the method is credited to moran ( ) , more modern derivatives exits today (matechou et al., ) . it is a commonly applied method today when analyzing age cohorts in fishery and wildlife management. the removal method that has three major advantages over simply reporting moving averages: • it does not relay any previous trend in the data, • we can generate prediction intervals for the uncertainty about daily true frequencies, the data is published on https://www.folkhalsomyndigheten. se/smittskydd-beredskap/utbrott/aktuella-utbrott/covid- / bekraftade-fall-i-sverige/. • the uncertainty estimates can be carried over to epidemiological models to help create more realistic models. a classic example where the method proposed to solve this problem has been used is in estimating statistics of trapping a closed population of animals (pollock, ) . each day the trapped animals are collected, and kept, and if there is no immigration the number of trapped animals the following days will, on average, decline. this pattern of declining number of trapped animals allows one to draw inference of the underlying population size. here we replace the animal population with the true number of deaths on a given day. instead of traps we have the new reports of covid- events. as the number of new reported deaths for a given day declines, we can draw inference on how many actually died that day. if we assume that the reporting structure is constant over time we can after a while quickly get good estimate of the actual number. suppose for example that on day one, individuals are reported dead for that day. on the second day, deaths are recorded for day two. then, with no further information, it is reasonable to assume that more people died on day two. if the proportion reported on the first day is %, the actual number of deaths would be for day one and for day two. if additionally, deaths are reported during the second day to have happened during day one, and on the third day, only are reported for day two, we now have conflicting information. from the first-day reports it seemed like more people had died during day two, but the second day-reports gave the opposite indication. the model we propose systematically deals with such data, and handles many other sources of systematic variation in reporting delay. in fact, the swedish reporting lag follows a calendar pattern. the number of events reported during weekends is much smaller. to account for this, we allow the estimated proportions of daily reported cases to follow a probability distribution taking into consideration what type of day it is. we propose a bayesian version of the removal model that assumes an overdispersed binomial distribution for the daily observations of deaths in sweden in covid- . we then calculate the posterior distribution, prediction median and % prediction intervals of the expected deaths from the reported deaths on each specific day. the method and algorithm is thoroughly described in the supplementary information. to get accurate estimates we apply two institution-specific corrections. first, we only count workdays as constituting reporting delay, as very few deaths are reported during weekends. second, we apply a constant bias correction to account for the fact that swedish deaths come from two distinct populations with different trends: deaths in hospitals, and in elderly care. in figure we apply the model to the latest statistics from sweden. the graph shows reported and predicted deaths (with uncertainty intervals) as bars, and a dashed line plots the -day (centered) moving average. a version without predictions is used in the public health agencyś daily press briefings. as expected, the model provides estimates of actual deaths considerably above the reported number of deaths. not how the model predicts additional deaths above the moving average line. to judge whether or not the model is accurate we need to compare it to a benchmark. the moving average of reported deaths is not useful, since it is biased for deaths that occurred within the last week. instead, we create a benchmark prediction by a normal distribution where the mean and standard deviation is taken from the historical lags from the last two weeks to the reported numbers . figure depicts four randomly chosen dates where the model is compared to the benchmark. the model and the benchmark are tasked with predicting the total number of individuals who have died at a given date and have been reported within days of that date. as time progresses, more deaths are reported and the dashed grey line approaches the horizontal line. meanwhile model uncertainty decreases. figure shows model performance compared to the benchmark for three difference performance metrics. all three graphs are based on predictions of reported deaths within days, and show how performance increases as more data has been reported. each data point is the average of all dates where predictions can be evaluated. scrps is a measure of accuracy that rewards precision, it is a proper scoring rule like the continuous probability rank score or the brier score (see definition in appendix) (bolin and wallin, ). the central plot shows the width of the prediction intervals, and the rightmost one the proportion of piś that cover the true value. benchmark and model point estimates are similarly close to the truth. the model produces tighter prediction intervals. for - days of reporting lag (see figure ), the intervals are too tight. this is likely because the public health agency queries the swedish death registry for covid- deaths only once or twice a week. since we do not know the process, it has not been explicitly modeled. the model proposed here can estimate the trends in surveillance data with reporting delays, such as the daily covid- reports in sweden. to generate accurate estimates of the actual event frequencies based on these reports is highly relevant and can have large implications for interpretations of the trends and evolution of disease outbreaks. in sweden, delays are considerable and exhibit a weekday and holiday pattern that need to be accounted for to draw conclusions from the data. the method and algorithm proposed overcomes major shortcomings in the daily interpretation and practice analyzing and controlling the novel corona virus pandemic. it also provides valuable measures of uncertainty around these estimates, showing users how large the range of possible outcomes can be. whenever case statistics are collected from multiple sources and attributed to its actual event date in the middle of a public health emergency, similar reporting delays to the ones in sweden will necessarily occur. the method described thus has implications and value beyond sweden, for any situation where nowcasts of disease event frequencies are of relevance to public health. nevertheless, the method also has its limitations. as presented, the model assumes that all deaths are reported in the same manner. given there exists many regions in sweden this is unlikely to be the case. for example, it is easy to see that the swedish region västra götland follows a different reporting structure than stockholm. building a model for each region separately would most likely give better results and make the assumptions more reasonable. unfortunately we do not currently have access to the high resolution data required to do so. moreover, deaths are reported from two distinct populations that seem to follow different trends. at the time of writing, the daily deaths in elderly care, reported with a longer delay, seem to be decreasing slower than hospital deaths. but statistics offer only aggregate numbers, prohibiting us from modeling two distinct processes. however, we have noted a clear decline in proportions of deaths reported the two first working days. for example the number of deaths occurring at the second of april ≈ % of deaths where reported within the first two working days whereas for the eighteens of may only ≈ % where reported during the two first working days. we address this by assuming that the deaths reported during the two first working days comes from a different population then the remainder of days. another limitation is that the model assumes that the number of new reported deaths for a given day cannot be negative, which is not actually true, due to miscount or misclassification of days. the number of such cases is very small, however, and its removal should not make much difference. the central assumption of the model is that the proportions deaths reported each day is fixed (up to the known covariates). if actual reporting standards change over time, the model will not be able to account for this. but reporting likely becomes faster as the crisis infrastructure improves. one can imagine that after a while the reporting improves, or is changed, if this is not accounted for by a covariate in the model, it will report incorrect numbers. of course, there might be unknown variables that we have failed to incorporate, but at the least the model is an improvement from the estimates using moving averages. when the covariates to the reporting delay pattern are known, the model can incorporate them and provide more accurate predictions. in this paper, we provide a method to accurately nowcast daily covid- statistics that are reported with delay. by systematically modelling the delay, policy makers can avoid dangerous illusory downward trends. our model also gives precise uncertainty intervals, making sure users of these statistics are aware of the fast-paced changes that are possible during this pandemic. death date r r · · · · · · r t r · · · · · · r t r · · · r t . . . . . . p ij , i.e. typically in removal sampling one would set the probability of reporting uniform, i.e. p i,j := p. however for this data this is clearly not realistic given weekly patterns in reporting -very little reporting during the weekends. instead we assume that we have k different probabilities. further, to account for overdispertion, we assume that each probability rather being a fixed scalar is a random variable with a beta distribution. the beta distribution has two parameters α and β. this resulting the following distribution for the probabilities ). here, if j ∈ h then day j is a holidays or weekends, and the parameters above are else. these extra parameters are created to account for the under-reporting that occurs during weekend and holidays. finally we add an extra mixture component that allows for very low reporting. for the α and β parameters we use an (improper) uniform prior. for the deaths, d, one could imagine several different prior ideally some sort of epidemiological model. however, here we just assume a log-gaussian cox processes (møller et al., ) , but instead of poisson distribution we use a negative binomial to handle possible over dispersion. the latent gaussian processes has a intrinsic random walk distribution (rue and held, ) i.e. this model is created to create a temporal smoothing between the reported deaths. for the hyperparameter σ we impose a inverse gamma distribution, this prior is suitable here because it guarantees that the process is not constant (σ = ) which we know is not the case. putting the likelihood and priors together we get the following hierarchical bayesian model where where and j ≤ i and i = , . . . , t . as the main goal to generate inference of the number of death d is through the posterior distribution of number of deaths d given the observations r. in order to generate samples from this distribution we use a markov chain monte carlo method (brooks et al., ) . in more detail we use a blocked gibbs sampler, which generates samples in the following sequence: • we sample α, β, α h , β h |d, r using the fact that one can integrate out p in the model, and then d|α, β, α h , β h , r, λ follows a beta-binomial distribution. here to we use an adaptive mala (atchadé, ) to sample from these parameters. • to sample d|α, β, α h , β h , r, λ, that each death, d i is conditionally independent, and we just use a metropolis hastings random walk to sample each one. • to sample λ|d, σ we again use an adaptive mala. • finally we sample σ |d,and p , π directly since this distribution is explicit, and φ using a mh-rw. in this section, we present additional comparison of the model to the benchmark. we first describe the benchmark model in detail. the benchmark model simply takes the sum of average historical reporting lags for the preceding days. as before r ij is the number of deaths that happened on day i and were recorded on day j. to predict the number of people that died on a given day, we first calculate lag averages: wherer i,i+l is the average number of deaths reported with a lag of l days, based on the reports closest preceding day i. if we are looking at data released − − and call this day , the latest death date that we have -day (l = ) reporting lag observation for is r − , . the average for lag( , ) is therefore taken over the days between r − ,− and r − , ( - - and - - ). for this reason, some of the earlier predictions will not have data from days. the average is then taken over all available reports. in the comparisons we aim at predicting the total number of deaths that will have been reported within days of the death date. to do so, we sum over the average lag that has yet to be reported. if we are predicting the number of people that have yet to be reported dead for day - , we already know the true values for r − ,− , r − ,− , r − ,− , and r − , so we only need to predict r − , . . . r − , . the prediction is then benchmark(i, j) = j l=i r i,l + l=jr i,l . ( ) as confidence interval we simply use a normal assumption with standard deviations of the reporting lags, assuming independence, i.e. this is just the square root of the sum of v ar(r). how will country-based mitigation measures influence the course of the covid- epidemic? an adaptive version for the metropolis adjusted langevin algorithm with a truncated drift scale dependence: why the average crps often is inappropriate for ranking probabilistic forecasts handbook of markov chain monte carlo evaluation of reporting timeliness of public health surveillance systems for infectious diseases an attempt to determine the absolute number of rats on a given area open models for removal data log gaussian cox processes a mathematical theory of animal trapping coronavirus: three things all governments and their science advisers must do now antalet virusdöda har underskattats review papers: modeling capture, recapture, and removal statistics for estimation of demographic parameters for fish and wildlife populations: past, present, and future gaussian markov random fields: theory and applications a appendix before presenting the model we describe some notation used through out the appendix. for a m × n matrix r we use the following broadcasting notation r k,j:l = [r k,j , r k,j+ , . . . , r k,l ]. further x|y ∼ π(.) implies that the random variable x if we conditioning on y follows distribution π(.). the relevant variables in the model are the following:variable name dimension descriptionlatent prior parameter for p α h × parameter for the probability, p for holiday adjustment. β h × parameter for the probability, p for holiday adjustment. µ t × µ i is the intensity of the expected number of deaths at day i. σ × variation of the random walk prior for the log intensity. φ × overdispersion parameter for negative binomial distribution. p × probability of reporting for a low reporting event. pi × probability of a low reporting event. the most complex part of our model is the likelihood, i.e. the density of the observations given the parameters. here the data consist the daily report of recorded deaths for the past days. this can conveniently be represented upper triangular matrix, r, where r i,j represents number of new reported deaths for day i reported at day j. this matrix is displayed on the left in table . we assume that given the true number of deaths at day i, d i , that each reported day j the remaining death d i − j− k= r i,k each recored with probability key: cord- -md om x authors: ketcham, scott w.; sedhai, yub raj; miller, h. catherine; bolig, thomas c.; ludwig, amy; co, ivan; claar, dru; mcsparron, jakob i.; prescott, hallie c.; sjoding, michael w. title: causes and characteristics of death in patients with acute hypoxemic respiratory failure and acute respiratory distress syndrome: a retrospective cohort study date: - - journal: crit care doi: . /s - - -w sha: doc_id: cord_uid: md om x background: acute hypoxemic respiratory failure (ahrf) and acute respiratory distress syndrome (ards) are associated with high in-hospital mortality. however, in cohorts of ards patients from the s, patients more commonly died from sepsis or multi-organ failure rather than refractory hypoxemia. given increased attention to lung-protective ventilation and sepsis treatment in the past years, we hypothesized that causes of death may be different among contemporary cohorts. these differences may provide clinicians with insight into targets for future therapeutic interventions. methods: we identified adult patients hospitalized at a single tertiary care center ( – ) with ahrf, defined as pao( )/fio( ) ≤ while receiving invasive mechanical ventilation for > h, who died during hospitalization. ards was adjudicated by multiple physicians using the berlin definition. separate abstractors blinded to ards status collected data on organ dysfunction and withdrawal of life support using a standardized tool. the primary cause of death was defined as the organ system that most directly contributed to death or withdrawal of life support. results: we identified decedents with ahrf, of whom ( %) had ards. the most common primary causes of death were sepsis ( %), pulmonary dysfunction ( %), and neurologic dysfunction ( %). multi-organ failure was present in % at time of death, most commonly due to sepsis ( % of all patients), and % were on significant respiratory support at the time of death. only % of patients had insupportable oxygenation or ventilation. eighty-five percent died following withdrawal of life support. patients with ards more often had pulmonary dysfunction as the primary cause of death ( % vs %; p = . ) and were also more likely to die while requiring significant respiratory support ( % vs %; p < . ). conclusions: in this contemporary cohort of patients with ahrf, the most common primary causes of death were sepsis and pulmonary dysfunction, but few patients had insupportable oxygenation or ventilation. the vast majority of deaths occurred after withdrawal of life support. ards patients were more likely to have pulmonary dysfunction as the primary cause of death and die while requiring significant respiratory support compared to patients without ards. conclusions: in this contemporary cohort of patients with ahrf, the most common primary causes of death were sepsis and pulmonary dysfunction, but few patients had insupportable oxygenation or ventilation. the vast majority of deaths occurred after withdrawal of life support. ards patients were more likely to have pulmonary dysfunction as the primary cause of death and die while requiring significant respiratory support compared to patients without ards. keywords: acute respiratory distress syndrome, acute hypoxemic respiratory failure, mortality, cause of death background acute hypoxemic respiratory failure (ahrf) is among the most common causes of critical illness, with a hospital mortality of approximately % [ ] . in patients meeting the definition of acute respiratory distress syndrome (ards), mortality is approximately % [ ] . however, while ahrf and ards are each defined by severe hypoxemia and associated with high mortality, death due to refractory hypoxemia is reportedly rare. in cohorts of ards patients treated in the s, only - % of deaths were due to refractory hypoxemia, while deaths due to multi-organ failure from sepsis were the cause of up to % of deaths [ ] . these findings suggested that therapies focused on reducing the complications of sepsis would have a greater impact at improving ards survival than therapies for severe hypoxia. since the s, however, cause of death specifically related to organ system dysfunction has not been described despite substantial evolution in critical care practices. ventilator management now focuses on minimizing ventilatorinduced lung injury, as opposed to normalizing oxygenation and ventilation [ ] , which may have led to further reduction in death due to refractory hypoxemia. in addition, there has been growing attention to minimization of sedation, early mobilization, and sepsis recognition and treatment, the latter of which may mitigate mortality due to sepsis [ ] [ ] [ ] [ ] . finally, there has been an increased focus on palliative care in the intensive care unit (icu), which may lead to earlier treatment limitations [ ] [ ] [ ] . because of these changes in practice and how they may affect causes of death in the icu, we hypothesized that causes of death among ahrf and ards patients may be different from historical cohorts. an updated understanding of the causes of death in these populations would help identify the most important targets for new therapies and help direct future investigation to improve survival. we sought to determine the causes and circumstances of death in a contemporary cohort of ahrf patients, and assess whether causes of death differed among patients with and without ards. we performed a retrospective cohort study of adult patients (aged ≥ years) hospitalized at michigan medicine (january , , to december , ) with ahrf who experienced in-hospital death. patients were identified via an electronic query tool of the electronic health record. as in prior studies [ , ] , patients were defined as having ahrf when the following criteria were met: ( ) receipt of invasive mechanical ventilation for at least h (to exclude routine post-operative ventilation) in the medical, surgical, cardiac, trauma, or neurologic icu, and ( ) a pao /fio ratio ≤ . lowtidal volume ventilation and protocols for daily awakening and spontaneous breathing trials for mechanically ventilated patients were employed [ ] . demographics, comorbidities, highest sequential organ failure assessment (sofa) score within the first h of ahrf onset, the lowest glasgow coma scale during the h prior to death, and icu setting were also collected from the electronic health record through use of the electronic query tool. patients were classified as having ards by multiple physician adjudication as part of a prior study [ ] . specifically, two critical-care trained physicians reviewed each ahrf hospitalization to determine whether patients met berlin criteria [ , ] for ards: ( ) new or worsening respiratory symptoms began within week of a known clinical insult, ( ) pao /fio ≤ while receiving a positive end-expiratory pressure ≥ cm h o, ( ) bilateral opacities on chest x-ray, ( ) unlikely to be cardiogenic pulmonary edema, and ( ) no other explanation for these findings. disagreement between physicians was resolved by a third physician in % of patients [ ] . in addition to ards status, specific ahrf or ards risk factors were collected as part of the prior study (pneumonia, aspiration, non-pulmonary sepsis, non-cardiogenic shock, major trauma, major surgery, transfusion, pancreatitis, major burn, inhalation injury, vasculitis, pulmonary contusion, drowning, or none) [ ] . patients transferred from another hospital were excluded as we were unable to reliably determine ards status, ahrf risk factors, or illness severity on presentation. patient data were reviewed by one of internal medicine-trained physicians who did not participate in the adjudication of ards and were blinded to adjudicated ards status. data regarding causes and circumstances of death were collected using a structured abstraction form (appendix , online supplement). specifically, we abstracted presence and severity of sepsis, presence and severity of organ system dysfunction, withdrawal of life-sustaining treatments, and cause of death, as described further below. all data required for abstractions were available in the electronic medical record. to ensure consistency across reviewers, excellent inter-rater reliability was demonstrated on an initial test set of charts (appendix , online supplement). for each patient, we assessed for sepsis and dysfunction of organ systems during the h prior to death. we classified sepsis and each organ dysfunction as severe or irreversible using definitions from a prior study by stapleton et al. [ ] , with the following changes (table ) . we changed the sepsis definition to align with sepsis- (appendix , online supplement). in addition, we changed the definition of severe pulmonary dysfunction from specific diagnoses (ards, bilobar pneumonia, bronchopleural fistula, or pulmonary embolism) to receipt of significant respiratory support (high-flow oxygen, invasive mechanical ventilation, or non-invasive positive-pressure ventilation) to better capture patients with severe pulmonary dysfunction. if a patient underwent withdrawal of life support before meeting any of the objective organ dysfunction criteria outlined in table , abstractors were instructed to assign irreversible dysfunction to the organ system primarily responsible for the decision to withdraw life support in order to accurately capture cause of death (appendix , online supplement). finally, as in stapleton et al., we defined multi-organ failure as organ dysfunction in at least two organ systems [ ] . for each patient, we assessed ( ) the primary organ system responsible for death, ( ) whether death was related to progression of an initial ahrf risk factor or a complication after ahrf, and ( ) whether withdrawal of life support occurred prior to death. the primary organ system responsible for death was defined as the organ dysfunction ( table ) that most directly resulted in the patient's death or the decision to withdraw life support (appendix , online supplement). for patients with a primary cause of death other than pulmonary dysfunction, cause of death was further classified as being due to progression of an ahrf risk factor (e.g., sepsis, aspiration) or a complication that arose after ahrf onset (appendix , online supplement). withdrawal of life support was determined from clinical documentation of intent to withdraw life support and/or not escalate life support in the event of clinical decompensation and subsequent removal or nonescalation of life-sustaining interventions. we present data as numbers (proportions) or medians (inter-quartile range). we compared characteristics of ards vs non-ards patients using chi-square and kruskal-wallis tests and considered p < . to be significant. data analysis was completed in r. the study was deemed exempt by the institutional review board since all patients were deceased. we identified adult patients with ahrf who died during a hospitalization in - , of whom ( %) had ards. the cohort was a median age of years ( - ), % female, % white, and had a median sofa score of ( ) ( ) ( ) ( ) ( ) at ahrf onset. most patients were admitted to a medical icu ( %). patients had a median of ( - ) risk factors for ahrf, most commonly non-cardiogenic shock ( % of patients), transfusion ( %), sepsis ( %), and pneumonia ( %, table ). patients with ards had a higher median sofa score within the first h of ahrf onset ( vs ; p = . ) and had higher prevalence of pneumonia ( % vs %; p < . ), aspiration ( % vs %; p = . ), and noncardiogenic shock ( % vs %; p < . ) compared to patients who did not meet the berlin definition of ards (table ) . among the patients, there were occurrences of organ system dysfunction in the h prior to death (etable , online supplement). there were ( . %) patients that had multiple organ systems with irreversible dysfunction. the most common organ system dysfunctions were pulmonary ( %), neurologic ( %), and cardiac ( %). sepsis was present in ( %) patients and patients ( %) had multi-organ failure prior to death. however, irreversible pulmonary dysfunction was only present in ( %) patients (table )- ( % of all patients) with insupportable oxygenation or ventilation, and patients with withdrawal of life support because of a poor pulmonary prognosis. patients with ards higher rates of sepsis ( % vs %; p < . ), pulmonary dysfunction ( % vs %; p < . ), irreversible pulmonary dysfunction ( % vs %; p = . ), and hematologic dysfunction ( % vs %; p = . ) compared to patients without ards. overall, the most common primary causes of death were sepsis ( %), pulmonary dysfunction ( %), and neurologic dysfunction ( %, fig. ). among the patients whose primary cause of death was not pulmonary dysfunction, ( % of all patients) died primarily due to progression of an ahrf risk factor and ( %) died primarily due to complications that arose after the onset of ahrf (table ). cause of death by icu setting can be found in etable in the supplementary appendix, with some variation in causes of death noted. ards patients were more likely to have a primary cause of death due to pulmonary dysfunction ( % vs %; p = . ) compared to patients without ards and less likely to have a primary cause of death from cardiac dysfunction ( % vs %; p = . , table ). in addition, ards patients were also more likely to die while receiving substantial respiratory support ( % vs %; p < . ). the majority of patients ( %) died after withdrawal of life support. the proportion of deaths that occurred after withdrawal of life support did not differ between patients with and without ards ( % vs %; p = . , table ). in this contemporary cohort of adult patients with ahrf, the most common primary causes of death were sepsis, pulmonary dysfunction, and neurologic dysfunction. the majority of patients had multi-organ failure prior to death, most commonly due to sepsis. more than half of patients were receiving substantial respiratory support at the time of death and the vast majority of patients died after withdrawal of life support. sepsis and pulmonary dysfunction were the top two primary causes of death among both patients with and without ards. our study is consistent with prior reports indicating that sepsis is the leading cause of death among patients with respiratory failure. stapleton et al. found that sepsis was the most common cause of death in ards patients pulmonary°inability to liberate from mechanical ventilation, non-invasive ventilation, or heated high flow nasal cannula due to inadequate oxygenation or ventilation without aforementioned support insupportable oxygenation or ventilation defined as pao < mmhg on fio - . for > h or respiratory acidosis with ph < . on maximum ventilator settings ∞ . option was given to apply irreversible dysfunction if care was withdrawn due to poor prognosis related to pulmonary organ system dysfunction. either cardiac output < . l/min/m or documented cardiogenic shock or reversible ventricular fibrillation or asystole cardiogenic shock or arrhythmia not responsive to treatment. option was given to apply irreversible dysfunction if care was withdrawn due to poor prognosis related to cardiac organ system dysfunction. glasgow coma scale < for ≥ days meets brain death criteria. option was given to apply irreversible dysfunction if care was withdrawn due to poor prognosis related to neurologic organ system dysfunction. microvascular bleeding with either fibrinogen < mg/dl, prothrombin time and partial thromboplastin time > . times control, or platelets < , /μl ongoing microvascular bleeding not surgically correctable with map < mmhg not reversible with blood products. option was given to apply irreversible dysfunction if care was withdrawn due to poor prognosis related to hematologic organ system dysfunction. hemorrhage map < mmhg for > h (or requiring vasopressors) necessitating blood transfusions and excluding other causes of hypotension uncontrollable "surgical" bleeding from a non-microvascular source. option was given to apply irreversible dysfunction if care was withdrawn due to poor prognosis related to hemorrhage. hepatic bilirubin > . mg/dl and albumin < . g/dl and prothrombin time or partial thromboplastin time > . times control severe criteria plus hepatic encephalopathy and/or hepatorenal syndrome not responsive to treatment. option was given to apply irreversible dysfunction if care was withdrawn due to poor prognosis related to hepatic organ system dysfunction. gastrointestinal resectable ruptured or necrotic bowel, or pancreatitis causing shock (map < mmhg for > h or requiring vasopressors) inoperable ruptured or necrotic bowel or pancreatitis causing irreversible shock. option was given to apply irreversible dysfunction if care was withdrawn due to poor prognosis related to gastrointestinal organ system dysfunction. either creatinine > . mg/dl or requiring hemodialysis renal failure with acidosis, hyperkalemia, and/or hypercalcemia causing irreversible cardiac arrest. option was given to apply irreversible dysfunction if care was withdrawn due to poor prognosis related to renal organ system dysfunction. *definition of sepsis changed to reflect current practices. please see appendix , online supplement for previous definition of severe and irreversible sepsis syndrome°d efinition of severe pulmonary organ system dysfunction changed to reflect current practices. previously defined by stapleton et al. as "[acute respiratory distress syndrome], bilobar pneumonia, bronchopleural fistula, or pulmonary embolism documented by high-probability ventilation/perfusion scan or pulmonary angiogram" ∞ pao arterial partial pressure of oxygen, fio fraction of inspired oxygen † blood pressure parameters previously described by stapleton et al. as "hypotension" for irreversible hematologic organ system dysfunction or "systolic bp < " for severe hemorrhagic and gi organ system dysfunction changed to "map < mmhg" treated in the s [ ] . despite increased attention to earlier identification and treatment of sepsis in the intervening decades [ , ] , our study found that sepsis remained the most common cause of death in ahrf patients. this is consistent with recent studies showing that sepsis is the leading contributor to death among patients hospitalized for any cause [ ] . sepsis was slightly more common in patients with ards than those without ards, which may reflect the higher rates of pneumonia and sepsis as risk factors for ards. however, it may also suggest that ards patients are at a heightened risk for secondary infections compared to patients without ards. these findings suggest that therapies targeting sepsis-induced multi-organ dysfunction may have the greatest impact on survival among ahrf patients. we found only small differences in the causes and circumstances of death among ahrf patients with and without ards. patients with ards were more likely to have a pulmonary dysfunction as the primary cause of death and more likely to die while receiving substantial pulmonary support than patients without ards. this indicates that the berlin ards definition identifies a subset of patients with ahrf who are more likely to die directly from respiratory failure and would benefit from therapies to enhance resolution of respiratory failure. however, the difference in rates of pulmonary dysfunction as the primary cause of death was relatively small among patients with and without ards. our study confirms the findings in prior studies indicating that insupportable oxygenation and/or ventilation is rare among patients with respiratory failure. one of the major findings of stapleton et al.'s study was the relatively low proportion of deaths due to insupportable ahrf acute hypoxemic respiratory failure, ards acute respiratory distress syndrome *sofa sequential organ failure assessment. represents the highest sofa score within the first h of ahrf onset † other risk factors for ards/ahrf, each present in < % of the cohort, include major trauma ( %), major surgery ( %), pulmonary contusion ( %), pancreatitis ( %), major burn ( %), inhalation injury ( %), vasculitis (< %), or drowning ( %) oxygenation or ventilation, occurring in only - % [ ] . given the increased awareness and effort to treat sepsis in the period after this original study, we hypothesized that pulmonary dysfunction may be a more common primary cause of death in a contemporary ahrf cohort. however, we found that only % of patients had pulmonary dysfunction as the primary cause of death, and only a handful of patients ( %) had insupportable oxygenation and/or ventilation. there are several potential explanations for these findings. first, with more consistent use of lung protective ventilation, contemporary ahrf patients may be less likely to develop ventilator induced lung injury and progressive respiratory failure [ ] . second, patients with severe ards may be more likely to be initiated on extra-corporeal membrane oxygen therapy prior to developing refractory pulmonary dysfunction [ ] . finally, other strategies such as prone positioning may prevent refractory hypoxemia [ ] . however, these hypotheses do not explain why a similar proportion of patients still ultimately die from respiratory failure despite not developing insupportable oxygenation and/or ventilation. while some patients may be supported through the initial phase of their respiratory failure, eventually life support is withdrawn when providers are unable to completely reverse their need for significant respiratory support. our study also highlights the increasing proportion of deaths that occur after a decision to withdraw or not escalate life support. stapleton et al. showed that from to , the proportion of ards deaths that occurred after withdrawal of life support rose from to % [ ] . similar trends have been reported for all-cause critically ill patients during this time period [ ] . our study suggests that this trend has continued, as we report that % of all deaths among ahrf are now occurring after a decision to withdraw or not escalate life support. our finding is also consistent with a recent study showing that % of deaths among critically ill patients treated in europe from to occurred in the setting of treatment limitations [ ] . there are likely several explanations for why a growing proportion of deaths occur after withdraw of life support. stapleton et al. hypothesized that icu clinicians have earlier and more frequent goals-of-care discussions [ ] , as is recommended in various clinical practice guidelines [ ] . indeed, early multidisciplinary meetings with patients and families may lead to an earlier transition to palliative care among patients likely to die [ , ] . more recently, there has been increased emphasis on family involvement in icu decision-making and treatment planning, for example, as recommended in the abcdef treatment bundle [ ] . overall, the greater emphasis on family involvement in early shared decision making may contribute to earlier transitions to palliation among patients who ultimately die in the icu [ ] . our study has several limitations. first, as a singlecenter study, it is possible that it may be lacking generalizability. however, we examined all deaths among patients with ahrf over a -year period who were treated in distinct icus with different practice patterns. as such, we believe these findings are more broadly applicable. second, while we tried to harmonize our study definitions to those of stapleton et al. to facilitate cross-study comparisons, some changes had to be made to account for interval changes in definitions (e.g., sepsis) and treatments (e.g., high-flow oxygen). we limited deviations in study definitions to those deemed absolutely necessary to reflect the current state of icu practice. third, patients were classified as having undergone withdrawal of life support regardless of the time lag between withdrawal and death. for patients in whom only minutes elapsed between withdrawal of support and death, death may be more accurately representative of the cessation of medical interventions due to futility. however, our approach for determining rates of withdrawal and the rates of withdrawal we observed are consistent with prior reports [ ] . fourth, given a high rate of withdrawal of life support, the most proximate cause of death is cessation of support. however, our methodology identifies which organ dysfunction or syndrome most directly led to that decision, thereby reflecting the primary pathophysiologic cause of death. fifth, there may be some subjectivity to assigning cause of death. however, we developed a standardized approach to assess causes of death based on the presence of irreversible and severe organ dysfunctions and confirmed excellent interrater reliability in identifying the primary cause of death among reviewers, which serves to strengthen the validity of our methodology. furthermore, chart review was performed by physicians only, as medical training may limit the subjectivity in identifying cause of death. in this contemporary cohort study of patients who died after ahrf, the most common primary causes of death were sepsis and pulmonary dysfunction. few patients had insupportable oxygenation or ventilation, but most received substantial respiratory support in the h prior to death. the vast majority of deaths occurred after a decision to withdraw or not escalate life support. patients with ards were more likely to have a primary cause of death of pulmonary dysfunction and to receive substantial respiratory support during the h prior to death. supplementary information accompanies this paper at https://doi.org/ . /s - - -w. additional file : appendix . redcap abstraction tool. appendix . inter-rater reliability. appendix . previous definition of severe and irreversible sepsis syndrome. appendix . examples. appendix . determining cause of death by organ system. etable . total organ system dysfunction. etable . cause of death by icu setting. abbreviations ahrf: acute hypoxemic respiratory failure; ards: acute respiratory distress syndrome; icu: intensive care unit; sofa: sequential organ failure assessment the epidemiology of acute respiratory failure in critically iii patients epidemiology, patterns of care, and mortality for patients with acute respiratory distress syndrome in intensive care units in countries causes and timing of death in patients with ards an official american thoracic society/european society of intensive care medicine/society of critical care medicine clinical practice guideline: mechanical ventilation in adult patients with acute respiratory distress syndrome ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome efficacy and safety of a paired sedation and ventilator weaning protocol for mechanically ventilated patients in intensive care (awakening and breathing controlled trial): a randomised controlled trial a binational multicenter pilot feasibility randomized controlled trial of early goal-directed mobilization in the icu early intensive care unit mobility therapy in the treatment of acute respiratory failure increasing incidence of withholding and withdrawal of life support from the critically ill palliative care in intensive care units: why, where, what, who, when, how the changing role of palliative care in the icu interobserver reliability of the berlin ards definition and strategies to improve the reliability of ards diagnosis differences between patients in whom physicians agree and disagree about the diagnosis of acute respiratory distress syndrome evaluating delivery of low tidal volume ventilation in six icus using electronic health record data acute respiratory distress syndrome: the berlin definition the berlin definition of ards: an expanded rationale, justification, and supplementary material surviving sepsis campaign: international guidelines for management of sepsis and septic shock the third international consensus definitions for sepsis and septic shock (sepsis- ) hospital deaths in patients with sepsis from independent cohorts comparison of the berlin definition for acute respiratory distress syndrome with autopsy extracorporeal life support organization registry report prone positioning in severe acute respiratory distress syndrome changes in end-of-life practices in european intensive care units from to an intensive communication intervention for the critically ill impact of a proactive approach to improve end-of-life care in a medical icu the abcdef bundle in critical care limitation of life-sustaining care in the critically ill: a systematic review of the literature publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations we thank daniel molling, ms, of va ccmr, for his careful data analysis. authors' contributions sk made substantial contributions to the conception and design of the work, acquisition, analysis, and interpretation of the data, and drafted and substantively revised the work. ys, hm, tb, aw, ic, dc, and jm made substantial contributions to the acquisition of data. hp made substantial contributions to the conception and design of the work, analysis, and interpretation of the data and drafted and substantively revised the work. ms made substantial contributions to the conception and design of the work, acquisition, analysis, and interpretation of the data and drafted and substantively revised the work. all authors have approved the submitted version and have agreed both to be personally accountable for the authors' own contributions and to ensure that questions related to the accuracy or integrity of any part of the work, even ones in which the author was not personally involved, are appropriately investigated, resolved, and the resolution documented in the literature. dr. prescott was supported in part by k gm from the nih/nigms. the datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. this study was approved by the university of michigan institutional review board. this study does not involve living individuals and therefore consent was waived. competing interests this material is the result of work supported with resources and use of facilities at the ann arbor va medical center. this manuscript does not represent the views of the department of veterans affairs or the us government. the authors declare that they have no competing interests. key: cord- -jveh w authors: rossen, lauren m.; branum, amy m.; ahmad, farida b.; sutton, paul; anderson, robert n. title: excess deaths associated with covid- , by age and race and ethnicity — united states, january –october , date: - - journal: mmwr morb mortal wkly rep doi: . /mmwr.mm e sha: doc_id: cord_uid: jveh w as of october , , deaths from coronavirus disease (covid- ) have been reported in the united states*; however, this number might underestimate the total impact of the pandemic on mortality. measures of excess deaths have been used to estimate the impact of public health pandemics or disasters, particularly when there are questions about underascertainment of deaths directly attributable to a given event or cause ( - ).† excess deaths are defined as the number of persons who have died from all causes, in excess of the expected number of deaths for a given place and time. this report describes trends and demographic patterns in excess deaths during january -october , . expected numbers of deaths were estimated using overdispersed poisson regression models with spline terms to account for seasonal patterns, using provisional mortality data from cdc's national vital statistics system (nvss) ( ). weekly numbers of deaths by age group and race/ethnicity were assessed to examine the difference between the weekly number of deaths occurring in and the average number occurring in the same week during - and the percentage change in . overall, an estimated , excess deaths have occurred in the united states from late january through october , , with two thirds of these attributed to covid- . the largest percentage increases were seen among adults aged - years and among hispanic or latino (hispanic) persons. these results provide information about the degree to which covid- deaths might be underascertained and inform efforts to prevent mortality directly or indirectly associated with the covid- pandemic, such as efforts to minimize disruptions to health care. determine the degree to which observed numbers of deaths differ from historical norms. in april , cdc's national center for health statistics (nchs) began publishing data on excess deaths associated with the covid- pandemic ( , ) . this report describes trends and demographic patterns in the number of excess deaths occurring in the united states from january , , through october , , and differences by age and race/ ethnicity using provisional mortality data from the nvss. § excess deaths are typically defined as the number of persons who have died from all causes, in excess of the expected number of deaths for a given place and time. a detailed description of the methodology for estimating excess deaths has been described previously ( ) . briefly, expected numbers of deaths are estimated using overdispersed poisson regression models with spline terms to account for seasonal patterns. the average expected number, as well as the upper bound of the % prediction interval (the range of values likely to contain the value of a single new observation), are used as thresholds to determine the number of excess deaths (i.e., observed numbers above each threshold) and percentage excess (excess deaths divided by average expected number of deaths). estimates described here refer to the number or percentage above the average; estimates above the upper bound threshold have been published elsewhere ( ) . observed numbers of deaths are weighted to account for incomplete reporting by jurisdictions ( states and the district of columbia [dc]) in the most recent weeks, where the weights were estimated based on completeness of provisional data in the past year ( ) . weekly nvss data on excess deaths occurring from january (the week ending february ), , through october , , were used to quantify the number of excess deaths and the percentage excess for deaths from all causes and deaths from all causes excluding covid- . ¶ non-hispanic american indian or alaska native [ai/an], and other/unknown race/ethnicity, which included non-hispanic native hawaiian or other pacific islander, non-hispanic multiracial, and unknown) were used to examine the difference between the weekly number of deaths occurring in and the average number occurring in the same week during - . these values were used to calculate an average percentage change in (i.e., above or below average compared with past years), over the period of analysis, by age group and race and hispanic ethnicity. nvss data in this report include all deaths occurring in the states and dc and are not limited to u.s. residents. approximately . % of decedents overall are foreign residents. r statistical software (version . . ; the r foundation) was used to conduct all analyses. from january , , through october , , an estimated , more persons than expected have died in the united states.** excess deaths reached their highest points to date during the weeks ending april ( . % excess) and august , ( . % excess) ( figure ). two thirds of excess deaths during the analysis period ( . %; , ) were attributed to covid- and the remaining third to other causes † † (figure ). the total number of excess deaths (deaths above average levels) from january through october ranged from a low of approximately in the youngest age group (< years) to a high of , among adults aged - years. § § however, the average percentage change in deaths over this period compared with previous years was largest for adults aged - years ( . %) ( figure ). overall, numbers of deaths among persons aged < years were . % below average, ¶ ¶ and among adults ** excess deaths over this period ranged from , to , . the lower end of this range corresponds to the total number above the upper bound of the % prediction intervals, and the upper end of the range corresponds to the total number above the average expected counts. deaths above the upper bound threshold are significantly higher than expected. https://www.cdc.gov/ nchs/nvss/vsrr/covid /excess_deaths.htm. † † excess deaths attributed to covid- were calculated by subtracting the number of excess deaths from all causes excluding covid- from the total number of excess deaths from all causes. these excess death estimates were based on the numbers of deaths above the average expected number. using the upper bound of the % prediction interval for the expected numbers (the upper bound threshold), an estimated , excess deaths occurred during this period, . % of which were attributed to covid- . § § weeks when the observed numbers of deaths were below the average numbers from to were excluded from the total numbers of excess deaths above average levels (i.e., negative values were treated as excess deaths). ¶ ¶ the total average percentage change in the number of deaths occurring from the week ending february , , through october , , included weeks where the percentage difference was negative (i.e., deaths were fewer than expected). this mainly affected the youngest age group, among whom, overall, deaths during this period were . % below average. excluding weeks with negative numbers of excess deaths results in overall percentage increases of . % for decedents aged < years. increases for other age groups were similar when excluding weeks with negative numbers of excess deaths, with the exception of those aged ≥ years, among whom the percentage increase was larger ( . %) when weeks with negative values were excluded. aged - , - years, - , and ≥ years were . %, . %, . %, and . % above average, respectively. when examined by race and ethnicity, the total numbers of excess deaths during the analysis period ranged from a low of approximately , among ai/an persons to a high of , among white persons. for white persons, deaths were . % higher when compared to average numbers during - . however, some racial and ethnic subgroups experienced disproportionately higher percentage increases in deaths ( figure ) . specifically, the average percentage increase over this period was largest for hispanic persons ( . %). deaths were . % above average for ai/an persons, . % above average for black persons, . % above average for those of other or unknown race or ethnicity, and . % above average for asian persons. based on nvss data, excess deaths have occurred every week in the united states since march . an estimated , more persons than expected have died since january , ; approximately two thirds of these deaths were attributed to covid- . a recent analysis of excess deaths from march through july reported very similar findings, but that study did not include more recent data through september ( ) . although more excess deaths have occurred among older age groups, relative to past years, adults aged - years have experienced the largest average percentage increase in the number of deaths from all causes from late january through october , . the age distribution of covid- deaths shifted toward younger age groups from may through august ( ); however, these disproportionate increases might also be related to underlying trends in other causes of death. future analyses might shed light on the extent to which increases among younger age groups are driven by covid- or by other causes of death. among racial and ethnic groups, the smallest average percentage increase in numbers of deaths compared with previous years occurred among white persons ( . %) and the largest for hispanic persons ( . %), with intermediate increases ( . %- . %) among ai/an, black, and asian persons. these disproportionate increases among certain racial and ethnic groups are consistent with noted disparities in covid- mortality.*** the findings in this report are subject to at least five limitations. first, the weighting of provisional nvss mortality data might not fully account for reporting lags, particularly in recent weeks. estimated numbers of deaths in the most recent weeks are likely underestimated and will increase as more data become available. second, there is uncertainty associated with *** https://www.cdc.gov/coronavirus/ -ncov/community/health-equity/ race-ethnicity.html. the models used to generate the expected numbers of deaths in a given week. a range of values for excess death estimates is provided elsewhere ( ), but these ranges might not reflect all of the sources of uncertainty, such as the completeness of provisional data. third, different methods or models for estimating the expected numbers of deaths might lead to different results. estimates of the number or percentage of deaths above average levels by race/ethnicity and age reported here might not sum to the total numbers of excess deaths reported elsewhere, which might have been estimated using different methodologies. fourth, using the average numbers of deaths from past years might underestimate the total expected numbers because of population growth or aging, or because of increasing trends in certain causes such as drug overdose mortality. finally, estimates of excess deaths attributed to covid- might underestimate the actual number directly attributable to covid- , because deaths from other causes might represent misclassified covid- -related deaths or deaths indirectly caused by the pandemic. specifically, deaths from circulatory diseases, alzheimer disease and dementia, and respiratory diseases have increased in relative to past years ( ) , and it is unclear to what extent these represent misclassified covid- deaths what is already known about this topic? as of october , , deaths from covid- have been reported in the united states; however, this might underestimate the total impact of the pandemic on mortality. what is added by this report? overall, an estimated , excess deaths occurred from late january through october , , with , ( %) excess deaths attributed to covid- . the largest percentage increases were seen among adults aged - years and among hispanic or latino persons. what are the implications for public health practice? these results inform efforts to prevent mortality directly or indirectly associated with the covid- pandemic, such as efforts to minimize disruptions to health care. directly or indirectly associated with the covid- pandemic and the elimination of health inequities. cdc continues to recommend the use of masks, frequent handwashing, and maintenance of social distancing to prevent covid- . † † † corresponding author: lauren m. rossen, lrossen@cdc.gov. new york city department of health and mental hygiene (dohmh) covid- response team. preliminary estimate of excess mortality during the covid- outbreak differential and persistent risk of excess mortality from hurricane maria in puerto rico: a time-series analysis estimation of excess deaths associated with the covid- pandemic in the united states estimating the number of excess deaths attributable to heat in excess deaths from covid- and other causes every body counts: measuring mortality from the covid- pandemic us department of health and human services, cdc, national center for health statistics us department of health and human services, cdc, national center for health statistics race, ethnicity, and age trends in persons who died from covid- -united states national center for health statistics, cdc.all authors have completed and submitted the international committee of medical journal editors form for disclosure of potential conflicts of interest. no potential conflicts of interest were disclosed. † † † https://www.cdc.gov/coronavirus/ -ncov/prevent-getting-sick/ prevention.html. week of death week of death week of death week of death week of death week of death week of death or deaths indirectly related to the pandemic (e.g., because of disruptions in health care access or utilization). despite these limitations, however, this report demonstrates important trends and demographic patterns in excess deaths that occurred during the covid- pandemic. these results provide more information about deaths during the covid- pandemic and inform public health messaging and mitigation efforts focused on the prevention of infection and mortality key: cord- -nth o ot authors: roy, satyaki; ghosh, preetam title: factors affecting covid- infected and death rates inform lockdown-related policymaking date: - - journal: plos one doi: . /journal.pone. sha: doc_id: cord_uid: nth o ot background: after claiming nearly five hundred thousand lives globally, the covid- pandemic is showing no signs of slowing down. while the uk, usa, brazil and parts of asia are bracing themselves for the second wave—or the extension of the first wave—it is imperative to identify the primary social, economic, environmental, demographic, ethnic, cultural and health factors contributing towards covid- infection and mortality numbers to facilitate mitigation and control measures. methods: we process several open-access datasets on us states to create an integrated dataset of potential factors leading to the pandemic spread. we then apply several supervised machine learning approaches to reach a consensus as well as rank the key factors. we carry out regression analysis to pinpoint the key pre-lockdown factors that affect post-lockdown infection and mortality, informing future lockdown-related policy making. findings: population density, testing numbers and airport traffic emerge as the most discriminatory factors, followed by higher age groups (above and specifically +). post-lockdown infected and death rates are highly influenced by their pre-lockdown counterparts, followed by population density and airport traffic. while healthcare index seems uncorrelated with mortality rate, principal component analysis on the key features show two groups: states ( ) forming early epicenters and ( ) experiencing strong second wave or peaking late in rate of infection and death. finally, a small case study on new york city shows that days-to-peak for infection of neighboring boroughs correlate better with inter-zone mobility than the inter-zone distance. interpretation: states forming the early hotspots are regions with high airport or road traffic resulting in human interaction. us states with high population density and testing tend to exhibit consistently high infected and death numbers. mortality rate seems to be driven by individual physiology, preexisting condition, age etc., rather than gender, healthcare facility or ethnic predisposition. finally, policymaking on the timing of lockdowns should primarily consider the pre-lockdown infected numbers along with population density and airport traffic. we process several open-access datasets on us states to create an integrated dataset of potential factors leading to the pandemic spread. we then apply several supervised machine learning approaches to reach a consensus as well as rank the key factors. we carry out regression analysis to pinpoint the key pre-lockdown factors that affect post-lockdown infection and mortality, informing future lockdown-related policy making. population density, testing numbers and airport traffic emerge as the most discriminatory factors, followed by higher age groups (above and specifically +). post-lockdown infected and death rates are highly influenced by their pre-lockdown counterparts, followed by population density and airport traffic. while healthcare index seems uncorrelated with mortality rate, principal component analysis on the key features show two groups: states ( ) forming early epicenters and ( ) experiencing strong second wave or peaking late in rate of infection and death. finally, a small case study on new york city shows that days-to-peak for infection of neighboring boroughs correlate better with inter-zone mobility than the interzone distance. states forming the early hotspots are regions with high airport or road traffic resulting in human interaction. us states with high population density and testing tend to exhibit during pre-and post-covid periods to show that the odds of mortality of whites and blacks are statistically equivalent [ ] . myers et al. analyzed the covid- positive patients in california to investigate its prognosis in the higher age groups and individuals with preexisting conditions [ ] . zoabi et al. applied ml on , covid- positive patients to understand the effect of gender, age and contact to show that close social interaction is a strong feature for covid- transmissibility [ ] . khan et al. applied regression tree, cluster analysis and principal component analysis on worldometer infection count data to study the variability and effect of testing in prediction of confirmed cases [ ] . finally, pan et al. studied the effects of the myriad public health interventions (such as lockdown, traffic restriction, social distancing, home quarantine, centralized quarantine, etc.) on , covid- patients, with respect to their age, sex, residential location, occupation, and severity [ ] . contributions: while it is evident that factors such as gender, race, age, testing, social contact and distancing have been analyzed in a piecemeal manner, there is no comprehensive study that combines the demographic, economic, and epidemiological, ethnic and health indicators for infection and mortality from covid- . to address this gap, we carry out a machine learning-based analysis with the following three objectives. . we curate a dataset of diverse features (detailed in sec. . ) from states of usa. this dataset is somewhat unique, since, in addition to the above features, it includes factors such as airport traffic, homeless and variations in lockdown dates. also, note that the lockdown was enforced on the us states at around the same time, when each state was at a different stage of the covid- infection cycle. . we analyze the variation of covid- infection spread and mortality rates using a set of standard supervised ml methods. we rank the key discriminatory factors based on the importance score calculated from randomized decision trees. we combine the findings to identify the most vulnerable age groups and us states. we also show the effect of testing and lockdowns on the infection spread dynamics. . we utilize multiple linear regression to gauge the extent to which the key pre-lockdown factors affect the post-lockdown infected and death numbers. this study assigns weights to features and drive mitigation efforts and large scale policymaking. our data-driven experiments using supervised methods demonstrate that population density, testing [ ] and airport traffic [ ] are key factors contributing to infection and mortality rates. furthermore, high age group ( and beyond, and specifically exceeding ) population are more vulnerable. principal component analysis on the key features show two groups: highly affected us states ( ) forming early epicenters and ( ) showing consistent or newly peaking rate of infection and death. multiple regression analysis shows that the postlockdown numbers are most influenced by the pre-lockdown infected and death numbers followed by population density and airport activity, while overall healthcare index of a state does not seem to play a part in the overall death count. similarly, the race of individuals did not play any significant role in the infection or mortality numbers. despite increased testing rates, the fraction of individuals tested positive drop approximately three weeks into the lockdown, suggesting that the social distance measures has had an impact on curbing spread. finally, we discuss the role of mobility and distance in infection spread. in the absence of large-scale inter-state mobility data, our case study on the boroughs of new york city show that peaks of infection correlate better with inter-zone mobility than the interzone distance. all the experiments have been performed using scikit-learn, which is a popular machine learning library in python [ ] . let us discuss the details of the two datasets used in this work. . . data from us states. our dataset has been carefully curated from several open sources to examine the possible factors that may affect the covid- related infection and death numbers in the states of usa. the individual open-access data sources as well as the integrated (curated) dataset has been shared on github (https://github.com/satunr/covid- /tree/master/us-covid-dataset). below, we discuss a summary of the features and output labels of the integrated dataset. • gross domestic product (in terms of million us dollars) for us states [ ] (filename: source/ gdp.xlsx, feature name: gdp). • distance from one state to another (is not measured in miles but the euclidean distance between their latitude-longitude coordinates between the pair of states [ ] ) (filename: source/data_distance.xlsx, feature name: d(state , state )). • gender feature(s) is a fraction of total population representing the male and female individuals [ ] (filename: source/data_gender.csv, feature name: male, female). • ethnicity feature(s) are the fraction of total population representing white, black, hispanic and asian individuals (we leave out other smaller ethnic groups) [ ] (filename: source/ data_ethnic.csv, feature name: white, black, hispanic and asian). • healthcare index is measured by agency for healthcare research and quality (ahrq) on the basis of ( ) type of care (like preventive, chronic), ( ) setting of care (like nursing homes, hospitals), and ( ) clinical areas (like care for patients with cancer, diabetes) [ ] (filename: source/data_health.xlsx, feature name: health). • homeless feature is the number of homeless individuals of a state [ ] (filename: source/ data_homeless.xlsx, feature name: homeless). the normalized homeless population of each state is the ratio between its homeless and total population. • total cases (and deaths) of covid- is the number of individuals tested positive and dead [ ] (filename: source/data_covid_total.xlsx, feature name: total cases and total death). the normalized infected/death is the ratio between the infected/death count to total population of the given state. • infected score and death score is obtained by rounding normalized total cases and deaths to discrete value between - (feature name: infected score, death score). • death-to-infected is a feature measuring impact of death in terms of the difference between death and infected scores. it is calculated as max(death score -infected score, ). • lockdown type is a feature capturing the type of lockdown (shelter in place: and stay at home: ) in a given state [ , ] (filename: source/data_lockdown.csv, feature name: lockdown). • day of lockdown captures the difference in days between st january to the date of imposition of lockdown in a region [ ] (filename: source/data_lockdown.csv, feature name: day lockdown). • population density is the ratio between the population and area of a region [ ] (filename: source/data_population.csv, feature name: population, area, population density). • traffic/activity of airport measures the passenger traffic (also normalized by the total traffic across all the states of usa [ ] (filename: source/data_airport.xlsx, feature name: busy airport score, normalized busy airport). • age groups ( - +) in brackets of year (also normalized by total population) [ ] (filename: source/data_age.xlsx, feature name: age_to_, norm_to_, e.g. age to ); we later group them in brackets of for the purposes of analysis. • peak infected (and peak death) measures the duration between first date of infection and date of daily infected (and death) peaks [ ] (feature name: peak infected, peak death). • testing measures the number of individuals tested for covid- (total number, before and after imposition of lockdown) [ , ] (filename: source/data_testing.xlsx, feature name: testing, pre-lockdown testing, post-lockdown testing). • pre-and post-infected and death count measures the number of individuals infected and dead before and after lockdown dates (feature name: testing, pre-infected count, pre-death count, post-infected count, post-death count). • days between first infected and lockdown date (feature name: first-inf-lockdown). the above features, their abbreviations and summary statistics (i.e., mean, standard deviation, maximum and minimum) are enlisted in table . note that, for gender and ethnicity we report the fraction of the total state population falling in each category. the new york city (nyc) datasets (https://github.com/ satunr/covid- /blob/master/us-covid-dataset/nyc_dist_mob.xlsx) show the inter-borough distance and mobility as well as covid- infected (https://github.com/satunr/covid- /blob/master/us-covid-dataset/nyc-inf.xlsx) and death counts (https://github.com/ satunr/covid- /blob/master/us-covid-dataset/nyc-dth.xlsx) for the boroughs of nyc, namely, manhattan, queens, brooklyn, bronx and staten island. table . summary of features and their statistics (i.e., mean, standard deviation (dev.), maximum (max.) and minimum (min.)). the features in the order shown under "feature name" are: gdp, inter-state distance based on lat-long coordinates, gender, ethnicity, quality of health care facility, number of homeless people, total infected and death, population density, airport passenger traffic, age group, days for infection and death to peak, number of people tested for covid- , days elapsed between first reported infection and the imposition of lockdown measures at a given state. factors affecting covid- infected and death rates inform lockdown-related policymaking • mobility data (based on traffic volume counts collected by dot for new york metropolitan transportation council (nymtc) [ ] ) shows the number of trips from one borough to another. • covid- data shows the number of covid- infected and death counts for each borough [ ] . we acquire the daily infected and testing counts across us from january-july, [ ] . this dataset is part of the covid tracking project that collect covid- statistics on the numbers on tests, cases, hospitalizations, and patient outcomes from every us state and territory by voluntary public participation. we use the scikit-learn library kbinsdiscretizer to group the continuous feature values into discrete values by creating balanced clusters using the quantile strategy [ ] . . . supervised learning methods. supervised machine learning algorithms learn a function that maps the input training data (i.e., features) to some output labels [ ] . in this work, we consider the following supervised learning techniques. (refer [ ] [ ] [ ] [ ] [ ] [ ] [ ] for the details on these ml approaches.) • support vector machine (svm) is used for classification and regression problems that maps the inputs to high-dimensional feature spaces. svm operates on hyperplanes-decision boundaries that help classify the data points. the objective is to maximize the separation between the data points and the hyperplane. svm is memory efficient and effective for datasets with fewer data samples [ ] . • stochastic gradient descent (sgd) is an iterative approach that fits the data to an objective function [ ] . as the name suggests, it is a stochastic variant of the popular gradient descent (gd) optimization model [ ] . in gd, the optimizer starts at a random point in the search space and reaches the lowest point of the function by traversing along the slope. unlike gd that requires calculating the partial derivative for each feature at each data point, sgd achieves computational efficiency by computing derivatives on randomly chosen data points. • nearest centroid (nc) is a simple classification model that represents each class by the centroid of its members. subsequently, it assigns each data point to the cluster whose centroid is the closest to it. nc is particularly effective for non-convex classes and does not suffer from any additional dependencies on model parameters [ ] . • decision trees (dts) are a classification and regression technique that assigns target labels based on decision rules inferred from data features [ ] . dt maintains the decision rules using a tree. a data point is assigned to a class by repeatedly comparing the tree root with the data point value to branch off to a new root. • gaussian naive bayes (nb) are a class of fast, probabilistic learning techniques that apply the bayes' theorem to assign labels to the data points [ ] . while supervised ml approaches generally yield reliable prediction accuracy, they often suffer from overfitting or convergence issues [ , ] . each of the above approaches has its own advantages and disadvantages. svm works well when the underlying distribution of the data is not known. however, it is prone to overfitting when the number of features is much greater than the number of samples. sgd needs low convergence time for a large dataset, but it may require to fit a number of hyperparameters. conversely, dt involves almost no hyperparameters, but often entails slightly higher training time. unlike dt, nb requires less training time but works on the implicit assumption that all the attributes are mutually independent. finally, nc is a fast method but is not robust to outliers or missing data. in the context of our work, we intuit that the discriminatory feature(s) will yield a high accuracy irrespective of the underlying supervised ml algorithm used. • accuracy function measures the fraction of matches between the predicted and actual labels in a multi-label classification, i.e., the ratio of correctly predicted observations to the total observations. it can be calculated as: in the above equation, tp, tn, fp, fn denote true positive, true negative, false positive and false negative, respectively. • extra trees classifier is an estimator that fits randomized decision trees (called extra-trees) on data samples. the memory and computation overhead of this approach can be controlled by regulating the size of the extra trees. the nodes in the tree are split into sub-trees resulting in high accuracy (i.e., drop in impurity). thus, feature importance is measured as total reduction in impurity affected by that feature [ ] . • multiple regression (mr) is a statistical tool to capture the linear relationship between the independent and the dependent variables x and y of a function y = g(x). in our context, mr generates a linear relationshipŷ where b fi is the coefficient that captures the contribution of feature f i towards the dependent variable y, while β and � are the intercept and error terms, respectively. given any pair of vectors v andv (jvj ¼ jvj ¼ n), we apply the following standard statistical operations: • mean centering subtracts the mean μ from each element of a vector v, i.e., v = v − μ(v). this standardization adjusts the scales of magnitude by making the new mean and helps compare data from varied sources or having different datatypes. • mean squared error (mse) is calculated as • pearson correlation coefficient (pcc) between v andv measures the strength of a linear association between two variables, where the value pcc = is a perfect positive correlation and − is perfect negative correlation. • positivity rate ρ is the ratio between the number of individuals tested positive to the number of tests performed daily [ ] . this section is classified into the following three subsections: ( ) and ( ) table . unless otherwise stated, the feature set comprises gdp, gender, ethnicity, health care, homeless, lockdown type, population density, airport activity, and age groups, whereas the output labels consist of infected and death scores on a scale of - . we apply supervised machine learning (ml) approaches to identify the key factors affecting covid- infected and death counts. for each supervised ml technique, we perform an exhaustive search of all possible combinations of any features and identify the feature subset (s) with the highest accuracy (discussed in sec. . ) as the most important features. fig shows the scores for different supervised methods. although proposing a machine learning algorithm that works best on covid- data is not the purpose of this study, it is worth reporting that decision tree classifier (dt) slightly outperforms the other algorithms for both cases of infected and death scores. we create a pool of all features participating in at least one combination for output labels of infected and death scores. fig shows a heatmap of the importance i for all such features against each supervised technique. for infected score as output label (top figure), homeless (home), population density (pd), airport activity (air), testing (test), white (wht), etc. have the highest i. for death score as output label, pd, air, test and age groups above years (age _ and age _ ) exhibit the highest importance. we apply the extra trees classifier to generate the impurity-based rank for the features (discussed in sec. . ). fig a shows the top important features corresponding to the infected and death scores, respectively. it is interesting that for both cases, the same set of features, namely, population density, days to peak, airport traffic, testing and high age groups, are identified. also note that the same features exhibit a very high participation in the -feature combinations shown in fig . next, as a validation exercise, we apply dimension reduction on the factors affecting covid- infected and death rates inform lockdown-related policymaking table we discussed in sec. . , that our initial dataset groups ages into brackets of ( - , - , and so on). our results from supervised learning (sec. . ) and extra trees (sec. . ) suggest that high age groups are important factors affecting the infected and death scores of covid- . to understand the effect of covid- infected and death scores on low and high age groups, we create two feature sets for population of age � and > . fig a shows that for both cases of infected and death, the accuracy (acc) is higher for higher age groups. we explore this by repeating the above experiment, this time, with a feature set of groups - and > . fig b depicts that acc for age group + is marginally higher, suggesting that the elderly are amongst the most vulnerable, however the difference in mortality rates in this case was not statistically significant. we carry out a study to identify the pre-lockdown factors of any region (us states in our case) that contribute to the overall post-lockdown infection and death numbers. we partition the total infected and death numbers for each state into pre-and post-lockdown infected and death counts. we then create a feature set consisting of population density, airport business, pre-lockdown infected, pre-lockdown death, days between first infected to lockdown and age group above . the features represent the set of observable factors for the administrative and health bodies and were already shown to possess high feature significance in the previous factors affecting covid- infected and death rates inform lockdown-related policymaking section. the output labels are the post-lockdown infected and post-lockdown death numbers. we perform the following experiments: . . identification of discriminating features. we carry out a simple preprocessing step to convert each feature entry to percentile (with respect to the feature vector) and rank the us states in the decreasing order of infected and death scores (fig ) . we calculate the weighted average percentile of features for the top and bottom k = us states using the formula where p(f i ) and ρ(f i ) are the percentile and rank of the i th feature value, while r is the number of us states (equal to maximum rank). we intuit that the feature exhibiting the maximum difference in weighted average percentile for top and bottom k covid- affected us states are the discriminating ones. fig a shows the percentile difference suggesting that airport and population density are the most significant, while days between first infected to lockdown and age group of + are the least discriminating. we apply multiple regression (mr) (see sec. . ) to measure the weightage of each of the above features in the observed post-lockdown infected (post_inf) and post-death numbers (post_dth). we eliminate the days between first infected to lockdown (fst-lock) and age group +, which are the least discriminating features from the percentile analysis (see fig a) . as a prerequisite for mr, we need to eliminate features that are mutually correlated. fig b shows that pre-inf and pre-dth are highly correlated, and hence we run two separate batches of mr: ( ) population density, airport business, pre-lockdown infected and ( ) population density, airport business, pre-lockdown death. we explore the effect of testing and lockdown on infection spread. we utilize positivity ratio ρ (defined in sec. . ) to gauge how widespread the infection spread is [ ] . we acquire the daily infected and testing count in us (see sec. . . ) and plot the mean daily ρ across all states over the period of february-july . fig a shows that the testing increased over a period time, while the positivity ratio dropped post lockdown (shown in red dotted line). while, testing (and, by extension, positivity ratio) is an effective epidemiological indicator, it cannot curb infection spread by itself. however, fig a shows that the ρ has dropped approximately three weeks into the lockdown, suggesting that the latter had an impact on curbing spread by minimizing social contact. table shows that pre-infected and pre-death with high coefficients contribute highly towards factors affecting covid- infected and death rates inform lockdown-related policymaking the post-lockdown infected and death numbers, followed by population density and airport traffic. this finding is further supported by the p values reported for the respective features. note that the r scores for all the four cases are > . , suggesting that the output features capture a high proportion of the variance in the input features. overall, pre-infected count has higher coefficient and r score and emerges as a marginally better discriminating feature of post-lockdown effects than the pre-death count. factors affecting covid- infected and death rates inform lockdown-related policymaking in sec. . , we perform pca on the feature set of the key factors to show that states with high infection and death numbers stand out of the cluster of other states. these states include some erstwhile hotspots forming group (such as new york city, new jersey, massachusetts, connecticut, rhode island) as well as states experiencing a steady infection and death count and also a strong second wave forming group (such as texas, washington, california, georgia, arkansas, utah and colorado) (fig b) . in the pca analysis, pc and pc account for % and % variance, respectively. we explore how each feature influences each component to show that pc is driven by factors such as airport activity and high age groups ( and beyond), while pc is dominated by population density, airport, age ( +) and testing. notice in fig b, though both groups and exhibit high spread across pc , group forms a slightly denser cluster than group , implying that it exhibits an even mix of pc and pc features. we intuit that the early peaking in infection in group states is due to high road and airport mobility leading to high mixing and infection spread that is manifested in the elderly population. group shows enduring infection spread due to high population density and testing, in addition to airport activity and populations with higher age group. we study how demographics affect covid- numbers to show that states with higher age groups (particularly and beyond) numbers are the most vulnerable. finally, we split the infected and death numbers on the pre-and post-lockdown epochs and apply multiple linear regression to show that pre-lockdown infected and death, population density and airport contribute highly to the post-lockdown numbers. this analysis can be particularly effective in pinpointing the most vulnerable states and recommending lockdown policies on starting dates and duration to curb pandemic spread. note that our present study pertains to the identification of the discriminatory features with respect to the date of lockdown. there exists several unanswered questions regarding the impact of length, scheduling strategies, lockdown types and extent of lockdowns on pandemic spread that need to be answered. such an analysis requires a richer feature set as well as a sound understanding of the dynamics of infection spread in terms of healthcare, distance, mobility, etc. as a preliminary study, we first explore whether there is any relationship between the health care index (health) of a us state and the number of transitions from infected to death (dth/inf) in this state. the pearson's correlation coefficient (see sec. . ) between the two factors is . , suggesting that the overall mortality numbers is largely unrelated to the healthcare facility and may solely depend on the infected individual's attributes, such as age, comorbidities, infection severity, etc. second, since proximity plays a role in infection spread, neighboring regions should peak at nearly the same time. we posit that mobility may play an even greater role in the spread, than a static measure like distance between a pair of regions. in the absence of a inter-state mobility dataset, we create two feature sets for the nyc boroughs dataset (see sec. . ): ( ) inter-borough distance and ( ) inter-borough mobility. each borough b has a distance and mobility vector d b = {d b , d b � � �} and m b = {m b , m b � � �} where d bi and m bi are the probabilistic measure of distance and mobility between a borough b with borough i. we calculate the correlation of the mean squared error (see sec. . ) of the distance/mobility vectors of any pair of boroughs b and b against the absolute difference of their peak to infected or peak-to-death features. fig b suggests that mobility yields a higher correlation ( . ) than distance ( . ) suggesting that mobility is a slightly more informative feature to analyze infection spread. we are currently working towards broadening the scope of this study in different directions. first, this work attempted to apply ml analysis on a wide range of features, making the the states of united states the ideal choice, specifically from the standpoint of data availability. in future we would like to extend this work by running these experiments on epidemiological, demographic and economic data of different countries. it would be interesting to report the variation in the discriminatory features identified for different countries. second, we identify population density, testing, airport activity and pre-lockdown infected count as key features driving the post-lockdown infection and death numbers. we plan to utilize these findings to design policies on the timing, duration and stringency of lockdown for future pandemics. third, all the input features discussed in this work are static or time invariant. it is imperative to analyze the evolution of dynamic features (such as gdp and unemployment rates) from the pre-covid to the post-covid timelines to uncover the long-term economic effects of covid- . machine learning is emerging as an important tool to predict the dynamics of spread of covid- and identify the key factors driving infection and mortality rates. while existing works study the effects of gender, race, age, testing, social contact and distancing separately, we present an unified analysis of the demographic, economic, and epidemiological, ethnic and health indicators for infection and mortality rates from covid- . we curate a dataset of us states comprising features (from varying sources discussed in sec. . ) that may potentially impact infection and death rates of covid- . we run several supervised machine learning techniques to identify and rank the key factors correlating with infection and fatality counts. population density, testing rate, airport traffic, high age groups emerge as significant, while ethnicity, gender, healthcare index, homeless and gdp have little or no impact on pandemic spread and mortality. coronavirus: what have been the worst pandemics and epidemics in history coronavirus world map: which countries have the most cases and deaths epidemiology, causes, clinical manifestation and diagnosis, prevention and control of coronavirus disease (covid- ) during the early outbreak period: a scoping review covid- and world economy. covid- and world economy covid-induced economic uncertainty is this the second wave of covid- in the u.s.? or are we still in the first? how will country-based mitigation measures influence the course of the covid- epidemic? the lancet in beijing it looked like coronavirus was gone. now we're living with a second wave daily covid- cases in india continue to soar, japan's tokyo in fears of nd wave of infections a fiasco in the making? as the coronavirus pandemic takes hold, we are making decisions without reliable data it's time to get real about the misleading data analysis of factors associated with disease outcomes in hospitalized patients with novel coronavirus disease factors affecting covid- transmission the origin, transmission and clinical therapies on coronavirus disease (covid- ) outbreak-an update on the status prediction models for diagnosis and prognosis of covid- infection: systematic review and critical appraisal artificial intelligence and machine learning to fight covid- machine learning using intrinsic genomic signatures for rapid classification of novel pathogens: covid- case study coronavirus (covid- ) classification using ct images by machine learning methods wrong but useful-what covid- epidemiologic models can and cannot tell us prediction of epidemic trends in covid- with logistic model and machine learning technics modified seir and ai prediction of the epidemics trend of covid- in china under public health interventions the association of race and covid- mortality. eclinicalmedicine, characteristics of hospitalized adults with covid- in an integrated health care system in california covid- diagnosis prediction by symptoms of tested individuals: a machine learning approach. medrxiv hossain countries are clustered but number of tests is not vital to predict global covid- confirmed cases: a machine learning approach. medrxiv randomized placebo-controlled trials of remdesivir in severe covid- patients: a systematic review and meta-analysis. medrxiv center for disease control and prevention. covid- testing overview scikit-learn: machine learning in python world population review. gross domestic product list of geographic centers of the united states population distribution by gender population distribution by race agency for healthcare research and quality. health care quality: how does your state compare? ahar: part -pit estimates of homelessness in the u cdc covid data tracker covid- cases covid us lockdown dates dataset united states census. state population by characteristics list of the busiest airports in the united states center for disease control and prevention. previous u.s. viral testing data nyc-covid borough level breakdown scikit-learn-preprocessing -kbinsdiscretizer machine learning: a review of classification and combining techniques support vector machine stochastic gradient descent scikit learn developers (bsd license) scikit learn developers (bsd license) scikit learn developers (bsd license). naive bayes scikit learn developers (bsd license) multiple linear regression support vector machine-a survey stochastic gradient descent an overview of gradient descent optimization algorithms a local mean-based k-nearest centroid neighbor classifier simplifying decision trees. international journal of man-machine studies an empirical study of the naive bayes classifier scikit-learn classifier tuning from complex training sets covid- testing: understanding the "percent positive the news tribune. washington state reports new covid- cases if trends persist, houston would become the worst affected city in the us, expert peter hotez says dph reports almost new cases of covid- in ga hundreds test positive for covid- at tyson foods plant in arkansas covid- cases rise as hospitalizations remain low in colorado utah confirms new coronavirus cases; more deaths on sunday the authors would like to acknowledge the editor/reviewers for critically assessing the materials and providing suggestions that significantly improved the presentation of the paper. furthermore, they acknowledge the department of computer science, virginia commonwealth university for its computational resources. validation: satyaki roy.visualization: satyaki roy. writing -review & editing: preetam ghosh. key: cord- -xxa hlpu authors: pyszczynski, tom; lockett, mckenzie; greenberg, jeff; solomon, sheldon title: terror management theory and the covid- pandemic date: - - journal: j humanist psychol doi: . / sha: doc_id: cord_uid: xxa hlpu terror management theory is focused on the role that awareness of death plays in diverse aspects of life. here, we discuss the theory’s implications for understanding the widely varying ways in which people have responded to the covid- pandemic. we argue that regardless of whether one consciously believes that the virus is a major threat to life or only a minor inconvenience, fear of death plays an important role in driving one’s attitudes and behavior related to the virus. we focus on the terror management theory distinction between proximal defenses, which are activated when thoughts of death are in current focal attention and are logically related to the threat at hand, and distal defenses, which are activated when thoughts of death are on the fringes of one’s consciousness and entail the pursuit of meaning, personal value, and close relationships. we use this framework to discuss the many ways in which covid- undermines psychological equanimity, the diverse ways people have responded to this threat, and the role of ineffective terror management in psychological distress and disorder that may emerge in response to the virus. although there are many disturbing aspects of the covid- pandemic, from the perspective of terror management theory (tmt; greenberg et al., ; solomon et al., ) , the enormous death toll and highly contagious nature of the virus play especially important roles in spawning the diverse forms of turmoil that have resulted from this crisis. we argue that the salience of death brought on by covid- plays a central role in driving the attitudes and behavior of even those who believe that the dangers of the virus have been vastly exaggerated. tmt is focused on the pervasive role death awareness plays in human affairs. much of what has been learned from the past years of research applying tmt to diverse aspects of human behavior is directly relevant to understanding responses to the pandemic, especially the distinction between proximal defenses, which are directly focused on the problem of death, and distal defenses, which bear no logical relationship to death but enable people to construe themselves as valuable contributors to a meaningful, significant, and permanent universe (pyszczynski et al., ) . in this article we discuss how the salience of death inherent in covid- influences diverse reactions to this pandemic. tmt (greenberg et al., ; pyszczynski et al., ) posits that an inherent consequence of humankind's sophisticated cognitive abilities is awareness of the inevitability of death. awareness of death in an animal with an inherent proclivity for self-preservation gives rise to an ever-present potential for existential terror. this potential for terror is managed by an anxiety-buffering system consisting of cultural worldviews, self-esteem, and close interpersonal relationships. cultural worldviews are shared beliefs about reality that provide answers to basic questions about life, standards for valued behavior, and the promise of literal or symbolic immortality to those who live up to these standards. literal immortality beliefs provide hope that life will continue after physical death, as exemplified by afterlife concepts such as heaven, reincarnation, or joining with ancestral spirits. symbolic immortality comes from contributing to something greater than oneself that will continue long after one has died, such as a family, nation, or the memories of others. selfesteem is a sense of personal value that results from believing that one is living up to the standards of one's cultural worldview. close relationships provide consensual validation of one's worldviews and self-esteem needed to maintain confidence in them, as well as providing security in their own right (mikulincer et al., ) . tmt posits that people manage the potential for anxiety inherent in awareness of the inevitability of death by maintaining faith in their cultural worldviews, self-esteem, and close relationships; these anxiety-buffering systems mitigate existential terror by imparting a sense that one is a person of value living in a meaningful world (for a more thorough presentation of these ideas, see solomon et al., ) . research has supported a network of converging hypotheses derived from tmt. this research shows that ( ) reminders of death (mortality salience) increase commitment to one's worldview, self-esteem, and relationships, and increase defense of these entities when threatened; ( ) bolstering self-esteem, worldview, or relationships makes one less prone to anxiety and anxiety-related behavior in response to threats; ( ) threats to worldview, self-esteem, and relationships increase the accessibility of deathrelated thoughts; and ( ) self-esteem striving, cultural worldview defense, or affirming close relationships in response to mortality salience reduce death thought accessibility and the need for further terror management defenses; this suggests that the three components of the anxiety buffer are psychologically interchangeable (hart et al., ) . for a recent review of the tmt literature, see pyszczynski et al. ( ) . meta-analyses have found strong evidence that reminders of death increase commitment to one's worldview (burke et al., ) and that threats to one's worldview increase the accessibility of death-related thoughts (steinman & updegraff, ) . tmt posits that people manage death anxiety with two distinct systems, referred to as proximal and distal defenses (pyszczynski et al., ) . when death-related thoughts are conscious (in current focal attention), proximal defenses are activated to suppress such thoughts or push death into the distant future by denying one's vulnerability to things that could kill, or intending to engage in healthier behavior to ensure a longer life. however, when deathrelated thoughts are on the fringes of consciousness (no longer in focal attention but still highly accessible), people activate distal defenses focused on maintaining faith in their cultural worldview and enhancing self-esteem. conscious awareness of death requires defensive maneuvers that "make sense," in that they imply that death is not a problem until many years in the distant future. but proximal defenses do little to quell anxiety stemming from the ultimate inevitability of death. these concerns are assuaged by distal defenses that are logically unrelated to death but imbue one's life with meaning, value, and the promise of either literal or symbolic immortality. research has shown that proximal defenses emerge shortly after reminders of death and that distal defenses emerge in response to death reminders only after a delay and distraction; however, distal defenses emerge immediately with no need for delay and distraction when death reminders are presented subliminally and thus bypass conscious attention. research has also shown that distal defenses reduce the accessibility of death-related thoughts, which is presumably how they manage anxiety (see arndt et al., ) . the personal, social, economic, and political costs of the covid- crisis are unprecedented. from the perspective of tmt, the root cause of all these problems is glaringly obvious-the risk of dying from the virus. regardless of how contagious and lethal the virus ultimately turns out to be, or what one consciously thinks about it, the possibility of dying from it is highly salient and evident in ever-increasing death toll statistics, vivid images of overburdened hospitals and makeshift morgues, and the testimonials to victims of the virus, both famous and unknown. the deadly disease is spawned by an invisible pathogen that is conveyed by droplets expelled in the breath of its victims and thus might be lurking almost anywhere. since early march, media coverage of the pandemic in the united states and europe has been virtually nonstop, interrupted only by coverage of a looming global economic collapse, lethal police violence against african americans, and the protests and social upheaval in response to it. george floyd's death and related tragedies triggered a reinvigorated focus on social and economic injustice in american society, including peaceful protests and organized campaigns calling for police reform and greater support for the black lives matter movement. however, media coverage has also highlighted the violence, vandalism, looting, and general disarray spawned by the protests and in response to them (johnson, ; kilgo, ) . what has emerged from the co-occurrence of the pandemic and social upheaval is a constant barrage of threatening information. moreover, it is impossible to visit a social media website without being inundated with new and often contradictory information on the virus. covid- thus poses a ubiquitous dramatic reminder of vulnerability and death. because of its potential lethality, attempts to stem the tide of the virus or "flatten the curve" have led to closing many businesses and public venues and, consequently, led to loss of income and jobs, falling stock market values, general economic chaos, and social isolation, all of which seriously undermined major resources for managing the potential terror of death (fitzgerald et al., ) . public gatherings of all kinds were initially prohibited and then later strictly regulated, creating a void in personal contacts and near-total isolation for some (banerjee & rai, ) . information provided by governments, scientists, and the health care community has been confusing and sometimes contradictory, with partisan media outlets exacerbating the problem by providing narratives tailored to their constituents and critical of those with different ideological affiliations (bermejo et al., ; jurkowitz & mitchell, ) . these side effects of the pandemic seriously undermined all three components of the anxiety-buffering system that people use to maintain equanimity. the world has suddenly become an even more chaotic, confusing, and hostile place, in which death lurks around every corner, and people struggle to maintain meaning and self-esteem. people are living with the very real threat of death from the pandemic, combined with challenges to their worldviews, loss of jobs, impediments to career goals, and isolation from friends and family who normally validate one's significance. from a tmt perspective, it is currently far more difficult for virtually all of us to manage the terror of death. people have responded to the pandemic in a wide variety of ways, some rational and some less so, some adaptive, and some destructive. the terror management health model (goldenberg & arndt, ) applies tmt and the distinction between proximal and distal defenses to health-related behavior. it suggests that thoughts of death can increase either motivation for healthy behavior or denial and avoidance when people are consciously focusing on them. however, when such thoughts are on the fringes of consciousness, they increase behavior oriented toward maintaining self-esteem and faith in one's cultural worldview, which could either facilitate or undermine health. we now consider some of the ways in which people employ proximal and distal tactics to cope with covid- . proximal defenses. tmt posits that when thoughts of death are in current focal attention, people attempt to remove them from their consciousness. this can entail simple suppression of such thoughts, denial of the threat, or engaging in behavior to reduce one's vulnerability. given the high level of media coverage, the changes in daily life that provide a constant reminder of the pandemic, and the extent to which virus-related concerns dominate conversations and media reporting, completely avoiding the issue is impossible. but there is evidence of increases in diversion-seeking behavior, such as alcohol consumption (furnari, ) , excessive eating (ammar et al., ) , and binge-watching television (dixit et al., ) . another form of proximal defense involves minimizing one's perception of the threat. this has taken the form of arguing that the virus is not nearly as contagious or lethal as health experts claim it to be (srikanth, ) , or that it is only lethal for the elderly or those already at-risk of dying from other diseases (fox et al., ) . others have trivialized the virus by comparing it to common illnesses such as the seasonal flu (ritter, ) , focusing on other common causes of death (mcginty, ), or viewing the publicity given the pandemic as originating in a politically motivated conspiracy (romano, ) . when sky-rocketing death statistics and vivid instances of contagion and mortality in the media make it hard to deny the problem outright, people sometimes claim that death rates are inflated to increase the funding hospitals receive (nunez, ) , or to bolster the aforementioned conspiracy to damage government leaders (brown, ; romano ) . another, likely more adaptive, form of proximal defense against covid- is to follow the prescriptions for avoiding infection provided by the medical community. this may be the most common proximal response to the pandemic; surveys suggest that % of people have followed guidelines for avoiding infection, to at least some extent (altman, ) . most people have engaged in some form of social distancing, increased sanitation practices such as hand washing and cleaning surfaces, wore masks in public places, and done other things to stay healthy (eanes, ) . but the economic and social effects of these measures interfere with feelings of value and connection with the world, the core way in which we distally quell concerns about our mortality. distal defenses. despite its ubiquitous nature, thoughts of the virus are not always the focus of conscious thought. this would be too disturbing for most people to bear and could lead to the emergence or exacerbation of psychological disorders. in addition, the proximal defenses that people employ are likely to be at least somewhat effective in removing thoughts of the pandemic from our consciousness. the bulk of the tmt literature suggests that distal defenses focused on affirming one's cultural worldview and maximizing selfesteem emerge when thoughts of death are highly accessible but not in focal attention; given the potential consequences of the virus and the enormous amount of attention the pandemic has attracted, this is likely to be the case for many people a great deal of the time. survey research provides clear evidence of a partisan divide in attitudes and behavior related to the virus. liberals tend to view the virus as much more dangerous than conservatives, report considerably more personal distress about it, and have greater confidence in what scientists and medical professionals have to say about it (funk et al., ; ritter, ) . conservatives, on the other hand, view the virus as less dangerous and are more likely to assign blame to china and other foreigners and view the virus as part of a conspiracy to discredit donald trump (romano, ) . the rapid emergence of polarized liberal and conservative narratives about the virus illustrate the dynamic interplay between individual psychological forces and cultural worldviews that is central to the tmt analysis of the relationship between individual and cultural psychology. this political divide is undoubtedly exacerbated by the accessibility of death-related cognition caused by the pandemic. though surveys have documented this divide since before president trump was elected, there is an even wider divergence regarding his overall handling of the pandemic (bycoffe et al., ) , attitudes toward those who have pushed back against some of his policies (spangler, ) , and dr. anthony fauci, who was at one time the major voice for the administration but later voiced some disagreements (brewster, ) . a political divide is also evident in attitudes toward easing restrictions and reopening businesses and public places, with conservatives much more in favor of such policies than liberals. whereas liberals tend to approve of societal restrictions to prevent the spread of the virus, conservatives tend to view them as unwarranted infringements on freedom and not worth the cost to the economy and individual incomes (shepard, ) . in many u.s. cities, protests against government restrictions were primarily attended by conservatives, some brandishing assault rifles and white nationalist symbols (mauger, ; perrett, ) . despite initial sentiment that "we're all in this together," the pandemic has become yet another domain for ideological division. from a tmt perspective, reminders of death motivate people to affirm their worldviews, and political ideology is a central element of worldviews for many people. though some studies show that mortality salience leads to a shift toward more conservative attitudes regardless of political orientation (cohen et al., ; landau et al., ) , others show it leads to polarization, with conservatives endorsing more conservative attitudes and liberals endorsing more liberal ones (kosloff et al., ) . a meta-analytic review of this literature concluded that there is evidence for both tendencies, with the evidence being somewhat stronger for polarization (burke et al., ) . we are seeing this polarization playing out in both proximal and distal reactions to the threat of the virus. one current example of intensified reactions may be the powerful and sometimes violent protests in response to the killing of george floyd. this was far from the first unjust killing of a black person by the police, but it has clearly led to the most intense and widespread outrage and protests of any of them. perhaps it is the final tragic straw that broke the proverbial camel's back that occurred at a time when people had more time to engage due to shutdowns in response to the virus. but we argue that the background of death thought accessibility due to the pandemic probably intensified these reactions. fueled by a greater need for terror management, many people jumped fervently onto this cause as a way to feel that they are doing something of value in their lives, when in reality their ability to feel that way has been so hampered by loss of jobs and income, social isolation, and difficulties in making sense of the tragedy and divisiveness that has emerged in the wake of the virus. meaning and significance derived from participating in these mass social protests may ironically increase death salience as protestors gather in large groups that exponentially increase their chance of exposure to the virus. protests are also threatening, in that-though most have been peacefulthere is a lurking potential for violence with police and counter-protesters. many people have suffered serious, life-altering injuries by "nonlethal weapons" used by police officers to control protests and riots; for example, some individuals have been permanently blinded after being shot with rubber bullets (bauerlein & calvert, ; sheikh & montgomery, ) . interestingly, the individuals who are risking their lives to protest racial inequality tend to hold the same political views as those who believe that social distancing should be practiced (diamond, ; nguyen, ) . thus, these individuals are in a precarious position, where bolstering one's cultural worldview by protesting racial inequality involves directly, consciously putting oneself in danger of violence and disease. tmt suggests that affirming one's worldview-along with one's self-esteem and close relationships-maintains psychological well-being by buffering death-related thoughts by providing outlets for symbolic immortality. the current crisis raises the intriguing question of what happens when affirming one's worldview involves putting oneself directly in harm's way and, consequently, never fully removing the salience of death from one's consciousness. consistent with the theoretical writings of becker ( ) , lifton ( ) , and yalom ( ) , tmt suggests that when people are not effectively managing their existential terror by building a meaningful and purposeful life, death anxiety and maladaptive ways of dealing with that anxiety are the common result. indeed, it has been argued that both death anxiety and ineffective or disrupted anxiety-buffer functioning are transdiagnostic vulnerability factors for psychological disorder (iverach et al., ; yetzer & pyszczynski, ) . if fear of death does indeed motivate the pursuit of meaning in life, selfesteem, and close relationships, then problems in managing death concerns exacerbated by the pandemic would leave people overwhelmed with anxiety and therefore more vulnerable to psychological disorder. experimental research has shown that reminders of mortality exacerbate phobias, obsessive-compulsive behaviors, depressive affect, and anxiety (e.g., finch et al., ; menzies & dar-nimrod, ; mikulincer et al., ; strachan et al., ) . this may help explain why a recent review found that the pandemic is associated with increased reports of anxiety, depression, and stress (torales et al., ; wang et al., ) . the covid- pandemic might cause psychological distress in two ways that correspond with proximal and distal defenses against death-related thought. first, the pandemic has directly increased death anxiety by raising awareness of personal vulnerability; recent research shows that the pandemic has increased anxiety and fear regarding one's physical well-being (jungmann & witthöft, ) . in addition, maladaptive proximal defenses may entail harmful practices aimed at avoiding the virus; some people have gargled bleach and cleaning supplies to reduce their chances of catching the virus (gharpure et al., ) . people are also employing a variety of unhealthy distractions to shift their focus away from the threat of the virus, including opiate use (https://www.ama-assn.org/system/files/ - /issue-briefincreases-in-opioid-related-overdose.pdf) and gambling (https://www.marketwatch.com/story/online-poker-betting-hits-a-record-high-during-the -pandemic- - - ). the pandemic has also undermined distal defenses by hampering or eliminating the anxiety-buffering outlets that people typically rely on to believe that they are valuable contributors to a meaningful world. when people lose their jobs and cannot pursue their financial, educational, and career goals, they are losing important sources of self-esteem. social relationships, which play such a major role in managing death fears, have also been hampered by the lockdown and social distancing measures. single people looking for a potential life partner have largely had to put this pursuit on hold. covid- related stress resulting from all of these aspects of the pandemic is associated with lower levels of meaning in life and life satisfaction (trzebiński et al., ) . inadequate distal defenses are likely to affect the need for proximal defenses and vice versa. increased death awareness associated with the threat of covid- is difficult to successfully manage because covid- has undermined access to many aspects of people's anxiety buffers; compromised anxiety buffers leave people vulnerable to experiencing higher levels of death anxiety than usual. how might one manage these overwhelmed death-related defenses? understanding the existential threats associated with the pandemic and reflecting on the proximal and distal defenses one uses to cope with them may help people develop new coping skills in these unprecedented times. for individuals experiencing high levels of death and health anxiety, managing one's engagement with virus-related information may help reduce explicit death anxiety. a recent study found that social media exposure during the pandemic is associated with poorer mental health as it likely contributes to the persistent salience of the virus and its mortality threat (gao et al., ) . engaging in and acknowledging the efficacy of best practices for avoiding infection is another potentially useful strategy for reducing anxiety. feelings of meaninglessness resulting from the loss of social relationships and selfesteem sources may be addressed by finding new sources of meaning, significance, and interpersonal connection; preferably ones that don't increase the threat of contracting or spreading the virus. reported increases in homebased hobbies such as baking bread or exercise have become popular as ways to derive a sense of meaning and value during the pandemic (vanderwerff, ). social events that have been cancelled have sometimes been redesigned as covid-friendly occasions-for example, the increasing popularity of drive-in theaters; online education; outdoor activities, such as hiking, where one can socialize while maintaining distance; and virtual get-togethers for parties, weddings, and funerals. "the new normal" has quickly become a cliché in light of the pandemic, yet it succinctly describes people's adaptations for both avoiding the threat of death from the virus while still being able to pursue goals that give life meaning, value, and connection with others. acknowledging one's emotional distress in response to the pandemic may encourage more creative and constructive ways of coping with it. the tension between measures to keep us safe from this oft-deadly virus and the desire to reopen the economy and resume "normal" life can be viewed as a battle between proximal and distal defenses against death. proximally, we want to forestall death and feel safe from it in the short term. distally, we want to maintain the view that life is meaningful and that we are valuable contributors to that meaningful life. the fundamental dilemma is that measures that keep us safe in the moment often interfere with our ability to find meaning and significance in our lives. both are important psychological concerns, and finding the right compromise to sufficiently meet both needs is the great challenge every culture is facing. one tragic example of this is that to keep hospitals and nursing homes safe, loved ones are often not allowed to be with their sicker or dying friends and family members. the result is people facing their own and their loved ones' imminent death without the support systems that provide them with their deepest psychological security. perhaps understanding these issues from the perspective of tmt can help societies determine the best versions of the many compromises with which they contend as they move forward with life in the face of the covid- pandemic. tom pyszczynski, phd, is distinguished professor of psychology at the university of colorado at colorado springs. he received his phd in psychology from the university of kansas in . with his colleagues jeff greenberg and sheldon solomon, he developed terror management theory, which explores the role of death in life and suggests that cultural worldviews, self-esteem, and close personal relationships function to manage the potential for existential terror that results from the uniquely human awareness of the inevitability of death. he has also conducted research on clinical problems such as anxiety, depression, and posttraumatic stress disorder. he is coauthor or coeditor of several books, including hanging on and letting go: understanding the onset, progression, and remission of depression ( ) , in the wake of / : the psychology of terror ( ) mckenzie's research endeavors focus on the social and emotional outcomes of trauma exposure. in particular, mckenzie often applies social psychological theories, including terror management theory and objectification theory, to understanding how trauma exposure relates to social and motivational processes that are typically only studied in nonclinical samples. most americans are practicing social distancing effects of covid- home confinement on physical activity and eating behavior: preliminary results of the eclb-covid international online survey mortality salience and the spreading activation of worldview-relevant constructs: exploring the cognitive architecture of terror management social isolation in covid- : the impact of loneliness serious eye injuries at protests spur calls to ban rubber bullets the denial of death coverage of covid- and political partisanship: comparing across nations fauci loses support from republicans after trump criticism, poll shows fact check: coronavirus did not spread in the us because of an anti-trump conspiracy death goes to the polls: a metaanalysis of mortality salience effects on political attitudes two decades of terror management theory: a meta-analysis of mortality salience research how americans view the coronavirus crisis and trump's response you're hired! mortality salience increases americans' support for donald trump suddenly, public health officials say social justice matters more than social distance binge watching behavior during covid pandemic: a cross-sectional, crossnational online survey around % of people are wearing face coverings because of coronavirus, survey finds. the news and observer terror mismanagement: evidence that mortality salience exacerbates attentional bias in social anxiety an instant economic crisis: how deep and how long? mckinsey & company covid- mainly kills old people: so do most other diseases. the print trust in medical scientists has grown in u.s., but mainly among democrats are americans drinking their way through the coronavirus pandemic? forbes mental health problems and social media exposure during covid- outbreak knowledge and practices regarding safe household cleaning and disinfection for covid- prevention-united states the implications of death for health: a terror management health model for behavioral health promotion the causes and consequences of a need for self-esteem: a terror management theory attachment, self-esteem, worldviews, and terror management: evidence for a tripartite security system death anxiety and its role in psychopathology: reviewing the status of a transdiagnostic construct what we're missing when we condemn "violence" at protests health anxiety, cyberchondria, and coping in the current covid- pandemic: which factors are related to coronavirus anxiety fewer now say media exaggerated covid- risks, but partisan gaps remain riot or resistance? how media frames unrest in minneapolis will shape public's view of protest. the conversation terror management and politics: comparing and integrating the "conservative shift" and "political worldview defense" hypotheses deliver us from evil: the effects of mortality salience and reminders of / on support for the broken connection: on death and the continuity of life protesters, some armed, enter michigan capitol in rally against covid- limits leading cause of death in the us? hint: it isn't covid- death anxiety and its relationship with obsessive-compulsive disorder the existential function of close relationships: introducing death into the science of love towards an anxiety-buffer disruption approach to depression: attachment anxiety and worldview threat heighten death-thought accessibility and depression-related feelings conservatives charge liberals with social-distancing hypocrisy trust index: are hospitals inflating covid- death count? click orlando why anti-lockdown protests are a "magnet" for white supremacists and far-right extremists a dual-process model of defense against conscious and unconscious death-related thoughts: an extension of terror management theory thirty years of terror management theory: from genesis to revelation republicans still skeptical of covid- lethality study: nearly a third of americans believe a conspiracy theory about the origins of the coronavirus rubber bullets and beanbag rrunds can cause devastating injuries. the new york times republican voters give trump and gop governors cover to reopen the worm at the core: on the role of death in life poll: michigan voters show support for gov why some people are still in denial over coronavirus-and how to convince them to take it seriously delay and death-thought accessibility: a meta-analysis terror mismanagement: evidence that mortality salience exacerbates phobic and compulsive behaviors the outbreak of covid- coronavirus and its impact on global mental health reaction to the covid- pandemic: the influence of meaning in life, life satisfaction, and assumptions on world orderliness and positivity how to bake bread immediate psychological responses and associated factors during the initial stage of the coronavirus disease (covid- ) epidemic among the general population in china existential psychotherapy. basic books terror management theory and psychological disorder: ineffective anxiety-buffer functioning as a transdiagnostic vulnerability factor for psychopathology phd, is a regents professor of the authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. the authors received no financial support for the research, authorship, and/or publication of this article. key: cord- -sgu ayvw authors: kolic, blas; dyer, joel title: data-driven modeling of public risk perception and emotion on twitter during the covid- pandemic date: - - journal: nan doi: nan sha: doc_id: cord_uid: sgu ayvw successful navigation of the covid- pandemic is predicated on public cooperation with safety measures and appropriate perception of risk, in which emotion and attention play important roles. signatures of public emotion and attention are present in social media data, thus natural language analysis of this text enables near-to-real-time monitoring of indicators of public risk perception. we compare key epidemiological indicators of the progression of the pandemic with indicators of the public perception of the pandemic constructed from approx. million unique covid- -related tweets from countries posted between th march and th june . we find evidence of psychophysical numbing: twitter users increasingly fixate on mortality, but in a decreasingly emotional and increasingly analytic tone. we find that the national attention on covid- mortality is modelled accurately as a logarithmic or power law function of national daily covid- deaths rates, implying generalisations of the weber-fechner and power law models of sensory perception to the collective. our parameter estimates for these models are consistent with estimates from psychological experiments, and indicate that users in this dataset exhibit differential sensitivity by country to the national covid- death rates. our work illustrates the potential utility of social media for monitoring public risk perception and guiding public communication during crisis scenarios. the covid- pandemic has brought about widespread disruption to human life. in many countries, public gatherings have been broadly forbidden, mass restrictions on human movement have been introduced, and entire industries have been paralysed in attempting to lower the peak stress on healthcare systems [ ] . however, the degree to which these restrictions have been enforced by law has varied over time and by location, and their success in mitigating public health risks depends on the extent of cooperation on the part of the public. a key determinant of the public's behaviour and their cooperation with state-imposed social restrictions is the public's emotional response to, and their perception of the the risk presented by, the pandemic. however, the evolution of emotions and risk perception in response to disasters is not well-understood, and there is a need for more longitudinal data on such responses with which this understanding can be improved [ ] . our goal is thus to contribute to bettering this understanding, and we do so by exploring the empirical relationships present between the progression of the covid- pandemic and the public's perception of the risk posed by the pandemic. we explain our findings in terms of the existing body of literature surrounding public perception of risk, disasters, and human suffering in cognitive psychology. in particular, we draw from psychophysics, the field that studies the relationship between stimulus and subjective sensation and perception [ ] . the search for psychophysical "laws" of perception has existed since at least the mid- th century with the proposing of the weber-fechner law [ ] , which posits that the smallest perceptible change ds in a physical stimulus of magnitude s is proportional to s. thus, the perceived magnitude p of such stimuli follows dp ∝ ds s . ( ) in the continuum limit, this implies that p grows logarithmically with the physical magnitude s of the stimulus. more recently, empirical studies by s. s. stevens [ ] supported, instead, a power law relationship between human perception of a stimulus and the physical magnitude of the stimulus: p ∝ s β . summers et al. [ ] extended this concept to human sensitivity to war death statistics and found that a power law with exponent β = . best fit the data. a number of further studies have corroborated the extension of these psychophysical laws describing the subjective perception of physical magnitudes to the subjective evaluations of human fatalities [ , , ] . in all of these, perception is a concave function of the stimulus, meaning that the larger the stimulus magnitude, the more it has to change in absolute terms to be equally noticeable. thus, perception is considered relative rather than absolute, implying that our judgments are comparative in nature. this observation has been shown to account for deviations from rationality in economic decision-making [ ] . these proposed psychophysical laws of human perception present an opportunity for monitoring a population's response to a disaster scenario such as the covid- pandemic. by evaluating the goodness of fit of these models to data on the perception of the progression of the pandemic, and determining the parameter values of such fits, we can describe the sensitivity of populations to the state of such crises, with important implications for risk communication and disaster management. to this end, we make use of a massive twitter dataset consisting of user-posted textual data to study the public's emotional and perceptual responses to the current public health crisis. twitter provides convenient access to the conversation amongst members of the public across the globe on a plethora of topics, and many authors are studying several aspects of the public's response to the pandemic with it. twitter is a particularly appropriate tool under conditions of physical distancing requirements and furlough schemes, where online communication has become more than ever a central feature of everyday life. moreover, results from psycholinguistics and advances in natural language processing techniques enable the extraction of psychologically meaningful attributes from textual data. with this dataset, our general approach is to offer a quantitative, spatiotemporal comparison between indicators of the state of the pandemic and the topics and psychologically meaningful linguistic features present in the discussion surrounding covid- on social media on a country-by-country basis, for a selection of countries. our work is novel in that, to our knowledge, it is the first to use a large social media dataset spanning multiple countries to model the perceptual response of countries' citizens to the pandemic in the context of risk perception. to date, empirical validation of the aforementioned psychophysical laws has largely taken place in controlled laboratory settings, in which decisions, actions, and scenarios are artificial or hypothetical. our work thus contributes to the body of literature surrounding risk perception by investigating these laws in a naturalistic setting. however, there have been numerous authors using social media to analyse the public response to the covid- pandemic. this includes work that has focused on the psychological burden of the social restrictions. for instance, stella et al. [ ] use the circumplex model of affect [ ] and the nrc lexicon [ ] to give a descriptive analysis of the public mood in italy from a twitter dataset collected during the week following the introduction of lockdown measures. in addition, venigalla et al [ ] has developed a web portal for categorising tweets by emotion in order to track mood in india on a daily basis. others have instead focused on negative emotions, as in the work of schild et al. [ ] , where they study the rise of hate speech and sinophobia as a result of the outbreaks. more specifically on perception, dryhust et al. [ ] measured the perceived risk of the covid- pandemic by conducting surveys at a global scale (n ∼ ) and compared countries, finding that factors such as individualistic and pro-social values and trust in government and science were significant predictors of risk perception. de bruin and bennett [ ] perform similar work in the united states. the closest work we have been able to find to our own is that of barrios and hochberg [ ] , in which the authors combine internet search data with daily travel data to show that regions in the united states with a greater proportion of trump voters exhibit behaviours that are consistent with a lower perceived risk during the covid- pandemic. despite the above, we have been unable to find work that combines large-scale social media data with linguistic analysis to offer a spatiotemporal, quantitative analysis of emotion and risk perception during the covid- pandemic across multiple countries. beyond the covid- pandemic, our work is related to a small but growing body of literature on the use of data science in understanding human emotion and risk perception. in such work, natural language analysis has succeeded in supporting established linguistic theories such as the importance of the distribution of words in a vocabulary as a proxy for knowledge [ ] , and regarding the relation between the uncertainty of events and the emotional response to their outcome [ , ] . for instance, using textual data from twitter, bhatia found that unexpected events elicit higher affective responses than those which are expected [ ] . in another instance, the same author conducted experiments with participants and predicted the perceived risk of several risk sources using a vector-space representation of natural language, concluding that the word distribution of language successfully captures human perception of risk [ ] . similar work has been conducted by jaidka et al. [ ] in the area of monitoring public well-being, in which they compare word-based and data-driven methods for predicting ground-truth survey results for subjective well-being of us citizens on a county-level basis using a . billion tweet dataset constructed from to . the remainder of this paper is laid out as follows. in section , we present the data set used in the subsequent analysis. in section , we provide further details on the approach followed to explore the relationships between indicators of the state of the pandemic and the public's perception of the pandemic, and discuss possible explanations for our observations by drawing on psychological literature. in section , we summarise and offer concluding remarks, along with a discussion of the limitations of the current work and suggestions for avenues of future work. in the following analysis, we make use of the set of tweets gathered by j. banda et. al [ ] , which are obtained and mantained using the twitter free stream api . at the time of writing, this data set consists of ∼ million original tweets spanning from march , to june , . data is collected according to the following query filters : "covid ", "coronavirus-pandemic", "covid- ", " ncov", "coronaoutbreak", "coronavirus" , "wuhanvirus", "covid ", "coronaviruspandemic", "covid- ", " ncov", "coronaoutbreak", "wuhanvirus". for our analysis, we consider only the english and spanish tweets with a non-empty selfreported location field. we process every self-reported location using openstreetmaps [ ] and remove non-sensical locations (e.g. "mars", "everywhere", "planet earth"). this allows us to group the remaining tweets by country and proceed with our analysis on a country-by-country the free stream api randomly samples around % of the total tweets for the given queries a number of publicly available twitter datasets have emerged in relation to the pandemic. we chose to work with this dataset since it used the most generic query terms among all the publicly available datasets we considered, and we wanted the least amount of bias possible for our analysis. basis. to assure the statistical significance of our analysis, we keep the countries with the highest number of tweets for each language, resulting in a geolocated twitter dataset of ∼ million tweets posted by ∼ million users on different countries, which we summarise in table . we measure the progression of the pandemic with the number of covid- confirmed cases and deaths for all the countries in our analysis. the data was made publicly available by our world in data repository [ ] . in particular, we take the daily covid- cases and deaths, both in linear and logarithmic scale, since these are four epidemiological indicators that are most frequently used to summarise the state of the pandemic, and are therefore frequently encountered by the public. in this section, we study the public's perception of the pandemic on a country-by-country basis, using the countries with the highest number of tweets in the observation period (see table ). we do this on a country-by-country basis since the pandemic has often invoked nation-level responses, making nation-level analysis the most natural geographic scale. our broad approach is to inspect and compare the linguistic features of the tweets released by users in the twitter dataset described in section . with the epidemiological data described in section . . our goal is to explore the public's perception of the pandemic. to do this, we analyse the linguistic features present in the textual data generated by twitter users, and map these features to psychologically meaningful categories that are indicative of the twitter users' perception. here, we are assuming that the words used by these twitter users are indicative of their internal cognitive and emotional states [ ] , which is supported in [ ] where they predict the perception of risk using text data. thus, we quantify the linguistic content of each tweet using the linguistic inquiry and word count (liwc) program [ ] . liwc has been widely adopted in several text data analyses, and it has proven successful in applications ranging from measuring the perception of emotions [ ] to predicting the german federal elections using twitter [ ] . liwc operates as text analysis program that reports the number of words in a document belonging to a set of predefined linguistically and psychologically meaningful categories [ ] . for our purposes, a document is a tweet d t i posted on date t and from a user based in country i. liwc represents documents as an unordered set of words, and a liwc category l is similarly a set of words associated with concept l. for a given document d t i , the linguistic score p l for category l is the percentage of words in d t i that belong to l: there are many such categories l, including family, work, and motion. we capitalise such category titles, and use the titles to refer to either the set of words associated with that category or to refer to the category itself. linguistic scores from eq. ( ) for individual tweets will be noisy, as they are short documents. moreover, we are interested in the average response of the population of a country. for this reason, we group the tweets by country i and by date t, and denote these sets of tweets as we then compute the national linguistic score (nls) for category l as the average of the linguistic scores over documents in d t i relative to an empirically observed twitter base rate p l b : the base rates p l b for the use of words on twitter associated with category l are given in [ ] . using eq. ( ) for all the selected linguistic categories, we construct multidimensional country-level time series that represent the evolution of the public perception of the pandemic, similar to the linguistic profiles introduced by tumasjan et al. [ ] . in figure , we show the collection of nlss for a selection of relevant linguistic categories. we observe clear trends that, in most cases, are synchronized between countries and languages. in particular, most categories associated with emotion -notably affect, anger, anxiety, positive emotion, negative emotion, and swear words (swearing is associated with frustration and anger [ ] ) -have their highest scores in mid-to-late march, when the world health organisation (who) announced the pandemic status of covid- and most western countries introduced more stringent social restrictions [ ] . these scores decay thereafter, indicating a relaxation of the emotional response in the conversation. this is consistent with results reported by bhatia regarding the affective response to unexpected events [ ] . a qualitatively similar trend can be seen in the social processes panel, the category involving "all non-first-person-singular personal pronouns as well as verbs that suggest human interaction (talking, sharing)" [ ] . we also observe that health-related categories such as death and health show an overall rising trend, with death rising most rapidly throughout march. these categories, with the exception of positive emotion and health, peak again in the united states at the end of may, coinciding with the murder of george floyd and the subsequent black lives matter protests. such universal trends are not apparent by visual inspection in the money, risk, and sadness panels. an additional feature of these plots is the absolute scale of these values: in all cases, there is a significant percentage change from their baseline values, with large percentage increases observed initially in the use of words associated with anxiety and later with death, and a moderate percentage increase in the use of words associated with risk. in this section, we explore the relationship between the nlss described in section . , which we use as a proxy for the public's perception, and the intensity of the pandemic, which we assume is the stimulus triggering this perception. our measure of the intensity of the pandemic is the number of covid- cases and deaths from the data described in section . . a straightforward way of approaching this relationship is by computing the correlations between the nlss and the epidemiological data in a per-country basis, and we show the average across countries of these per-country correlations in figure . on the one hand, we observe significant negative correlations in emotionally charged categories (eg. swear words, anger, anxiety, affective processes), indicating a decay in emotion as the pandemic intensifies. conversely, categories related with health and mortality (death, health) and analytical thinking (analytic) show significant positive correlation . we believe the trends we observe in fig. and the correlations we observe in fig. are consistent with the notion of psychophysical numbing. this term was introduced by robert jay lifton [ ] , and developed by paul slovic [ , ] in the context of human perception of genocides and their associated death tolls, to describe the paradoxical phenomenon in which people exhibit growing indifference towards human suffering as the number of humans suffering increases. by inspecting the correlations between the nlss and the epidemiological indicators, we find that as the pandemic intensifies -in the sense of an increasing number of cases and deaths reported daily -our emotional response diminishes, as expected from a psychophysical numbing phenomenon. specifically, we observe negative correlations between almost all components of the nlss associated with affect -affective processes, anger, anxiety, negative emotion, positive emotion, and swear words -and the epidemiological data . by inspecting figure , we see that every country exhibits similar downward trends in these components and, with the exception of anxiety, are all significantly lower than their baseline values throughout the observation period. this unusually low and decreasing affect word count is accompanied, conversely, with a growing awareness of the morbidity of the situation in that we observe significant positive correlations between the death nlss and the daily national cases and deaths, indicating that the decrease in affect occurs simultaneously with and despite an attentional shift towards covid- related mortality. we also observe a simultaneous increase in the analytic component of each english-language dataset over this same period, indicating a movement towards more logical and analytical, rather than intuitive and emotional, thinking. the potential implication of this is that the public is less perceptive of the risk that the pandemic poses to public health, since their emotional response is reduced and reducing [ ] . when analysing these correlations, we found that, overall, the cumulative cases and deaths correlate better with most linguistic categories than the daily data. however, while this is sensible in the early stages of the pandemic, it is unlikely to remain the case over a long time horizon due to humans' finite memory. we therefore proceeded with our comparison using the daily epidemiological data alone for this reason. the only exception is the cross-country average of the sadness component of the nlss, which is positively correlated with the epidemiological indicators and appears to be driven only from argentina's, chile's, and colombia's increasing use of words related to sadness. the remaining countries remain stationary at a lowerthan-baseline value for this component. unfortunately, the spanish liwc dictionary does not yet have an analytic category. for example, van bavel et al. [ ] and loewenstein et al. [ ] describe that risk perception is driven more by association and affect-based processes than analytic and reason-based processes, with the affect-based processes typically prevailing when there is disagreement between the two modes of thinking. the negative correlations between the intensity of the pandemic and affective processes, together with its positive correlation with the prevalence of analytic processes, suggests that public risk communication could be adjusted to re-balance the degree of affective and analytic thinking amongst members of the public to achieve favourable risk avoidance behaviour and, consequently, favourable public health outcomes. to support our claim that these observations are attributable to psychophysical numbing, we construct word co-occurrence networks using tweets in our dataset. given a set t of tweets, the word co-occurrence network g(t ) is represented by a weighted adjancency matrix a(t ) in which the nodes are words belonging to the death and affect liwc dictionaries. entry a ij (t ) counts the number of co-occurrences between words i and j across all tweets in t , and is computed as where b tk (t ) counts the number of instances of word k in tweet t ∈ t . we ignore self-edges by imposing a ii = , since it is the relationship between distinct words that is of interest. (see appendix b. for further details on the construction of these networks.) if the psychophysical numbing effect is legitimate, we expect that words in the death dictionary co-occur more frequently with other death-related words and less frequently with words in the affect dictionary. we construct three such networks by aggregating the word co-occurrences over three distinct periods: th march to th april , th april to rd may, and th may to th june. as we discussed previously, the first period coincides with the pandemic status of covid- declared by the who and has a high affect score but a low and increasing death score; the second one has a high and relatively stable death score and a decreasing affect score; and the third has a high death score but one in which the affect scores and some of its subcategories (e.g. anger, anxiety, negative emotion) increase again, which we attribute -at least partly -to the public response to the murder of george floyd and the subsequent black lives matter protests. in constructing these networks, we weight each country equally by taking a random sample of approximately , tweets from each country. th march - th april in this network (see fig. a ) we see two main clusters emerging. the first consists mostly of words associated with death (left), and the second of words associated with affect (right). the appearance of some of the affect-related words in the death cluster can be explained given the context of the pandemic. for example, the word "positive" is likely used in reference to the number of people that have tested positive for covid- , which is closely related to the conversation around covid- cases and deaths. similarly, the word "panic" is likely reflecting the early conversations around panic-buying of household goods, for example toilet paper and hand-sanitiser, and the word "isolat*" is likely used in calls for symptomatic individuals to self-isolate. thus, while some instances of affect-related words that appear in this predominantly death-based community are harder to explain without appealing to the existence of a true subjective experience of affect amongst the twitter users (e.g. "risk*"), the most important (in terms of node degree) of these affect-related words are more likely being used here in an affect-free sense and are appropriately grouped with death-related words here given the context of the pandemic. thus the community structure we observe is consistent with our hypothesis of a separation between words belonging to the death and affect dictionaries. th april - rd may in this network (see fig. b ), the two-cluster structure seen in the previous snapshot remains, with the cluster more centered on death on the left and a cluster corresponding to almost exclusive use of affect-related words on the right. the size of the death-related cluster has increased relative to the affect-based community, reflecting the higher death nls during this period. two new and important affect-related nodes appear in the death-based community for this time period: "care" and "fail*". these can once again be plausibly explained by the context of the pandemic. for example, the appearance of the word "care" in the death-related community can be explained in terms of the conversation surrounding the health care system and death care industries, the number of covid- patients being admitted to intensive care units, andparticularly for the united kingdom -the number of deaths that have occurred in care homes for the elderly. these are all clearly related to covid- deaths, and the word "care" in this context most likely constitutes part of the noun and topic of conversation rather than any expression of emotion. the word "fail*" could reflect the discussion around failures on the part of governments to respond with sufficient vigor to the public health crisis -e.g. in terms of a failure to impose social restrictions in a timely manner or to meet testing quotas or quotas on the provision of personal protective equipment for key workers. for example, the polling company yougov finds that approximately % of respondents felt during this period that the us and uk governments had been handling the pandemic well, and that these numbers decreased throughout this period to approximately % [ ] . this does not however exclude the possibility that the appearance of "fail*" indicates a subjective emotional experience: it is possible that twitter users that fixate on government failures are doing so as a result of a sense of outrage with regard to these perceived failures. whether such outrage is motivated specifically by the human fatalities themselves or is merely a manifestation of broader political hostilities and polarisation in modern society remains open. thus, while the appearance of "sure*", "fail*", and some other minor affect-related terms in the death-community may be truly indicative of emotion in the conversation around covid- fatalities, the presence of many of the most highly co-occurring affect-related words in this predominantly death-related community could be explained by their appearance in common phrases related to covid- fatalities, e.g. the "death care" and "health care" industries, "care homes", "testing positive" for the virus etc. these words, therefore, do not necessarily reveal emotion in the current context. we thus argue once again that this co-occurrence network and its community structure shows that death-and affect-based words are well-separated, consistent with our claim of psychophysical numbing. th may - th june our argument remains unchanged for this period (see fig. c ). the only notable difference for this period is that a significant proportion of the conversation surrounding death is focused on the political issues that inspired the black lives matter protests and the protests themselves. this is apparent from the appearance of the word "protests" in the left-hand side's death-related community. altogether, this analysis demonstrates that words indicating a subjective emotional/affective experience and words related to death are well-separated in this twitter data, which is consistent with the notion of psychophysical numbing as an explanation for the trends and correlations observed in figures and . for completeness, we include the equivalent co-occurrence graphs for the spanish-language tweets in appendix b. , from which similar conclusions can be drawn. in the previous section, we demonstrated our finding that as the pandemic intensifies, the proportion of words that appear in the set of tweets posted in each country that indicate emotion diminishes over time. this indicates that the actual emotional response to the pandemic diminishes as the intensity of the pandemic increases, implying a psychophysical numbing effect. we supported this explanation by showing that the word co-occurrence networks induced by our set of tweets host a community structure that separates words in the death and affect dictionaries, suggesting that people do not talk about covid- deaths in a highly emotional tone. the following sections model the relationship between the progression of the covid- pandemic and the twitter users' perception using grounded theories of psychophysical numbing. our analysis suggests that the public's perception of the progression of the pandemic is logarithmic or, at least, sublinear. from figure , we observe that the correlation magnitudes between nlss and epidemiological data are generally larger in absolute value whenever the latter are taken in logarithmic scale. to exemplify this observation, we show in figure the z-scores of the death nlss and of the logarithm of the daily number of deaths and cases within each country. the general correspondence between all three normalised features in each country is striking . we propose that this can be explained in terms of the weber-fechner law [ ] , which is a quantitative statement with its origins in psychology and psychophysics regarding humans' perceived magnitude p of a stimulus with physical magnitude s. it states that a human's perception of the magnitude of a stimulus varies as the logarithm of the physical magnitude s of the stimulus, meaning we are more sensitive to ratios when comparing different physical magnitudes than we are to absolute differences. in the continuum limit, eq. ( ) gives the following functional form for the weber-fechner law: recall that the z-score of a sequence of observations y = (y , · · · , y t ) is given by z = (y − µ y )/σ y , where µ y and σ y are the mean and standard deviation of y, respectively. we note that the correspondence is weaker for australia, nigeria, and south africa due to the relatively low number of cases in these countries (see fig. in the appendix for reference). the correspondence is also weaker in spain, for two reasons: due to its revision of the number of cases in late may, resulting in a day of "negative deaths"; and due to their having recorded a day with no covid- -related deaths, which was a significant event given that spain had seen many deaths until that point. (c) may th to june th, . figure : snapshots of the word co-occurrences associated with death (green labels) and affect (red labels) for english-language tweets aggregated across all analyzed countries in three different time windows (see sub-captions). the nodes are coloured according to their community label as obtained by maximising modularity with the louvain algorithm [ ] . we filtered edges with weight below co-occurrences for visualisation purposes. (t) and the national daily death rate is given in parentheses for each country. data is smoothed with a -day moving average and standardized with their z-score to make them visually comparable. vertical lines represent peaks in the death discourse caused by exogenous events (see main text for details) which we remove from the time series. table : results from the fit of the weber-fechner law to the observed relationship between the death nls and the logarithm of the daily number of deaths in each country (see figure ) . overall, this model best describes the relationship between the daily number of deaths local to each country and the death nls. where k and s are real-valued parameters and r(t) the residual. parameter k determines the sensitivity of perception to changes in the stimulus s, while s determines the minimum threshold that the stimuli s must overcome in order to be perceived. the residual term r(t) is a random variable representing noise not directly captured by the stimulus. for instance, exogenous events can trigger abrupt peaks in the death score. this is the case, for example, with the murder of george floyd in the united states, or the peak in nigeria around april th , triggered by a number of prominent african figures dying from covid- around that day, including the nigerian president's top aide (see [ ] ). in order to test the weber-fechner law, we fit a linear regression model to p death i (t), the death nls time series in country i, and log s i (t), the daily number of deaths in the same country, and summarize the results of these fits in table . we find that eq. ( ) accurately models the data, with significant coefficients (p-value < . ) for all countries except spain. the sensitivity parameter k has the same order of magnitude for all significant countries. however, the country with the lowest k is ∼ times less sensitive than the highest, indicating that twitter users in different countries may react differently to the evolution of the pandemic. the minimum stimuli threshold s , in the other hand, is always small: most countries, except for the united states and the united kingdom, need only one covid- death in a given day in order to be perceived. conversely, the united states and united kingdom need approximately and deaths to be perceived, which is small compared to the thousands of daily deaths registered in these countries during the observation period. table : the results from the fit of a power law to the relationship between the death nls and the national daily death count. this is the best model in some cases, though is outperformed by the weber-fechner law most times. *while we fit this model assuming a log-log relationship between p and s, we compute r with linear p to make it comparable to the model implied by the weber-fechner law (see eq. ( ) in appendix a for details). this may cause negative values of r as is the case for spain. an alternative functional form for the relationship between human perception p of a stimulus and the physical magnitude s of the stimulus is a power law relationship where ν and β are parameters determining the perception from a stimulus of unit magnitude and the growth rate of the perception as a function of the stimulus magnitude, andr(t) is a residual term. this form has been shown to outperform the weber-fechner law in characterising human perception in a number of empirical studies [ ] . we also therefore report the results of this model fit to the relationship between the death nls p death i (t) and national daily death counts s i (t) for each country i, reporting our results in table . in all cases, we observe sublinear exponents β for the perception of the daily deaths data, with significant exponents (p-value < . ) ranging between . and . . these exponents are of the same order of magnitude as the β of . reported in [ ] , where in several laboratory experiments they measure psychophysical numbing in participants' perception of death statistics. as discussed previously, the data for spain is unusual for a number of reasons, thus the model does not accurately describe the data in this instance. these results suggest that twitter users in certain countries are more sensitive to change in the number of deaths than others. both the weber-fechner law and power-law relationships between the death nls and the daily number of reported deaths accurately model the data. each captures the phenomenon in which "the first few fatalities in an ongoing event elicit more concern than those occurring later on" [ ] . by way of comparison, we present in table (nrmse), defined as for these models, in addition to a linear model between p death i (t) and s i (t) as a baseline "null" model. here, e(t) = p(t) −p(t) is the model residual, and n is the sample size. the models are directly comparable in this sense, since each involves only two parameters. bhatia [ ] made a similar model comparison to test psychophysical laws for subjective probability judgements of real-world events, in that case finding that the linear relationship was the best. in our case, however, a linear relationship between s and p is significantly worse than the present concave models of perception (see appendix a for the results of the linear model), reinforcing our hypothesis of psychophysical numbing. while the weber-fechner law is better than the power law model overall, the difference in their goodness of fit -as measured by the nrmse -is marginal. both are reasonable descriptions of the observed relationship, and similar conclusions can be drawn from both. in particular, the parameters k and β from the weber-fechner law and power law, respectively, are analogous in their interpretation as the measure of the sensitivity of the nation's twitter users to changes in the national covid- daily death rate. to illustrate this, we rank the countries in our dataset in order of sensitivity to changes in the local death rate, as measured separately by these two parameters, and plot the correlation between the countries' ranks in figure . here, low rank indicates high sensitivity to changes in the number of daily deaths nationally. the correlation between the two methods of ranking -according to k, the weber-fechner law slope parameters, and according to β, the power law model exponents -is high, with correlation coefficient . . this shows that the sensitivity of each country is relatively robust between models. by both measures, therefore, twitter users tweeting in english and spanish from australia and argentina, respectively, appear to be the most sensitive to changes in the national daily death rate, while twitter users posting in english from south africa, india, and nigeria and in spanish from spain and chile appear to be the least sensitive to these changes. figure : comparison of the rank of each country as determined by their k and β parameters in the weber-fechner and power-law fits, respectively, which determine the sensitivity of twitter users tweeting from each country to changes in the number of daily reported deaths. low rank indicates high sensitivity relative to the remaining countries. the correlation between countries' ranks from both measures is high at . . we explored the country-by-country relationship between the linguistic features present in a large set of tweets posted in relation to the covid- pandemic, and the progression/intensity of the pandemic as measured by the daily number of cases and deaths in each country we consider. by considering the change, relative to a baseline, in the percentage of words present in each tweet that are associated with a number of psychologically meaningful categorieshere called linguistic scores -we observed significant trends that we believe are indicative of a psychophysical numbing effect [ ] . we found that the national linguistic scores (nlss, see eq. ( )) associated with emotion and affect decrease as the pandemic intensifies. this is in spite of a greater attentional focus on death and mortality and a simultaneous increase in use of words indicating analytic reasoning. we showed, by constructing word co-occurrence networks on different time periods of the pandemic, that words related to death co-occur more frequently with other words related to death than they do with words indicating affect and emotion, and that this separation of affect from the conversation around death is also revealed by the community structure of this network. this is consistent with the notion of psychophysical numbing, which we believe explains these observations. we also showed that the psychophysical laws of weber-fechner and of power law perception in humans accurately model the relationship between the frequency of words related to death and the actual daily number of covid- deaths in each country. we estimated sub-linear exponents in the power law perception function that are of similar values to values previously estimated from psychological experiments [ ] . these exponents, together with parameter k of the weber-fechner law (see eq. ( )), tell us how sensitive the twitter users in each country are to their national covid- daily deaths, and were seen to vary by country, indicating intercountry differences in risk perception and sensitivity to death rates. such sensitivities were consistent across models (see fig. ) suggesting that these measures of a nation's twitter users' sensitivities to changes in the national death rate are robust features of the data. our findings illustrate the signaling power of twitter, and demonstrate its potential use as a tool for monitoring public perception of risk during large-scale crisis scenarios. with the modelling and visualisation approaches we employ in this paper, policy-makers and public officials could track in near-to-real-time the public's attitudes towards threats to public well-being and the prevalence of factors important to public perception of risk, including degree of outrage and relative attentional focus on the threat. our findings also imply a functional form for agent perception of the system state in models of opinion dynamics. this will be instrumental for developing coupled opinion dynamics-epidemiological models, in which the bidirectional relationships between human perception, human behaviour, and epidemic progression are modelled endogenously. a natural extension to this work would involve nowcasting and/or forecasting of certain economic indicators. it has also been limited in that we assumed that only the national death rate is a significant predictor of perception. a more complete analysis should account for the effect of other countries' death statistics as a driver of local perception, or more broadly an advancement of a process-level explanation of the cross-cultural differences we observe in the sensitivity to death statistics. this analysis could also be enhanced by relating these measures of risk perception to behavioural data, which -since "people's behavior is mediated by their perceptions of risk" [ ] -may be useful for understanding the role of emotions in driving behaviours that are conducive to public health during crises. further, a deconstruction of the aggregate indicators we have developed to the state and regional level may be necessary to more accurately characterise the relationship between local crisis progression and human risk perception. we also stress that the results presented in this paper may be indicative only of the responses of twitter users posting from each of these countries in each of these languages, so extrapolating these results to the broader population will only be possible with a better understanding of the biases present in, and representativeness of, the dataset at hand. bk acknowledges funding from the conacyt-sener: sustentabilidad energtica and jd acknowledges funding from the epsrc industrially focused mathematical modelling centre for doctoral training centre. the authors declare that they have no competing interests. the twitter data used in the manuscript is collected and maintained by banda et al. at the panacea lab [ ] , and it is available at their website http://www.panacealab.org/covid . the data on covid- confirmed cases and deaths were obtained from the "coronavirus pandemic (covid- )" page of the our world in data website [ ] , and the stable url for this data is https://covid.ourworldindata.org/data/owid-covid-data.csv. bk and jd both conceived the idea, carried out the analysis, and wrote, read, and approved the final manuscript. • liwc: linguistic inquiry and word count. • who: world health organization. • nls: national linguistic score. in this section, we present further results of our models to give a more complete overview of their quality. besides the weber-fechner law and power law models (see eqs. ( ) and ( )), we use the following linear relationship between s and p as our benchmark model where a and b are parameters. we summarize our results for the linear model in table . for all models, we compute the r values where e(t) = p(t) −p(t) is the model residual, σ p = n t= (p(t) − µ p ) /(n − ) is the variance of p(t), and n is the sample size. the r values for all models are summarized in table . (note that as the power law model implies a log-normal residual, the r values can be negative.) from this table we see that, once again, the weber-fechner law is generally a better fit to the data across all countries, but that the power law and weber-fechner models are often comparable and significantly better than the linear model. we also show in figures and scatterplots of the death nlss against the logarithm of the daily number of deaths in each country, with the y-axis in linear-and log-scales, respectively. red lines indicate the line of best fit, with the slope equal to k and β in eqs. and , respectively. b word co-occurrence analysis in constructing the word co-occurrence networks presented in section . . , we preform basic text preprocessing, including taking the lower-case form of all letters, removing urls, removing punctuation, and removing the following small set of stopwords from the vocabulary: to, today, too, has, have, like. we retain hashtags, since liwc also recognises hashtags and because hashtags are an essential aspect to communications on twitter. it is also necessary to account for the fact that a number of "words" appearing in the liwc dictionary are in fact regular expressions to which many complete words in the twitter dataset map. for example, the "word" "isolat*" appears in the english liwc dictionary, to which each of the following words would map: "isolate", "isolated", "isolating". thus, construction of the word co-occurrence networks g i involves a two-step procedure: first, constructing the raw word co-occurrence networks g i , in which the nodes are words exactly as they appear in the twitter dataset; and then reducing this to a quotient graph g i by contracting nodes in g i that are matched by the same regular expression in the liwc dictionary. more formally: the liwc dictionary implies an equivalence relation ∼ on the vocabulary v implied by the twitter dataset, such that v ∼ u for words v, u ∈ v if both v and u are matched by the same regular expression in the liwc dictionary. the weights of edges between nodes v ⊂ v and u ⊂ v in g i are then taken to be where w g (x, y) is the weight of edge (x, y) in g. note that w g (x, y) = w g (y, x) and w g (x, y) = if (x, y) is not an edge in g. for completeness, we provide here the word co-occurrence graphs for the spanish language tweets. we omit a discussion of the results, since similar conclusions can be drawn from these as in the english counterparts. we include this section as a reference for the actual number of deaths in each country for the period we analysed throughout the paper, which we present in fig. . covid- government response tracker, blavatnik school of government risk perception and behaviors: anticipating and responding to crises psychophysical numbing: an empirical basis for perceptions of collective violence if i look at the mass i will never act": psychic numbing and genocide insensitivity to the value of human life: a study of psychophysical numbing psychophysical numbing: when lives are valued less as the lives at risk increase perception matters: psychophysics for economists #lockdown: network-enhanced emotional profiling at the times of covid- the circumplex model of affect:an integrative approach to affective neuroscience, cognitive development, and psychopathology emotions evoked by common words and phrases: using mechanical turk to create an emotion lexicon mood of india during covid- -an interactive web portal based on emotion analysis of twitter data go eat a bat risk perceptions of covid- around the world relationships between initial covid- risk perceptions and protective health behaviors: a national survey risk perception through the lens of politics in the time of the covid- pandemic distributional structure the effect of differential failure on expectation of success, reported anxiety, and response uncertainty discrepancy from expectation in relation to affect and motivation: tests of mcclelland's hypothesis affective responses to uncertain real-world outcomes: sentiment change on twitter predicting risk perception: new insights from data science estimating geographic subjective well-being from twitter: a comparison of dictionary and data-driven language methods a largescale covid- twitter chatter dataset for open scientific research -an international collaboration openstreetmap contributors coronavirus pandemic (covid- ) the psychological meaning of words: liwc and computerized text analysis methods the development and psychometric properties of liwc anxious or angry? effects of discrete emotions on the perceived helpfulness of online reviews predicting elections with twitter: what characters reveal about political sentiment the pragmatics of swearing beyond psychic numbing: a call to awareness responding to community outrage: strategies for effective risk communication using social and behavioural science to support covid- pandemic response risk as feelings covid- : government handling and confidence in health authorities fast unfolding of communities in large networks africa's top virus deaths the cognitive psychology of sensitivity to human fatalities: implications for life-saving policies vector space semantic models predict subjective probability judgments for real-world events snapshots of the word co-occurrences associated with death ("muerte", green labels) and affect ("afecto", red labels) for spanish-language tweets aggregated across all analyzed countries in three different time windows (see sub-captions). the nodes are coloured based on the community labels obtained by maximising modularity using the louvain algorithm we filtered edges with weight below co-occurrences for visualisation purposes the authors would like to thank mirta galesic, rodrigo leal cervantes, rita maria del rio chanona, françois lafond, and j. doyne farmer for helpful feedback, and to the oxford inet complexity economics group for stimulating discussions. key: cord- - kpupt authors: imre, gergely title: the involvement of regulated cell death forms in modulating the bacterial and viral pathogenesis date: - - journal: int rev cell mol biol doi: . /bs.ircmb. . . sha: doc_id: cord_uid: kpupt apoptosis, necroptosis and pyroptosis represent three distinct types of regulated cell death forms, which play significant roles in response to viral and bacterial infections. whereas apoptosis is characterized by cell shrinkage, nuclear condensation, bleb formation and retained membrane integrity, necroptosis and pyroptosis exhibit osmotic imbalance driven cytoplasmic swelling and early membrane damage. these three cell death forms exert distinct immune stimulatory potential. the caspase driven apoptotic cell demise is considered in many circumstances as anti-inflammatory, whereas the two lytic cell death modalities can efficiently trigger immune response by releasing damage associated molecular patterns to the extracellular space. the relevance of these cell death modalities in infections can be best demonstrated by the presence of viral proteins that directly interfere with cell death pathways. conversely, some pathogens hijack the cell death signaling routes to initiate a targeted attack against the immune cells of the host, and extracellular bacteria can benefit from the destruction of intact extracellular barriers upon cell death induction. the complexity and the crosstalk between these cell death modalities reflect a continuous evolutionary race between pathogens and host. this chapter discusses the current advances in the research of cell death signaling with regard to viral and bacterial infections and describes the network of the cell death initiating molecular mechanisms that selectively recognize pathogen associated molecular patterns. abstract apoptosis, necroptosis and pyroptosis represent three distinct types of regulated cell death forms, which play significant roles in response to viral and bacterial infections. whereas apoptosis is characterized by cell shrinkage, nuclear condensation, bleb formation and retained membrane integrity, necroptosis and pyroptosis exhibit osmotic imbalance driven cytoplasmic swelling and early membrane damage. these three cell death forms exert distinct immune stimulatory potential. the caspase driven apoptotic cell demise is considered in many circumstances as anti-inflammatory, whereas the two lytic cell death modalities can efficiently trigger immune response by releasing damage associated molecular patterns to the extracellular space. the relevance of these cell death modalities in infections can be best demonstrated by the presence of viral proteins that directly interfere with cell death pathways. conversely, some pathogens in the late s, it was recognized that cell death could be considered as a regulated process. for the genetically programmed, regulated cell demise the term apoptosis (from greek: "falling off") was coined, which exhibited distinctive morphological and molecular patterns (kerr et al., ) . that time necrosis was interpreted as a total counterpart of apoptosis, a nonphysiological and non-regulated, lytic cell death modality. in the recent decades, additional newly discovered regulated cell death modalities have been added to the list of cell death types (galluzzi et al., ) , and it became evident that necrotic-like, lytic cell death can also take place in a regulated manner. apoptosis is a non-lytic form of cell death characterized by nuclear condensation, shrinkage of the cell, formation of apoptotic bodies and preservation of the cytoplasm membrane integrity until the late stages of the process (kerr et al., ) . importantly, apoptosis actively labels itself by translocating phosphatidylserine (ps) to the outer leaflet of the membrane lipid bilayer. the ps exposure on the outer surface provides a so-called "eat me" signal for the neighboring monocytes and macrophages and ensuring the clearance of the apoptotic debris (fadok et al., ) . these two main attributes qualify apoptosis as a non-inflammatory cell death. in contrast, necroptosis and pyroptosis are lytic cell death modalities, accompanied by early membrane rupture and osmotic imbalance, which provoke inflammatory signals by releasing damage associated molecular patterns (damps) to the extracellular space (galluzzi et al., ) . it is important to note that the immunogenicity of the cell death is a more complex process than above outlined. the lytic property of a cell demise does not necessarily result in an immune stimulation, while the caspase- driven apoptosis may lead to immunogenic processes (yatim et al., ) . this also clearly indicates that the release of damps is a signaling-dependent, actively regulated process. the human body developed elaborate strategies to efficiently avoid and eliminate infectious pathogens. two corner stones of this response are the innate and the adaptive immune signaling, which also include controlled killing/self-killing mechanisms of the infected host cells ( jorgensen et al., ) . this review focuses on the infection related aspects of regulated cell death forms, which in many instances provide benefit for the hosts by destroying the intracellular niche of the pathogen. nevertheless, it is evident that some extracellular pathogens-and also intracellular ones-can benefit from the death of the host cells. in this chapter, based on the current advances in the research, we give a detailed description about the key cell death modalities, including apoptosis, necroptosis and pyroptosis emerging in response to pathogenic insults, and we discuss how bacterial and viral infections can modulate these signaling pathways. apoptosis is conducted by caspase activation. caspases are cysteine-driven and aspartate-directed peptidases, which present in inactive forms, and become cleaved and activated upon selective apoptotic stimulation (cohen, ) . the initiator caspases (for instance caspase- and - ) are the upstream components of the caspase-cascade. their structurally unique feature is the large pro-domain, that enables the recruitment to multimeric protein complexes (mace and riedl, ) . extrinsic and intrinsic pathways represent the two major routes of initiator caspase activation. in the extrinsic pathway, extracellular ligands from the tumor necrosis factor (tnf) superfamily interact with transmembrane death receptors (dr) of the tnf superfamily. consequently, by the help of an adaptor protein, caspase- and - are recruited into the dr containing cell death inducing signaling complex (disc), which leads to proximity induced cleavage and activation of these initiator caspases (wilson et al., ) . the intrinsic pathway can be triggered by intracellular perturbations, for instance by genotoxic stress, starvation and er-stress. upon stimulus, two pro-apoptotic members of the b cell lymphoma (bcl- ) protein family bcl- -associated x protein (bax) and bcl- -antagonistic killer (bak) go through conformational changes and translocate into the mitochondria (tait and green, ) . in turn, these proteins oligomerize and form large pores throughout the mitochondrial membrane (große et al., ) . this event initiates the mitochondrial outer membrane permeabilization (momp). caspase- activation takes place in the apoptosome protein complex. formation of this complex is dependent on the momp-driven release of mitochondrial proapoptotic proteins, such as cytochrome-c (riedl and salvesen, ) . the activation of initiator caspases in both pathways culminates in the cleavage of executioner caspase- , - and - . in turn, the effector caspases cleave selective substrates, including poly adp-ribose polymerase (parp), inhibitor of caspase activated dnase (icad) and lamins calling forth the biochemical and morphological features of apoptosis (cohen, ; tewari et al., ) . viruses are obligatory intracellular pathogens and their propagation is fully dependent on the intracellular milieu of the host cell. the host organisms developed various strategies to detect and destroy the invading viral pathogens. the immunological responses, involving the transcription of anti-viral genes (e.g., interferons and cytokines) given to a specific virus infection are well characterized. it is widely accepted view that apoptosis can be beneficial for the host by destroying the intracellular niche of the pathogen (evavold et al., ) . the most striking evidence for the anti-viral advantage of apoptosis is the observation that many viruses produce anti-apoptotic/ necroptotic proteins that block cell death signals (thome et al., ; zhou et al., ) . however, based on recent studies, it is obvious that the picture is more complex than originally expected, and whether the cell death is beneficial or detrimental for the host seems to be dependent on multiple factors, including timing, cell type and the effectiveness of the cell death modality to trigger or suppress immune response. the tnf ligand superfamily acts via transmembrane tnf receptors. specific members of these receptor family, also called death receptors (dr), possess a cytoplasmic amino acid motif, termed death effector domain. these receptors, including tnfr , first apoptosis signal (fas, also called cd ) and the tnf-related apoptosis-inducing ligand receptor- / (trailr / ) can unleash various signaling pathways, such as proliferation, survival, immune response and apoptosis (wilson et al., ) . in response to ligand binding, the trimerized drs can recruit one of the two death domain (dd) containing adaptor proteins: tnfr associated dd (tradd) or fas associated dd (fadd). recruitment of tradd leads to the assembly of a complex containing several proteins, including the dd containing receptor interacting protein kinase- (ripk ) (hsu et al., ) , the cellular flice/casp inhibitory protein (cflip), the tnfr associated factor- (traf ) and the e ubiquitin ligase cellular inhibitor of apoptosis (ciap) (varfolomeev and vucic, ) . this protein complex conveys a non-apoptotic signal via the activation of the nf-κb signaling. in contrast, the binding of fadd triggers the interaction of caspase- and - , leading to apoptotic signaling either via the mitochondrial amplification loop or via direct caspase- /- cleavage in immune cells. tnfr stimulation primarily results in the recruitment of tradd, leading to non-apoptotic signals via nf-κb activation and expression of pro-survival genes such as cflip. once this primary signal is compromised by the inhibition of the expression of anti-apoptotic cflip, the apoptotic signal can proceed. conversely, activation of fas and trailr predominately transduces fadd dependent apoptotic signaling (wilson et al., ) . signaling through drs represents one of the major forms of the immune cell directed host cell killing in infection. cytotoxic t cells (ctl) are key components of the adaptive immune response and natural killer (nk) cells play analogous role in the innate immune response. ctls express fas ligands (fasl) on their surface ( fig. ) (halle et al., ) . the trimerized ligands bind the drs expressed on the cell surface, leading to dr clustering and binding of the adaptor protein fadd, which consequently results in the recruitment of initiator caspase- /- . this cell killing mechanism can reduce the spreading of the virus by eliminating the intracellular niche of the pathogen (halle et al., ) . this occurs in human vascular endothelial cells in response to dengue virus (dv), where the cells exhibit increased fasl and fas expression and increased number of fas on the cell surface. this phenomenon is accompanied by elevated rate of apoptosis (liao et al., ) . similarly, influenza a virus (iav) infection leads to the expression of fasl, fadd and caspase- , and downregulates the expression of cflip in human lung epithelial cell lines. however, it is not entirely clear if the host or the virus profits from the apoptosis induction. for instance, the replication of the iav can be amplified by overexpression of proapoptotic genes and is blocked by the upregulation of anti-apoptotic genes . this study suggests that an initially protective apoptotic over-activation can turn against the host in some instances. tnfα, the ligand of tnfr can also be expressed paracrine or in autocrine manner as the part of the anti-viral immune response. hepatitis b virus x protein (hbx) sensitizes host cells to tnf driven apoptosis by interfering the anti-apoptotic function of cflip, thereby resulting in caspase- over-activation upon stimuli (kim and seong, ) . in addition, ctls and nk cells induce apoptosis in virus infected target cells through the action of secreted effector molecules. granzyme-b enters the cells along with another secreted protein called perforin ( fig. ) (lowin et al., ; shresta et al., ) . granzyme-b is a serine protease that has been shown to induce mitochondrial apoptosis by inducing bid cleavage and subsequent mitochondrial cytochrome-c release, leading to caspase- activation (pinkoski et al., ) . apart from the significant role of immune system driven apoptotic responses, a plethora of evidence demonstrates the relevance of apoptosis, fig. apoptosis signaling in response to infection. extracellular and intracellular viral rna (vrna) fragments (left) activate tlr and rig-i, respectively, resulting in irf- activation. cytoplasmic viral, bacterial or mitochondrial dna (mtdna) is detected by cgas, which in turn activates the sting pathway, leading to apoptotic signaling via a putative irf- driven mechanism. apoptosis can be triggered by effector cells of the immune system. ligation of the fasr (right) leads to the assembly of the adaptor fadd and caspase- containing disc, resulting in the activation of caspase- . activation of caspase- either results in the direct cleavage of caspase- / or leads to the bid cleavage (tbid) driven amplification loop of the intrinsic pathway. the secretion of the pro-apoptotic perforin and granzyme-b (middle) triggers the activation of the intrinsic apoptotic pathway. unleashed directly by the viral pathogens intracellularly. enterovirus (ev ), a virus in the family of picornaviridae, is one of the main causative agents of hand, foot and mouth disease in infected infants and young children. ev can directly induce the intrinsic apoptotic pathway by activating bax and the subsequent momp, leading to the caspase- driven apoptotic signal. importantly, depletion of caspase- has no influence on the cell death initiated by the virus, excluding the role of a dr driven external stimulation (han and cong, ) . analogously, coxsackievirus a (ca ) can trigger both the extrinsic and the intrinsic way of apoptosis in human neuronal and muscle cell lines accompanied by the typical apoptotic features: caspase cleavage, dna fragmentation, and ps translocation (li et al., ) . the west nile virus (wnv) is a mosquito born singlestranded rna virus, causing west nile encephalitis in humans. the non-structural proteins b and (ns b and ns ) conduct the proteolytic cleavage and the processing of the large viral polyprotein, that is formed directly from the rna of the virus. besides, ns initiates apoptosis in a caspase- dependent way, which then results in typical apoptotic features in multiple human cell lines (ramanathan et al., ) . the virus-host co-evolution is well demonstrated by the examples, where the pathogen has developed strategies to hijack the originally protective apoptotic signaling of the host. chikungunya virus (chikv) was reported to infect macrophages and induce apoptosis by triggering the intrinsic apoptotic pathway via the caspase- /- axis (nayak et al., ) . intriguingly, chikv virus exploits the formation of the apoptotic blebs to transport the virus particles into neighboring macrophages (krejbich-trotot et al., ) . the apoptosis can be beneficial for the pathogen, if the defensive immune cells are selectively deleted by the virus-driven cell death signals. the human immunodeficiency virus (hiv) protease cleaves procaspase- to a kda fragment, called casp p , which lacks the catalytic site of the caspase. interestingly, the catalytically inactive fragment can facilitate bax oligomerization and apoptosis in the host cd cells (sainski et al., ) . probably one of the most thrilling question, emerging from the studies on the host-pathogen interactions, is how the host organisms differentiate between friend and foe. detection of intracellular pathogens in mammalian cells is a multifaceted mechanism, which involves the recognition of various signals deriving from the pathogens, collectively termed pathogen associated molecular patterns (pamps). for instance, the viral rna and dna fragments represent a prominent group of pamps. pamps are detected by pamp recognition receptors (prr) (chow et al., ) , which mediate the expression and activation of immunomodulatory cytokines, chemokines, and interferons. furthermore, the signaling through prrs leads to cell death signals primarily via the transcriptional activation of apoptotic genes. depending on the spatial localization of the viral rna fragments, rna viruses can either stimulate toll-like receptors (tlrs) or activate the retinoic acid inducible gene-i (rig-i) pathway. tlr , which resides in the cytoplasmic and endosomal membranes (fig. ) , recognizes extracellular doublestranded viral rna (vrna) fragments (alexopoulou et al., ) . tlr utilizes the adaptor protein toll/interleukin- receptor (tir) domain containing adaptor inducing interferon (trif) (yamamoto et al., ) to recruit the inhibitor of kappa b kinase (ikk), subsequently leading to the activation of transcription factor nuclear factor kappa-light-chain-enhancer of activated b cells (nf-κb) and the tank binding kinase (tbk ) (fitzgerald et al., ) , which in turn induces the interferon regulatory factor- (irf ). intracellular dsrna, produced by replicating viruses is detected by cytoplasmic rna helicases, rig-i and melanoma differentiation-associated gene (mda ) (fig. ) . rig-i plays important role in recognizing rnas from rna viruses, including newcastle disease virus (ndv), sendai virus (sev) and vesicular stomatitis virus (vsv) (kato et al., ) . they recruit members of the traf family via the mitochondrial anti-viral-signaling protein (mavs). in turn, mavs, analogously to the tlr pathway activates the transcription factors irf and nf-κb. importantly, cells lacking the irf gene are not susceptible to apoptosis upon sev infection (peters et al., ) . conversely, mavs is shown to induce apoptosis via caspase- but not via irf upon semliki forest virus infection (el maadidi et al., ) . infection of a genetically engineered vaccinia virus strain, lacking the viral bcl- analog f l protein induces apoptosis via the activation of irf , which in turn mediates the expression of the pro-apoptotic protein noxa (from latin: "damage") in monocytes and lymphocytes (eitz ferrer et al., ) , whereas the wild type strain, possessing the f l protein cannot stimulate the same response, indicating that the vaccinia virus successfully adapted in an evolutionary race to overcome the apoptotic inducing capacity of the host. it was recognized that tlr stimulation was able to trigger apoptosis via the adaptor protein trif (kaiser and offermann, ) and that tlr mediated direct caspase- activation, primarily depending on the presence of cflip isoforms (feoktistova et al., ; weber et al., ) . along with this, immunoprecipitation experiments of tlr demonstrate the formation of a tlr associated disc-like complex, including trif, ripk , fadd, cflip and caspase- in response to iav infection. the administration of interleukin- (il- ) facilitates the caspase- activation by decreasing the presence of cflip in the complex and renders the cells to apoptosis (weiss et al., ) . the relevance of tlr- /caspase- dependent apoptosis is further confirmed in mouse intestinal epithelial cells, where the apoptosis is independent from the tnfr signaling and from the effect of external cytokines (mcallister et al., ) . the cyclic gmp-amp (cgamp) synthase (cgas) has been shown to interact with viral dna (vdna) molecules. upon interaction, cgas produces its second messenger cgamp and this in turn activates the stimulator of interferon gene (sting) protein ( fig. ) , which is located in the endoplasmic reticulum (er). consequently, sting can activate irf (ishikawa and barber, ; sun et al., ) and result in mitochondrial apoptosis (diner et al., ) . intriguingly, mitochondrial dna (mtdna) that is released upon the apoptotic mitochondrial membrane permeabilization can also activate cgas. this can trigger irf activation and consequently result in anti-viral responses even in circumstances where viruses efficiently block apoptosis downstream of momp, for example at the level of caspase activation. it has been demonstrated that apoptotic caspases not only mediate an anti-inflammatory cell death pathway, but also directly block the mtdna triggered sting-mediated ifn production in mice (white et al., ) . a recent study provides evidence that apoptotic caspases cleave the cytoplasmic anti-viral sensors cgas, mavs and irf to limit cytokine production. in this study caspase- deficient mice exerts increased resistance to virus infection, suggesting that caspase activation may contribute to virus propagation by cutting off the pro-inflammatory signals deriving from the infected cells (ning et al., ) . the viruses developed elaborate countermeasures to avoid or delay a premature apoptotic demise, which consequently would eliminate the essential environment of the virus replication. these strategies include the expression of apoptosis inhibitor proteins. the majority of the antiapoptotic factors are encoded by large dna viruses, which generally replicate with slower kinetics than rna viruses. a broad group of inhibitors attack the extrinsic pathway of apoptosis, which is triggered by an anti-viral immune response. viral flice-like inhibitory protein (vflip) is encoded by the molluscum contagiosum virus and several gamma herpes viruses, including kaposi's sarcoma-associated herpes virus (kshv). the vflip exerts its inhibitory activity by binding the adaptor protein fadd and forming inactive caspase- /vflip heterodimers (thome et al., ) . human cytomegalovirus (cmv) is a common infection in humans, which does not lead to disease in healthy individuals, whereas it causes severe infections or death in patients with immunodeficiencies. the mouse cmv (mcmcv) is the experimental model of human cmv infection. the m protein of the virus blocks caspase- driven apoptosis and thus contributing to the pathogenicity of the microbe (ebermann et al., ; skaletskaya et al., ) . the large subunit of the herpes simplex virus (hsv) ribonucleotide reductase icp demonstrates caspase- inhibitory potential by directly interacting with the initiator caspase (dufour et al., ; langelier et al., ) . the cytokine response modifier a (crma), a product of the cowpox virus, prevents apoptosis by blocking the activity of caspase- (zhou et al., ) . other group of viral anti-apoptotic proteins modulates the expression of cellular anti-apoptotic proteins. the human t-cell leukemia virus type- (htlv- )-associated adult t-cell leukemia/lymphoma (atl) is a malignancy of infected mature cd t cells in humans (matsuoka and jeang, ) . the transactivator protein tax of htlv- upregulates the expression of cflip by binding to the inhibitor of κb kinase (ikk) and inducing the nf-κb pathway in the host cells, and thereby the virus efficiently amplifies the apoptosis resistance in infected cells (okamoto et al., ) . a further group of virally encoded anti-apoptotic factors target the intrinsic apoptosis signaling. the vaccinia virus encoded f l protein inhibits the mitochondrial apoptotic signaling in two distinct ways. f l functions both as an inhibitor of pro-apoptotic bcl- family proteins and as a direct caspase- inhibitor (zhai et al., ) . human papillomavirus (hpv), a double-stranded dna virus has been associated with cervical cancer developments in humans (d€ urst et al., ) . the early gene-coding region protein (e ) of hpv is reported to interfere with apoptosis signaling at the level of p and by interacting with the pro-apoptotic bak protein. in contrast to viruses, bacteria are not essentially dependent on the intracellular environment. moreover, intact cellular layers exhibit physical barriers against bacterial penetration. therefore, extracellular bacteria developed strategies to efficiently destroy these barrier functions. a novel study presents the bacterial qourum-sensing autoinducer of the pseudomonas aeruginosa n-( -oxo-dodecanoyl) homoserine lactone ( oc), a small chemical released to control microbial communication, as an inducer of tnfr . it is demonstrated that oc directly disrupts the lipid domain structures, containing sphingolipids and cholesterol, and induces the translocation of tnfr into the disordered lipid phase of the membrane, which in turn triggers the trimerization of the tnfr and leads to subsequent apoptosis in human and mice monocytes (song et al., ) . this type of tnfr activation ensures a ligand independent induction of the extrinsic pathway. another type of cell killing involves neighboring immune cells. the gram-positive bacteria listeria monocytogenes conducts the sorting of the bacterial dna into extracellular vesicles, which in turn are delivered to bystander t cells, where they trigger the dna sensor cgas-sting pathway mediated apoptosis (nandakumar et al., ) . yersinia pestis, the causative agent of pest, induces significant cytotoxicity. the type three secretion system (t ss) is a host sensor and effector system, which can be found in several gram-negative bacteria. this system promotes the virulence of the bacteria by injecting effector proteins into the cytosol (ashida et al., ) . structurally, tt s consists of a base embedded in the bacterial membrane, the needle, which directly interacts with the host membrane, and the rod, which connects the base to the needle. yersinia outer protein j (yopj), an effector protein of the t ss, rapidly kills macrophages in a caspase- dependent way, which occurs independent of the canonical inducers of extrinsic apoptosis. in this model, tlr stimulation by the pathogen activates the trif/ripk /caspase- axis, which leads to caspase- cleavage and cell death. consistently, mice lacking caspase- and ripk are more susceptible to the infection, pinpointing the importance of caspase- and ripk driven signaling in tackling this disease (gr€ obner et al., ; weng et al., ) . mycobacterium (m.) tuberculosis triggers caspase- dependent apoptosis in human monocytes, which is induced by extracellular bacterial rna fragments detected by tlr (obregón-henao et al., ) . the polymorphic gc-rich repetitive sequence containing pe_pgrs , a surface exposed protein and the -kda glycolipoprotein (p ) of the m. tuberculosis can both engage tlr and initiate apoptosis signal-regulating kinase- (ask ) driven tnf and tnfr expression in mouse macrophages (basu et al., ; lópez et al., ) . finally, group b streptococcus, a pathogen causing neonatal meningitis, induces apoptosis via the activation of tlr , which engages its adaptor protein myeloma differentiation primary response- (myd ), and consequently results in caspase- activation in microglial cells (lehnardt et al., ) . caspase- is unique among caspases. it exhibits initiator characteristics structurally, since it has an n-terminal caspase activation and recruitment domain (card). in addition, caspase- represents executioner features functionally, because it can selectively cleave substrates similar to that of caspase- or - (olsson et al., ) . caspase- , along with caspase- plays a significant role in brucella abortus and brucella suis induced cell death (bronner et al., ; chen et al., ) . one possible mechanism of activation is derived from the studies on bacterial pore forming toxins (pft), alpha toxin and aerolysin secreted by the human pathogens staphylococcus aureus and pseudomonas aeruginosa, respectively. based on this study, caspase- activation can be triggered by the pft driven potassium efflux in epithelial cells, which facilitates the assembly of a caspase- containing high molecular weight protein complex (imre et al., ) . importantly, the low potassium concentration triggered activation of caspases is not without precedent in the literature. though, the role of potassium efflux has been demonstrated in terms of the pro-inflammatory caspase- activation in immune cells (see pyroptosis), it is not clear whether caspase- activation takes place in an inflammasome-like complex in response to pathogen insults in epithelial cells. several pathogenic bacteria developed strategies to evade the apoptotic demolition. francisella tularensis, a highly virulent gram-negative pathogen inhibits apoptosis by retaining mitochondrial integrity via the inhibition of bid processing and bax translocation, and successfully extends the lifespan of the infected neutrophils (mccracken et al., ) . helicobacter pylori leads to a persistent, yet asymptomatic infection, which, however, can increase the incidence of the chronic gastritis and the gastric adenocarcinoma. this pathogen inhibits the anti-bacterial and cell death responses simultaneously by enhancing the level of ubiquitin-editing enzyme a , which in turn bi-functionally blocks both the nf-κb activity and the caspase- activity (lim et al., ) . enteropathogenic escherichia coli (epec) transports effector proteins directly into the cytosol of infected cells by utilizing the t ss. one of these effector proteins is the non-locus of enterocyte effacement encoded effector b (nleb ). nleb transfers an n-acetyl-glucosamine residue to arg in the death domain of fadd and inhibits the fadd dependent caspase- activation (pearson et al., ) . chlamydia trachomatis is an obligate intracellular pathogen with medical relevance, which interferes with apoptosis by blocking the internalization of the cell death receptor tnfr . this mechanism renders the specific blockage of the apoptotic signaling but preserves the function of the nf-κb signaling of the tnfr (waguia kontchou et al., ). in the previous chapter, we have seen several examples for apoptosis being induced in response to pamps. nevertheless, apoptosis in many circumstances does not support the immune stimulation, which on the long run assists the pathogen to evade the counterattacks of the immune system. necroptosis represents an immunogenic cell death modality. this nonapoptotic programmed cell death form, that morphologically resembles necrosis, has gained attention in the past decades (degterev et al., ; vercammen et al., ) , and its relevance in infection has been recognized only recently (pan et al., ) . in contrast to apoptosis, necroptosis is characterized by swelling of the cytoplasm, osmotic perturbations and the early rupture of the cytoplasm membrane, resulting in the release of damps into the extracellular space. receptor interacting protein kinase- and - (ripk and ripk ) play indispensable role in conducting necroptosis signaling. ripks interact via their ripk homotypic interaction motif (rhim) domains. the most studied pathway, by which ripk dependent necroptosis can be triggered is the tnfr induced pathway. tnfr activation primarily results in nf-κb dependent pro-proliferative response. upon the ligation of tnfr the formation of the so-called complex-i takes place, consisting of the adaptor proteins tradd, traf , ciap and ubiquitinated ripk (fig. ) (wilson et al., ) . once, the pro-proliferative signal is blocked, the fadd containing apoptotic signal takes over, leading to caspase- activation. surprisingly, the inhibition of the tnfr induced apoptosis by caspase inhibitors results in a necrotic-like cell demise. this programmed necrotic cell death, also termed necroptosis can be inhibited by ripk inhibitor necrostatin- (xie et al., ) . it is also acknowledged that the proteolytic activity of caspase- acts against the necroptotic activity of three key components of the pathway: ripk , ripk and cylindromatosis (cyld) (o'donnell et al., ; tenev et al., ) , hence inhibition of caspase- activity is a prerequisite for the induction of the tnfr driven necroptosis (fig. ) . once caspase- activity is blocked, ripk is enabled to recruit ripk . consequently, ripk activation can lead to the phosphorylation of the necroptotic effector mixed lineage kinase-like protein (mlkl) (fig. ) (sun et al., ) . upon the phosphorylation driven conformational changes, mlkl oligomerizes and forms pores throughout the cytoplasm membrane, resulting in membrane integrity loss. in the past decade, it became clear that ripk is a multifaceted scaffold protein. if ubiquitinated, ripk inhibits the caspase- driven apoptosis and the ripk /mlkl dependent necroptosis. however, as a deubiquitinated protein, ripk can promote the stimulation of both cell death signaling pathways (varfolomeev and vucic, ) . apart from the decisive role of ubiquitination, phosphorylation at ser by protein kinase a, c, and phosphorylation at several sites by ikks have been found to be important to attenuate the cytotoxic potential of this signaling protein (dondelinger et al., ; mcquade et al., ) . the current observations indicate that the majority of the human pathogen viruses developed strategies to avoid the necroptotic signaling of the host. conversely, human immunodeficiency virus type (hiv- ) induces tnfα dependent necroptotic cell death in cd + t lymphocytes, which can be inhibited by necrostatin- . unlike apoptosis, this cell death occurs in infected cells and spares the bystander cells (pan et al., ) . nevertheless, the hiv triggered necroptosis is considered detrimental for the host by targeting the defensive mechanisms of the immune system. another example, where necroptosis is triggered by viruses, takes places in response to human coronavirus (hcov) infection. hcov is a neuroinvasive respiratory pathogen, which induces a cell death pathway dependent on ripk and mlkl in human neuroblastoma cells (meessen-pinard et al., ) . the protein z-dna/rna binding protein- (zbp ; also known as dai) recognizes z-rna or z-dna structures, which are common as a consequence of a rapid viral rna and dna synthesis, but distinct from the b structure of the eukaryotic rna or dna. zbp can activate ripk (upton et al., ) , since it is capable to interact the ripks via its rhim (fig. ) . intriguingly, mefs, liver cells, lymphoid cells and skin cells isolated from mice exhibiting catalytically inactive ripk die early and exert ripk auto-phosphorylation, a hallmark of the necroptotic activation, whereas ripk and mlkl deficient mice are protected from the cell death ex vivo and in vivo. furthermore, the zbp -ripk interaction can be blocked in macrophages from wild type mice, whereas the interaction takes place in mouse macrophages that express a rhim deficient ripk . these results indicate that ripk acts as a negative regulator of zpb -ripk interaction during development in order to prevent premature necroptosis (newton et al., ) . to date, iav infection is the only example, where zbp dependent necroptosis occurs in response to viral infections employing non-engineered (wild type) virus strains. iav induces ripk and mlkl dependent, but tnfα and tlr independent necroptosis . this pathway is mediated through zbp by sensing viral genomic rna segments . a recent study (wang et al., b) performed in mice further confirms the role of zbp- as the key necroptosis modulator protein in response to iav infection in vivo. aside from the necroptosis stimulatory effects of the viral rna fragments, the ns protein of the iav is suggested as a direct interactor of mlkl (gaba et al., ) . finally, the transcriptional regulation of zbp seems to be dependent on the interferon regulatory factor (irf ), which is activated in mouse bone marrow-derived macrophages in response to iav infection (kuriakose et al., ) . consequently, depletion of irf can reduce the level of zbp and diminish the iav induced necroptotic features, including mlkl phosphorylation and membrane integrity loss. murine cytomegalovirus (mcmv) lacking the viral m -encoded inhibitor of ripk (vira) triggers premature cell death in mice. importantly, ripk knockout can rescue the premature death phenotype, indicating that the mcmv infection induces a ripk dependent necroptotic cell death (upton et al., ) . similar to iav infection, the ripk driven necroptosis is independent of the effect of ripk and the tnf stimulation, but zbp is activated via its putative rna binding domain zα (sridharan et al., ) . analogously, the vaccinia virus (vacv) encoded innate immune evasion protein- (e ) can directly interact with the zα domain of the zbp . along with this, e deleted vacv strain triggers immediate necroptosis in interferon treated mouse l and human embryonal kidney (hek t) cells. the cytotoxic effect of the engineered vacv strain is attributed to necroptotic signaling, because the pathogenicity of the virus can be restored in vivo by the depletion of either ripk or zbp in mice (koehler et al., ) . the large subunit of the viral ribonucleotide reductase icp (fig. ) is capable to engage both ripk and ripk through its rhim-like domain upon herpes simplex virus type (hsv- ) infection and to result in necroptosis, which also contributes to the limited viral propagation in mice. intriguingly, icp inhibits the necroptosis signaling in its natural host, in human cells. this observation points out the species-specific differences in cell death signaling pathways and indicates that viruses in their natural hosts evolved to overcome host defense mechanisms . hsv- induces necroptosis not only via its icp protein, but also by triggering the zbp pathway in mice. conversely, in order to induce the necroptotic pathway in human cells, the expression of zbp has to be first enhanced . moreover, icp , the virally encoded large subunit of the ribonucleotide reductase of hsv- , can efficiently prevent necroptosis that is induced by tnfα . one possible explanation for the difference between mice and human responses might be provided by a recent study on hsv- demonstrating that icp blocks tnfr induced mlkl necrosome assembly in membrane vesicles (ali et al., ) . this compartmentalization process of the necrosome represents a critical step for the progression of necroptosis in human cells. consistent with this, rhim domain deficient ripk can efficiently phosphorylate mlkl but does not lead to necroptosis in human cells. thus, one can speculate that icp can trigger necrosome formation in mice, because the compartmentalization step may not be required for the effector mechanisms of necroptosis. additionally, inositol phosphate (ip) kinases largely contribute to the oligomerization of mlkl. ip kinase mutants are unable to achieve the mlkl oligomerization and membrane localization despite a functioning ripk in response to a genetically engineered icp- mutant hsv- infection . this study suggests that the ripk dependent phosphorylation at the c-terminal ser and the subsequent events, such as oligomerization and membrane localization are separately regulated in human cells. accordingly, in order to accomplish necroptosis, mlkl activation requires a highly phosphorylated ip product to bind to the n-terminal domain. it is currently not clear what the exact role of the highly phosphorylated product can be, nevertheless the stabilization of the oligomerized mlkl complex can serve as one plausible explanation. the latent membrane protein- (lmp ) of the γ-herpes virus epstein-barr virus (ebv) is considered as the major transforming factor of the pathogen. initially, this transmembrane protein has been reported to exert antiapoptotic effects in b cells (henderson et al., ) . later, lmp has been associated with necroptosis inhibition in ebv-infected human nasopharyngeal epithelial cells and nasopharyngeal carcinoma cells. lmp can interact both ripks and can successfully mediate the polyubiquitination of them, which in turn leads to the inhibition of the kinase function of the ripk and activates the pro-survival pathway . as previously discussed, the vira protein of mcmv can efficiently block ripks and hamper the necroptotic signaling. though, distinct from the necroptosis inhibitory mechanism of the vira protein, the human cmv has also developed a strategy to suppress necroptosis upstream of mlkl phosphorylation . furthermore, rhinoviruses, the causative agents of common cold can actively suppress both apoptosis and necroptosis. the viral c protease directly interacts and cleaves ripk and blocks necroptosis stimulated by poly(i:c) in human airway epithelial cells (l€ otzerich et al., ) . it is therefore somewhat surprising that the viral protease still triggers a lytic cell death pathway, which is distinct from necroptosis and apoptosis. whereas the relevance of necroptosis initiation in viral infection has been proven only in few cases, for instance in response to iav, the connection between necroptosis and bacterial infection seems to be more obvious. the significance of type i interferons in initiating necroptosis is demonstrated during salmonella typhimurium infection. mice deficient in the receptor for type i interferons (ifnar ) exhibit improved survival, and macrophages derived from these mice exert resistance to the bacteria triggered cell death. consistently, ripk deficient macrophages can similarly evade the cell death upon infection. this study gives an example for a necroptosis induction, which is detrimental for the host and beneficial for the bacteria (robinson et al., ) . similarly, staphylococcus (s.) aureus infection induced cytotoxicity can be prevented by ripk depletion or inhibition of ripk and mlkl in primary human macrophages. surprisingly, necroptosis induction by pore forming toxins (pft) of s. aureus occurs without employing any external caspase inhibition, which suggests that the caspase- signal remains inactive throughout the whole infection process (kitur et al., ) . in concordance with this, several pft producing bacterial strains, including streptococcus pneumoniae and staphylococcus marcescens induce necroptotic cell death in mouse macrophages, which can be arrested by necroptosis inhibitors (e.g., necrostatins) but not by the depletion/inhibition of caspases (gonzález-juarbe et al., ) . surprisingly, a pft secreted by s. marcescens can trigger mlkl dependent necroptosis in mouse lung epithelial cells in the absence of tnfr / or tlr signaling (gonzález-juarbe et al., ) . potassium efflux, as a key event, has been previously studied in necroptosis initiation. however, how this osmotic imbalance orchestrates the necroptotic process still needs to be elucidated. an in vivo study describes the pft driven activation of caspase- , - and - , which facilitates the release of alarmins to the extracellular space . mycobacterium tuberculosis infection induces nad+ depletion triggered necroptosis via the ripk -mlkl axis, which is initiated by the tuberculosis necrotizing toxin (tnt) in human macrophages (pajuelo et al., ) . in contrast to the observation in human macrophages, the pathogenesis of mycobacterium tuberculosis infection is independent from the role of ripk and necroptosis in mice (stutz et al., ) . yersinia pestis outer protein j (yopj) is required for the lymphatic spread of the bacteria, which leads to the formation of swollen and necrotic structures called buboes. yopj suppresses the downregulators of the ripk driven necroptosis, resulting in necroptotic spread of the pathogen. in concordance with this, necrostatin treatment efficiently protects the mice from lethal infections of the bacteria (arifuzzaman et al., ) . it is demonstrated that yopj interferes with the pro-survival activity of the flips in the disc, thereby promoting the necroptotic feature of the ripks. in contrast to the previous examples, some bacteria have developed countermeasures to inhibit the necroptotic signals of the host. the enterophatogenic e. coli (epec) directly interferes with the elements of the necroptotic signal transduction. espl, an effector component of the epec t ss is identified as a protease, that can target and cleave host proteins containing a rhim domain, including ripk / , trif and zbp (pearson et al., ) . further example for necroptosis inhibition is demonstrated in a study employing the salmonella induced colitis model. the salmonella outer protein b (sopb) attenuates the necroptotic cell death by decreasing the level of mlkl phosphorylation in the cells (hu et al., ) . tlr- and tlr- are pamp recognition receptors and engage dsrna and bacterial lipopolysaccharide (lps), respectively, which in turn initiate the recruitment of toll/il- r (tir) domain containing adapters to activate gene expression via transcription factors, such as nf-κb and irf / . over the last years, the role of tlr receptors in mediating cell death pathways has been intensively studied. induction of tlrs can induce caspase- activation (kaiser and offermann, ) and the inhibition of caspase- activity leads to a ripk dependent necrotic-like cell death in mouse macrophages . the adaptor protein trif is necessary to transduce signals from tlr stimulation, whereas activation of tlr can engage both trif and myd . these pathways proceed independent of ripk in mouse fibroblasts, but can efficiently trigger mlkl activation downstream of ripk (kaiser et al., ) (fig. ) . in response to tnfr ligation, the ubiquitination status of ripk is defined as the key factor in the decision making between survival and death. upon tnf stimulation, the activated caspase- cleaves cyld, the deubiquitinase enzyme of ripk (o'donnell et al., ) . the lps induced tlr- ligation recruits fadd and caspase- in a complex and initiates the cleavage of cyld (legarda et al., ) . the non-histone chromatin binding protein high mobility group box- (hmgb ) is released into the extracellular space (lu et al., ) during necroptosis and functions as a pro-inflammatory damp (scaffidi et al., ) . interestingly, hmgb can form a complex with lps upon infection, which leads to an amplified activation of trif dependent immune responses . pyroptosis is an inflammatory and regulated cell death modality (fink and cookson, ) that partially shares features with apoptosis, because it requires caspase activation. on the other hand, the pyroptotic membrane rupture occurs rapidly in the affected cells, which is a common feature shared with necroptosis. pyroptosis has been observed in monocytes, macrophages, dendritic cells, lymphocytes, epithelial cells, and endothelial cells (man et al., ) . contrary to apoptosis, induction of pyroptosis activates inflammatory caspases, which, similarly to apoptotic initiator caspases, possess a large prodomain structure. the inflammatory caspase- , caspase- / in humans and caspase- in mice are being activated in large molecular weight complexes, termed inflammasomes (van opdenbosch and lamkanfi, ). in general, inflammasomes consist of sensors that can recognize pathogenic insults, also termed pamps. depending on the nature of a pamp signal, different sensors are triggered. the nucleotide binding oligomerization domain (nod)like receptors represent the best studied inflammasome components, that recognize pamps (fig. ) , but various other pamp receptors have also been identified in the last years. these pathogen sensors can recruit caspases either directly or via an adaptor molecule. once, the inflammatory caspase- is activated, it can process interleukin (il) β and into their mature forms. furthermore, both caspase- and caspase- can process gasdermin-d (gsdmd), a member of the less studied gasdermin protein family (kayagaki et al., ) . the n-terminal leaflet of gsdmd inserts into the lipid bilayer and forms oligomer pores with an inner diameter of approximately nm (sborgi et al., ) , leading to il- β/- release (he et al., ) and to osmotic imbalance driven swelling of the cells (fig. ) . several studies conclude the role of gsdmd as a key factor for il secretion. in this setup, the pyroptotic membrane integrity loss, the swelling and the cell death are interpreted as the collateral damage of the pore formation. in fact, gsdmd driven il release can proceed without any cell destruction (evavold et al., ) . in contrast, a recent study on mice infected with burkholderia thailandensis leads to the conclusion that the action of gsdmd on bacterial killing occurs in a pyroptosis dependent manner fig. inflammasome activation and pyroptosis signaling. components of the bacterial t ss and bacterial flagellins can activate the naips. in turn, naips stimulate the nlrc- inflammasome (top left), leading to caspase- activation. anthrax lethal toxin (top middle) directly interacts with nlrp b and triggers the recruitment of asc. intracellular perturbations culminating in potassium efflux stimulate the assembly of the nlrp inflammasome (top right), leading to the recruitment of the adaptor protein asc. nlrp requires priming signals via the ligation of tnf or tlrs. the ifn-stimulated gbps facilitate the release of the pathogenic dna fragments into the cytosol, which in turn can stimulate the aim inflammasome. caspase- and caspase- / (left) can directly detect lps, released by the gbps into the cytosol. active caspase- can process pro-il β/il and cleave gsdmd, subsequently leading to gsdmd pore formation and release of the ils into the extracellular space. and is entirely independent from an indirect effect of il production (wang et al., a) . interestingly, gsdmd is also capable to insert into the bacterial membrane and to kill extracellular bacteria (liu et al., ) . thus, it seems that gsdmd is represented at three different layers of defense depending on the localization of the pathogen. besides, gasdermin e, another member of the gasdermin family, is also reported to participate in pyroptosis and in post-apoptotic secondary necrosis (rogers et al., ) . whereas, virus triggered inflammasome activation is a common event, inflammasome induced pyroptosis cannot be observed in all circumstances. nevertheless, inflammasome dependent pyroptosis has been reported in response to infection by several viral pathogens, including hiv, dv (tan and chu, ) , iav , coxsacivirus (cv) , parvovirus (deng et al., ) , hepatitis c virus (kofahi et al., ) , and enterovirus (ev ) zhu et al., ). an in vivo study on attenuated rabies strain provides evidence for the beneficial role of pyroptosis against viral infections. the infection with the attenuated rabies virus strain, which normally leads to the onset of mild symptoms, can be aggravated by caspase- /- depletion, but not by caspase- / or by il- β/il- depletion (kip et al., ) . these observations suggest that the pyroptosis can be necessary to limit the rabies infection. further argument for the anti-viral nature of pyroptosis is provided by the observation that some viruses developed strategies in order to inhibit the effector mechanisms of this lytic cell death modality. ev triggers gsdmd by the viral protease c, and the cleaved fragments fail to unleash pyroptosis in thp- macrophages, hek t embryonal kidney and rhabdomyosarcoma cell lines . the nature of the viral pamps determines which inflammasome mediated pyroptotic pathway is activated in response to a specific stimulus. in accordance with this, viral nucleotide patterns can lead to the activation of absent in melanoma- (aim ) and interferon-γ-inducible protein (ifi ) inflammasome activation, whereas various virally triggered stress signals can result in the assembly of the nod-like receptor (nlr) family pyrin domain containing (nlrp ) inflammasome (fig. ) . the role of nlrp inflammasome nlrp protein is the major component of the nlrp inflammasome, which is the most studied inflammasome to date. nlrp consist of a pyrin domain, a nucleotide binding site, and a leucine-rich repeat motif. upon activation by pamps, nlrp interacts with the apoptosis-associated speck-like protein (asc) via its pyrin domain (fig. ) . this in turn induces asc polymerization and the assembly of the asc specks. the adaptor protein asc possesses a caspase recruitment card domain and is therefore capable to recruit pro-inflammatory caspases, such as caspase- (martinon, ) . consequently, caspase- is dimerized and auto-processed based on the principle of the induced proximity. several pamp signals have been demonstrated to culminate in nlrp activation. for instance, lytic cell death induced potassium efflux is proposed as the major mechanism of the nlrp inflammasome activation following encephalomyocarditis virus (emcv) and vsv infection (da costa et al., ) . in these infections, the lytic cell death precedes the inflammasome formation, whereas the infection with iav results in rapid nlrp dependent pyroptosis in dendritic cells (fernandez et al., ; kuriakose et al., ) , indicating that other mechanism can also trigger nlrp assembly. it is of note, that iav infection results in pyroptosis with a delayed kinetic in respiratory epithelial cells . in the activation of nlrp inflammasome in dv infection, the cytoplasmic membrane surface c-type lectin a (clec a) is found to be critical, which interacts directly the dv virion and serves as a pathogen sensing receptor in human inflammatory macrophages . a highly pathogenic avian influenza a (h n ) results in pneumonia and acute respiratory distress syndrome in humans. the depletion of galectin-b, a b-galactosidase binding protein leads to a striking reduction of inflammation in lungs of the infected mice compared to that of the wild type mice. consequently, co-immunoprecipitation experiments demonstrate the interaction of galectin-b, nlrp and asc in bone marrow-derived macrophages upon infection . the gp envelop protein of the hiv is the major contributing factor of the hiv-associated neurocognitive disorder accompanied by neuroinflammation. the gp triggered neuropathy is nlrp dependent and the administration of a selective nlrp inhibitor can alleviate the neuronal death and the impaired cognitive functions in gp transgenic mice. in a recent study, the role of the purinergic x (p x ), a cationic ion channel involved in neuropathic pain, is shown to mediate the gp dependent inflammasome activation and the caspase- cleavage in glial cells . most of these studies indicate the role of osmotic imbalances, such as potassium efflux as one of the major causative of nlrp inflammasome stimulation, however, how the low potassium concentration renders changes in the nlrp structure and whether this is a direct or an indirect effect still needs to be clarified. the role of ifi in the hiv induced t lymphocyte pyroptosis hiv triggers bystander pyroptosis in non-infected t lymphocytes . the bystander cd t cells, that are not permissive to hiv infection, play a crucial role in the acquired immunodeficiency syndrome (aids). these cells go through an incomplete virus replication cycle, resulting in aberrant transcripts of the viral genome. these transcripts have been demonstrated to stimulate the dna sensor interferon-γ-inducible protein (ifi ), leading to caspase- activation mediated pyroptosis (fig. ) . the immune stimulatory dying of the t lymphocytes initiates a pathogenic positive feedback loop by recruiting more immune cells to the site of infections, which leads to even more accelerated propagation of the virus. cell free virions are unable to trigger pyroptosis. thus, a cell to cell transmission of the virions is required for the induction of this lytic inflammatory cell death (galloway et al., ) . a clinical antiretroviral therapy study in humans demonstrates that the non-responding individuals have increased level of nlrp and caspase- . in addition, the non-responding group exhibits significantly lower levels of cd t lymphocytes in comparison to that of the responder patients. though, this patient data supports the previous in vitro and mouse experiments, which suggest the relevance of cd lymphocyte loss in the pathogenesis of hiv infection (bandera et al., ) , it does not provide direct evidence for an ifi driven pyroptotic cell death. aim has been shown to bind dna and engage the caspase- -activating adaptor protein asc to form an inflammasome. aim conducts the maturation of il- β/il- and induces pyroptosis in response to exposure of synthetic double-stranded dna (fig. ) . in natural situations, aim dependent pyroptosis is observed in response to mcmv, vaccinia virus (rathinam et al., ) and ev infections (yogarajah et al., ) . zika virus (zikv), an rna virus, has gained increased attention over the last years due to its association with severe birth defects, including microcephaly. infection with this flavivirus triggers higher expression levels of nlrp , , aim and caspase- in brain tissues collected from fatal human cases (de sousa et al., ) . it is, however, not clear whether the accompanied brain tissue damage is the direct consequence of the gsdmd driven pyroptosis or the detected higher inflammasome expression levels represent a completely unrelated phenomenon. virus induced pyroptosis can be triggered cell type specifically for instance in rotavirus infection, a leading cause of gastroenteritis and diarrhea in small children. nlrp is a member of the nlr family that is specifically expressed in intestinal epithelial cells in the site of the rotavirus infection. nlrp can recognize dsrna fragments and form inflammasome with asc to initiate caspase- cleavage and gsdmd driven pyroptosis (zhu et al., ) . moreover, caspase- , the human homolog of the mouse caspase- has been reported to function as a pamp receptor. in response to dv infection, caspase- is activated and can process caspase- in order to accomplish pyroptosis (cheung et al., ) . the first report describing infection driven pyroptosis was performed in salmonella enterica serovar typhimurium (also called salmonella typhimurium) infected mouse macrophages (fink and cookson, ) , although caspase- dependent cell death in response to bacterial infection was mentioned in the literature earlier (chen et al., ; monack et al., ) . the detection of pamps in bacterial infections includes the recognition of bacterial lps, flagellins, components of the bacterial t ss and the detection of cytoplasmic bacterial dna. the components of the bacterial t ss trigger inflammasome formation and pyroptotic cell death in shigella infected macrophages through the activation of the nlr family card domain containing protein (nlrc ) (fig. ) . nlrc possess a card domain, therefore this nlr family member can directly engage and process caspase- (suzuki et al., ) . besides, flagellins of various bacterial strains, including listeria (l.) monocytogenes (cervantes et al., ; warren et al., ) , legionella pneumophila and burkholderia thailandensis (miao et al., ) can lead to the assembly of the nlrc inflammasome (case et al., ) . the activation of nlrc inflammasome has been shown to lead to bacterial clearance independent from the effect of il- and il- production, since this process functioned efficiently in il- / il- knockout mice (miao et al., ) , thus implicating the primary role of pyroptosis in the bacterial clearance. this process involves the pyroptotic release of the intracellular pathogens from the macrophages and leads to the uptake of the bacteria by neutrophil granulocytes. in contrast, pyroptosis can be also detrimental for the host organisms. the extracellular opportunistic pathogen pseudomonas aeruginosa infection in mice leads to nlrc inflammasome activation, pyroptosis and increased mortality, whereas depletion of nlrc results in improved bacterial clearance and increased survival (cohen and prince, ) . the nlrc inflammasome utilizes various sensors in order to recognize pamps. the human nlr family apoptosis inhibitory protein (naip ) and the mouse counterpart naip prevent apoptosis by inhibiting caspases. the flagellin of the legionella pneumophila can result in caspase- activation and cell death in a naip dependent manner in macrophages ( fig. ) (katagiri et al., ) . importantly, not all bacterial strains express flagellins constantly. nevertheless, the inner rod of the t ss needle complex of the non-flagellated shigella bacteria results in the engagement of mouse naip and leads to the nlrc inflammasome mediated pyroptosis in primary mouse macrophages (suzuki et al., ) . the hnaip and the mouse counterpart naip (fig. ) can recognize the needle component of the salmonella t ss in dendritic cells, and to less extent in macrophages . nlrc activation is mediated by naip independent phosphorylation events, which suggest a biphasic model for the stimulation of this intracellular pathogen sensor (matusiak et al., ; qu et al., ) . the existence of a second stimulatory signal is further supported by the observations on the t ss effector protein yopj of the yersinia pestis, which translocates into the cytoplasm upon infection and induces a rapid apoptosis in macrophages. in contrast, a preceding activation by lps or by the ligation of the prrs tlr , and leads to a yopj independent pyroptosis (bergsbaken and cookson, ) . furthermore, interferon regulatory factor (irf ) is required for the activation of the nlrc inflammasome in bone marrow-derived macrophages infected with salmonella typhimurium, burkholderia thailandensis and pseudomonas aeruginosa. in these infections, irf conducts the transcription of the naips (karki et al., ) . these results implicate that the activation of nlrc relies on a second pamp signal, which then initiate transcriptional priming and can mediate further stimulatory phosphorylation steps. . . the nlrp inflammasome nlrp indirectly senses pathogen associated signals such as extracellular atp and uric acid crystals (martinon et al., ) , which then leads to a conformational change, resulting in the binding and the oligomerization of the adaptor molecule asc. bacterial pfts, including the hemolysin of the bacteria vibrio cholerae, or the m surface protein of the group a streptococcus can efficiently induce potassium efflux, which has been defined as the unified trigger of the nlrp inflammasome activation (muñoz-planillo et al., ; queen et al., ; valderrama et al., ) . alternatively, disruption of the glycolysis, leading to mitochondrial ros production is also proposed as the intracellular mechanism of the nlrp activation (sanman et al., ) . besides, a tlr triggered second priming event by ubiquitination is necessary to stimulate the nlrp inflammasome in mice myeloid cells (humphries et al., ) . anthrax lethal toxin (lt), the critical virulence factor of the bacillus anthracis induces caspase- activation and lytic cell death through the cytosolic nlr family member nlrp b in macrophages (boyden and dietrich, ) and dendritic cells (fink et al., ) (fig. ) . in humans, nlrp b possess a pyrin domain and lacks the card domain, therefore asc is required as an adaptor protein to activate caspase- . it has been demonstrated that lt was transported to the cytosol, where it could interact with additional factors, however the question whether nlrp b directly interacted the lt remained elusive. this question was answered few years later in a study where the direct proteolytic cleavage of nlrp b could activate nlrp b driven pyroptosis in mice and rat cells (chavarría-smith and vance, ). the role of caspase- and caspase- /- the mouse caspase- and its duplicated human homologs caspase- and - are often referred as non-canonical inflammasome, since they function as direct lps sensors in the cytosol (fig. ) . the card domain of the caspase- is capable to bind lps molecules and, surprisingly, can directly cleave lps micelles . the advantage of the caspase- and caspase- / inflammasome can be understood in situations in which the pathogen attempts to evade caspase- activation, for instance by blocking the nlrc- inflammasome activation. it has been recently shown that the shiga toxin from the pathogenic enterohemorrhagic escherichia coli could lead to the activation of caspase- and result in the cleavage of gsdmd upstream of the nlrp activation in human thp- cells (platnich et al., ) . furthermore, the gram-negative burkholderia thailandensis infection triggers the activation of caspase- in mice (aachoui et al., ) and legionella (l.) pneumophila similarly elicits the activation of caspase- , which in this scenario is dependent on the ifn signaling. interestingly, loss of ifns does not lead to the inhibition of caspase- expression, but it hampers the caspase- dependent pyroptosis in macrophages (case et al., ) . the requirement of ifns leads us to the problem of the cytosolic lps detection. the gram-negative bacteria salmonella typhimurium and l. pneumophila are normally located in vacuoles, hence they are protected from cytoplasmic pathogen sensors as for instance caspase- . the guanylate-binding proteins (gbps) represent a family of ifn-inducible gtpases. it has been demonstrated that ifn-stimulated gbp members were capable to bind the vacuoles and initiate the rupture of the membrane, which in turn resulted in the release of the bacterial lps, exposing it to the cytoplasmic caspase- (meunier et al., ; rupper and cardelli, ) . conversely, ifns can also directly regulate the transcription of the pro-inflammatory caspases. based on a new genome-wide screen, the interferon regulatory factor (irf ) seems to act as the main transcription factor directly regulating caspase- expression upon lps stimulus in human monocytes (benaoudia et al., ) . the role of the aim inflammasome in the recognition of the intracellular bacteria has been demonstrated in response to listeria monocytogenes, francisella (f.) tularensis (kim et al., ; rathinam et al., ) and brucella abortus infection (costa franco et al., ) . since the recognition by aim is fully dependent on the availability of cytosolic dna, strains engineered to lyse in the cytosol can induce pyroptosis more efficiently (sauer et al., ) . conversely, in response to a naturally-occurring infection by non-lysing strains, including f. novicida , f. tularensis (man et al., ) and chlamydia trachomatis (finethy et al., ) the cytosolic release of the bacterial dna is facilitated by the tlr mediated gbps, hence supporting the relevance of the ifn dependent priming events to activate this inflammasome. . . cross talk between inflammatory and apoptotic cell death pathways a study on caspase- deficient murine macrophages implicates the existence of the cross talk between pyroptosis and apoptotic caspases. in the absence of caspase- , the stimulation of the aim /asc axis by f. tularensis leads to caspase- cleavage and apoptosis, which can restrict bacterial replication in vitro (pierini et al., ) . the mechanistic details of the activation, however, still need to be elucidated. surprisingly, the loss of caspase- results in faster kinetics of the aim- speck formation, indicating the presence of a caspase- dependent negative feedback loop ( juruj et al., ) . we therefore speculate that an over-activation of the aim speck formation may result in the loss of specificity, consequently leading to the recruitment of other initiator caspases into the complex. in fact, the recruitment of caspase- into the nlrp -asc inflammasome has been demonstrated in lps and nigericin stimulated murine bone marrow-derived dendritic cells in the absence of caspase- and - (antonopoulos et al., ) . furthermore, the yersinia effector protein yopj has been shown to inhibit tak and ikk, which in turn led to ripk and caspase- activation in macrophages. this and another group simultaneously published the finding that caspase- could directly cleave gsdmd, leading to pyroptotic cell death (orning et al., ; sarhan et al., ) . finally, in a third study the nlrc dependent assembly of the asc caspase- inflammasome has been demonstrated in salmonella typhimurium infected cells (man et al., ) . in this review, we discussed the role of three major cell death pathways: apoptosis, necroptosis, and pyroptosis in the context of viral and bacterial infections. the past few years of research in the field of host-pathogen interaction demonstrates us that the connection between the pathogens and the host cell death signaling is more complex than previously expected. the complexity of the apoptotic, necroptotic, and pyroptotic signaling pathways, the presence of the cross talks and the feedback mechanisms reflect the significance of a continuously running evolutionary race between the pathogens and their natural hosts. this is best proven by the observations that successful human pathogens developed counter measures to attenuate the cell death induction potential of the host, whereas other pathogens developed strategies to hijack the cell death machinery in order to accomplish the targeted destruction of the defense mechanism of the host. thus, looking it from an evolutionary perspective, the different layers of cell death pathways can be interpreted as different evolutionary steps built upon each other in response to a selective pathogen insult. the better understanding of these mechanisms can provide us novel strategies to more efficiently treat infectious diseases in the future. caspase- protects against bacteria that escape the vacuole recognition of doublestranded rna and activation of nf-kappab by toll-like receptor herpes simplex virus icp impedes tnf receptor -induced necrosome assembly during compartmentalization to detergent-resistant membrane vesicles caspase- disaggregates lipopolysaccharide micelles via lps-card interaction caspase- as an effector and regulator of nlrp inflammasome signaling necroptosis of infiltrated macrophages drives yersinia pestis dispersal within buboes bacteria and host interactions in the gut epithelial barrier the nlrp inflammasome is upregulated in hivinfected antiretroviral therapy-treated individuals with defective immune recovery execution of macrophage apoptosis by pe_pgrs of mycobacterium tuberculosis is mediated by toll-like receptor -dependent release of tumor necrosis factor-alpha a genome-wide screen identifies irf as a key regulator of caspase- in human cells macrophage activation redirects yersinia-infected host cell death from apoptosis to caspase- -dependent pyroptosis nalp b controls mouse macrophage susceptibility to anthrax lethal toxin caspase- mediates a brucella abortus rb -induced hybrid cell death having features of apoptosis and pyroptosis asc and ipaf inflammasomes direct distinct pathways for caspase- activation in response to legionella pneumophila caspase- stimulates rapid flagellin-independent pyroptosis in response to legionella pneumophila intracytosolic listeria monocytogenes induces cell death through caspase- activation in murine macrophages direct proteolytic cleavage of nlrp b is necessary and sufficient for inflammasome activation by anthrax lethal factor a bacterial invasin induces macrophage apoptosis by binding directly to ice proinflammatory caspase- -mediated macrophage cell death induced by a rough attenuated brucella suis strain galectin- enhances avian h n influenza a virusinduced pulmonary inflammation by promoting nlrp inflammasome activation involvement of caspase- in il- beta production and pyroptosis in human macrophages during dengue virus infection prrs are watching you: localization of innate sensing and signaling regulators caspases: the executioners of apoptosis activation of inflammasome signaling mediates pathology of acute p. aeruginosa pneumonia aim senses brucella abortus dna in dendritic cells to induce il- β secretion, pyroptosis and resistance to bacterial infection in mice rna viruses promote activation of the nlrp inflammasome through cytopathogenic effect-induced potassium efflux situ inflammasome activation results in severe damage to the central nervous system in fatal zika virus microcephaly cases chemical inhibitor of nonapoptotic cell death with therapeutic potential for ischemic brain injury human parvovirus infection of human airway epithelia induces pyroptotic cell death by inhibiting apoptosis viral dna sensors ifi and cyclic gmp-amp synthase possess distinct functions in regulating viral gene expression, immune defenses, and apoptotic responses during herpesvirus infection cell death by pyroptosis drives cd t-cell depletion in hiv- infection serine phosphorylation inhibits ripk kinase-dependent cell death in models of infection and inflammation mlkl requires the inositol phosphate code to execute necroptosis the ribonucleotide reductase r subunits of herpes simplex virus types and protect cells against tnfα-and fasl-induced apoptosis by interacting with caspase- a papillomavirus dna from a cervical carcinoma and its prevalence in cancer biopsy samples from different geographic regions block of death-receptor apoptosis protects mouse cytomegalovirus from macrophages and is a determinant of virulence in immunodeficient hosts induction of noxa-mediated apoptosis by modified vaccinia virus ankara depends on viral recognition by cytosolic helicases, leading to irf- /ifn-β-dependent induction of proapoptotic noxa a novel mitochondrial mavs/caspase- platform links rna virus-induced innate antiviral signaling to bax/bak-independent apoptosis the pore-forming protein gasdermin d regulates interleukin- secretion from living macrophages exposure of phosphatidylserine on the surface of apoptotic lymphocytes triggers specific recognition and removal by macrophages ciaps block ripoptosome formation, a rip /caspase- containing intracellular cell death complex differentially regulated by cflip isoforms ion efflux and influenza infection trigger nlrp inflammasome signaling in human dendritic cells guanylate binding proteins enable rapid activation of canonical and noncanonical inflammasomes in chlamydia-infected macrophages caspase- -dependent pore formation during pyroptosis leads to osmotic lysis of infected host macrophages anthrax lethal toxin and salmonella elicit the common cell death pathway of caspase- -dependent pyroptosis via distinct mechanisms ikkepsilon and tbk are essential components of the irf signaling pathway the ns protein of influenza a virus participates in necroptosis by interacting with mlkl and increasing its oligomerization and membrane translocation cell-to-cell transmission of hiv- is required to trigger pyroptotic death of lymphoid-tissue-derived cd t cells pore-forming toxin-mediated ion dysregulation leads to death receptor-independent necroptosis of lung epithelial cells during bacterial pneumonia bacterial pore-forming toxins promote the activation of caspases in parallel to necroptosis to enhance alarmin release and inflammation during pneumonia absence of toll-like receptor signaling results in delayed yersinia enterocolitica yopp-induced cell death of dendritic cells bax assembles into large ring-like structures remodeling the mitochondrial outer membrane in apoptosis herpes simplex virus suppresses necroptosis in human cells species-independent contribution of zbp /dai/dlm- -triggered necroptosis in host defense against hsv mechanisms and dynamics of t cell-mediated cytotoxicity in vivo enterovirus induces apoptosis by directly modulating the conformational activation of pro-apoptotic protein bax toll-like receptors activate programmed necrosis in macrophages through a receptor-interacting kinase- -mediated pathway gasdermin d is an executor of pyroptosis and required for interleukin- β secretion induction of bcl- expression by epstein-barr virus latent membrane protein protects infected b cells from programmed cell death tnf-dependent recruitment of the protein kinase rip to the tnf receptor- signaling complex salmonella outer protein b suppresses colitis development via protecting cell from necroptosis rip /rip binding to hsv- icp initiates necroptosis to restrict virus propagation in mice the e ubiquitin ligase pellino mediates priming of the nlrp inflammasome caspase- is an initiator caspase responsible for pore-forming toxin-mediated apoptosis sting is an endoplasmic reticulum adaptor that facilitates innate immune signalling programmed cell death as a defence against infection caspase- activity affects aim speck formation/stability through a negative feedback loop apoptosis induced by the toll-like receptor adaptor trif is dependent on its receptor interacting protein homotypic interaction motif toll-like receptor -mediated necrosis via trif, rip , and mlkl irf regulates transcription of naips for nlrc inflammasome activation the human apoptosis inhibitor naip induces pyroptosis in macrophages infected with legionella pneumophila differential roles of mda and rig-i helicases in the recognition of rna viruses caspase- cleaves gasdermin d for non-canonical inflammasome signalling apoptosis: a basic biological phenomenon with wide-ranging implications in tissue kinetics listeria monocytogenes is sensed by the nlrp and aim inflammasome pro-apoptotic function of hbv x protein is mediated by interaction with c-flip and enhancement of death-inducing signal impact of caspase- / , - , - , or il- β/il- deficiency on rabies virus-induced macrophage cell death and onset of disease toxin-induced necroptosis is a major mechanism of staphylococcus aureus lung damage inhibition of dai-dependent necroptosis by the z-dna binding domain of the vaccinia virus innate immune evasion protein, e hepatitis c virus infection of cultured human hepatoma cells causes apoptosis and pyroptosis in both infected and bystander cells chikungunya virus mobilizes the apoptotic machinery to invade host cell defenses zbp /dai is an innate sensor of influenza virus triggering the nlrp inflammasome and programmed cell death pathways irf is a transcriptional regulator of zbp promoting nlrp inflammasome activation and cell death during influenza virus infection the r subunit of herpes simplex virus ribonucleotide reductase protects cells against apoptosis at, or upstream of, caspase- activation influenza a virus infection triggers pyroptosis and apoptosis of respiratory epithelial cells through the type i interferon signaling pathway in a mutually exclusive manner cyld proteolysis protects macrophages from tnf-mediated auto-necroptosis induced by lps and licensed by type i ifn tlr and caspase- are essential for group b streptococcus-induced apoptosis in microglia enterovirus inhibits pyroptosis through cleavage of gasdermin d coxsackievirus a infection induces neural cell and non-neural cell apoptosis in vitro fasl/fas pathway is involved in dengue virus induced apoptosis of the vascular endothelial cells pathogen-induced ubiquitin-editing enzyme a bifunctionally shuts off nf-κb and caspase- -dependent apoptotic cell death inflammasome-activated gasdermin d causes pyroptosis by forming membrane pores epstein-barr virus encoded latent membrane protein suppresses necroptosis through targeting ripk / ubiquitination the -kda mycobacterium tuberculosis protein induces macrophage apoptosis through toll-like receptor- rhinovirus c protease suppresses apoptosis and triggers caspase-independent cell death a null mutation in the perforin gene impairs cytolytic t lymphocyte-and natural killer cell-mediated cytotoxicity novel role of pkr in inflammasome activation and hmgb release molecular cell death platforms and assemblies salmonella infection induces recruitment of caspase- to the inflammasome to modulate il- β production the transcription factor irf and guanylate-binding proteins target activation of the aim inflammasome by francisella infection irgb liberates bacterial ligands for sensing by the aim and caspase- -nlrp inflammasomes molecular mechanisms and functions of pyroptosis, inflammatory caspases and inflammasomes in infectious diseases detection of immune danger signals by nalp gout-associated uric acid crystals activate the nalp inflammasome human t-cell leukaemia virus type (htlv- ) infectivity and cellular transformation flagellin-induced nlrc phosphorylation primes the inflammasome for activation by naip tlr , trif, and caspase determine double-stranded rna-induced epithelial cell death and survival in vivo francisella tularensis modulates a distinct subset of regulatory factors and sustains mitochondrial integrity to impair human neutrophil apoptosis positive and negative phosphorylation regulates rip -and rip -induced programmed necrosis pivotal role of receptor-interacting protein kinase and mixed lineage kinase domain-like in neuronal cell death induced by the human neuroinvasive coronavirus oc high mobility group box enables bacterial lipids to trigger receptor-interacting protein kinase (ripk )-mediated necroptosis and apoptosis in mice caspase- activation requires lysis of pathogen-containing vacuoles by ifninduced gtpases caspase- -induced pyroptosis is an innate immune effector mechanism against intracellular bacteria salmonella-induced macrophage death: the role of caspase- in death and inflammation ifi dna sensor is required for death of lymphoid cd t cells abortively infected with hiv k + efflux is the common trigger of nlrp inflammasome activation by bacterial toxins and particulate matter intracellular bacteria engage a sting-tbk -mvb b pathway to enable paracrine cgas-sting signalling regulation of viral replication, apoptosis and proinflammatory responses by -aag during chikungunya virus infection in macrophages ripk inhibits zbp -driven necroptosis during development apoptotic caspases suppress type i interferon production via the cleavage of cgas, mavs, and irf ripk activates parallel pathways of mlkl-driven necroptosis and fadd-mediated apoptosis to protect against influenza a virus stable extracellular rna fragments of mycobacterium tuberculosis induce early apoptosis in human monocytes via a caspase- dependent mechanism caspase inhibits programmed necrosis by processing cyld human t-cell leukemia virus type-i oncoprotein tax inhibits fas-mediated apoptosis by inducing cellular flip through activation of nf-kappab caspase- : the reinvented enzyme suppression of rip -dependent necroptosis by human cytomegalovirus pathogen blockade of tak triggers caspase- -dependent cleavage of gasdermin d and cell death nad+ depletion triggers macrophage necroptosis, a cell death pathway exploited by mycobacterium tuberculosis necroptosis takes place in human immunodeficiency virus type- (hiv- )-infected cd + t lymphocytes a type iii effector antagonizes death receptor signalling during bacterial gut infection espl is a bacterial cysteine protease effector that cleaves rhim proteins to block necroptosis and inflammation irf- activation by sendai virus infection is required for cellular apoptosis and avoidance of persistence aim /asc triggers caspase- -dependent apoptosis in francisella-infected caspase- -deficient macrophages granzyme b-mediated apoptosis proceeds predominantly through a bcl- -inhibitable mitochondrial pathway shiga toxin/lipopolysaccharide activates caspase- and gasdermin d to trigger mitochondrial reactive oxygen species upstream of the nlrp inflammasome phosphorylation of nlrc is critical for inflammasome activation mechanisms of inflammasome activation by vibrio cholerae secreted toxins vary with strain biotype host cell killing by the west nile virus ns b-ns proteolytic complex: ns alone is sufficient to recruit caspase- -based apoptotic pathway the aim inflammasome is essential for host defense against cytosolic bacteria and dna viruses the apoptosome: signalling platform of cell death type i interferon induces necroptosis in macrophages during infection with salmonella enterica serovar typhimurium cleavage of dfna by caspase- during apoptosis mediates progression to secondary necrotic/pyroptotic cell death induction of guanylate binding protein by gamma interferon increases susceptibility to salmonella enterica serovar typhimurium-induced pyroptosis in raw . cells casp p generated by hiv protease kills cd t cells through direct bak activation disruption of glycolytic flux is a signal for inflammasome signaling and pyroptotic cell death. elife caspase- induces cleavage of gasdermin d to elicit pyroptosis during yersinia infection listeria monocytogenes triggers aim -mediated pyroptosis upon infrequent bacteriolysis in the macrophage cytosol gsdmd membrane pore formation constitutes the mechanism of pyroptotic cell death release of chromatin protein hmgb by necrotic cells triggers inflammation natural killer and lymphokine-activated killer cells require granzyme b for the rapid induction of apoptosis in susceptible target cells a cytomegalovirus-encoded inhibitor of apoptosis that suppresses caspase- activation pseudomonas aeruginosa quorumsensing metabolite induces host immune cell death through cell surface lipid domain dissolution murine cytomegalovirus ie -dependent transcription is required for dai/zbp -mediated necroptosis is receptor-interacting protein kinase a viable therapeutic target for mycobacterium tuberculosis infection? mixed lineage kinase domain-like protein mediates necrosis signaling downstream of rip kinase cyclic gmp-amp synthase is a cytosolic dna sensor that activates the type i interferon pathway differential regulation of caspase- activation, pyroptosis, and autophagy via ipaf and asc in shigella-infected macrophages shigella type iii secretion protein mxii is recognized by naip to induce nlrc inflammasome activation independently of pkcδ mitochondria and cell death: outer membrane permeabilization and beyond dengue virus-infected human monocytes trigger late activation of caspase- , which mediates pro-inflammatory il- β secretion and pyroptosis the ripoptosome, a signaling platform that assembles in response to genotoxic stress and loss of iaps yama/cpp β, a mammalian homolog of ced- , is a crma-inhibitable protease that cleaves the death substrate poly(adpribose) polymerase dai senses influenza a virus genomic rna and activates ripk -dependent cell death viral flice-inhibitory proteins (flips) prevent apoptosis induced by death receptors virus inhibition of rip -dependent necrosis dai/zbp /dlm- complexes with rip to mediate virus-induced programmed necrosis that is targeted by murine cytomegalovirus vira group a streptococcal m protein activates the nlrp inflammasome caspases in cell death, inflammation, and disease un)expected roles of c-iaps in apoptotic and nfkappab signaling pathways intracellular regulation of tnf activity in health and disease dual signaling of the fas receptor: initiation of both apoptotic and necrotic cell death pathways infection of epithelial cells with chlamydia trachomatis inhibits tnf-induced apoptosis at the level of receptor internalization while leaving non-apoptotic tnf-signalling intact novel pandemic influenza a (h n ) virus infection modulates apoptotic pathways that impact its replication in a cells cathepsin b aggravates coxsackievirus b -induced myocarditis through activating the inflammasome and promoting pyroptosis gasdermin d protects from melioidosis through pyroptosis and direct killing of bacteria influenza virus infection induces zbp expression and necroptosis in mouse lungs multiple nodlike receptors activate caspase during listeria monocytogenes infection proapoptotic signalling through toll-like receptor- involves trif-dependent activation of caspase- and is under the control of inhibitor of apoptosis proteins in melanoma cells il- sensitizes tumor cells to tlr -mediated apoptosis caspase- and rip kinases regulate bacteria-induced innate immune responses and cell death apoptotic caspases suppress mtdna-induced sting-mediated type i ifn production death receptor signal transducers: nodes of coordination in immune signaling networks clec a is critical for dengue virus-induced inflammasome activation in human macrophages structural basis of rip inhibition by necrostatins role of adaptor trif in the myd -independent toll-like receptor signaling pathway human naip and mouse naip recognize bacterial type iii secretion needle protein for inflammasome activation ripk and nf-κb signaling in dying cells determines cross-priming of cd + t cells aim inflammasome-mediated pyroptosis in enterovirus a -infected neuronal cells restricts viral replication vaccinia virus protein f l is a caspase- inhibitor involvement of p x receptor in gp -induced pyroptosis in dorsal root ganglia target protease specificity of the viral serpin crma. analysis of five caspases nlrp b inflammasome restricts rotavirus infection in intestinal epithelial cells pyroptosis induced by enterovirus a infection in cultured human neuroblastoma cells the author is grateful to madeleine eichler for the graphical design of the figures presented in the manuscript and to usman nasrullah for the critical reading of the manuscript. key: cord- -h qy z authors: kontis, v.; bennett, j. e.; parks, r. m.; rashid, t.; pearson-stuttard, j.; asaria, p.; guillot, m.; blangiardo, m.; ezzati, m. title: age- and sex-specific total mortality impacts of the early weeks of the covid- pandemic in england and wales: application of a bayesian model ensemble to mortality statistics date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: h qy z background: the covid- pandemic affects mortality directly through infection as well as through changes in the social, environmental and healthcare determinants of health. the impacts on mortality are likely to vary, in both magnitude and timing, by age and sex. our aim was to estimate the total mortality impacts of the pandemic, by sex, age group and week. methods: we developed an ensemble of bayesian models that probabilistically estimate the weekly number of deaths that would be expected had the covid- pandemic not occurred. the models account for seasonality of death rates, medium-long-term trends in death rates, the impact of temperature on death rates, association of death rates in each week on those in preceding week(s), and the impact of bank holidays. we used data from january through mid-february (i.e., week starting th february ) to estimate the parameters of each model, which was then used to predict the number of deaths for subsequent weeks as estimates of death rates if the pandemic had not occurred. we subtracted these estimates from the actual reported number of deaths to measure the total mortality impact of the pandemic. results: in the week that began on st march, the same week that a national lockdown was put in place, there was a > % probability that there were more deaths in men and women aged [≥] years than would occur in the absence of the pandemic; the probability was % from the subsequent week. taken over the entire period from mid-february to th may , there were an estimated [~] , ( , - , ) or % ( - ) more deaths than would be expected had the pandemic not taken place. , ( , - , ) of these deaths were in females ( % ( - ) higher than if there had not been a pandemic), and , ( , - , ) in males ( % ( - ) higher). the largest number of excess deaths occurred among women aged > years ( , ; , - , ), followed by men aged > years ( , ; , - , ) and - years ( , ; , - , ). the cause of death assigned to the majority ( , ) of these excess deaths was covid- . there was nonetheless a > . % probability that there has been an increase in deaths assigned to other causes in those aged [≥] years. however, by the th of may, the all-cause excess mortality had become virtually equal to deaths assigned to covid- , and non-covid excess deaths had diminished to close to zero, or possibly become negative, in all age-sex groups. interpretation: the death toll of covid- pandemic, in middle and older ages, is substantially larger than the number of deaths reported as a result of confirmed infection, and was visible in vital statistics when the national lockdown was put in place. when all-cause mortality is considered, the mortality impact of the pandemic on men and women is more similar than when comparing deaths assigned to covid- as underlying cause of death. the covid- pandemic has led to tens of thousands of deaths among patients with confirmed infection in the uk. the pandemic has also profoundly changed the social, economic, environmental and healthcare determinants of morbidity and mortality. these changes are likely to impact public health beyond the deaths caused directly by infection through a number of routes including delayed disease prevention and procedures for acute and chronic medical care; loss of jobs and income; disruption of social networks; changes in crime and self-harm; changes in quantity and quality of food, and the use of tobacco, alcohol and other drugs; and changes in mobility and transport patterns with potential impacts on road traffic injuries and air pollution. how changes in these social, environmental and healthcare determinants impact mortality is likely to vary, in both magnitude and timing, by age and sex. an understanding of the mortality impacts beyond deaths assigned to covid- infection is needed to understand the overall public health impacts of the pandemic and control policies, titrate and adjust the response, and put in place mitigation mechanisms to minimize the adverse impacts as the pandemic and response continue beyond early weeks. we developed and applied methodology to quantify the weekly mortality impacts of the covid- pandemic and associated responses by age group and sex in england and wales. the methodology can also be used for comparable cross-country analysis on a real-time basis. we used data on the weekly number of deaths in england and wales, by age group and sex released by the office for national statistics (ons). weekly death files include all deaths registered from saturday through the subsequent friday. they also include data on the number of deaths involving covid- , which are deaths with a mention of covid- anywhere on the death certificate, including in combination with other health conditions. we used data from first week in january through the week starting on saturday nd may (ending on th may ). we used data on mid-year population by age group and sex from the ons. we calculated weekly population through interpolation, as done by the ons for quarterly population. we obtained data on temperature from era , which uses data from global in situ and satellite measurements to generate a worldwide meteorological dataset, with full space and time coverage over our analysis period. we used gridded estimates measured four times daily at a resolution of km to generate weekly temperatures for each local authority district, which we weighted by local authority population to create national level summaries. the total mortality impact of the covid- pandemic should be calculated as the difference between the observed number of deaths and the number of deaths had the pandemic not occurred, which is not directly measurable. the most common approach to addressing this issue has been to use the average number of deaths over previous years, e.g., the most recent five years, for the corresponding week or month when the comparison is made. this approach however does not take into account changes in population size and age structure, nor longand short-term trends in mortality, which are particularly pronounced for some age groups. , nor does this approach account for time-varying factors like temperature, that are largely external to the pandemic, but also affect death rates. in addition, bank holidays such as easter that do not occur in the same week of the year affect number of deaths and death registration. we developed an ensemble of short-term bayesian mortality projection models that each make an estimate of weekly death rates that would be expected if the covid- pandemic had not occurred. we used multiple models because there is inherent uncertainty in the choice of model that best predicts death rates in the absence of pandemic. these models were formulated to incorporate features of weekly death rates as follows: • first, death rates may have a medium-to-long-term trend. [ ] [ ] [ ] we developed two sets of models, one with no trend and one with a linear trend term over weekly deaths. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) • second, death rates have a seasonal pattern, which varies by age group and sex. [ ] [ ] [ ] [ ] we included weekly random intercepts for each week of the year. to account for the fact that seasonal patterns "repeat" (i.e., late december and early january are seasonally similar) we used a seasonal structure , for the random intercepts. the seasonal structure allows the magnitude of the random intercepts to vary over time, and implicitly accounts for timevarying factors such as annual fluctuations in flu season. • third, death rates in each week may be related to those in preceding week(s). we formulated four sets of models to account for this relationship. the weekly random intercepts in these models had a first, second, fourth or eighth order autoregressive structure , the higher order autoregressive models allow death rates in any given week to be informed by those in a progressively larger number of preceding weeks. further, trends not picked up by the linear or seasonal terms would be captured by these autoregressive terms. • fourth, beyond having a seasonal pattern, death rates depend on temperature, and specifically on whether temperature is higher or lower than its long-term norm during a particular time of year. [ ] [ ] [ ] [ ] [ ] [ ] the effect of temperature on mortality varies throughout the year, and may be in opposite directions for different times of year. we used two sets of models, one without temperature and one with a weekly term for temperature anomaly, defined as deviation of weekly temperature from the local average weekly temperature over the entire analysis period. the coefficients of temperature anomalies were specified as a random effect with a random walk prior of order one, so that temperature effect is more similar in adjacent weeks. the random effect had a circular structure so that late december and early january are treated as adjacent. • fifth, reported death rates in weeks that contain or follow a holiday may be different from other weeks. we included effects (as fixed intercepts) for the week containing and the week after each of the following holidays: christmas and/or boxing day; good friday; . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) easter monday; new year's day; early-may bank holiday; spring bank holiday; and summer bank holiday. • we also tested, but did not include, model terms for the weeks that coincided with a change to and from daylight saving time because the effect was negligible. these choices led to an ensemble of short-term bayesian mortality projection models ( trend options x autoregressive options x temperature options). we used the time-series of weekly reported deaths from january through mid-february (i.e., week starting th february ) to estimate the parameters of each model, which was then used to predict death rates for subsequent weeks as estimates of the counterfactual death rates (i.e., if the pandemic had not occurred). for the projection period, we used recorded temperature so that our projections take into consideration actual temperature in . this choice of training and prediction periods assumes that the number of deaths that are directly or indirectly related to the covid- pandemic was negligible through mid-february , which is about two weeks after the first confirmed case in the uk, but it allows for impacts to have appeared in subsequent weeks. we used weakly informative log gamma priors on log precision with both shape and rate equal to . . we tested the sensitivity of the results to the choice of prior through the use of penalized complexity priors and found that the results were similar. all models were fitted using integrated nested laplace approximation (inla), implemented in the r-inla software (version . ). we took , draws from the posterior distribution of age-specific deaths under each model, and pooled the , draws to obtain the posterior distribution of agespecific deaths if the covid- pandemic had not taken place. the reported credible intervals represent the . th and . th percentiles of the posterior distribution of the draws from the entire ensemble. this approach incorporates both the uncertainty of estimates from each model and the uncertainty in the choice of model. we did all analyses separately by sex and age group ( - years, - years, - years, - years, - years and + years) because death rates, and how they are impacted by the pandemic, vary by age group and sex. to obtain estimates across age groups and both sexes, we summed draws from age-sex-specific estimates. for the purpose of reporting, we rounded results on number of deaths that are ≥ to the nearest hundred to avoid giving a false sense of precision in the presence of uncertainty; results < are rounded to the nearest ten. we tested the performance of the projections from the model by withholding data for weeks starting from mid-february (i.e., the same projection period as done for ) for an earlier year and used the preceding time-series of data to train the models. we then projected death rates for the weeks with withheld data, and evaluated how well the model ensemble projections reproduce the known-but-withheld death rates. we repeated this for three different years: (i.e. train model using data from january to mid-february and test for the subsequent weeks), (i.e., train model using data from january to mid-february and test for the subsequent weeks), and (i.e., train model using data from january to mid-february and test for the subsequent weeks). we report the projection error (which measures systematic bias) and absolute forecast error (which measures any deviation from the data). additionally, we report coverage of the projection uncertainty; if projected death rates and their uncertainties are well estimated, the estimated % credible intervals should cover % of the withheld data. weekly mortality varied substantially over time in all ages with evidence of seasonal pattern above years of age ( figure ). from nd february through th march , the observed number of deaths in every age group was well within the credible interval of what was expected to have occurred if the covid- pandemic had not taken place (figures and ) . in the week that began on st march, the same week that a national lockdown was put in place, there was . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint already a > % probability that there were more deaths in both sexes and all age groups ≥ years than would occur in the absence of the pandemic; the probability was % (i.e., every one of the , draws were positive) from the subsequent week (figures and ) . the same phenomenon occurred in those aged - years in the first week of april . taken over the entire period from mid-february to th may and across all age groups, , deaths were registered in england and wales. this represents an estimated ~ , ( , - , ) or % ( - ) more deaths than would be expected had the pandemic not taken place. , ( , - , ) of these deaths were in females ( % ( - ) higher than if there had not been a pandemic), and , ( , - , ) in males ( % ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) higher). the largest overall (i.e. from any cause) number of excess deaths occurred among women aged > years ( , ; , - , ), followed by men aged > years ( , ; , - , ) and - years ( , ; , - , ). the cause of death assigned to the majority ( , ) of these excess deaths was covid- ( figures and ) . nonetheless, there was a > . % probability that there has also been an increase in deaths assigned to other causes in those aged ≥ years. the share of total excess deaths not assigned to covid- was higher in those aged years and older and was particularly high in + year olds. specifically, over the entire period from nd february to th may , the share of non-covid excess deaths was %, %, % and % in men aged - , - , - and + years, respectively, and %, %, % and % in women aged - , - , - and + years. the number of non-covid excess deaths also increased with age, reaching , ( - , ) and , ( - , ) in men and women aged - years, respectively, and , ( - , ) and , ( , - , ) in men and women aged + years. however, by the th of may, non-covid excess deaths had diminished to close to zero or possibly become negative in all age-sex groups. in boys and young men aged - and - years, there may have been a short-term decline, lasting two weeks in mid-late march, in non-covid deaths but the probability that deaths were lower than would be . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . expected without the pandemic in different weeks was < %. similarly, in the first week of may, deaths in boys/men and girls/women aged < years may have dropped below what would be expected without the pandemic with posterior probabilities ranging from % to %. the results of model validation (table ) show that the estimates of how many deaths would be expected in different weeks had the pandemic not occurred had mean projection errors < % in all age-sex groups. the mean absolute error was also < % in all age-sex groups except in those aged - years where the number of deaths is small. % coverage, which measures how well the posterior distributions of projected deaths coincide with withheld data was > % for all age and sex groups which shows that the posterior distribution is wellestimated. we applied a robust probabilistic method to coherently and consistently estimate the total death toll of covid- pandemic from the nd february to th may. at this stage, the covid- pandemic was responsible for over , excess deaths in england and wales. we also found that when all-cause mortality is considered, the mortality impact of the pandemic on men (~ % increase in deaths) and women (~ % increase in deaths) is more similar than when comparing deaths assigned to covid- . deaths that were not assigned to covid- made up % of all excess deaths, but had diminished in most age-sex groups by the first week of may. we also rule out, with virtual certainty (posterior probability > . %), the hypothesis that the death toll of the pandemic may be smaller than direct covid- deaths either because some of those dying of covid- would have died in this same time period from other underlying conditions even if the pandemic had not occurred, , . further, although children and young adults may have experienced shot-term declines in deaths, taken across all age groups any "positive" impacts of the lockdown (e.g., reductions in deaths due to air pollution or traffic injuries) are dwarfed by the negative impacts of the epidemic. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . our overall estimates are similar to those reported by the ons and the financial times but our findings reveal important details on excess deaths by age group and sex which these sources do not. euromomo does not report country-specific excess deaths and hence could not be compared with our results. the main strength of our work is the systematic use of time-series data from to early to estimate how many deaths would be expected in the absence of pandemic. by modelling death rates, rather than simply the number of deaths as is done in most other analyses, we account for changes in population size and age structure. the models incorporated important features of mortality, including seasonality of death rates, how mortality in one week may depend on previous week(s) and the seasonally-variable role of ambient temperature. the use of a modelling framework, as we have done, allowed us to make estimates by age group and sex, and, as data become available, will allow doing so for specific causes of death and subnational geographies which would, because of smaller numbers, not be possible for other methods. we used an ensemble of models which typically leads to more robust projections and represent both the uncertainty associated with each individual model and that of model choice. finally, this framework, specifically the inclusion of seasonality and ambient temperature, is well suited for more robust estimation and standardised comparisons of excess deaths across countries on a real-time basis. the main limitation of our work is that we did not have data on underlying cause of death beyond the distinction between covid- and non-covid deaths. having a breakdown of deaths by underlying cause will help develop cause-specific models and understand which causes have exceeded or fallen below the levels expected. we also could not access agespecific data by region because the ons only releases aggregate numbers for regions. nor did we have data on total mortality by socio-demographic status to understand inequalities in the impacts of the pandemic beyond deaths assigned to covid- as the underlying cause of death. releasing these data will allow more granular analysis of the impacts of the pandemic, . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . which can in turn inform resource allocation and a more targeted approach to mitigating both the direct and indirect effects of covid- , now and for future waves of the pandemic. further, weekly mortality files from the ons cover deaths registered in any given week. these include some deaths from prior weeks and leave out some deaths in the reporting week. however, the approach is consistent over time and does not affect year to year comparisons, including for as the lags in registration of deaths assigned to covid- seem to be the same as those from other causes. it is likely that some of the apparently non-covid excess deaths are due to undetected covid- infections. an example of such deaths are the likely covid- deaths in care homes. , other such deaths may be those who called the nhs helpline or the ambulance service, were advised to self-isolate because their symptoms were not deemed sufficiently severe to be admitted, and died at home. , that the share of excess deaths from non-covid causes became smaller over time may be because with increasing awareness of, and attention to, clinical symptoms more of such deaths are assigned to covid- as the underlying cause. it is also possible that many excess deaths have been caused, and may continue to do so in coming months, by the pandemic due to changes in personal and family economic and employment circumstances, and in healthcare provision, access and utilisation, as evidenced by reductions in a&e attendance and procedures for a diverse range of acute and chronic conditions. [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] while the official position on this has been to encourage people to seek care, the situation is more complex: for some accessing care becomes more restricted because their family members are infected and cannot continue supporting them. others, typically those in limited and marginalised housing and employment, may not do so in fear of losing their livelihood. the large death toll of the pandemic, from deaths assigned to covid- as well as other causes, together with the fact that excess death toll was already happening when a national lockdown was announced indicate that the cessation of community contact tracing in early . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . march, hesitation in putting a lockdown in place earlier, and sub-optimal identification and management of those with complex conditions in the community at increased risk from both the virus and effects of lockdown, is likely to have contributed to the substantial excess deaths. minimising these impacts requires a coherent strategy that supresses the epidemic and strengthens the social safety net and healthcare provision, in facilities as well as in the community and at home, together with transparent communication to encourage resumption of care seeking. we thank giulia mangiameli for help with background materials and references. all authors contributed to study design. vk and jeb developed and tested statistical methods with input from other authors. vk, rmp, tr and jeb accessed, harmonised and analysed data. vk conducted analysis and prepared results. me wrote the first draft of the paper and other authors contributed to the paper. me reports a charitable grant from the astrazeneca young health programme, and personal fees from prudential and scor, outside the submitted work. jp-s is vice-chair of the royal society for public health and reports personal fees from novo nordisk a/s and lane, clark & peacock llp, outside of the submitted work. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q cumulative deaths. the grey shading shows the levels of credible intervals around the median prediction, from % (dark grey) to % (light grey). the red points show the number of deaths assigned to covid- as underlying cause of death. the difference between these points and the curves is excess non-covid- deaths. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q mitigating the wider health effects of covid- pandemic response and year-end review reanalysis datasets excess uk deaths in covid- pandemic top , financial times the future of life expectancy and life expectancy inequalities in england and wales: bayesian spatiotemporal forecasting contributions of diseases and injuries to widening life expectancy inequalities in england from to : a population-based analysis of vital registration data contribution of six risk factors to achieving the x non-communicable disease mortality reduction target: a modelling study national and regional seasonal dynamics of all-cause and cause-specific mortality in the usa from seasonality of deaths in the u.s. by age and cause excess winter deaths in europe: a multicountry descriptive analysis deaths in winter: can britain learn from europe? forecasting: principles and practice applied bayesian modelling high ambient temperature and mortality: a review of epidemiologic studies from relation between elevated ambient temperature and mortality: a review of the epidemiologic evidence mortality risk attributable to high and low ambient temperature: a multicountry observational study impact of ambient temperature on morbidity and mortality: an overview of reviews vulnerability to the mortality effects of warm temperature in the districts of england and wales anomalously warm temperatures are associated with increased injury deaths bayesian model averaging: a tutorial approximate bayesian inference for latent gaussian models using integrated nested laplace approximations (with discussion) two thirds of coronavirus victims may have died this year anyway, government adviser says the telegraph office for national statistics. comparison of weekly death occurrences in england and wales: up to week ending coronavirus death in california came weeks before first known u.s. death. the new york times coronavirus involved in quarter of care home residents' deaths in england and wales the guardian the health foundation. care homes have seen the biggest increase in deaths since the start of the outbreak london woman dies of suspected covid- after being told she was 'not priority'. the guardian the uber driver evicted from home and left to die of coronavirus coronavirus crisis could lead to , more cancer deaths, experts warn the guardian more than m operations cancelled as nhs fights covid- . the guardian how coronavirus is impacting cancer services in the uk record drop in a&e attendance in england 'a ticking timebomb', say doctors the guardian patients with heart attacks, strokes and even appendicitis vanish from hospitals. the washington post covid's other casualties. reuters investigates response of cardiac surgery units to covid- : an internationally-based quantitative survey key: cord- - voe r f authors: kim, moon-young; cheong, harin; kim, hyung-seok title: proposal of the autopsy guideline for infectious diseases: preparation for the post-covid- era (abridged translation) date: - - journal: j korean med sci doi: . /jkms. . .e sha: doc_id: cord_uid: voe r f with the rapidly spreading coronavirus disease (covid- ) pandemic over the past few months, the world is facing an unprecedented crisis. innumerable lives have been lost to this novel infectious disease, the nature of which supersedes conventional medical understanding. the covid- pandemic is not just a global health crisis, several aspects of life in the post-covid- era are also being contemplated. experts in unison are warning that the upcoming changes in all areas of life could potentially be far more drastic than ever experienced in the entire human civilization. the medical community is no exception, and therefore, personnel involved in forensic medicine also need to be adequately prepared for the future. forensic medicine is a branch of medicine dedicated to one of the most important stages of the human lifecycle and has always been at the forefront in times of unprecedented social change. the autopsy, one of the most important tools of forensic medicine, is also useful to infectious diseases because it identifies the causal relationship between death and infection, reveals medical and epidemiological knowledge, and provides objective evidence for legal disputes. we present new autopsy guidelines in forensic medicine, formulated based on the various infectious diseases that we presently live with and may encounter in the future. in formulation of these guidelines several considerations have been taken into account, namely, the role forensic pathologists should play in the post-covid- era and the necessary preparations as well as the support needed from society to fulfill that role. the present covid- outbreak should be a starting point for formulating improvements in current practices in forensic science, including autopsy biosafety practices and the medicolegal death investigation system. despite the development of medical science, as the complexity of our society increases, various microorganisms that have the potential to be infectious agents constantly threaten humanity. through accumulated mutation, even well-known microorganisms are becoming new species, resulting in stronger transmission or higher number of fatalities. korea is also in a situation where management is required of both the interior spread and the foreign inflow of various infectious diseases. for examples of the former in korea, there are respiratorymediated infectious diseases, such as tuberculosis, which is known to be endemic, and bloodmediated infectious diseases, such as hepatitis b, hepatitis c, and acquired immune deficiency syndrome (aids), which need continuous monitoring. , numerous foreign infectious diseases are newly emerging as a result of changes in climate and biological distribution due to environmental degradation, and the collapse of interspecies barriers. as international exchange increases, they can flow into other countries at any time. recently, several respiratory diseases caused by novel viruses, such as severe acute respiratory syndrome ( in a situation where enormous social and economic losses are caused by the periodic outbreak of novel infectious diseases, the national quarantine system requires improvement to cope with the public health crisis. we believe that autopsy can provide the basic data for establishing appropriate quarantine and preventive measures. the autopsy is the most direct approach to a disease or other medical abnormalities. historically, a wide range of information on pathogenesis, epidemiology, and the natural course of numerous diseases has been collected through autopsy, leading to the development of medicine. also, the autopsy identifies legal problems related to death and prevents potential disputes, the necessity for which has been recognized across many sectors of society. it should be considered more important in a death related to an infectious disease. while the clinical environment is ready for infectious diseases under administrative and financial support, the death investigation system in korea does not seem to be comparable. although many infectious diseases are diagnosed only postmortem through autopsy, the personnel related to the autopsy are exposed to the risk of infection, due to insufficient clinical information, lack of facilities or equipment for protection, and injury accidents. therefore, a guideline for the standard autopsy for infectious diseases is stated here, which aims to: ) provide scientific grounds to establish appropriate plans for the prevention and treatment of infectious diseases, ) contribute to improving national health by controlling the spread of pathogens within the community, and ) protect human resources engaged in autopsy-related work from the risk of infection. several autopsy guidelines, including more recent ones focusing on covid- , have been adopted here. [ ] [ ] [ ] [ ] [ ] most of them suggest that the principles of handling covid- during autopsy are not different from that of the handling of other infectious diseases. this guideline does not present the current modus operandi, but indicates the way in which we need to operate from now on, and which needs our continuous effort dedicated to forensics, as well as support from the related social systems. the pathogen of infectious diseases includes various microorganisms, such as bacteria, viruses, fungi, parasites, and even prions. among various routes of transmission, direct contact of blood or body fluids, and aerosol transfer via droplet or its nuclei are considered as important during the autopsy. patients with active infection could have symptoms of acute, subacute, or chronic status, which is called clinical disease, or have no apparent symptoms, which is called subclinical or occult disease. infection by some agents could be inactive for a certain period, which is called latent infection. the diagnosis of an infectious disease could be considered based on ) medical history, from the statements of his or her acquaintances or formal medical records; ) postmortem tests for the detection of microorganisms, such as serologic, genetic, or culture tests using blood, secretion, fluid, or tissue; ) pathologic findings, using conventional and special stains; and ) epidemiologic information about the deceased or his or her close contacts, such as the location of residence and workplace, occupation, travel history, and recent whereabouts. the infectious disease control and prevention act of korea designated some infectious diseases with epidemiologic importance as 'legal infectious diseases.' these diseases were classified into four classes according to their severity, infectiousness, and isolation level (appendix ). an emerging infectious disease with the possibility of severe symptoms or rapid transmission is considered an 'emerging infectious disease syndrome' in class . covid- , caused by sars-cov- , is an example of this temporary classification, which should be classified properly after the pathogenesis and clinical features are further revealed. according to the act, a doctor who identified an infectious disease from a living patient or a dead body should report to the regional public health center. the director of the korea centers for disease control and prevention (kcdc) may order an autopsy of the deceased who is suspected of having died from an infectious disease, to confirm the final diagnosis. the autopsy process should be conducted by a specialist in infectious disease, human anatomy, pathology, or forensic medicine, in a facility with an adequate level of biosafety. the kcdc has suggested a revised classification of the risk groups of infectious agents (appendix ) in , which is based on the classification for the biology laboratory published by the who in . according to this classification, risk group includes the pathogens that are unlikely to be a serious hazard, such as hbv and hcv, while risk group includes the pathogens that usually cause serious diseases, such as mycobacterium tuberculosis, sars-cov, and hiv. for both groups, effective treatment and preventative measures are available in general. the infectious disease control and prevention act of korea classifies the safety control measures of the facilities handling high-risk pathogens into four grades (appendix ), which correspond to the biosafety levels (bsls or bls) suggested by the who. they could be applied to all the pathogens identified so far. registration with the kcdc is required for handling high-risk pathogens of grade and , while permission from the kcdc is required for those of grade or . the classification is as follows: • grade : facilities that handle high-risk pathogens that are unlikely to cause diseases to healthy adults. • grade : facilities that handle high-risk pathogens that can cause human diseases unlikely to be a serious hazard and for which effective treatment and preventive measures are available. • grade : facilities that handle high-risk pathogens that usually cause serious human diseases and for which effective treatment and preventive measures are available. • grade : facilities that handle high-risk pathogens that usually cause serious human diseases and for which effective treatment and preventive measures are not usually available. in addition, the same act designates some infectious agents as 'high-risk pathogens' that require special attention from the nation and society, because of the potential of serious risk to public health if used for biological terrorism, or spread to the outside by accident (appendix ). some agents in risk group , , and recommended by the who and some causative agents of a recent outbreak, such as sars-cov and mers-cov, are included in this list. it is anticipated sars-cov- will be added here in the near future by revision of the act. autopsy plays a critical role in ) determining the situation and specific causes of death, ) excluding other causes of death when a patient dies during treatment or isolation for a confirmed infection, and ) evaluating the medical relationships between infection and death if the infection is not a direct cause of death. autopsy is able to provide crucial information for ) the establishment of an appropriate treatment plan based on the pathological mechanisms by confirming the clinical course, symptoms, histology, and prognosis, and ) scientific evidence to control and prevent the spread of pathogens within the community, by identifying the path of transmission, and the prevalence of the target population. since immediately after a particular death, the possibility of legal disputes related to the death are often unclear, and the bereaved family are often confused, the conducting of an autopsy should be decided under careful consideration of the circumstances surrounding the death. potential legal disputes may be related to the validity and relevance of medical treatment or administrative actions, compensation claims against industrial accident insurance or commercial medical insurance, or professional negligence of a business owner. most of the situations are already covered by the criteria for the decision of unnatural deaths suggested by the kslm (appendix ), or the instructions for handling unnatural deaths declared by the korean national police agency (appendix ). most of the medical history provided before the autopsy by the police is limited to the statements of bereaved families or acquaintances, or concise data from the national health insurance corporation. obtaining the medical records of the deceased needs an additional effort of the police or the bereaved families. however, the medical information is mandatory in identifying the health status of the deceased, and preparing against the potential risk of infection. the incidence of tuberculosis among autopsy workers is known to be - times that of the general population, while it has never been investigated in korea. a smallgroup survey in korea indicated that the prevalence of tuberculosis and hepatitis b among medical workers was suspected to be very high. because hepatitis c and aids are difficult to cure and have a poor prognosis, serologic tests are performed on all surgical patients to protect medical personnel. but currently in korea, no particular tests are required to be conducted in the routine for a dead body before the autopsy. the purpose of the medico-legal autopsy may be divided into a judicial one, to confirm criminal relevance, and an administrative one, to manage public issues related to infectious diseases, accidents, or disasters, while that of the clinical autopsy is usually focused on medical evaluation. in korea, the legal basis for all forms of autopsy is prepared. for example, in the cases of infectious diseases that are not expected to be related to crime, the autopsy may be conducted by the minister of health and welfare, the mayor, the governor, the director of the kcdc, or the head of the quarantine office. but in practice, the autopsy is always requested by the police, which inevitably limits its purpose. although autopsy rooms are installed at the national forensic service (nfs) and its local branches, and some medical schools having forensic or pathology departments, their bsls are in different situations. for example, the headquarters of the nfs has a special autopsy room of bsl , while some medical schools have only bsl rooms. in principle, if the deceased is known to be a tuberculosis patient, the autopsy should be conducted in the bsl autopsy room, because mycobacterium tuberculosis belongs to risk group with sars-cov and hiv. but this principle is hard to follow, due to the high prevalence of tuberculosis in korea, and a lack of medical history, as mentioned above. each institution is in charge of the management of personal protective equipment (ppe) required for the autopsy, without sharing a standardized protocol. to assess the risk of infection caused by autopsy, the medical conditions of autopsy personnel should be checked periodically, especially after the autopsy of a high-risk person. throughout the branches of the nfs and the universities, there are no principles for the list of target pathogens, the method and frequency of surveillance test, and the criteria for subjects who need such monitoring. to conduct an autopsy, the sequence of procedures should be involved, of ) transfer from the funeral home, ) receive the body at the autopsy room, ) check the identity of the body with the police or bereaved family, ) perform the autopsy, ) return the body to the funeral home, and ) transfer the samples for postmortem test to other departments. the workers who will be involved before and after the autopsy should be guided and trained in the use of ppe and hygiene control, because during the wrapping and transporting processes, there is a possibility of contagion from the deceased. laboratory personnel dealing with samples taken from the body during the autopsy should be aware of the potential risk of infection in all autopsy samples, and receive the same level of health support as the autopsy personnel. in particular, all laboratories dealing with the initial sample that has not been chemically treated or biologically inactivated, must have bsl or higher level of facilities and appropriate ppe. throughout the autopsy-related facilities in korea, there is only a low level of safety considerations for these types of personnel who are not directly involved in the autopsy, and the level is insufficient to deal with a body or samples infected by a highrisk pathogen. all bodies should be considered to be infected by unspecified microorganisms, until they are diagnosed as negative by a medically verified examination, using appropriate samples. the autopsy personnel have the right to be protected from infection by the body, for which the affiliated agencies should make appropriate efforts. even if an infectious disease is newly diagnosed after the autopsy, the risk of infection to the autopsy personnel should be low level. a 'standard autopsy' for infectious diseases is defined as an autopsy conducted by an agreed procedure for this purpose, which should always be observed, regardless of the prevalence of the infectious disease. for confirmed cases, some conditions could be added for the optimal protection of the autopsy personnel. in contrast, if any of the facilities, personnel, equipment, or procedure did not meet the standard, an autopsy shall be considered as an 'ordinary autopsy'. the risk of transmission during an autopsy could be assessed according to the infection status of the body ( table ) . to conduct an autopsy for a confirmed case, the biosafety levels of the facilities for autopsy and laboratory tests should be equivalent to or higher than that of the pathogen. during the prevalence of a certain infectious disease, all the unknown cases should be regarded at least as suspected cases. however, considering the realistic restrictions, if there were reasonable compensations, such as preliminary tests before the autopsy, adequate ventilation and disinfection of the facilities, or additional use of ppes, the autopsy could be conducted by substandard protocols. even for the negative cases that are allowed for the ordinary autopsy, a higher level of protection is recommended, because there is always the possibility of a false-negative. considering the prevalence and biological risks, a list of infectious pathogens should be selected, and periodically evaluated for surveillance. preliminary tests for these pathogens are recommended. during the prevalence of high-risk pathogens (appendix ) or their equivalents, preliminary tests are mandatory for clinically suspected cases to determine the conduct and the coverage of the autopsy. the autopsy can be postponed until the results of the preliminary tests are available. even for the cases in which the preliminary test was negative, if suspicious findings were found during the autopsy, it is recommended to repeat the test with the autopsy samples. the possibility of false-negatives should always be considered, because the results could be affected by the infection period, sampling methods, status of the samples or bodies, or the characteristics of the test itself. place the body in a leak-proof transparent plastic bag with a thickness of μm, and seal it. do not use pins or clips that can damage the sealing conditions. put the plastic bag into another opaque body bag, and wipe its outer surface with sodium hypochlorite diluted : (e.g., % sodium hypochlorite ml + water ml mix), and dry. attach an identification tag to both the body and its bag, respectively, and make sure that they are not lost. refrigerate the body at °c. at the beginning of the autopsy, disinfect the outer and inner surface of the body bag and the skin of the body with % alcohol or sodium hypochlorite diluted : (e.g., % sodium hypochlorite ml + water ml mix). the biosafety standard of the autopsy-related facilities may correspond to the bsl in general, although a little modification is required to reflect the procedure and equipment of the autopsy. the concept of bsl is also adopted in the 'standards for the installation and operation of facilities handling high-risk pathogens (ministry of health and welfare notice no. - )' (appendix ) , which the korean institutes should follow for handling microorganisms with potential biologic risk. a bsl autopsy room is required for the ordinary autopsy, while a bsl or higher level is required for the standard autopsy, according to the risk group of the confirmed or suspected pathogen. in an autopsy room that does not meet the above criteria, at least ) the air inside the autopsy room should not escape to other spaces in the building, ) the route of exhaust should avoid other intake vent or public spaces, and ) additional devices or ppes should be utilized to compensate insufficiently met requirements. considering the environment of the autopsy room and the prevalence status of the time, the preliminary test of all the requested bodies should be considered for certain pathogens, and be referred to the decision of the conduct and coverage of the autopsy. waste generated in all processes related to the body correspond to medical waste. they should be immediately disposed of in a dedicated envelope or containerboard box. in particular, sharp tools, such as injection needles, suture needles, or scalpels, should be discarded in a dedicated plastic container. waste are sealed, disinfected, and then refrigerated in a dedicated warehouse. they should be transported to a medical waste incinerator within days, and disposed of within days. if a surface is contaminated, wipe it with sodium hypochlorite diluted : (e.g., % sodium hypochlorite ml + water ml mix), and leave it for - minutes, before wiping it again with water. if a metal surface is to be disinfected, wipe it with % alcohol (e.g. % alcohol ml + water ml mix). if a surface is visibly contaminated by blood and body fluids, wipe it with sodium hypochlorite diluted : (e.g., % sodium hypochlorite ml + water ml mix), and leave it for minutes, before wiping it again with water. the sodium hypochlorite solution should be newly mixed each time. after disinfection is finished, thorough ventilation is required. reusable surgical garments (e.g., gown, mask) made of cotton could be included in the alternative list of ppes. the cotton contaminated with blood or body fluids should be washed with hot water at °c or higher. if unavailable, soak them in sodium hypochlorite diluted : (e.g., % sodium hypochlorite ml + water , ml mix), and leave them for minutes before washing. handle as gently as possible, to avoid aerosols. aiming to protect the whole body of the autopsy personnel, including respiratory tract, eyes, and hands, from the infection, ppe should be selected in consideration of the nature and the infection route of the pathogen, and the expected situation of possible exposure. ppe should in principle be disposable or single-use, but some items (e.g., powered air-purifying respirators (papr), goggles, face shields, surgical garments made of cotton, and boots or shoes) may be designed for reuse, which should be disinfected or sterilized according to the manufacturer's instructions. keep hair from flowing down, and remove personal accessories, like watches, in advance. to prevent unnecessary contamination, each manual of the ppes, including dressing and undressing orders, should be understood in advance, and properly applied. once the ppes are used, they shall be discarded or disinfected, being regarded as contaminated. hand hygiene shall be carried out before and after dressing or undressing. damaged or contaminated ppes should be discarded, without being reused or stored again. cross-check between the autopsy personnel is recommended of whether the ppes are worn properly, or not. the dressing order of ppe should be as follows: the undressing order of ppe should be as follows: it is recommended to disinfect inner gloves at each step, as during the undressing they may become contaminated. if ppes are found to be damaged, these cases should be considered as exposure to the pathogen, followed by proper management for the personnel. all the processes should be supervised by an experienced forensic pathologist. the number of people who participate in the autopsy should be minimized. however, it is recommended that at least two people be present in the autopsy room, in case of an emergency. to prevent cutting injuries, the dissection of each body part should be conducted by only one person at a time. a person who is not directly participating in the autopsy, such as the bereaved family member or police, is restricted from entering the autopsy room. during the prevalence of certain infectious diseases or the autopsy for confirmed cases, the access of trainees, such as medical students or residents, is also restricted. if necessary, observation through a window or a monitor is recommended, in a completely separate space from the autopsy room. all the autopsy personnel should be cautious with sharp objects, such as scalpels, knives, needles, or bone sections, which can cause cutting injuries. damaged or contaminated ppes should be immediately discarded, and replaced with new ones. in the case of exposure to infection source, disinfect the exposed area immediately in a proper way; and if there is medical evidence, start prophylactic treatment. if the body is suspected to have an airborne disease, the following should be operated with special caution, to prevent aerosols: ) cutting bone with electronic saws, for which replacement if a full-body suit is not available, surgical cap and long boots can be used to minimize exposed parts. also, if the suit or gown is not made of waterproof material, the waterproof function of the ppes can be supplemented with a plastic apron or arm covers; b although they are not truly cut-proof, work gloves made of cotton may interrupt the movement of blades. by manual saws or additional use of vacuum inhalers is recommended, ) opening the containers or centrifugation of samples, ) body movement during transportation or postural adjustment during autopsy, which may cause spout of oral and nasal contents, ) incision of the bronchus or lung parenchyma, which may expose the secretion inside, and ) washing the body with a showerhead, which may spray its body fluids or adhesives together. for the suspected cases or the unknown/negative cases with suspicious findings in the autopsy, a medically verified test at the time for each pathogen or disease should be requested, with appropriate samples. in these cases, the autopsy personnel and facilities should be managed as if they participated in the autopsy for confirmed cases, until the test result is assured to be negative. the initial sample, which is not chemically treated nor biologically inactivated, should be handled within the biosafety cabinet installed in the bsl laboratory by experienced personnel wearing the ppes equivalent to that used in the autopsy room. meanwhile, after chemical treatment or biologic inactivation, the samples can be handled on an ordinary bench. purified dna or protein can be handled in the bsl laboratory, but the use of a biosafety cabinet or its equivalent is recommended. to transport the autopsy samples, they should be prepared in the following order: ) put the samples into the primary container, and seal it, ) disinfect the outer surface of the primary container with % alcohol, and label it with an identification tag, ) wrap the primary container with an absorbent (e.g., paper towel), ) put the primary container into the secondary container, and seal it, ) put the secondary container into the tertiary container and seal it, and then label it with a tag. the personnel who pack or open the containers should wear the ppes equivalent to that used in the autopsy room or laboratory. any work that requires contact with the containers, for example, simple transportation in sealed status, requires at least the wearing of gloves. the affiliated agency should recognize the major infection history of all personnel who participate in autopsy or handle postmortem samples, and take necessary measures to prevent infections. if the standard autopsy was conducted without any damage of ppes, the risk of infection is generally low. however, if the biologic nature of the pathogen or the epidemiology and pathophysiology of the disease are not fully identified, all the participants should be alert during the expected incubation period, even though they are not obviously exposed, with self-monitoring of the symptoms and the minimizing of face-to-face contacts. all the personnel who have accessed the autopsy room should be recorded: not only the direct participants in the autopsy, but also assistants for the maintenance of the facilities. considering the prevalence and biological risk, a list of infectious pathogens should be selected, and periodically evaluated for surveillance. if there is clinical evidence, prophylaxis, like vaccination, is recommended. in particular, each participant in the autopsy of confirmed cases should check whether he or she is already infected with the pathogen or not, so that if he or she is infected during the autopsy, the infection source could be traced. in the case of personnel who participated in the autopsy of a body confirmed to be infected but the standard protocol was followed, there is no possibility of exposure, so only selfmonitoring of the symptoms and the minimizing of face-to-face contacts during the expected incubation period are required. however, if the autopsy procedures failed to meet the standard protocol, or the ppes were damaged, infection should be suspected. in this case, self-isolation during the expected incubation period, and if available, prophylaxis, is required. the relevant personnel should be tested for the pathogen at the time point when related symptoms are shown, or the isolation period is nearly ended. if a test was requested after the autopsy, but the results are pending, the same actions are required in the interim. if the autopsy procedures failed to meet the standard protocol, or the ppes were damaged, the forensic pathologist in charge of the autopsy may consider adjusting the participant members, or discontinuing and delaying the autopsy schedule, to protect the autopsy personnel. if a test for certain infection was requested after the autopsy, the process and the result should be shared with all the personnel who had, or would have, contact with the body, including the police, the bereaved family, the person who has discovered, reported, inspected, or transported the body, and the funeral staff. they are required to minimize faceto-face contacts, until the test result is confirmed. covid- is a respiratory syndrome caused by the infection of sars-cov- , which belongs to the coronavirus family. currently, in korea, covid- is regarded as an 'emerging infectious disease syndrome,' which is included in class legal infectious disease (appendix ), and sars-cov- is considered a high-risk pathogen, which needs 'urgent management' (appendix ) . it is known to be transmitted through aerosols, droplets, or direct contact, while the viruses have also been found in tears and feces. , the incubation period is up to days, and symptoms were expressed within . days after exposure in % of the infected. according to the studies published so far, the survival period of sars-cov- is hours in aerosols, hours on copper surfaces, hours on cardboard surfaces, and - days on plastic or iron surfaces, which indicate that sars-cov- can survive for a considerable period outside of the host. ) clinical and pathologic findings sars-cov- patients show diverse symptoms, ranging from asymptomatic to severe respiratory failure. major symptoms are fever, fatigue, dry cough, muscle ache, and shortness of breath; and a few cases included sputum, headache, hemoptysis, and diarrhea. recently, the cdc of the united states and the kcdc added ageusia and anosmia as major symptoms of covid- . the patients are frequently diagnosed with viral pneumonia, regardless of the actual severity of their symptoms. as of yet, there is no specific therapeutic agent or vaccine. severe patients suffer from respiratory failure, septic shock, and multiple organ failure. the median time to respiratory failure was . days from symptom onset, while that to mechanical ventilation was . days. according to a few reports of autopsy or histopathology test, microscopic findings included diffuse alveolar damage, fibromucinous exudates, inflammatory infiltration in the interstitium or intra-alveolar area, viral cytopathic-like change, and thrombogenic vasculopathy. - a diagnostic test for sars-cov- could be considered based on ) medical history or symptoms, which are mainly fever or respiratory symptoms, and also include headache, abdominal pain, and fatigue, ) epidemiologic connection, such as temporal, spatial, or geographical relationships with an epidemic region or confirmed patient, and ) gross pathologic findings of the lungs, such as consolidation, thick exudates, excessive mucus, or other findings suggestive of acute or severe pneumonia, regardless of the clinical symptoms. the autopsy of confirmed and suspected cases should be conducted at bsl or equivalent facilities. for unknown cases, the autopsy could be conducted under bsl facilities, but there should be reasonable compensations, such as preliminary tests before the autopsy, adequate ventilation and disinfection of the facilities, or additional use of ppes. the management of autopsy related facilities follows the standard autopsy protocol as mentioned above. even though the generation of droplets or aerosols from bodies is unlikely, it is recommended to minimize direct contact with the bodies or postmortem samples, and prevent damage of ppes. for pathologic study, the respiratory system, including proximal and distal trachea, pulmonary hilum, main and segmental bronchi, pulmonary parenchyma, and other organs, such as the heart, the liver, kidney, spleen, and intestines, could be sampled, according to the purpose of the study. fix them with % formalin for - days. ( ) the autopsy for confirmed cases the management of facilities, environment, and human resources follows the standard autopsy protocol suggested above. the initial samples should be sent to the bsl laboratory. if there is a risk of infection due to damage of ppes, skin exposure, aerosol-prone manipulation, or cutting injuries, the relevant autopsy personnel and his or her contacts should be provided with proper medical treatment, including disinfection and virus test. also, for weeks from the exposed time point, which is the expected incubation period of covid- , they should be isolated and excluded from the work, even though the initial test result is negative. the affiliated agency should monitor his or her symptoms. ( ) the autopsy for suspected or unknown cases if a virus test for the body is carried out during the autopsy, the participated autopsy personnel should minimize face-to-face contact, until the results are notified. if the result of the virus test is positive, a postmortem test, such as toxicology (except for alcohol), biochemical, or genetic test should be requested, after the disinfection of the samples by mixing with % alcohol for (sample): (alcohol) ratios. then the samples should be transported according to the standard protocol above. the management of the facilities, environments, and human resources generally follows the standard protocol above, while the disinfection process of the autopsy room and related facilities should refer to the kcdc guideline. if an infection is suspected, for example, due to a substandard autopsy procedure or damaged ppes, the autopsy personnel and their contacts should be provided with proper medical treatment, including disinfection and virus test, with isolation and monitoring for weeks, as mentioned above in section )- ( ) . when the body is confirmed to be negative for the virus test, the isolation and monitoring could be discontinued. since the autopsy personnel are under constant risk of infection, there should be consistent effort for the implementation of the standard autopsy guidelines. first of all, expecting the periodic spread of infectious diseases in the future, the preparation of adequate level of ppes / https://jkms.org https://doi.org/ . /jkms. . .e and bsl of the autopsy related facilities, and the establishment of a health monitoring and surveillance system are required. to compensate for the problems of the current death investigation system, which is focused on the judicial purpose, in the short term, the range of unnatural death considered as the subject of judicial autopsy should be expanded as wide as possible, in consultation with the police and the prosecution. in the long term, the autopsy request ordered by the directors of the ministry of health and welfare or the kcdc should be encouraged with systemic supports. also, the ministry of health and welfare or the kcdc should be in charge of the management of biosafety requirements in the autopsy facilities and the arrangement of the qualified human resources and financial support, so that this guideline could be satisfactorily implemented. forensic medicine has developed and gradually improved over a long period, despite all the difficulties such as unfavorable environment and systemic constraints. however, in the upcoming post-covid- era, there should be more integrated and organized provision, especially against the risk of infectious diseases. health authorities and forensic pathologists should work together to improve the autopsy environment and the death investigation system, so that a better national health system can be established in the near future. trends in infectious disease mortality the korean society of infectious diseases. guidelines for potential emerging infectious diseases in korea. seoul: the korean society of infectious diseases republic of korea biosafety considerations for autopsy interim guidance for collection and submission of postmortem specimens from deceased persons under investigation (pui) for covid- briefing on covid- : autopsy practice relating to possible cases of covid- guide to forensic pathology practice for death cases related to coronavirus disease (covid- ) (trial draft) central disaster management headquarters and central disease control headquarters. corona virus infection- response guideline. - th ed. cheongju: central disaster management headquarters and central disease control headquarters mycobacterium tuberculosis at autopsy--exposure and protection: an old adversary revisited occupational infections of health care personnel in korea review article: gastrointestinal features in covid- and the possibility of faecal transmission evaluation of coronavirus in tears and conjunctival secretions of patients with sars-cov- infection aerosol and surface stability of sars-cov- as compared with sars-cov- pathological findings of covid- associated with acute respiratory distress syndrome covid- autopsies withdrawn: mortality of a pregnant patient diagnosed with covid- : a case report with clinical, radiological, and histopathological findings complement associated microvascular injury and thrombosis in the pathogenesis of severe covid- infection: a report of five cases death in the course of judicial execution, such as arrest, interrogation, detention, etc death at the accommodation for health, welfare, and nursing, etc . the term "unnatural death" means a death falling under any of the following whose cause is unclear:a. death suspicious, or confirmed to be related to a crime b. accidental death due to natural disaster, traffic accident, safety accident, industrial accident, fire, drowning, etc. c. suicide, or death suspected as suicide d. death in the course of judicial execution, such as arrest, interrogation, detention, etc. e. death at the accommodation for health, welfare, and nursing f. death suspected as acute poisoning by drug, pesticide, alcohol, gas, etc. g. other death with unknown cause . the term "unnatural death case" means a case in which one or more bodies that correspond to, or are suspected of unnatural death, are found. . the director of an unnatural death case shall apply for a warrant for an autopsy in any of the following cases (referred to as a "priority control case"), unless there are special circumstances: a. death suspected to be by murder b. unidentified body, despite the investigation of belongings, fingerprints, etc. at the scene c. death that is expected to draw social attention, such as collective death, child abuse, etc. d. severely decomposed body, so hard to identify injuries or cause of death . the director of an unnatural death case shall consider a warrant for an autopsy in any of the following cases (referred to as a "autopsy-considered case"), to confirm the relation to a crime: a. unexpected death of infant or child b. death in the course of judicial execution, such as arrest, interrogation, detention, etc. c. death suspected as acute poisoning by drug, pesticide, alcohol, gas, etc. d. death suspected to be drowning or falling, for which eyewitness or cctv footage is unavailable e. body that is carbonized or skeletonized f. death for which the bereaved family harbors suspicions about the cause g. death by traffic accident suspicious for the relationship to other crime h. death of a person with excessive death benefit, compared to his or her property i. death with disagreement about the cause between the inspection doctor, the investigators, or the director of the case j. other death for which autopsy is required to confirm its cause or circumstance key: cord- - vrb law authors: bao, richard; chen, august; gowda, jethin; mudide, shiva title: pecaiqr: a model for infectious disease applied to the covid- epidemic date: - - journal: nan doi: nan sha: doc_id: cord_uid: vrb law the covid- pandemic has made clear the need to improve modern multivariate time-series forecasting models. current state of the art predictions of future daily deaths and, especially, hospital resource usage have confidence intervals that are unacceptably wide. policy makers and hospitals require accurate forecasts to make informed decisions on passing legislation and allocating resources. we used us county-level data on daily deaths and population statistics to forecast future deaths. we extended the sir epidemiological model to a novel model we call the pecaiqr model. it adds several new variables and parameters to the naive sir model by taking into account the ramifications of the partial quarantining implemented in the us. we fitted data to the model parameters with numerical integration. because of the fit degeneracy in parameter space and non-constant nature of the parameters, we developed several methods to optimize our fit, such as training on the data tail and training on specific policy regimes. we use cross-validation to tune our hyper parameters at the county level and generate a cdf for future daily deaths. for predictions made from training data up to may th, we consistently obtained an averaged pinball loss score of . on a day forecast. we finally present examples of possible avenues for utility from our model. we generate longer-time horizon predictions over various -month windows in the past, forecast how many medical resources such as ventilators and icu beds will be needed in counties, and evaluate the efficacy of our model in other countries. we used the county-level cumulative and daily death reporting from the new york times [ ] , as well as the county-level active cases reporting from johns hopkins university [ ] . we also used county-level population statistics from the american community survey and county-level policy dating information from johns hopkins university. specifically, we loaded the policy dating information regarding the implementation of the stay at home orders. we loaded these data into a data frame by matching the county fips codes from each source, and computed a moving average death statistic as an additional feature, using a window of size three days. we dealt with missing values in the fips codes and death reporting by removing the corresponding observations. the only exception to this was the county , which represents new york city. this county had a missing fips code, but had well reported data. since new york city is the most active covid- hotspot in the united states, its manual inclusion was necessary. in addition to the aforementioned data sources, we experimented with mobility data, but this did not make it into our final epidemiological model because there was no simple mapping to any of the model variables, and we believed the policy dating information was sufficient to establish distinct regimes in the data. the traditional sir epidemiological model breaks up a regionâĂŹs population into three separate groups: susceptible, infected, and removed [ ] . the primary flaw of applying this model to the current covid- pandemic is the poor approximation of assuming that all people in each group have uniform experiences. clumping the population into only three groups is an example of omitted-variable bias, excluding major realities caused by the scale of the pandemic. the pecaiqr model that is described below takes each group in the sir model and breaks them down into another level of classification. the variables in our model are developed following simple logical arguments based on current global realities and widely accepted scientific and epidemiological results. we first break down the susceptible class of the sir model. due to policies implemented in the us, many individuals have been self-quarantining [ ] . it is sensible then that each day only a fraction of the susceptible population are exposed to potentially getting the virus from the outside world. following this logic, our model breaks up the susceptible population into three classes: protected, exposed, and carriers (explained in detail below). next, we consider the infected group of sir. there is evidence from the large data sets gathered in south korea and other well-respected global scientific efforts that a significant fraction of people infected with covid- do not display visible symptoms [ ] . using this line of logic our model breaks the infected group into two classes: asymptomatic and infectious. an important note is that we assume these asymptomatic and infectious people still actively participate in the community, enabling them to come in contact with exposed people. finally, we are left with the removed group. our model incorporates two lines of logic in the breakdown of this group. first, it is sensible that a portion of the members of the asymptomatic and infectious classes end up self-quarantining (at least in effect) as a result of: getting tested, showing initial symptoms, or having an âĂIJintuitionâĂİ they contracted the disease. second, much of the evidence about covid- so far shows that an individual who contracts the disease cannot get the disease again (at least on the order of a few months) [ ] . following these arguments, the model breaks the removed group into: quarantined and removed classes. removed further has subclasses of dead people and recovered people, who are assumed to be immune from the virus. each of the variables we describe change continuously with time. we describe the variables for a given time t (associated with some given day t ). protected : people who did not go outside to expose themselves to any infected individuals (asymptomatic or infectious) on t, though they have a chance of coming in contact with a carrier. any person in protected also has a chance of joining the exposed class at a later time t + δ (interpreted as the next day: t + ), as they may wish to travel outside that day for any reason. exposed: people who went outside on t and therefore had a chance of coming into contact with an infected individual (asymptomatic or infectious) to become a carrier. any member also has a chance of joining the protected class on the following day t + δ, as they may wish to self-isolate the next day for any reason. exposed : people who went outside on t and therefore had a chance of coming into contact with an infected individual (asymptomatic or infectious) to become a carrier. any member also has a chance of joining the protected class on the following day t + δ, as they may wish to self-isolate the next day for any reason. carrier : people in the model in a purposely temporary position. they are people in the exposed class who came into contact with an asymptomatic or infectious person and got the disease âĂIJon their handsâĂİ to some degree. carriers returning home on t have a chance of spreading the disease to some protected people living at their home, making those people either asymptomatic or infectious on t+δ. carriers at time t can also have a chance to either âĂIJtouch their faceâĂİ and contract the disease themselves (becoming asymptomatic or infectious on t + δ), or âĂIJwash their handsâĂİ and return to being a member of the exposed class on t + δ. asymptomatic: people who were infected by covid- and are contagious on t, but show no symptoms. these people are assumed to be active in the public, in that they have a chance to spread the virus to exposed people in public areas on t. an asymptomatic person on t + δ has a chance of becoming a member of infectious on after showing initial symptoms, a member of quarantined after somehow finding that they contracted the virus, or a member of removed after having the disease pass through their immune system. here, all the people going from asymptomatic to removed would go to the recovered subclass. infectious: people who were infected by covid- , are contagious on t, and show symptoms. these people are assumed to be active in the public, in that they have a chance to spread the virus to exposed people in public areas on t. an infectious person on t + δ has a chance of becoming a member of quarantined after figuring that they contracted the virus and a chance of becoming a member of removed after having the disease pass through their immune system. when these infectious people become a member of removed they have a chance of dying, to become a member of the dead, or living and becoming a member of recovered. quarantined : people who were originally asymptomatic or infectious and subsequently removed themselves from the public and are self-quarantining on t . these people are assumed to not be able to spread the disease to any people, so this group includes people who have covid- but are in a non-contagious stage. a quarantined person on t + δ has a chance of becoming a member of removed. when these quarantined people become a member of removed they have a chance of dying, to become a member of the dead, or living and becoming a member of recovered. removed : people who were originally asymptomatic, infectious, or quarantined, and had the disease fully pass through their immune system on or before t. these people belong to one of two subclasses: dead and recovered. recovered people are assumed to have survived the disease, and will not get infected again. to fit the deaths data to the system of differential equations in the pecaiqr model, we performed numerical integration using the scipy odeint package [ ] , and traversed the parameter space to find a set of parameters that minimized the least squares error of each fit variable in relation to its observed variables. due to the size of the parameter space, this requires an initial guess for the parameters and the initial conditions of each of the pecaiqr variables, as well as defined ranges to restrict the size of the parameter space. of course, we can simply examine the entire logical parameter space (with allowed values ranging from to ) for the each of the parameters, and for each of the pecaiqr variables as well, and then use a random guess within that space to initialize the least squares minimization. however, this is unnecessarily inefficient. the actual parameters may differ widely from county to county, but they all share a similar order of magnitude. a better approach is to use this exhaustive, unconstrained search only once, on a relatively mature curve like lombardi in italy or new york in the united states, to extract a reasonable guess for these orders of magnitude, and then simply feed in this guess for the other counties as well, to provide a more logical starting point in the least squares minimization. due to the complexity of our model, however, there are a few more nuanced details that we will discuss later in section. first, we will elaborate more on the fitting procedure. initially, we only fit the death variable (d) to the observed death reporting. early on, we decided not to fit our infection curves to daily case statistics, as they have inconsistent and unreliable reporting, so this would only diminish the accuracy of the fit to observed deaths, which is a much more reliable statistic. however, later on, we realized that we could also effectively fit the quarantined variable (q) in the pecaiqr model to the active cases reporting. the intuition behind this is the assumption that those who test positive would either self quarantine at home or be forcefully quarantined in a hospital if their condition is severe enough. of course, this does not capture all the self-quarantined individuals, so we permitted a large degree of fuzziness in the fit of the quarantined variable (q) to active cases. we achieved this with a bias scaled factor that gave much more weight to observed deaths in the fit of the death variable (d). this causes the model to prioritize the deaths fit over the active cases fit, so the active cases fit becomes more of a suggestion rather than a constraint. the hope is that the least squares error on the active cases is not large enough in magnitude to compromise the deaths fit and force the parameter space into a different minimum, but will rather provide a subtle correction around the local minimum discovered by the minimization of the least squares error on the deaths. unfortunately, a fit to active cases is not helpful for most counties, as most counties do not maintain their active cases reporting well, and moreover we discovered that the criteria for an active case may vary widely across different states. we also realized that the pecaiqr epidemiological model parameters are not static, due to the dynamic and rapidly evolving state of the pandemic, caused mainly by external forces such as social distancing protocols and lock down policies. a naive fit on the data would only yield some sort of average of the parameters. but since we only care about the most recent characteristics of the death and infection curves when making predictions into the future, we can do better than this by affording the more recent data points more weight in the least squares error calculation. this forces the minimization to favor a solution that fits more heavily on a more recent window of time, while still retaining the effects of the past data to some degree. we can achieve this with a geometric progression of weights, as well as a bias term that sets a maximum weight for all data points before a certain cutoff. we called this method training on the tail. another method to separate more recent parameters from the data is to use the assumption that there are distinct parameter regimes correlating to the start of policies -the stay at home orders in particular. note that these policies directly affect the infection curves because they limit the spread of disease. for the death curve, there is typically an offset between the date of policy implementation and the date at which the effects become apparent in the death data, this offset approximately equal to the average time till death for covid- . with this assumption, we separate the deaths data into two training regimes -one for dates before the policy implementation plus the offset, and one for dates after the policy implementation plus the offset. then, we train one fit on the the first regime, and feed the fitted parameters and predicted variables on the date of the policy implementation as the guesses and initial conditions, respectively of a second fit on the second regime. we called this method training on the policy regime. we perform procedures described above on a county-level, and then use the fitted parameters with the numerical integration to extrapolate into the future to get county-specific predictions. note that the predictions and their errors are in cumulative deaths, but for the sake of visualization, we converted to daily deaths later. to get the errors, we initially used a method that calculated error bounds by finding the parameter variance from the residual variance using the covariance matrix of residuals and the jacobian around the fitted parameters. having obtained the mean and standard deviation for each of the parameters, we assumed that each parameter had normally distributed values, and so we sampled parameter sets. we could not simply apply calculate the parameters for each confidence interval from their mean and standard deviation because it is not obvious which parameters are positively correlated or negatively correlated with the deaths predictions. this method turned out not to be ideal in all cases, as there are certain regions in the parameter space that are invalid and cause the error bars to spike. we then developed a bounding method that allowed a reliable way to accurately tighten our confidence intervals. this method infers the error bars from the deviation relative to the predicted fit, of a smoothed version of the residuals calculated from the moving average of daily deaths, which is equivalent to the moving average of the slope of the cumulative deaths. generate a pdf of deaths around the best fit prediction. for each time t in the training range, compute the difference in âĂIJslopeâĂİ between the fit and the actual data. the slope in the fit is simply the current predicted death minus the previous death on the fit curve. for the slope in the actual data, to mitigate the dominating effects of outliers, we find the slope as the difference between consecutive points of the moving average (window of days) instead. we then define a normal distribution of slope ratios. for each time t in the training data, we find the ratio of the actual (moving average) slope and the fit slope, and fill a list of these ratios. next we find the mean and standard deviation of the distribution as the mean and (sample) standard deviation of the ratios list. outliers, which we remove, are defined as values that are above three standard deviations from the mean. this is necessary because the ratios are unrealistically high in the small number limit. for each time t in the extrapolated prediction fit we generate a confidence range of points. we find the slope at the extrapolated time as the predicted death minus the predicted death at the previous time. this slope is multiplied by a random scaler sampled from the normal distribution of ratios and denoted as s. we multiply the predicted death at the previous time t − in the fit by + s and add this as a point in the pdf at time t. this multiplication is repeated times so that points are generated. we get the discrete , , , , , , , , cdf percentiles by sampling the pdf. the methods described in the past two subsections are implemented as options that can be activated with hyper parameters, and collectively they provide several different ways to fit the pecaiqr model and generate the confidence intervals. due to the time constraint of the competition, we were not able to develop a sophisticated blending method to optimize on the hyper-parameter space. however, we were able to sample a few different combinations of hyper-parameters and determine on a county-level which one works the best for each county, by modifying the evaluation script provided by the tas. this script essentially computes pinball [ ] loss for the submission file of predictions, scored against the most recent data. to determine a good set of hyperparameters for each county, we simply trained using a two week cutoff in the data, and scored the predictions for the subsequent two weeks using the evaluation script. for predictions made from training data up to may th, we were able to obtain an average pinball loss of . for all counties with scoring on a day forecast. our greatest weakness was the lack of a second working model with which we could cross validate, as well as the lack of a sophisticated blending method to optimize on the hyper-parameter space for the single model. we were not able to develop an alternative working model due to the time constraints of our group members, but a detailed description of our attempts is available in section failed models. having two different models would have allowed us to mitigate their individual weaknesses and account for their individual edge cases with the other's strengths. the epidemiological model has several major weaknesses. although it works well for counties with well recorded data, it fails for the vast majority of counties, which have low or noisy death statistics. we called these counties the "non-convergent counties" because the epidemiological model was not able to converge to a parameter set that yielded predictions which could consistently score better than the naive all zeros prediction. the curves for the pecaiqr variables in the left column of the preceding figures make sense intuitively. we expect expect to see that conversion between the protected and exposed populations, and that the removed population to eventually dominate. we also expect the carrier population to peak before the infected population, which should have a similar shape to the asymptomatic population, and lastly we expect the infected and asymptomatic populations to peak before the deceased population peaks. lastly, the second figure, which uses the "fit on tail" method, more closely captures the uncertainty of the tail end of the curve, as expected. (a) (b) (c) (d) (e) (f) (g) (h) for the "nonconvergent counties," we attempted to skip over the parameter fitting step by guessing parameters based on the parameters of similar counties. in order to do this, we need to first establish proof of concept that there is some correlation between the non-covid features of a county and its pecaiqr parameters. so we visualized the parameters for all the "convergent" counties by using dimension reduction algorithms. we used principal component analysis [ ] , singular value decomposition [ ], and t-distributed stochastic neighbor embedding figure : several different visualizations of the parameter space using dimension reduction algorithms. each point represents a count, and its size correlates with the county population. for svd, we plotted the first two columns of the left matrix against each other, which produces an visualization how similar the counties are by distance, based on their abstract preferences in the parameter space. although we did not have time to further explore the possible correlation between parameter clusters and non-covid county-specific features, we were excited to see that there are some distinct intrinsic patterns in the parameter space. further research would involve classifying these clusters with a clustering algorithm, and then attempting to predict the cluster labels or quantified distances of specific counties with a regression algorithm from the non-covid feature space. we also realized that, even for the larger counties, there was some uncertainty caused by the complexity of our model. since our parameter space has so many dimensions, there is an issue with degeneracy. the least squares minimization settles on a parameter set that yields a minimum in the least squares error metric, but this is not necessarily the only minimum or even the best minimum. we were convinced of this when we discovered that for each county, we could get distinct parameter regimes with different initial guesses, and that these solutions are similarly valid. we will show this in the figures below, by repeating the procedure from the previous section, but with a different set of parameters as the initial guess for the least squares minimization. following these observations, we hypothesized that there could be better parameter guesses that we have not encountered. in order to get better parameter guesses, further research should focus on deriving reasonable values for the more intuitive parameters of the model, such as those concerning the rate of infection, from existing data. this also illuminates the issue of ode stiffness. we believe the functionality of the scipy odeint package is quite limited. given more time, we would explore other statistical packages like stan, which has a state of the art implementation of th order runge kutta numerical integration for stiff odes [ ] . another improvement would be to use a numerical integrator that solve handle delay differential equations, which would allow us to have more control over the time delay in expression between infected and death states. we believe that overall the pecaiqr model is very promising, and reveals the benefit of attempting more ambitious, complex epidemiological models. the epidemiological model's differential equations establishes intuitive rules by which it operates, so it has more long term predictive power than most other models. the shape of the solutions of the pecaiqr are also quite interesting, as they resemble heavy tail distributions similar to the frechet/weibull distribution. the heavy tail is especially important for pandemic forecasting, as we expect the daily deaths to fluctuate around a low value for some time, instead of decaying immediately to zero. in this respect, the tail end of the curve is almost stochastic in nature, once the infection curves lose enough momentum. due to the high variability found in the data for reported deaths for each county, with some counties reporting unrealistic jumps in the numbers of deaths, it seemed sensible to try using a stochastic model to make some predictions of death counts for counties. in order to best preprocess the county data for training, we explored possible clustering methods that would be able to cluster time series of varying lengths. dynamic-time warping (dtw) was picked to be the distance measure of choice as it is able to compare time series of different lengths and provide a âĂIJwarped distanceâĂİ between each series of daily reported deaths. the daily reported deaths were preprocessed by first removing the series with all âĂŹs, and then performing a z-normalization on each data series. the data series were then compared with dynamic-time warping, and a hierarchical clustering was developed based on the dtw distance matrix. the number of clusters was determined using the elbow method. the series that had all âĂŹs are then added back into the clustering list, with cluster id âĂŸ âĂŹ. it was desired that further clustering be done, specifically clustering involving hmm-based clustering. the ideal setup would be a clustering based on an iterative dtw-hmm clustering algorithm, in order to fully extract similarities between county reported deaths. unfortunately we did not have time to fully develop this idea. once the clusterings were developed, each cluster was used to train an hmm model. we used hmmlearnâĂŹs gaussianhmm model in order to have the gaussian emissions needed for this kind of data (a multinomial model with discrete output would be stretched thin with ∼ states). the number of states each gaus-sianhmm was given per cluster was chosen by doing a modified version of the elbow method, to ensure that the hmm is complex/simple enough to match the variance in reported deaths. once these hmms were trained, we developed a method to initiate an hmm close to a particular starting emission and allow it to generate subsequent emissions, to make -day predictions for the counties in the cluster. these predictions could be made thousands of times for each hmm, allowing each cluster to effectively create a probability distribution of predictions. further work needed to be done to make this stochastic model effective. as the model only allowed for predictions based off of an entire cluster, there needed to be a secondary layer of scaling to allow a cluster prediction to be mapped to each individual county. an idea of attempting to do some time-series comparison and do some sort of series stretching/scaling was developed, but was never fully formed. there are definite rooms for improvement and optimization in this model. one fundamental issue is that perhaps the small number of states of the gaussianmm limits the complexity of each hmm per cluster. while this is true, this level of simplicity mixed with stochasticity could capture the random element seen in the reported deaths data. if this random element is able to be removed or smoothed away,however, perhaps this model would not be so useful any longer. when attempting to use statistical regression to model nonlinear correlations in data, a common approach is to employ a bayesian non-parametric strategy, such as the gaussian process. bayesian non-parametric strategies like the gaussian process are essentially extensions of bayesian inference on an infinite-dimensional parameter space. very loosely, this allows us to model the data as the combination of many different gaussians (each quite accurate in its local region), stitched together to create a single model [ ] . this technique was attempted in the latter stages of the course to create predictions for counties where the predictions from the pecaiqr did not converge, an issue at the time for counties with poor data. we used a custom implementation of gaussian processes, using a mean function of and an exponential squared kernel. the data used was the rolling average of deaths over a three day window vs time. in this model, there are three parameters: l which is the length parameter for the kernel, σ f which is vertical variation parameter for the kernel, and σ y which is the noise parameter. for each county, the optimal hyperparameters for that county were found by searching for the set of parameters within a given range that minimized the error. here, the bounds on the parameters . ≤ l ≤ . , . ≤ σ f ≤ m . , . ≤ σ ≤ m . where m was the maximum value of rolling three day average deaths over the past days. the error function was the root mean squared error (rmse) over the last days. these bounds and the error function were determined with validation procedures. the gaussian process seems to fit quite accurately in the short term, but the predictions quickly drop to zero, as demonstrated in the plots above. unlike the pecaiqr model, it does not retain a heavy tail. therefore, gaussian process does not have long term predictive power. we can possibly improve the gaussian process and overcome the issue by setting a custom mean instead of using the default zero mean, which likely contributes to the rapid decay of the tail. we did attempt to so, inspired by the pymc tutorial posted in the cs b piazza [ ] . it was attempted since a custom mean function could be used [ ] ; similarities were noted between the graph of the number of deaths vs time in new york county and the weibull distribution, so the weibull distribution was this custom mean. however, it was not pursued further since it could not successfully run -when training the model on just one particular county, the program timed out and crashed jupyter on the computer it was run on. since the final deadline was already quite close, this attempt was abandoned. given how this model can fit to a custom mean, if it was attempted earlier it might have proved to be useful. we verify that the issue of a rapidly decaying tail is not specific to county as shown above. we can conclude that the gaussian process fits very well on the training data, but fails to present long term predictive power. again, these failures in the gaussian process model may be overcome by setting a custom mean. but we believe that the epidemiological model, though less accurate, has intrinsic advantages that the gaussian process cannot match. this is because the epidemiological model has some sort of intuition in the form of the rules established by its differential equations, while the gaussian process is purely curve fitting. perhaps, given more time, we could have combined these two models to utilize both the short term accuracy of the gaussian process and the long term predictive power of the epidemiological model. analysis of model predictions for different cutoff dates in the training data shows that the model is quite stable and consistent when predicting on the region past the peak. however, predicting before the peak is much harder, as we are no longer operating with the assumption that the infection curves are dying down. in sub-figure d) we see that the peak daily deaths value predicted by the model is significantly less than the actual peak that is revealed with more data. however, the location of the peak is correct. we realized that the training data at this early cutoff is largely dominated by data in the regime before the first stay at home order, but we were training using an initial parameter guess that accounted for the effects of a stay at home order, and so in sub-figure e), we tried a different set of parameters for the initial parameter guess in the fitting procedure, inspired from the parameters obtained from a fit on similar early curves in italian regions. this modification yielded a more accurate prediction of the peak daily deaths value, but a less accurate placement of the peak. the prediction curve becomes extended when we use this alternative parameter set because the model no longer assumes that there will be a stay at home order, and therefore the curve will not flatten to the same degree. analysis of model predictions for different cutoff dates in the training data shows that the model is quite stable and consistent when predicting on the region past the peak. one thing to note is that sub-figure a), trained on the most recent data, seems to have a much longer tail. this is a result of the high level of noise in the more recent data points, which are not included in the other cutoffs. predicting before the peak is much harder, as we are no longer operating with the assumption that the infection curves are dying down. in sub-figure d), we see that the fit is quite different from the fits with later cutoffs. the data before this cutoff is quite noisy, so the model cannot accurately predict when the infection curves will begin to die off. to fix this, in sub-figure e) we fit the model on active cases as well. the model is able to use the active case statistic to determine that the infection curve dies down earlier than it otherwise would predict. here, we fit each of the three counties using the policy regime method described in section . fitting. the faint gray curve shows the predicted infection curve on the data regime before the date of the stay at home order, plus the time of death, and the dark gray curve shows the predicted infection curve on the data regime after the stay at home order. in counties and , the stay at home order seems to have flattened the curve, but in county the effects are more ambiguous. the is likely relates to factors specific to new york that worsened the outbreak, such as the high population density of new york city, and the shortage of hospital resources later on. the model can be easily adapted to train on international data. at the bare minimum, the model only requires death reporting and population. the predictions made from the may th cutoff, roughly a month before the june rd cutoff, shows that the prediction fit is quite stable and is consistent. (a) (b) (c) (d) (e) (f) figure : the first column of figures shows daily deaths and currently hospitalized, daily hospitalized, and currently in icu plotted, respectively, plotted against time. the second column shows the linear regression for daily deaths vs currently hospitalized, daily deaths vs daily hospitalized, and daily deaths vs currently on ventilator, respectively. to smooth out the data, we used a moving average with a window of data points. we then aligned the peaks for each hospital statistic to match the peak of the daily deaths, in order to account for the offset term that corresponds with the average time between hospital admittance and death. the first column shows the hospital statistics before they were aligned with the death statistic, and there is clearly an shift, although only by a few days. this makes sense, as patients who are admitted to the hospital are likely patients who have already developed a severe condition. note that the hospital statistics and the death statistics are both gaussian. this suggests that we can find some scaling factor once we align their peaks. for this, we perform a linear regression against the death statistic, revealing a definitive linear correlation for all the statistics. note that there is a slight nonlinearity in the regression between currently hospitalized and daily deaths, as the daily deaths increases beyond a certain point. this may indicate that we are nearing the hospital capacity, and so the change in number of currently hospitalized patients begins to lag behind the change in daily deaths. also note that there is an extreme outlier in the regression between currently on ventilator and daily deaths at the peak daily deaths value. similarly, this is also likely caused by some limit on the number of ventilators available. indeed, there seems to be a ceiling to the number of people on ventilators past a certain number of daily deaths. using the linear regression fit, we can now directly convert any of our model predictions, as well as their confidence intervals, to predictions for hospital resources. we noticed that the vast majority of counties have inconsistent reporting of deaths. some counties even had cumulative death statistics that decreased on certain dates. clearly this is not possible, as death is permanent. other counties report constant values for cumulative deaths (zero values for daily deaths) for an extended period of time, followed by a quick spike. it is doubtful that all deaths suddenly occur in a single day, so this observation suggests that the deaths reporting might not be distributed correctly, perhaps due to administrative lag. for counties with low numbers of deaths, this creates large amounts of variance in the data, which makes it hard to fit. it seems like the deaths reporting also seems to be correlated with the day of the week. for many counties there is an interesting trend that the daily deaths increase over consecutive days before the weekend. again, this might be due to administrative lag in the hospitals that report deaths. the active cases statistics also seem to be poorly recorded, and this makes sense given how ambiguous the classification of an active case can be, especially with the limitations on testing. many counties completely lack useful active case statistics, and in other cases, it is only available very late into the infection curve. in general, statistics that attempt to quantify the number of cases, whether active, cumulative, or daily, are intrinsically unreliable, as they depend on the availability of tests, which can vary greatly over time. perhaps a more reliable statistic is the ratio of positive tests to administered test on any given day. another issue was with the non time-series data. some files, such as the age race file from [ ] and the aggregate jhu file from [ ] , had valuable information but also numerous holes. while this data was not used in creating the predictions, it was used heavily in section creative data visualizations which details an attempt on clustering based on categories in these datasets. such holes proved difficult to fill in. the filling in was done by using the data from the nearest county if a given entry was missing for a certain county. however this could be ineffective if this certain county and its nearest neighbor are very different in nature, as then the data for this certain county would not be particularly representative. while the data holes were not particularly impactful for the models described above, they certainly could be for models that took many non time-series features from those datasets into consideration. a contribution to the mathematical theory of epidemics unique epidemiological and clinical features of the emerging novel coronavirus pneumonia (covid- ) implicate special control measures positive rt-pcr test results in patients recovered from covid- table? q= american\ % community\ % survey\ % \ % acs\ % \ % \ &tid= acsdp y . dp \ &y= \ & vintage= key: cord- -kc t y o authors: ray, soumi; roy, mitu title: susceptibility and sustainability of india against covid : a multivariate approach date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: kc t y o purpose: we are currently in the middle of a global crisis. covid pandemic has suddenly threatened the existence of human life. till date, as no medicine or vaccine is discovered, the best way to fight against this pandemic is prevention. the impact of different environmental, social, economic and health parameters is unknown and under research. it is important to identify the factors which can weaken the virus, and the nations which are more vulnerable to this virus. materials and methods: data of weather, vaccination trends, life expectancy, lung disease, number of infected people in the pre-lockdown and post-lockdown period of highly infected nations are collected. these are extracted from authentic online resources and published reports. analysis is done to find the possible impact of each parameter on covid . results: covid has no linear correlation with any of the selected parameters, though few parameters have depicted non-linear relationship in the graphs. further investigations have shown better result for some parameters. a combination of the parameters results in a better correlation with infection rate. conclusions: though depending on the study outcome, the impact of covid in india can be predicted, the required lockdown period cannot be calculated due to data limitation. the entire world has almost stopped theoretically in the month of march . this is one of the most unexpected and unbelievable situation in world's history. a virus, starting its journey from wuhan city of china in december , has now reached almost all major cities and has created colonies very rapidly. as per world health organization (who), the first case of covid was identified on th december [ ] . initially the disease was misunderstood as some variation of influenza. scientists, researchers and doctors, after doing continuous analysis, then came up with information about this novel corona virus. though the drug is not yet in the market, the structural details of the virus are now known to us [ ] . because of its similarities with the behavior of severe acute respiratory syndrome (sars) corona virus, this virus was named as severe acute respiratory syndrome coronavirus (sars-cov- ) and the disease was identified as coronavirus disease (covid ) by who [ ] . seasonal diseases which have higher mortality rate usually belong to sars category. observing the high infection rate, on th march , who declared covid as pandemic [ ] . in , asian countries were badly affected by sars epidemic which originated in china. the worldwide death toll was , having a ratio of : of registered cases [ ] . in , asian flu virus claimed around thousand lives in india [ ] . one of the severest global pandemic in the history of recent past was flu. this flu was first reported in a spanish newspaper, and it infiltrated india through bombay port. the disease was contagious. it claimed million lives worldwide [ ] . it is claimed that one third of the population was infected. different articles claimed china as the origin of the flu [ ] [ ] [ ] . but death in china itself was very few, whereas india lost almost one fifth of its population [ ] . the estimated death toll was million [ ] . the flu was so deadly, that it brought the population down for first time as well as the last time, till date in the history of india. but the virus disappeared almost suddenly after few months. it is assumed that like any other pathogen, this virus also rapidly mutated to a lesser lethal strain and then finally died out [ ] . covid has many similarities with these flu and sars. all are viral and contagious infections, epidemic in nature, turning into pandemic within few weeks, transmitted through droplets and very easily transmissible from human to human. it is also suspected that coronavirus can be transmitted through air and it can survive in environment without any decrease in its efficiency for a long time and thus the chance of infection increases [ ] . if coronavirus is transmissible and airborne, this is an alarming situation for india. normally during season change, indians suffer from different common ailments like cold and cough, nasal congestion, conjunctivitis etc. many of these diseases are infectious and transmissible. a large part of the population suffers from one or more of these issues commonly. along with those common epidemics, covid has to be taken care of. due to nor'westers during march/april the chances of fast spreading of covid is also high. many people are already infected, among which all are not having significant symptoms and hence not identified. finding ways to stop community transmission, case identification at initial stage and controlling the death rate is an emergency. even the cure will not be easy to save the world if the infection is not prevented. faster and easier worldwide transportation system has turned into a curse in case of viral epidemics. due to international travels, almost all the countries in the world have got infected by this transmissible virus within a very short duration simultaneously. covid has affected countries and territories around the world as of th april . from the information of the impact of pandemic in india, we have tried to understand the underlying facts. a large population who lived near and below poverty line got affected by the pandemic [ ] . apparently, it seems that, the sanitation had a significant relation with the disease infection, spread and severity. but the high mortality rate in case of covid even in developed countries raises a question on this easy assumption. light from a different angle may have some answer to it. people from lower economic segment not only failed to maintain sanitation but also suffered from improper diet. lack of consumption of proper and healthy food weakened their immunity. before , no vaccine other than for smallpox was available. no vaccine or antibacterial was invented to prevent the pandemic diseases. it is not very difficult to assume that the practice of vaccination among the poor people in india was also low. so, hygiene was not solely responsible for the devastating death toll of . immunity also played a bigger role in it. in this article, we have tried to give some insight from available worldwide information, in order to understand the nature of this new coronavirus infection which is causing covid . we have tried to inspect its dependencies on other known parameters. a measure of the possible effects of different parameters on the outbreak and infection growth has also been estimated in order to understand the risk in india. the scientists, from different parts of world, already have done researches and have published valuable information. all the important outcomes have been considered and examined before concluding our findings. a common drawback is associated with most of the reported works. they have discussed impact of single dimensions like temperature, vaccination or life cycle of virus. this has restricted the scope of understanding of the virus's overall activities and limited the chance of prediction and prevention. we have tried to overcome this limitation by replacing univariate analysis with multivariate approach. this article has considered different possible aspects to get a robust outcome. to keep the result as unbiased as possible, we have collected data of multiple cities and countries all over the world having different geographical locations and climatic conditions. we have conducted analysis of several relevant factors to look into the situation from all possible corners. possible dependency of the number of total infection and death has been examined against the environmental conditions taking different weather parameters as independent variables. for this purpose, data of cities all over the world has been considered. these cities are significantly affected by sars-cov . we have collected the covid related data from who site and other data from particular websites for each individual type of parameter to minimize the biases. the duration considered to check dependency of weather parameters is from st to th march, as because by st of march, a large number of countries got significantly infected. the days have been limited to th , as by rd week of march majority of the countries applied social distancing and isolation. data beyond that time may have significant impact of isolation. isolation includes noise in the measure of infection and death, as it puts restriction in virus transmission due to lower availability of hosts. the weather information of all those locations have been collected from a single website [ ] to keep it uniform even if the information includes any noise or bias. we also have taken a measure of the population to compare the infection rate. human to human infection depends on the population for transmission. other than these affecting parameters, another checking has been done on the impact due to the lockdown. how the duration of lockdown has been affecting the number of new infection, have also been examined to understand its importance. we have considered life expectancy also to inspect its impact on the number of infected cases and deaths. the life expectancy includes the impact of different parameters like average living standard, socioeconomic situation, health service qualities, natural calamities etc. a comparison with life expectancy refers to be have a relation with all those hidden parameters though the insights of each are not accessible. the data is collected from united nations development program reports [ ] . as proposed in a paper [ ] , vaccination may have great impact on death. in this article, the data of bacillus calmette-guérin (bcg) vaccination has been compared with present death rate of different countries. this observation has a significant impact in our final outcome. this data is collected from who and review articles [ ] [ ] [ ] . the additional factors included in this study are the impact of lung cancer (lc), chronic obstructive pulmonary disease (copd) and lower respiratory infect (lri). these diseases have shown an impact on death rate in many countries which are badly affected by coronavirus. the required data are retried from online resources [ ] . . cc-by-nc-nd . international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . the total number of cases per million and total death per million are very much correlated as per our examination (ρ= . , p= . e- ) as of th april . because of the good correlation, any of these two parameters can be used for prediction analysis and the outcome will remain comparable. in this article we have used either of these parameters to understand the impact of all other parameters on covid and the results are compared later. we have divided our test result into different parts. in the first part we have discussed the impact of different weather parameters on the number of infected cases. we have not considered active cases because the number is ever changing with continuous addition of new cases, elimination of recovery numbers and deaths. initially the number of infected cases increased rapidly due to lack of awareness, availability of more hosts, free movement of hosts etc. and then gradually decreased for enforcement of quarantine, growing awareness like washing our hands regularly, social distancing etc. these qualitative parameters are not traceable but have significant impact on the number of new cases. different cities, selected in this article, have different geographical locations with widely varying atmospheric condition. if any of the considered weather parameters have a significant effect on infection transmission, then that will impact the transmission equally, irrespective of the location. the list of the cities with parameters details is given in the appendix. we have taken data per million to nullify the population bias. as the virus can be transmitted from human to human and can travel a small distance through droplets with airflow, higher population increases the availability of new hosts and hence the rate of infection. our target is to inspect the significance of that impact. this study is very important for india as its population is very high. the correlation between number of registered infected cases per million and the different weather parameters like temperature measures, humidity, dew points and precipitation have been found. we have used pearson's correlation to check the dependency of identified case numbers with the other parameters. we have used linear regression correlation to understand the linearity in the relationship, its statistical significance and to make a prediction. the results are presented in tabular form in table . . cc-by-nc-nd . international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . the p-value in each case is high enough to support null hypothesis and reject any considerable correlation. but we decided to inspect further. each parameter is examined separately to find any possible impact of covid deaths. this method offers exciting information about lowest temperature and highest temperature. we have considered death of . % of population as the threshold. above this rate is considered as a matter of concern. as per central intelligence agency, present population of india is , , , [ ] . the . % of this population is about , , which is a few thousand more than present death count all over world and not an ignorable count. the countries which experienced higher death rate (as well as death) had the minimum temperature below °c as shown in figure . the y-axes are changed to logarithmic scale to enhance the datapoint visibility. the linear trend lines are also shown in the figures for visual depiction of our understanding, that is with increase in minimum temperature the death rate and total death (hence infection rate) decreases. on contrary, for highest temperature of a place, a range of the temperatures has shown higher risk of death due to covid . the countries with high death rate had highest day temperature in between to degree centigrade as per figure . we have given a try to understand the effect of lockdown. how this lockdown is impacting the community transmission is an important study. the entire world is depending on this policy in absence of vaccines. a detailed study of this factor will help us to predict the required lockdown duration for india. we have taken the number of new registered cases of different countries at the starting of lockdown and then after completion of each week till th april . we have considered the total change per week to reduce the noise of a temporary change in rate. a sudden change in the number for one or two days without any consistency does not portray any significant change of the overall situation of a country. the result is presented as graph in figure . only norway and italy have entered into the th week of lockdown and both countries have shown a drop in new cases after th week. but the steadiness is which is yet to be known, and it is highly required to conclude in a positive note. italy has completed days of th week on th april . the new cases registered in this duration are , which is a little lower rate than the rate per day of th week of lockdown. austria and australia have shown notable reduction after weeks of lockdown. after th day of the th week, the average per day new cases is still very low in austria. australia has just completed day of next week as on th april. though per day rate is lower than last week but it is higher for new cases registered in each of the last days. though no conclusion can be drawn about prevention of infection through lockdown, a prospective evidence of slowdown in the rate of spread of covid pandemic is available. for a decisive interpretation the impact of lockdown needs to be observed for few more weeks. cc-by-nc-nd . international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/ . / . . . doi: medrxiv preprint life expectancy has no direct impact on infection growth and death due to that. but life expectancy is connected with different facts as discussed in introduction. along with those, depression, low average education and unemployment also have an impact on the life span of a community. to have a measure of all such hidden factors, we have checked the relation between life expectancy and death rate. here we have taken death per million to compare the change in it with respect to life expectancy. the graph is shown in figure . the surprising observation is that, the high death rate is mostly associated with high life expectancy. inspection of causation is required to understand the actual impact of this finding. this is beyond the scope of our present study and can be considered during advance analysis. not all the affected countries are equally prone to covid as per the report of who. though our preceding analyses have failed to find any strong relationship between the discussed parameters and this pandemic, the presence of some factors influencing the rate of spread of this infection and death rate is obvious. a literature [ ] showed some hope in its study of relation between covid and bacillus calmette-guérin (bgc) vaccination. we have repeated that study to find the relation with present scenario. the work was reported in the middle of march when many countries were not as affected as of today. the scenario in usa has changed dramatically in the last weeks; india also has shown significant increase in the number of infected cases in the last days. hence, the repetition of the analysis is necessary before concluding any decision. our study has shown that vaccination has good impact in most of the cases but there are many exceptions too. the comparison is presented in the figure . france, iran, ireland, portugal and sweden have significant death toll even after having good history of vaccination. on the contrary, australia and canada have quite low death rate without vaccination program. hence the hypothesis of any direct relation between bcg vaccination and covid has been rejected. a further, in depth analysis including more details of vaccination program, coverage of the population, specially in the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . below poverty level population, is required to find the exact relation. a low level dependency of death rate on bcg vaccination is visible in the graph which is supported by statistical analysis of correlation resulting in ρ = . and p-value = . . covid is a severe acute respiratory syndrome (sars) disease. it attacks human lungs, creates trouble in breathing and gradually becomes deadly to claim lives. to understand its relation with other lungs disease, a study has been conducted. top diseases of each country are examined to see the burden of other lung diseases. after preliminary scrutiny, we decided to consider lung cancer (lc), chronic obstructive pulmonary disease (copd) and lower respiratory infect (lri). data cleaning and preprocessing was done before analysis. the rank of these diseases which are among the top and primarily responsible for death in a country has been considered for further analysis. if a disease is not listed in a country's top diseases, a numerical value has been considered. the correlation and p-value of these diseases with death rate are ρ = . , p-value = . . the p-value depicts a higher tendency towards null hypothesis though the correlation is average. the relation lc alone is better with death rate. with a ρ = . , p-value = . lc demands a command over death rate due to covid . this finding does not offer any way of prevention of the pandemic but gives an idea about the risk of a country. none of the above discussed parameters has significant effect covid except lungs cancer. though few have shown impact in increase in number of cases (or deaths) in some cities, nothing convincing has been found. it seems like the disease gets transmitted with almost equal potential in different atmosphere. impact of lockdown is also not considerably good to get any definite suggestion. hence the question remains unsolved. how india is going to response to this pandemic? negative minimum temperature, a specific range of maximum temperature, lack of bcg vaccination and tendency of other lungs diseases have shown some positive impact in increasing the number of covid cases and death. we have combined all these four parameters to see their combined effect on death rate. before statistical analysis, we have done preprocessing of each parameter to create four distinguished features. the temperature data are analogous in nature having no significant impact in case of a minute change. using the already acquired knowledge from the previous analysis, we have classified the temperature into two different classes. in case of minimum temperature, values equal to or below °c are considered as one class which has strong negative impact on death and hence represented by . rest of the temperature values which have less impact or no impact on death are considered as . similarly, maximum temperatures between to °c are converted to and rest are to . disease scores are divided into classes - to , to and . score greater than , represents disease rank within top . for these ranks a disease is represented by . scores from to means the disease is ranked between to and rank below is presented by . when a disease is not listed in top diseases of the corresponding country, it has been represented by . this data preprocessing steps are important for better insight. vaccination is represented by number of the years the program is continued in a country since as mentioned before. after creating a clean dataset with these four features, statistical analysis was done to check if any useful correlation is present. this analysis offers a prominent correlation, ρ = . with high acceptance, p = . . the features are plotted in figure along with increasing death rate. in the figure, singapore, australia and france have shown exception in vaccination impact. the probable reason can be the temperature. both singapore and australia are hot countries. australia is not prone to any lungs disease and singapore has low tendency of lc. on the other hand, france has low temperature and high impact of lc in country's death rate. hence, a significant impact of these parameters can be assumed, and this needs further research for definite conclusion. in april, temperature remains significantly high in india (non-hill zones). in most of the areas, specially in the cities which are reporting high rate of cases and death, the lowest temperature remains higher than °c and maximum temperature goes well above °c. india also have different vaccination program for years and bcg vaccination is done for almost % of the population. the negative factor is that the lung diseases are very common in india. copd (rank ), lri (rank ) and tuberculosis (rank ) are major causes of death here, though lc is not that common like other countries which are badly affected by covid pandemic. this information and data dependent statistical analysis is not self-sufficient to understand the nature of coronavirus. along with this geographic, demographic and meteorological analysis, information from other branches of science like virology, biotechnology must be considered. an important finding about such pandemic is their sudden disappearance after few months. it happened every time in the world's history. not considering the expected life of sars-cov with respect to environmental conditions and continuous mutation will be impractical. the faster the virus will selflimit itself, the earlier the rate of infection will go down. india has imposed quarantine through locked down for days starting from rd of march . in last days the number of cases as well as death has increased significantly. the lockdown do restrict the community transmission but the rate is increasing may be because of detection of already infected cases. fast identification of old cases is required to access the effect of isolation on infection transmission rate. the mortality and morbidity ratio may be affected by the immune system of a population and will vary with different life style factors starting from food habits, common diseases, vaccination programs etc. in the high altitude areas (mainly the himalayan region) with low temperature throughout the year, the risk is higher as per our analysis. february, march are not tourism season of these himalayan region of india because of chilling cold, road blockage due to snowfall and for academic session ending with examinations. probably due to little tourist flow, these regions are still not reporting cases of covid . once the severity of pandemic will fall and the country will start resuming its normal life, the free movement of tourists can be a threat for those hilly areas and it could ignite a reappearance of the disease in cities too through the returned tourists. a proper protection plan and . cc-by-nc-nd . international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . restricted movement for reasonably longer time after the pandemic is required to keep the citizens safe. to summarize the risk of india with more accuracy, we have to consider the socioeconomic condition too. a good vaccination program, history of having seasonal endemics ensures better immunity against similar diseases. sars-cov- is a new virus and hence the old antibodies cannot completely prevent it. but whether any antibody has any significant impact on its growth or not is under research. cyclicity is ubiquitous for acute infectious diseases [ ] . each disease is unique on its own. the sars diseases occupy the span of two months, march and april, of indian epidemic calendar. the reason is that the seasonal variation has significant impact on transmission of infectious diseases. this is known as seasonal forcing. if the vaccine of tuberculosis is resistive for covid , then the already present antibodies in the blood will reduce the severity of the disease and death rate in india. the epidemiological characteristics of an outbreak of novel coronavirus diseases (covid- ) in china features, evaluation and treatment coronavirus (covid- ). in: statpearls [internet covid- )-and-the-virus-that-causes-it] . who director-general's opening remarks at the media briefing on covid- - sars (severe acute respiratory syndrome the pandemic of influenza in india did the - influenza pandemic originate in china? population and development review what happened in china during the influenza pandemic? flu pandemic that killed million originated in china, historians say flu experts warn of need for pandemic plans mortality from the influenza pandemic of - : the case of india aerosol and surface stability of sars-cov- as compared with sars-cov- estimation of potential global pandemic influenza mortality on the basis of vital registry data from the - pandemic: a quantitative analysis correlation between universal bcg vaccination policy and reduced morbidity and mortality for covid- : an epidemiological study connecting bcg vaccination and covid- : additional data global, regional, and national burden of tuberculosis, - : results from the global burden of diseases, injuries, and risk factors soper he: the interpretation of periodicity in disease prevalence the first author, soumi ray, ph.d. in image analysis from the indian institute of technology roorkee, has done the data analysis, explored useful insights, prepared the manuscript and concluded the work. the second author, mitu roy, b.tech in computer science from haldia institute of technology, conceived the content and retrieved the data to help the first author. none. both the authors declare to have no conflict of interest. none. key: cord- - jmnwpq authors: medina, marie-jo title: pandemic influenza planning for the mental health security of survivors of mass deaths date: - - journal: exploring the security landscape: non-traditional security challenges doi: . / - - - - _ sha: doc_id: cord_uid: jmnwpq influenza a pandemics have been documented to occur at - to -year intervals—an average of three events per century, dating back from the th century. each recorded pandemic has resulted in an increase in annual mortality rates in the infected population, with mass deaths in one pandemic wave equalling fatalities sustained over six months of an epidemic season. this chapter aims to rectify the oversight in pandemic preparedness plans by presenting a compendium of guidelines and recommendations by international health organisations, pandemic fatality experts, and experienced mass death management professionals. its objective is to have available a mass fatality framework to complement the who pandemic influenza preparedness and response ( ) guideline, from which individual national pandemic preparedness plans are based. it is written in a format that incorporates who’s emphasis on finding the ethical balance between human rights and successful plan implementation; the assimilation of national pandemic plans with existing national emergency measures; and the ‘whole group’ system of engaging individuals, families, localities, and business establishments in the process. this chapter is also written such that it can be made applicable to analogous infectious disease outbreaks such as sars and ebola, as well as comparable mass fatality events. influenza a pandemics have been documented to occur at -to -year intervals -an average of three events per century, dating back from the th century (kasowski ; taubenberger ; who ) . each recorded pandemic has resulted in an increase in annual mortality rates in the infected population, with mass deaths in one pandemic wave equalling fatalities sustained over six months of an epidemic season (hardin ). the three pandemics in the th century occurred in , , and . the latter two have been estimated to have resulted in increased deaths totalling up to four million in people in at-risk groups worldwide, while the former resulted in the mass deaths of approximately million in the otherwise healthy groups (hardin ; kasowski ; taubenberger ; who ) . the pandemic remains the most fatal pandemic in history; a novel influenza subtype of equivalent virulence is anticipated to result in deaths in approximately % of the current global population (ibid). there has so far been one pandemic this st century, caused by the h n influenza subtype in . although its attack rate was characterised as mild, it nonetheless resulted in the global deaths of up to , people who would not have otherwise perished at that time (dawood ) . approximately % of the fatalities were in populations younger than those who generally decease during influenza epidemics, and the burden was most pronounced in the poorer african and southeast asian countries (ibid). in , who published a guidance on pandemic influenza preparedness as a framework for who member-nations, in their attempts to develop a plan against the risk of the occurrence of an influenza pandemic, and to introduce the six phases in the declaration of a pandemic (who ) . in , improvements to the guidance were incorporated in keeping with the international health regulations (ihr). in , further revisions were made to consolidate developments that have transpired since the enactment of the framework (who ). pertinent to this discourse is the revision accentuating the prevailing of ethical principles when finding a balance between human rights and successful pandemic plan implementation. upholding ethical principles include respecting both the dead and the bereaved throughout the course of the event (morgan (morgan , ; handling and disposing of bodies in a dignified manner; and respecting cultural and religious conventions (ibid). further, it encompasses the acknowledgement of the diversified vulnerabilities and capabilities of individuals and groups, so that nobody experiences marginalisation and disavowal of support (sphere ) . vulnerabilities may be physical, such as: gender; age; physical or mental impairment; and hiv/aids status. they may also be social, including: ethnicity; religious affiliation; political leanings; and residency status (ibid). published literature in psychology suggests that disasters can induce mental illnesses among survivors (bonanno ; gibbs ) . the most often affiliated mental health illness in disasters is posttraumatic stress disorder (ptsd). however, several individual symptoms, as well as syndromes, have also been associated with the trauma, albeit not given a specific name (ibid) . some research promote that the amount of trauma sustained in a disaster is directly proportional to the severity of the psychological illness. others assert, on the other hand, that ancillary factors may also contribute to mental health risks. these may include the specific context with which the survivor identifies with the disaster; the emotional and physical distance an individual has from the situation; and the quality and accessibility of the support available (ibid). further, there are those who argue that ptsd may be overly estimated; while other, less characterised, symptoms are under-estimated (bonanno ) . this dubiousness in the literature has been attributed to the difficulty encountered in assessing psychological consequences sustained in disasters, because of the chaotic nature of the event; and because of the methodological impediments to psychoanalysis (ibid). to provide a more cohesive portrait of 'typical' mental health illnesses following a disaster, george bonanno and colleagues (bonanno ) compiled data from high quality research and summarised their findings in five categories. the first category relates to the severity of mental illness brought on by disaster. it was determined that, although consequences of trauma from disasters may range from grief and ptsd to depression and suicidal tendencies, more extreme presentations of the disease have only been observed in a small number of cases. in adults, this accounts for only % of all subjects studied. in youths, acute symptoms in the initial aftermath tend to be severe; however, chronic symptoms tend to be more similar in the adults, not exceeding %. the second category pertains to differences in psychological outcomes and resilience. it is suggested that some survivors overcome the traumas within two years post-disaster; while the more resilient only experience transient symptoms and recover fairly quickly. the third refers to the factors relating to outcomes, already alluded to above, and theorises that there is no single predictor of outcome. this is because individuals have different risk factors for mental health illness, as well as varied mechanisms for coping with trauma. the penultimate category specifies the risk to interpersonal and community relationships. it acknowledges that, although some affiliations are made stronger by shared traumatic experiences, several indicators suggest that most relationships actually do not survive the experience. incidentally, the status of their post-traumatic interpersonal relationships also influences their coping mechanisms. finally, in examining the mental health effects to populations located at a distance from the disaster scene, it has been determined that transient grief may be experienced by these individuals; however, psychological disorders may only be recognisable in those with prior experience in disasters, including those who lost loved ones under similar circumstances (bonanno ) . lastly, literature suggests that the emotional and psychological traumas among survivors of multiple deaths are compounded when the bodies of their loved ones are not processed with care; this is true irrespective of the age, race, or nationality of the deceased (gibbs ; morgan ) . poorly managed deaths therefore, present a perceivable global mental health risk. however, despite the globally acknowledged increase in deaths due to infection with novel influenza a subtypes, and all that is recognised about risks to mental health security in mass fatalities, pandemic preparedness plans remain disproportionately focused on preventing the manifestation of a pandemic and on mitigating morbidities and mortalities, rather than equally addressing mass fatality management preparedness plans. mass fatality management preparedness planning is paramount in any influenza pandemic preparedness plan if business continuity is to be expediently achieved, and survivor grief and psychological trauma can be mitigated through the honourable and respectful handling of the remains of the dead. this chapter aims to rectify the oversight in pandemic preparedness plans by presenting a compendium of guidelines and recommendations by international health organisations; pandemic fatality experts; and experienced mass death management professionals. its objective is to have available a mass fatality framework to complement the who pandemic influenza preparedness and response guideline, from which individual national pandemic preparedness plans are based. it is written in a format that incorporates who's emphasis on the assimilation of national pandemic plans with existing national emergency measures; the 'whole group' system of engaging individuals, families, localities, and business establishments in the process; and on finding the ethical balance between human rights and successful plan implementation. sources for the guidelines include: . hardin and ahrens ( ) (hardin hereafter) authored a chapter specific to influenza pandemic mass fatality management. it delineates the facts from the myths and provides a guideline for mass fatality planning. . the integrated regional information networks ( ) (irin), whose purposes are to promote the understanding of regional affairs; to advocate competent humanitarian response; and to advance knowledge-based media reporting. fatalities seminar summary report' ( ) (homeland hereafter). this report focused on the lessons learned by multiple sectors, based on their experiences with mass fatality response. . oliver morgan's 'management of dead bodies after disasters: a field manual for first responders,' ( ) (morgan henceforth) whose aims are to advocate decent and respectful dead body management; and to increase the likelihood of a successful victim identification. . the sphere project: humanitarian charter and minimum standards in humanitarian response (sphere hereafter). it developed the 'universal minimum standards' in humanitarian aid, based on the cumulative experiences of disaster teams and agencies. . the uk home office 'guidance on dealing with fatalities in emergencies' (home office henceforward). this is a joint publication of the uk home office and cabinet office, from which was based the london olympics pandemic plan, the most successful olympics yet. this chapter is written such that it can be made applicable to analogous infectious disease outbreaks such as sars and ebola, as well as comparable mass fatality events. mass fatality is defined as an event where the number of the dead exceeds available local capacities for appropriate management of human remains (morgan ; ralph ) . they may ensue from natural or man-made disasters, or infectious disease pandemics. mass fatality management planning is highly relevant because of the psychological effects improper handling of dead bodies can have on the survivors (ibid); and because initial stages of fatality management will determine the final outcome in the unequivocal identification of dead bodies, and the subsequent return of their remains to the rightful relatives (ibid). the survivors' utmost desire, in disasters, is to unequivocally ascertain the circumstances of their missing loved ones (morgan ). however, this desire may run contra-parallel to the disaster teams' priority-mitigating further consequences of the event (ibid). a balance between practicality and empathy would therefore, need to be established. formulating preparedness plans is made difficult by the necessity of predicting scenarios for which the plans can be rationally devised. undoubtedly, human imagination will fail to predict every possible scenario, and the disaster that eventually unfolds will be one too unbelievable to conceptualise. nonetheless, it is imperative that certain assumptions are made, if only to provide planners with a point of reference. when developing pandemic plans, hardin and ahrens ( ) suggest five assumptions that would be invaluable. they are: . the local community would need to be able to support itself, particularly during a pandemic, when similar events are simultaneously occurring elsewhere, and aid will tend to be diffused. . funeral homes will be rapidly overwhelmed. . resourcefulness will be needed in acquiring inventory essential for body management. . funeral and memorial practices may need to be altered to ensure the expeditious processing of bodies. . friends and family from near and far will be desperate for information. chaos is the immediate aftermath of a disaster (morgan ). therefore, a coordinated plan put into operation as soon as practicable will be invaluable in managing the disaster area. it is likely that local emergency personnel will be first at the scene, and will already have coordinated disaster plans in operation (ibid). however, it is important to note that stakeholders, leadership structure and operational procedures in pandemic planning may differ from these and other mass fatality plans (hardin ; morgan ). hence, it is essential that: (a) a comprehensive list of stakeholders is included in the plan. these may include: . emergency management teams . public health authorities . medical and veterinary teams . medical examiners and coroners . police . death registry . funeral directors . cemetery and crematorium administrators . legal professionals . religious officials and community support groups . schools . social well-being advisers . mental health professionals (b) establish a structure of leadership, with absolute authority ascribed to the entity presiding over the management of the dead. a flowchart with names, responsibilities and emergency contact numbers will be beneficial. (c) specify each stakeholder's duties and responsibilities. provide timelines and benchmarks for the successful completion of each task. (d) coordinate resources. a system of real-time stock-taking will be beneficial in the sharing and distribution of essential goods and services. stipulate how reimbursement for the use of shared resources will be managed, including realistic timelines for monetary disbursement. (e) coordinate with regional and national fatality management plans. their resources and expertise will be of considerable value, particularly in matters relating to funeral homes, mass communication, logistics, and national and international jurisprudence and aid. (f) coordinate with international aid organisations. they have the experience, expertise and resources to respond on short notice. coordinating resources beforehand (in (d) above) should prevent stockpiling of necessities with shortened expiration dates that may later go to waste. it is suggested that funeral directors have stock in circulation that is proportionate to a six-month supply for standard operations, the assumed length of the first pandemic wave. it is necessary to note that (hardin ; irin ; morgan ): (a) embalming fluids tend to have a protracted shelf life. (b) affordable caskets will be in great demand, particularly in instances when death occurs in more than one family member. (c) cremations will require large amounts of fuel. copious amounts of information are compiled on the dead and missing, regardless of the size of the disaster. appropriate management of all information will require human and technical expertise, which may be beyond the capabilities of local communities. regional authorities are more likely to have trained personnel and modernistic technologies, and may therefore, be best placed to take the lead in information management (homeland ; morgan ). mass media are indispensable in communicating with a wide audience during a disaster, and both amateur and seasoned journalists will be among the first at the scene. however, the content of the information they provide as well as the manner in which they dispense their knowledge of the scene may induce stress and anxiety among the survivors. therefore, it is paramount that members of the press be given every possible opportunity to communicate responsibly and to the best of their abilities (homeland ; morgan : ) . effective information management reduces stress and anxiety among survivors, and augments efforts in successfully recovering remains and identifying the dead. listed below are the matters that need to be considered (homeland ; morgan ): (a) coordinating information . information hubs need a local and regional presence and should be established in the first instance. . determine who would need to be informed, and what the best method of communication would be, to ensure that information reaches as much of the appropriate target groups as possible. . local centres are best for collecting and providing information on the dead and the missing, and for relaying information on the immediate needs of the grieved. . impose upon humanitarian and aid agencies since they will have first-hand knowledge of the state of the scene, and the kind of support the survivors will need. . all information needs to be centralised and synchronised for accuracy, and for promoting the successful tracking of the dead and missing. (b) the information . foremost is the protection of the privacy of those afflicted and their families. . take advantage of already established methods of gathering information (e.g. surveillance networks; automatic alert systems). ascertain whether expanding the scope of these systems will be beneficial and can be implemented rapidly. . use a template that covers all the essential information, and that could easily be updated. this would include what is being done; what is known; what is yet to be determined; and where further information will be provided when they become available. . an informed decision needs to be taken on when it would be appropriate to report the number of dead, missing and displaced. too soon, and the numbers are likely to be grossly inaccurate; too late, and the media could be disposed towards exaggeration. . information on the system of search and rescue, and body retrieval, identification, interment and disposal must be provided. . photographs and other identifying information should only be released to the media if it has been determined that doing so would enhance the identification process. (c) the media . designate a representative with whom the media may liaise. . install an office specific for media relations, preferably as close to the scene as possible. . provide journalists with accurate, confirmable, and up-to-date information as close to real time as practicable, to advance factual reporting and mitigate rumour-mongering. this may be facilitated through regular press briefings or short interviews. . social media is a double-edged sword. knowledge will be available immediately and in real-time; however, the material will tend to be unedited and prone to bias. if not managed appropriately, it may disrupt fatality plans already in progress. (d) the public . determine the most appropriate method of providing information to different age groups and social, cultural and economic strata, to avoid marginalisation. . circulate concise information on what procedures need to be adhered to, immediately following a disaster. . vigilance in social media trends is essential. (e) the survivors . impress upon survivors that help is available. enumerate what support can and cannot be provided, and where they need to go to receive the specific aid they need. . provide an emergency contact number strictly for the relatives of the missing and the dead. . provide specific information on where relatives need to go and what documents they would need to bring, to facilitate the efficient and expeditious management of enquiries. . specify the process for procuring a death certificate, so that they may be able to make legal and funeral arrangements. (f) the humanitarians . ensure that humanitarian and aid agencies are provided with accurate information, particularly in regard to the risks from dead bodies, and that they themselves are sharing accurate information to those at the scene. . relief agencies such as the international committee of the red cross may be able to help trace missing persons, if given sufficient information. (g) the dead bodies . standard pro forma containing basic information should be completed for all bodies. . in the absence of an electronic system of data-gathering, hand-written forms may be used. however, extreme care would be needed in writing and in the subsequent transfer onto an electronic format. . all manner of original forms must be readily available, should data confirmation be necessary. . all items of a personal nature, including photographs, may be included in the database. . all information must be accompanied by a chain-of-custody. in the early stages of a pandemic, scientific intelligence gathered through already established surveillance systems would need to be rapidly apprised of the nature of the virus and the manner of death, through the investigation of the index case. it is recommended that the role of investigator be entrusted to the jurisdictional medical examiner or coroner (me/c) in two capacities (hardin ): . limited jurisdiction over the dead body in cases when: (i) death fits the profile for an emerging disease that needs laboratory confirmation from body fluids and tissues. (ii) death of a poultry worker from influenza-like illness (ili). (iii) death from ili of family members or contacts of poultry workers. (iv) death due to recent travel to a country where pandemic flu strain is circulating. (v) first death case in a hospital, requiring tissue samples for virus characterisation. (i) there is no listed attending physician. (ii) the deceased is unknown and decedents have not been found. (iii) sudden deaths and fatalities uncharacteristic of those due to a flu virus. (iv) death of incarcerated persons. (v) it is essential to public health. (b) search for the missing death from pandemic influenza generally occurs at home or in group care facilities. in the event that an exceedingly virulent pandemic strain also kills its victims with haste, more will be unable to seek hospital admissions prior to death (hardin (ii) numbering and photographing the dead (or body parts for non-intact bodies). (iii) a mechanism for immediate confirmation of death by me/c. existing laws may need to broaden the stipulations on who has legal powers to pronounce death. (iv) record the date, time and place of death, as well as the testifier's name and contact information, and their affiliated organisation's name and address. . in the community (hardin ) (i) designate a phone number for the missing persons' hub where inquiries can be made about the well-being of certain individuals. this hub must be interfaced with hospital and healthcare centre systems of admissions and discharges, and with me/cand death registry logs. (ii) there must be a system for the regular advertisement of the hub number in several mass media formats. (iii) it is essential that the hub's database be unrestrictedly shared with the police and emergency missing persons' divisions. (c) recovery and transport of bodies dead body management begins when the remains of the deceased are being recovered (morgan ). recovery commences immediately after searching of the scene has been completed (ralph ) . it could last for days or weeks, but may be protracted in more severe disasters (morgan ). its priority is the rapid location and retrieval of bodies or body parts, and the deceased's personal effects. speed in recovery aids in identifying the dead; reducing the psychological impact on survivors; and diminishing the distress often associated with the image and odour of death (irin ; morgan ). the recovery scene is often chaotic and uncoordinated because there is an abundance of groups and individuals trying to help, including locals; aid agencies; and military and civilian search and rescue operatives (morgan ). in order that body recovery does not impede the simultaneous assistance offered to survivors, the following should be considered (hardin (i) use of photographic equipment and standard documentation materials such as body tags with unique references. documenting the exact place and date of recovery would augment the identification process. (ii) impermeable body bags are ideal for recovery, and double-bagging is preferential; however, sheets of any material may be used if nothing else is at hand. each body part must be collected in separate bags and no attempts must be made to match them at the scene. (iii) personal items ought not to be separated from the owner, and all documentation must remain with the body. (iv) establish two teams: one to take bodies to a holding area prior to delivery; the other to deliver them for either immediate identification or temporary storage for subsequent identification. (v) the holding area will have rapid turn-over. hence, it is best situated within close proximity of the scene; preferably stretched across the inner scene cordon. the holding area is a private and secured space where documents can be cross-checked and evaluated for completeness. at no point must this area be used as a mortuary; a facility for victim identification; or as a temporary storage facility. (vi) transport can be achieved by using the body bags or sheets with which they are covered, or by trucks and trailers; however under no circumstances must ambulances be used, as the living are best served by them. . disaster areas may be hazardous. it is paramount that recovery teams not be exposed to undue risks in performing already stress-filled tasks. risk assessments are requisite and basic health and safety measures must be in place (home office ; morgan ). (i) ventilate enclosed spaces before attempting recovery. (ii) at the minimum, protective clothing would include disposable biohazard suits; sturdy boots and durable gloves. face masks may be provided, if only to alleviate anxiety from odours and from fear of aerosol infections. (iii) personnel need appropriate training in donning, doffing and decontaminating protective equipment. (iv) a mechanism of hand-washing, disinfection and decontamination should be available. (v) first aid and emergency treatments will be needed on-site. (vi) the need for vaccination and prophylaxis would have to be evaluated. mass fatalities are expected to overwhelm local surge capacities which will invariably result in delays in victim identification. further identification delays can result from the logistics of assembling a forensics team, which can take weeks; and from natural decomposition. places in hot climates are especially vulnerable to decomposition, resulting in bodies becoming unrecognisable within - h. to maximise every opportunity of successfully identifying bodies, temporary storage facilities are compulsory. these can be in the form of cold storage or transitory interment (hardin ; irin ; morgan ; sphere ) . it is imperative that bodies or body parts are stored in the bags or sheets in which they were recovered and that their associated unique identifying tags are written on water-impermeable labels, rather than on the bodies or bags themselves (ibid). (iv) shortage of refrigerated storage at the scene is to be expected. establish a back-up plan until more coolers become available. (v) dry ice may be used in the interim . overlaying dead bodies with dry ice creates forensic artefacts, and should therefore, be avoided. instead around small groups of bodies, construct a wall of dry ice approximately . m in height, and secured with durable plastic sheeting. . ventilate areas where dry ice is in use. (vi) the use of ice is impractical and problematic. . a large inventory is required, particularly in instances when rapid melting occurs. . melted run-offs may pose concerns about diarrheal infections. . appropriate disposal of ice water will complicate management plans. . water may distort bodies and destroy personal properties. . interment is the burial of bodies underground when there are no other alternatives, and when temporary storage is needed for longer periods. (i) efficient disinterment will be aided by proper grave construction. . use a familiar and protected plot of land. . bury bodies individually if at all possible. otherwise, use trenches. . local practices may dictate how bodies are positioned (e.g.: facing mecca). . burials should only have one level; be at least . m in depth; and have parallel spaces . m in between bodies. . bottoms of graves with less than occupants should be at least . m away from ground water. this space should be increased to at least . m if buried in sand, and at least m if many more bodies are interred. . tag each body, and record their positions above the grave. use of gps systems will be invaluable. (ii) selecting burial sites . assess soil characteristics, height of water table, and available tracts of land. . situate in land acceptable to local communities. . establish in areas easily accessible to mourners. . sites should be at a distance of at least m from developed land, and m from sources of water, depending on local topographical conditions. (iii) unceremonious burial in mass graves does not satisfy any public health interests; is socially unacceptable; and may waste inventory. (iv) avoid rushed and unmannerly cremations. (v) it is disrespectful to gather the dead using backhoes, diggers, or bulldozers. (vi) sphere international standards mandate that: . bodies are disposed of with dignity . cultural and religious practices be honoured . public health practices be upheld. (vii) where burial is inconceivable due to frozen tracts of land or lack of solid ground, it may be necessary to store bodies for the duration of a pandemic wave. (viii) survivors are more likely to spread infectious diseases than dead bodies, except in cases where diarrheal diseases and haemorrhagic fevers are indicated. . tuberculosis, hepatitis b and c, and diarrhoeal diseases can survive for up to days in dead bodies. . hiv may survive for up to days. establishing the identity of the deceased is the second major function of incident response teams, following search and recovery, and is generally the responsibility of the me/c (ralph ) . identification is accomplished by making a match between the information collected about the deceased, and the information documented on the missing and presumed dead (morgan ). the sooner a positive id is accomplished, the better for the relatives waiting to bury their dead and to go through the legal procedures (ibid). visual identification through decedent recognition or photography is the most basic method of identification (home office ; morgan ; ralph ) . however, mistaken identity is common with this practice, particularly when the dead is soiled or already decomposed (ibid). further, viewing multiple dead bodies may have psychological effects on the witness, thereby diminishing the legitimacy of the identification. errors in identification cause embarrassment to all involved; distress to the relatives; and difficulties with legal issues (morgan ). therefore, forensic methods would also need to be employed. the success of forensic identification is enhanced by the initiative and diligence of the death management team (ibid). identification is carried out in the morgue, where the cause and manner of death are also determined. the me/c determines where the incident morgue is eventually established (hardin ; home office ; ralph ) ; it may be that a temporary facility is constructed, or that an already existing structure is expanded to accommodate the surge (ibid). the benefits and drawbacks of each type of facility would need to be judiciously considered (ibid). . things to consider: (i) determine how soon temporary mortuaries can be commissioned for use, compared to how quickly expanded space in already built mortuaries can be made available in disasters. (ii) commissioning time will have a direct impact on body recovery, storage, and transport. (iii) temporary facilities need to be operational as soon as h post-disaster. (iv) the use of previously functional morgues may mean that storage already contain bodies; hence, surge capacity will be unknown until such time as the disaster occurs. (v) the disaster scene will be instrumental in determining the necessity of constructing temporary facilities. information on the projected number of afflicted; the disposition of the dead; and the estimated . autopsies are not needed to confirm death caused by influenza. however, if they are performed, some guidelines apply: (i) in the interests of public health, respiratory tract and tissue samples for laboratory analyses may be collected. (ii) liaising with public health laboratories on the current guidelines for collecting and transporting influenza specimens will save time and effort. (iii) next-of-kin will generally need to give permission for the autopsy to be performed in a hospital. (iv) me/cs do not need permission if the autopsy is within their remit. (i) release dead bodies only when a definitive identification has been made. (ii) expedited release may be necessary where cultural or religious customs are indicated. (iii) some laws stipulate who has the authority to perform this task. (iv) the name and contact details of the claimants need to be collected and filed along with other documents associated with the body. (v) unidentified bodies, foreign nationals, undocumented migrants, and homeless persons need to be stored or interred for further identification at a later time. (vi) release of bodies with missing parts may later impede the management of severed body parts. to minimise complications, family members' wishes regarding future identification of other body parts should be documented. choices may include: . to postpone body release until all body parts have been found. . to proceed with the funeral but be apprised of other parts that are later found. . to proceed with the funeral and consider the matter closed. (vi) a death certificate is provided with the release of the body. . death certificates (i) the death certificate is a legal document; hence, the law stipulates the signatory on the certificate. (ii) the document specifies the cause and manner of death; where death occurred; when it was pronounced; and the name and contact details of the signatory. (iii) in pandemics, it is essential that hospitals and care facilities assign this task to specific individuals in order to mitigate chaos. (iv) funeral directors with policies against collecting bodies unaccompanied by a certificate of death need to allow for flexibility during pandemics. . this should be addressed in the planning stages. . all stakeholders must be in agreement. (b) funeral homes and crematory operations funeral directors are responsible for the recovery and transport of dead bodies; preservation of the integrity of the chain-of-custody; and assistance in disposal of the remains. although they are not qualified grief counsellors, they are nonetheless tasked with conversing with individuals on the most discomfiting day of their lives. this therefore, also makes them the best people to facilitate the return of the dead to their bereaved relatives (homeland ). once a body has been released to the decedents, it is generally their responsibility to contact the funeral director of their choosing, for the transport of bodies to funeral homes and the subsequent burial or cremation, according to their culture or religious beliefs (hardin ; homeland ; irin ; morgan ; sphere ) . pandemics could result in funeral homes overseeing months' worth of dead bodies within a - week period (hardin ; homeland ; irin ; morgan ). therefore, it may be prudent for individual funeral homes to plan for employing more trained personnel who can be available on short notice (ibid). (i) funeral directors will be responding to requests from families to transport bodies to funeral homes, and from me/cs to provide conveyance to mortuaries or storage facilities. plans for the inclusion of more licensed funeral directors and transport services is therefore essential. (ii) safeguard lawful body transport by ensuring that funeral home personnel are licensed and trained in recovery and transport, and that their vehicles are approved and registered for carrying dead bodies. (i) burials are more practicable in disasters, because they enable future identification of persons yet unknown. (ii) it is not good practice to cremate the remains of unidentified bodies. . there is no public health benefit in cremating those who die of influenza. . cremation will not allow identification in future. (iii) cremating one body takes h and produces to pounds of ash and partially incinerated body parts; thereby, creating logistical difficulties when the number of bodies rapidly mount. (iii) ensure the area is secure and private. (iv) it needs to be accessible for h within the first days, after which it can be scaled down to operate for - h a day. (v) anticipate approximately kinsperson for every victim and plan accordingly. (vi) multiple facs may be necessary, but movement of families from one area to another must be avoided; instead, fac personnel should go to where the survivors are situated. (vii) facilities must be scalable. . support and assistance (i) prioritising the needs of the vulnerable. (ii) personal and private meetings with family members as soon as practicable to initiate the collection of ante mortem information for the mortuary. (iii) system for reporting and providing information on the missing. (iv) emotional and psycho-social support for survivors befitting their needs, culture, and the context of the disaster. (v) systematic, up-to-the-minute information on the missing and the dead. families ought to be the first informed of the status of their loved ones. (i) each support agency within facs needs a command post; a separate area for staff preparation and duty operation; and the capability to deploy staff to fac. (ii) the nature of the disaster will determine which agencies are involved. frequently in force are family assistance services; mental health assistance; and child agencies. (iii) aid agencies and faith groups may be present. (iv) fac staff must be vetted. (v) flexibility is essential in order to accommodate the changing needs of the families as time progresses. based on the history of influenza a pandemics, this century may be due for, at most, two more pandemics. if even one of them is as deadly as that of , then approximately % of the global population will die. however, even if the future st century pandemics are atypically mild as that of , still many more people will die than normally would. the who provided a framework for influenza pandemic preparedness planning. however, its focus is skewed towards the prevention of the event from happening, and a bit remiss on planning for the management of the surge in deaths. having a fatality management plan incorporated in pandemic plans is relevant because mishandling of dead bodies is a mental health risk for their loved ones. mass fatalities may ensue from natural or man-made disasters, or infectious disease pandemics. regardless of how they may transpire, conflict will invariably come to pass between the fatality management team, and the surviving relatives of the missing and the dead. conflict is inevitable, because each group contextualises the event from different perspectives; fatality management personnel perceive the event as something that needs immediate oversight, in order that they may mitigate further calamitous consequences; survivors, on the other hand, are more single-minded in their overwhelming desire to determine the circumstances of their missing loved ones (morgan ). however fatality management ultimately eventuates, respect; sympathy; and caring are due the dead and their relatives throughout the event (ibid). weighing the costs of disaster: consequences, risk, and resilience in individuals, families, and communities estimated global mortality associated with the first months of pandemic influenza a h n virus circulation: a modelling study disasters, a psychological perspective ed) pandemic influenza emergency planning and community preparedness managing mass fatalities seminar: a summary report guidance on dealing with fatalities in emergencies influenza pandemic epidemiologic and virologic diversity: reminding ourselves of the possibilities mass fatality management following the south asian tsunami disaster: case studies in thailand ) management of dead bodies after disasters: a field manual for first responders mass fatality management', disaster resource guide analysis: why dead body management matters', irin humanitarian news and analysis influenza: the mother of all pandemics who global influenza preparedness plan: the role of who and recommendations for national measures before and during pandemics pandemic influenza preparedness and response: a who guidance document time of recovering their remains, all need to be considered. (vi) in the event that a pandemic is caused by a cbrn attack, the mortuary will be fundamental in criminal investigations; hence, standard operating procedures must be such that substantiation does not fail under legal scrutiny. family assistance is one of the most sensitive undertaking in mass fatality management. family assistance centers (fac) are generally established near mass fatality scenes, where survivors can congregate to wait to hear about the status of their missing, and to receive much-needed support (homeland ; morgan ; ralph ) . facs are secure, private, and multi-sectorial, so that all the support and assistance needed can be provided under one facility (ibid). things to be considered in establishing facs include: . function of fac (i) to provide families with information on their missing and dead. (ii) to provide shelter from media intrusion and from the newsmongers. (iii) to enable investigators and me/cs to gather information from families about the missing and the deceased. (i) situate facs near the disaster scene, where ingress and egress can easily flow. (ii) avoid locating facs near the morgue. key: cord- -n fbzae authors: richmond, peter; roehner, bertrand m. title: coupling between death spikes and birth troughs. part : evidence date: - - journal: physica a doi: . /j.physa. . . sha: doc_id: cord_uid: n fbzae in the wake of the influenza pandemic of – jacques bertillon, a pioneer of medical statistics, noticed that after the massive death spike there was a dip in birth numbers around months later which was significantly larger than that which could be explained by the population change as a result of excess deaths. in addition it can be noticed that this dip was followed by a birth rebound a few months later. however having made this observation, bertillon did not explore it further. since that time the phenomenon was not revisited in spite of the fact that in the meanwhile there have been several new cases of massive death spikes. the aim here is to analyze these new cases to get a better understanding of this death–birth coupling phenomenon. the largest death spikes occurred in the wake of more recent influenza pandemics in and , others were triggered by the earthquakes in tokyo and the twin tower attack on september , . we shall see that the first of these events indeed produced an extra dip in births whereas the / event did not. this disparity highlights the pivotal role of collateral sufferers. in the last section it is shown how the present coupling leads to predictions; it can explain in a unified way effects which so far have been studied separately, as for instance the impact on birth rates of heat waves. thus, it appears that behind the apparent randomness of birth rate fluctuations there are in fact hidden explanatory factors. this paper is about a remarkable case of birth and death fluctuations which, apart from its own interest, may give new insight in the more general problem of vital rate fluctuations. we begin with the following observation. • in the early th century european populations were increasing at an annual rate of around %. under the assumption of a constant λ birth numbers will see the same annual change. thus, for monthly changes the rate will be times smaller, i.e. around . %. • in contrast, the monthly birth rate fluctuations of actual observed monthly birth numbers were about %. this is times larger than the fluctuations under constant birth rate. however remarkable, the case considered in this paper is possibly just one of several similar processes leading to predictable birth rate changes. the topic of short-term fluctuations has so far attracted much less attention than the study of medium-or long-term changes. for instance the demographic transition in developed countries was brought about by changes of vital rates over a time interval of several decades. the present study shares several important features with the research field that is concerned with the fluctuations of sex ratio at birth, see [ , ] . • both researches rely crucially on comparative analysis, either in space across different countries or in time over past centuries. • both investigations focus on short-term effects, for instance the changes that occur in the months following an epidemic or a war, see [ ] and [ ] . • the respective key-variables, monthly birth rates on one hand and sex ratio fluctuations on the other, can be used as markers, in other words as measurement devices which give insight into abnormal situations. for instance, james [ ] documents how offspring sex ratios can reveal endocrine disruptions; similarly mortality sex ratios can be used to explore anomalies of the immune system, see [ ] and [ ] . in any country the time series of monthly births display substantial fluctuations. there is usually a seasonal pattern which is country dependent; in addition for the same month (say october) in different years there are annual fluctuations of about the same magnitude. it is customary to say that these are random fluctuations but are they really random? in this respect one can observe that by saying that a phenomenon is random one gives up ipso facto all attempts to understand it. it is probably for this kind of reason that albert einstein supported the ''hidden variable'' interpretation of quantum physics. in , i.e. some years after quantum mechanics was introduced, einstein et al. [ ] suggested that the wave function description of quantum objects was incomplete in the sense that it did not include some hidden parameters. in jacques bertillon, a pioneer of medical statistics and one of the designers of the ''international classification of diseases'', published an analysis of the influenza pandemic of november -february in which he showed that approximately months after the climax of the epidemic a temporary birth rate trough (of an amplitude of about %) was observed in all countries where the pandemic has had a substantial impact, particularly austria, france, germany or italy. while writing his paper bertillon did not know that some years later there would be a massive influenza pandemic through which his discovery could be tested. we will show below that it was indeed confirmed in all countries where the pandemic has had an impact. apparently, there has been no further studies of this phenomenon ever since. actually, even in bertillon's paper the effect is discussed fairly briefly. in particular its mechanism remains to be uncovered. this is the purpose of the present paper. while in - and - the effect can be detected very clearly, is it not natural to assume that it exists also in less spectacular cases. this raises the following question. in most countries the death rate presents a winter peak in january fig. a shows that the population changes based on the summation of monthly births and deaths approximates correctly the observed annual population changes. the curve of birth numbers based on a constant birth rate is naturally parallel to the population curve; it is displayed in the inset graph of fig. b ; the variation interval of births numbers (as shown on the vertical axis of the inset) is ( . , . ). these variations are much smaller than the actual fluctuations of birth numbers; as a result the broken line of predicted birth numbers looks almost completely flat when displayed on the same graph as the actual births fluctuations. source: bunle [ ]. or february. according to the bertillon coupling one would expect a trough of births months later that is to say in october or november. is this indeed the case? if not, in what respect do exceptional death surges differ from regular winter surges? this point will be discussed in the second one of the two papers devoted to this question. in an attempt to get a clearer idea of the mechanism of the coupling effect the present paper will address the following questions. ( ) first, we will review and expand the analysis presented in [ ] . the question which comes immediately to mind is whether the same effect can be observed in other pandemics. ( ) if the answer to the previous question is affirmative it raises another interrogation, namely is this effect restricted to pandemics or does it also exist for other large-scale mortality shocks, for instance famines, earthquakes and so on. wars should not be included in our study for in this case the separation between husbands and wives interferes with the coupling that we wish to observe. ( ) a possible mechanism that one can imagine is through a transient impact on marriages. if the marriage rate is reduced during the time of the epidemic, one would indeed expect a reduction in birth numbers some months later. such an explanation can easily be tested provided one can find monthly marriage data (see below). before starting this investigation we need to answer an obvious objection which can be stated as follows. is it not natural that following a reduction in population one sees a fall in the number of births? after this fall the birth numbers will go up again as the population resumes its ascending movement. qualitatively the argument seems satisfactory. if correct, the effect would become trivial. however, in what follows we show that quantitatively the argument is not correct for it explains less than one tenth of the observed birth changes. before coming to more elaborate arguments one can make a simple remark. what we see after months is a fairly narrow dip. on the contrary, a fall in population would produce a permanent reduction, in other words not a dip but a heaviside step. it is true that because of the upward trend due to the overall population growth after a while the births would resume their ascending progression, however the resulting shape would be a broad trough rather than a narrow dip; this difference can be seen clearly in fig. a versus b . we present the reasoning in two forms. while straightforward, the first is perhaps not very transparent; the second is more theoretical but intuitively clearer. the first argument is very simple. representation of the bertillon coupling effect in the form of an input/output system. if one interprets the death spike as an impulse function, the birth trough can be seen as the impulse response of the system. an important point is that, as explained in the text, the loss of lives during the death spike cannot by itself account for the observed birth trough. its direct effect is at least times too small which means that there must also be an indirect effect; it is the purpose of the paper to identify it more closely. the statistical records provide monthly birth and death numbers, b e (t) and d e (t). thus, starting from a known population at initial time t we can forecast the population p f (t) in all subsequent months simply by summing up the monthly population increases b e (t) − d e (t). it can be checked (fig. a ) that p f (t) is indeed consistent with the observed population evolution p e (t) (which amounts to say that in this time interval emigration and immigration do not play a great role). then, under the assumption of a constant birth rate λ the predicted monthly birth numbers will be: b f (t) = λp f (t). when these numbers are compared with the observed monthly birth numbers b e (t) one sees a huge difference in the magnitude of the fluctuations (fig. b) . to double check that there is no flaw in the data on which the graph is based consider the start and end points. according to flora et al. [ , p. ]: • in when the mid-year population was thousands a (constant) annual birth rate of per thousand gives , births for the year, i.e. an average of births per month. • in when the mid-year population has increased to the same birth rate gives , births, i.e. an average of per month. this represents a total predicted birth number increase of . % whereas the observed monthly birth numbers show fluctuations of the order of %. the following statement (explained in appendix a) summarizes the situation. proposition under the assumption of a constant monthly birth rate λ the fall in births resulting from a monthly death excess e is given by: ∆b = λe; for european countries in the early th century λ ≃ %/ which leads to: ∆b ≃ e/ . the proposition has a fairly clear intuitive interpretation. to make things simple let us assume that the death-excess variable e takes place in one month, as is for instance the case for earthquakes. in a virtual world where everybody conceives once every month, for each missing person there would be a missing baby months later. under this assumption the birth trough would be of same magnitude as the death spike. however we know that actually the probability to conceive in a given month is λ/ = . where λ is the annual birth rate. in other words in , persons only may conceive in the relevant month. thus, this direct effect contributes very little to the birth trough. the birth rate reduction can be attributed to one of the factors mentioned in fig. a . the objective of the present paper is to see more clearly which one of these factors plays a leading role. what real-life mechanisms can one think of that may explain the temporary birth rate fall? in fig. a there is a distinction between social and biological factors. it is by comparing different case studies that we came to the conclusion that social factors play a key-role. how? as famine and influenza have biological effects, at first sight the resulting birth reduction might also be attributed to such effects. on the contrary, birth reductions after earthquakes can hardly by accounted for by biological factors. for that reason the case of the tokyo earthquake of played a key role in our understanding. it convinced us that the suffering of the many people who are affected but do not die was of central importance. in examining successive case-studies it will also be seen that the magnitude of the birth reduction is compatible with the existence of a group of people that we call collateral sufferers. this term refers to persons who are affected by the event under consideration (whether epidemic, disease or anything else) but do not die. in order to generate the observed birth dips the size of this group must be several times larger than the number of fatalities. this is illustrated in fig. b . in sweden although there were only , excess deaths in , it is estimated that about one third of the population, that is to say almost millions ( times more than those who died), was affected by the disease. naturally, this fraction is difficult to define precisely because there is a continuum between the persons who were not affected at all and those who experienced very mild forms of the disease. the two following points need to be emphasized. • among the persons affected by the disease (or even in the whole population), there may have been a more restrained sexual behavior for instance by fear of a possible contagion. this may have produced a reduction in the number of conceptions. • among the persons mildly affected by the disease there may have been biological effects leading to less fecundation or early miscarriages in the or first months of pregnancy. the fact that in what follows we will observe the bertillon birth effect not only for diseases but also for earthquakes may seem to speak against such biological effects. however, one cannot exclude possible psychosomatic effects resulting from a stressful situation. an example is the so-called famine amenorrhea [ ] [ ] [ ] [ ] . finally it should be mentioned that the birth phase represented in fig. a comprises in fact two cases: natural birth and medically assisted birth through caesarean or induced delivery. such medical interventions are much more frequent on ordinary working days than on holidays. for that reason, there is a drastic reduction in daily birth numbers on saturdays, sundays and on holidays. for instance in the united states; january. july, labor day, thanksgiving, - december are marked by a reduction of about %. this effect is of great importance when one uses daily data; at the level of monthly data the effect is ''diluted'' because the deliveries are simply postponed by a few days, thus the total monthly figure should not be affected. at first sight one might think that birth dates may be critically affected by marriage dates. however, in the discussion of appendix b it is shown that this effect is fairly weak. one reason for that is because only first born children are concerned. thus, in the th or early th century when having four or five children was common, only a small fraction of the births were affected. it is currently said that, as indicated in fig. a , pregnancy lasts months which corresponds to × / = average days. however this statement raises two questions. ( ) for this -month estimate what starting point is considered? ( ) what is the dispersion of pregnancy around its average? this question is of importance because it affects the width of the birth trough. these questions are discussed in appendix c; it appears that the average time interval between sexual intercourse and birth can be taken equal to: the fact that the standard deviation is of the order of a few days shows that the dispersion of the births (that is to say the width of the dip) will be mostly due to the dispersion of the conceptions. even for a sharply defined event such as an earthquake, the behavior of collateral sufferers may be modified for several weeks. actually, it is through the width of the dip that we can know how the survivors reacted to the shock in terms of intercourse frequency. in finland the spring of was unusually cold; in may the average temperature was only . • c, which is some degrees below the long-term average. this made it very difficult to grow spring cereals or potatoes. then, in autumn the winter started early and was also colder than average. at the end of and early relief grains were imported thanks to a loan of the rothschild bank in frankfurt, however, as often in such cases, the inadequacy of the transportation network hampered delivery. in the death rate was per thousand, already % above average; then in it climbed to per thousand. just as an element of comparison, it can be mentioned that this rate was . times higher than the rate of per thousand reached during the famine of in china. in the breakdown of the deaths according to their causes only famine deaths were recorded whereas , deaths were attributed to tuberculosis, dysentery, smallpox and whooping cough. an additional , were attributed to ''various fevers'' [ , p. - ] . these numbers illustrate the zones of fig. b ; food scarcity was the triggering factor but it killed very few people directly. with respect to a population of . million, the excess-deaths of , correspond to a rate of per thousand. of the mortality cases that we are going to examine, this one is the most massive. thus, one expects the bertillon birth effect to be clearly visible and indeed it is. for the respective peaks the birth-death ratio is approximately %/ % = . . in other words, if one considers the mortality increase as the signal and the change in births as the response of the system we have here an output signal which is about / of the input signal. incidentally, it can be noted that these data were available in bertillon's time but he did not use them. the pandemic of december -january is described in [ ] . in paris the first cases occurred in the week of november . on the basis of death certificates in paris influenza was given as the direct cause of death for only persons. however, during the time of the epidemic there were about more deaths than in the corresponding period of the previous years. thus, zone of fig. b was larger than zone . with the population of paris numbering about million, one gets a fatality rate of . per thousand. whether the epidemic can be called a pandemic is a matter of definition. it is true that there was a death spike in many places (saint petersburg, berlin, vienna, paris, london, new york) but with the exception of paris and london, in all other places it was not more severe than the common annual winter death spike. of the pages of bertillon's report, only two are devoted to the effect on birth numbers months later. from the weekly data that he gives for berlin and paris (as well as some other european capitals) one can draw the following observations. fig. a shows the two curves in normal graphical representation; the death scale is on the left and the birth scale on the right (both are expressed in thousands). in fig. b the birth curve was shifted months to the left and turned upside down (by turning to negative numbers). as a result this new series represents the conceptions (at this point we do not know why the time lag is rather months instead of months). the bertillon birth effect is clearly visible in the fact that the maximum of the deaths coincides with a minimum of the conceptions. in fig. b the correlation of the two series is . . a correction was performed on the monthly data so as to give all months the same length of / = . days. all monthly data used in the rest of the paper were corrected in this way before being used. source: finland [ ] . fig. a the death scale is on the left and the birth scale on the right (both are expressed in thousands). in fig. b the birth curve was shifted months to the left and turned upside down. the bertillon birth effect is again clearly visible, this time with the expected time lag of months. in fig. b the correlation of the two series is . . source: statistique de la france, nouvelle série. statistique annuelle, various years. • in paris the peak of the epidemic occurred in the first week of january whereas the lowest point in the trough of birth numbers occurred in the st week of . in both cases the time lag is close to the weeks of a normal pregnancy. although bertillon examines the timing with great accuracy he does not discuss the question of the amplitude of the troughs. in particular, he does not show that the troughs cannot be explained solely by the deaths due to the epidemic. that is of course a key point which is why in the previous section we discussed it with great care. although the death toll was much smaller than in the case of finland, the bertillon birth effect appears fairly clearly. what makes the observation more convincing is of course the fact that the trough can be identified not just in a single city (where it might occur almost by chance) but simultaneously in several cities having non-identical seasonal birth fluctuations. in addition to paris and berlin (see twenty eight years after the pandemic of - there was the great influenza pandemic of . the death-birth coupling can be observed in all countries where this disease had a substantial impact. however, we will not examine these cases here because they will be studied closely in the second one of this series of two papers. the particular interest of this case comes from the fact that it was neither due to famine nor to a disease. although called the ''great kanto earthquake'' in japan, it concerned in fact mainly two of the prefectures which constitute the kanto region, namely tokyo and kanagawa (i.e. yokohama just south of tokyo). some % of the fatalities were concentrated in these two prefectures. the number of fatalities can be estimated by subtracting the average death numbers of and from the death number of , i.e. (in thousands): − = [ , p. ]. for the whole of japan the death rate was . per thousand, in tokyo it was . per thousand and in yokohama it was per thousand. the earthquake was accompanied by a tsunami with a wave up to meter high and, especially in tokyo, by fire tornadoes. the birth trough is of smaller amplitude than in previous cases and in fact there are two circumstances which play a crucial role in its identification. ( ) although the seasonal births have wide fluctuations (in fact much larger than in other countries) their annual repetitions are very regular. ( ) as one knows exactly where the trough is expected even a small signal can be identified. a purely statistical analysis that would fail to take into account these circumstances would result in overestimating the size of the confidence interval. incidentally, if monthly data were available at prefecture level one could get a better accuracy. two airliners belonging respectively to united airlines and american airlines were crashed into the north and south towers of the world trade center complex in new york city. within about two hours both buildings collapsed. there were some fatalities including some firefighters and police officers. with respect to the million population of new york city this corresponds to a rate of . per thousand. there have been many odd stories and speculations about this attack. however, there is one well documented, indisputable and nevertheless not often mentioned aspect which is the huge amount of put options bought in the days preceding the attack. for stock owners, put options provide a protection against the fall of a stock price because it gives them the right to sell their stock at a predetermined price which may be much higher than the current price. moreover, because the price of a put option increases when the price of the stock declines, speculators can make a profit by selling their put options. more details can be found on the following webpage: http:// research.wtc .netsept /stockputs.html. fig. is remarkable because it shows that / did not produce any birth trough whatsoever. it shows that the death-birth coupling is by no means commonplace. here the zones and of fig. b can be merged into one which corresponds to the total death toll. zone would correspond to persons (some ) injured but not killed. zone can be seen as comprising the families and close relatives of the persons killed or injured. based on the average size of us households which is of the order of . , one gets for zone a total number of: . × = , . of this number only the fraction in the age interval - would contribute to the birth trough. based on the us population pyramid this fraction is of the order of %. for the pandemics considered so far, it is almost impossible to estimate the size of zone . however, for an earthquake one can posit that zone corresponds to the persons whose houses have been destroyed or damaged. the outbreak of sars (severe acute respiratory syndrome) in hong kong is interesting for two reasons. it is not only the number of deaths which was small but also the number of cases, namely (i.e. per million population). worldwide it was the same picture; there were only some deaths compared with the , who died from influenza in the same year. nonetheless, the city took drastic measures. • primary and secondary schools were shut for a month beginning in late march. • various public places were closed. • financial companies asked their employees not to come to their office and work from home. • a whole residential complex called amoy gardens was put under an emergency quarantine. the residents were sent to a vacancy center and the building was closed. altogether in this complex people were infected and died. • unlike the influenza pandemic of , sars was particularly severe for elderly people. none of the infected females under died whereas among males older than the death rate was %. thus, although healthcare workers accounted for % of all infected persons [ , p. ] only few died. despite the small death toll, the death effect of the sars epidemic can be identified fairly clearly because it occurs two months later than the standard winter death outbreaks (fig. ) . the conception effect can also be identified clearly the trough occurs of shifted births occurs in march rather than in august. the epicenter of the earthquake was under sea near the city of sendai which is km north-east of tokyo. the death toll (including the missing) was about , and a further were injured. moreover some , buildings collapsed or half-collapsed. in other words we can take h = , as an estimate of the number of people who were directly affected. we will see below that under appropriate assumptions one can use this estimate to derive the expected birth number reduction. the conception trough of march due to the earthquake appears fairly clearly because it is distinct from the seasonal troughs seen in other years (see fig. ). in all epidemics and disasters, apart from the fatalities, there is a group of persons which is directly affected. for an influenza epidemic this would include the persons who fell seriously ill but did not die. for an attack like / it would be the family members of those who died or were injured. naturally, the previous dichotomous picture is a simplification. actually, there is a gradual transition from those highly affected to those lightly affected. under such an assumption the reduction ∆b in the number of conceptions would be written: here λ is the ''normal'' conception rate, x is the zone-dependent conception rate and h(x)dx is the number of persons having a conception rate comprised between x and x + dx. the dichotomous argument which gives ∆b = λh would correspond to: h(x) = hδ(x) where δ(x) denotes the dirac delta distribution and h the total number of people affected. the function h(x) describes the severity of the shock among the surviving population. if h(x) is concentrated in a narrow interval (λ − ϵ, λ) it means that the surviving population is not much affected. on the contrary, a function h(x) concentrated in a narrow interval ( , ϵ) means a severe incidence among those affected (their reproduction rate falls from λ to almost zero). in the next section we discuss shortly how information about the function h(x) can be derived from observation. table summarizes the death rates of the various case studies considered above. the function h(x) describes the incidence of the death spike in terms of suppressed conceptions and suppressed births. how can one derive information about it from observation? one can offer the following suggestions. • first, is suppressed conceptions identical to suppressed births? not necessarily. there can be a biological regulation which leads to early elimination of embryos which are not in good shape because they were generated in conditions of illness, scarcity or famine. this is a medical question for which one should be able to find reliable information in the medical literature. • the fraction of the population directly affected by the mortality spike comprises the following subgroups. (i) persons killed (k ) (ii) persons injured or who were seriously ill (h ) (iii) persons whose living conditions were affected in a major way, for instance because they lost a family member (h a ) or because their home was destroyed (h b ). so far, we have focused on death spikes. however, according to the view outlined above there could also be cases displaying birth troughs without any (or at least very few) fatalities but instead a substantial number of persons who become ill, injured or were otherwise severely affected. as possible example we investigated west nile fewer in the united states. this disease which is due to a virus carried by mosquitoes peaks in august-september which would give a birth trough in may. however, we could not identify any significant birth trough even in the states most affected, e.g. texas in . the number of incapacitated persons may be too low. there is no birth trough following / but there is one in the hong kong sars epidemic whose death rate is about times smaller. this seems fairly puzzling but the two events were of very different kinds. whereas / was a one-day event, the sars epidemic lasted a few months and, especially during early times, the spread of the disease as well as its severity (in terms of number of deaths per cases) were a matter of uncertainty and concern. thus, even persons who did not become ill were affected. we have shown that sudden death spikes are almost always followed months later by a birth trough. we have also seen that the spike of / did not lead to the expected birth reduction. this observation is particularly intriguing when compared with the sars outbreak in hong kong in which there were less deaths than in / and which was nonetheless followed by a birth dip. although a tentative explanation was proposed it is clear that in order to get a better understanding it would be useful to examine other cases in which a death spike is not followed by a birth dip. for that purpose the normal procedure is to propose predictions in the hope that for some of them the expected dip will not materialize. this strategy is very much in line with what was done in the development of physics. every time an expectation happened to be contradicted by observation, this was an opportunity for new progress. a well-known case was the non-observation of the aether by michelson and morley which led to the theory of relativity. there are two possible kinds of predictions. • the first kind consists in what we may call standard predictions; they are instances very similar to those already analyzed but less well known because of smaller amplitude. the predictions of birth troughs in the netherlands in - and in chile in are of this kind. both graphs are shown in appendix e which is included in the arxiv version of this paper. although in the case of chile we do not yet know the reason of the death spike of july-august , the birth trough could indeed be observed in the month in which it was expected. • in the second kind of predictions one deliberately considers a cause of death that has not yet been tested. one case of that kind are deaths due to heat waves. for instance in france in august a heat wave caused , excess deaths. this was a fairly exceptional case. most other heat waves caused only of the order of one or a few thousands excess deaths. in contrast with diseases, for heat waves there is no contagion effect; in contrast with earthquakes there are no collateral destructions. however, in contrast with / , all persons (at least those who do not have air conditioning) are directly affected to some degree. are there birth dips in the wake of heat waves? because of the fairly low amplitude of most death spikes the identification of the troughs turns out to be more difficult but they are nevertheless present [ , ] . ] ∆p in the case of a death spike the population change will be given by the excess-death number ∆p = −[d(t) − b(t)] = −e (for the sake of simplicity we ignore the -month time-lag between conception and birth). this leads to the proposition given in the text. as a case in point in order to illustrate the previous argument with real data we consider again sweden during the influenza epidemic of . it is by purpose that we selected a country which did not take part in the first world war so as to avoid any interference. the data are taken from [ , p. , ] and [ , p. ]. in early the swedish population numbered . millions; on average its annual birth and death rates were . % and . % respectively. in other words, the birth number reduction cannot simply be a ''mechanical'' consequence of the death spike. it can only be explained by a drastic reduction in the birth rate. average time interval between intercourse conducive to conception and birth. the familiar estimate of (mean) months corresponds to an interval of days, that is to say one week longer than the (more accurate) estimate given in the figure. the standard day figure refers to the time interval between the last menstrual period and birth; it overestimates the time between sexual intercourse and birth by two weeks. the day dispersion of ovulation refers to its standard deviation. source: wilcox et al. [ , ] , bhat and kushtagi [ ] . each duration, which gives an average of d = . the distribution of the fertility window is approximately gaussian with a standard deviation of days. in summary: d = ± σ , σ = days. this estimate is confirmed by the following result based on a sample of pregnancies in which ''the median time from ovulation to birth was days'' [ ] . remark. in the previous results ovulation time was determined by urinary hormone measurements or estimated through ultrasound observation performed later on in the pregnancy. needless to say each of these methods involve some uncertainty. it might seem that medically assisted conception would afford a direct and henceforth more accurate method. the difficulty here is that such pregnancies are know to lead to an inflated proportion of preterm deliveries. cycle day of insemination, coital rate and sex-ratio the inconstancy of human sex ratios at birth. eletters to the editor sex ratio at birth and war in croatia the variations of human sex ratio at birth during and after wars, and their potential explanations offspring sex ratios at birth as markers of paternal endocrine disruption sex differences in measles mortality: a world review sexist diseases can quantum-mechanical description of physical reality be considered complete? influence des épidémies de grippe sur la fécondité the influenza epidemic in paris and in some other cities in western europe le mouvement naturel de la population dans le monde de à state, economy, and society in western europe - . a data handbook in two volumes famine amenorrhea (seventeenth to twentieth century) biologie des menschen in der geschichte. beiträge zur socialgeschichte der neuzeit aus frankreich or skandinavia les crises de subsistence et la démographie dans la france d'ancien régime the french subtitle of this official publication is: Éléments démographiques principaux de la finlande pour les années - , ii: mouvement de la population. the finnish title is: suomen [finland] väestötilastosta the swedish title of this official publication is: befolknings-statisk [population statistics] ii. . underdåniga berättelse [report to the king] för åren med statistique internationale du mouvement de la population d'après les registres d'État civil. résumé rétrospectif depuis l'origine des statistiques de l'État civil jusqu'en epidemiology of sars in the hong kong epidemic Évolution de la saisonnalité des naissances en france de à nos jours [changes in the seasonal birth pattern in france from vagues de chaleur, fluctuations ordinaires des températures et mortalité en france depuis [heat waves, temperature fluctuations and mortality in france from timing of sexual intercourse in relation to ovulation. effects on the probability of conception, survival of the pregnancy, and sex of the baby the timing of the ''fertile window'' in the menstrual cycle: day specific estimates from a prospective study a re-look at the duration of human pregnancy length of human pregnancy and contributors to its natural variation we wish to express our sincere thanks to ms. ela klayman-cohen of the swedish ''statistical central agency'' (statistika central-byråns), ms. maija maronen of ''statistics finland'', mr. chihiro omori of the japanese ministry of internal affairs for their kind help in guiding us through the rich datasets of their respective countries. in this appendix we examine the implication of a constant birth rate on population and birth changes.the monthly birth rate is defined as: λ = b(t)/p(t) where b(t) is the number of births in month t. now let us apply changes ∆b and ∆p to the numerator and denominator respectively. how these changes must be connected if λ is to remain constant is shown by the following calculation. an explanation based on the number of marriages may be considered but in fact it can be quickly discarded. one may say that during the time of the epidemic people postponed planned marriages. as in sweden in any normal month there were about marriages a drastic reduction could in principle account nine months later for a fall in births of the same magnitude, that is to say of the size actually observed. however, it appears that in most countries there was only a slight reduction in the number of marriages or even none at all. thus, in sweden in october , at the height of the influenza epidemic, there were marriages compared to an average of in the same month of october october , october , even when there is a fall in marriages it has not necessarily an impact on the number of births months later. an illustration is provided by new york city in september . as a result of the world trade center attack the number of marriages fell from in august to in september. however the correlation between the time series of monthly percentage variations of marriages and conceptions over the months from to is as low as . (which, for a confidence level of . , is not a significant correlation). the corresponding linear regression reads: ∆b/b = . (∆m/m) + . which shows that the observed fall of % of marriages will translate in a fall of less than % for the births.in short, in some cases where there is both a substantial reduction in the number of marriages (which is rare) and a significant marriage-birth correlation, this factor may contribute but it cannot be the root factor that we are looking for. the standard length of pregnancy, namely weeks ( days) is counted from the woman's last period, not the date of conception which generally occurs two weeks later. for the present study we rather need the time d between the sexual intercourse which led to the conception and birth. the conception results from the encounter between a spermatozoon and an ovum (also called ovule or egg cell). ovulation means that the egg is released from the ovary into the fallopian tube; it remains there in good shape for only one day which means that conception and ovulation must take place almost on the same day. in contrast the spermatozoa can stay alive in the fallopian tube for days with similar conception probabilities during those days. once fertilized, the egg starts its journey down the fallopian tube and into the uterus where it will get implanted (see fig. c. ) .the key-question then is: when does ovulation occur? it is often said that it occurs at day in the menstrual cycle. in fact, this depends upon the length of the menstrual cycle [ ] . when the length of the cycle is days, ovulation occurs on average at day . when the cycle lasts more than days, the ovulation takes place on day . the global average for all cycles is days. coupled with the standard estimate of days starting from the beginning of the cycle one gets: − = days following ovulation. with respect to intercourse one gets time intervals of d = , , days with same probability for actually they can remain alive for about days but the probability of conception during these three extra days is only one third of the conception probability during the first days [ ] . key: cord- -x k apav authors: li, d.; croft, d. p.; ossip, d. j.; xie, z. title: are vapers more susceptible to covid- infection? date: - - journal: medrxiv : the preprint server for health sciences doi: . / . . . sha: doc_id: cord_uid: x k apav background covid- , caused by severe acute respiratory syndrome coronavirus (sars-cov- ), was declared a global pandemic in march . electronic cigarette use (vaping) rapidly gained popularity in the us in recent years. whether electronic cigarette users (vapers) are more susceptible to covid- infection is unknown. methods using integrated data in each us state from the behavioral risk factor surveillance system (brfss), united states census bureau and the point acres.com website, generalized estimating equation (gee) models with negative binomial distribution assumption and log link functions were used to examine the association of weighted proportions of vapers with number of covid- infections and deaths in the us. results the weighted proportion of vapers who used e-cigarettes every day or some days ranged from . % to . % for us states. statistically significant associations were observed between the weighted proportion of vapers and number of covid- infected cases as well as covid- deaths in the us after adjusting for the weighted proportion of smokers and other significant covariates in the gee models. with every one percent increase in weighted proportion of vapers in each state, the number of covid- infected cases increase by . ( % ci: . - . ) and the number of covid- deaths increase by . ( % ci: . - . ) in log scale in each us state. conclusions the positive associations between the proportion of vapers and the number of covid- infected cases and deaths in each us state suggest an increased susceptibility of vapers to covid- infections and deaths. novel coronavirus disease (covid- ) outbreak was declared a global pandemic by the world health organization (who) on march , . as of april , , there were over three million covid- infected cases and over , deaths globally. in the united state, the total number of infected covid- cases exceeded one million, with over , deaths reported by april , . covid- infection presents with cough, dyspnea and fever among other systemic symptoms and can lead to pneumonia and acute hypoxemic respiratory failure. electronic cigarettes (e-cigarettes), promoted as an alternative for cigarette smoking, rapidly gained popularity in recent years in the us. in , the prevalence of current e-cigarette use (vaping) in us adults was . %. recent studies on the associations of vaping and health symptoms/diseases have observed associations between vaping and symptoms of wheezing and self-reported chronic obstructive pulmonary disease (copd), along with increased inflammation in bronchial epithelial cells and alterations in the pulmonary immune response to infection. [ ] [ ] [ ] [ ] [ ] [ ] tobacco control researchers have raised concerns that vapers may be more susceptible to covid- infections and could develop more severe covid- symptoms. however, there is very limited evidence on the association between vaping and covid- infection. we will examine the association of vaping with covid- infections and deaths, using the integrated state-level weighted proportions of current e-cigarette users (vapers) from the behavioral risk factor surveillance system (brfss) survey data, the population size and land area in in each state from united states census bureau, and the daily number of covid- infected cases and deaths in each state from the point acres.com website during the time period from january , to april , in the united states. our study is the first one to provide evidence on the association of vaping with covid- infections and deaths at the us population level. we integrated the behavioral risk factor surveillance system (brfss) survey data at state level, the population size and land area in each state from the united states census bureau, and the covid- infected cases and deaths data from the point acres.com website at available dates from each state through the unique two letter state abbreviations. from the brfss survey, states provided information on the vaping status variable. the population size in each state in and land area in each state were obtained from the united states census bureau website. the covid- infected cases and deaths counts were available for each state from january , to april , . reports of negative numbers of infected cases and deaths were excluded from the covid- data. after integrating the brfss data and the census data with the covid- infected case and deaths from different dates at the state level, there were observations in the final analysis data. the vaping status variable was defined by the answers to the question "do you now use ecigarettes, every day, some days, or not at all?" in the brfss survey. subjects who now use e-cigarettes every day or some days were classified as vapers and subjects who responded all rights reserved. no reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint (which this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint that they use e-cigarettes "not at all" or "not applicable" were classified as non-vapers. the weighted frequency of vapers in each state was obtained using the proc surveyfreq procedure in sas version . (sas institute inc., cary, nc), considering the complex sampling design of the brfss survey. the weighted proportion of vapers in each us state was calculated using the ratio of weighted frequency of vapers and weighted frequency of total number of subjects in each state. the outcomes used in current analysis are the number of covid- infected cases and deaths. covariates considered in the current investigation include population size, population density (calculated using population size divided by land area), age, gender, race/ethnicity, education, income, mental health, physical health, obesity, respiratory disease (including asthma and copd), heart disease, cancer, stroke, diabetes, kidney disease, and smoking (currently smoke every day or some days). the number of covid- infected cases was also used as a covariate when modeling the covid- deaths. generalized estimating equation (gee) models with negative binomial distribution assumptions and log link functions were used to examine the association of weighted proportion of vapers with the number of covid- infections and deaths, after adjusting for the confounding effects from significant covariates. , the correlations of number of covid- infections and deaths from different dates within the same state were considered through the autoregressive (ar ( )) variance-covariance structure within the gee model framework. the purposeful covariates all rights reserved. no reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint (which this version posted may , . . selection method was used to select significant covariates for the gee models. variance inflation factor (vif) was used to examine the multicollinearities among the predictor variables in the gee models. a vif value of five or less was considered to indicate multicollinearity in the fitted gee model. all statistical analyses were conducted using statistical analysis software sas version . (sas institute inc., cary, nc) and r (r core team, ). significance levels for all tests were set at % for two-sided tests. the weighted proportion of vapers ranged from . % to . % for us states. male gender, poor physical health, cancer, and obesity had a negative association with the daily all rights reserved. no reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint (which this version posted may , . using integrated state level data obtained from the brfss survey, the united states census bureau and the point acres.com website, we were able to investigate the association of weighted proportion of vapers with covid- infected cases and deaths in the united states. we found a significant positive association of vaping with covid- infections and deaths at the state level controlling for sociodemographic and health related covariates. this finding all rights reserved. no reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint (which this version posted may , . thus, we cannot determine causality between vaping and covid- infections and deaths. however, prior research supports the biological plausibility of a relationship between vaping and an increased susceptibility to respiratory infection. multiple mouse models have observed an increased severity in infection associated with vaping exposure related to dysregulation of lung epithelial cells and an impaired immune response to both viral , and bacterial infection. bacterial superinfection of viral illnesses like influenza and covid- is especially dangerous, as this leads to an increased severity in illness a human cell based model of exposure to nicotine-free flavored e-liquid observed immunosuppressive effects and impaired respiratory innate immune cell function (alveolar macrophages, neutrophils, and natural killer cells). in humans, bronchoalveolar lavage samples from the airways of active vapers also revealed dysregulation of the airway's innate immune response including neutrophilic response and mucin a wide variety of flavorings are used by vapers, many of which, such as diacetyl, acetoin, pentanedione, o-vanillin, maltol, and coumarin in nicotine-free e-liquid, could also trigger inflammatory responses in human monocytes. finally, previous human studies found vaping is associated with increased risk of chronic bronchitic symptoms (chronic cough, phlegm, or all rights reserved. no reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint (which this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint bronchitis). [ ] [ ] [ ] and epidemiologic studies observed an increased risk of self-reported wheezing and copd. , with the ongoing covid- pandemic, particular health concerns have been raised regarding vaping, such as whether vapers have higher risk for covid- infection and could develop more severe symptoms once contracted covid- . to our best knowledge, this is the first population-based study to empirically examine and find an association between vaping and covid- . the existing literature on the increased risk of respiratory infection in combustible cigarette smokers was summarized by a prior meta-analysis finding an increased risk of current smokers for influenza infection compared to non-smokers. as covid- is a novel condition, the literature examining the risk of smokers for covid- is scant. a recent study based on covid- patients found smoking history was associated with covid- severity. a recent systematic review on covid- and smoking concludes that smoking is likely associated with worse outcomes in covid- . however, other studies indicate that smoking might not be associated with the incidence and severity of covid- , for example, a recent meta-analysis based on chinese patients suggests that active smoking is not associated with severity of covid- . it remains unclear whether nicotine has a role in the either the increased or decreased severity of illness for smokers with covid- . our study did not find a significant association between the weighted proportion of smokers and the number of covid- infections and deaths at state level. due to the incomplete testing and tracking of home deaths, it is possible that a percentage of older smokers with comorbidities are dying at home from covid- and therefore are not captured into the reported covid- infections and death data , another possibility is that smokers with comorbidities are homebound and more likely to be all rights reserved. no reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint (which this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint strictly following social/physical distancing guidelines, and are therefore reducing their risk for covid- . currently, there is limited evidence on the susceptibility of smokers to covid- infection and whether smokers have a worsened course in the setting of covid- . additionally, it is unknown whether or not covid- virus could be transmitted to those surrounding smokers through passive smoking and vaping. more epidemiological and clinical studies are needed to investigate the association of smoking with covid- infections and deaths. we found states that had a larger weighted proportion of subjects who had less than high school education had a higher number of covid- infections. researchers from the university of southern california found that americans who had less than high school education had a lower perceived risk of exposure to covid- and a higher perceived risk of deaths than those who have college or higher degrees. this might explain, in part, the positive association between the weighted proportion of less than high school education with the number of covid- infections. we also found states that had a larger proportion of non-hispanic blacks and hispanics had a larger number of covid- deaths. the covid- deaths rate data from washington d.c. and us states reported through april , showed that non-hispanic blacks ( . %) and hispanics ( . %) had the highest covid- deaths rate. this could be related to the higher proportion of chronic conditions such as hypertension, heart disease and diabetes in non-hispanic blacks and hispanics. , we also found that states having a higher proportion of respiratory disease such as asthma or copd had a higher number of covid- deaths, which indicated that respiratory disease such as asthma and copd could potentially increase the risk of covid- deaths. there are several limitations in current study. one limitation is that the weighted proportions of vapers, smokers, and other demographic and chronic diseases are from the brfss data, which might differ from the estimates. the reported covid- infected cases and deaths obtained from point acres.com website could be subject to some reporting errors as we noticed some negative number of covid- infected cases and deaths, which we excluded from further analysis. however, we compared the covid- data obtained from point acres. was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint (which this version posted may , . the content is solely the responsibility of the authors and does not necessarily represent the official views of the nih or the fda. the authors declare no competing interests. the behavioral risk factor surveillance system (brfss) survey data are publicly available from the centers for disease control and prevention website (https://www.cdc.gov/brfss/annual_data/annual_ .html). the state population in and the land area in each state were obtained from the united states census bureau website all rights reserved. no reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint (which this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint (https://www.census.gov/). the covid- infected cases and deaths data were requested and obtained from the point acres.com website (https://coronavirus. point acres.com/en). • current e-cigarette use is positively associated with covid- infections. • current e-cigarette use is positively associated with covid- deaths. • this study emphasizes the importance of studying the susceptibility of current e-cigarette users to covid- infection and death. who declares covid- a pandemic an interactive web-based dashboard to track covid- in real time current status of epidemiology, diagnosis, therapeutics, and vaccines for novel coronavirus disease (covid- ) prevalence of e-cigarette use among adults in the united states vaping away epithelial integrity electronic cigarette vapour increases virulence and inflammatory potential of respiratory pathogens e-cigarette flavored pods induce inflammation, epithelial barrier dysfunction, and dna damage in lung epithelial cells and monocytes electronic cigarette liquid increases inflammation and virus infection in primary human airway epithelial cells association of smoking and electronic cigarette use with wheezing and related respiratory symptoms in adults: cross-sectional results from the population assessment of tobacco and health (path) study, wave use of electronic cigarettes and self-reported chronic obstructive pulmonary disease diagnosis in adults public health concerns and unsubstantiated claims at the intersection of vaping and covid- a generalized estimating equations approach to mixed-effects ordinal probit models modelling correlated data: multilevel models and generalized estimating equations and their use with data from research in developmental disabilities purposeful selection of variables in logistic regression applied linear-regression models -neter public health consequences of e-cigarettes electronic cigarettes disrupt lung lipid homeostasis and innate immunity independent of nicotine electronic cigarette inhalation alters innate immunity and airway cytokines while increasing the virulence of colonizing bacteria influenza and bacterial superinfection: illuminating the immunologic mechanisms of disease flavored e-cigarette liquids and cinnamaldehyde impair respiratory innate immune cell function e-cigarette use causes a unique innate immune response in the lung, involving increased neutrophilic activation and altered mucin secretion what are the respiratory effects of ecigarettes? electronic cigarette use and respiratory symptoms in chinese adolescents in hong kong electronic cigarette use and respiratory symptoms in adolescents cigarette smoking and the occurrence of influenza -systematic review clinical characteristics of coronavirus disease in china covid- and smoking: a systematic review of the evidence sex difference and smoking predisposition in patients with covid- active smoking is not associated with severity of coronavirus disease (covid- ) what is happening to non-covid deaths the association of smoking status with sars-cov- infection, hospitalisation and mortality from covid- : a living rapid evidence review racial/ethnic disparities in prevalence, treatment, and control of hypertension among us adults following application of the american college of cardiology/american heart association guideline age-stratified prevalence, treatment status, and associated factors of hypertension among us adults following application of the acc/aha guideline key: cord- -u u mjz authors: wang, jiwei; lai, bin; nanayakkara, gayani; yang, qian; sun, yu; lu, yifan; shao, ying; yu, daohai; yang, william y.; cueto, ramon; fu, hangfei; zeng, huihong; shen, wen; wu, susu; zhang, chunquan; liu, yanna; choi, eric t.; wang, hong; yang, xiaofeng title: experimental data-mining analyses reveal new roles of low-intensity ultrasound in differentiating cell death regulatome in cancer and non-cancer cells via potential modulation of chromatin long-range interactions date: - - journal: front oncol doi: . /fonc. . sha: doc_id: cord_uid: u u mjz background: the mechanisms underlying low intensity ultrasound (lius) mediated suppression of inflammation and tumorigenesis remain poorly determined. methods: we used microarray datasets from ncbi geo dataset databases and conducted a comprehensive data mining analyses, where we studied the gene expression of cell death regulators that regulate different cell death types (cell death regulatome) in cells treated with lius. results: we made the following findings: ( ) lius exerts a profound effect on the expression of cell death regulatome in cancer cells and non-cancer cells. of note, lius has the tendency to downregulate the gene expression of cell death regulators in non-cancer cells. most of the cell death regulator genes downregulated by lius in non-cancer cells are responsible for mediating inflammatory signaling pathways; ( ) lius activates different cell death transcription factors in cancer and non-cancer cells. transcription factors tp- and srf- were induced by lius exposure in cancer cells and non-cancer cells, respectively; ( ) as two well-accepted mechanisms of lius, mild hyperthermia and oscillatory shear stress induce changes in the expression of cell death regulators, therefore, may be responsible for inducing lius mediated changes in gene expression patterns of cell death regulators in cells; ( ) lius exposure may change the redox status of the cells. lius may induce more of antioxidant effects in non-cancer cells compared to cancer cells; and ( ) the genes modulated by lius in cancer cells have distinct chromatin long range interaction (clri) patterns to that of non-cancer cells. conclusions: our analysis suggests novel molecular mechanisms that may be utilized by lius to induce tumor suppression and inflammation inhibition. our findings may lead to development of new treatment protocols for cancers and chronic inflammation. ultrasound, alone or combined with contrast agent microbubbles, have numerous applications ranging from being a well-established diagnostic tool ( , ) to a method of drug delivery ( ) . the application of microbubbles and ultrasound to deliver nanoparticle carriers for drug and gene delivery is a research area that has expanded greatly in recent years. recent studies reported that utilization of ultrasound contrast microbubbles causes the so-called "sonoporation" effect ( , ) , which has been recognized as a significant factor in transient disruption of cell membrane permeability ( ) that allows easier transport of extracellular compounds into the cytoplasm of viable cells ( ) . ultrasound therapy now is widely used in clinical practice, and clinical/translational research in the treatment of various human malignancies and pathologies including breast cancer, leukemia, lymphoma, melanoma, pancreatic neuroendocrine tumors ( ) , hepatic cancer, nasopharyngeal cancers, glioma, ovarian cancer, colon cancer, gastric cancer ( ), sarcoma ( ) ( ) ( ) , stroke ( ) , prostatic hyperplasia, renal masses ( ), treatment of abdominal subcutaneous adipose tissue ( ) , bone repair ( ) , osteoarthritis ( ) , and carpal tunnel syndrome ( ) . so far, several therapeutic ultrasound formats have been developed including high intensity focused ultrasound ( ) and low-intensity pulsed ultrasound ( ) . recently, several clinical trials and experimental data verified the ability of ultrasound to elicit anti-inflammatory and tissue repair/regeneration responses ( , ) , suggesting the potential of using ultrasound as a novel therapeutic method ( , ( ) ( ) ( ) ( ) . cell death pathways have gained attention as novel therapeutic targets for treatment of cancers ( , ) and inflammation ( , ) . we and others have reported that activated t cells survive by upregulating anti-cell death proteins such as bclxγ ( ) , translationally controlled tumor protein (tctp) ( ) and inhibiting severe acute respiratory syndrome (sars) virus e protein ( ) . moreover, cd +cd +foxp + regulatory t cells readily undergo apoptosis ( ) by upregulating proapoptotic protein bax ( ) and downregulating anti-apoptotic regulator tctp ( , ) . in addition, metabolic disease-risk factors such as hyperlipidemia ( ) ( ) ( ) , hyperhomocysteinemia ( ) and chronic kidney disease ( ) accelerate vascular diseases by inducing inflammatory cell death (pyroptosis) or pyrop-apoptosis ( , ) . interestingly, lius was reported to induce apoptosis, autophagy or necrosis in cancer cells including nasopharyngeal cancer cells ( ) , laryngeal cancer cells ( ) , hepatocellular cancer cells ( , ) , leukemia cells ( ) , lymphoma cells ( ) , and osteosarcoma cells ( ) . in contrast, lius affects various non-cancer immune cells and other cells by regulation of inflammation. for example, ultrasound promotes vasodilation, enhances blood flow, promotes fibroblast and osteoblast proliferation, and increases other cellular components leading to wound healing ( ) . moreover, lius was reported to suppress synovial cell proliferation ( ) , affect mesenchymal stem cell migration ( ) , enhance the regeneration of myofibers ( ) , reduce the expression of inflammatory mediators ( ) , promote skin fibroblast proliferation ( ) , and chondrocyte and osteoblast proliferation ( ) . however, it remains unclear how lius can distinguish cancer cells from non-cancer cells and induce differential biological responses. further, molecular and cellular mechanisms underlying lius mediated inflammation inhibition and cancer suppression effects are not understood. in order to broaden our understanding of lius-mediated effects in cellular context, we hypothesized that lius may induce differential gene expression patterns of cell death regulators in cancer cells and non-cancer cells. therefore, in this study, we analyzed the expression pattern of cell death regulators in lius-treated cancer cells and non-cancer cells. these genes were responsible for regulating different types of cell death mechanisms including apoptosis, mptdriven necrosis, necroptosis, ferroptosis, pyroptosis, parthanatos, entotic cell death, netotic cell death, lysosomal dependent cell death (lcdc), autophagy dependent cell death (adcd), immunogenic cell death (icd), mitotoic death, and anoikis. herein, we will refer to the cell death regulators of the different types of cell death pathways that we included in this study as the cell death regulatome. our data indicated that lius exerts a differential gene expression pattern of the cell death regulatome in cancer cells and non-cancer cells. furthermore, our data implies that thermal effects and osmotic shear stress (oss) associated with lius may potentially play a role in inducing the differential gene expression patterns of the cell death regulatome that we observed. also, we observed that lius has the tendency to induce antioxidant effects in non-cancer cells, which may also contribute to the differential gene expression patterns of the cell death regulatome in lius-treated cancer cells and non-cancer cells. most interestingly, we observed that the cell death regulator genes modulated by lius in cancer cells and non-cancer cells have unique chromatin long range interaction (clri) sites. chromatin looping enables clris, that gives the opportunity to gene promoters to interact with distal regulatory elements ( ) . rapid development of technologies such as chromosome conformation capture-sequencing ( c-seq) ( ) , circularized chromosome conformation capture-sequencing ( c-seq) ( , ) and chromosome conformation capture carbon copy-sequencing ( c-seq) ( ) that capture chromosome conformation allow determination of interactions between the target genes and clri sites. the clri may enhance and modulate the expression of genes of interest. differences in chromatin long-range interaction patterns between genes have previously been hypothesized to influence alternative splicing ( ) and the transcription of inflammatory genes such as cytokine ( ) , cytokine receptor ( ) and cardiovascular disease-causative genes ( ) . therefore, we suggest that the unique clri seen in genes modulated by lius in cancer cells and non-cancer cells may play a role in producing a differential response. expression profile of cell death genes in ultrasound-treated, mild hyperthermia-treated, and oscillatory shear stress-treated cells microarray datasets were collected from national institutes of health (nih)-national center for biotechnology information (ncbi) -geo dataset (https://www.ncbi.nlm.nih.gov/gds/) databases and analyzed with geo r (https://www.ncbi.nlm. nih.gov/geo/geo r/). the numbers of geo datasets that are as follows: gse , gse , gse , gse , gse , gse , gse , gse , gse . the detailed information of these geo datasets was shown in table . two hundred and ninety nine cell death regulators that participate in different types of cell death were analyzed in our study, and cell death regulators regulate multiple cell death types. of note, several cell death forms are reported recently, including mpt-driven necrosis, parthanatos, entotic cell death, netotic cell death, lysosome dependent cell death, immunogenic cell death, mitotic death, and anoikis. all detailed information of these cell death regulators and types of cell death was shown in table . as we reported ( , ) , we applied a statistical method similar to that meta-analysis and analyzed the expression of four housekeeping genes (chmp a, psmb , actb, and gapdh) in all geo datasets regardless of species that were chosen for this study. the house-keeping gene list was extracted from related report ( ) . briefly, the variations between the expression of housekeeping genes between treatment and control groups vary from − . to . . as this variation was very narrow, we concluded that the datasets ( table ) are of high quality. the target genes with expression change more than . fold were defined as the upregulated genes, while genes with their expression decreases more than . -fold were defined as downregulated genes. we utilized ingenuity pathway analysis (ipa, ingenuity systems, http://pages.ingenuity.com/rs/ingenuity/images/ ipa_data_sheet.pdf) to characterize clinical relevance, and molecular and cellular functions related to the identified genes in our microarray analysis. the differentially expressed genes were identified and uploaded into ipa for analysis. the core and pathways analysis was used to identify molecular and cellular pathways as we have previously reported ( , ) . the chromatin long-range interaction data were collected from the hi-c data deposited in the d genome database (https:// dgenome.research.chop.edu) as a tabulated text file ( ) . as we reported ( , ) , we extracted the data related to interact gene and interaction sites interacting with lius regulated-cell death genes, then calculated the distances between the interaction sites and lius-modulated gene promoters. the resulting filtered data was imported into microsoft excel and raw interaction distances calculated as the differences between gene start coordinates. an table s ldcd ros , dram , stat , ctsb, ctsl, serpina g, bid, bax, bcl , xiap, prtn , hspa a see table s adcd table s total gene number - = the latest definitions, morphological features, main molecular features, and detection methods of different cell death types (see tables s , s ). awk script was used to determine whether the lius-modulated gene promoters were downstream or upstream of its partner in each interaction pair; and to add this information to the data file. the signs of distance values were then updated, with downstream entries designated as positive; and upstream values designated as negative. distance distributions for all upregulated and all downregulated lius-modulated genes were compared by groups overall, respectively. these housekeeping genes were extracted from related report (pmid: ). chmp a, charged multivesicular body protein a; psmb , proteasome subunit beta ; actb, actin beta; gapdh, glyceraldehyde- -phosphate dehydrogenase. exerts other therapeutic effects in non-cancer cells such as modulation of cell proliferation, regulation of cell migration, enhancement of regeneration etc. however, how cancer cells and non-cancer cells produce a differential response to lius remain unknown ( ) . a previous study reported a comparison of the effects of lius treatment in cancer cells including breast cancer melanoma, lung cancer with non-cancer cells such as foreskin fibroblasts, and amniotic fluid epithelial cells using cell death assay. however, as we pointed out in table , this comparison gives limited information; and the mechanisms that induce differential signaling response in cancer cells and non-cancer cells were not examined in detail ( ) . we hypothesized that lius induces differential responses in cell death pathways in cancer cells vs. non-cancer cells by modulating the gene expression of cell death regulators. to examine this hypothesis comprehensively, we collected genes that regulate all the cell death types including apoptosis, mitochondrial permeability transition (mpt)-driven necrosis, necroptosis, ferroptosis, pyroptosis, parthanatos, entotic cell death, netotic cell death, lysosome dependent cell death (ldcd), autophagy dependent cell death (adcd), immunogenic cell death (icd), mitotic death, and anoikis ( ) ( table ) . as mentioned above, all the cell death regulators that we included in this study are collectively referred to as cell death regulatome. of note, among genes, genes regulate multiple types of cell death. in addition, we found three microarray datasets deposited in the nih-ncbi geodatasets database, which were conducted on human lymphoma cells treated with lius and two non-cancer cells such as mouse mc t -e pro-osteoblast cell and rat bone marrow cell, treated with lius ( table ) . as shown in table , among the genes analyzed, lius induced upregulation of genes in lymphoma cells including bok, casp , cyld, dpp , ezr, atg , atg l , vamp , cxcl , anxa , fpr , panx , and tp ; and downregulated genes such as hk , casp , tlr , map k , cd , atm, cdc c, ttk, src, pdgfra, itgb , and itgb . when conducted ingenuity pathway analysis (ipa) on upregulated and downregulate genes by lius treatment in cancer cells, it revealed that none of the signaling pathways modulated by the two groups were shared (figure ). therefore, this may suggest that the cell death regulators modulated by lius treatment activate intricate signaling mechanisms that may cause an impact on cell death in cancer cells. lius induced the expression of cell death regulator genes in non-cancer cells including myc, srf, nt e, bag , numa , ireb , ptk , pdgfra, and akt ; and downregulated genes among genes examined ( table ). similar to the observation seen in cancer cells, the ipa analysis demonstrated that lius induced upregulated and downregulated cell death regulators in non-cancer cells activate distinct cell signaling pathways. of note, the cell death regulators that were downregulated by lius treatment seem to attenuate several pro-inflammatory pathways including neuroinflammation signaling pathway, inflammasome pathway, nf-κb signaling, trem signaling, role of pattern recognition receptors in recognition of bacteria and viruses, role of macrophages, fibroblasts and endothelial cells in rheumatoid arthritis, hmgb signaling, toll-like receptor signaling, altered t cell and b cell signaling in rheumatoid arthritis, communication between innate and adaptive immune cells, hepatic fibrosis/hepatic stellate cell activation, type i diabetes mellitus signaling, and role of osteoblasts, osteoclasts and chondrocytes in rheumatoid arthritis (figure ). therefore, this data suggests that lius treatment may significantly impact the inflammatory status of non-cancer cells. taken together, these results suggest that first, lius exerts a profound effect on cell the gene expression of the cell death regulatome in cancer cells and non-cancer cells; second, lius treatment has the tendency to reduce the expression of cell death regulators than activating them in non-cancer cells; third, in non-cancer cells, the death regulators that are downregulated by lius attenuate several inflammatory signaling pathways. these findings are well-correlated with our previous publication where we reported that lius mediated upregulation of antiinflammatory regulators ( ) . furthermore, this data suggest that the majority of lius-downregulated cell death regulators in noncancer cells are functional as inflammatory regulators, which confirmed our updated understanding that inflammation and cell death pathways are often inter-connected ( ) . for example, the binary classification of mammalian cell death regulators, such as caspases, as either apoptotic or inflammatory ( ) is now obsolete. emerging data indicate that all mammalian caspases are intricately involved in the regulation of inflammation and immunity ( ) . as mentioned above, we observed that cell death regulatome is differentially regulated in cancer cells and non-cancer cells when exposed to lius treatment. in order to explain this observation, we hypothesized that lius may modulate the expression of cell death-related transcription factors in both cancer and non-cancer cell groups. we found that tumor suppressor gene tp ( ) was upregulated by lius treatment in cancer cells; and that serum response factor (srf) was upregulated by lius in noncancer cells ( table ) . to determine whether lius-induced tp plays roles in regulating the cell death regulator genes modulated by lius, we examined the expression of cell death regulators in tp gene deficiency cell microarray datasets. as shown in table , we found that a list of lius-downregulated cell death regulators such as cd , protein kinase ataxiatelangiectasia mutated (atm) ( ), caspase- (casp ) ( ), casp , cd , and bcl were upregulated in human tp knock-down (kd) mesenchymal stem cell datasets. these results suggest that lius has the potential to regulate cell death regulator expression in cancer cells partially via inducing tumor suppressor transcription factor tp as we reported previously ( ) . moreover, we found that srf deficient cardiomyocytes express higher levels of hepatocyte growth factor (hgf), which was attenuated by lius treatment in non-cancer cells. hgf plays a central role in metabolic disorders such as insulin resistance and in diabetes pathophysiology ( ) . therefore, upregulation of srf may serve as the suppressive mechanism underlying attenuation of hgf seen in lius treated noncancer cells. taken together, our results have demonstrated that modulation of transcription factors such as tp- and srf- may play a role in altered cell death regulator gene expression seen in lius-treated cancer cells and non-cancer cells, respectively. it is well-accepted that the therapeutic applications of ultrasound depend on the propagation of ultrasound waves through tissues to produce biological effects (figure ) . the biological effects of ultrasound are separated into thermal and non-thermal effects. the thermal effects of ultrasound that arise from the absorption of ultrasonic energy and creation of heat depend on ultrasound exposure parameters, tissue properties and beam configuration. cavitation, acoustic radiation force, radiation torque, acoustic streaming, shock wave and shear stress are considered non-thermal effects of ultrasound ( ) although there are some different opinions in classification of cavitation ( , , ) . lius is a form of ultrasound that delivered at a much lower intensity (< w/cm ) than high intensity focus ultrasound, and it has been considered as removed thermal component or minimal thermal effects due to its low intensity mode ( , ) . however, some publications have reported an increment of temperature by approximately to • c after lius treatment ( ) , meaning that thermal effect is inevitable during lius treatment. we also examined the expressions of heat shock proteins in the three lius-treated microarrays and found that dnaj (hsp ) heat shock protein (gak) is upregulated ( . -folds, p = . ) in one of the microarray datasets (human lymphoma, gse ) but is not significantly modulated in other two lius-treated microarray datasets. these results suggest that lius treatment induces the upregulation of heat shock protein gene and potential thermal stress; and that the differences in the expressions of gak in three microarrays may be due to the potential differences of used lius methods and parameters as well as cell types ( table ) . we then hypothesized that thermal effects partially underlie the lius induced modulation of the cell death regulatome. to examine this hypothesis, we analyzed the microarray datasets conducted on cancer cells and non-cancer cells treated with mild hyperthermia ( table ) . table table ) . interestingly, we found that similar to our observation in lius treated cancer cells, mild hyperthermia induced the expression of transcription factor tp- in cancer of note, lius has the tendency to downregulate cell death regulators in non-cancer cells. frontiers in oncology | www.frontiersin.org the red highlights indicate increased fold changes, and green highlights indicate decreased fold changes. the biological effects exerted by ultrasound therapy includes thermal and non-thermal effects. the thermal effects of ultrasound that arise from the absorption of ultrasonic energy; and creation of heat depend on ultrasound exposure parameters, tissue properties, and beam configuration. cavitation, acoustic radiation force, radiation torque, acoustic streaming, shock wave, and shear stress are considered non-thermal effects of ultrasound. cavitation is perhaps the most widely studied biological effect and is described as the formation and oscillation of a gas bubble. in addition, the oscillation of the bubble can also result in heat generation. cells. additionally, we observed that the expression of cell death regulators such as cyld, ezr, vamp , anax , and fpr was increased in mild hyperthermia treated cancer cells, which was similar to lius treated cancer cells. further, we observed that expression of tlr , map k , and src were downregulated in both data sets. therefore, there is a possibility that mild hyperthermia associated with lius may regulate the expression of a few cell death regulators in cancer cells. also, upregulation of the transcription factor tp- may occur due to mild hyperthermia associated with lius in cancer cells. similarly, we observed that the expression of gclc, nlrp and cd are downregulated in both groups of non-cancer cells treated with either lius or mild hyperthermia. furthermore, we analyzed whether there are similarities between the signaling pathways that are regulated by genes that were modulated by lius and mild hyperthermia. our analysis revealed that there are three common signaling pathways that were shared by cell death regulator genes that were upregulated by lius and mild hyperthermia in cancer cells (figure ) . these cell signaling pathways are granulocyte adhesion and diapedesis, huntington's disease signaling, and molecular mechanisms of cancer. in addition, the down regulated cell death regulators in cancer cells treated by lius or mild hyperthermia shared signaling pathways such as macropinocytosis signaling and virus entry via endocytotic pathway (figure ) . despite the fact that there are no common upregulated genes attributed to both groups, in non-cancer cells cell death regulators that are modulated by lius and mild hyperthermia shared five signaling pathways. these are pten signaling, myc mediated apoptosis signaling, molecular mechanisms of cancer, il- signaling, and pedf signaling. furthermore, cell death regulators that were downregulated by lius treatment or mild hyperthermia in non-cancer cells affected four signaling pathways including neuroinflammation signaling pathway, inflammasome pathway, trem signaling, role of pattern recognition receptors in recognition of bacteria and viruses, and dha signaling regulation. taken together, these results suggest that unlike what was reported previously, thermal effect may play an indispensable role for the lius induced modulation of cell death regulators. we also hypothesized that non-thermal effects such as oscillatory shear stress (oss) partially underlie lius induced modulation of cell death regulators. to examine this hypothesis, we utilized microarray datasets performed on non-cancer cells treated with oss ( table ) . of note, we did not find any microarray datasets that were conducted on cancer cells treated with oss. oss-induced upregulation of out of cell death regulators involved in four cell death types such as apoptosis, ferroptosis, icd and anoikis and downregulated out of cell death regulators in non-cancer cells involved in ten out of cell death types ( table ). these results suggest that similar to lius, oss has the tendency to downregulate the expression of cell death regulators in non-cancer cells. nevertheless, we could not find a similarity between the genes that were upregulated by oss and lius treatment in non-cancer cells. however, genes including birc , cdk , e f , casp , and bub were downregulated in both datasets. similar to lius treatment, we observed that oss treatment did not affect the expression of cell death regulators involved in three cell death types such as mpt-driven necrosis, parthanatos, and netotic cell death. despite the fact the oss and lius treatment did not modulate similar set of genes in noncancer cells, we found that the cell death regulator genes that were upregulated by both treatments share three signaling pathways (figure ) . these three pathways are egf signaling, endometrial cancer signaling, and molecular mechanisms of cancer. interestingly, the cell death regulators, that are downregulated by lius or oss treatment in non-cancer cells, downregulate nine common pathways. these signaling pathways are trem signaling, role of pattern recognition receptors in recognition of bacteria and viruses, cyclins and cell cycle regulation, pancreatic adenocancer signaling, estrogen-mediated s-phase entry, communication between innate and adaptive immune cells, p signaling, and role of chk proteins in cell cycle checkpoint control. taken together, these results suggest that non-thermal effects of lius such as oss may be responsible for inducing biological effects observed in noncancer cells. oxidative stress is well-known to be involved in a wide variety of human diseases including cardiovascular diseases, inflammatory disorders, immune system dysfunction, diabetes, cancer, aging and neurodegenerative disorders ( ) ( ) ( ) ( ) . low grade reactive oxygen species (ros) generation helps to mediate signaling that maintain the homeostasis of the cells, however, excess production of ros can be detrimental to the cells by activating cell death pathways. previously, it was shown that lius exerts cytoprotective effects against oxidative injury in human aortic endothelial cells and retinal pigment epithelium cells ( , ) . in contrast, lius exerts cancer suppressing effects by inducing ros generation, which result in mitochondrial damage and subsequent cell death ( , , , ) . therefore, we analyzed whether lius treatment exerts any expression changes of genes in our datasets that has the potential to regulate oxidative stress (table ) . interestingly, we observed that lius treatment in cancer cells induced the expression of genes that regulate ros generation while downregulating the expression of gpx that has the potential to exert antioxidant effects. therefore, this suggests that lius may cause changes in the redox status in cancer cells. further, we observed that lius has the tendency to induce antioxidant effects in non-cancer cells by attenuating the gene expression of nos and nos , which promote ros generation, while promoting the expression of antioxidant genes gpx and gpx . it is well-known that ros can indeed impact the cell viability by affecting activity figure | the common signaling pathways that were shared by cell death regulators that were modulated by mild hyperthermia, oss, and lius. the top pathways that are modulated by mild hyperthermia and oscillatory shear stress in cancer cells and non-cancer cells were shown in figures s a-f. of caspases, mitochondrial function and activating apoptotic pathways depending on the cellular contexts ( ) . therefore, it is a possibility that lius-mediated changes in the redox status of the cancer and non-cancer cells may have an impact on cell death regulator gene expression leading to activation of various cell death pathways. further, we analyzed whether mild hyperthermia and oss has the potential to change the redox status by regulating ros generating and antioxidant genes. we observed that there were changes in genes that regulate ros generation and exert antioxidant effects in cancer cells; and to our surprise, we observed an induction of ros generating genes in non-cancer the gene list of chromosome conformation regulators was extracted from related report (pmid: ), detailed information (see table s ). the red highlights indicate increased fold changes, and green highlights indicate decreased fold changes. cells with treated with mild hyperthermia. however, our data revealed that oss has the tendency to exert anti-inflammatory effects in non-cancer cells, similar to our observation in lius treated non-cancer cells. therefore, it can be postulated that oss may impact the lius-mediated redox status in noncancer cells. the results from this study, our previous study ( ) and others' reports indicated that lius regulates gene expression presumably at transcription levels. our recent further reports showed that histone modification enzymes are significantly modulated in response to disease risk factor stimulations ( ) ; that il- suppresses endothelial cell activation by inhibiting mitochondrial reactive oxygen species-mediated site specific acetylation of histone lysine ( ) ; and that dna damage factors and dna repair factors serve as an integrated sensor and cell fate determining machinery for all the intracellular stresses and dangers ( ) . our reports suggest that various nuclear programs control gene expression responses to endogenous and exogenous damps and other stimuli including lius. we hypothesized that newly characterized chromatin long range interactions (clri) differentially regulate the gene promoters to differentiate lius-modulated gene expression in cancer cells vs. non-cancer cells. to test this hypothesis with respect to lius effects in modulating chromatin remodeling, we examined the expression changes of chromatin insulator-binding factors such as ctcf and rad and other promoter-binding factors and non-promoter binding factors in lius-treated cancer cells and non-cancer cells ( ) . as shown in table , lius did not change the expression of two insulator-binding factors, promoter-binding factors but changed the expression of one of six non-promoter binding factors in cancer cells. in addition, lius changed the expressions of two out of promoter binding factors in non-cancer cells. lius upregulated the non-promoter-binding factors in carcinoma cells, but upregulated the promoter-binding factors in non-carcinoma cells. the gene list of chromosome conformation regulators was extracted from related report (pmid: ), detailed information (see table s ). we further hypothesized that the differential gene expression seen between lius treated cancer cells and non-cancer cells were due to differences in the clris of the modulated genes figures a-c . therefore, to analyze this, we obtained chromatin long-range interaction data for all the significantly modulated cell death regulator genes from the dgenome database. this is a well-accepted database, which contains information on a huge collection of , , experimentally-derived chromatin longrange interactions ( ) . we then calculated the distances between interacting sites ( figure d ) with respect to lius-modulated gene promoters. if the lius-modulated gene promoter was located downstream of its long-range interaction partner, we designated the clri as negative. if the clri site is located downstream of the target gene promoter, we designated the interaction as positive. the two-sample kolmogorov-smirnov test of the chromatin long-range interaction distances between gene promoters corresponding to lius downregulated and upregulated genes indicated some significant differences between the two distance distributions (p < . ) ( figure e) . our data indicated that the majority of clri sites for cell death regulator genes that were modulated by lius treatment were concentrated between − and − base pairs (bp) upstream or and bp downstream in cancer cells. interestingly, we also found that the clri sites of the genes that were upregulated by lius were spanned between a narrower range compared to that of the downregulated genes in cancer cells ( figure e) . however, in non-cancer cells, most of the clri sites of cell death regulator genes modulated by lius were located downstream of the target genes. similar to the observation seen in genes modulated by lius in cancer cells, the clris of upregulated cell death regulators spanned across a comparatively narrower region than that of downregulated genes in lius treated non-cancer cells. of note, lius induced upregulation of rna polymerase iii subunit d (polr d) and rna polymerase iii subunit f (polr f), two out of promoter binding factors in non-cancer cells in table , are associated lius-induced higher concentrations of clriss. of note, a common feature of the clriss with lius modulated genes in cancer cells is associated with lius upregulation of gata in table . future experiments will be needed to verify these interesting associations between clri sites and the genes that were modulated by lius treatment in both cancer and non-cancer cells. since the dgenome database contains the experimental data derived from human non-aortic endothelial cells ( ) , the future work will be needed to use circular chromosome conformation capture sequencing ( c-seq) to examine liustreated cancer cells and non-cancer cells to map the specific upstream interaction sites for modulation of cell death regulator expression in cancer cells and non-cancer cells. taken together, our results have demonstrated for the first time that, lius induced a differential gene expression pattern in the cell death regulatome in cancer cells and non-cancer cells, and that these genes have unique clri sites. therefore, our results may suggest that optimal clriss may serve as new therapeutic targets in the future to enhance lius-mediated cancer cell suppression and lius anti-inflammatory functions in noncancer cells. therapeutic applications of ultrasound in addition to its use in diagnosis, have been accepted to be clinically beneficial. these benefits include that lius functions in suppressing cancers via inducing cell death pathways ( table ) . as we pointed out previously ( ), the anti-inflammatory effects are responsible for inducing the clinical benefits mediated by lius in non-cancer cells ( ) ( ) ( ) . however, the molecular mechanisms underlying the anti-cancer cell functions and anti-inflammatory effects of lius remain poorly defined. determination of the novel molecular mechanisms underlying the anti-cancer cell functions the interacting sites mostly located in upstream zone of promoters in cancer cells, but higher percentages located in downstream zone of promoters in non-cancer cells. (e) the long-range interaction sites of lius regulated genes in cancer cells mostly located between − and − upstream, but in non-cancer cells, the long-range interaction sites of lius regulated genes were concentrated between and bp downstream. and anti-inflammatory properties of lius in non-cancer cells would significantly improve our understanding on this important issue, and allow for the improvement of lius-based therapeutics. to fill in this important knowledge gap, in this study, we used cutting-edged molecular database mining approaches that we pioneered in ( , ( ) ( ) ( ) . our data analyses have made for the first time the following significant findings: ( ) lius differentially regulates cell death regulatome gene expression in cancer cells and in non-cancer cells. however, in non-cancer cells, lius treatment has the tendency to downregulate the expression of cell death regulators more. most of the cell death regulator genes downregulated by lius in non-cancer cells are responsible for mediating inflammatory signaling pathways; ( ) lius inhibits the expression of several inflammatory cell death regulators potentially via tp -, and srf-, mediated pathways in cancer cells and non-cancer cells, respectively; ( ) thermal effects and osmotic shear stress associated with lius may play a role in altering the cell death regulator expression patterns; ( ) lius has the tendency to induce antioxidant effects specifically in noncancer cells; and ( ) the genes that were modulated in cancer cells by lius have unique clri patterns, different from that of non-cancer cells. it is not clear how lius exposure may transmit signals to the nucleus to modulate the gene expression in both cancer and non-cancer cells. previously, it was shown that lius can overstretch the cell membrane and cause reparable submicron pore formation ( ) . this phenomenon is called sonoporation. such effects may lead to disruption of cytoskeleton in tandem because this network of subcellular filaments is physically interconnected with the plasma membrane ( ) . therefore, sonoporation associated with lius may be responsible for inducing important biological effects in cells. in addition, ultrasound at low diagnostic power can cause stable oscillations of the microbubbles, resulting in a transient increase in membrane permeability for ca + ( , ) . we previously reported that lius may make use of natural membrane vesicles as small as exosomes which are derived from immunosuppressor cells to fulfill its anti-inflammatory effects by upregulating the expression of extracellular vesicle/exosome biogenesis mediators and docking mediators ( ) . taken together, all these factors may profoundly affect cellular sensors that can activate various downstream signaling pathways in tissues exposed to lius. however, our findings suggest that cancer cells and non-cancer cells may use distinct signaling mechanisms to activate downstream targets when exposed to lius. our analysis revealed that lius can activate more of antioxidant effects in non-cancer cells compared to cancer cells. such changes in redox status of the cellular environment may lead to activation of sensors that may produce distinct gene expression patterns in non-cancer cells relative to cancer cells exposed to lius. our data indicated that different cell death regulatory transcription factors are induced in cancer cells and non-cancer cells treated with lius. for an example, we observed that tp and srf- genes were induced in lius treated cancer cells and non-cancer cells, respectively. therefore, it can be hypothesized that lius produces differential biological responses in different cellular contexts by activating distinct transcription factors, thus activating specific gene expression patterns. most interestingly, we observed unique patterns in clri sites in genes modulated by lius in cancer cells and non-cancer cells. therefore, lius may have the ability to modulate clri sites in cancer cells and noncancer cells, leading to distinct gene expression patterns. also, mild hyperthermia and oss associated with lius may also play a role in generating distinct transcriptome profiles depending on the cellular context. based on our findings, we propose a new working model on lius-mediated cancer-suppressing and anti-inflammatory mechanisms as shown in figure . our new model integrates the follow findings: first, lius induces cell death gene expression potentially via transcription factors tp -, and srf-, mediated pathways. second, the therapeutic applications of lius may depend on the propagation of ultrasound waves through tissues to produce thermal and non-thermal mechanic effects. third, lius may modulate chromatin long-range interactions to differentially regulate the cell death regulatome gene expression in cancer cells and non-cancer cells. upstream chromatin long-range interaction sites (clriss) are more favorable than downstream clriss for lius modulation of cell death regulator expression in cancer cells; and in contrast, downstream clriss play more important roles than upstream clriss for lius downregulation of inflammatory pathways in non-cancer cells. one limitation of the current study is the unavailability of biological data obtained from lius-treated patients' biopsies. we acknowledge that carefully designed in-vitro and invivo experimental models will be needed to further verify the lius mediated cancer-suppressing and anti-inflammatory mechanisms we report here. these experimental models will enable to consolidate the efficacy of lius in various pathological conditions as well. however, our analyses provide a significant insight in to lius-mediated modulation of the cell death regulatome via newly-defined nuclear programs to induce cell death in cancer cells and downregulate more inflammatory pathways in non-cancer cells. once again ( ), our findings provide molecular readouts that can be used to determine optimal ultrasound intensity and duration, and will provide guidance for the development of the future lius therapeutics for cancers, inflammations, tissue regeneration, and tissue repair. our report allows us to propose a new molecular working model for lius therapies for the treatment of cancers and inflammation: first, lius differentially upregulates cell death regulators in cancer cells, and downregulates inflammatory pathways in noncancer cells potentially via transcription factors tp -, and srf-, mediated pathways; second, the therapeutic applications of lius may depend on the propagation of ultrasound waves through tissues to produce thermal and non-thermal mechanic effects; third, lius may modulate chromatin long-range interactions to differentially regulate cell death gene expressions in cancer cells and non-cancer cells. our findings provide a significant insight in lius-mediated modulation of the expressions of cell death regulators via newly-defined nuclear programs to induce more cell death in cancer cells and downregulate more inflammatory pathways in non-cancer cells. also, our findings provide molecular readouts that can be used to determine optimal ultrasound intensity and duration; and will provide guidance for the development of the future lius therapeutics for cancers, inflammation, tissue regeneration, and tissue repair. jw and bl carried out the data gathering, data analysis, and prepared tables and figures. gn, qy, ysu, yl, ysh, dy, wy, rc, hf, hz, ws, sw, cz, yl, ec, and hw aided with analysis of the data. xy supervised the experimental design, data analysis, and manuscript writing. all authors read and approved the final manuscript. this work was supported by the hospital fellowships to bl and jw. tissue doppler echocardiography detects preclinical markers of cardiac lesion in mds patients low-intensity ultrasound-induced anti-inflammatory effects are mediated by several new mechanisms including gene induction, immunosuppressor cell promotion, and enhancement of exosome biogenesis and docking ultrasound, liposomes, and drug delivery: principles for using ultrasound to control the release of drugs from liposomes cellular mechanisms of the blood-brain barrier opening induced by ultrasound in presence of microbubbles development of a theoretical model describing sonoporation activity of cells exposed to ultrasound in the presence of contrast agents lepikh ia i, shegur pa. [anti-inflammatory effect of therapeutic and low-frequency ultrasound on a rat model of inflammatory process mechanical and biological effects of ultrasound: a review of present knowledge evolving diagnostic and treatment strategies for pancreatic neuroendocrine tumors pathological complete response after neoadjuvant chemotherapy with trastuzumabcontaining regimen in gastric cancer: a case report high-intensity focused ultrasound: current status for image-guided therapy a review of low-intensity ultrasound for cancer therapy sonodynamic therapy: concept, mechanism and application to cancer treatment roberts ww. development and translation of histotripsy: current status and future directions a review of the aesthetic treatment of abdominal subcutaneous adipose tissue: background, implications, and therapeutic options stimulation of bone repair with ultrasound: a review of the possible mechanic effects therapeutic ultrasound for osteoarthritis of the knee or hip therapeutic ultrasound for carpal tunnel syndrome low-intensity pulsed ultrasound rescues insufficient salivary secretion in autoimmune sialadenitis changing paradigms in cranio-facial regeneration: current and new strategies for the activation of endogenous stem cells nonthermal effects of therapeutic ultrasound: the frequency resonance hypothesis lipus suppressed lps-induced il- alpha through the inhibition of nf-kappab nuclear translocation via at -plcbeta pathway in mc t -e cells anti-inflammatory effects of ultrasound therapy: evidence for a major placebo effect the anti-inflammatory effects of dexamethasone and therapeutic ultrasound in oral surgery antiinflammatory effect of low intensity ultrasound (lius) on complete freund's adjuvant-induced arthritis synovium targeting mutant p for efficient cancer therapy a matter of life and death: stem cell survival in tissue regeneration and tumour formation regulated cell death and inflammation: an auto-amplification loop causes organ failure beyond tnf: tnf superfamily cytokines as targets for the treatment of rheumatic diseases a novel bcl-x isoform connected to the t cell receptor regulates apoptosis in t cells an n-terminal region of translationally controlled tumor protein is required for its antiapoptotic activity bcl-xl inhibits t-cell apoptosis induced by expression of sars coronavirus e protein in the absence of growth factors factors regulating apoptosis and homeostasis of cd + cd (high) foxp + regulatory t cells are new therapeutic targets higher expression of bax in regulatory t cells increases vascular inflammation expression of tctp antisense in cd (high) regulatory t cells aggravates cuff-injured vascular inflammation cd high t cells with a prolonged survival inhibit development of diabetes early hyperlipidemia promotes endothelial activation via a caspase- -sirtuin pathway inhibition of caspase- activation in endothelial cells improves angiogenesis: a novel therapeutic potential for ischemia caspase- mediates hyperlipidemia-weakened progenitor cell vessel repair caspase- inflammasome activation mediates homocysteine-induced pyrop-apoptosis in endothelial cells caspase- plays a critical role in accelerating chronic kidney diseasepromoted neointimal hyperplasia in the carotid artery low-intensity ultrasound-induced cellular destruction and autophagy of nasopharyngeal carcinoma cells caveolin- mediates low-intensity ultrasound-induced apoptosis via downregulation of signal transducer and activator of transcription phosphorylation in laryngeal carcinoma cells bioeffects of low-intensity ultrasound in vitro: apoptosis, protein profile alteration, and potential molecular mechanism apoptosis induced by the sonomechanical effects of low intensity pulsed ultrasound in a human leukemia cell line study of cellular response induced by low intensity ultrasound frequency sweep pattern on myelomonocytic lymphoma u cells -aminolevulinic acid-based sonodynamic therapy induces the apoptosis of osteosarcoma in mice low-intensity pulsed ultrasound activates integrin-mediated mechanotransduction pathway in synovial cells low intensity pulsed ultrasound enhanced mesenchymal stem cell recruitment through stromal derived factor- signaling in fracture healing using low-intensity pulsed ultrasound to improve muscle healing after laceration injury: an in vitro and in vivo study molecular mechanisms of low intensity pulsed ultrasound in human skin fibroblasts lowintensity pulsed ultrasound accelerated bone-tendon junction healing through regulation of vascular endothelial growth factor expression and cartilage formation genomic interactions: chromatin loops and gene meeting points in transcriptional regulation capturing chromosome conformation the inactive x chromosome adopts a unique three-dimensional conformation that is dependent on xist rna multiplexed chromosome conformation capture sequencing for rapid genome-scale high-resolution detection of long-range chromatin interactions chromosome conformation capture carbon copy ( c): a massively parallel solution for mapping interactions between genomic elements long range chromatin organization dynamic long-range chromatin interaction controls expression of il- in cd + t cells long-range genomic interactions epigenetically regulate the expression of a cytokine receptor a promoter interaction map for cardiovascular disease genetics increased acetylation of h k in the genomic regions that encode trained immunity enzymes in lysophosphatidylcholine-activated human aortic endothelial cells -novel qualification markers for chronic disease risk factors and conditional damps analyses of caspase- -regulated transcriptomes in various tissues lead to identification of novel il- beta-, il- -and sirtuin- -independent pathways human housekeeping genes, revisited novel extracellular and nuclear caspase- and inflammasomes propagate inflammation and regulate gene expression: a comprehensive database mining study dgenome: a comprehensive database of chromatin interactions increasing upstream chromatin long-range interactions may favor induction of circular rnas in lysopc-activated human aortic endothelial cells distinct sensitivity of normal and malignant cells to ultrasound in vitro molecular mechanisms of cell death: recommendations of the nomenclature committee on cell death diverse ubiquitin linkages regulate rip kinases-mediated inflammatory and cell death signaling inflammasomes: sensors of metabolic stresses for vascular inflammation an updated view on the functions of caspases in inflammation and immunity therapeutic targeting of p : all mutants are equal, but some mutants are more equal than others the atm protein kinase: regulating the cellular response to genotoxic stress, and more the role of hepatocyte growth factor (hgf) in insulin resistance and diabetes a review of therapeutic ultrasound: biophysical effects ultrasound-based triggered drug delivery to tumors the effects of lipus on soft-tissue healing: a review of literature clinical applications of lowintensity pulsed ultrasound and its potential role in urology intramuscular heating characteristics of multihour low-intensity therapeutic ultrasound the role of oxidative stress, antioxidants and vascular inflammation in cardiovascular disease (a review) oxidative stress and cancer: an overview erythrocyte deformability and oxidative stress in inflammatory bowel disease apoptosis and oxidative stress in neurodegenerative diseases low-intensity pulsed ultrasound prevents the oxidative stress induced endothelial-mesenchymal transition in human aortic endothelial cells effects of low-intensity ultrasound on oxidative damage in retinal pigment epithelial cells in vitro in vitro study of low intensity ultrasound combined with different doses of pdt: effects on c glioma cells -aminolevulinic acid enhances ultrasound-mediated antitumor activity via mitochondrial oxidative damage in breast cancer mechanisms of cell death in oxidative stress metabolic diseases downregulate the majority of histone modification enzymes, making a few upregulated enzymes novel therapeutic targets-"sand out and gold stays il- (interleukin- ) suppresses endothelial cell activation by inhibiting mitochondrial reactive oxygen species-mediated site-specific acetylation of h k (histone lysine ) dna checkpoint and repair factors are nuclear sensors for intracellular organelle stresses-inflammations and cancers can have high genomic risks integration of hi-c and chip-seq data reveals distinct types of chromatin linkages identification of genes responsive to low-intensity pulsed ultrasound stimulations comparison of continuous vs. pulsed focused ultrasound in treated muscle tissue as evaluated by magnetic resonance imaging, histological analysis, and microarray analysis identification of genes responsive to low intensity pulsed ultrasound in a human leukemia cell line molt- increased noncanonical splicing of autoantigen transcripts provides the structural basis for expression of untolerized epitopes inflammasomes are differentially expressed in cardiovascular and other tissues il- is a novel responsive anti-inflammatory cytokine-a new system of categorizing anti-inflammatory cytokines reparable cell sonoporation in suspension: theranostic potential of microbubble herpes simplex virus- transmission probability estimates based on quantity of viral shedding transient permeabilization of cell membranes by ultrasound-exposed microbubbles is related to formation of hydrogen peroxide modulation of intracellular ca + concentration in brain microvascular endothelial cells in vitro by acoustic cavitation the supplementary material for this article can be found online at: https://www.frontiersin.org/articles/ . /fonc. . /full#supplementary-material key: cord- -ptvaopgj authors: li, jing; wang, lishi; guo, sumin; xie, ning; yao, lan; cao, yanhong; day, sara w.; howard, scott c.; graff, j. carolyn; gu, tianshu; ji, jiafu; gu, weikuan; sun, dianjun title: the data set for patient information based algorithm to predict mortality cause by covid- date: - - journal: data brief doi: . /j.dib. . sha: doc_id: cord_uid: ptvaopgj the data of covid- disease in china and then in south korea were collected daily from several different official websites. the collected data included death cases in wuhan city of hubei province during early outbreak as well as confirmed cases and death toll in some specific regions, which were chosen as representatives from the perspective of the coronavirus outbreak in china. data were copied and pasted onto excel spreadsheets to perform data analysis. a new methodology, patient information based algorithm (piba) [ ], has been adapted to process the data and used to estimate the death rate of covid- in real-time. assumption is that the number of days from inpatients to death fall into a pattern of normal distribution and the scores in normal distribution can be obtained by observing death cases and analysing the data [ ]. we selected scores in normal distribution of these durations as lagging days, which will be used in the following estimation of death rate. we calculated each death rate on accumulative confirmed cases with each lagging day from the current data and then weighted every death rate with its corresponding possibility to obtain the total death rate on each day. while the trendline of these death rate curves meet the curve of current ratio between accumulative death cases and confirmed cases at some points in the near future, we considered that these intersections are within the range of real death rates. six tables were presented to illustrate the piba method using data from china and south korea. one figure on estimated rate of infection and patients in serious condition and retrospective estimation of initially occurring time of corid- based on piba. the data of covid- disease in china and then in south korea were collected daily from several different official websites. the collected data included death cases in wuhan city of hubei province during early outbreak as well as confirmed cases and death toll in some specific regions, which were chosen as representatives from the perspective of the coronavirus outbreak in china. data were copied and pasted onto excel spreadsheets to perform data analysis. a new methodology, patient information based algorithm (piba) [ ] , has been adapted to process the data and used to estimate the death rate of covid- in real-time. assumption is that the number of days from inpatients to death fall into a pattern of normal distribution and the scores in normal distribution can be obtained by observing death cases and analysing the data [ ] . we selected scores in normal distribution of these durations as lagging days, which will be used in the following estimation of death rate. we calculated each death rate on accumulative confirmed cases with each lagging day from the current data and then weighted every death rate with its corresponding possibility to obtain the total death rate on each day. while the trendline of these death rate curves meet the curve of current ratio between accumulative death cases and confirmed cases at some points in the near future, we considered that these intersections are within the range of real death rates. six tables were presented to illustrate the piba method using data from china and south korea. table subject death rate estimation using normal distribution, of mean, standard deviations and formulas. the data estimation focuses on the early estimation of death rate of infectious diseases, in particular, the disease covid- caused by -ncov. collected data are formatted on excel spreadsheets for analysing. data include the total number of patients, total number of deaths, daily numbers of new patients, daily number of new deaths, from starting data of official report to the presented time, e.g., march , . data were collected through the cyberlinke of each official websites and copied and pasted the desired data onto excel spreadsheets.  these data provide the scientific community with a new methodology to estimate the death rate and then predict the death cases during an epidemic.  scientific researchers, cdc employees, government officers for disease control and management, and public population, will benefit from these data.  these data will be very useful for the studies with the purpose either of disease control management or of related sources preparation to combat against an outbreak.  due to the limited amount of data samples collected in this article, some factors, such as the phases of an outbreak and the measurements issued by the department of disease control that might impact the death rate of an epidemic, could be taken into for further insights and development of experiments with a large amount of data. chd-coronary heart disease the data of death cases in table have been collected from the official website of the health commission of hubei province in china, which include the date that patients have onset of symptoms, the date that patients began to be taken into icu and the date of decease. with these data, the days both from symptoms appearance to death and from icu intake to death can be calculated. following normal distribution, the mean score μ and standard deviation σ can be calculated either. thus the selected scores (μ, μ ± σ and μ ± σ) in normal distribution can be obtain as the basic elements for the following estimation and prediction of death rate, which are respectively , , , , days. the disease information in table has been collected from the public media before we resume data analysing with the same method of death rate estimation and prediction in south korea as in china [ ] . we have collected accumulative confirmed cases and deaths and then new confirmed cases and new deaths in south korea. death rate from the date symptoms - - - - - - - - - - each score we selected in normal distribution has a specific possibility when we take them into consideration of representatives in bell curve [ ] . when we weighted each death rate on a day with their corresponding possibilities and then sum, the total death rate on each day can be obtained. each curve consisting of several death rate will have a trendline and thus a formula to describe this trend as well as the current ratio between accumulative death cases and confirmed cases on each day (table ). - - - - - - - - - - current ratio between accumulative death cases and confirmed cases the current ratio between accumulative death cases and confirmed cases is calculated by dividing accumulative death cases with accumulative confirmed cases on each day. the intersect points of three trendlines intersect intersect death rate in south korea . % . % the trendlines of death rate and death rate tend to intersect with the trendline of the current ratio finally, because the current ratio will be the real death rate at the end of epidemic. we considered that the intersection value of three trendline (death rate and , current ratio) will drop in the range of real death rate. when we calculated the death rate separately with the corresponding formula of their trendlines, two intersections have been acquired (table -b) . we pick the maximum value between them to predict new death cases in the following days (table ). tables are produced based on the patient information based algorithm (piba) [ ] . piba has been adapted when estimating the death rate of covid- in real-time with publicly posted data. following normal distribution, the different durations with different possibilities between symptom appearance and death have been derived from analysing death cases in wuhan city of hubei province in china [ ] . based on these results, the total death rate in regions can be calculated specifically by putting in the different death rates with different durations together. while the trendline of these death rate curves meet the curve of current ratio between accumulative death cases and confirmed cases at some points in the near future, we considered that these intersections are within the range of real death rates. the data analysis was all following normal distribution, either in calculating the possibility of every selected score or in estimating the death rate. after collection of data of covid patients from south korea, the data was analysed with piba method as indicated above ( table ). the death rate was first estimated ( table ). the death rate then was calculated (table ). following estimations, the piba method then was used to predict the number of deaths in the following week (table ). the predicated death numbers then were compared to the real death numbers (table ). considerably lower than expected. prior expectation has been much higher, based on multiple infectious routes [ ] [ ] . using our formula, the results indicate that the current infectious rate is even lower than the rate based on the total numbers (see fig. a ). the infectious rate in hubei province is currently around %, although previously the rate was as high as %. on average, the infectious rate overall in china is about %, while in hubei it is %. in the rest of the country, it is . %. among the inpatients, the rate in serious medical condition ranges from % to % (see fig. b ), while it averages at % in china, % in hubei, and % in the rest of country (except hubei). based on the estimated death rate, on january , there should be a total of to inpatients (see fig. c ). based on the rate of patients who are severely ill among all patients, on january , there should be to patients. based on the effective infection rate and based on the assumption of one week or days from close contact to the onset of symptoms, there might be , to , people who were infected around december , . if we believe the epidemic doubling time is approximately days, the initial infection source may date back to as early as november or october . dianjun sun. real-time estimation and prediction of mortality caused by covid- with patient information based algorithm clinical features of patients infected with novel coronavirus in wuhan nowcasting and forecasting the potential domestic and international spread of the -ncov outbreak originating in wuhan, china: a modelling study epidemiological and clinical characteristics of cases of novel coronavirus pneumonia in wuhan, china: a descriptive study this work was partially supported by funding from merit grant i bx to wg from the the authors declare that they have no known competing financial interests or personal relationships which have, or could be perceived to have, influenced the work reported in this article. key: cord- -yi bh js authors: o'brien, mauria a.; kirby, rebecca title: apoptosis: a review of pro‐apoptotic and anti‐apoptotic pathways and dysregulation in disease date: - - journal: j vet emerg crit care (san antonio) doi: . /j. - . . .x sha: doc_id: cord_uid: yi bh js objective – to review the human and veterinary literature on the biology of apoptosis in health and disease. data sources – data were examined from the human and veterinary literature identified through pubmed and references listed in appropriate articles pertaining to apoptosis. human data synthesis – the role of apoptosis in health and disease is a rapidly growing area of research in human medicine. apoptosis has been identified as a component of human autoimmune diseases, alzheimer's disease, cancer, and sepsis. veterinary data synthesis – research data available from the veterinary literature pertaining to apoptosis and its role in diseases of small animal species is still in its infancy. the majority of veterinary studies focus on oncologic therapy. most of the basic science and human clinical research studies use human blood and tissue samples and murine models. the results from these studies may be applicable to small animal species. conclusions – apoptosis is the complex physiologic process of programmed cell death. the pathophysiology of apoptosis and disease is only now being closely evaluated in human medicine. knowledge of the physiologic mechanisms by which tissues regulate their size and composition is leading researchers to investigate the role of apoptosis in human diseases such as cancer, autoimmune disease and sepsis. because it is a multifaceted process, apoptosis is difficult to target or manipulate therapeutically. future studies may reveal methods to regulate or manipulate apoptosis and improve patient outcome. all tissues must be able to tightly control cell numbers and tissue size and to protect themselves from rogue cells that threaten homeostasis. in the early s, kerr et al, observed a single-cell-death phenomenon that occurred in the dying cells of healthy tissues, as well as in cells associated with teratogenesis, neoplasia, tumor regression, atrophy, and involution. the term apoptosis, from greek origins (apo for, ptosis falling), was chosen to describe the cellular process of programmed cell death. [ ] [ ] [ ] apoptosis is a tightly regulated intracellular program in which cells destined to die activate enzymes that degrade the cell's dna and nuclear and cytoplasmic proteins. programmed cell death eliminates unwanted cells or potentially reactive cell lines either before or after maturation. this process is vital to fetal and embryonic development and to tissue remodeling. cell populations that normally have a high rate of proliferation, such as the intestinal epithelium, depend upon apoptosis to maintain the necessary number of cells. the number of activated immune cells must be controlled to contain the inflammatory response. , hormone-dependent apoptosis occurs during estrus and causes prostatic atrophy after castration. kerr et al observed that apoptotic cells share many morphologic features distinct from those in necrotic cells. cells undergoing apoptosis exhibit or more of the following: cell shrinkage, chromatin condensation and nucleosomal fragmentation, and bubbling of the plasma membrane (blebbing). biochemical features of apoptosis include dna fragmentation, protein cleavage at specific locations, increased mitochondrial membrane permeability, and the appearance of phosphatidylserine on the cell membrane surface. , there is an increase in mitochondrial permeability leading to the release of pro-apoptotic proteins and subsequent formation of apoptotic bodies. the resulting membrane-bound apoptotic bodies are consumed by neighboring cells or by macrophages. apoptosis is a single-cell event, and does not induce an inflammatory reaction. apoptosis must be distinguished from necrosis, which is also a form of cellular death. in contrast to apoptosis, necrosis is not a genetically programmed function, it affects groups of neighboring cells, and produces an inflammatory response. the death of a cell by necrosis leads to the release of alarm signal molecules that stimulate or more pattern-recognition receptors on macrophages, dendritic cells, and natural killer cells. the presence of necrotic cells in a tissue is frequently interpreted by the immune system as dangerous and therefore acts as a signal to initiate an immune response. unlike apoptosis, with necrosis there is cellular swelling with loss of cell membrane integrity, organelle swelling, and lysosomal leakage. the degradation of dna is random and lysed cells are ingested by macrophages. whether a cell survives or dies by apoptosis is determined by the balance between pro-apoptotic (stress or death) signals and anti-apoptotic (mitogenic or survival) signals within and around the cell (see tables and ). cell injury via oxygen deprivation, heat stress, chemical agents, radiation, infectious agents, genetic derangements, nutritional imbalances, immunologic reactions (eg, anaphylaxis), and other types of severe cell stress will initiate the pro-apoptotic pathways. dysregulation of apoptosis can affect the equilibrium between cell growth and cell death, resulting in organ dysfunction. apoptosis in health is a finely balanced process. too much or too little apoptosis contributes to disease. apoptosis of infected cells is part of the host's defense mechanism. some viruses and bacteria, however, have developed the ability to inhibit the infected cell's apoptotic mechanisms and protect their environment. inhibition of apoptosis is linked to uncontrolled cell growth and the formation of many types of cancer. in humans, excessive apoptosis is linked to stroke and alzheimer's disease. the activation or restoration of apoptosis is emerging as a key strategy for treatment of cancer and other diseases. [ ] [ ] [ ] [ ] [ ] [ ] our aim is to provide a basic review of the literature regarding the mechanisms and regulation of apoptosis. extracellular ligand-directed or intracellular stressinduced stimuli can activate this highly regulated process. caspases play a central role by initiating and executing the intracellular cascade of events that result in protein and nucleic acid cleavage, and ultimately, cell death. many of the key apoptotic proteins have been identified, however there is still much to learn regarding the molecular mechanisms of action or activation of these proteins. knowledge of the pro-apoptotic and anti-apoptotic cell pathways is important to understanding the mechanisms of many life-altering diseases in humans and animals and realizing the potential for novel therapeutics. apoptosis can be genetically encoded or can occur in response to cellular or external stimuli. there are features that characterize apoptosis: protein cleavage or hydrolysis, breakdown of nuclear dna, and recognition of the apoptotic cell by phagocytic cells. the cleavage of proteins primarily occurs with the activation of a family of cysteine proteases called caspases (cysteine aspartate-specific proteases). caspases are synthesized in an inactive form and activated by specific initiation mechanisms. programmed cell death can also result from caspase-independent mechanisms triggered by cell membrane receptor-ligand binding or damage to cell organelles. [ ] [ ] [ ] [ ] [ ] [ ] initiation of caspase cascades there are known pathways that initiate the activation of caspase cascades and the programmed death of a cell. the route utilized is dependent on the initial death signal, the cell type involved, and the balance between pro-apoptotic and anti-apoptotic signals. one initiating path may lead to another with cross-talk between them possible. two of the pathways, the death receptor (dr) (extrinsic) and mitochondrial (intrinsic), have been detailed in the literature. , , [ ] [ ] [ ] the third is an bcl- , b-cell lymphoma ; bcl-xl, bcl- -associated protein xl; bax, bcl- -associated protein x; bak, bcl- -associated protein k; c-flip, flice-like inhibitory protein; nf-kb, nuclear factor-kb; ikb, inhibitory-kb; iaps, inibitor of apoptosis proteins; xiap, x-linked inhibitor of apoptosis protein; jak, janus kinase; stat, signal transducers and activators of transcription; mapk, mitogen-activated protein kinase; pkr, protein kinase r; cdk, cyclindependent kinase. intrinsic pathway involving the endoplasmic reticulum (er) and is the least understood. [ ] [ ] [ ] [ ] [ ] this pathway is believed to be a pathologically relevant form of apoptosis occurring in response to cellular stress. , extrinsic (dr) pathway: the extrinsic pathway (see figure ) begins with pro-apoptotic receptors on the cell's surface activated by a pro-apoptotic molecule or ligands specific for that receptor. these cell drs belong to the tumor necrosis factor (tnf) receptor superfamily, with the fas receptor and tnfr as the most intensely studied members. fas is present on a variety of cell types including activated b cells and t cells. ligands that activate pro-apoptotic receptors include the fas ligand (fasl) and tnf-a [ ] [ ] [ ] [ ] [ ] [ ] (table ). fasl is expressed by a variety of cell types, including activated t cells and natural killer cells. tnf-a is produced predominantly by activated monocyte/macrophages and lymphocytes. the intracellular portion of the dr is known as the death domain (dd). once or more dr-ligand complexes bunch, their dds are brought into close proximity and a binding site for an adaptor protein is formed. the adaptor protein is specific for that receptor (eg, fas-associated dd [fadd] or tnf receptor-associated dd [tradd] ). this complex of ligand-receptor-adaptor protein is called the death-inducing signaling complex (disc), leading to the recruitment and assembly of initiator caspases and . - these caspases can now undergo self-processing and release active caspase enzyme molecules into the cytosol. here, they activate the effector caspases , , and . , , figure illustrates the sequence of events that trigger the extrinsic pathway. the extrinsic or death receptor (dr) pathway. pro-apoptotic ligands, death signals, and fas bind to fas or tnfrs. the intracellular portion of the dr is known as the death domain (dd). bunching of the receptor-ligand complexes groups their dds and a binding site for an adaptor protein is formed. this ligand-receptor-adaptor protein complex is called the death-inducing signaling complex (disc). it recruits and assembles initiator caspase- that releases active caspase enzyme molecules into the cytosol. here, they activate the effector caspases- and - , resulting in nuclear protein cleavage and the initiation of apoptosis. fasl, fas ligand; tnfr tumor necrosis factor receptor; fadd, fas-associated death domain; tradd, tnf-associated death domain; c-flip, flicelike inhibitory protein; disc, death-inducing signaling complex. intrinsic mitochondrial pathway: the intrinsic mitochondrial pathway (see figure ) is initiated from within the cell in response to cellular stresses such as dna damage, radical oxygen species, radiation, hormone or growth-factor deprivation, chemotherapeutic agents, cytokines, and glucocorticoids. initiation of this pathway eventually results in the release of pro-apoptotic proteins from the mitochondria that will activate caspase enzymes and trigger apoptosis. [ ] [ ] [ ] [ ] the success of the pathway in inducing apoptosis depends on the balance of activity between pro-apoptotic and anti-apoptotic members of the b-cell lymphoma- (bcl- ) superfamily of proteins (table ) . bcl- superfamily of proteins derives its name as the second member of a range of proteins found in follicular lymphoma. , all of the bcl- family members are present on the outer mitochondrial membranes as dimers where they control membrane permeability in ion channel fashion or through the creation of pores. the permeability of the mitochondrial outer membrane determines whether or not there is release of the pro-apoptogenic substances from the mitochondria. this bcl- family of proteins is subdivided into groups based on structural similarities and functional criteria. group i possess anti-apoptotic activity while groups ii and iii promote cell death. the bcl- family the mitochondrial or intrinsic pathway. activation of the pro-apoptotic proteins bax and bak occurs through conversion of bid to tbid by caspase- or- and through activation of puma, noxa, or other bh initiator proteins when p is induced by dna damage. activated bax and bak oligomerize at the mitochondrial membrane and cause the release of several mitochondrial factors. cytochrome c combines with apaf- and procaspase- forming an apoptosome. also released from the mitochondria are smac/ diablo, proteins that inactivate iaps. activated caspase- then is able to activate caspase- or - allowing apoptosis to proceed. also released from the mitochondria are endog and aif that stimulate apoptosis independent of caspases. bcl- and bcl-xl block the activation of bax and bak. bcl- , b-cell lymphoma- ; iap, inhibitor of apoptosis protein; apaf- , apoptosis-activating factor- ; smac, second mitochondrial-derived activator of caspases; diablo, director inhibitor of apoptosis-binding protein with low pi; bh , bclhomology- ; tbid, truncated bid; endog, endonuclease g; aif, apoptosis-inducing factor; bax, bcl- -associated protein x; bak, bcl- associated protein k. share or more of characteristic domains of homology crucial for function. the anti-apoptotic bcl- family proteins, such as bcl- and bcl-xl, contain all domains and exert their control of mitochondrial permeability by stimulating adp/atp exchange, stabilizing the mitochondrial inner transmembrane potential, and preventing the opening of a permeability transition pore. overexpression of bcl- and bcl-xl is known to be associated with a number of human malignancies , , ( table ). these proteins also act by inhibiting the action of the pro-apoptotic proteins, bax and bak. pro-apoptotic proteins of the bcl- family initiate apoptosis by blocking the anti-apoptotic activity of bcl- and bcl-xl by binding to their mitochondrial binding sites or by triggering the activation of pro-apoptotic bax/ bak. a third type of pro-apoptotic activity is through the cytoplasmic protein, bid. this molecule is found in the cytoplasm in an inactive form. when cleaved by activated caspase- from the extrinsic pathway, bid (once activated, referred to as t-bid) causes a structural change to bax making it similar to the structure of the anti-apoptotic molecule, bcl- , allowing bax to translocate to the mitochondria. [ ] [ ] [ ] [ ] this is but one method of cross-talk that occurs between the intrinsic and extrinsic pathways. each bcl- family member can interact with other bcl- members, so that large numbers of heterodimer combinations within a cell are possible. cells with more prodeath proteins are sensitive to death and cells with an excess of protective family members are usually apoptosis-resistant. there are at least current theories describing the exact mechanism by which the bcl- pro-apoptotic proteins lead to increased mitochondrial permeability. the first theory describes the insertion of bcl- proteins into the mitochondrial membrane and directly forming a channel. , a second theory explains the bcl- proapoptotic proteins interacting with other mitochondrial membrane proteins, possibly voltage-dependent anion channel, to form large pores. the size of the voltage-dependent anion channel is too small for proteins to pass, so this model assumes that there is a conformation change with bcl- binding. the third theory describes the bcl- proteins modulating the mitochondrial proteins resulting in the formation of a permeability transition pore and loss of membrane potential, organelle swelling, and loss of cytochrome c from the pore. , once the permeability of the membrane has been compromised, cytochrome c is released and combines with a cytosolic molecule called apoptosis activating factor- (apaf- ). cytochrome c and apaf- combine with procaspase- for activation of this caspase ( figure ). the binding of these substances forms an apoptosome, which then activates procaspase- . alternate substances can initiate the intrinsic and extrinsic pathways (table ) . phosphoprotein p is a transcription factor that regulates the cell cycle and functions as a cell stress sensor molecule capable of inducing the intrinsic mitochondrial pathway. factors that damage dna, such as ionizing radiation, genotoxic drugs, and free radicals, activate p . activated p promotes apoptosis primarily through its ability to suppress the transcription of anti-apoptotic factors like bcl- or induce the manufacture of pro-apoptotic factors like bax, insulin growth factor binding protein- and upregulation of the fas receptor. , bcl-xl can inhibit p , and inactivation or loss of p is a common abnormality of many human cancers. , one veterinary study showed an increased risk in cats, diagnosed with oral squamous cell carcinoma and exposed to secondhand tobacco smoke, to overexpress p . in human cancer, p is the most commonly disrupted gene and is also the most frequently mutated gene in human oral cancer. intrinsic er pathway: the third and least understood pathway is referred to as the endoplasmic reticulum or er pathway. it involves caspase- and is said to be able to function independently of the mitochondria. cellular stresses such as hypoxia, glucose starvation, disturbances in calcium homeostasis, and exposure to free radicals injure the er, resulting in the unfolding of proteins and reduction in protein synthesis. in normal cells, an adaptor protein, tnf receptorassociated factor (traf ), is bound to procaspase- , rendering it inactive. stress of the er leads to the dissociation of traf , activation of caspase- . , , once activated, caspase- cleaves procaspase- , which then cleaves procaspase- . this mechanism is independent of the mitochondria although there is evidence that caspase can cause the release of cytochrome c from the mitochondria and thus stimulate the intrinsic pathway (table ) . there are other mechanisms of cross-talk between the intrinsic pathways that initiate apoptosis. ito et al, demonstrated that er stress can cause activation of a positive apoptosis regulator, c-abl tyrosine kinase, known for its tumorogenic characteristics and results in the release of cytochrome c from the mitochondria. , in addition, the mitochondrial pathway pro-apoptotic bcl- family protein, bak, has been implicated in causing er depletion of calcium, which can induce caspase- activation (table ) . the central executioners of apoptosis are part of a large protein family known as the caspases. , to date, caspases have been identified. specific caspases are found in relatively large amounts as inactive precursors called procaspases within the cytoplasm. procaspases can be activated by of methods: ( ) exposure to another activated caspase, ( ) autocatalysis, or ( ) association with an activator protein, such as caspase- , apaf- , and cytochrome c. the caspases involved in apoptosis are subdivided into initiator caspases ( , , , ) and effector (or executioner) caspases ( , , ) . the initiator caspases are activated by adaptor-mediated self-cleavage. the specific interaction of caspases and activator protein promotes the formation of a multimeric complex that is necessary to bring caspase precursors together to activate each other and produce an active tetramer. , this caspase cascade strategy of activation is used by the initiator caspases to cleave and activate the effector caspases. when activated, the effector caspases selectively cleave a restricted set of target proteins that follow an aspartate residue. in most cases, this results in inactivation of the target protein. however, they can also activate proteins, either directly by cleaving off a negative regulatory domain, or indirectly, by inactivating a regulatory subunit. , effector caspases cause cytoskeletal filament aggregation, clumping of ribosomal particles and rearrangement of rough er to form a series of concentric whorls as seen on electron microscopy (table ) . , caspases are also responsible for cleaving nuclear lamins required for nuclear shrinking and budding, and for loss of cellular shape and membrane blebbing. , , caspases will activate caspase-activated dnaases (cad) that breakdown nuclear dna, the second feature of apoptosis. cad exists in an inactive form (icad) in the nucleus when it is bound to a subunit. once the effector caspase- is activated it migrates to the nucleus and cleaves the inhibitory subunit thus activating cad. cad is the nuclease responsible for breaking down the dna into - -kb pieces that are then cleaved into - -bp fragments by endonucleases. it is these fragments that compose the dna ladders visualized by agarose gel electrophoresis, a biochemical hallmark of apoptosis. alternate pathways granzyme b: there is an accessory method of triggering apoptosis by the serine protease granzyme b, a lymphocyte granular enzyme expressed by activated cytotoxic t lymphocytes and natural killer cells. granzyme b will bind to its cell surface receptor, an insulin-like growth factor ii receptor, causing endocytosis of the protease. it remains in the endocytic vesicle until stimulation by an activated cytotoxic t cell. activation of almost all of the activator and effector cas-pases can occur through the action of the granzyme b pathway. a number of caspase sensitive targets, as well as other unique proteins not normally cleaved by caspases, can be cleaved directly by granzyme b. , this mechanism is dependent on mitochondrial disruption, as overexpression of anti-apoptotic bcl- will halt this process. granzyme-mediated apoptosis is integral to the immune surveillance machinery that rid the body of virally infected or malignantly transformed cells (table ) . mitochondrial factors: increased mitochondrial outer membrane permeability can result in the release of mitochondrial pro-death substances in addition to cytochrome c, such as apoptosis inducing factor, smac/ diablo - (second mitochondrial-derived activator of caspases/direct inhibitor of apoptosis-binding protein with low pi), endonuclease g, htra /omi, and several procaspases (eg, procaspase- , - , and - ). smac/diablo and htra /omi bind to cytoplasmic inhibitor of apoptosis proteins (iaps), neutralizing their anti-apoptotic activity. the migration of apoptosis inducing factor from the mitochondria to the nucleus induces caspase-independent chromatin condensation and dna fragmentation. endonuclease g can also directly cause the break up of dna independent of caspases (table , figure ). ceramide: the ceramide/sphingomyelin pathway can proceed with or without caspase interaction. ceramides are sphingolipid-signaling mediators involved in the regulation of differentiation, growth suppression, cell aging, stress responses, and apoptosis. various stresses, such as ultraviolet radiation, radical oxygen species, chemotherapeutics, il- b, tnf-a, or fas activation, initiate this pathway. sphingomyelinases are activated by binding to tnf receptor family dd, fadd, and tradd and cleave sphingomyelin, a member of the phospholipid bilayer, into ceramide. ceramide can also be generated on lysosomes, er, and mitochondrial membranes, when stimulated by a variety of cytotoxic agents. [ ] [ ] [ ] increased levels of ceramide initiate an apoptotic program involving mitochondrial membrane disruption (table ) . ceramide-induced apoptosis is mediated through the mitochondria when ceramide accumulates in the mitochondrial membrane. elevation of ceramide and sphingosine result in increased mitochondrial membrane permeability, cyctochrome c release, and activation of caspase- . the susceptibility of a cell for ceramide-induced apoptosis is reliant on the cell's bax to bcl- ratio. ceramide production, dna damage, and radical oxygen species also stimulate cellular lysosomal release of cathepsins, which leads to release of mitochondrial factors, activation of procaspase- and - , and activation of bid. , inhibitors of apoptosis the most well-known anti-apoptotic factors are members of the bcl- family (table ). there is a dynamic equilibrium between the anti-apoptotic members and pro-apoptotic members of the bcl- family. additional inhibitors include iaps, flice-like inhibitory protein (c-flip) and nuclear factor-kb (nf-kb). iaps: a failsafe inhibitory mechanism exists in the intrinsic pathway. pro-apoptotic activity is counterbalanced by a family of at least proteins, known as iaps (eg, survivin and xiap). the iaps compromise the effector phase of apoptosis through blocking or inactivating caspases [ ] [ ] [ ] [ ] [ ] [ ] (see table ). they have been found associated with the activated tnf receptors where they block the activation of caspase- and are upregulated by nf-kb activation. , iaps also act downstream of the mitochondrial release of cytochrome c to prevent activation of caspase- . , caspase- uses a peptide from one of its end subunits to attract an iap family molecule. this binding allows caspase- to remain dormant even though the initial steps for activation have taken place. , the overexpression of iaps is associated with a drug-resistant phenotype of cancer cells. c-flip: c-flip is a protein-deficient caspase homolog. this inhibitor prevents both the binding of caspase- to various drs and its activation. the ratio of c-flip to caspase- is critical for the assembly of the disc. formation of the disc recruits c-flip to bind to its target molecule. this works as a built-in safety system and alludes to the integral balance between proapoptotic and anti-apoptotic mediators. upregulation of c-flip has been associated with diverse hematologic cancer cell lines. , the sensitization of many cancer cells to death ligand-mediated apoptosis appears to be mediated by c-flip downregulation. , there does not appear to be inactivation of caspase- mechanisms because c-flip does not prevent apoptosis induced by granzyme b or by chemotherapeutic drugs and irradiation. nf-jb: nf-kb is a transcription factor and an antiapoptotic gene regulator composed of a p /p heterodimer. the p subunit provides the gene regulatory function. nf-kb is kept quiescent in the cytoplasm as a dimer bound to its repressor, inhibitors of nf-kb (ikb) family. phosphorylation of ikb by upstream kinases frees nf-kb, which then translocates to the nucleus. the p subunit is eventually released from the dna and binds to newly synthesized ikb a, which complexes with nf-kb. this complex translocates back into the cytoplasm. stimulation of nf-kb activation has been associated with accelerated growth, resistance to ap-optosis, and propensity to form metastases. inversely, inhibition of nf-kb activation produces an increase in apoptosis, indicating that the balance of cell viability versus cell death is maintained in some degree to nf-kb activation. recognition of apoptotic cells the third characteristic of apoptosis is recognition of the dying cells by phagocytic cells. the apoptotic cell expresses markers on their membrane that are recognized by phagocytes. through internal cellular signals phosphatidylserine, a phospholipid component, is shifted from the inner to the outer layer of the plasma membrane. this allows for early recognition and removal of a dying cell without release of pro-inflammatory mediators, as occurs with necrosis. control of whether the pro-apoptotic or anti-apoptotic pathway is chosen is subject to positive and negative genetic and environmental regulators. pro-apoptotic gene activation will lead to cell death while deactivation of the gene will block apoptotic pathways. genetic regulation can also be modified by exogenous stimuli from the cell's immediate environment. a cell destined or started on a death pathway can receive a survival signal that can save the cell from apoptosis. genetic regulators (mostly pro-apoptotic) include the c-myc gene family, the p tumor suppressor gene, drs and the caspase family. the c-myc gene is a member of the janus kinase or jak family, and is involved in both cell proliferation and apoptosis. in the presence of anti-apoptotic cytokines (eg, insulin-like growth factor- ) or negative regulators of apoptosis (eg bcl- ), c-myc drives cellular proliferation. in the absence of these factors, the c-myc family promotes apoptosis. survival signals, such as growth factors and other soluble mediators, are often released by neighboring cells. hematopoietic cell lines and differentiated cells are dependent on survival factors like granulocyte macrophage colony stimulating factor, granulocyte colony stimulating factor, il- , and erythropoietin. t and b lymphocytes are dependent on certain interleukins, such as il- and il- , to mature properly or lead to cellular differentiation. the role of environmental survival substances leads to the possibility of dysregulation of apoptosis in the presence of inappropriate survival signals, as observed in tumors and sepsis. , , apoptosis in disease dysregulated apoptosis best describes pathologic disease states that induce or inhibit cell death inappropri-ately. in humans, excessive apoptosis is linked to stroke and alzheimer's disease and reduced apoptosis to cancer and autoimmune disease. , , apoptosis also plays a key role in sepsis. a review of the current literature in companion animal medicine yields research pertaining to apoptosis in the fields of oncology, orthopedics, and virology, as well as in other disease states. table lists some of the articles of veterinary origin that investigate apoptosis in disease. a feature characteristic of cancer cells is the uncoupling of cell division and cell death; cells that should have died were not properly signaled to do so. oncologic studies are investigating the enhancement of apoptosis to arrest the growth of tumor cells. one mechanism through which normal cellular numbers are maintained is through p and its control over-apoptosis. deficiencies in p can lead to reduced apoptosis and tumor development. , some cancers harbor mutations in the p genome or disrupt normal p functions whereas others increase or overexpress bcl- proteins leading to a cessation of the normal cellular death program. apoptosis in sepsis cellular demise in sepsis can occur through apoptosis as well as necrosis. in , bone first proposed that apoptosis contributes to the multiple organ dysfunction in sepsis. , most treatments had been aimed at blunting the over-reactive or pro-inflammatory response. bone proposed that the anergic or hypoimmune aspect of sepsis, when apoptosis becomes most critical, must also be addressed, apoptotic loss of b cells, t cells, and dendritic cells in sepsis decreases antibody production, macrophage activation, and antigen presentation, respectively. leukocytes are responsible for opsonization and phagocytosis of infected cells and antigens at the site of inflammation. neutrophils produce highly toxic and unstable reactive oxygen species and release bactericidal substances. a hallmark of sepsis is the loss of normal apoptosis of neutrophils. this prolonged life produces neighboring cell damage and contributes to activation of pro-inflammatory cytokines. normally, pro-inflammatory mediators (tnf-a, il- b, il- , il- , and ifn-g) released from macrophages and neutrophils have overlapping effects and function to limit damage, combat pathogenic organisms, eliminate foreign antigens, and promote repair. anti-inflammatory cytokines (il- , il- , tgf-b, soluble receptors and receptor antagonists) are also quickly released to try to reduce and locally contain the inflammatory response. many of these inflammatory components are the key factors responsible for the dysregulated apoptosis of immune cells in sepsis. key cells involved in the inflammatory process (neutrophils, macrophages, dendritic cells, and lympho- glucocorticoid immunosuppressive effects on canine lymphocytes may involve apoptosis cytes) are also cells targeted for apoptosis. apoptosis of immune cells is normally not a pathologic process because inflammatory cells must be eliminated so that inflammation does not continue unabated. , however, in sepsis or other overwhelming inflammatory processes (like trauma and severe burns) there is extensive cell death of lymphocytes and dendritic cells and delayed cell death of neutrophils. this leads to a blunted immune response coinciding with increased cellular damage. neutrophils are one of the first cells to migrate to the site of inflammation with an average half life of - hours when unstimulated. , once a neutrophil is released into circulation, its apoptotic program has been activated. studies have shown that sepsis can shorten as well as prolong the life span of the activated neutrophils (early or delayed apoptosis, respectively). [ ] [ ] [ ] [ ] early apoptosis of neutrophils dampens respiratory burst activity and may lessen secondary tissue injury. delayed apoptosis of neutrophils contributes to increased cellular damage, especially in the lung, liver, kidneys, and gastrointestinal tract. acute respiratory distress syndrome (ards) is marked by significant pulmonary accumulation of neutrophils, , and is considered to be a direct effect of neutrophil-induced injury. cells retrieved from the lungs of septic patients show reduced rates of neutrophil apoptosis with the degree of inhibition paralleling the severity of sepsis. , additional studies have shown that increased apoptosis (via fas/fasl-dependent mechanism) of pulmonary epithelial cells will lead to permeability changes characteristic of ards. , the cytokines produced by activated neutrophils summon macrophages to the area of inflammation; cytokines are also produced by tissue macrophages in response to foreign invasion. macrophages are antigenpresenting cells (apc) capable of engulfing foreign material, infected cells, and apoptotic cells through recognition of specific cell surface molecules. dendritic cells, another type of apc, are viewed as the sentinels of the immune system. like macrophages, mature dendritic cells are able to activate lymphocytes through the presentation of antigen. lymphocytes need signals to stimulate differentiation and initiation of the immune response. the first signal is the presentation of antigen thus accounting for the specificity of the response. the second signal involves costimulatory molecules on apcs or apc secretion of cytokines. the costimulatory molecules interact with specific t cell sites producing a pro-apoptotic or anti-apoptotic state. failure of the appropriate second signal after interaction with an apc results in anergy or apoptosis of the lymphocyte. , anergy is a state of unresponsiveness to antigen. this is also the mechanism for self-tolerance. immature dendritic cells are capable of ingesting apoptotic cells but this will render them incapable of maturing and stimulating t cells. macrophages and dendritic cells will secrete il- after engulfing apoptotic cells. il- , considered an anti-inflammatory cytokine, selectively blocks the maturation of dendritic cells. this has been shown to suppress the phagocytic activity as well as pro-inflammatory cytokine production of alveolar macrophages. il- is also secreted by dendritic cells after ingesting apoptotic cells, leading to autocrine blockage of maturation. this lack of maturation leads to a tolerogenic state with no further stimulation of the immune system. after ingesting apoptotic cells, dendritic cells can mature only when there are danger signals expressed by the apoptotic cell or when dendritic cells are engulfing an excessive number of apoptotic cells. this leads to secretion of pro-inflammatory cytokines il- b and tnf-a by these signaled dendritic cells. inadequate clearance of surplus apoptotic cells results in these cells becoming necrotic and inducing a pro-inflammatory response. studies have demonstrated apoptosis of intestinal and splenic b cells, cd t cells and dendritic cells in sepsis. [ ] [ ] [ ] overwhelming infection should lead to massive clonal expansion of b and t lymphocytes but instead there is significant loss of these cell lines in sepsis. lack of stimulation by apc cells experiencing apoptosis leads to poor b cell and t cell stimulation. these unstimulated t cells are removed by apoptosis. in as many as % of bacteremic patients who die from systemic inflammatory response syndrome or multiple organ dysfunction syndrome, no focus of infection is identified. premature b cell and intestinal epithelial cell death through apoptosis in the intestines is one theory to explain intestinal bacterial translocation and loss of the first line of intestinal defense. treatment strategies current human treatment strategies for manipulating apoptosis focus mainly on cancer and sepsis. a pubmed literature search at the time of writing revealed a total of citations for apoptosis and veterinary. most of the studies are experimental at this point but many oncologic studies are manipulating apoptosis in the treatment of their patients (table ) . in human cancer research, peptidomimetics are in the early stages of experimental study. they are synthetic peptides that are resistant to enzymatic degradation and are being used for their pro-apoptotic effects on bid as well as their antagonism of iaps. , in cancer treatments, there are trials attempting to block the overexpression of bcl- because it is the cessation of normal apoptosis that leads to the growth of tumors. in addition to the direct effects on the apoptotic programs, discoveries are being made that allow chemotherapeutics to act synergistically with various anti-apoptotic therapeutics. treatment strategies for sepsis had previously targeted the hyperimmune phase rather than the hypoimmune phase. addressing apoptosis can be therapeutically challenging because targeted blockade of apoptosis in lymphocyte populations must be specific enough to primarily target those cell populations undergoing increased apoptosis and to be sufficiently transient to prevent the risk of malignant transformation associated with prolonged blockade of apoptosis. this poses a challenge to try to target specific pathways. attempts at blocking the circulating mediators and cytokines that induce apoptosis have not been successful because of the inherent redundancy and fail-safe mechanisms of the apoptotic pathways. caspase inhibitors show promise because caspases are common factors to many apoptotic pathways. broad-spectrum caspase inhibitors have been shown to prevent lymphocyte apoptosis and improve survival in animal models of sepsis. , , caution must be expressed, though, because increased dosages of caspase inhibitors can cause cytotoxicity and tnf-a induced injury. , gene therapy studies have shown that overexpression of bcl- delays or blocks apoptosis and improves survival in septic mice. , , other therapies target akt, a regulator of cell proliferation and death. mice overexpressing akt have reduced lymphocyte apoptosis and increased survival after cecal ligation and puncture. fas fusion proteins, and attempts at altering gene expression of members of the dr pathways, are also areas of ongoing apoptotic research. , [ ] [ ] [ ] [ ] recombinant human activated protein c, a product being used successfully in some septic patients, , may counteract the induction of apoptosis in monocyte and endothelial cell lines, modulating the inflammatory and coagulation cascades during sepsis. [ ] [ ] [ ] [ ] [ ] [ ] it also helps to attenuate the levels of pro-and anti-apoptotic proteins in favor of survival. in contrast to anti-apoptotic strategies, recent studies have addressed the hypothesis that apoptosis in sepsis may in some cases be beneficial by downregulating the inflammatory response. earlier onset of apoptosis may, in fact, favor survival. giarmarellos-bourboulis et al have associated monocyte apoptosis at the onset of sepsis to a favorable outcome due to a decrease in the amount of pro-inflammatory cytokines produced. apoptosis is a normal biologic process necessary to maintain cellular homeostasis. there are characteristic pathways that lead to this form of cell death continually influenced by local cellular events, growth factors, and neighboring stresses. this complicated system has numerous built-in avenues and failsafe mechanisms, including pro-apoptotic and anti-apoptotic factors. during sepsis and cancer, just of the many diseases causing dysregulation of the apoptotic process, cells are either killed too quickly or survive too long. newer therapies, designed to manipulate apoptosis depending on the pathology involved, can promote or delay this form of cellular demise. although an extremely complicated process, understanding the relationship between sepsis and apoptosis will undoubtedly lead to new treatment modalities. apoptosis: a basic biological phenomenon with wide-ranging implications in tissue kinetics the biochemistry of apoptosis robbins and cotran pathologic basis of disease regulationan of cell number in the mammalian gastrointestinal tract; the importance of apoptosis apoptosis and programmed cell death in immunity apoptosis and the immune system apoptosis: a different type of cell death morphological features of cell death cellular apoptosis and organ injury in sepsis: a review apoptosis: controlled demolition at the cellular level immunology of apoptosis and necrosis role of apoptotic cell death in sepsis the mitochondrion: is it central to apoptosis targeting apoptosis pathways in cancer targeted induction of apoptosis in cancer management: the emerging role of necrosis factor-related apoptosisinducing ligand receptor activating agents promoting apoptosis as a strategy for cancer drug discovery new approaches and therapuetics targeting apoptosis in disease drug insight: cancer therapy strategies based on restoration of endogenous cell death mechanisms apoptosis-based therapies for hematologic malignancies cd (fas/apo- ) signals ceramide generation independent of effector stage of apoptosis sphingosine kinase signalling in immune cells activation of bax by ceramide is independent of caspases regulatory role of cathepsin d in apoptosis role of mitochondria as the gardens of cell death cell death signalling pathways in the pathogenesis and therapy of haematologic malignancies: overview of apoptotic pathways a conserved xiap-interaction motif in caspase- and smac/diablo regulates caspase activity and apoptosis cd 's deadly mission in the immune system ways of dying: multiple pathways to apoptosis textbook of critical care mediators of endoplasmic reticulum stress-induced apoptosis caspase- and er-stressmediated apoptosis: the story so far coupling endoplasmic reticulum stress to the cell death program activation of caspase- , an endoplasmic reticulum (er) resident caspase, through tumor necrosis factor receptor associated factor -dependent mechanism in response to er stress an endoplasmic reticulum stress-specific caspase cascade in apoptosis activation and differentiation of autoreactive b- cells by interleukin induce autoimmune hemolytic anemia in fas-deficient antierythrocyte immunoglobulin transgenic mice the role of cap in cd signaling: new insights into the mechanism of procaspase- activation protein kinase c regulates fadd recruitment and death-inducing signaling complex formation in fas/cd -induced apoptosis structural requirements for signal-induced target binding of fadd determined by functional reconstitution of fadd deficiency a mechanism for death receptor discrimination by death adaptors death receptors death receptor signaling a novel protein that interacts with the death domain of fas/apo contains a sequence motif related to the death domain fadd, a novel death domain-containing protein, interacts with the death domain of fas and initiates apoptosis cytotoxicity-dependent apo- (fas/cd )-associated proteins form a death-inducing signaling complex (disc) with the receptor caspase- is an initiator caspase in death receptor signaling targeting death and decoy receptors of the tumornecrosis factor superfamily caspases: pharmacological manipulation of cell death the role of the bcl- protein family in cancer pharmacological manipulation of bcl- family members to control cell death caspases: enemies within extrinsic versus intrinsic apoptosis pathways in anticancer chemotherapy bcl-x(l) forms an ion channel in synthetic lipid membranes the bcl- protein family: arbiters of cell survival the role of bcl- family members in the progression of cutaneous melanoma death and anti-death: tumour resistance to apoptosis apoptosis in the development of the immune system bid, a bcl interacting protein, mediates cytochrome c release from mitochondria in response to activation of cell surface death receptors cleavage of bid by caspase mediates the mitochondrial damage in the fas pathway of apoptosis caspase cleaved bid targets mitochondria and is required for cytochrome c release, while bcl-xl prevents this release but not tumor necrosis factor-r / fas death bid-dependent and bid-independent pathways for bax insertion into mitochondria double identity for proteins of the bcl- family bcl- family proteins regulate the release of apoptogenic chytochrome c by the mitochondrial channel vdac in vivo mitochondrial p translocation triggers a rapid first wave of cell death in response to dna damage that can precede p target gene activation cell death induced by acute renal injury: a perspective on the contributions of apoptosis and necrosis new tricks of an old molecule: lifespan regulation by p apoptosis inducing factor (aif): a phyologenticially old, caspase-independent effector of cell death p expression and environmental tobacco smoke exposure in feline oral squamous cell carcinoma targeting of the c-abl tyrosine kinase to mitochondria in endoplasmic reticulum stress-induced apoptosis bax and bak can localize to the endoplasmic reticulum to initiate apoptosis human ice/ ced- protease nomenclature autoproteolytic activation of pro-caspases by oligomerization membrane oligomerization and cleavage activates the caspase- (flice/machalpha ) death signal granzyme b directly and efficiently cleaves several downstream caspase substrates: implications for ctl-induced apoptosis lamin proteolysis facilitates nuclear events during apoptosis caspase-dependent proteolysis of integral and peripheral proteins of nuclear membranes and nuclear pore complex proteins during apoptosis apoptotic dna fragmentation apoptosis: a basic biological phenomenon with wide-ranging implications in human disease autoantigens as substrates for apoptotic proteases: implications for the pathogenesis of systematic autoimmune disease identification of di-ablo, a mammalian protein that promotes apoptosis by bindingn to and antagonizing iap proteins smac, a mitochondrial protein that promotes cytochrome c-dependent caspase activation by eliminating iap inhibition mitochondria, the killer organelles and their weapons the mitochondrion in cell death control: certainties and incognita mitochondrial effectors in caspase-independent cell death a mitochondrial pool of sphingomyelin is involved in tnfalpha-induced bax translocation to mitochondria ceramide induces bcl dephosphorylation via a mechanism involving mitochondrial pp a ceramide induces mitochondrial activation and apoptosis via a bax-dependent pathway in human carcinoma cells the tnfr -traf signaling complex contains two novel proteins related to baculoviral inhibitor of apoptosis proteins suppression of tumor necrosis factor-induced cell death by inhibitor of apoptosis c-iap is under nf-kappab control nf-kappab antiapoptosis: induction of traf and traf adn c-iap and c-iap to suppress caspase- ativation iaps block apoptotic events induced by caspase- and cytochrome c by direct inhibition of distinct caspases human iap-like protein regulates programmed cell death downstream of bcl-xl and cytochrome c caspase activation, inhibition and reactivation: a mechanistic view bax-induced caspase activation and apoptosis via cytochrome c release from mitochondria is inhibitable by bcl-xl bcl- , bcl-xl and adenovirus protein e b kd are functionally equivalent in their ability to inhibit cell death a conserved xiap-interaction motif in caspase- and smac/diablo regulates caspase activity and apoptosis baiting death inhibitors cell death signalling pathways in the pathogenesis and therapy of haematologic malignancies: overview of apoptotic pathways inhibition of fas death signals by flips adhesion-mediated intracellular redistribution of c-fas-associated death domain-like il- -converting enzyme-like inhibitory protein-long confers resistance to cd -induced apoptosis in hematopoietic cancer cell lines constitutive expression of c-flip in hodgkin and reed-sternberg cells an inducible pathway for degradation of flip protein sensitizes tumor cells to trail-induced apoptosis selective inhibition of flice-like inhibitory protein expression with small interfering rna oligonucleotides is sufficient to sensitize tumor cells for trail-induced apoptosis flip prevents apoptosis induced by death receptors but not by perforin/granzyme b, chemotherapeutic drug, and gamma irradiation structure, regulation and funtion of nf-kappa b an essential role for nf-kappab in preventing tnf-alpha-induced cell death apoptosis, cross-presentation, and the fate of the antigen specific immune response cellular response to oxidative stress: signaling for suicide and survival clearance of apoptotic and necrotic cells and its immunologic consequences p biological network: at the crossroads of the cellular-stress response pathway and molecular carcinogenesis sir isaac newton, sepsis, sirs, and cars the pathophysiology and treatment of sepsis anti-inflammatory cytokines apoptosis in the development and maintenance of the immune system negative selection -clearing out the bad apples from the t-cell repertoire trauma: the role of the innate immune system accelerated lymphocyte death in sepsis occurs by both the death receptor and mitochondrial pathways apoptosis in sepsis: a new target for therapeutic exploration pathological aspects of apoptosis in severe sepsis and shock? dysregulated expression of neutrophil apoptosis in the systemic inflammatory response syndrome circulating mediators in serum of injured patients with septic complications inhibit neutrophil apoptosis through up-regulation of protein-tyrosine phosphorylation interleukin- counterregulates proinflammatory cytokine-induced inhibition of neutrophil apoptosis during severe sepsis upregulation of reactive oxygen species generation and phagocytosis, and increased apoptosis in human neutrophils during severe sepsis and septic shock impairment of function in aging neutrophils is associated with apoptosis neutrophils in the pathogenesis of sepsis the acute respiratory distress syndrome immune protection against septic peritonitis in endotoxin-primed mice is related to reduced neutrophil apoptosis neutrophil apoptosis in acute respiratory distress syndrome soluble fas ligand induces epithelial cell apoptosis in humans with acute lung injury (ards) silencing of fas, but not caspase- , in lung epithelial cells ameliorates pulmonary apoptosis, inflammation, and neutrophil influx after hemorrhagic shock and sepsis dendritic cell regulation of th -th development apoptosis in sepsis homeostasis and self-tolerance of the immune system: turning lymphocytes off homeostasis and self-tolerance in the immune system: turning lymphocytes off natural adjuvants: endogenous activators of dendritic cells cutting edge: bystander apoptosis triggers dendritic cell maturation and antigen-presenting function alveolar macrophage deactivation in murine septic peritonitis: role of interleukin immunoregulation of dendritic cells phagocytosis of apoptotic cells and immune regulation sepsis induces apoptosis and profound depletion of splenic interdigitating and follicular dendritic cells sepsis-induced apoptosis causes progressive profound depletion of b and cd t lymphocytes in humans depletion of dendritic cells, but not macrophages, in patients with sepsis activation of apoptosis in vivo by a hydrocarbon-stapled bh helix a small molecule smac mimic potentiates trail-and tnfalpha-mediated cell death considering immunomodulatory therapies in the septic patient: should apoptosis be a potential therapeutic target? blockade of apoptosis as a rational threapeutic strategy for the treatment of sepsis q-vd-oph, a broad spectrum caspase inhibitor with potent anti-apoptotic properties caspase activation is not death caspase inhibition causes hyperacute tumor necrosis factor-induced shock via oxidative stress and phospholipase a overexpression of bcl- in transgenic mice decreases apoptosis and improves survival in sepsis mitochondrial membrane potential and apoptosis peripheral blood monocytes in severe human sepsis akt decreases lymphocyte apoptosis and improves survival in sepsis inhibition of fas/fas ligand signaling improves septic survival: differential effects on macrophage apoptotic and functional capacity inhibition of fas signaling prevents hepatic injury and improves organ blood flow during sepsis in vivo delivery of caspase- or fas sirna improves the survival of septic mice leukocyte apoptosis and its significance in sepsis and shock hospital mortality and resource use in subgroups of the recombinant human activated protein c worldwide evaluation in severe sepsis (prowess) trial the effect of drotrecogin alfa (activated) on long-term survival after severe sepsis inuence of drotrecogin alpha (activated) infusion on the variation of bax/bcl- and bax/ bcl-xl ratios in circulating mononuclear cells: a cohort study in septic shock patients the apoptotic pathway as a therapeutic target in sepsis gene expression profile of antithrombotic protein c defines new mechanisms modulating inflammation and apoptosis recombinant human activated protein c attenuates the inflammatory response in endothelium and monocytes by modulating nuclear factor-kappab leukocyte and endothelial cell interactions in sepsis: relevance of the protein c pathway activated protein c blocks p -mediated apoptosis in ischemic human brain endothelium and is neuroprotective apoptosis: target for novel drugs early apoptosis of blood monocytes in the septic host: is it a mechanism of protection in the event of septic shock? key: cord- -vhuw gwn authors: demertzis, nicolas; eyerman, ron title: covid- as cultural trauma date: - - journal: am j cult sociol doi: . /s - - -z sha: doc_id: cord_uid: vhuw gwn this paper has two aims. the first is to introduce the concept of compressed cultural trauma, and the second is to apply the theory of cultural trauma in two case studies of the current covid- pandemic, greece and sweden. our central question is whether the pandemic will evolve into a cultural trauma in these two countries. we believe the pandemic presents a challenge to cultural trauma theory, which the idea of compressed trauma is meant to address. we conclude that, while the ongoing covid- pandemic has had traumatic consequences in sweden and greece, it has not evolved into cultural trauma in either country. this paper has two aims. the first is to introduce the concept of compressed cultural trauma and the second is apply the theory of cultural trauma in two case studies of the current covid- pandemic, greece and sweden. our central question is whether the pandemic will evolve into a cultural trauma in these two countries. we believe the pandemic presents a challenge to cultural trauma theory, which the idea of compressed trauma is meant to address. we begin with a short presentation of cultural trauma theory, and then discuss the idea of a compressed cultural trauma before moving into our two cases. there can be little doubt that the ongoing covid- pandemic poses a global threat that has created crisis on many levels, from local communities to states and nations. as eric woods et al. ( ) puts it, 'it has significant potential to trigger multiple, cascading crises in nearly every aspect of our lives. in addition to the presence of a threat, crises typically involve systemic disruption, uncertainty and stress'. a cultural trauma is a form of crisis, a crisis of identity that affects individuals and collectives. both greece and sweden experienced severe crisis during the first stages of the pandemic yet neither, to this point, have developed into cultural traumas. no matter how severe, not all crises become cultural traumas and the point of our comparison is to explain why. a cultural trauma occurs as the taken-for-granted foundations of individual and collective identity are shattered, setting in motion a discursive process to understand what happened, assign blame, and find pathways to repair an interpreted situation. in this ensuing meaning struggle various actors propose answers to these questions, articulating trauma narratives that compete for attention and acceptance. cultural traumas are not the aggregate of individual traumas, nor are they determined by gradations of suffering. cultural trauma is a specific form of collective trauma, affecting collective identity, where groups of individuals feel similarly affected by a fracturing of the existential security that a firm sense of identity affords. a step towards regaining that security, a human requirement, is to understand what happened and who or what was responsible, and then to act accordingly. cultural traumas imply anxiety and suffering, but also opportunity. the latter stems from the human capacity to learn and adjust to new conditions, to remake the world as well as to live in it. cultural traumas are usually studied in retrospect from a distant point in time, allowing one to isolate a point of origin, often a cataclysmic incident, and then trace the ensuing meaning struggle through a range of forums and media, toward memorialization and the impact with regard to collective memory. it is also possible, though more speculative, to study cultural trauma as it is occurring. an example is neil smelser's essay on / that appeared as an epilogue in cultural trauma and collective identity (alexander et al. ) . written just months after the event, smelser ( , p. ) suggests that the distinctive culture of the united states shaped the cultural trauma process in a way that was 'fused, telescoped and undifferentiated'. explaining this, smelser writes: 'with respect to the dimension of time alone, the traumatic process was truncated… the moment of the attacks to the recognition that they constituted a national trauma was a matter of short days, if not hours…the scope of the trauma and the identity of the victims were established immediately… there was an instant consensus that it was a trauma for everybody, for the nation… there was no significant divergence in the reactions of government and community leaders, the media, and the public in assigning meaning to the events as a national tragedy and outrage…there was little evidence of social division around the trauma' (smelser , p. ) . in this article, we look at the ongoing covid- pandemic through the lens of cultural trauma theory, asking what kind of trauma is this, for whom, and what are the possible outcomes? we will illustrate this further through the examples of greece and sweden, addressing the issue raised by smelser about how national cultures and a compressed time/space postmodern condition shape the trauma process. heuristically speaking, the pandemic covid- erupted as a total social fact manifested as an unexpected chain of ruptures (kasuga ) in all socioeconomic, political and cultural institutions and every aspect of everyday life at local, national and international levels. millions are experiencing the greatest disruption of their lifetimes. even during wwii businesses and schools remained open, restaurants served customers, major cultural and sporting events took place and life went on. there has never been anything affecting modern daily life as the current pandemic. it is imposing itself physically and virtually, as an image, with unprecedented and expanding force, condensing time and space worldwide in the most critical way, a phenomenon known as 'time-space compression' (harvey ) . some historians assure us that the coronavirus is a juncture where the river of history changes direction that could lead to the best of times or to the worst of times (macmillan ; garton ash ). as total social fact, the pandemic forces itself into existence in at least four interrelated ways: (a) it is global: it encompasses almost half of the world's population since trillion people are under lockdown as we write. due to international transportation and trading this virus has been transmitted with unusual speed in all continents causing very serious damage to most national economies and world finance markets. also, it deepened the antagonism between the usa and china with regards to g technology, the discovery of the vaccine, and the blame game over the cause of the pandemic. (b) it is risky and uncertain: the pandemic instantiates the quintessence of risk society; a hazardous situation systematically spread through global interconnectivity. it is individually experienced as an anticipated threat to be realized or not, eliciting negative and positive emotionality: anxiety, fear, sadness, and grief, for one's own vulnerability. at the same time, it evokes a sense of loss and anger; feelings of resentment, hope, confidence and trust in a government's efficiency or its opposite. for the time being, insofar as the genomic attributes of the virus are not fully discovered, we claim that the pandemic generates an emotional climate of uncertainty, not as in 'we know that we don't know', but as in 'we don't know what we don't know'. (c) it is highly mediatized: in variable proportions, individuals form a synthetic experience of the pandemic; either as infected or as locked down at home. they live the pandemic through firsthand experience and through the information disseminated by the communications media. the aesthetics of media coverage (dramatization, personalization, fragmentation) are likely to affect the way viewers perceive the health crisis and the concomitant risks. with a lack of effective medical care and the much sought-after vaccine, an individual's need for orientation increases all the more. this need for orientation is contingent upon the relevance of an issue for that individuals' life interests, and the degree of certainty they have concerning their knowledge about it. whenever an issue is highly relevant to one's personal agenda and one's information and knowledge about it is limited, the need for orientation increases. therefore, a person's media dependency increases and the media agenda setting effect rises with conspiracy theories being a possible side effect. with both the radical diversity in media access and representation, the conflict over what happened and who is responsible intensifies, adding more uncertainty with regard to orientation. who is one to believe? (d) it is traumatic: several hundreds of thousands of people around the world are grieving due to the unexpected and sudden loss of loved ones. those deaths often occurred under dire circumstances, in poorly equipped and crowded hospitals, or isolated and abandoned in caretaking facilities. in many countries, normal grieving rituals are prohibited for fear of further contagion. millions of middle aged or elderly people infected by the coronavirus are at home, vulnerable, insecure and isolated, while many others, tired of 'social distancing', harbor anxiety about the upcoming months and an uncertain future. photographs of empty and unpopulated metropolitan areas, in new york, london or paris, may be comprehensible, yet also inconceivable. this is certainly a situation full of traumatic potential. who could imagine the long lines of trucks carrying coffins in italy or the innumerable makeshift caskets stored in refrigerated vehicles outside new york hospitals, or the unclaimed victims buried in mass graves? one can rightly ask, is this europe? is this america? as we noted at the outset, studies of cultural trauma are carried out in retrospect, at the end of a process where discursive themes and central actors can be identified and studied over time. the covid- pandemic is ongoing, but as a total event, it already has many of the characteristics that circumscribe cultural trauma. these include a fundamental disruption of the taken for granted in daily life, a potential loss of trust in leaders and social institutions, negative attribution in the media, a contentious meaning struggle to determine what happened and who is responsible, with many competing accounts aired in various forums. there is constant reference to collective memory, the search for comparable historical incidents, as grounds for understanding, and various carrier groups have formed to articulate and defend their interpretation. this is certainly a collectively aggregated trauma: innumerable people are experiencing the pandemic as traumatic, characterized by a loss of existential security, a biopolitical condition that can potentially create new modalities of subjection and subjectivation, shaping both collective and individual subjectivities. the global dimensions of a public health crisis, its rapid expansion, and the instant circulation of images depicting deep human tragedy have already initiated a trauma drama process. the public discourse about the coronavirus and its effects is multifaceted, antagonistic, and replete with emotionality; it revolves around the meaning of normality, discipline, and self-discipline, trust, confidence and distrust toward institutions and political authority. in many countries, the covid- trauma drama has triggered new forms of artistic expression, such as online concerts, musical and theatrical performances, humorous offline and online creations, the production of books about pandemics throughout the centuries, novels and poetry. blame attribution and the politics of fear through the designation of the virus as 'invisible enemy' and bellicose metaphors of the efforts to cope with it as 'war', permeate public discourse in the same way warfare language was used to tame tb and cancer in the nineteenth and twentieth century. globalization and the accompanying ubiquitous presence of the internet and digital social media have contributed to an intensified time-space compression, whereby the trauma drama that is the core of cultural trauma phenomenon is deepened. in previous theorizing and comparative analysis, cultural trauma studies have noted a belated reaction to a triggering incident or series of incidents that leave indelible marks on collective memory and group identity. as with / and hurricane katrina, this seems no longer the case with regard to covid- , should this be generalized it would expand the trauma potential of related incidents. in many places, a sense of crisis, with a trauma potential, began immediately. under what is identified as time-space compression, some-if not all-elements of the tenets of cultural trauma are visibly present at once: (a) emotionality (mostly negative); (b) blame attribution (carrier groups, media claims etc.); (c) identity formation processes; (d) defense mechanisms (artistic creation). if the indelible marks on social body (i.e. change in personal and group identity, alteration in value priorities) are always noted at a later time, then the coronavirus crisis is not a cultural trauma proper. the 'period effect ' inglehart , pp. - ; norris and inglehart , p. ) of the health crisis has to be traced retroactively, not in anticipation. another issue to cope with is the distribution of the disease and its trauma potential. at first glance, the pandemic seems to be inherently universal because it threatens everyone, irrespective of class, ethnicity and religion. it might also be possible to be perceived differently from the way individuals perceive existential threats, war and terrorism for example, which are unambiguously anthropogenic and particularistic. however, the extant differences in death rates and different patterns of the disease dispersal inevitably poses once more the question 'trauma for whom?' and at what level. as already mentioned, the pandemic reveals the dynamics of risk society. yet, we should take heed of the fact that despite the genomic indeterminacy of the sarscov- there is a background assumption and a reasonable expectation about the upcoming cure and protection after a vaccine becomes readily available. the current pandemic is encountered by disease-experienced lay people and a knowledgeable community of experts. as a species, we are more or less confident that this is not a repetition of the black death. the covid- is not entirely surrounded with horrible and unsolvable mystery and it is not regarded as a totally intractable and capricious disease like tb or cancer before finding their cure (sontag , pp. - ) . this might mitigate the prospect of the pandemic to become a future point of reference for collective and/or global memory. not infrequently, harsh pandemics leave no trace in social memory. although the pandemic influenza virus identified in hong kong in caused one million casualties, it passed almost unnoticed, since people in many countries had other priorities (keck ; keck and lachenal ) . on the other hand, it might be plausible to argue that even if the number of deaths will not be comparable at all, for example, that of the great influenza or aids, it will mark collective memory and identities because it forcefully damaged the illusion of invulnerability modern science can confer. by turning a spotlight on the dark sides of the present-day technocracy and technopoly (postman ) , this pandemic may point to a threshold in the 'imaginary institution of society' creating an anticipation of human extinction. building up a 'pandemic imaginary' drawn from the 'apocalyptic catastrophism' of risk society, people are seen as unable to selfcreate as before (lynteris , pp. , ) ; as a total social fact, this pandemic like other pandemics and epidemics in the past, instills a feeling of being 'lost in the world' and as such is deeply traumatic. a pertinent question is whether we should think of the pandemic not only as a total social fact, but also in terms of a trauma-ridden 'historical event'; namely, as a ramified sequence of rupturing occurrences that is recognized as notable by contemporaries, and that results in a durable transformation of structures (sewell ) . currently, at least as far as most of the eu countries and the usa are concerned, it seems that the pandemic meets most of the criteria sewell ( , p. ff) identified as relevant in the conceptualization of historical events: dislocation and re-articulation of socioeconomic structures; cultural-aesthetic transformation; heightened emotionality; institution of rituals; conducive 'structure of the conjunction'. these are traits to be found in the accelerated trauma drama already unfolding in terms of time-space compression and ensuing presentism. the case in point, however, is not only the pace but also the depth and the possible duration of these traits, as we will discuss in our case studies. how then does this compression affect an unfolding trauma drama? eyerman ( , p. ) identified several factors that influence the development of a cultural trauma: timing, political context, performance of authority, mass-mediated representations, carrier groups, and collective memory. the relative weight of these factors is determinant in the appearance of cultural trauma. refers to the relative proximity of triggering incidents, the closer in time the more chance of a cumulative collective emotional response. under the conditions of compressed trauma, the triggering responses are continuous, with local, national and international reports of cases and deaths flowing together, intensifying anxiety and fear. everything seems to be happening at once and threats everywhere, undermining the existential security of individuals. the very air breathed cannot be trusted, turning family members and neighbors, as well as strangers, into potential carriers. basic notions of humanity, morality, and empathy are undermined; all that's solid melts into air. one of the first casualties of the pandemic was the cooperative relations between nations. if by globalization, one refers to the systematic interconnectedness and indepth interdependency of nations, the pandemic revived the nation-state as the natural site of collective reference. similar to the way the outbreak of the first world war fragmented the international socialist movement in a wave of emotional nationalism, the first response to the pandemic was to withdraw behind national boarders, to turn inward for protection. this was especially prominent in the european union, where borders were shuttered and international travelers looked upon with suspicion. like city walls built to keep out the plague, airports and train stations were manned with border guards. as the enemy was invisible, national identification became the main means of determining purity from danger. nevertheless, the cunning of history herein consists in the extant global cooperation between bio-laboratories and pharmaceutic big companies for the production of a vaccine, on the one side, and, on the other, the instant global transmission of news items about the disease which leaves room for empathetic feeling and acting towards distant others. in polarized political contexts, such as the u.s. the possibility of radically different accounts is more likely to appear. this is aided by the presence of alternative means of communication to distribute such accounts. as a rule, whenever there is dispute among elites over a public issue the media are prone to disseminate divergent accounts of it. central to the pandemic are public health authorities, usually medical professionals and epidemiologists who, more often than not, are represented as professional heroes. how well they argued for various control measures and how they interact with political authorities is crucial; as is the quality and availability of health care systems. in antagonistic political cultural contexts and discordant public spheres, the performance of national authorities and the relative strength of traditions of confidence in the institutions they represent call for more attention as drivers of cultural trauma. there is frequent use of war metaphors, bombast and fear mongering on the one side and cold, factual, scientific representations on the other. the general public is dependent on media for information and on authorities for dealing with an invisible and unknown enemy. it is through such means that what is happening and who is responsible is articulated and transmitted. the mass media create story lines, highlight victims and heroes and generally construct meaningful accounts of what is happening. articulate the alternative narratives, to interpret what is happening, who is responsible and what is to be done. besides the authorities mentioned above and those working in the mass media, important actors here are professional organizations, trade unions and other interest organizations that speak in the name of their members. in polarized contexts, the relative strength of these carrier groups, their access to mass media, and their support from political and religious authorities is crucial. those nations that have experienced trauma appear to be better prepared, provided with a memory and a means to deal with crisis (keck and lachenal , pp. , ) . 'this' is like 'that', we survived that and can survive this. in the process, various historical instances are recalled through previous trauma, from earlier pandemics to aids. such associations are meant to offer hope as well understanding. we can illustrate the interplay of these factors under the intense conditions of ongoing compressed cultural trauma through two short case studies. we chose these countries for practical reasons, from firsthand knowledge and because they represent varied responses to the pandemic. both countries had the advantage of some advance knowledge of the virus and the means of its spread. sweden has been held up as a unique approach, for some exemplary, for others dangerous. however, despite initial high levels of anxiety the pandemic did not evolve into cultural trauma in either, even with an exceptionally high death rate in sweden and the great difference in trust in authority that distinguishes the two countries. we hope to explain why. like many south and east european countries, greece was much less exposed to global transactions during the - winter and that might be a reason why the disease spread and the death toll remained at quite low levels compared to other eu countries. the first covid- case was diagnosed in greece on february th. the virus came through those returning from travel to holy sites in jerusalem and from a group of fur trade businessmen who visited milan a few days before. contact tracing was initiated on the first and all subsequent confirmed cases, with all contacts being tested and isolated. on february th, the annual carnival in patra (an event which draws large crowds from all over the country) was cancelled. on march th, with officially cases and deaths, all schools and universities across the country were closed. on march th, libraries, movie theaters, gyms and courtrooms were closed. on march th, with confirmed cases and death, malls, cafés, restaurants, bars, beauty parlors, museums and archaeological sites were closed. on march th, supervised beaches and ski resorts were also closed. all stores but big food markets were closed four days later and on march rd, a nation-wide lockdown was enforced, whereby citizens could leave their house only for specific reasons and with a special permit and under strict time limits. as we write confirmed cases amount to since the start of the outbreak with the death toll at . as a country that was pulling out of a decade-long financial crisis, greece is considered surprisingly successful in containing the pandemic. with its national health care system weakened by the crisis and with one of the oldest populations in the eu the outbreak of the disease alarmed the authorities and the majority of the people to the extent that greece would be the next italy or spain. at the outset, a gloomy atmosphere of fear and anxiety was documented in numerous opinion polls. soon after the curve began to flatten, hope, pride and confidence improved in the emotional agenda of the public sphere. since may th the lockdown measures are gradually loosening in view of approaching the greek summer. to avoid overcrowded hospitals with a limited number of intense care units, public authorities acted in a very cautious and unusually efficient way. mobilization had to be very fast because the health care system could not have otherwise handled the outbreak. the government has been strictly following the strategic directions given by a committee of experts in public health, composed of prestigious professors of epidemiology and other disciplines. the pandemic has been handled by a rather strong center-right government that took office in july , with a good number of technocratic members. in a polity traditionally permeated by fierce antagonism, apart from minor reservations opposition parties raised no substantive objections to the lockdown measures. there were however disputes and skirmishes on resource allocation policies, especially in view of the expected economic fallout caused by the pandemic. however, there has been consensus over the need to keep tourism alive during the summer in order to shield the per cent drop in gdp, a dire prospect that would make the greek economy the hardest hit across the entire eurozone. the necessity of the lockdown measures was also reinforced by the extremely poor reserves of test kits for the general population. until substantive amounts of kits were imported, less than one percent of the population had been tested for the virus. testing was focused on people admitted to hospitals with symptoms and those closely connected to confirmed cases, as well on all those returning from abroad. this raised concerns about the accuracy of the case numbers announced during briefings held by sotiris tsiodras, the president of the experts committee, a soft-spoken, infectious disease specialist at the university of athens, and nicholas hardalias, the deputy minister for civil protection. since the death toll was small, concerns were alleviated and these two became the most trusted authorities and their briefing became a sort of news event, thus crediting television with a trace of its old rhapsodic fun action (fiske and hartley ) . consecutive opinion polls during the spring of indicated an impressive increase in public confidence towards most social and political institutions of the country, as well as the raising popularity of political figures. for a good many years-if not decades-political institutions and the political authorities in greece were much discredited and this change in public mood might be taken as a signal of the remaking public trust in the face of a collective trauma. yet, that upswing was brought about more because of an ensuing 'rally round the flag' reaction, and was not a sign of a regained and restored institutional credibility. if anything, as documented by a social attitude research directed by nicolas demertzis in may , the covariance between institutional credibility and social trust was limited (r = . ), which means that amidst the trauma drama there was a paradox of trustworthy institutions and a misanthropic-like attitude. this paradox is contingent upon the duration of the rally round the flag effect. in this respect, if a deep recession is imminent, much of the confidence in political and social institutions will evaporate and the trauma drama will deepen, especially because any new economic crisis will overlap with the previous one, occurring no more than a year ago. after the crash test of the gradual removal of restrictions in may , the 'stay home' message of authorities was replaced by a 'stay safe' logo. yet, in many instances social distancing and other safety measures were not observed. in some cases where young people from extremist social networks initiated festive gatherings in public spaces violating any precaution, police had to intervene. in greece's over-politicized public culture, these incidents were enough to trigger grievances and controversy. more or less, however, they sprung more from spontaneous reactions to stringency and not as organized massive protest like the mobilizations in the usa, australia, poland, and germany, where lockdown is viewed as an unjustified tyrannical policy. it is striking that due to the customarily weak civil society in greece (demertzis ) , organizations representing the elderly and pensioners haven't been given (or taken) the opportunity to get involved in the debate concerning the repercussions of social distancing. after all, the chief objective of the entire strategy was to avoid a massive loss of older people who, nevertheless, in greece are not placed in care homes due to strong family ties. national and international ngos are at pains to warn authorities about the risk of the disease spreading among the overcrowded clusters of asylum seekers in identification and reception centers. although the living conditions in these camps are unacceptable, according to official sources only a few tested positive. yet, right-wing voices often decry them as 'health-bombs'. apart from the community of doctors and experts whose voice are respectfully heard, other carrier groups include think tanks organizing webinars on the socioeconomic impact of the pandemic, major cultural organizations like the national theater, the national opera, the onassis foundation and the stavros niarchos cultural center delivering a vast number of artistic creations and performances online. another influential carrier group has been the greek orthodox church. drawing from the byzantine tradition of lesser caesaropapism and its legendary role as the only religious and juridical-political institutional mediation of the orthodox populations under the ottoman rule, the orthodox church accommodated itself to the greek nation-state as a claimant of the national identity. we cannot delve into its impact and to greek religiosity except to stress its perpetual stronghold on public life to roughly understand the fierce dispute between political parties, secularized civil society groups, health experts groups, and religious cycles when the greek orthodox church announced on march that coronavirus could not be transmitted by communion wine or water. as the greek easter approached, a dispute grew with the government hesitant to ban public rituals due to its political cost, especially when orthodox hardliners and rogue bishops made public declarations and followed practices undercutting government efforts to come to grip with the pandemic through social distancing in the first crucial weeks in march. eventually, the government regulated the issue and easter church services were held without the participation of parishioners. after the gradual loosening of restrictions on may, church attendance was permitted as long as social distancing was observed. however, despite its marginal and grotesque character, it is striking that on may a media-based former bishop excommunicated the prime minister, the minister of education, and the deputy minister for civil protection. although this would have been much more serious if that gesture was undertaken by an ordinary bishop or the archbishop himself, it is indicative of the spiritual and political power the church entertains during the pandemic trauma drama. sweden's current statistics as of july , were , tested positive, deaths, of a population of . million. sweden's death toll, with . deaths per , , remains among the worlds highest. most of these deaths, %, were persons over years of age, many living in state run care facilitates. by comparison, greece's population is . million, cases and deaths, or . deaths per , . sweden has experienced few traumas of national proportion. it remained neutral during the two world wars and, as opposed to scandinavian neighbors, avoided the trauma of occupation and as opposed to greece its handling of recent financial crises has been exemplary. the two political assassinations that rocked the nation in and did not shake the foundations of collective identity as similar events did in the us and the netherlands (eyerman ) . the virus came to sweden through those returning from travel (over million returned to sweden during these early weeks), many from ski vacations in italy and other parts of europe, as this was winter sport vocation for students, and from china. the news of the treat however preceded the arrival. once the source was identified, identification, tracking of contacts and isolation was put in place. sweden's landbased access to the european continent closed when denmark shut its borders on march , setting up roadblocks on the bridge made famous by a television show. international rail and air traffic shut down for all travel deemed 'unnecessary'. the many swedes who commuted to work in denmark were forced to show proof of the necessity of their daily commute. the same was true for truckers carting food and other goods into sweden. effectively cut off from the rest of the world, swedish authorities designed their own path of response to the internal spread of the virus, though communication with other members of the european union continued. close contact with other scandinavian countries was also maintained. at the center of this strategic planning was the national public health authority (folkhalsomyndighten) working in close contact with elected political authorities. as distinctive from other countries, even its nordic neighbors, these exists a very tight institutional connection between heath and political authorities in sweden. however, there also exists distinctive regional autonomy. through its daily press conferences, anders tegnell, the 'state epidemiologist', became the public face of this strategy. at these televised press conferences, tegnell and other authorities offered statistics over the 'state of the cov- virus' in sweden and the rest of the world. interviewed for the journal nature, tegnell described the daily planning sessions and meetings with regional authorities that lay the foundation for the swedish response to the pandemic. at this point (april ) he noted that the death rates amongst older swedes was not yet 'traumatic', indicating that there was some critical threshold, presumably connected to public sensitivity, where a response by the authorities would become necessary. sweden is unique in the sense that officials held fast with this strategy of containment, with a minimum of enforced restrictions. the idea of managing the spread of the virus and 'flattening the curve' so that hospitals never became overwhelmed was generally discussed, including in the united states. however, most countries changed to much stricter restrictions after death rates in italy and spain climbed seemingly out of control. along with strict restriction of movement, germany and other countries began general testing in order to identify new cases and then trace contact networks in order to isolate those infected. such measures were accompanied by strict hygiene measures, such as wearing masks in public places. what was (perhaps wrongly) called the 'south korea solution' was followed by many european countries. sweden did not waiver however; it followed the original strategy of minimal restriction to manage the spread without overwhelming the health care system to reach 'herd immunity'. at the same time, after considering who was dying in other countries recommendations were made to protect those over the age of . defined as a risk group, those over were cautioned to stay at home, with food shopping and other essential activities suggested for special hours of the day. these were recommendations however, with no enforced restrictions or penalties, as in other parts of europe. this stems in part from the stipulations of the swedish constitution, which prohibit restrictions on mobility, but primarily on the belief that citizens would follow the stipulations, especially social distancing, on their own. as the very young appeared the least at risk, day care centers and grade schools remained open. an additional rationale was that those who worked in the vital health care system, doctors, nurses and others, were likely to have young children and in need of child care. high schools, colleges and universities shut down and their activities shifted to the internet. stores and restaurants stayed open, with social distancing requirements put in place. several restaurants were temporarily closed for not following social distancing recommendations. crowds of over persons were banned, thus cancelling sporting events, though teams were permitted to practice and, later, to play to empty stadiums. people were encouraged to exercise, especially outdoors, gyms, voluntarily closed at first, soon re-opened. as death rates climbed to alarming proportions compared to neighboring countries, it became clear those working in care facilities as well as those they cared for were especially vulnerable and visitors were forbidden. the specificity of the death rates, especially in care facilities, remains unclear and a matter of concern. a formal inquiry is now ongoing. as for testing, sweden began by opening public booths to administer tests, but soon abandoned this policy in favor of testing only those displaying specific symptoms, as well as professionals working in the care sector. there is no general testing policy or program, in part because general testing was deemed unreliable, as well as costly. anyone showing symptoms was advised to call a hot line and to stay at in an op-ed in the major national newspaper dagens nyheter in april signed by researchers with the headline 'the national board of public health has failed, now it time for politicians to step in' (https ://www.dn.se/debat t/folkh alsom yndig heten -har-missl yckat s-nu-maste -polit ikern a-gripa -in/), it was argued that the strategy being followed was not working and that elected officials ought to step in to change it. the authors pointed to italy as evidence, arguing that sweden should follow other european countries in imposing stronger restrictions. the public health authorities met this criticism by denying that reaching 'herd immunity' through such callous means was their strategy, rather the aim was to flatten the curve, to slow the spread of the virus but not eliminate it. the issue of herd immunity was the subject of another article in dagens nyheter on may . the paper's science editor defined herd immunity as a state where a sufficient number of a community is immune to an illness to prevent further spread. this can be reached, she pointed out, by vaccination or immunity from having the illness and carrying antibodies. the issue with covid- is twofold, one that it remains uncertain if one can become immune through having the virus and ) even if the body does produce antibodies, it is still uncertain how long they last. the article points out that the price of achieving this state too quickly is the death of many elderly. home until symptoms became acute. the population was told that the vast majority of those who fall ill with the virus will have only mild symptoms, and that the real danger is infecting others. sweden's death rate is significantly higher than its scandinavian neighbors and they have been some of its most vocal critics. the alleged 'herd immunity' policies were called immoral and utilitarian. while opening its boarder to germany, denmark has refused to do the same with sweden; greece announced an opening to european tourists, but excluded sweden as well. in defense, swedish authorities have publicly acknowledged failures with regard to elderly care facilities and new policies and finances are promised. the swedish prime minister explained the high death rate as 'communications problem' within the care sector, rather than a problem stemming from the overall strategy (interview in sydsvenska dagbladet july : a ). public debate has begun addressing this issue, though at this point relative political unity prevails even with a weak coalition government in place. all parties remain united behind the leadership, even the anti-immigrant party (though its leader has called the death rates among the elderly a 'massacre'). trust remains high and the daily press conferences of the public authorities are exemplary in their fact-based seriousness. these civil servants and the science-based expertise they represent continue to be held in high regard (though polls noted a % drop in approval between april and june), and the approval ratings given the social democratic prime minister have been rising, with an . % increase since the last survey in november . if the death rates continue to rise however, how long this trust in leadership and collective solidarity will last is an open question. at this point the feared 'second wave' has not occurred. even at this date, one can identify stages in an ongoing trauma drama. concerning the question of who is to blame: there was first denial, the hope that the virus could be contained in asia or other parts of the world. with the sudden rise in italy, there was the recognition that it could come to sweden. the presumptive early carriers were swedes returning from vacations and business travel; thus, they could not be outwardly blamed or stigmatized as outside carriers, however there were a few reported incidents of harassed asian travelers and immigrants as outside carriers. the right-wing internet has focused on immigrants as carriers and what they consider their 'over representation' in the number of cases. another form of denial, with the fact of rising death rates in sweden, is the claim that other countries are less accurate in their reporting, or that they are 'at a different place in the curve' and will soon catch up. both such claims have been made by swedish health authorities. following denial came acceptance. swedish authorities turned to modeling science and to managing the spread of the virus from that standpoint. it is a form of bargaining with natural forces and predicting that most of the cases will be mild, that primarily only the elderly are seriously at risk. this freed most of the population from some anxiety, but could induce guilt or worry about older relatives, who were being 'sacrificed for the herd', which is to say, the nation. the greater good for the greatest number. given these assumptions, daily life proceeded almost normally for a good portion of the population. there are several features of the swedish national culture that have influenced this strategy and its general acceptance. the first is a tradition of identification and trust in government and representative authorities in general. sweden lacks the strong opposition between the people and the state that exists in some other countries. there is also an underlying faith in science and technology and a good health care system available to all. all of which have contributed to a general following of the suggested regulations. the regulations are viewed as rational responses to an exceptional incident, not as restrictions imposed by an alien authority over the natural rights of individuals. swedish political life builds around consent and consensus and the general populace has been willing to give up some of the most basic human interactions, including the possibility to grieve their dead. survey results concerning trust in the authorities (https ://www.msb.se/conte ntass ets/ f a c fa c b e d / -msb_resul tat-coron aunde rsokn ng_ .pdf) with the highest ranked being the health care system ( % positive), the national public health authority ( %) the police %) lower down the government at %. one exception that became apparent during the pandemic can be found in some immigrant communities, where there remain traces of different social traditions of communication and interaction. an example is that of the tightly knit somali community in stockholm that was overrepresented among those infected, that either did not receive or ignored the imposed regulations. there is also evidence that trust in political authority is lower in immigrant communities. all this has been fodder for the extreme right and their anti-immigrant rhetoric, at least behind the scenes, on the internet, and on social media. the problem of social segregation has been further illuminated through the pandemic; one recent study revealed that death rates among middle-aged and older immigrants from syria, iraq and somalia were % higher for the months during the pandemic than for the same period last year. the differences for ethnic swedes for the same period was very much less (cited in sydsvenska dagbladet july a: ). in answering the question 'trauma for whom?', one would look to immigrant communities and minority groups, both for their being stigmatized as well as being disproportionally impacted by the virus itself. as in many parts of europe, immigrant groups and minorities work in some of the most vulnerable and exposed occupations, under the most precarious conditions. like the elderly, they are a group at risk, but unlike the elderly (an abstract category), they, as a stigmatized and stereotyped group, are not as respected. there are also specific demographic factors that have affected the levels of contagion in sweden, including the countries low population density, high share of single person households (not the cross generation extended family of italy or greece for example), generally high life expectancy, low level of chronic illness, low levels of obesity ( % as compared to % in the us, and low rates of diabetes ( . % as compared to . in the us). the high death rate however remains to be explained and public debate has begun. faith in the system and its representatives remains high, for covid- to lead to cultural trauma the sense of existential security in the general populace would have to be fractured. for the moment, the pandemic is viewed as a public health crisis, not a threat to basic values or collective identity(/ies). for that to happen trust in leadership and institutions and faith in science-based medicine would have to be broken. foundational issues, such as what it means to be a swedish citizen and the relationship between individual freedom and responsibility to the collective have been opened by the pandemic. an example being the need to follow directives that restrict individual mobility for the collective good. also being debated is the question of who should bear for the extra (financial, political, and moral) burden in a health care in a system where there are wide differences between regions affected by the virus; are localities where there are few cases of the virus responsible for those where there are many; should those living in southern sweden where there have been relatively few deaths be held accountable (by other countries) for those in other regions where there have been many? sweden is a nation very much aware of how it is viewed by the rest of the world. with its handling of the pandemic now the subject of global discussion and the resulting stigmatizing of swedish citizens as dangerous, potential carriers, strikes at the heart of the basic trust in political and expert authority and the relationship between state and citizen. if trusting authority leads to others distrusting swedish citizens, what then? with the proviso that the pandemic crisis and responses to it continue to change from day to day, we argue that the covid- pandemic can be usefully studied as a cultural trauma, but with modifications. previous studies have been retrospective accounts of the cultural trauma process, while here we offer a means to use the framework for a trauma that is ongoing. in this final section, we summarize these modifications, ending with comments on the outcome. in retrospective studies, timing has to do with the sequence of incidents that initiate and propel the trauma process. the flow of information is constant and confusing but the distance of time (and theory) one knows what to look for. in studying the trauma process as it occurs one does not have a clear notion of an underlying logic or an end in sight. in the current pandemic, nations face the future with uncertainly. there is no surety concerning the immunity of those who have been ill and no sure knowledge with regard to the availability of a vaccine. strategies of containment have had relative and varied success, but deciding when to end restrictions, to open borders and permit international travel is a guessing game, with frightening consequences. the fear of new waves of infection and rising death rates is real and most be included in all proposals. as in war, one needs exit strategies. all this increases uncertainly and anxiety amongst leaders and within the general population bringing to the fore a painful interplay between normality and exceptional. added to this is the prospect of future pandemics of similar consequence though with different cause, raising the possibility of cumulative affect with regard to collective trauma. retrospective studies can isolate relatively distinct incidents; compressed trauma faces the simultaneity of incidents without clear order or value. it is not only the fear of infection and death; there is also the collapsing economies and rising rates of unemployment that create another level of fear and anxiety within a population. these fears must be balanced in the strategies formulated by elites and is conditioned not only by levels of trust and faith in leadership but also by the type of rule. authoritarian or competitive authoritarian systems have a different relation to those they represent than democracies, where election cycles are an important rhythm and consideration in the choices made by those holding power. sitting leaders do not have the luxury of trial and error to the dimensions available in authoritarian systems, adding to the pressure imposed by time. in the midst of the pandemic, many countries have experienced collective solidarity and political unity, a real question is how long this will prevail? as a global historical event, the covid- pandemic appears as facilitator and accelerator of structural calibrations and cultural shifts. in all likelihood, it condenses social time and identity formation processes within the terms of late modern time-space compression. as with / and hurricane katrina, it triggered a trauma process from the very beginning giving us the opportunity to comparatively study it in vivo with the aid of cultural trauma theory. yet, a caveat to keep in mind is necessary at this point; time-space compression is both a result of technological and informational globalization and a globalizing driver of shifting our sense of time towards presentism. it brings about a 'nowist culture' (bertman ) , the 'tyranny of the moment' (eriksen ) , and an experience of a 'continuous present' in the sense that, severing the present from history. one lives in a 'flat collection and arbitrary sequence of present moments' (bauman , p. ) . therefore, it is not only a pandemic induced trauma that is deeply inscribed in the time-space compression, but also those commentators (like ourselves) who are hermeneutically dealing with it. this might express itself in hasty postures like those of giorgio agamben who, on february , , denounced the measures taken by the italian government as 'absolutely unwarranted', as a pretext to the imposition of a state of exception. to avoid the error of theorizing too quickly under the pressure of presentism, it is better to consider the pandemic as cultural trauma in the making. the lure of presentism can cause one to jump to conclusions, even if all the components of cultural trauma-with the possible exception of collective memoryappear at hand. a possible conceptual means to avoid hasty theorizing is to look at the pandemic not only as compressed, but also as virtual cultural trauma. as long as it 'is not the result of an event but the effect of a sociocultural process … the result of an exercise of human agency' (alexander , p. ), cultural trauma is not meant to be historically realized, but rather socially constructed. ontologically it is understood as a status nascendi, the specificity of which is conditioned by the interplay between virtuality and actuality (lévy ) . contrary to the predetermined correspondence of potentiality to reality, where all possible attributes of the real are already inherent in the potential, virtuality is replete with openness and contingency so that the actual outcome of an historical event emerges from unpredictable agentic action. as historical constructs cultural traumas may or may not emerge out of the discursive mediation of abrupt fractures of the societal fabric, and in this respect, as a compressed shocking experience, the pandemic trauma process signals a virtual cultural trauma. relatedly, as customarily held, traumas break and remake societal bonds regarding in-group and out-group relations. among others, the social marketing campaign for diminishing the pandemic in greece was a first-class opportunity for the restitution of the damaged national pride driven by the debt crisis. for nearly ten years greece was depicted in the international media as the black sheep of europe but now, with the successful handling of the first wave of the pandemic and employing an 'we and them' schema, a repetitive message campaign claims that 'this time others can learn from us'. in a retroactive way, the present trauma drama presents an opportunity to symbolically heal the past trauma of economic collapse. the greek government and the party in power are doing their best to regain trust through this success, however as mentioned above, this is a precarious and ephemeral enterprise due to the long-standing distrust of the greek public with regard to political institutions. if anything, according to certain estimations, implemented policies against the pandemic were more successful in low-trust societies than in high-trust nations. although efficient coping with the pandemic is a multi-variable task, the cases of greece and sweden seem to support this estimation. in the main, greeks complied with the stringency rules because they were afraid that a discredited political and administrative system would be unable to protect them. ultimately, greeks began feeling confident in the system as soon as it worked efficiently. this resulted from elites addressing the public in a convincing manner and keeping the death toll at a relatively low level. to the contrary, the allegiance of the swedish populace to the loose restrictions was premised on their prior well-established confidence in political institutions and scientific discourse. should the death toll continue to rise dramatically, the likelihood is that distrust will increase in sweden, as well as instigate a contentious political debate; yet this short-term entropic tendency will likely be counterbalanced by the long-term culture of trust in national institutions. as explained above, quite the opposite could be expected in greece. comparatively speaking the countries are almost entirely different: high trust versus low trust tradition; economic success and security versus the opposite; rational and secular value orientations versus traditional and religious. there is also a sharp contrast in their death and infection rates, with greece incredibly low and sweden the reverse. unless major upheavals in public trust, economic performance, and the number of deaths occur, the likelihood of a cultural trauma unfolding in either country is small for however different reasons. the compressed condition stemming from covid- is unlikely to end up doing serious harm to habitual value patterns and collective identification in either these two countries. how do we explain this? with regard to sweden, one can point once again to long-standing traditions of trust in national authorities and institutions. the performance of these authorities in articulating and managing the crisis potential was swift and transparent. with the pandemic framed and understood as a public health emergency, not a political crisis, the management of the threat was turned over to non-partisan health authorities. their televised press conferences, organized around factual presentations and preventive recommendations, were models of authoritative representation in their non-dramatic tone and format. these daily performances had a large and receptive audience. those entrusted the public good spoke with one voice. that the policy recommendations were minimally invasive and appeared to work was essential to their being followed. as the unusually high death rates became apparent, these were acknowledged as policy failures, with accompanying explanation and the promise of reparation. that these deaths could be explained as largely restricted to identifiable groups, the elderly and ethnic minorities, also freed the majority of the population from a degree of anxiety. that all political parties and mass media outlets accepted this definition of the situation is also an important factor in explaining the absence of collective trauma. had there been political contestation and mass-mediated rhetorical challenges to the prevailing framing, the trust in leadership and the management of the threat would have been more difficult. this is to speak of cultural trauma at the national level. there is greater potential for such trauma at the group level, most particularly amongst the elderly, immigrant groups and minorities, those, that is, that bear the statistical brunt of the virus. the isolation felt by the elderly, especially those in care facilities, has not been remedied and a sense of uncertainly remains as restrictions have not been lifted, nor has any time frame been noted. this, however, is largely a group without voice and representation, a statistical rather than social or political group. it is also a fleeting group, with little possibility for narrative collective identification or collective memory formation. this is not the case for ethnic minority groups, where the impact of the virus is great and where marginalization and stigmatization are prevalent. here there is more potential for individual trauma to find collective voice and representation. finally, with regard to collective memory, one can ask if the pandemic will leave an indelible mark on collective memory. the spanish flu left little or none. the theory of cultural trauma is a heuristic framework that offers little grounds for prediction. that said, given the secular nature of swedish society and the absence of strong commemorative traditions, there mostly likely will not be any collective memorialization for those who died of the coronavirus. a recent national recognition of the dead occurred after the tsunami, where over swedish tourists died. in , the swedish church and embassy organized a th anniversary ceremony and a memorial has been erected in stockholm. at a ceremony marking the th anniversary of the estonia ferry disaster of , the swedish prime minister called it 'a trauma for the entire nation' and representatives of the swedish royal family offered flowers during the memorial ceremony. the individualized deaths resulting from the covid- pandemic, visible primarily through statistics, do not appear to lend themselves to collective commemoration in the same way. against this however, one should point out that in both the other cases initiative for the memorial celebrations and national recognition of the victims came from the bottom up, from survivors and family members. the emergence of such carrier groups cannot be excluded in the covid- case. as for greece, the low death toll is unlikely to make a traumatic impact on public memory since it is represented as a mark of elite success, to further boost the rebound of the economy predicted for . most probably, any traumatic memory will be set aside in public discourse and mourning made a private matter, as is the case with the people who suffered to death amidst a wild fire in attica on july after a blatant failure of civil protection. albeit in that incident, there was clearly someone to blame for the hecatomb of burn victims, yet no indelible mark was left on social identity or public memory; this time there is virtually nowhere to place blame. insofar as losses remain low, individual family tragedies will not turn into a collectively shared suffering that could initiate a cultural trauma. a collective trauma might virtually arise if families were to lose the grandparents living at home. as family ties remain very strong in greece, living in care facilities or on their own is the exception, not the rule as it is in sweden and other countries. in a country with a shame-oriented culture, the mass loss of generational predecessors would inflict painful shame and guilt, strong emotions that when shared shatter individuals and collectives alike. such might be sufficient cause for the unfolding of a cultural trauma, especially if a reform in the protection measures was not forthcoming. for the time being however, in greece as in sweden, there has been a basic consensus among the elites regarding the handling of covid- ; therefore, the mainstream media were unable to amplify strategic differences and polarize public opinion. in greece however, this served to regain national pride, while it sweden it preserved it. we are not claiming an absence of trauma in these countries. as we indicated, despite great differences in the number of cases and the death tolls, both countries have their victims and large segments of the population that face the future with great anxiety. these are troubles confined and maintained in the private sphere. by and large, albeit references to 'trauma' are made in both countries, on a societal level the covid- is perceived and framed in terms of crisis and crisis management, rather than a traumatic catastrophe that fractures value priorities and collective representations. compressed cultural trauma means that the trauma drama begins immediately, but makes not prediction that it will result in indelible marks on the social body. if anything, the core idea of compressed cultural trauma idea is its virtuality, implying that the constellation of meaning making processes will shift when and if societal variables change. trauma. a social theory cultural trauma and collective identity berkeley from pilgrim to tourist-or a short history of identity hyperculture: the human cost of speed the political sociology of emotions. essays on trauma and ressentiment greece german cinema-terror and trauma: cultural memory since tyranny of the moment: fast and slow time in the information age the cultural sociology of political assassination a better world can emerge after coronavirus. or a much worse one the condition of postmodernity hegel's logic (trans: wallace, w.) total social fact: structuring, partially connecting, and reassembling the new emerging viruses are unpredictable. kathmerini simulations of epidemics: techniques of global health and neo-liberal government qu' est-e que le virtuel human extinction and the pandemic imaginary the world after covid- social capital. summing up the debate on a conceptual tool of comparative politics and public policy technolopy. the surrender of culture to technology logics of history. social theory and social transformation cultural trauma and collective identity illness as metaphor covid- as cultural trauma covid- , nationalism, and the politics of crisis: a scholarly exchange publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations nicolas demertzis is professor at the department of communications and media studies at the university of athens. since , he has been director and president of the board of administrators of the national centre for social research (ekke) we thank the editors for the opportunity to approach this subject and especially jeffrey alexander for the constructive criticism he offered. key: cord- -p cyzjwf authors: schneble, marc; nicola, giacomo de; kauermann, goran; berger, ursula title: nowcasting fatal covid- infections on a regional level in germany date: - - journal: nan doi: nan sha: doc_id: cord_uid: p cyzjwf we analyse the temporal and regional structure in mortality rates related to covid- infections. we relate the fatality date of each deceased patient to the corresponding day of registration of the infection, leading to a nowcasting model which allows us to estimate the number of present-day infections that will, at a later date, prove to be fatal. the numbers are broken down to the district level in germany. given that death counts generally provide more reliable information on the spread of the disease compared to infection counts, which inevitably depend on testing strategy and capacity, the proposed model and the presented results allow to obtain reliable insight into the current state of the pandemic in germany. in march , covid- became a global pandemic. from wuhan, china, the virus spread across the whole world, and with its diffusion, more and more data became available to scientists for analytical purposes. in daily reports, the who provides the number of registered infections as well as the daily death toll globally (https://www.who.int/). it is inevitable for the number of registered infections to depend on the testing strategy in each country (see e.g. cohen and kupferschmidt, ) . this has a direct influence on the number of undetected infections (see e.g. li et al., ) , and first empirical analyses aim to quantify how detected and undetected infections are related (see e.g. niehus et al., ) . though similar issues with respect to data quality hold for the reported number of fatalities (see e.g. baud et al., ) , the number of deaths can overall be considered a more reliable source of information than the number of registered infections. the results of the "heinsberg study" in germany point in the same direction (streeck et al., ) . a thorough analysis of death counts can in turn generate insights on changes in infections as proposed in flaxman et al. ( ) (see also ferguson et al., ) . in this paper we pursue the idea of directly modelling registered death counts instead of registered infections. we analyse data from germany and break down the analyses to a regional level. such regional view is apparently immensely important, considering the local nature of some of the outbreaks for example in italy (see e.g. , france (see e.g. massonnaud et al., ) or spain. the analysis of fatalities has, however, an inevitable time delay, and requires to take the course of the disease into account. a first approach on modelling and analysing the time from illness and onset of symptoms to reporting and further to death is given in jung et al. ( ) (see also linton et al., ) . understanding the delay between onset and registration of an infection and, for severe cases, the time between registered infection and death can be of vital importance. knowledge on those time spans allows us to obtain estimates for the number of infections that are expected to be fatal based on the number of infections registered on the present day. the statistical technique to obtain such estimates is called nowcasting (see e.g. höhle and an der heiden, ) and traces back to lawless ( ) . nowcasting in covid- data analyses is not novel and is for instance used in günther et al. ( ) for nowcasting daily infection counts, that is to adjust daily reported new infections to include infections which occurred the same day but were not yet reported. we extend this approach to model the delay between the registration date of an infection and its fatal outcome. we therefore analyse the number of fatal cases of covid- infections in germany using district-level data. the data are provided by the robert-koch-institute (www.rki.de) and give the cumulative number of deaths in different gender and age groups for each of the administrative districts in germany together with the date of registration of the infection. the data are available in dynamic form through daily downloads of the updated cumulated numbers of deaths. we employ flexible statistical models with smooth components (see e.g. wood, ) assuming a district specific poisson process. the spatial structure in the death rate is incorporated in two ways. first, we assume a spatial correlation of the number of deaths by including a long-range smooth spatial death intensity. this allows to show that regions of germany are affected to different extents. on top of this long-range effect we include two types of unstructured region specific effects. an overall region specific effect reflects the situation of a district as a whole, while a short-term effect mirrors region specific variation of fatalities over time and captures local outbreaks as happened in e.g. heinsberg (north-rhine-westphalia) or tirschenreuth (bavaria). in addition we include dynamic effects to capture the global changes in the number of fatal infections for germany over calendar time. this enables us to investigate the impact of certain interventions, such as social distancing, school closure, complete lockdowns and lockdown releases, on the dynamic of the infection and hence on the number of deaths. modelling infectious diseases is a well developed field in statistics and we refer to held et al. ( ) for a general overview of the different models. we also refer to the powerful r package surveillance . since our focus is on analysing the district specific dynamics of fatal infections we here make use of poisson-based models implemented in the mgcv package in r, which allows to decompose the spatial component in more depth. the paper is organized as follows. in section we describe the data. section highlights the results of our analysis. the remaining sections provide the technical material, starting with section where we motivate the statistical model, which is extended by our nowcasting model in section . extended results as well as model validation are given in section , while section concludes the paper. we make use of the covid- dataset provided by the robert-koch-institute for the districts in germany (which also include the twelve districts of berlin separately). the data are updated on a daily basis and can be downloaded from the robert-koch-institute's website. we have daily downloads of the data for the time interval from march , until today. the subsequent analysis was conducted on may , , and was performed considering only deadly infections with registration dates from march , until may , (the day before the day of analysis). the data contain the newly notified laboratory-confirmed covid- infections and the cumulated number of deaths related to covid- for each district of germany, classified by gender and age group. each data entry has a time stamp which corresponds to the registration date of a confirmed covid- infection. this means that the time stamp for a fatal outcome always refers to the registration date and not to the death date. due to daily downloads of the data we can derive the time point of death (or to be more specific, the time point when the death of a case is included in the database). we obtain the latter by observing a status change from infected to deceased when comparing the data from two consecutive days. the robert-koch-institute collects the data from the district-based health authorities (gesundheitsämter). due to different population sizes in the districts and certainly also because of different local situations, some health authorities report the daily numbers to the robert-koch-institute with a delay. this happens in particular over the weekend, a fact that we need to take into account in our model. we refrain from providing general descriptive statistics of the data here, since these numbers can easily be found on the rki webpage, which also gives a link to a dashboard to visualize the data (see also https://corona.stat.uni-muenchen.de/maps/) before we discuss our modelling approach in detail, we want to describe our major findings. first, table shows that age and gender both play a major role when estimating the daily death toll. as is generally known, elderly people exhibit a much higher death rate which is for the age group + around times higher than for people in the age group - . a remarkable difference is also observed between genders, where the expected death rate of females is around % (≈ − exp(− . )) lower than the death rate for males. furthermore, we see that significantly less deaths are attributed to infections registered on sundays compared to weekdays, due to the existing reporting delay during weekends. our model includes a global smooth time trend representing changes in the death rate since march th. this is visualized in figure . the plotted death rate is scaled to give the expected number of deaths per . people in an average district for the reference group, i.e. males in the age group - . overall, we see a peak in the death rate on april rd and a downwards slope till end of april. however, our nowcast reveals that the rate remains constant since beginning of may. note that this recent development cannot be seen by simply displaying the raw death counts of these days. the nowcasting step inevitably carries statistical uncertainty, which is taken into account in figure by including best and worst case scenarios. the latter are based on bootstrapped confidence intervals, where details are provided in section . later in the paper. our aim is to investigate spatial variation and regional dynamics. to do so, we combine a global geographic trend for germany with unstructured region-specific effects, where the latter uncover local behaviour. in figure we combine these different components and map the fitted nowcasted death counts related to covid- for the different districts of germany, cumulating over the last seven days before the day of analysis (here may , ). while in most districts of germany the death rate is relatively low, some hotspots can be identified. among those, traunstein and rosenheim (in the south-east part of bavaria) are the most evident, but greiz and sonneberg (east and south part of thuringia) stand out as well, to mention a few. a deeper investigation of the spatial structure is provided in section , where we show the global geographic trend and provide maps that allow to detect new hotspot areas, after correcting for the overall spatial distribution of the infection. on the day of analysis, we do not observe the total counts of deaths for recently registered infections, since not all patients with an ongoing fatal infections have died yet. we therefore nowcast those numbers, i.e. we predict the prospective deaths which can be attributed to all registration dates up to today. this is done on a national level, and the resulting nowcast of fatal infections for germany is shown in figure . for example, on may , there are deaths reported where the infection was registered on may th (red line on may th). we expect this number to increase to about when all deaths due to covid- for this registration date will have been reported (blue line on may th). naturally, the closer a date is to the present, the larger the uncertainty in the nowcast. this is shown by the shaded bands. details on how the statistical uncertainty has been quantified are provided in section below. the fit of this model has been incorporated into the district model discussed before, but the nowcast results are interesting in their own right. the curve confirms that the number of fatal infections is decreasing since the beginning of april. note that the curve also mirrors the "weekend effect" in registration, as less infections are reported on sundays. further analyses and a detailed description of the model are given in the following sections. let y t,r,g denote the number of daily deaths due to covid- in district/region r and age and gender group g with time point (date of registration) t = , . . . , t . here t = t corresponds to the day of analysis, which is may , and t = corresponds to march , . note that time point t refers to the time point of registration, i.e. the date at which the infection was confirmed. even though the time point of infection obviously precedes that of death, registration can also occur after death, e.g. when a post mortem test is conducted, or when test results arrive after the patient has passed away. we set the day of death to be equal to the day of registered infections in this case. the majority of fatalities with registered infection at time point t have not yet been observed at time t, as these deaths will occur later. we therefore need a model for nowcasting, which is discussed in the next section. for now we assume all y t,r,g to be known. we model y t,r,g as (quasi-)poisson distributed according to where we specify λ t,r,g through λ t,r,g = exp{(β + age g β age + gender g β gender + weekday t β weekday the linear predictor is composed as follows: • β is the intercept. • β age and β gender are the age and gender related regression coefficients. • β weekday are the weekday-related regression coefficients. • m (t) is an overall smooth time trend, with no prior structure imposed on it. • m (s r ) is a smooth spatial effect, where s r is the geographical centroid of district/region r. • u r and u r are district/region-specific random effects which are i.i.d. and follow a normal prior probability model. while u r specifies an overall level of in the death rate for district r over the entire observation time, u r reveals region specific dynamics by allowing the regional effects to differ for the last days. • pop r,g is the gender and age group-specific population size in district/region r and serves as an offset in our model. we here emphasize that we fit two spatial effects of different types: we model a smooth spatial effect, i.e. m (s r ), which takes the correlation between the death rates of neighbouring districts/regions into account and gives a global overview of the spatial distribution of fatal infections. in addition to that we also have unstructured district/region-specific effects u r = (u r , u r ) , which capture local behaviour related to single districts only. the district specific effects u r are considered as random with a prior structure for r = , . . . , . the prior variance matrix Σ u is estimated from the data. the predicted values u r (i.e. the posterior mode) exhibit districts that show unexpectedly high or low death tolls when adjusted for the global spatial structure and for age-and gender-specific population size. model ( ) belongs to the model class of generalized additive mixed model, see e.g. wood ( ) . the smooth functions are estimated by penalized splines, where the quadratic penalty can be comprehended as a normal prior (see e.g. wand, ) . the same type of prior structure holds for the region-specific random effects u r . in other words, smooth estimation and random effect estimation can be accommodated in one fitting routine, which is implemented in the r package mgcv. this package has been used to fit the model, so that no extra software implementation was necessary. this demonstrates the practicability of the method. the above model cannot be fitted directly to the available data, since we need to take the course of the disease into account. for a given registration date t, the number of deaths of patients registered as positive on that day, y t,r,g , may not yet be known, since not all patients with a fatal outcome of the disease have died yet. this requires the implementation of nowcasting. we do this on a national level, and cumulate the numbers over district/region r and gender and age groups g. this allows to drop the corresponding subscripts in the following and we simply notate the cumulated number of deaths with registered infections at day t with y t . let n t,d denote the number of deaths reported on day t + d for infections registered on day t. assuming that the true date of death is at t+d, or at least close to it, we ignore any time delays between time of death and its notification to the health authorities. we call d the duration between the registration date as a covid- patient and the reported day of death, where d = , . . . , d max . here, d max is a fixed reasonable maximum duration, which we set to days (see e.g. wilson et al., ) . the minimum delay is one day. in nowcasting we are interested in the cumulated number of deaths for infections registered on day t, which we define as the total number of deaths with a registered infection at t is apparently unknown at time point t and becomes available only after d max days. in other words, only after d max days we know exactly how many deaths occurred due to an infection which was registered on day t. we define the partial cumulated sum of deaths as on day t = t , when the nowcasting is performed, we are faced with the following data constellation, where na stands for not (yet) available: we may consider the time span between registered infection and (reported) death as a discrete duration time taking values d = , . . . , d max . let d be the random duration time, which by construction is a multinomial random variable. in principle, for each death we can consider the pairs (d i , t i ) as i.i.d. and we aim to find a suitable regression model for d i given t i , including potential additional covariates x t,d . we make use of the sequential multinomial model (see agresti, ) and define π(d; t, x t,d ) = p (d = d|d ≤ d; t, x t,d ) let f t (d) denote the corresponding cumulated distribution function of d which relates to probabilities π() through f t (d) = p t (d ≤ d) = p(d ≤ d|d ≤ d + ) · p (d ≤ d + ) = ( − π(d + ; ·)) · ( − π(d + ; ·)) · . . . · ( − π(d max ; ·)) for d = , . . . , d max − and f t (d max ) = . the available data on cumulated death counts allow us to estimate the conditional probabilities π(d; ) for d = , . . . , d max . in fact, the sequential multinomial model allows to look at binary data such that where • s (t) is an overall smooth time trend over calendar days, • s (d) is a smooth duration effects, capturing the course of the disease, • x t,d are covariates which may be time and duration specific. assuming that d, the duration between a registered fatal infection and its reported death, is independent of the number of fatal covid- infections, we obtain the relationship note further that if we model y t with a quasi-poisson model as presented in the previous chapter, we have no available observation y t for time points t > t − d max . instead, we have observed c t,t −t , which relates to the mean of y t through ( ). including therefore log f t (t − t) as additional offset in model ( ), allows to fit the model as before, but with nowcasted deaths included. that means, instead of λ t,r,g as in ( ), the expected death rates are now parametrized by λ t,r,g = λ t,r,g exp(log f t (t −t)), where the latter multiplicative term is included as additional offset in the model. we fit the nowcasting model ( ) with parametrization ( ). we include a weekday effect for the registration date of the infection with reference category "monday". the estimates of the fixed linear effects are shown in table . the fitted smooth effects are shown in figure , where the top panel shows the effect over calendar time, which is very weak and confirms that the course of the disease hardly varies over time. this shows that the german health care system remained stable over the considered period, and hence survival did not depend on the date on which the infection was notified. the bottom panel of figure shows the course of the disease as a smooth effect over the time between registration of the infection and death. we see that the probabilities π(d; ·) decrease in d, where this effect is the strongest in the first days after registration. thus, most of the covid- patients with fatal infections are expected to die not long after their registration date. the effect of d becomes easier to interpret by visualizing the resulting distribution function f t (d). this is shown in figure for two dates t, i.e.. april th and may th. the plot also shows how the course of the disease hardly varies over calendar time: in fact, the small differences between the two distribution functions is dominated by the weekday effect, since the red curve is related to a tuesday while the blue one is from a wednesday. in figure above we have shown the nowcasting results along with uncertainty intervals shaded in grey. these were constructed using a bootstrap approach as follows. given the fitted model, we simulate n = times from the asymptotic joint normal distribution ( ), where c t −t is the observed partial cumulated sum of deaths at time point t − t. the pointwise lower and upper bounds of the % prediction intervals for the nowcast for y t are then given by the . and the . quantiles of the set { y (i) t , i = , . . . , n}, respectively. in section we presented the fitted death rate, which is the convolution of a smooth spatial effect as well as region specific effects. it is of general interest to disentangle these two spatial components. this is provided by the model. we visualize the fitted global geographic trend figure : long term region specific level (left hand side) and short term dynamics (right hand side) of the covid- infections m (·) for germany in figure . the plot confirms that up to may the northern parts of the country are less affected by the disease in comparison to the southern states. the two plots in figure map the region specific effects, i.e. the predicted long term level of a district u r (left hand side) and the predicted short term dynamics u r (right hand side). both plots uncover quite some region-specific variability. in particular, the short term dynamics captured in the right hand side plot (u r ) pinpoint districts with unexpectedly high nowcasted death rates in the last two weeks, after correcting for the global geographic trend and the long term effect of the district. some of the noticeable districts have already been highlighted in section above, but we can detect further districts, which are less pronounced in figure . for instance, steinfurt (in the north-west of north rhine-westphalia), olpe (southern north rhine-westphalia) or gotha (center of thringen) presently show a high rate of fatal infections. a large number of the registered deaths related to covid- stem from people in the age group +. locally increased numbers are often caused by an outbreak in a retirement home. such outbreaks apparently have a different effect on the spread of the disease, and the risk of an epidemic infection caused by outbreaks in this age group is limited. thus, the death rate of people in the age group + could vary differently across districts when compared to regional peaks in the death rate of the rest of the population. in order to respect this, we decompose the district-specific effects u r in ( ) into u − r = (u − r , u − r ) for the age group -and u + r = (u + r , u + r ) for the age group +, where the age group -consists of the aggregated age groups - , - and - . we put the same prior assumption on the random effects as we did in ( ), but now the variance matrix that needs to be estimated from the data has dimension by . the fitted age group-specific random effects are shown in figure , where the u − r are shown in the top panel and the u + r in the bottom panel. most evidently, the variation of the random effects is much higher in the age group + when compared to the younger age groups, as more districts occur which are coloured dark blue or dark red, respectively. when comparing the district-specific short term dynamics of the last days (u r ) in figure to those in figure , we recognize that in most of the districts which recently experienced very high death intensities (with respect to the whole period of analysis), these stem from the age group +. as mentioned before, this can often be explained by outbreaks in retirement homes. when fitting the mortality model ( ) we included the fitted nowcast model as offset parameter. this apparently neglects the estimation variability in the nowcasting model, which we explored via bootstrap as explained in section . and visualized in figure . in order to also incorporate this uncertainty in the fit of the mortality model, we refitted the model using (a) the upper end and (b) the lower end of the prediction intervals shown in figure . it appears that there is little (and hardly any visible) effect on the spatial components, which is therefore not shown here. but the time trend shown in figure does change, which is visualized by including the two fitted functions corresponding to the . % and . % quantile of the offset function. we can see that the estimated uncertainty of the nowcast model mostly affects the last ten days, with a strong potential increase in the death rate mirroring a possible worst case scenario. in figure we show a normal qq-plot of the pearson residuals in the nowcasting model. apart from some observations in the lower tail, the pearson residuals are distributed very closely to a standard normal distribution when considering the estimate φ = . of the dispersion parameter in the quasi-poisson model ( ). overall, the model seems to fit to the available data quite well. the paper presents a model to monitor the dynamic behaviour of covid- infections based on death counts. it is important to highlight that the proposed model makes no use of new infection numbers, but only of observed deaths related to covid- . this in turn means that the results are less dependent on testing strategies. the nowcasting approach enables us to estimate the number of deaths following a registered infection today, even if the fatal outcome has not occurred yet. moreover, the district level modelling uncovers hotspots, which are salient exclusively through increased death rates. a differential analysis of the number of current fatal infections on a regional level allows to draw conclusions on the current dynamics of the disease assuming a constant case fatality rate, i.e. a stable proportion of death compared to the true number of infections when adjusting for age and gender. a natural next step would now be to consider the nowcasted deaths in relation to the number of newly registered infections, which is, in contrast, highly dependent on both testing strategy and capacity. we consider this as future research, and the proposed model allows us to explore data in this direction. this might ultimately help us in shedding light on the relationship between registered and undetected infections as well as on the effectiveness of different testing strategies. there are several limitations to this study which we want to address as well. first and utmost, even though death counts are, with respect to cases counts, less dependent on testing strategies, they are not completely independent from them. this applies in particular to the handling of post-mortem tests. we therefore do not claim that our analysis of death counts is completely unaffected by testing strategies. secondly, a fundamental assumption in the model is the independence between the course of the disease and the number of infections. overall, if the local health systems have sufficient capacity and triage can be avoided, this assumption seems plausible, but it is difficult or even impossible to prove the assumption formally. finally, the nowcasting itself is not carried out on a regional level, though the model focuses on regional aspects of the pandemic. while it would be desirable to fit the nowcast model regionally, the limited amount of data simply prevents us from extending the model in this direction. analysis of ordinal categorical data real estimates of mortality following covid- infection countries test tactics in 'war' against covid- estimating the number of infections and the impact of non-pharmaceutical interventions on covid- in european countries critical care utilization for the covid- outbreak in lombardy, italy: early experience and forecast during an emergency response baseline characteristics and outcomes of patients infected with sars-cov- nowcasting the covid- pandemic in bavaria probabilistic forecasting in infectious disease epidemiology: the th armitage lecture bayesian nowcasting during the stec o :h outbreak in germany real-time estimation of the risk of death from novel coronavirus (covid- ) infection: inference using exported cases adjustment for reporting delays and the prediction of occurred but not reported events substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (sars-cov ) incubation period and other epidemiological characteristics of novel coronavirus infections with right truncation: a statistical analysis of publicly available case data covid- : forecasting short term hospital needs in france. medrxiv spatio-temporal analysis of epidemic phenomena using the r package surveillance quantifying bias of covid- prevalence and severity estimates in wuhan, china that depend on reported cases in international travelers infection fatality rate of sars-cov- infection in a german community with a super-spreading event smoothing and mixed models case-fatality risk estimates for covid- calculated by using a lag time for fatality generalized additive models: an introduction with r we want to thank maximilian weigert and andreas bender for introducing us to the art of producing geographic maps with r. moreover, we would like to thank all members of the corona data analysis group (codag) at lmu munich for fruitful discussions. key: cord- -pua ogz authors: coker, eric s.; cavalli, laura; fabrizi, enrico; guastella, gianni; lippo, enrico; parisi, maria laura; pontarollo, nicola; rizzati, massimiliano; varacca, alessandro; vergalli, sergio title: the effects of air pollution on covid- related mortality in northern italy date: - - journal: environ resour econ (dordr) doi: . /s - - - sha: doc_id: cord_uid: pua ogz long-term exposure to ambient air pollutant concentrations is known to cause chronic lung inflammation, a condition that may promote increased severity of covid- syndrome caused by the novel coronavirus (sars-cov- ). in this paper, we empirically investigate the ecologic association between long-term concentrations of area-level fine particulate matter (pm( . )) and excess deaths in the first quarter of in municipalities of northern italy. the study accounts for potentially spatial confounding factors related to urbanization that may have influenced the spreading of sars-cov- and related covid- mortality. our epidemiological analysis uses geographical information (e.g., municipalities) and negative binomial regression to assess whether both ambient pm( . ) concentration and excess mortality have a similar spatial distribution. our analysis suggests a positive association of ambient pm( . ) concentration on excess mortality in northern italy related to the covid- epidemic. our estimates suggest that a one-unit increase in pm( . ) concentration (µg/m( )) is associated with a % ( % confidence interval: – %) increase in covid- related mortality. valley, an extension of flat river lands enclosed between the alps and apennines mountains, which causes the stagnation of pollutants due to low ventilation (giulianelli et al. ) . these factors help to characterize the po valley's peculiarity with respect to different european areas with comparable urban and industrial density levels (eeftens et al. ) . moreover, in addition to the urbanized and industrial areas, the remainder of the valley presents an intensive agricultural activity. local studies on emission sources highlight a varying composition of the final concentration values depending on the position of monitoring stations and with different sources acting as local or diffused ones (for instance having high emissions from traffic close to cities, while having background biomass burning diffused in the whole region) (bigi and ghermandi ; larsen et al. ) . indeed, given the eu ambient air quality directives that sets the air quality standards for the protection of health at μg/m for the averaging period of a calendar year, the po valley shows values consistently near or above the threshold. these values often range in the - μg/m interval with peaks of > μg/m , which in europe are only matched in southern poland and other smaller eastern european clusters (eea ). compared to its overall representation in the population, lombardy is disproportionately impacted by covid- related mortality, with approximately % of italy's covid- deaths as of april , (odone et al. ) . lombardy is also the most impacted italian region as far as the total number of deaths in excess in the first quarter of compared to the same period of the previous years. comparing the official covid- death data with registry deaths, it emerges that the latter is almost % larger than the former in lombardy, % larger in emilia-romagna and % and % in veneto and piemonte, respectively. it is, therefore, imperative to consider the role that pm may have played in such disproportionate covid- deaths in northern italy. there are a number of plausible pathways by which airborne pm may impact covid- related morbidity and mortality. existing data already finds a strong positive correlation between viral respiratory infection incidence and ambient pm concentrations (ciencewicki and jaspers ; sedlmaier et al. ). one plausible pathway for this phenomenon is the fate and transport of the virus itself within the environment. a recent position paper by the italian society of environmental medicine argues that pm may act as both a carrier and substrate of the virus and thus influence the virus' fate and transport in the environment and reaching susceptible receptors (setti et al. ) . another pathway is the increase in susceptibility to covid- mortality caused by long term exposure to pm. fine pm is already known to affect cardiovascular and respiratory morbidity and mortality (cakmak et al. ; jeong et al. ; mcguinn et al. ; yin et al. ) . moreover, among italian covid- patients who died in the hospitals, and for whom it was possible to analyze clinic charts, data showed substantial comorbidities including ischemic heart disease ( . %); atrial fibrillation ( . %); heart failure ( . %); stroke ( . %); hypertension ( . %), and chronic obstructive pulmonary disease ( . %) (istituto superiore di sanità ). biologically, long-term pm exposure may be responsible for a chronic inflammation status that causes the hyper-activation of the immune system and the life-threatening respiratory disorders caused by covid- (shi et al. ) . some preliminary evidence is now emerging about covid- that shows a positive relationship between air pollution and morbidity and mortality. beyond qualitatively describing the european air quality index for northern italy to argue the causal role of air pollution and the relatively high covid- mortality observed in that region, conticini et al. ( ) review the most recent existing toxicological and epidemiological literature. based on existing evidence from other empirical studies, they clarify the relationship between air pollution, prolonged inflammation and immune system hyper-activation and immune suppression, and the link between the latter and acute respiratory distress syndrome, and respiratory mortality. their paper is important in that it suggests a clinical and biologically plausible explanation to our analysis, but does not provide statistical evidence in support of the hypothesis. a separate empirical analysis by becchetti et al. ( ) finds preliminary evidence that confirms such a positive effect of air pollution on mortality in italy based on the analysis of death data at the province level. similarly, wu et al. ( ) show a positive association between long term pm exposure and covid- related deaths in us counties. ogen ( ) recently analysed data from administrative regions in france, spain, italy, and germany, and found that the highest covid- deaths in these regions were associated with five regions of northern italy that also corresponded with the highest levels of atmospheric nitrogen dioxide (no ). cole et al. ( ) estimate the same relationship using netherlands municipality data and find pm . positively associated with covid- cases, hospitalization, and deaths. in this paper, we follow this emerging stream of the empirical literature and test the hypothesis that a higher average long-term exposure to pm . is positively associated with the current extraordinarily high death toll in northern italy. we decided to focus on pm . because, given the complexity of air pollution, it is quite common in air pollution epidemiology studies to focus the analysis on a single pollutant (wu et al. ) , although multipollutant analyses are certainly warranted. we selected pm . for a variety of important reasons, including policy implications and evidence in the literature in terms of chronic health effects. regarding its policy implications, we selected pm . as opposed to pm because the former is more correlated with human activities than the latter, and it correlates with stronger health effects than pm does. with respect to respiratory mortality effects from the existing air pollution literature, the most robust evidence points to pm . as opposed to other gaseous air pollutants (bowe et al. ) . mortality data are collected at the municipality level for the period january-april . given that mortality data are not disaggregated by mortality cause, death counts are measured as the difference from the last five-years mean to reflect the abnormal number of deaths caused by the spreading of the pandemic. since pm . can be associated to generic mortality even in the absence of the pandemic outbreak (dominici et al. ; katsouyanni et al. ; samet et al. ) , we also estimate the impact of pm . on the excess mortality in the sample using data, a time in which the coronavirus epidemic had presumably not yet begun. data on pm . concentration at the municipality level refer to the years prior to account for long-term population exposure. we assign municipality pm . concentration by a set of different methods of spatial interpolation (kriging) of monitoring station data related to the years - . we estimate a negative binomial model of excessive deaths on historical pm . concentrations and a series of control variables that may plausibly affect both pm . concentration and mortality, including population density; the spatial concentration of the industrial manufacturing sites; climatic conditions observed during the first quarter of ; and the demographic composition of the municipal population among others. in addition, we consider spatial heterogeneity in the distribution of the number of deaths related to regional and local unobservable factors. we account for region-specific effects because regions, in italy, are the administrative units in charge of managing the health systems and the measures taken to trace and contrast the spreading of the pandemic varied greatly among even contiguous regions. we also account for local effects common to functionally linked clusters of municipalities (the local labour systems-lls). we deem this part of the identification strategy crucial because the relationship between pm . and covid- mortality may be confounded by several other factors, some of which were not observable or measurable, but are nevertheless intrinsically related to the geographical location of the observed units. the remainder of the paper is organized as follows. the next section introduces the empirical strategy and describes the dataset. the results are presented and discussed in section three, considering the total number of (excess) deaths. section four draws the conclusions and highlights the limitation of the study and the indications for future research. our analysis is restricted to the study area of northern italy (fig. ) , which encompasses the sub-regions of valle d'aosta, piemonte, liguria, lombardia, emilia-romagna, veneto, friuli-venezia giulia and trentino-alto adige/südtirol. official territorial data on covid- mortality in italy are available at the rather aggregate regional or provincial level, corresponding to the levels and , respectively of the european nomenclature units for territorial statistics (nuts). in addition, these official data refer to the deaths of patients tested positive for severe acute respiratory syndrome coronavirus (sars-cov ) only and do not include (potential) patients without covid- diagnosis because they were not tested and died at home or elsewhere. hence, the officially reported deaths are likely underestimated. because testing policies vary among regions in italy, the induced measurement error is also non-randomly distributed among the provinces. ciminelli and garciamandicó ( ) compare the official covid- fatality rates with historical death data and working under the assumption that covid- deaths are underestimated in italy, the choice is made in this paper to use the total deaths from the official registries, accordingly, and to scale the analysis at the municipality level, the smallest administrative units, to have a more granular representation of the spatial dimension of the phenomenon. since we are interested in excess deaths, we take the difference between the number of deaths in the period january -april , , and the average number of deaths in the same period of the previous years (exdeaths) and use this metric as the dependent variable in our statistical model. figure displays the geographical distribution of the above-described data among the municipalities for which data is available. the variable is assumed to follow a negative binomial distribution, a generalization of the poisson distribution that avoids the restrictive mean-variance equality of the latter, and is modelled as follows: where is the overdispersion parameter to be estimated and i is the municipality-specific expectation conditional on the value of the covariates. among the covariates, pm is the concentration of fine particulate matter in municipality i and is the associated parameter, which we expect positive and statistically different from zero; x is a vector of control variables that adjusts for the potential confounding effects and includes the (log of) total population as the offset while is a normally-distributed error term. our main source of pm . data is the european environmental agency's (eea) air monitoring database, which is provided to eea by the institute for environmental protection and research (ispra). ispra conducts ground-level air measurements of pm . air concentrations (µg/m ) collected at monitoring sites throughout italy. specifically, we use the eea's e a and e a datasets, which are primary validated assessment data and primary up-to-date assessment data reported by the european member states, respectively. although the measurements come both in hourly and daily averaging formats, we work ( ) does not include a time component, we further compute a six-year averaging time to obtain a metric of long-term (chronic) pm . concentration levels throughout different spatial units of northern italy. the number of years for the reference period is sufficiently long to account for long-term exposure while being not too long to be affected by the mobility of people among municipalities, and it is in line with existing literature assessing long-term effects of pm exposure (yorifuji et al. ). since the air monitoring stations provide only partial spatial coverage for municipality-level pm . concentration data, we impute missing observations using a spatial interpolation model. specifically, we fill in the gaps using a mean stationary ordinary kriging (see bivand et al. , p ) defined through an exponential covariance function with nugget, partial sill and range parameters estimated through (restricted) maximum likelihood methods. figure displays the resulting pm . concentration data. comparing figs. and , it is possible to visually appreciate a spatial coincidence between higher levels of excess mortality and higher levels of pm . , in particular in the lombardia region which notably is the region with both the highest particulate concentration and the highest number of excess mortality. the hypothesis that pm . concentration affected covid deaths, that is ̂> , is tested among several possible specifications. in model ( ) we include regional effects j . these effects are expected to capture the aspects related to the management of the outbreak, which may have systematically influenced covid- mortality and that are common to all the municipalities in the same region. italy has a national health system that ensures equal access to healthcare to all citizens. the system is managed by regions at the local level, and, in the specific case of this pandemic, regions were responsible for defining the testing and contact-tracing protocols and implementing the necessary measures to contain the outbreak, among which the measure to protect healthcare workers. in model ( ), we include lls-specific effects e k . lls are spatial clusters of contiguous municipalities related by commuting flows that share a common specialization in a specific sector of manufacturing production and correspond to the conceptualization of marshallian districts (becattini ) . the number of lls clusters per-region and the total number of municipalities belonging to clusters are reported in table , along with the minimum, maximum, and average cluster size. the use of lls captures the interlinkages within neighbouring municipalities that may have favoured the geographical spreading of coronavirus around specific hotspots. mortality data are then expected to vary among municipalities in different lls, but differences are expected to be non-systematic in this case. in model ( ) we include both the regional fixed effects and the lls random effects. control variables to be included in the model were chosen to avoid any potential spatial confounding effect and considering as well the emerging literature on the impact of pm on covid- related deaths (cole et al. ; wu et al. ). the population density and per-capita income account for urbanisation level. the most densely populated and wealthy municipalities are among the most polluted due to the spatial concentration of manufacturing and service activities but are also the places where the contagion could have been easier, with a potential impact on mortality. in addition to the density of population, the shares of municipality area occupied by industrial sites and the average size of manufacturing firms are included in the regression because they are related to pollutant concentration and possibly to mortality. national measures to stop the spreading of the viral infection (lockdown) involved the service sector to the largest extent while many manufacturing activities, being considered necessary, were left open and, in the absence of social distance and individual protection measures, the geographical concentration of these activities in a municipality with their complex logistics and transport interconnections, and the size of plants, may have influenced mortality. average temperature, for which an association with covid- deaths has also been found , is also included in the regression. moreover, covid- incidence has proven to be higher among men than women and people aged or more. hence these two variables are considered in the model, even though these aspects are not necessarily connected with the average pm . exposure in a municipality. underlying socioeconomic conditions can also play a role in covid- related mortality (goutte et al. ) . brandt et al. ( ) and mukherji (n.d.) have shown that, in the us, covid- is more threatening for ethnic minorities, and we believe that the share of migrants, identified as non-eu citizens, can control for this aspect influencing the observed excess mortality. on the other hand, mukherji ( ) and goutte et al. ( ) also find that places with a higher share of the population with a low level of education have higher deaths. in our paper, given the lack of updated data on education at the municipal level, we proxy it with the percentage of university students on the total population. the distance from the closest airport is a proxy for the functional and relational linkage between a municipality and a place of highly frequent national and international connections and potential sources of coronavirus spreading. finally, we consider the number of hospital beds as a proxy for the supply of health services to account for the fact that many people died at home without being diagnosed for coronavirus due to the shortage of beds in public structures. the full details of the variables in the model, including sources and summary statistics, are presented in table . having accounted for the confounding effect due to the omission of relevant information from the empirical specification, we exclude any other potential source of endogeneity considered in similar papers. in particular, we exclude endogeneity due to measurement error in the outcome variable and the main independent variable. concerning the outcome variable, the relationship between deaths and cases with fine pm could be spurious because more cases could be registered, and more individuals tested in highly polluted areas as people there are more likely to show covid- symptoms due to the chronic inflammation induced by pm. the high toll of deaths of people diagnosed with covid- would be a natural consequence of that. in contrast, the number of deaths in excess, used in this paper, is not affected by testing problems since it considers all the potential covid- deaths. concerning the pm variable, measurement errors are likely to occur when using satellite data or modelled data. we preferred to use pm . levels observed from monitoring stations to avoid such a measurement error. some caution is needed in the spatial interpolation because the method chosen to fill the missing data may underestimate the value in locations farther from the monitoring stations. with this concern in mind, we test the robustness of our results using pm . data obtained from different interpolation approaches. as indicated in table , the overall average of pm . for the study area between and is roughly µg/m , as most municipalities in norther italian regions belong to industrial and agricultural intensive locations. the average mortality between and for the period of interest (january -april ) was deaths, while it grew to in . that results in an average excess death of , with standard deviation four times as larger (see table ). estimation results from the negative binomial models are summarised in table for the four different specifications of the model ( -no geographical effects; -regional fixed effects; -lls random effects; -regional fixed effects and lls random effects). in the lower part of the table, the estimated overdispersion parameter, the akaike information criterion (aic), and the moran's test for the null hypothesis of absence of spatial autocorrelation in the residuals (moran ) are reported. the four specifications provide consistent results in terms of the direction and significance of pm . coefficients. the overall effect of pm . on covid- -related excess mortality is positive and statistically significant in all models. the estimated incidence rate ratios, reported in table with their confidence interval, for model , , and are . %, . %, . %, and . %, respectively. in model , the regional fixed effects coefficients are statistically significant. they indicate that other things being equal, the number of deaths in municipalities in lombardy and emilia romagna has been systematically higher compared to base category and in municipalities in veneto it has been systematically lower. the significance of the coefficient for emilia romagna, however, drops after including the random effects in the model. since the first three models are nested into model it is also possible to compare the models in terms of aic. model performs substantially better than the other three. in general, the inclusion of re in models and leads to a decrease in the value of the aic. in models and the residuals appear spatially autocorrelated, as the null hypothesis of no spatial autocorrelation is rejected in both cases (p < . ). the introduction of the lls random effects appears to solve the issue of autocorrelation. based on the estimates of model , we compute the expected value of excess deaths conditional on covariates (taken at the average level) in the average city for varying levels of pm . and show how the expected number of deaths by region varies at different concentration levels in fig. . notably, emilia-romagna and liguria are the regions in which a a reduction of average fine pm from the highest level to the lowest would have benefited the most. for robustness check of the pm . metric used in our study, we explored the influence that other alternate pm . metrics may have on the direction and magnitude of the observed associations. figure depicts the point estimates and the % confidence interval for table estimation result of main regressions, dependent variable: excess deaths during the period january -april , municipalities in northern italy ***p < . , **p < . , *p < . model ( ) model ( ) model ( ) model ( we find that while data from satellite elaborations (modis and dimaq ), and monitoring stations' interpolation eea pm . models result in irrs trending in the same direction, the point estimates for irrs are lower than our primary analysis which was based on a combination of ground monitoring and kriging. the lower irr point estimates are unsurprising because the underlying data for the alternate pm . metrics do not have the same temporal coverage as the ground-level monitoring data ( ) ( ) ( ) ( ) ( ) . this lack of temporal coverage contributes to non-differential exposure misclassification, which, in turn, would lead to suppressing effect estimates towards the null. despite this, it is encouraging to find that regardless of the pm . metric used, the direction of the observed associations remains, and so does statistical significance. as previously anticipated, we re-estimate model ( ) using different specifications of the kriging interpolator. in particular, we first relax the mean-stationarity assumption of ordinary kriging by modelling the mean function of the process through both a linear and a quadratic trend in latitude and longitude. next, we replace the simple exponential function with a spherical model and a more flexible matérn kernel with the characteristic parameter set at / (to preserve mean-square differentiability). all these specifications still assume covariance stationarity. figure and table in the appendix report the estimated incidence rate ratios (irr) regression coefficients for the pm variable in model ( ) under these multiple setups: both point estimates and % confidence intervals indicate that there are no substantive differences between using different trend or covariance models, indicating that our result is robust to alternative specification of the interpolation method. in each of the four specifications presented, the coefficient related to pm . is always of the hypothesized direction and statistically different from zero. precisely and consistently with previous results for the original sars-coronavirus during the outbreak (cui et al. ) , an increase in air pollution exposure is associated with increased mortality for covid- . the first panel in fig. , as well as table , suggests that, when using interpolated data from ispra monitoring stations, the increase in mortality rate due to a one-unit increase in pm . concentration varies between % (model -highest rate) and nearly % (model -lowest rate). the % confidence interval for the point estimate in model lies between roughly % and %. our findings fall in line with both wu et al. ( ) and cole et al. ( ) papers. specifically, both papers find a positive ecological relationship between pm . and covid- mortality. in relation to a µg/m increase in pm . , wu et al. find % change in covid- mortality, cole et al. find the same increase associated to additional covid- deaths (almost % if compared to their sample mean), and our paper finds % increase in covid- related excess mortality. despite this similarity in results, the two key differences between our study and the others relate to the exposure assessment method and the outcome assessment method. in our study we use a spatial interpolation method (kriging) from ground-level monitoring data, whereas these other two studies utilize pm . gridded surfaces such as chemical transport modelling in the case of cole et al. and a hybrid approach using chemical transport, aerosol optical depth and land use regression modelling in the case of wu et al. with respect to covid- mortality data, wu et al. use county-level data from the johns hopkins university, center for systems science and engineering coronavirus resource center, which is comprised of covid- deaths tabulated by the us centers for disease control and prevention and state health departments. in cole et al., researchers obtained covid- deaths by residential address and aggregated these to the municipality level. the obvious difference between their study and ours is that we used a surrogate excess mortality measure due to the issues of reliability for covid- death data, as we have already discussed. the other relevant difference between our study and the wu et al. and cole et al. studies is that we subsample the total cohort of italian municipalities to only regions with a very high mortality rate, which are also the regions most affected by the air quality problems. on the other hand, when satellite data are used, our estimate yields lower incidence ratios. although ground-level concentration metrics come with fewer measurement errors, satellite data proves nevertheless useful in corroborating both the direction and the significance of the effect of interest. this redundancy is particularly relevant in light of the relatively few stations capable of detecting the finest particulate. with reference to model ( ) and the remaining covariates, we observe no effect related to population density or income or the extent of industrial areas in the municipality. likewise, there is no evidence suggesting significant links between the share of non-eu residents, the female to male ratio (which disappears after we incorporate the random effects), and the level of education (proxy by the percentage of university students) on the dependent variable of interest. on the other hand, our results suggest a negative association between temperatures and mortality due to covid- . finally, as expected, we find that municipalities with higher shares of the population aged or more have been most affected. the distance from the closest airport, a measure of relational connectedness that also proxy for the exposure to the contagion process, deserves a last comment. we find that municipalities closer to an airport experience a higher number of deaths in excess. we speculate that the result could be related to a higher likelihood for these municipalities to become clusters of contagion in the initial phase of the pandemic, but a causal link cannot be inferred based on our result ad the topic needs more research to be addressed adequately. we conclude our analysis by checking the consistency of our results to different choices of the dependent variable. existing evidence (dominici et al. ; pascal et al. ; samet et al. ; yin et al. ) associates fine pm to severe cardiovascular and respiratory diseases and mortality. in european cities, in particular, an estimated increase in the number of daily deaths of . % is associated with an increase of µg/m of pm (katsouyanni et al. ) . this evidence suggests that long term pm exposure may have had an overall effect on deaths even before the outbreak in the sample municipalities, making it more difficult to isolate the real effect on covid- deaths. we thus run model ( ) using the total number of deaths in the same observation period of as the dependent variable to understand whether the effect of fine pm . on mortality has been more severe during the pandemic. we find no evidence of an effect of pm . on total deaths for in the sampled municipalities, suggesting that the effect of pm exposure on the mortality rate is closely connected to the novel coronavirus outbreak (see table in the appendix). however, since the dependent variable in this "placebo" regression cannot be directly compared to the excess mortality, we repeat the test using total mortality for the year . although the latter includes both covid- related and unrelated deaths, these two variables represent data generating processes of the same nature. as expected, both the regression coefficients and irrs calculated regressing total deaths in suggest a positive and statistically significant effect of exposure to fine particulate on mortality, even though its magnitude is greatly reduced if compared to the estimates in tables and (see table in the appendix). presumably, the effect of pm . concentration on covid- related mortality becomes muted by the noise introduced when accounting for other causes of death. this would also explain the non-significant pm . coefficient in the first "placebo" regression. italy is among the countries most severely affected by the new coronavirus, with more than thousand confirmed cases and more than thousand deaths as of the end of may. yet, the spatial distribution of confirmed cases and deaths suggest that the effects of the viral infection spreading largely vary across the regions of the country but also within regions. in this work, we examined the role of ambient pm . in explaining the spatial variation in deaths that occurred throughout the most extreme time period of the epidemic. the results in the paper, that suggest a positive relationship between pm . concentration and covid- related excess mortality, are robust to different specifications pm . and estimation strategies, even after controlling for additional confounder factors. coherently with previous findings in the literature, we highlight a strong positive correlation between viral respiratory infection incidence and ambient pm . concentrations and the increase in susceptibility to covid- mortality caused by long term exposure to pm . , consistent with evidence for the original sars-coronavirus during the outbreak. in fact, fine pm is already known to affect cardiovascular and respiratory morbidity and mortality. however, we are aware that the phenomenon and the cause and effect relationships are very complex and that our work can only address part of the problem. the cross-sectional nature of the dataset and the use of geographically aggregated information in the epidemiological model does not allow concluding a causal effect exists. in our opinion, the robust evidence in the paper shows that the relationship between pm . and covid- related excess deaths goes far beyond a simple geographical correlation, and further research is needed to explore the causal effect more in depth, when reliable time series data are available. in fact, our paper does not deal with the spread of contagion and the dynamics linked to it, also because, as we underlined, such analysis would require time-series data, a different econometric methodology, and the identification of the exogenous coronavirus insurgence in northern italy. to the latter purpose, the spread of the pandemic incorporates two different dynamics: ( ) on the one hand, the dynamics of the spread of the contagion requires further information to be investigated such as its genesis, the type of virus, and setting of the first outbreak; ( ) the effects of the lockdown changed (or partially blocked) the contagion in an asymmetric way. in addition to this, of course, there are other elements that should be investigated, such as additional variables about health data, mobility, and so forth. our results reinforce the need to adopt environmental policies that would not only reduce the impact of pollution on the health of citizens and workers but would contribute to smooth the negative effects of a (future) pandemics, avoiding collapses of health systems. indeed, recent studies show that in addition to chronic lung inflammation, environmental air pollutant concentrations can exacerbate the effects of increasingly frequent one-shot systemic shocks, which in turn are also caused by environmental factors. in this regard, sustainable and decarbonization policies such as the green new deal, conceived as long-term policies, should be accelerated. td-i (b) for the updated sample (column ) and td-i (c) for the latter release. it turned out that the number of istat excess deaths increased by % with this revision (compare column and column ). it further increased by % on april (compare column and column ) due to corrections, enlargement and new cases. d-pc stands for number of deaths with or from covid- reported by the national department protezione civile italiana (italian civil protection, which is a department of the presidency of the italian council of ministers) over january -march , (column ) and january -april , in column . the official d-pc data are available only at the regional level and they are officially released by the health departments of the regional administrations. in the period march and april , pc registered a % increase in covid- deaths. istat and pc department therefore collect data independently. the key difference between ed-i and d-pc lies in how pc recognizes fatalities as covid- related: only patients known to have tested positive to sars-cov- got registered under this nomenclature. on the other hand, ed-i is just a mathematical construct that takes into account all deaths in a given municipality, regardless the cause. due to difficulties in providing timely screenings and accurate testing during the peak of the pandemics, it is likely that official figures from pc might have been underestimated between early january and mid-april. this is particularly true for the most affected areas. consequently, the discrepancies between ed-i and d-pc can be either moderate or strong, as we observe for lombardy (where ed-i is roughly % higher than d-pc) and trentino alto-adige. last, the unlikely event of ed-i lying below d-pc is due the fact that istat initially collected figures for only a subsample of municipalities (i.e. those recording a percentage of excess deaths in greater than %) leaving out quite a lot of statistical units (for example, friuli-venezia giulia, which had initially the biggest discrepancy (ratio , ) reported data for a very small number of municipalities to istat). however, with some delay, istat has been upgrading the sample, such that in june it covered about all municipalities. in our paper, we use the updated data ed-i j- a (column ), disaggregated at the municipality level. column ( ) shows that the ratio of total ed-i and d-pc is about . , with mean ratio equal to . and standard deviation equal to . . underestimation of cases in the initial sampling base concerned all northern italian regions except emilia romagna, lombardy and veneto (the biggest regions in terms of population and covid- cases). the ratio of total ed-i (b) and d-pc (b) in column ( ) is about . , with a mean ratio equal to . and standard deviation . . one month later, the total ratio decreases to . column ( )-with mean ratio . and standard deviation . . in the latter case, a slight underestimation of the registered cases concerns only fvg and valle d'aosta. in any case, since lombardy has the biggest discrepancy on april (ratio , ) we also run our regressions excluding lombardy (i.e. taking all lombardy municipalities off) and the results are robust, reporting a significant higher-than- relative risk ratio for the exposure variable. the latter results are available upon request. from marshall's to the italian "industrial districts". a brief critical reconstruction understanding the heterogeneity of adverse covid- outcomes : the role of poor quality of air and lockdown decisions trends and variability of atmospheric pm . and pm - . concentration in the po valley, italy applied spatial data analysis with r burden of cause-specific mortality associated with pm . air pollution in the united states air pollution, racial disparities, and covid- mortality associations between long-term pm . and ozone exposure and mortality in the canadian census health and environment cohort (canchec), by spatial synoptic classification zone air pollution exposure, cause-specific deaths and hospitalizations in a highly polluted italian region air pollution and respiratory viral infection covid- in italy: an analysis of death registry data covid- in italy air pollution exposure and covid- can atmospheric pollution be considered a co-factor in extremely high level of sars-cov- lethality in northern italy? air pollution and case fatality of sars in the people's republic of china: an ecologic study airborne particulate matter and mortality: timescale effects in four us cities absorbance and pmcoarse concentrations between and within european study areas and the relationship with no -results of the escape project air quality in europe fog occurrence and chemical composition in the po valley over the last twenty years the role of economic structural factors in determining pandemic mortality rates: evidence from the covid- outbreak in france european air quality maps for characteristics of covid- patients dying in italy report based on available data on epidermal growth factor receptor (egfr)-mapknuclear factor(nf)-κb-il : a possible mechanism of particulate matter(pm) . -induced lung toxicity confounding and effect modification in the short-term effects of ambient particles on total mortality: results from european cities within the aphea project sources for pm air pollution in the po plain, italy: ii. probabilistic uncertainty characterization and sensitivity analysis of secondary and primary sources dia.it/pages /quali tà-dell'aria-duran te-l'emerg enza-covid - ,-l'anali si-di-arpa-lomba rdia-.aspx effects of temperature variation and humidity on the death of covid- in wuhan fine particulate matter and cardiovascular disease: comparison of assessment methods for long-term exposure notes on continuous stochastic phenomena the social and economic factors underlying the incidence of covid- cases and deaths in us counties covid- deaths in lombardy, italy: data in context assessing nitrogen dioxide (no ) levels as a contributing factor to coronavirus (covid- ) fatality short-term impacts of particulate matter (pm , pm - . , pm . ) on mortality in nine french cities pm . source apportionment in lombardy (italy): comparison of receptor and chemistry-transport modelling results long-term concentrations of fine particulate matter and impact on human health in fine particulate air pollution and mortality in u.s. cities, - generation of avian influenza virus (aiv) contaminated fecal fine particulate matter (pm . ): genome and infectivity detection and calculation of immission evaluation of the potential relationship between particulate matter (pm) pollution and covid- infection spread in italy. sima position paper data integration model for air quality: a hierarchical approach to the global estimation of exposures to ambient air pollution covid- infection: the perspectives on immune responses global estimates of fine particulate matter using a combined geophysical-statistical method with information from satellites exposure to air pollution and covid- mortality in the united states long-term fine particulate matter exposure and nonaccidental and cause-specific mortality in a large national cohort of chinese men long-term exposure to fine particulate matter and natural-cause and cause-specific mortality in japan acknowledgements open access funding provided by università cattolica del sacro cuore within the see table . see table . see table . data comparison from different official sources. table reports the excess deaths data as reported by the italian bureau of statistics (istat) along with the official covid statistics as indicated by protezione civile italiana (pc). ed-i stands for excess deaths reported by istat, calculated as the sum of deaths (from all causes) between january and march (column ) or april . (column ) with respect to the average value in - (same months). the difference between column and column involves the sampling base. in fact, as long as istat was upgrading the deaths data, it was both enlarging the sample of municipalities and correcting the past figures; see column to : td-i (a) stands for total deaths reported by istat as the sum of total deaths from january to march , in the initial sample (column ), table estimation results for the placebo regression, dependent variable: total number of deaths during the period jan -april , municipalities in northern italy ***p < . , **p < . , *p < . model ( ) model ( ) model ( ) model ( ) estimate publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. key: cord- -nfl z c authors: slavova, svetla; larochelle, marc r.; root, elisabeth; feaster, daniel j.; villani, jennifer; knott, charles e.; talbert, jeffrey; mack, aimee; crane, dushka; bernson, dana; booth, austin; walsh, sharon l. title: operationalizing and selecting outcome measures for the healing communities study date: - - journal: drug alcohol depend doi: . /j.drugalcdep. . sha: doc_id: cord_uid: nfl z c background: the helping to end addiction long-term (healing) communities study (hcs) is a multisite, parallel-group, cluster randomized wait-list controlled trial evaluating the impact of the communities that heal intervention to reduce opioid overdose deaths and associated adverse outcomes. this paper presents the approach used to define and align administrative data across the four research sites to measure key study outcomes. methods: priority was given to using administrative data and established data collection infrastructure to ensure reliable, timely, and sustainable measures and to harmonize study outcomes across the hcs sites. results: the research teams established multiple data use agreements and developed technical specifications for more than study measures. the primary outcome, number of opioid overdose deaths, will be measured from death certificate data. three secondary outcome measures will support hypothesis testing for specific evidence-based practices known to decrease opioid overdose deaths: ( ) number of naloxone units distributed in hcs communities; ( ) number of unique hcs residents receiving food and drug administration-approved buprenorphine products for treatment of opioid use disorder; and ( ) number of hcs residents with new incidents of high-risk opioid prescribing. conclusions: the hcs has already made an impact on existing data capacity in the four states. in addition to providing data needed to measure study outcomes, the hcs will provide methodology and tools to facilitate data-driven responses to the opioid epidemic, and establish a central repository for community-level longitudinal data to help researchers and public health practitioners study and understand different aspects of the communities that heal framework. the helping to end addiction long-term (healing) communities study (hcs) is a multisite, parallel-group, cluster randomized wait-list controlled trial evaluating the impact of the communities that heal intervention to reduce opioid overdose deaths and other associated adverse outcomes (walsh et al., in press) . the intervention includes three components: ( ) a community-engaged and data-driven process to assist communities in selecting and implementing evidence-based practices to address opioid misuse and opioid use disorder (oud), and reduce opioid overdose deaths (martinez et al., in press) ; ( ) the opioid reduction continuum of care approach which contains a compendium of evidence-based practices and strategies to expand opioid overdose education and naloxone distribution, medications for opioid use disorder (moud), and safe opioid prescribing (winhusen et al., in press) ; and ( ) community-based health communication campaigns to increase awareness and demand for the evidence-based practices and reduce their stigma (lefebrve et al., in press) . a total of communities across four highly affected states (kentucky, massachusetts, new york, ohio) were recruited to participate in the hcs and randomized to one of two waves in a wait-list, controlled design. the communities were randomized to receive either the intervention (referred to as wave communities) or a waitlist control (referred to as wave communities). the hcs has one primary hypothesis (h ) and three secondary hypotheses (h , h , h ) (walsh et al., in press) . it is hypothesized that during the evaluation period (january , to december , ), wave communities compared with wave communities, will: h : reduce opioid overdose deaths (primary outcome); h : increase naloxone distribution; j o u r n a l p r e -p r o o f h : expand utilization of moud; and h : reduce high-risk opioid prescribing. quality data are needed to measure the study outcomes and assess the impact of the integrated intervention and the specific evidence-based practices. data are also an important component of the intervention because communities can use data on opioid overdose mortality and morbidity supplemented with data on community resources and needs to develop a datadriven action plan to expand the utilization of evidence-based practices. communities also need timely and accurate data for visualization in data dashboards designed to monitor the uptake and success of the selected evidence-based practices and strategies, and respond to emerging challenges and community needs (wu et al., in press) . this article describes the process for using administrative data to develop the hcs outcome measures aligned with the primary and three secondary hypotheses of the study. each research site developed collaborations and partnerships with state agencies and other data owners to understand the regulations and policies governing the use of administrative data for research. an hcs data capture work group was formed and included representatives from the four research study sites, the data coordinating center at the rti international, and the sponsors (the national institute on drug abuse and the substance abuse and mental health services administration [samhsa]). a structured consensus decision-making strategy was used to: a. identify data sources to measure the primary, secondary, and other study outcomes; c. develop data governance strategy and data use agreements; d. develop study measure definitions, technical specifications, programming code, procedures for data quality control, common data model, and schedule for data transfer to the data coordinating center; during development, priority was given to use of existing state-level administrative data sources with regulated and sustained data collections and established infrastructures for quality assurance and control. this is an efficient and cost-effective way to study community-level changes, capitalizing on the federal and state investments for collecting standardized surveillance data, and adopting, when possible, validated surveillance definitions. in addition, using multiple administrative data sources allowed for the construction of measures at the community/population level (i.e., unit of analysis being hcs community) by aggregating individual-level data (e.g., unit of measurement being a community resident or a provider practicing within an hcs community) that best matched hcs outcomes. priority also was given to data sources with timely reporting, preferably with less than a -month lag between the occurrence of events and data availability. timeliness and near-realtime access to data was critical for three reasons: ( ) the community engagement component of the intervention is data-driven and dependent on providing ongoing data feedback to community partners throughout the process (walsh et al., in press); j o u r n a l p r e -p r o o f ( ) it is imperative that the study results are made publicly available quickly because of the magnitude and impact of the opioid crisis on us communities; and ( ) the hcs was designed as a four-year study. this study protocol (pro ) was approved by advarra inc., the healing communities study single institutional review board. the study is registered with this section presents the results from the selection and operationalization of administrative data measures for study hypotheses testing (table ) , as well as study measures for secondary analyses and monitoring the progress in implementing evidence-based practices ( the primary hcs outcome is the number of opioid overdose deaths among residents in hcs communities. the traditional data source for capturing drug overdose deaths are death certificate records (isw , ; hedegaard et al., ; warner et al., ) . suspected drug overdose deaths are considered unnatural deaths and are subject to medicolegal death j o u r n a l p r e -p r o o f investigation before the death is certified by a coroner or a medical examiner (hanzlick, ; hanzlick and combs, ) , and a completed death certificate is filed with the office of vital statistics in the state where the death occurred (nchs, a, b) . selected fields from the death certificate record are then sent to the national center for health statistics where the cause-ofdeath information is coded with one underlying and up to multiple (i.e., supplementary) cause-of-death codes using the international classification of diseases, tenth revision (icd- ) coding system (who, ) . the cdc definition for identifying drug overdose deaths with opioid involvement in icd- -coded death certificate records is commonly accepted by researchers and public health agencies. using icd- -coded death certificate data, drug overdose deaths are identified as deaths with an underlying icd- cause-of-death code x - previous research has identified several methodological challenges for identification of opioid involvement in drug overdose deaths (e.g., lack of routinely performed postmortem toxicology testing, especially for fentanyl and designer opioids; challenges to detection and quantification of new designer opioids; variation in jurisdictional office policy in completion of drug overdose death certificates; differences in the proportion of drug overdose death certificates completed by different jurisdictions that do not list the specific contributing drugs) (buchanich et al., ; ruhm, ; slavova et al., ; slavova et al., ; warner and hedegaard, ; j o u r n a l p r e -p r o o f warner et al., ) . prior to the evaluation period, the research sites are administering surveys among the coroners, medical examiners, and toxicology labs serving both wave and wave communities to collect information related to death investigations of suspected drug overdose deaths (including postmortem toxicology testing, timelines for death certificate completion, and possible covid- -related changes in these processes that could have lasting effects during the hcs evaluation period) in order to understand possible limitations and changes in the completeness and accuracy of the primary outcome measure. each hcs research site will use death certificate records from their state office of vital statistics to identify hcs resident deaths with opioid contribution. one challenge in using death certificate data for the primary study outcome is the lag between the death date and the date when death certificate records are available for analysis (rossen et al., ) . sites have been working with local coroners, medical examiners, and state vital statistics offices to improve the timeliness of data availability across all hcs communities. in , almost all the death certificate records in kentucky, massachusetts, new york, and ohio were available for analysis within months after the overdose death (cdc, ). the following steps describe the hcs operational definition for capturing opioid overdose deaths for testing the primary study hypothesis:  step : all sites will use state death certificate files captured months after the end of the evaluation period to identify the death certificate records for residents of hcs communities with a date of death within the evaluation period, an underlying cause-of- this process will ensure a quality harmonized measure that is captured consistently across the four research sites. number of naloxone units distributed in an hcs community as measured by the sum of the naloxone units ( ) the us surgeon general's advisory on naloxone emphasized that expanding naloxone availability in communities is a key public health response to the opioid crisis (hhs, ) . research has shown that opioid overdose death rates were reduced both in communities that implemented overdose education and naloxone distribution programs (walley et al., ) and in jurisdictions enacting laws allowing direct pharmacist dispensing of naloxone (abouk et al., ) . there are three limitations of this data source: ( ) no information is provided about the number of pharmacies dispensing naloxone prescriptions; ( ) suppression rules preclude reporting of data for geographic areas with fewer than four pharmacies; and ( ) prescriptions are assigned to communities based on the location of the pharmacy rather than the customer's residence. suppression rules impacted three communities in massachusetts; this was resolved by requesting the total for the three communities and dividing it relative to the community populations. the assignment of a pharmacy to a community based on pharmacy address may result in an overcount of naloxone in a community with pharmacies that serve residents of non-hcs communities or an undercount if a pharmacy is just outside an hcs community but serves hcs residents. a limitation of the measure is that it may not capture naloxone distributed in hospitals, correctional facilities, or other venues when the naloxone is purchased with support from private donations, foundations, or locally awarded federal funding. the number of hcs residents receiving buprenorphine products approved by the food there are three fda-approved moud products: buprenorphine, methadone, and naltrexone. multiple randomized controlled trials (krupitsky et al., ; mattick et al., mattick et al., , have demonstrated that moud can reduce cravings and illicit opioid use. observational studies have identified that buprenorphine and methadone are associated with reduced mortality sordo et al., ) . thus, as part of the opioid reduction continuum of care approach, communities are required to expand moud with buprenorphine and/or methadone (winhusen et al., in press) . access to moud is geographically heterogeneous and differs by patient population (haffajee et al., ; pashmineh azar et al., ) . for example, opioid treatment programs providing methadone are less common in rural than urban areas (joudrey et al., ) . criminal justice-involved populations, where there has been a historical j o u r n a l p r e -p r o o f preference toward naltrexone (krawczyk et al., ) , are less likely to receive buprenorphine and methadone. there also is a great deal of variation in billing and documentation of the type of moud, administration modality (e.g., office-based administration as compared with prescription filled at pharmacy by patient), provider type, state policies, and insurance coverage. accurate estimation on the prevalence of oud in hcs communities is important for planning and scaling of the moud uptake. however, estimating the population at need for moud is a challenge for the hcs. the hcs team is working on developing improved estimations for oud prevalence in each hcs community using a capture-recapture statistical methodology previously applied by barocas et al. (barocas et al., ) . five potential sources for measurement of moud were identified: medicaid claims, allpayer claims databases, pdmps, opioid treatment program central registries, and pharmacy dispensed prescriptions (iqvia). the disparate data sources vary in completeness and timeliness. all-payer claims databases are large state databases that typically include medical claims across multiple settings (e.g., hospitalizations, emergency departments visits, outpatient visits), pharmacy claims, and eligibility and provider files. data are collected from both public and private payers and reported directly by insurers to a state repository. all-payer claims databases are structured similarly to medicaid claims data and allow for linking of individuals across claims to identify individuals with oud and their treatments. the key advantage is the inclusion of private insurance, allowing more accurate estimation of prevalence of individuals with diagnosed oud and treatment with moud in a state. all-payer claims have been used previously in oud-related research (burke et al., ; freedman et al., ; lebaron et al., ; saloner et al., ) . seventeen states have all- (grecu et al., ) . opioid treatment programs are the only facilities allowed to deliver methadone for oud but may also offer buprenorphine and naltrexone along with behavioral therapy. they must be certified by samhsa and an independent, samhsa-approved accrediting body to dispense moud (samhsa, ). they also must be licensed by the state in which they operate and must be registered with the drug enforcement administration. the registries are established to prevent patient's simultaneous enrollment in multiple locations (e-cfr, a). number of enrolled patients, aggregated at the hcs community level, as permitted by section § . research, cfr part (e-cfr, a) can be used as a measure for methadone treatment uptake, but central registries were not available in all four research sites. iqvia data capture pharmacy dispensed naltrexone. however, naltrexone is indicated for treatment of oud and for alcohol use disorder. because pharmacy records do not include diagnose-related information for making this distinction, this data source may overestimate the uptake of naltrexone for oud. defined as ≥ mg mme over calendar months; or ( ) incident overlapping opioid and benzodiazepine prescriptions greater than days over calendar months. high opioid dosages, co-prescribing opioids with benzodiazepines or other sedative hypnotics, and receipt of opioid prescriptions from multiple providers or pharmacies are associated with opioid-related harms (bohnert et al., ; cochran et al., ; dunn et al., ; rose et al., ) . characteristics of opioid initiation are also important. for example, initiating opioid treatment with extended-release/long-acting opioids (miller et al., ) is associated with increased risk of overdose, and longer prescription duration is associated with transition to long-term opioid use (shah et al., ) . based on available evidence, in the cdc published guidelines (dowell et al., ) for prescribing opioids for chronic pain. numerous quality measures have been developed to encourage and measure progress toward improving the safety of opioid prescribing. after decades of increases, rates of opioid prescribing peaked and are now declining, although they remain historically high (guy et al., ; schieber et al., ) . developing safe and patient-centered approaches for individuals receiving long-term opioid therapy has been a challenge to address in underlying evidence or guidelines. increasing two constructs with the best available evidence to support decreases in opioid-related harms were targeted with the intention of reducing the number of individuals initiating high-risk opioid prescribing and the likelihood that new opioid prescribing episodes develop into longterm episodes (shah et al., ) . state pdmps were identified as the best available data source for these measures across all four research sites. a limitation of these data is the lack of clinical context-such as diagnostic codes for disease or condition-associated with the prescribed medication. as a result, at the patient level, it is difficult to assess the appropriateness of a high-dose opioid prescribing episode, such as that needed for management of severe pain for patients with cancer or end-of-life care. another limitation of this measure is the lack of automated data sharing among state pdmps on prescriptions filled across state boundaries. a benefit of the pdmp data source is that it is timely and captures dispensed prescriptions for controlled substances paid for by both insurance and cash. medicaid claims and all-payer claims databases are alternative data sources. the main advantage of claims data compared with pdmp data is the clinical context. however, claims data lack information on prescriptions paid by cash or alternative insurance coverage, which is associated with increased risk of opioid-related harms (becker et al., ) . medicaid claims are common across the sites, whereas all-payer claims databases exist in only two of the four states. claims data lag by at least months, making them less useful for timely monitoring of progress. existing measures were identified through a review of the literature, including existing measures from cdc and national quality forum, and national committee for quality alliance, which were subsequently adapted to the constructs identified above. all opioid agonist medications, including tramadol, were included, with the exception of antitussive codeine formulations and buprenorphine formulated for pain. the reasons for their exclusion are a lack of clear guidance for conversion to morphine milligram equivalents and a lack of evidence that buprenorphine, a partial opioid agonist, conveys the same risk as this of full opioid agonists. to maintain consistency across sites, the team developed a standardized list of national drug codes for opioids, benzodiazepines, and moud using the medi-span electronic drug file (med-file) v and the drug inactive data file (wolters kluwer, ) . the standardized study drug list is updated quarterly. the med-file includes product names, dosage forms, strength, the ndc, and generic product identifier (gpi). the gpi is a -digit number that allows identification of drug products by primary and secondary classifications and simplifies identification of similar drug products from different manufacturers or different packaging. because our study requires baseline data on opioid utilization, the inactive date file is used to include drugs that may be currently inactive but were used during the baseline period. all gpis beginning with the classification " "-which identifies any drug product containing an opioid or combination-are included in the opioid list. next, opioid products that are not likely to be used in the outpatient/ambulatory pharmacy setting-such as bulk powder, bulk chemicals, and dosage forms typically used in hospitals or hospice settings (e.g., epidurals, ivs)-are excluded. products classified as cough/cold/allergy combinations, cough medications, j o u r n a l p r e -p r o o f antidiarrheal/probiotic agents, buprenorphine products used for oud and pain, and methadone products used for oud were also excluded. the cdc file that identifies oral mmes (cdc, b) was used to add mmes to each opioid product and to identify products as long-acting or short-acting. to ensure the hcs list includes all current and inactive products, the cdc list was cross-referenced with the list of all gpi products. the benzodiazepine products are identified using the gpi classification " ", which identifies any drug product containing a benzodiazepine or combination. products that are not likely to be used in the outpatient/ambulatory pharmacy setting-such as bulk powder, bulk chemicals, and dosage forms typically used in hospitals or hospice settings-were excluded. the full list of gpis for opioids, benzodiazepines, and buprenorphine are included in the appendix. the success of the intervention relies on the community's ability to assess the complexity and specifics of the local opioid epidemic and identify the best ways to implement and promote evidence-based practices locally. a set of additional measures was developed, to be shared with the intervention communities as counts and/or rates over time and visualized as trends on community-tailored dashboards (wu et al., in press) . these measures monitor the complexity of the opioid-related harms as well as the progress in the three main evidence-based practices from the opioid reduction continuum of care approach (winhusen et al., in press) . a list of selected study measures is provided in table working closely with state stakeholders, the research sites also developed standard operating procedures for data quality assurance and control, and improved data collection (e.g., improved timelines of an existing data sources or development of new administrative data collections). the hcs data coordinating center created a common data model to match the complexity and scale of the clinical trial design and measures and the conditions of the data use agreements. the common data model consisted of ( ) an internal identification number for each hcs measure outcome; ( ) frequency of reporting (i.e., daily, monthly, quarterly, semi-annually, or annually); ( ) display features for dashboards and visualization (i.e., display date, display value, research cite/research community identification number, label); and j o u r n a l p r e -p r o o f ( ) internal usage information (i.e., is estimate, is suppressed [per data use agreement suppression requirements], notes, stratification, and version number). the common data model allows coordinated presentation of data to communities to aid with decision making and monitoring of progress and allows the hcs consortium and trial sponsors to routinely monitor progress. during the first year of the hcs, the data capture work group evaluated more than administrative data sources across the four states for their ability to support study measures in multiple relevant domains. the research site teams established multiple data use agreements with data owners to support the calculation for more than study measures based on administrative data collections, such as death certificates, emergency medical services data, inpatient and emergency department discharge billing records, medicaid claims, syndromic surveillance data, pdmp data, drug enforcement administration data on drug take back collection sites and events, data waivered prescriber data, hiv registry, naloxone distribution and dispensed prescription data. there were many challenges related to state variations in data timeliness and content that needed to be addressed, and compromises were made to achieve harmonization across research sites. the harmonization on medicaid measure specifications required participation from the state partners because individual states have some unique codes or code bundles for capturing specific services. collaborative workgroups with participation from state partners were formed with specific focus on medicaid data, pdmp data, and emergency medical services data. another challenge is the lack of quality validation studies for many of the measures, so the degree of possible misclassification of diagnosis or service codes used in some specifications j o u r n a l p r e -p r o o f is unknown. one example is attempting to identify oud prevalence using diagnosis codes in medical claims or other administrative data sources knowing that oud is often underdiagnosed. massachusetts also is seeking to partner with emergency medical services agencies to improve timeliness of data reporting and completeness of race/ethnicity data. new york developed a cloud-based application to facilitate data aggregation and sharing both for hcs and future research projects. in ohio, the hcs team partnered with the innovateohio platform, which was established by executive order a few weeks prior to the hcs project start date. the hcs has been a highly successful "test case" for how a single technology platform could be leveraged to provide necessary data quickly and efficiently for a large study involving multiple state agencies. the platform facilitated a multi-agency data use agreements, and curates, cleans, and links data sets across multiple ohio state agencies monthly. this allowed the ohio hcs team to sign one data use agreement to cover all project data activities. the hcs will provide methodology and tools to facilitate data-driven responses to the opioid epidemic at the local, state, and national levels. number of opioid overdose deaths among hcs residents during the evaluation period as measured by deaths with an underlying cause-of-death being drug overdose (i.e. an underlying cause-of-death icd- code in the range x -x , x -x , x , y -y ) where opioids, alone or in combination with other drugs (i.e. a multiple cause-of-death icd- code in the range t . -t . , or t . ), were determined to be contributing to the drug overdose death. data source: drug overdose deaths are captured by death certificate records; additional medicolegal death investigation records can be used (per established protocol) to determine opioid involvement when specific drugs contributing to the overdose deaths are not listed on the death certificate. number of naloxone units distributed in an hcs community during the evaluation period as measured by the sum of ( ) the naloxone units distributed to community residents by overdose education and naloxone distribution programs with support from state and federal funding, including dedicated hcs funding, and ( ) the naloxone units dispensed by retail pharmacies located within hcs communities. data source: data are captured from state administrative records and supplemented by study records to include naloxone funded through hcs, as well as iqvia xponent® database. number of hcs residents receiving buprenorphine products approved by the food and drug administration for treatment of opioid use disorder as measured by the number of unique individuals residing in an hcs community who had at least one dispensed prescription for these products during the evaluation period. data source: state prescription drug monitoring program data. number of hcs residents with new incidents of high-risk opioid prescribing during the evaluation period as measured by the number of residents in an hcs community who met at least one of the following four criteria for a new high-risk opioid prescribing episode after a washout period of at least days: ( ) incident opioid prescribing episode greater than days duration (continuous opioid receipt with no more than a day gap); ( ) starting an incident opioid prescribing episode with extended-release or long-acting opioid formulation; ( ) incident high-dose opioid prescribing, defined as ≥ mg morphine equivalent dose over calendar months; or ( ) incident overlapping opioid and benzodiazepine prescriptions greater than days over calendar months. association between state laws facilitating pharmacy distribution of naloxone and risk of fatal overdose the medicaid outcomes distributed research network (modrn) innovative solutions for state medicaid programs to leverage their data, build their analytic capacity, and create evidence-based policy claims database council estimated prevalence of opioid use disorder in massachusetts multiple sources of prescription payment and risky opioid therapy among veterans association between opioid prescribing patterns and opioid overdose-related deaths the effect of incomplete death certificates on estimates of unintentional opioid-related overdose deaths in the united states trends in opioid use disorder and overdose among opioid-naive individuals receiving an opioid prescription in massachusetts from annual surveillance report of drugrelated risks and outcomes-united states opioid overdose. data resources. analyzing prescription data and morphine milligram equivalents (mme) centers for disease control and prevention (cdc), . nchs data quality measures an examination of claims-based predictors of overdose from a large medicaid program kasper controlled substance reporting guide, . kentucky cabinet for health and family services nonfatal opioid overdose standardized surveillance case definition no shortcuts to safer opioid prescribing cdc guideline for prescribing opioids for chronic pain--united states opioid prescriptions for chronic pain and overdose: a cohort study electronic code of federal regulations, a. § . security for records electronic code of federal regulations, b. e-cfr website all-payer claims databases -uses and expanded prospects after gobeille mandatory access prescription drug monitoring programs and prescription drug abuse vital signs: changes in opioid prescribing in the united states characteristics of us counties with high opioid overdose mortality and low capacity to deliver medications for opioid use disorder a perspective on medicolegal death investigation in the united states medical examiner and coroner systems: history and trends drug overdose deaths in the united states consensus recommendations for national and state poisoning surveillance drive times to opioid treatment programs in urban and rural counties in us states only one in twenty justice-referred adults in specialty treatment for opioid use receive methadone or buprenorphine injectable extended-release naltrexone for opioid dependence: a double-blind, placebocontrolled, multicentre randomised trial medication for opioid use disorder after nonfatal opioid overdose and association with mortality: a cohort study opioid epidemic or pain crisis? using the virginia all payer claims database to describe opioid medication prescribing patterns and potential harms for patients with cancer health communication campaigns to drive demand for evidence-based practices and reduce stigma in the healing communities study pharmacy reporting and data submission methadone maintenance therapy versus no opioid replacement therapy for opioid dependence buprenorphine maintenance versus placebo or methadone maintenance for opioid dependence prescription opioid duration of action and the risk of unintentional overdose among patients receiving opioid therapy medical examiners' and coroners' handbook on death registration and fetal death reporting u.s. standard certificate of death new york department of health ohio data submission dispenser guide rise and regional disparities in buprenorphine utilization in the united states prescription drug monitoring program training and technical assistance center df#:~:text=in% % c% doj% began% the% harold% rogers% prescri ption,were% interested% in% establishing% c% implementing% c% and% enhancing% pdmps. accessed on prescription drug monitoring program training and technical assistance center (pdmp ttac), . pdmp policies and practices potentially inappropriate opioid prescribing, overdose, and mortality method to adjust provisional counts of drug overdose deaths for underreporting. division of vital statistics corrected us opioid-involved drug poisoning deaths and mortality rates patterns of buprenorphine-naloxone treatment for opioid use disorder in a multistate population variation in adult outpatient opioid prescription dispensing by age and sex -united states drug and opioid-involved overdose deaths -united states characteristics of initial prescription episodes and likelihood of long-term opioid use -united states methodological complexities in quantifying rates of fatal opioid-related overdose drug overdose deaths: let's get specific mortality risk during and after opioid substitution treatment: systematic review and meta-analysis of cohort studies medications for opioid use disorder substance abuse mental health services administration (samhsa), . certification of opioid treatment programs (otps) us surgeon general's advisory on naloxone and opioid overdose department of health and human services (hhs), . hhs guide for clinicians on the appropriate dosage reduction or discontinuation of long-term opioid analgesics fda identifies harm reported from sudden discontinuation of opioid pain medicines and requires label changes to guide prescribers on gradual opioid overdose rates and implementation of overdose education and nasal naloxone distribution in massachusetts: interrupted time series analysis identifying opioid overdose deaths using vital statistics data state variation in certifying manner of death and drugs involved in drug intoxication deaths evidence-based practices in the healing communities study drug alcohol depend drug data. www.wolterskluwercdi.com. accessed on community dashboards to support datainformed decision making in the healing communities study table . healing communities study primary and secondary outcome measures for hypothesis testing all authors contributed to the development of the hcs measures, the development of the framework for the manuscript, and the editing of the manuscript. s. slavova, j. villani, and s.l.walsh drafted the introduction, s. slavova drafted the methods, m.r. larochelle developed the table, and each author participated in drafting parts of the results or discussion sections. key: cord- - mue eiv authors: bertozzi, giuseppe; maglietta, francesca; baldari, benedetta; besi, livia; torsello, alessandra; di gioia, cira rosaria tiziana; sessa, francesco; aromatario, mariarosaria; cipolloni, luigi title: mistrial or misdiagnosis: the importance of autopsy and histopathological examination in cases of sudden infant bronchiolitis-related death date: - - journal: front pediatr doi: . /fped. . sha: doc_id: cord_uid: mue eiv pediatrics, among all the branches of medicine, is a sector not particularly affected by a high number of claims. nevertheless, the economic value of the compensation is significantly high, for example, in cases of children who suffered multiple disabilities following perinatal lesions with a long life expectancy. in italy, most of the claims for compensation concern surgical pathologies and infections. among these latter, the dominant role is taken by respiratory tract infections. in this context, the purpose of this manuscript is to present a case series of infant deaths in different emergency-related facilities (ambulances, emergency rooms) denounced by relatives. following these complaints, the autopsy was performed, and subsequent histological examinations revealed the presence of typical and pathognomonic histological findings of acute viral bronchiolitis, whose morphological appearance is poorly reported in the literature. the analysis of these cases made it possible to highlight the following conclusions: the main problems in diagnosing sudden death causes, especially in childhood, are the rapidity of death and the scarce correlation between the preexistent diseases and of the cause of death itself. for all these reasons, the autopsy, either clinical or medicolegal, is mandatory in cases of sudden unexpected infant death to manage claim requests because only the histological examinations performed on samples collected during the autopsy can reveal the real cause of death. pediatrics, among the branches of medicine, is not particularly affected by a high number of claims. nevertheless, the economic value of compensation is significantly high, for example, in cases of children who suffered multiple disabilities following perinatal lesions with a long life expectancy. according to carroll and buddenbaum ( ) and moriani et al. ( ) , examining the data collected by an association of several american insurance companies (physician insurers association of america), it was found that only % of the cases resulted in compensation; among these, in cases where no damages had been paid, the average cost per the only defense was $ , , while it results in $ , for paid claims ( ) . the medical diagnoses, most commonly involved in civilian pediatric trials in the united states, were brain damage (average damages $ , ) and meningitis ($ , ) . respiratory problems in newborns account for $ , . in italy, data from the insurance company carige spa, in the period - , excluding neonatology, highlighted how the main pathologies for which legal action was referred to surgery (gastrointestinal and testicular) and infections (more respiratory ones). moreover, the data on litigation have also shown a different stratification of the number of requests for compensation, which were greater in the north and minors in central italy, mostly involving the public health system ( ) . according to the italian study performed during the - period, neonatology has also shown an overlapping geographical stratification, with the greatest interest for the public sector. the claims for damages following death, concerning the neonatal intensive care unit (nicu), mainly concern respiratory diseases ( . % of cases) ( , ) . both in the pediatric population in general, but especially in the neonatological one, the dominant role is taken by the respiratory tract infections. in this context, the purpose of these case series is to demonstrate how the identification of the correct cause of death in the sudden unexpected infant deaths (suids) allowed evaluating the absence of medical liability. particularly, the definition of gold standard methods in similar cases could be considered very important to avoid the compensation in unjustified claim requests. all procedures performed in the study were in accordance with the ethical standards of the institution and with the helsinki declaration and its later amendments or comparable ethical standards. written informed consent was obtained from the first-degree relatives. a -month-old male infant died during emergency medical services (ems) transport to the hospital. when parents had been asked for any modification in their child habits, a mild "rhinitis" for a few days was told. for this reason, they went to their trusted pediatrician days earlier, who suggested saline nasal rinses and a short turn check. the medical examiner documented no relevant external sign to explain death. therefore, the parents sued the pediatrician for both penal and civil liability. during the forensic autopsy, the macroscopic examination was unremarkable except for mild edema affecting both lungs. on the contrary, histological examination showed in both lungs a diffuse transmural inflammation in the bronchiolar wall. other tissue sections showed chronic inflammation, and bronchiolar wall fibrosis primarily restricted to bronchioles (figure ). an ambulance was called for a -month-old female infant, who lived in a nomad camp; her parents referred that suddenly she did not respond to external stimuli. relatives did not refer to any symptoms neither clinical signs in the previous days. during the resuscitation maneuvers in place, the infant died. thus, the prosecutor ordered the autopsy for alleged medical liability; her parents demanded the civil compensation to the local health insurance, thinking that during the ambulance transportation, there was medical responsibility. the macroscopic examination, both at the external corpse and internal organs, only showed severe pulmonary edema. the histology was characterized by lymphocytic infiltration of the bronchioles (figure ). an -month-old female infant was admitted to the emergency room of a pediatric hospital for severe cough and pharyngitis; she died after a few hours. symptoms onset occurred the day before hospitalization. she was a preterm infant ( . weeks, birth weight , g) who suffered from severe respiratory distress at birth (apgar score ′ = ) and needed a long period of hospitalization. after discharge, she showed neurodevelopmental impairment; moreover, a month before death, she suffered from many viral infective pathologies such as influenza and mononucleosis: all pathologies were successfully treated with standard pharmacological therapies. in this case, the judicial authority disposed of the forensic examination, suspecting medical liability to clarify penal and civil aspects: indeed, at the time of death, a claim for damages has been made to the hospital by the family of the patient. the autopsy showed congestion of tracheal and bronchial mucosa. at histological examination, focal edema and diffuse congestion of both lungs, acute emphysema, and peribronchial and intrabronchial wall leukocyte infiltrates were found; the same results involved nearby septal vessels (figure ). the method used in these cases, as in all cases of sudden death, consists of a rigorous and multidisciplinary methodological approach ( , ) : -anamnestic collection and clinical findings: the clinical symptomatology presented by the subject before his death or in close chronological concurrence, clinical history of the case, and previous medical records; -anatomical evidence: a macroscopic examination of all organs, their weight, consistency, and color at the autopsy, such as the appearance of fluids (blood, urine, vitreous humor). -histomorphological examination h&e of the organ samples to study any alteration due to pathological condition; -immunohistochemistry: to evaluate, in particular, the presence and the location of the main white blood cells via antibody anti-leukocyte common antigen (cd ). the adherence to this diagnostic procedure not only allows you to check and evaluate a larger quantity of data but allows a complete evaluation on all fronts, from the study of which can confirm the suspicion and/or an unexpected result but crucial for investigations. in the discussed cases, following both the autoptic and especially the microscopic examination, the cause of death was identified in all investigated cases: a rapidly progressive acute bronchiolitis was ascertained. these findings allowed exonerating doctors from any penal liability. the bronchiolitis was defined in , from a collaboration between the american academy of pediatrics (aap) and the european respiratory society (ers), as "a constellation of clinical symptoms and signs including a viral upper respiratory prodrome followed by increased respiratory effort and wheezing in children < years of age" ( ) . acute viral bronchiolitis (avb) is a lower respiratory tract infective disorder, typically affecting infants < years old ( % of cases). respiratory syncytial virus (rsv) is involved in up to % of cases, followed by rhinovirus (up to %); the remaining cases are related to coronavirus, adenovirus, influenza, and parainfluenza virus, and human metapneumovirus. coinfections are common ( ) . however, the seasonality of bronchiolitis, generally more frequently found during the winter months, coincides with the seasonal pattern of rsv diffusion ( ) . the infection starts in the upper respiratory tract, spreading to the lower airways in a few days. the bronchiolar damage is determined by the direct action of the virus on the epithelium of the same tract; alternatively, it was indirectly immunemediated, and it was characterized by a peribronchial infiltration of white blood cell types, mainly mononuclear cells, with edema of the submucosa and adventitia ( ) . the pathophysiological continuation is caused by a mixture of edema, increased production of mucus, and progressive damage of the epithelium even to necrosis, which determines obstruction of the airflow, entrapment of distal air, atelectasis, and alteration of the ventilation/perfusion. the results are hypoxemia and increased respiratory work, which in turn worsens hypoxemia ( ) . the most important extrapulmonary symptoms involve the brain (apnea, epileptic status) and heart (ventricular tachycardia, ventricular fibrillation, cardiogenic shock, complete heart block, and pericardial tamponade) and are common in children with severe infections ( ) . the most dreadful complications of bronchiolitis are central apnea, a respiratory pause with bradycardia, cyanosis, pallor, and hypotonia that often requires hospitalization ( ) . bronchiolitis represents a disease with high morbidity but low mortality. death from respiratory failure in bronchiolitis is rare and varies from deaths from . : , in the uk to . : , in the us, for children under months, with a relationship that goes hand in hand, reducing itself to the improvement of good intensive practices ( , ( ) ( ) ( ) . in all these cases, in the absence of the clinical-anamnestic data that can guide the clinical diagnosis, the external examination data and the autopsy macroscopic data could point toward a diagnosis of suid. in the case of Özdemir et al. ( ) , in fact, on the totality of the cases of malpractice claims, . % of the children had died and . % were subjected to autopsy. in these cases, the causes of death reported before and after the autopsies were different in %, and the medical staff was found to be responsible for . % of the claims. therefore, the determination of the exact cause of death assumes fundamental importance to ascertain the causal link of any conduct of both health facilities and individual professionals in determining death ( ) ( ) ( ) . this allows not only a measurement of the quality of care provided by promoting public trust for the health system but also as a measure of clinical governance; moreover, it is possible to better manage the medicolegal disputes as a guarantee of ascertaining the truth. indeed, in italy, the data on the frequency of adverse events (aes), preventable adverse events (paes), and negligent adverse events (naes) are available; nevertheless, data about malpractice claims are not available both under the penal and civil points of view. furthermore, the epidemiological purposes cannot be forgotten, considering that it is the only reliable method of data collection. indeed, a complete methodological approach, integrating clinical data, autopsy, and histological findings could be considered the best way to solve similar cases. in fact, in the reported case studies, histopathologic diagnostics identified pathognomonic signs of acute bronchiolitis characterized by edema, congestion, leukocytic infiltration in the bronchiolar wall, leukocytes in the peribronchial interstitial pulmonary space, allowing the identification of the exact cause of death. therefore, these pieces of evidence have allowed excluding the medical responsibility in the reported cases, demonstrating that there are events not related to the supplied health care. the analysis of the presented cases shows that the autopsy is mandatory in suid occurrence, in which the absence of anamnestic data and/or acute clinical signs does not allow to identify the cause of death. hypothesizing medical negligence in each case, the autopsy was performed following the judicial appointment after the relative's complaint. the subsequent histological examinations revealed the presence of typical and pathognomonic histological findings of avb, whose morphological appearance is poorly described in the literature. only the postmortem examinations have allowed excluding medical liability and therefore the compensation for damage. in light of these findings, it could be considered essential an accurate evaluation of similar cases, collecting all data to avoid compensation in unjustified claims made against the hospital. in this way, it is possible to contain the hospital costs related to this kind of accident. for all these reasons, the autopsy combined with the subsequent examination represents a gold standard method to identify the absence of the hospital's responsibility in suid cases. all datasets generated for this study are included in the article/supplementary material. written informed consent was obtained from the first-degree relatives for the publication of this case report. gb, fm, ma, and lc contributed to the conception of the study and wrote the manuscript. fs, bb, lb, at, and cd contributed significantly to literature review and manuscript preparation. gb, fm, fs, bb, lb, at, cd, ma, and lc helped perform the analysis with constructive discussions and approved the final version. malpractice claims involving pediatricians: epidemiology and etiology suicide by sharp instruments: a case of harakiri medical diagnoses commonly associated with pediatric malpractice lawsuits in the united states pediatric claims in italy during a -years survey neonatal malpractice claims in italy: how big is the problem and which are the causes? post-mortem magnetic resonance foetal imaging: a study of morphological correlation with conventional autopsy and histopathological findings a multidisciplinary approach is mandatory to solve complex crimes: a case report italian mafia: a focus on apulia mafia with a literature review acute bronchiolitis in infants, a review prospective multicenter study of viral etiology and hospital length of stay in children with severe bronchiolitis effect of prematurity on respiratory syncytial virus hospital resource use and outcomes extrapulmonary manifestations of severe respiratory syncytial virus infection -a systematic review frequency of apnea and respiratory viruses in infants with bronchiolitis clinical predictors of radiographic abnormalities among infants with bronchiolitis in a paediatric emergency department variation in the management of infants hospitalized for bronchiolitis persists after the american academy of pediatrics bronchiolitis guidelines risk factors for bronchiolitis-associated deaths among infants in the united states medical malpractice claims involving children medical records quality as prevention tool for healthcare-associated infections (hais) related litigation: a case series personalised healthcare: the dima clinical model healthcare-associated infections: not only a clinical burden, but a forensic point of view key: cord- - lkcdisr authors: asirvatham, edwin sam; sarman, charishma jones; saravanamurthy, sakthivel p.; mahalingam, periasamy; maduraipandian, swarna; lakshmanan, jeyaseelan title: who is dying from covid- and when? an analysis of fatalities in tamil nadu, india date: - - journal: clin epidemiol glob health doi: . /j.cegh. . . sha: doc_id: cord_uid: lkcdisr background: as the number of covid- cases continues to rise, public health efforts must focus on preventing avoidable fatalities. understanding the demographic and clinical characteristics of deceased covid- patients; and estimation of time-interval between symptom onset, hospital admission and death could inform public health interventions focusing on preventing mortality due to covid- . methods: we obtained covid- death summaries from the official dashboard of the government of tamil nadu, between th may and july , . of the deaths, we included cases for analysis. results: the mean age of the deceased was . years (sd: . ). the crude death rate was . per , population; the age-specific death rate was . among above years and . among less than years, and it was higher among men ( . vs . per , population). around % reported having any one or more comorbidities; diabetes ( %), hypertension ( . %) and cad ( . %) were the commonly reported comorbidities. the median time interval between symptom onset and hospital admission was days (iqr: , ); admission and death was days (iqr: , ) with a significant difference between the type of admitting hospital. one-fourth of ( . %) deaths occurred within a day of hospital admission. conclusion: elderly, male, people living in densely populated areas and people with underlying comorbidities die disproportionately due to covid- . while shorter time-interval between symptom onset and admission is essential, the relatively short time interval between admission and death is a concern and the possible reasons must be evaluated and addressed to reduce avoidable mortality. tamil nadu, india as of th july , the sars-cov- has infected around , individuals with , deaths in india ( ). tamil nadu, a south indian state with a population of around million ( ) , reported around , cases and , deaths which was % of total confirmed cases and . % of total deaths in india ( ). the rapid spread of the disease has undoubtedly become a burden to health systems in several countries, as a significant proportion of elderly, immunosuppressed and those with underlying metabolic, cardiovascular or respiratory diseases continue to develop severe forms of the covid- and are at an increased risk for adverse outcomes ( ) . at the same time, evidence is emerging to caution that young and adult general population are also at considerable risk for critical illness and adverse outcome ( ) . as the number of cases continues to increase, public health efforts must focus on preventing avoidable fatalities. when a health system is burdened beyond its capacity and the morale of health care workers are affected, the standard of care would be compromised, leading to negative health outcomes. current therapeutic strategies to deal with the covid- infection are only supportive, and prevention efforts aimed at reducing transmission in the community is considered as the most effective method ( ) . however, the fatality due to covid- could be reduced, if there is early and accurate diagnosis, identification of clinical features of severe risks, prediction of disease progression and appropriate clinical intervention. further, early seeking of medical care by people with exposure and symptoms, especially the most vulnerable, could substantially reduce the spread of infection, severity and fatality due to this disease and produce better clinical outcome ( ) . currently, the covid- infection has its presence across the globe, generating new information, fresh knowledge and evidence continuously. however, there are still many unknowns and ambiguity about the demographic and clinical characteristics of deceased covid- patients; the different time intervals between the time of infection and outcomedeath or recovery ( ) . the currently available literature indicated, information varying contextually, across regions and countries, emphasizing the need for generating evidence for a specific geography, population, and context. this study aims to understand the demographic and clinical characteristics of deceased covid- patients; and estimate the time-interval between symptom onset, hospital admission and death, which could inform public health interventions focusing on preventing mortality due to covid- . . we obtained covid- death summaries from the official dashboard of the government of tamil nadu (https://stopcorona.tn.gov.in/), a south indian state. each death summary consisted of information such as district, age, gender, type of admitting hospital, presence of comorbidities, presenting symptoms, number of days with symptoms, date of hospital admission and date of death. we collected information from th may to th july . a total of , deaths were reported during the period. we excluded brought dead cases ( ) and finalised , cases of deaths for analysis. for the analysis of comorbidities, we included only the cases that reported the presence or absence with details of comorbidities. for the analysis of presenting symptoms, we included only the death summaries that indicated the presence of symptoms with details, as the absence of symptoms j o u r n a l p r e -p r o o f on admission is not provided in death summaries. for the estimation of time intervals between symptom onset and hospital admission; admission and death, we excluded the cases referred from other hospitals due to the potential influence of the time intervals; and cases who did not have all the information. we summarised the categorical variables as frequency and percentages; and continuous variables as mean, standard deviation (sd), median, and interquartile range (iqr) as appropriate. we analysed the continuous variables using the independent t-tests or mann-whitney test/ kruskal-wallis h test. the proportions for categorical variables were analysed using chi-squared test (χ ) test. the analyses were performed using spss version . (ibm corp., armonk, ny, usa). among cases ( , ) that reported the presence or absence with details of comorbidities, . % reported any one or more comorbidities at the time of hospital admission (table- ). diabetes was found to be the most common comorbidity associated with % of the deceased; hypertension and cad were present among . % and . % of the deceased respectively. the coexistence of diabetes and hypertension; diabetes, hypertension and cad were found among . % and . % of the individuals respectively. as expected, the study found a significantly higher presence of comorbidities among the elderly compared to the younger age groups (p< . ), in terms of the presence of any one or more comorbidities, diabetes, hypertension, chronic obstructive pulmonary disease (copd), coronary artery disease (cad), the coexistence of diabetes and hypertension, and the coexistence of diabetes, hypertension and cad. as age increased, the presence of comorbidities seems to increase significantly. the presence of any one or more comorbidities (p= . ) and comorbidities categorised as others (p< . ) were found to be higher among women. private hospitals reported to have treated a significantly higher proportion of deceased patients with comorbidities such as diabetes (p= . ), hypertension (p< . ), cad (p< . ), asthma (p< . ), the coexistence of diabetes and hypertension (p< . ), and j o u r n a l p r e -p r o o f the coexistence of diabetes, hypertension and cad (p= . ), whereas public hospitals reported to have treated higher proportion of patients with ckd (p< . ). around % of deceased reported having other comorbidities such as hypothyroidism, dementia, encephalopathy, cerebrovascular diseases (cvs), hepatitis etc., however, the presence of these as independent comorbidity was found to be very low ( . %, ). around deaths summaries consisted of the details of presenting symptoms on admission (table- ) . fever was the most common presenting symptom ( . %) reported by . % of the deceased men and . % of women. breathing difficulty was reported by . % of the patients ( . % of men, % of women); around % had cough ( . % of men and . % of women) and . % had fever, cough and breathing difficulty together ( . % of men and . % of the women), however, the differences are not statistically significant. diarrhoea and generalised weakness/myalgia were reported by . %, and . % of the patients respectively, with women reporting significantly higher than men (p= . ; p= . ). fever was reported to be significantly higher among the older population (p= . ). the median time interval between onset of symptoms and hospital admission was days (iqr: , ) without significant difference among gender and age groups (table- ). the patients who were admitted in private hospitals had a median of days of symptoms as compared to public hospital patients with days of symptoms (p< . ). the median time interval between hospital admission and death was (iqr: , ) and there are no significant differences among gender and age groups. however, it was significantly higher, days (iqr: , ) in private hospitals compared to days (iqr: , ) in public hospitals (p< . ). around one fourth of ( . %) of the reported deaths occurred within a day of hospital admission (public . %; private . %), . % between to days (public . %; private . %), . % between to days (public . %; private . %) and . % occurred after days of admission (public . %; private . %) (p< . ). it is well documented that the covid- pandemic takes different shapes and forms with varying mortality levels across geographic regions and countries. though there are several studies from other countries that explained the characteristics of covid- deaths, there is a dearth of peer-reviewed and published literature from india. our study analysed the individual death summaries, and described the demographic and clinical characteristics of deceased covid- patients; and estimated the time intervals between symptoms onset to hospital admission and death, which are critical for developing context and geographicspecific public health interventions focusing on reducing the mortality. our study findings indicated a disproportionate death rate among the categories of age, gender and geography. it is well demonstrated that age is the most significant risk factor for death due to covid- and our study confirms the existing evidence. the increasing death rate with age is expected and it could be due to the higher prevalence of comorbidities, the reduced and less responsive innate and adaptive immune system among the elderly ( , ). the study reported a higher proportion of deaths ( %) among men, though the proportion of j o u r n a l p r e -p r o o f total confirmed male cases was only % in the state ( ) ; and death per , male population was . as compared to . deaths per , female population. the less mortality among women has been reported in many studies which could be due to the protection of x chromosome and sex hormones, which play an important role in providing innate and adaptive immunity ( ) . the higher mortality among men could be due to the behavioral risk factors such as smoking, and alcohol consumption, which are relatively higher among men in india ( ) . around % of deaths occurred in the capital district/city that recorded % of the total cases and just . % of the population of the state. it is also the smallest and densest of all the districts in the state ( ). the possible reasons could be, lack of timely access to healthcare facilities, delayed seeking of care, overwhelming of health system due to sudden surge of cases due to larger and exponential spread of the virus in the city. the disease severity, increased admission rate into the intensive care unit (icu), and increased risk of mortality of covid- are strongly associated with comorbidities such as diabetes, hypertension, obesity, cardiovascular disease, and respiratory system diseases and our study results confirm the previous findings ( , , ) . a study showed a hazard ratio of . among patients with two or more comorbidities compared to . among patients with one comorbidity( ). the centers for disease control and prevention reported times higher deaths among patients with reported underlying conditions compared with those without reported underlying conditions ( . % versus . %) ( ) . in our study, the prevalence of any one or more comorbidities among the deceased was found to be around %, and a significant proportion of deceased having other comorbidities such as diabetes, hypertension, and cad respectively, with a strong association with age. studies in china reported around % of deaths with any one comorbidity ( , ) , south korea and brazil reported % and j o u r n a l p r e -p r o o f . %; and these studies have reported hypertension, cad, and diabetes as the main comorbidities among deaths ( , ) . in our study, fever and breathing difficulty each was reported among / th of the deceased, cough among half of the patients and around / rd had fever, breathing difficulty and cough together, which are in line with the existing literature that fever, dry cough, shortness of breath and fatigue were the common symptoms on admission among the deceased patients ( , ) a meta-analysis of covid- patients, showed fever ( . %) as the most common symptom, followed by dry cough ( %) and fatigue ( ) . another study that reviewed , cases across the world showed % of the cases with fever and % with cough ( ) . though the study found an association between fever as a presenting symptom and age of the deceased, other symptoms and multiple concurrent symptoms did not indicate any association with age, gender and admitting hospital. the median time interval between symptom onset and hospital admission was found to be days, which is within the range mentioned in studies from china ( , days), singapore ( days), italy ( days) that indicated a range of to days ( , ( ) ( ) ( ) . the duration might change during the different phases of the epidemic due to rapid changes in the level of knowledge and awareness, stigma and discrimination, fear of the disease, programmatic interventions, health-seeking behaviors, and access to health care services. as the epidemic progresses and there is continuous scale-up of public health and social interventions, the duration is expected to decrease, which could potentially limit the infection to others. according to our study, the median time interval between hospital admission and death was days with a significant difference between patients admitted in private and public hospitals. other countries reported a slightly higher, but a wide range of to days of time interval j o u r n a l p r e -p r o o f between hospital admission and death ( , ( ) ( ) ( ) . a study that reviewed the length of stay across the world reported a shorter length of stay for those who died in hospital compared to those were discharged alive, with medians between and days compared to and days, respectively ( ) . the shorter period of hospital stay with negative outcome could be due to the delayed seeking of care due to economic reasons and lack of awareness about the disease, shortage of icu facilities and ventilators, and overwhelming of hospitals that affect the capacity to deliver services effectively. the knowledge and awareness level of health workers, especially about the rapid progression to severe illness, treatment protocols, and the availability of effective drugs could also alter the duration of hospital stay. the relatively longer time interval between hospital admission and death in private hospitals, despite the higher proportion of deceased patients with comorbidities and presence of multiple concurrent symptoms could be due to the availability advanced technology and facilities to provide advanced life-supporting critical care, the affordability of those seeking health care and less burden of patients especially for covid- management as compared to public hospitals studies reported an average incubation period of to days, which is the time interval between the exposure/infection and onset of symptoms ( , , ) . considering an average of days between the exposure/infection and onset of symptoms, the estimated time interval between exposure/infection and death is days in the state, whereas the time interval between symptom onset and death is just days. two studies have reported a higher average time interval of . and . days between the first recorded symptoms and death ( , ) , as compared to our study findings. j o u r n a l p r e -p r o o f the study provides evidence from india, emphasizing that the elderly, male and people living in densely populated areas, and patients with underlying comorbidities die disproportionately due to covid- . the study estimated a time interval of days between exposure to infection and death, with days each for symptoms onset to hospital admission; and admission to death. none of the potential factors significantly alter the time interval between the onset of symptoms and hospital admission and death, except the type of treating hospital. the shorter time interval between the onset of symptoms and hospital admission would be critical as early diagnosis, supportive care and treatment could substantially reduce the mortality especially among the elderly and vulnerable population. however, the shorter time interval between admission and death is a concern and the possible reasons must be elucidated and addressed. essentially, as the number of deaths from covid- continues to increase, early diagnosis and timely treatment for moderate and severe cases are of crucial importance to reduce mortality. the analysis is based on data available in the public domain and it is limited to only deaths that were reported in the state, through hospital admission. hospital care can vary from general ward care to intensive care and we do not have disaggregated data for this. the j o u r n a l p r e -p r o o f - , j o u r n a l p r e -p r o o f government of india covid- active cases in tamil nadu chennai: health & family welfare department, government of tamil nadu comorbidity and its impact on patients with covid- in china: a nationwide analysis medical vulnerability of young adults to severe covid- illness&#x ;data from the national health interview survey evaluation and treatment coronavirus (covid- ): in: statpearls duration of symptom onset to hospital admission and admission to discharge or death in sars in mainland china: a descriptive study rapid progression to acute respiratory distress syndrome: review of current understanding of critical illness from covid- infection why does covid- disproportionately affect older people? aging (albany ny) estimates of the severity of coronavirus disease : a model-based analysis. the lancet infectious diseases daily report on public health measures taken for covid- chennai: directorate of public health and preventive medicine, health and family welfare department, government of tamil nadu molecular mechanisms of sex bias differences in covid- mortality national family health survey- (nfhs- ) new delhi: ministry of heallth and family welfare,government of india clinical characteristics of deceased patients with coronavirus disease : retrospective study comorbid chronic diseases are strongly correlated with disease severity among covid- patients: a systematic review and meta-analysis covid- ) -united states clinical features of fatal cases of covid- from wuhan. a retrospective observational study chronic heart diseases as the most prevalent comorbidities among deaths by covid- in brazil korean society of infectious d, korea centers for disease c, prevention. analysis on mortality cases of coronavirus disease in the republic of korea from comorbidity and its impact on patients with covid- the prevalence of symptoms in covid- ): a systematic review and meta-analysis of studies from countries clinical progression of patients with covid- in shanghai investigation of three clusters of covid- in singapore: implications for surveillance and response measures -day mortality in patients hospitalized with covid- during the first wave of the italian epidemic: a prospective cohort study clinical course and outcomes of critically ill patients with sars-cov- pneumonia in wuhan, china: a single-centered, retrospective, observational study clinical characteristics and outcomes of patients undergoing surgeries during the incubation period of covid- infection predictors of mortality for patients with covid- pneumonia caused by sars-cov- : a prospective cohort study covid- length of hospital stay: a systematic review and data synthesis the incubation period of coronavirus disease (covid- ) from publicly reported confirmed cases: estimation and application clinical course and risk factors for mortality of adult inpatients with covid- in wuhan, china: a retrospective cohort study. the lancet the authors declare that they have no conflict of interest. this research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. not required j o u r n a l p r e -p r o o f key: cord- -zouscolw authors: wu, jianhua; mamas, mamas a; mohamed, mohamed o; kwok, chun shing; roebuck, chris; humberstone, ben; denwood, tom; luescher, thomas; de belder, mark a; deanfield, john e; gale, chris p title: place and causes of acute cardiovascular mortality during the covid- pandemic date: - - journal: heart doi: . /heartjnl- - sha: doc_id: cord_uid: zouscolw objective: to describe the place and causes of acute cardiovascular death during the covid- pandemic. methods: retrospective cohort of adult (age ≥ years) acute cardiovascular deaths (n= ) in england and wales, from january to june . the exposure was the covid- pandemic (from onset of the first covid- death in england, march ). the main outcome was acute cardiovascular events directly contributing to death. results: after march , there were acute cardiovascular deaths of which . % related to covid- , and an excess acute cardiovascular mortality of (+ %). deaths in the community accounted for nearly half of all deaths during this period. death at home had the greatest excess acute cardiovascular deaths ( , + %), followed by deaths at care homes and hospices ( , + %) and in hospital ( , + %). the most frequent cause of acute cardiovascular death during this period was stroke ( , . %), followed by acute coronary syndrome (acs) ( , . %), heart failure ( , . %), pulmonary embolism ( , . %) and cardiac arrest ( , . %). the greatest cause of excess cardiovascular death in care homes and hospices was stroke ( , + %), compared with acs ( , + %) at home and cardiogenic shock ( , + %) in hospital. conclusions and relevance: the covid- pandemic has resulted in an inflation in acute cardiovascular deaths, nearly half of which occurred in the community and most did not relate to covid- infection suggesting there were delays to seeking help or likely the result of undiagnosed covid- . cardiovascular disease (cvd) is one of the most prevalent underlying condition associated with increased mortality from covid- infection. [ ] [ ] [ ] [ ] [ ] yet, we and others have shown a substantial reduction in presentations to hospitals with acute cardiovascular (cv) conditions including acute coronary syndrome, heart failure, cardiac arrhythmia and stroke during the pandemic. [ ] [ ] [ ] [ ] [ ] [ ] this would be expected to result in a much higher number of deaths, unless there has been an actual decrease in the incidence of these acute conditions. the detailed impact on mortality from acute cvd has, however, not been studied at national level. we now report, with high temporal resolution, cv-specific mortality during covid- in england and wales. in particular, we have evaluated the location of cv deaths (eg, hospitals, home or care homes), their relation to covid- infection and the specific cv fatal events that contributed directly to death. this information is vital for the understanding of healthcare policy during the pandemic and to assist governments around the world reorganise healthcare services. the analytical cohort included all certified and registered deaths in england and wales ≥ years of age, between january and june recorded in the civil registration deaths data of the office for national statistics (ons) of england and wales. the primary analysis was based on any of the th revision of the international statistical classification of diseases and related health problems (icd- ) codes corresponding to the immediate cause of death and contributed causes registered, as stated on the medical certificate of cause of death (mccd). the mccd is completed by the doctor who attended the deceased during their last illness within days unless there is to be a coroner's postmortem or an inquest. cv events directly leading to death (herein called acute cv deaths) were categorised as acute coronary syndrome (st-elevation myocardial infarction (stemi), non-stemi, type myocardial infarction, reinfarction) abbreviated as acute coronary syndrome, heart failure, cardiac arrest, ventricular tachycardia (vt) and/ or ventricular fibrillation (vf), stroke (acute ischaemic stroke, acute haemorrhagic stroke, other non-cerebral strokes, unspecified stroke), cardiogenic shock, pulmonary embolism, deep venous thrombosis, aortic disease (aortic aneurysm rupture and aortic dissection) and infective endocarditis (online supplemental table ). icd- codes 'u ' (confirmed) and 'u ' (suspected) were used to identify whether a death was related to covid- infection on any part of the mccd. the place of death as recorded on the mccd was classified as home, care home and hospice, and hospital. baseline characteristics were described using numbers and percentages for categorical data. data healthcare delivery, economics and global health were stratified by covid- status (suspected or confirmed, not infected), age band (< , - , - , - , + years)), sex and place of death (home, hospital, care home or hospice). the number of daily deaths was presented using a -day simple moving average (the mean number of daily deaths for that day and the preceding days) from february up to and including june , adjusted for seasonality. the expected daily deaths from february up to and including june were estimated using farrington surveillance algorithm for daily historical data between and . the algorithm used overdispersed poisson generalised linear models with spline terms to model trends in counts of daily death, accounting for seasonality. the number of non-covid- cv deaths each day from february were subtracted from the estimated expected daily deaths in the same time period to create a zero historical baseline. deaths above this baseline maybe interpreted as excess mortality, which were calculated as the difference between the observed daily deaths and the expected daily deaths. negative values, where the observed deaths fell below the expected deaths, were set to zero. the rate of excess deaths was derived from dividing excess mortality by the sum of the expected deaths between march and june . for the categories of acute cv death, the icd- code on the mccd was counted only once per deceased. thus, the overall rate of acute cv death represents the number of people with a direct cv cause of death. given that, people may have more than one of the predefined cv events leading to death, analyses for each of the predefined cv categories represent the number of events (not people) per category. for the purposes of this investigation, cvd that contributed, but did not directly lead to death were excluded from the analyses. all tests were two sided and statistical significance considered as p< . . statistical analyses were performed in r v. . . , and the farrington surveillance algorithm was fitted using r package 'surveillance'. between january and june , there were deaths from all causes among adults. the proportion of deaths increased with increasing age band and there were ( . %) in women (table ) . people dying from any of the directly contributing cv categories accounted for ( . %) of all deaths, of which . % had at least two of the predefined cv categories that directly contributed to death. most deaths occurred in hospital ( . %) followed by home ( . %) and at care home ( . %). following the first uk death from covid- on march , there were acute cv deaths of which . % related to covid- ( . % suspected; . % confirmed), and an excess acute cv mortality of (a proportional increase of %) compared with the expected historical average in the same time period of the year. covid- deaths accounted for ( . %) of all excess deaths after this date (figure , , table ) . compared with deaths prior to march , covid- related acute cv deaths were more likely to occur in hospital ( . % vs . %), much less at home ( . % vs . %) and remained of similar proportions to non-covid- -related acute cv deaths in care homes ( . % vs . %). the rate of covid- -related excess cv deaths was higher in hospitals than care homes (a proportional increase of % vs + %) and less at home (a proportional increase of %). excess covid- related acute cv deaths occurred in similar proportions for men and women (a proportional increase of % vs %), and the rate of excess covid- -related acute cv deaths was comparable across the age bands (table ) . the greatest proportional increase of excess covid- -related acute cv death was due to pulmonary embolism ( , a proportional increase of %) followed by stroke ( , a proportional increase of %), acute coronary syndrome ( , a proportional increase of %), cardiac arrest ( , a proportional increase of %) and heart failure ( , a proportional increase of %) (figure , table ). the most frequent causes of excess acute cv death in care homes and hospices were stroke ( , a proportional increase of %) and heart failure ( , a proportional increase of %), which compared with acute coronary syndrome ( , a proportional increase of %) and heart failure ( , a proportional increase of %) at home, and pulmonary embolism ( , a proportional increase of %) and cardiogenic shock ( , a proportional increase of %) in hospital ( figure , table ). for stroke, acute coronary syndrome, heart failure and cardiac arrest, the numbers of deaths in hospital were lower than the historical baseline ( figure ) . we show for the first time, in a nationwide complete analysis of all adult deaths, the extent, site and underlying causes of the increased acute cv mortality during the covid- pandemic compared with previous years. this shows that the pandemic has resulted in an abrupt inflation in acute cv deaths above that expected for the time of year. nearly half of the deaths occurred outside of the hospital setting, either at home or in care homes, with people's homes witnessing the greatest proportional increase in excess acute cv deaths. the most frequent cause of acute cv death during the covid- pandemic in england and wales was stroke followed by acute coronary syndrome and heart failure. this is key information to optimise messaging to the public, as well as for allocation of health resources and planning. numerous international studies have reported the decline in hospital presentations for a range of cv emergencies. [ ] [ ] [ ] [ ] [ ] [ ] to the best of our knowledge, this is the first study to show that this is associated with an adverse overall cv impact. while stroke and acute coronary syndrome accounted for the vast majority of acute cv deaths, the number of deaths in hospital due to these conditions fell below that expected for the time of year and it increased in the community, and particularly in people's homes. this 'displacement of death', most likely, signifies that the public either did not seek help or were not referred to hospital during the pandemic-a finding supported by the fact that the majority of acute cv deaths were not recorded as related to infection with covid- . given the times series plots show that the excess in acute cv mortality began in late march and peaked in early april , government directives at the time including the onset of the uk lockdown on march could have accentuated a maladaptive public response. the major causes of acute cv death were different between hospital and community settings. this 'differential of cause of death by place' provides an understanding of how the infection and public response to the pandemic played out. the most frequent cause of excess acute cv death in people's homes was acute coronary syndrome, in care homes and hospices it was stroke and in hospital it was pulmonary embolism. assuming that the public did not seek help for medical emergencies for fear of contagion in hospital or to prevent hospitals being overwhelmed, then it is not surprising that there were deaths from acute coronary syndrome at home. complications of untreated acute myocardial infarction include cardiac arrest, arrhythmia and acute heart failure. we found that in people's homes there were increases in excess acute cv deaths from cardiac arrest (in line with others' findings ) and heart failure, and in hospitals there were increases in excess deaths from cardiogenic shock and vt and vf-all of which are complications of late presentation myocardial infarction. in hospital, we also found an inflation of deaths from infective endocarditis and aortic dissection and rupture, indicating perhaps a more advanced (and for some, irreversible) stage of disease presentation during the pandemic, akin to the situation with acute myocardial infarction. care homes and hospices witnessed a substantial increase in excess acute cv deaths. herein, stroke, heart failure, acute coronary syndrome and pulmonary embolism were the the most common causes of acute cv death. this finding highlights the susceptibility of the elderly and comorbid to the wider implications of covid- crisis. that is, not only were care home residents prone to the respiratory effects of covid- infection but they will also have been exposed to the acute cv complications of covid- and decisions not to go to hospital for fear of becoming infected. this situation will have been exacerbated by several factors, including the discharge of unknowingly infected patients from hospitals to care homes early in the course of the pandemic (where the virus could easily spread and actions to reduce the spread of the virus in social care were too late and insufficient ), a lack of systematic antibody testing for the sars-cov- virus, the efficient person-to-person transmission of the virus, and its propensity to death in the vulnerable. while previous reports have described an elevated risk of death among the elderly and people with cv disease during the covid- pandemic, none have characterised the cv events directly leading to death and few quantified the excess in acute cv mortality. to date, insights have been derived from small series of cases, regional or national death records data-each reporting elevated mortality rates, but none by the type and place of cv death together. - the unique strengths of this investigation include full population coverage of all adult deaths across places of death. most previous reports have been confined to hospitals deaths and have not captured the full extent of the impact of the pandemic, including deaths figure time series of acute cardiovascular (cv) deaths by covid- , by cause of death. the number of daily cv deaths is presented using a day simple moving average (indicating the mean number of daily cv deaths for that day and the preceding days) from february up to and including june , adjusted for seasonality. the number of non-covid- excess cv deaths each day from february were subtracted from the expected daily death estimated using farrington surveillance algorithm in the same time period. the green line is a zero historical baseline. the red line represents daily covid- cv death from march to june ; the purple line represents excess daily non-covid- cv death from march to june and the blue line represents the total excess daily cv death from february to june . vf, ventricular fibrillation; vt, ventricular tachycardia. time series of acute cardiovascular (cv) deaths by cause of death and place of death. the number of daily cv deaths is presented using a -day simple moving average (indicating the mean number of daily cv deaths for that day and the preceding days) from february up to and including june , adjusted for seasonality. the number of non-covid- excess cv deaths each day from february were subtracted from the expected daily death estimated using farrington surveillance algorithm in the same time period. the green line is a zero historical baseline. the red line represents excess daily death at hospital; the purple line represents excess daily cv death at care home and hospice and the blue line represents excess daily cv death at home. vf, ventricular fibrillation; vt, ventricular tachycardia. outside of hospitals in people who may not have been tested for the disease. nonetheless, our study has limitations. during the covid- pandemic, emergency guidance enabled any doctor in the uk (not just the attending) to complete the mccd, the duration of time over which the deceased was not seen before referral to the coroner was extended from to days, and causes of death could be 'to the best of their knowledge and belief ' without diagnostic proof, if appropriate and to avoid delay. this may have resulted in misclassification bias, with under-reporting of the deaths directly due to cv disease in preference to covid- infection (which is a notifiable disease under the health protection (notification) regulations ) or respiratory disease. in fact, we found that mccds with covid- certification less frequently contained details of acute cv events directly leading to death. although the mccd allows the detailing of the sequence of events directly leading to death, we found that after march few ( . %) had multiple acute cv events recorded, and therefore the categorisation of the acute cv events effectively represents per-patient events. the lower proportion of deaths with covid- at home and in care homes may represent the lack of access to community-based covid- testing. equally, because there was no systematic testing of the uk populace for the presence the covid- , deaths associated with the infection may have been underestimated. this analysis will have excluded a small proportion of deaths under review by the coroner, though typically these will have been unnatural in aetiology. in addition, we did not include the spatial information in the farrington surveillance algorithm, which may affect the accuracy of the estimates for the expected death. to date, there is no whole-population, high temporal resolution information about acute cv-specific mortality during the covid- pandemic. through the systematic classification of all adult deaths in england and wales, it has been possible to show that there has been an excess in acute cv mortality during the covid- pandemic, seen greatest in the community and which corresponds with the onset of public messaging and the substantial decline in admissions to hospital with acute cv emergencies. cardiogenic shock ( %) ( %) ( %) ( %) (+ %) (+ %) vt and vf ( %) (+ %) ( %) ( %) (+ %) (+ %) deep vein thrombosis ( %) (+ %) ( %) ( %) (+ %) (+ %) *the positive excess rate in hospital was due to setting those daily deaths below the expected historical average to zeros. cv, cardiovascular; vf, ventricular fibrillation; vt, ventricular tachycardia. what is already known on this subject? ► cardiovascular disease is one of the most prevalent underlying conditions associated with increased mortality from covid- infection, along with dementia and alzheimer's disease. ► at the same time, there has been a substantial reduction in presentations to hospitals with acute cardiovascular (cv) conditions. ► our study of all adult deaths in england and wales between january and june has quantified the cv mortality impact of the covid- pandemic, be this related to contagion and/or the public response. ► it shows that during the pandemic there has been an inflation in acute cv deaths above that expected for the time of year. ► home death had the greatest increase in excess acute cv death, and the most frequent cause of acute cv death during this period was stroke, followed by acute coronary syndrome. how might this impact on clinical practice? ► these contemporary nationwide cause and place of mortality data provide key information to optimise messaging to the public, as well as for allocation of health resources and planning. digital with the mortality data and takes responsibility for the integrity of these data. the programme was endorsed the british heart foundation collaborative, which also includes health data research uk, hsc public health agency, national institute for cardiovascular outcomes research, cancer research uk, public health scotland, nhs digital, sail databank and uk health data research alliance. contributors cpg and jw was responsible for the study design and concept. jw performed the data cleaning and analysis. jw and cpg wrote the first draft of the manuscript, and all authors contributed to the writing of the paper. funding jw and cpg are funded by the university of leeds. mam is funded by the university of keele. competing interests none declared. patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. patient consent for publication not required. ethics approval ethical approval was not required as this study used fully anonymised routinely collected civil registration deaths data. the data analysis was conducted through remote access to nhs digital data science server. provenance and peer review not commissioned; externally peer reviewed. estimating excess -year mortality associated with the covid- pandemic according to underlying conditions and age: a populationbased cohort study case-fatality rate and characteristics of patients dying in relation to covid- in italy features of uk patients in hospital with covid- using the isaric who clinical characterisation protocol: prospective observational cohort study pre-existing conditions of people who died with covid- peop lepo pula tion andc ommunity/ birt hsde aths andm arriages/ deaths/ articles/ anal ysis ofde athr egis trat ions noti nvol ving coro navi rusc ovid en glan dand wale s d ecem ber t o ma y / technicalannex# characteristics the covid- pandemic and the incidence of acute myocardial infarction collateral effect of covid- on stroke evaluation in the united states emergency hospital admissions and interventional treatments for heart failure and cardiac arrhythmias in germany during the covid- outbreak: insights from the german-wide helios hospital network covid- pandemic and admission rates for and management of acute coronary syndromes in england patient response, treatments and mortality for acute myocardial infarction during the covid- pandemic user guide to mortality statistics an improved algorithm for outbreak detection in multiple surveillance systems out-of-hospital cardiac arrest during the covid- outbreak in italy let's be open and honest about covid- deaths in care homes staggering number" of extra deaths in community is not explained by covid- sars-cov- : virus dynamics and host response sex-differences in mortality rates and underlying conditions for covid- deaths in england and wales excess uk deaths in covid- pandemic top , real estimates of mortality following covid- infection covid- ) mortality rate use of all cause mortality to quantify the consequences of covid- in nembro, lombardy: descriptive study guidance for doctors completing medical certificates of cause of death in england and wales tackling uk's mortality problem: covid- and other causes acknowledgements we acknowledge the intellectual input of professor colin baigent, university of oxford. jw had full access to all the data in the study and takes responsibility for the accuracy of the data analysis. the ons provided nhs this article is made freely available for use in accordance with bmj's website terms and conditions for the duration of the covid- pandemic or until otherwise determined by bmj. you may use, download and print the article for any lawful, non-commercial purpose (including text and data mining) provided that all copyright notices and trade marks are retained. key: cord- -ykfvbypo authors: mclaughlin, r.; kelly, c. j.; kay, e.; bouchier-hayes, d. title: the role of apoptotic cell death in cardiovascular disease date: journal: ir j med sci doi: . /bf sha: doc_id: cord_uid: ykfvbypo background: programmed cell death, or apoptesis, is a distinct, managed form of cell death. it is fundamentally different from necrosis. it is a genetically controlled, energy-dependent method of cellular deletion without inflammation. in the cardiovascular system, apoptosis occurs as a primary and secondary event in disease pathogenesis. this review addresses our current understanding of the initiation, propagation and significance of apoptosis in the cardiovascular system, as well as assessing therapeutic potentials arising therefrom. methods: a medline search was performed and relevant publications reviewed. further articles were obtained from the references of these publications. results and conclusions: apoptotic cell death is a key element in the pathogenesis and progression of ischaemia-reperfusion (ir) injury, cardiac failure, myocardial infarction, atherosclerosis, endothelial dysfunction and the clinical syndromes which these situations produce. our increased understanding of the role of apoptosis in the pathogenesis of cardiovascular disease offers potential to develop new therapeutic strategies. the term and concept of apoptotic or programmed cell death first appeared in the medical literature in .' since then, abnormalities of programmed cell death, or manipulation of the processes of apoptosis, have become recognised as being of central importance in a multitude of pathological states. the concept that cells possess a modifiable capacity to dictate whether or not they enter into a controlled lethal process is significant in understanding cellular responses to adverse circumstances. apoptosis is a physiological process for the deletion of unwanted or senescent cells without inflammation. apoptosis works synchronously with, but opposite to, mitosis in counterbalancing and regulating tissue kinetics.= the regression of interdigital webs is an example of physiological apoptotic cell death. the process of apoptosis is an energy-dependent, tightly regulated and genetically encoded mechanism of individual cell disposal. the ubiquitous process of apoptosis is seen in its physiological context, actively deleting cells during embryogenesis, metamorphosis, tissue atrophy, t-cell killing tumour regression and turnover of intestinal epithelial cells.'• ' ' cellular resistance to apoptosis is exhibited in abnormalities of these processes and particularly in carcinogenesis. apoptosis is exhibited pathologically in response to stresses such as ischaemia-reperfusion (ir) hyperthermia, radiotherapy and pharmacological agents. . . however, pathological or inappropriate apoptosis has significant structural and functional implications for the involved end organs. in mature organisms, ongoing apoptosis maintains cellular homeostasis in renewing cell types such as intestinal epithelium and leucocytes. in contrast, apoptosis is not a feature of non-renewable parenchymal cells in the normal course of events. this is particularly so in terminally differentiated cells such as cardiomyocytes. in the myocardium, the finding of apoptosis after reperfusion is particularly significant as it implies late and potentially avoidable cell death. in cardiovascular disease, developments in molecular biology have facilitated an increasing appreciation of the role of apoptosis, not only in heart failure and myocardial infarction but also in ir injury, ageing, vascular wall remodelling and atherosclerotic plaque destabilisation.'' - apoptosis is now recognised to not only contribute to the initiation of these disease states but also to influence long-term outcome in terms of eventual myocardial pump failure or arterial occlusion. apoptosis occurs acutely but also persists after an initial inflammatory or ischaemic insult, thus perpetuating the injury sustained. as apoptosis is a process that can be potentially attenuated, recognition of its contribution to cardiovascular disease opens exciting new therapeutic avenues. apoptotic cell death is a morphologically characteristic process identified by a series of discrete distinguishable steps. initially, detachment of the affected cell from its surrounding cells occurs. apoptosis proceeds in an orderly controlled fashion with characteristic cell volume reduction followed by convolution and blebbing of the cell membrane. condensation of the cytoplasm is reflected in an increased cell density. the nuclear chromatin condenses, in association with activation of an endogenous endonuclease, and gathers beneath the nuclear membrane. the entire cell may now split into membranebound structures with preserved organelle structures. these apoptotic bodies may be visualised within tissues at this stage. the cell now becomes subject to tissue phagocytic cell recognition and action. membrane integrity is maintained throughout the process, in contrast to the swelling and subsequent rupture of a cell dying by necrosis. the destructive spilling of intracellular contents into the surrounding tissues evoking the classical inflammatory responses of necrosis is absent in apoptosis. apoptotic cells are phagocytosed by neighbouring cells without the inflammation associated with, or need for, neutrophil recruitment.'-' the process is an inducible phenomenon with a key biological event being the activation of an endogenous endonuclease, resulting in internucleosomal cleavage of genomic dna. the extensive nucleosomal dna fragments resulting from this endonuclease activation are - bp oligonucleosomal units, or multiples thereof, and are a pathological hallmark of apoptosis, which may be detected as a ladder on gel electrophoresis. single cell microgel electrophoresis, the `comet' method, can also demonstrate these distinctive dna strand breaks in individual apoptotic cells by their migration into a specific `tail' pattern. a' the specific cleavage pattern of genomic dna by endonucleases yields strand breaks that can be identified by labelling of the free '-oh dna ends with modified nucleotides. terminal deoxynucleotidyl transferase (tdt) catalyses polymerisation of nucleotides to free '-oh dna ends in a template manner in order to label these breaks in the tunel (tdt-mediated dutp nick end labelling) assay for apoptosis. ' the regulation of intracellular ph is integrated into the progression of apoptosis, with a fall in ph preceding the terminal subcellular events. preconditioning of cells using ischaemia protects against cell death through protein kinase c (pkc) mechanisms. • ° the pkc protective mechanism involves diminution of intracellular acidification and inhibition of this pathway in vitro. this has been shown to protect cardiomyocytes from apoptotic death in response-simulated ir injury. although the apoptotic cell does not evoke inflammatory responses to stimulate its removal, it does signal its death to resident tissue phagocytes and neighbouring cells. this is achieved by cell membrane surface expression of the normally exclusively cytoplasmic facing aminophospholipid phosphatidylserine. [ ] [ ] [ ] this phosphatidylserine extrusion facilitates phagocytic removal of the apoptotic cell. it also facilitates phosphatidylserine detecnon on the cell surface by annexin v, thus allowing for another method of identifying apoptotic cells.tm as phosphatidylserine surface expression is an early event. in the apoptotic pathway, annexin v labelling allows identification of apoptotic cells at this timepoint. cell surface phosphatidylserine expression promotes thrombin generation in vitro. this finding raises the possibility that apoptosis seen in atherosclerotic plaques may contribute to their thrombogenicity in vivo. myocyte apoptosis is difficult to recognise as it affects inchvidual myocytes and is scattered across the muscle or myocardium. the process of apoptosis is rapidly completed in as little as minutes (as assessed by time lapse photography'') and may not be detectable beyond hours.'#-" given this fact, the detection of a relatively low prevalence ( . %) of apoptotic myocytes in the myocardium at a given time point is highly significant, implying that if sustained for one year this would lead to an unsustainable loss of more than % of the myocardium. various investigators have reported great variability in the extent of apoptosis in diseased myocardium, some reporting as high as % in ischaemic cardiomyopathies.'° the immunohistochemical finding of apoptosis in the myocardium can be further verified by demonstrating dna laddering on electrophoresis of comparable specimens. biochemical control of apoptosis the process of apoptosis is orchestrated by the interactions of numerous proto-oncogenes and environmental factors. initial clues about the genetic command structure of apoptosis were gained from landmark studies on the tiny nematode caenorhabditis elegans showing that there are , cell births and cell deaths in the construction of the adult worm.`-` further work identified two genes ced- and ced- which are essential for the programmed death of these cells. mutations of these genes permit these cells, which would normally die, to survive. a third gene, ced- , is necessary to sustain life through tonic inhibition of the. pro-apoptotic genes ced- and ced- . the ced- gene shares structural and functional homology with the human bcl- gene. the c. elegans ced- gene equivalent product in humans is a family of cystine proteases associated with interleukin (il) bconverting enzyme (ice). these cystine proteases, referred to as caspases, are key mediators of apoptosis.' ," a characteristic feature of the caspases (of which have been identified to date) is the presence of a cysteine residue within a highly conserved pentameric sequence in the catalytic centre. ' the caspase family appears to play a key role in the initiation of apoptotic cell death, being present during apoptosis and their inhibitors protecting against apoptotic cell death. caspase activation produces the structural nuclear changes in apoptosis via protein cleavage of cytoskeletal (actin, fodrin and the familial polyposis protein apc), dna repair (dna-pk, parp), nuclear envelope (lamins) and cell cycle (retinoblastoma protein) elements.' in addition, one of the endonucleases thought to be involved in the chromatin changes of apoptosis is activated by proteolytic cleavage. the activation of caspases may be initiated via receptordependent mechanisms, specifically fas for two members of the family, caspase- and - . ' -{' tumour necrosis factor (tnf) and fas ligand-mediated apoptotic death rely upon caspase activation as a prerequisite to cell deletion.'' there are a number of characterised death receptors, including fas, tnfri, death receptor (dr)- (apo- ) and tnf-related, apoptosis-inducing ligand (trail) receptors dr- and - . binding of the respective ligands to these receptors results in recruitment -of dedicated intracellular adapter proteins to the cellular membrane. for tnfr and dr , this protein is tnfr associated death domain protein (tradd).' " fas-associated death domain (fadd) protein fulfils this role for fas and dr- . these proteins then interact with caspases, precipitating fatal cellular proteolytic cleavage. work by enari et al'° on apoptosis induced by fas receptor interaction suggested that the caspases are sequentially activated. the ultimately lethal proteolytic effects of caspase activation are mediated downstream by caspases - and - . other upstream caspases with long prodromes have regulatory functions." bcl- family the bcl- (b-cell lymphoma- ) gene was isolated from the breakpoint of a translocation between chromosomes and . the gene is found in % of follicular lymphomas and % of diffuse non-hodgkin's lymphomas.$ • this translocation results in the bringing together of the bcl- gene and the immunological heavy chain locus with increased bcl- production and expression. bcl- expression in myeloid precursors and pro-b cells dramatically improves survival on withdrawal of growth factors. the c. elegans equivalent to bcl- was found to be ced- , whose loss of function mutation precipitated apoptosis, thus characterising it as a negative regulator of apoptosis. bcl- gene expression in humans correlates with resistance to apoptotic cell death. bcl- overexpression confers resistance to the action of gamma radiation and chemotherapeutic drugs that act by inducing dna damage. it does not prevent damage to the cells from these agents, but rather inhibits apoptosis occurring in response in the heart, diminished bcl- expression in the right ventricle has been associated with apoptotic cell death assisting the transition to an adult circulatory system. in coronary artery ligation models and in vivo with heart failure, bcl- expression is markedly upregulated. this upregulation is suggested to be a compensatory mechanism by the myocardium in the face of widespread cell death. the exact mechanism whereby bcl- inhibits apoptosis is uncertain; however, its co-localisation with bax on the mitochondrial membrane suggests this may be a key site." the release of mitochondrial proteins, cytochrome c and apoptosis-inducing factor (aif) into the cytosol activates caspases and thus cell death. , a caspase - , - and - activation are downstream from bcl- . however, bcl- blocks the release of cytochrome c and aif from mitochondria, thus suggesting a potential site for its antiapoptotic activity. , bcl- has also been suggested to function as an anti-apoptotic agent through its anti-oxidative role.°' however, bcl- is also anti-apoptotic under anaerobic cell culture conditions and apoptosis occurs in similar circumstances without reactive oxygen species being present bcl- still remains protective. ' therefore, bcl- cannot function through purely antioxidant-dependent mechanisms, nor can the generation of reactive oxygen species be an exclusive effector of programmed cell death. the extended bcl- family contains nine members, defined by the presence of three structural motifs in the protein sequence. ' these proteins may be pro-or anti-apoptotic and appear to interface after cell surface signals induce caspase activation. bax is a pro-apoptotic member of this family with % homology to bcl- . it appears that the formation of bax/bcl- heterodimers is an important control point in apoptosis. mutations of bcl- may allow it to evade the blocking effect of binding to bax and thus allow anti-apoptotic signals to assume dominance within the cell. thus, if the bax/bcl- ratio is altered, this allows the dominant pro-or anti-apoptotic influence to prevail. the overexpression of bcl- and the anti-apoptotic proto-oncogene bcl-xl have been shown to be associated with an alteration in the onset of toxin-induced apoptosis.^•t he bcl-x member of this family offers protection from apoptosis by binding to bax, thus preventing the formation of bax homodimers which accelerate the apoptotic process.'° the observed reduction in bcl-x seen in experimentally injured arteries may therefore allow bax homodimerisation to occur, precipitating smooth muscle cell apoptosis. ' bcl- expression is markedly upregulated in humans with heart failure. however, bax expression is unaffected. this upregulation occurs in the context of an increase in cardiomyocyte apoptosis. ` the increase in bcl- expression appears in this context to be a secondary compensatory mechanism to promote survival of the remaining cardiomyocytes. the tumour suppressor gene p is pro-apoptotic. loss of p function is the result of the most commonly mutated gene in human malignancies. it acts as a cellular sensor of dna damage, inappropriate oncogene activation, hypoxia and the presence or absence of certain cytokines, and induces apoptosis in response to such stimuli. p has a role in chaperoning dna, inducing apoptotic death when defects emerge. $ in the cardiovascular system, gene transfer-induced p overexpression in normoxic cultured rat neonatal cardiomyocytes can precipitate apoptotic cell death.'; hypoxia in cultured cardiomyocvtes, which induces apoptosis, is also associated with increased p expression.° however, in p deficient mice, apoptosis may also occur in cardiomyocytes by forced entry into the s phase of the cell cycle.' ' ' ' ° in these p deficient mice, ligation of the left coronary artery does not alter the degree of apoptosis in hypoxic areas when compared with wild type mice. °k irshenbaum and de r%,ioissac'' and kirshenbaum have shown that p interacts dynamically with other proto-oncogenes involved in apoptotic regulation. p expression causes an upregulation of pro-apoptotic bax, but this effect and its own apoptotic potency are downregulated by bcl- expression. ' nitric oxide nitric oxide (no) is involved in vessel relaxation, inhibition of smooth muscle and endothelial cell proliferation, and reduction of platelet adhesion. ' ° in the myocardium, recent work has shown that constitutively expressed no synthase (cnos), inducible no synthase (inos) and endothelial no synthase (enos) have functional autocrine and paracrine effects similar to their effects on vascular cells.$` no release precipitates myocardial depression in response to systemic disturbance and part of this dysfunction is mediated by noinduced cardiomyocyte apoptosis. in a rat model of heterotopic cardiac transplantation, it was shown that rejection was associated with significant induction of the mrna, protein and enzyme activity of inos. ' this was detectable in endothelial cells, cardiomyocytes and infiltrating monocytes. further work expanded the understanding of the role of no in this context by demonstrating that cytokine induction of inos was associated with no-mediated death of cardiomyocytes in vitro." szabolcs et al subsequently confirmed that the mechanism whereby no-induced cardiomyocyte death in vivo was indeed apoptotic, thereby demonstrating directly the link between no and cardiac apoptosis.`°°i n vivo gene transfer studies of endothelial cell no synthase (ecnos) activity to the myocardium resulted in cardiomyocyte apoptosis. the suggestion from this study was that the inflammation and cvtokine release associated with myocardial infarction, myocarditis and . cardiomyopathies may result in similar pro-apoptotic effects on cazdiomyocytes via no-dependent mechanisms. in the myocardium, no activates cgmp-dependent protein kinases and cgmp-modulated phosphodiesterases, thus regulating calcium current and contraction. in the course of oxidative injury (such as ir), the simultaneous production of no and superoxide leads to the formation of peroxynitrite. it now appears that peroxynitrite is a significant mediator of injury previously attributed solely to no or superoxide. peroxynitrite production results in single strand dna breaks, with consequent activation of poly adp-ribose synthetase ( pars) which is cytotoxic. this mechanism has now been shown to cause endothelial cell apoptosis in vitro. ' no appears to play a major role in vessel wall apoptosis by inducing programmed cell death in smooth muscle cells and infiltrating monocytes. $' no activity is reduced in the vessel wall in hypercholesterolaemia suggesting a potential mechanism whereby diminished apoptotic control of cellular homeostasis allows progression of atherosclerosis. nf-kb has a potent anti-apoptotic role. in the vascular endothelium, it is expressed constitutively and can be upregulated by stimuli such as il- , thrombin, platelet-derived growth factor (pdgf) and basic fibroblast growth factor. , in addition, it is essential for in vitro proliferation of vascular. smooth muscle cells. expression of vascular cell adhesion molecule (vcam)- is also nf-kb dependent. "° accumulating evidence has shown nf-kb is activated in both vascular cell injury and atherosclerosis. ' in a rat model of balloon catheter carotid injury, nf-kb regulated genes, monocyte chemoattractant protein (mcp)- and vcam- were apparent in smooth muscle cells within four hours. this expression occurred in parallel with macrophage infiltration. this finding links nf-kb to the injury and lesion formation responses of the endothelium." nf-kb plays a key role in regulating vascular smooth muscle cell proliferation in normal and damaged endothelium. abnormalities in the regulatory processes of smooth muscle and endothelial cell proliferation clinically produce neointimal hyperplasia. its anti-apoptotic role in such circumstances may ultimately be harmful. its upregulation in atherosclerotic plaques also suggests a potential role in their pathogenesis. , tumour necrosis factor tumour necrosis factor (tnf) is a potent inducer of apoptotic cell death in many cell types, subsequent to binding to its receptor. tnfa has been shown to induce apoptosis in cardiomyocytes in vitro and in vivo. ' two cell surface receptors for tnfa, the kd and kd tnfr and tnfr , respectively, are both functionally expressed by cardiomyocytes in the failing myocardium. ' fas expression in the murine myocardium has been found to be enhanced in models of viral congestive cardiac failure and this was associated with apoptosis of inflammatory cells and cardiomyocytes. fas expression is also upregulated in coronary artery ligation.'$ apoptosis in the normal cardiovascular system in the cardiovascular system, apoptosis is first seen during embryogenesis with remodelling of the bulbis cordis and atrioventricular cushions by programmed cell death.' postnatal cardiac apoptotic cell death is expressed preferentially in the right ventricle, with the resultant reduced right chamber muscle mass accompanying the transition to an adult circulatory system." this cardiac remodelling is associated with and facilitated by diminished expression of the anti-apoptotic _ gene bcl- . the development of the cardiac conductive system of pathways also employs apoptosis in deleting unnecessary cells. aberrant persistent atrioventricular conductive pathways, which would normally be removed by apoptosis, have been suggested as a potential cause of abnormalities such as wolff-parkinson-white syndrome.' ' the postnatal development of the vascular system involves cell death by apoptosis in vessel remodelling." , '°° in an analogous situation, the vascular endothelial post-injury response relies on homeostasis between proliferation and remodelling by apoptotic cell deletion in order to prevent excessive neointimal hyperplasia.'°' apoptosis in ischaemia-reperfusion ( r) the vascular endothelium, by virtue of its location at the blood-tissue interface, is in the front line of ir injury. reperfusion of ischaemic tissues is a consistent aim of clinical therapies. however, as parks and granger showed, reperfusion is a double-edged sword which can exacerbate the initial hypoxic injury. reactive oxygen species such as superoxide, no, hydrogen peroxide, lipid peroxides and the hydroxyl radical generated in excess during ir injury can subject a cell to oxidative stress and subsequent apoptosis. in addition to the reactive oxygen species, other elements of the reperfusion injury which are known to cause apoptosis include alterations in intracellular calcium homeostasis, , '° an inflammatory reactiontr" and increased mechanical stretch.'°' the presence of similar mediators in other contexts leads on to a potential role for apoptosis in mediating some of the clinically relevant effects of local and systemic ir injury, as well as the systemic inflammatory response syndrome (sirs). the synergistic dual effect produced under such circumstances results in initial vascular endothelial damage and ultimately end organ injury. an analogous situation occurs in the failing myocardium where reactive oxygen species, inflammatory cytokines, no, hypoxia, reperfusion and mechanical stretch exert pro-apoptotic influences upon the endothelium and cardiomyocyte. ""° autopsy studies of post-myocardial infarction patients, who had initially successful thrombolysis and therefore a patent infarct-related artery, showed a clear subset of cardiomyocytes dying by apoptosis as a result of ir. this finding is also seen in experimental animal models of ir. ` because apoptosis after infarction is an energy-dependent process relying on circulatory nutrient supply, it will occur primarily in reperfused or watershed areas of the infarct. studies have shown that myocardial apoptosis does not occur in purely ischaemic areas, but will occur upon reperfusion of these areas. therefore, while one may attempt to deal with an atherosclerotic thrombus after the event occurs, acute apoptotic cardiomyocyte death, while it may be prompted, begins when flow is partially or fully restored. therefore, reperfusion by recanalisation, thrombolysis or angioplasty provides a window from which time onwards the contribution of apoptotic cell death to the pathological process is potentially amenable to attenuation. antioxidants and free radical scavengers such as superoxide dismutase have been used to counteract the pro-apoptotic effects of oxidative stress on tissues exposed to ir. ° , " vascular endothelial cells in vitro are protected from apoptotic cell death in a sirs analogous model using the amino acid taurine. apoptosis in the diseased myocardium cardiomyocyte cell loss and scar formation are integral components of cardiac dysfunction from numerous aetiologies. experimental and autopsy studies have confirmed that cardiomyocytes undergo cell death by apoptosis as a component of hypoxia, ir, heart failure , , , , and myocardial infarction.l$ following on from this work, the traditional viewpoint of heart failure as a purely haemodynamic continuum has been modified by an increasing awareness of the fact that the interaction of cytokines, neurohormones and apoptotic mediators play a significant role in the evolution and progression of this disease process."° as the myocardium has no regenerative capacity, prevention of apoptosis-induced cardiomyocyte loss has potentially significant clinical implications. the changes in myocardial loading that accompany heart disease of ischaemic or non-ischaemic origin activate various cellular responses. apoptotic myocyte death is a sequel of such physical stresses." animal studies of pressure overload hypertrophy models, such as aortic banding, loading of isolated papillary muscle and genetically determined hypertension, have all demonstrated resultant apoptotic cardiomyocyte losses. , ' . the progression of cardiac hypertrophy resulting from pressure overload involves myocyte loss with hypertrophy of the remaining myocytes and proliferation of non-muscle cells. in a rat model of pressure overload, the secondary cardiac hypertrophy and remodelling was initiated by a wave of apoptotic cardiomyocyte death, thus implicating apoptosis in the pathogenesis of these events." leri et al' ` have suggested a mechanism whereby myocyte stretch-induced autocrine release of angiotensin ii is associated with activation of p , thus resulting in prolonged upregulation of myocyte apoptosis. this suggestion was reinforced by kajstura et al"' who noted an increased incidence of apoptosis after treatment of isolated adult cardiomyocytes with angiotensin ii. furthermore, in myocytes subjected to stretch, the bcl- to bax ratio was lowered, thus increasing myocyte apoptotic susceptibility. thus, the possibility arises that apoptosis may be involved in the pathogenesis of overall cardiac remodelling. ' postmortem examination studies of human hearts days post-myocardial infarction showed an incidence of apoptosis of . % in areas distant from the infarct. there was no apoptosis in control non-infarcted hearts.¢ this study implicates apoptosis in a wider myocardial role, not limited to just the infarcted or hypoxic area. the ventricular myocardial adaptation and remodelling that occurs in response to pathological stimuli, while it may be compensatory in the short term, may initiate changes leading ultimately to pump failure. recently, evidence has been accumulating that apoptotic cellular deletion participates, and may be a significant determinant, in this pathological transition process.' ' • ' if this rationale holds true, then the gradual progression of myocardial dysfunction and ultimate failure could potentially be halted by the arrest of apoptosis in the myocardium. the range of cardiac diseases in which inappropriate apoptosis has been demonstrated also includes cardiomyopathic failure of ischaemic, viral and idiopathic origin, as well as in arrhythmogenic right ventricular dysplasia.' ' " . " in myocardial ischaemia, apoptosis rather than necrosis has been shown to be the critical determinant of eventual myocardial infarct size and impact. `°, in addition, when reperfusion of the myocardium occurs this prompts a wave of apoptotic cell death, aggravating the ischaemic insult. ` in vitro studies have shown that cardiomyocytes are primed for apoptosis, which may be triggered by cytokines such as tnfa. tnfa is a particularly clinically relevant cytokine, being present during local and systemic ir, sepsis and sirs. chronic infusion of tnfa in vivo has been reported to cause rapid onset of dilated cardiomyopathy with widespread myocyte apoptosis. tnfa levels are elevated in heart failure, as well as in reperfused and infarcted myocardium."' -"' the finding that oxidative stress induced apoptosis in isolated cardiomyocytes is significant as it suggests a direct mechanism whereby ir can induce apoptosis. , ' therefore, cardiomyocytes are primed for apoptosis which may be triggered by cytokines, mechanical stretch or ir. interestingly, it has been suggested that the readiness with which cardiomyocytes undergo apoptosis may also explain the rare incidence of primary cardiac tumours. investigators attempting to promote cellular regeneration in the myocardium have used recombinant adenoviruses, delivering the adenoviral protein s e a, and have induced dna synthesis in terminally differentiated cardiomyocytes. however, this was followed by widespread apoptosis in the absence of a second adenoviral protein e b, a structural and functional homologue of bcl- . introduction of the bcl- homologue allowed cardiomyocvte proliferation and suppressed apoptotic cell death. the interaction of tnfa with cd /apol/fas is directly implicated in the death of cardiomyocytes post jr. ' the evidence above suggests that the expression of apoptosis in the myocardium is a frequent pathological event with significant clinical implications. atherosclerotic plaque destabilisation has been shown to be associated with apoptotic cell death, principally of inflammatory cells and subendothelial smooth muscle cells in the fibrous cap.' - bennett et al ' have shown that human vascular smooth muscle cells derived from normal and atherosclerotic endothelium have different thresholds for expression of apoptosis in vitro. apoptosis in normal vascular endothelial smooth muscle cells occurred only on removal of serum growth factors. however, cells from atherosclerotic plaques died even with high serum conditions. in addition, bcl- expression induced by gene transfer and the cytokines igf and pdgf protected against smooth muscle cell apoptosis.' " this suggests that the prevailing influences in determining the occurrence of apoptosis in vascular smooth muscle cells include proto-oncogene expression and local cytokine interactions. the highest concentration of apoptotic cells in an atherosclerotic plaque appears to be in regions enriched with macrophages. this may be as a result of induction of apoptotic pathways via macrophage tnfa release, reactive oxygen species generation or directly by the production of oxidised low density lipoprotein (ldl). * ldl penetration of the intima is an initiating pathogenic effect in atherosclerosis. •' its subsequent oxidative modification by macrophages and other cells precedes foam cell formation. the effect of oxidised ldl in directly precipitating apoptosis in cultured endothelial cells is significant as it implies a pathway for the participation of apoptosis in the initial pathogenesis and progression of atherosclerosis.' this effect has been exploited therapeutically in ldl receptor knockout mice by using the no precursor l-arginine to prevent and induce regression of atherosclerosis in hypercholesterolaemic animals, possibly by an antioxidant effect or alternatively directly via no.' advanced human atherosclerotic plaques contain sclerotic hypocellular regions. this has led to the suggestion that these areas arise due to apoptotic cell death. as part of the coronary angioplasty versus excisional atherectomy trial (caveat), % of the primary atherosclerotic plaques obtained by directional atherectomy were found to be hypocellular. in addition, apoptotic cell death of proliferating smooth muscle cells has been shown to control their numbers after injury such as atherectomy. fas is widely expressed in atherosclerotic lesions, suggesting a potential receptor-mediated mechanism of apoptotic death within such lesions. ' dong et al have also shown fas-mediated apoptotic cell death to be associated with post-transplant coronary artery disease." the endothelial cell fulfils an interactive role inhibiting proliferation of adjoining cells, thrombus formation and leucocyte adhesion while simultaneously regulating vasomotor tone in response to tissue requirements or systemic stimuli. endothelial dysfunction contributes to abnormalities of these processes. in the endothelium, loss of integrin-mediated cell matrix contact results in apoptotic cell death known as `anoikis'. the term anoikis is derived from the greek word for homeless and is used in this context to denote a cell dying by apoptosis having being displaced from its normal environment. the death of endothelial cells after detachment is beneficial as it prevents detached or abnormal cells from reattaching and growing in a dysplastic fashion .' neutrophil endothelial interaction is a key element in the development of the adult respiratory distress syndrome (ards), often as a component of sirs. the endothelial dysfunction, the role of apoptotic cell death in card iovascular disease manifested as increased capillary permeability in sirs, is partly as a consequence of endothelial cell apoptosis. ' ' it has been suggested that, as a therapeutic strategy in sepsis and sirs, attenuation of the cellular response to this process may be more beneficial than dealing with the extracellular mediators of this process. it has been demonstrated in vitro that induction of heat shock proteins can protect endothelial cells against subsequent apoptosis in response to a sirs insult model. the vascular endothelium response to injury involves smooth muscle cell proliferation and neointimal formation. this proliferation continues for up to weeks. if abnormally regulated proliferation proceeds unchecked, then an overexuberant thickening of the intima will result. however, the total number of cells present after such injury is typically maximal at two weeks.'" the rate-limiting factor in this context is smooth muscle cell deletion by apoptosis, which intercedes to control cell numbers. the inference, therefore, is that this apoptotic control allows neointimal thickness to remain constant despite a continued smooth muscle cell proliferation response to injury. ' pearlman et al showed that extensive apoptosis is detectable in an animal model of intimal injury as early as minutes post-injury. however, no apoptosis was seen in normal vessels. in neointimal hyperplasia, proto-oncogene controls of apoptosis act as molecular thermostats balancing cell death versus proliferation, which determines cell numbers and survival. apoptosis is observed more frequently in restenotic than primary atherosclerotic lesions. ° this is again consistent with the dynamic interaction of cell proliferation and apoptosis exerting homeostatic control on cell numbers in the context of endothelial repair and remodelling. this increased frequency of apoptosis in restenotic rather than primary atherosclerotic lesions is supported by autopsy reports of patients dying after multiple angioplasty procedures."' studies of specimens obtained by percutaneous transluminal atherectomy have provided further evidence of hyperplasia being homeostatically counterbalanced by apoptosis, with both elements being more prominent in restenotic lesions.-°, '- an even greater frequency of smooth muscle cell proliferalion and apoptosis is seen in neointimal hyperplasia complicating stent restenosis. , . the combined significance of these observations is that if apoptosis can be upregulated within these lesions, then the homeostatic mechanisms will dampen down the excessive proliferation of smooth muscle cells. gene therapy delivering the thymidine kinase gene via an adenoviral vector has been successfully used to reduce restenosis after angioplasty in an atheromatous rabbit model by inducing apoptosis of smooth muscle cells."" there is now clear evidence that apoptosis plays a key role in the pathogenesis of cardiovascular disease. its effects range from reduced apoptosis contributing to neointimal hyperplasia to excessive apoptosis inducing cardiac failure. the process of apoptosis may be modulated to attenuate, delay, avoid or precipitate cell death. these modulations are used by neoplastic and viral infected cells to promote their own survival. however, in the context of cardiovascular apoptotic cell -death, similar manipulations have the potential to be used to induce favourable clinical outcomes. genetic transfers of particular cellular profiles of proto-oncogene expression, such as bcl- and p , hold the potential to suppress undesirable apoptosis in cardiac ischaemia or to beneficially promote apoptosis in neointimal hyperplasia. the inflammatory response is a dynamic process and apoptosis is recognised as being a key element in this continuum. it occurs early"° and late" , " in reperfusion as a part of ir injury. the mechanisms involved in these processes are gradually being elucidated in the myocardium and endothelium.' ' in the myocardium and vascular endothelium, reactive oxygen species, inflammatory cytokines, no, hypoxia, reperfusion and mechanical stretch exert pro-apoptotic influences upon cells. the clinical syndromes produced by these influences have established apoptotic contributions to their pathogenesis and significance. the apoptotic cpmponent contributing to these disease processes is potentially amenable to beneficial therapeutic manipulation. this study was funded by a health research board clinical research fellowship and the north-eastern health board. apoptosis: a basic biological phenomenon with wide ranging implications in tissue cell kinetics stretch induced programmed myocyte cell death the cellular basis of dilated cardiomyopathy in humans acute myocardial infarction in humans is associated with activation of programmed myocyte cell death in the surviving portion of the heart apoptosis: an overview cytological and cytochemical studies on cell death and digestion in the foetal rat foot: the role of macrophages and hydrolytic enzymes cell death by apoptosis and its protective role against disease apoptosis and carcinogenesis programmed cell death: apoptosis and oncogenesis the role of apoptosis in the response of cells and tumours to mild hyperthermia programmed cell death: apoptosis and oncogenesis activation of programmed cell death (apoptosis) by cisplatin, other anti-cancer drugs and hyperthermia the importance of oxidative stress in apoptosis reperfusion injury induces apoptosis in rabbit cardiomyocytes apoptosis in the heart apoptosis and the failing human heart myocyte nuclear and possible cellular hyperplasia contribute to ventricular remodeling in the hypertrophic senescent heart in humans apoptotic and necrotic cell deaths are independent contributing variables of infarct size in rats apoptosis is abundant in human atherosclerotic lesions especially in inflammatory cells and may contribute to the accumulation of gruel and plaque instability apoptosis in human atherosclerosis and restenosis apoptosis in the infarcted human heart apoptosis; the role of the endonuclease tumour necrosis factor alpha induced apoptosis in cardiac myocytes microelectrophoretic study of radiation induced dna damage in mammalian cells the comet assay: a comprehensive review the cell cycle related differences in susceptibility of hl- cells to apoptosis induced by various antitumor agents differentiation between cellular apoptosis and necrosis by the combined use of in situ tailing and nick translation techniques events in apoptosis. acidification is downstream of protease activation and bcl- protection preconditioning protects ischemic rabbit heart by protein kinase c activation preconditioning rabbit cardiomyocytes: role of ph, vacuolar proton atpase, and apoptosis role of translocases in the generation of phosphatidylserine asymmetry different populations of macrophages use either the vitronectin receptor or the phosphatidylserine receptor to recognise and remove apoptotic cells exposure of phosphatidylserine on the surface of lymphocytes triggers specific recognition and removal by macrophages binding of vascular anticoagulant to planar phospholipid bilayers thrombin generation by apoptotic vascular smooth muscle cells studies on cell agony and death: an attempt at classification death of a cell inflammation in the course of early myocardial ischemia apoptosis in the pathogenesis and treatment of disease apoptosis in myocytes in end stage heart failure the ins and outs of programmed cell death during c. elegans development genetic control of programmed cell death in the nematode c. elegans elegans survival gene ced- encodes a functional homologue of the mammalian protooncogene bcl- purification of a -kd protease from apoptosis tumour cells that activates dna fragmentation caspase activation is required for commitment to fas-mediated apoptosis signal transduction by dr , a death domain containing receptor related to tnfr and cd fadd, a novel death domain containing protein interacts with the death domain of fas and initiates apoptosis a death domain containing receptor that mediates apoptosis sequential activation of ice-like and cpp -like proteases during fas-mediated apoptosis the bcl- family of proteins cloning of the chromosome breakpoint of neoplastic b cells with the t( ; ) chromosome translocation constitutive expression of human bcl- modulates nitrogen mustard and campothecin induced apoptosis bcl- oncoprotein blocks chemotherapy induced apoptosis in a human leukemic cell line programmed cell death and expression of the proto-oncogene bcl- in myocytes during post-natal maturation of the heart bh and bh domains of bcl- are required for inhibition of apoptosis and heterodimerisation with bax induction of apoptotic program in cell free extracts: requirements for datp and cvtochrome c molecular ordering of the cell death pathway: bcl- and bcl-xl function upstream of the ced- like apoptotic proteases fas induced activation of the cell death related protease cpp is inhibited by bcl- and by ice family proteases the release of cytochrome c from mitochondria: a primary site for bcl- regulation of apoptosis bcl- functions in an anti-oxidant pathway to prevent apoptosis programmed cell death and bcl- protection in very low oxygen prevention of hypoxia induced cell death by bcl- and bcl -xl bik, a novel death inducing protein shares a distinct structural motif with bcl- family proteins and interacts with viral and cellular survival promoting proteins molecular thanatopsis: a discourse on the bcl- family and cell death regulation of apoptosis by bcl- family proteins and its role in cancer and chemoresistance checkpoints of dueling dimers foil cell death evidence for the rapid onset of apoptosis in medial smooth muscle cells after balloon injury stabilised p facilitates ancuploid clonal divergence in colorectal cancer p and the hypoxia induced apoptosis of cultured neonatal rat cardiac myocytes adenovirus e a represses cardiac gene transcription and reactivates dna synthesis in ventricular cardiomyocytes, via alternate pocket protein and p binding domains adenoviral delivery of e f- directs cell cycle re-entry and p independent apoptosis in post-mitotic adult cardiomyocytes in vitro myocyte apoptosis during acute myocardial infarction in the mouse localizes to hypoxic regions but occurs independently of p the bcl- gene product prevents programmed cell death of ventricular myocytes regulators of apoptosis in the heart: a matter of life or death the discovery of endothelium-derived relaxing factor and its importance in the identification of nitric oxide the role of chemical mediators released by the endothelium in the control of the cardiovascular system nitric oxide synthases and cardiac muscle: autocrine and paracrine influences induction of myocardial nitric oxide synthase by cardiac allograft rejection cellular and molecular cardiovascular disease: apoptosis of cardiac myocytes during cardiac allograft rejection: relation to induction of nitric oxide synthase number induced nitric oxide in cardiac myocytes are blocked by nitric oxide synthase antagonism or transforming growth factor beta in vivo gene transfection of human endothelial cell nitric oxide synthase in cardiomyocytes causes apoptosis like cell death; identification using sendai virus coated particles nitric oxide attenuates vascular smooth muscle activation by interferon gamma mechanisms and genes of cellular suicide vascular smooth muscle cells express a constitutive nf-kb like activity nf-kb is induced in the nuclei of cultured rat aortic smooth muscle cells by stimulation of various growth factors expression of an nf-kb like activity is essential for proliferation of cultured bovine vascular smooth muscle cells activated transcription factor nuclear factor-kb is present in the atherosclerotic lesion expression of nf-kb and ik-b by aortic endothelium in an arterial injury model tumour necrosis factor can induce both apoptotic and necrotic forms of cell lysis evidence of cardiocyte apoptosis in myocardium of dogs with chronic heart failure tumour necrosis factor-( and tumour necrosis factor receptor in the failing human heart apoptosis in congestive heart failure induced by viral myocarditis in mice apoptosis and dna fragmentation in the bulbis cordis of the developing rat heart normal and abnormal consequences of apoptosis in the human heart: from post-natal morphogenesis to paroxysmal arrhvthmias apoptosis in arteries of the neonatal lamb the tunica media in aging rats evidence for apoptosis in human atherogenesis and in a rat vascular injury model contributions of ischaemia and reperfusion to mucosal lesion formation introduction. oxidative stress does reperfusion injury exist in humans stretch induced programmed myocyte cell death forging a path to cell death basic mechanisms in heart failure: the cytokine hypothesis overexpression of manganese superoxide dismutase promotes the survival of tumour cells exposed to interleukin , tumour necrosis factor selected anti-cancer drugs and ionising radiation oxidative stress induced apoptosis prevented by trolox cellular mechanisms of endothelial cell death during the systemic inflammatory response syndrome apoptosis in pressure overload induced heart hypertrophy in the rat apoptosis in target organs of hypertension stretch mediated release of angiotensin ii induces myocyte apoptosis by activating p that enhances the local renin angiotensin system and decreases the bcl- to bax protein ratio in the cell angiotensin ii induces apoptosis of adult ventricular myocytes in vitro evidence of apoptosis in arrhythmogenic right ventricular failure cytokine mrna expression in post-ischaemic/reperfused myocardium serum interleukin- levels become elevated in acute myocardial infarction elevated circulating levels of tumour necrosis factor-alpha in severe chronic heart failure oxidative stress activates extracellular signal-regulated kinases through src and ras in cultured cardiac myocytes of neonatal rats myocardial reperfusion injury: role of oxygen radicals and potential therapy with antioxidants evidence for apoptosis in advanced human atheroma: co-localization with interleukin- -(-converting enzyme apoptosis of human vascular smooth muscle cells derived from normal vessels and coronary atherosclerotic plaques necrosis and apoptosis induced by oxidized low density lipoproteins occurs through two calcium dependant pathways in lymphoblastoid pathways role of oxidized low density lipoprotein in atherogenesis the oxidation hypothesis of atherosclerosis suppression of apoptosis by nitric oxide via inhibition of interleukin-l beta converting enzyme(ice)-like and cysteine protease protein (cpp)- -like proteases l-arginine prevents xanthoma development and inhibits atherosclerosis in ldl receptor knockout mice core pathology lab findings in patients undergoing directional atherectomy for a primary coronary artery stenosis and relationship to subsequent outcome: the caveat study integrins and anoikis disruption of endothelial cell matrix interactions induces apoptosis proliferative activity in peripheral and coronary atherosclerotic plaque among patients undergoing percutaneous revascularization heat shock induces ikappab-alpha and prevents stress-induced endothelial cell apoptosis kinetics of cellular proliferation after arterial injury, l: smooth muscle growth in the absence of endothelium nf-kb is induced in the nuclei of cultured rat aortic smooth muscle cells by stimulation of various growth factors complex causes of fatal myocardial infarction use of human tissue specimens obtained by directional atherectomy to study restenosis histopathology of instent restenosis in patients with peripheral artery disease reduction of restenosis after angioplasty in an atheromatous rabbit model by suicide gene therapy involvement of cd /apo- /fas in cell death after myocardial infarction inhibition of apoptosis induced by ischaemia reperfusion prevents inflammation the cardiac fas (apo- /cd ) receptor/fas ligand system tumour necrosis factor-( contributes to ischaemia-and reperfusion induced endothelial dysfunction in isolated hearts two distinct cytokines released from a human aminoacyl-t-rna synthase pro and antiinflammatory mechanisms in renal reperfusion in mice: modulation by endogenous tumour necrosis factor alpha and interleukin key: cord- - zurc t authors: imre, gergely title: cell death signalling in virus infection date: - - journal: cell signal doi: . /j.cellsig. . sha: doc_id: cord_uid: zurc t apoptosis, necroptosis and pyroptosis represent three major regulated cell death modalities. apoptosis features cell shrinkage, nuclear fragmentation and cytoplasm-blebbing. necroptosis and pyroptosis exhibit osmotic imbalances in the cell accompanied by early membrane ruptures, which morphologically resembles necrosis. importantly, these two lytic cell death forms facilitate the release of damage associated molecular patterns into the extracellular space leading to inflammatory response. whereas, during apoptosis, the membrane integrity is preserved and the apoptotic cell is removed by neighbouring cells ensuring the avoidance of immune-stimulation. viruses comprise a versatile group of intracellular pathogens, which elicit various strategies to infect and to propagate. viruses are recognized by a myriad of pathogen recognition receptors in the human cells, which consequently lead to activation of the immune system and in certain circumstances cell-autonomous cell death. importantly, the long-standing view that a cell death inducing capacity of a virus is equal to its pathogenic potential seems to be only partially valid. the altruistic cell death of an infected cell may serve the whole organism by ultimately curbing the way of virus manufacturing. in fact, several viruses express “anti-cell death” proteins to avoid this viral-defence mechanism. conversely, some viruses hijack cell death pathways to selectively destroy cell populations in order to compromise the immune system of the host. this review discusses the pros and cons of virus induced cell death from the perspective of the host cells and attempts to provide a comprehensive overview of the complex network of cell death signalling in virus infection. to infection ultimately blocks the way of virus propagation and therefore protects the rest of the cell population from an increasing viral burden. the strongest argument for this theory comes from experimental observations on viruses which, in fact, actively evade cell death pathways by encoding viral "anti-death" proteins (in detail see . , . and table .). if, however, the relation of cell death and virus infection is scrutinized at organism level, the answer for the question is more complex, particularly, since viruses represent a broad group of pathogens with different tissue-tropisms and various strategies to infect and to replicate. in general, the interplay of three key factors can determine the final outcome of cell death in virus infection: timing, immunogenic capacity and tissue specificity of the cell destruction. for instance, manifestation of a cell death with delayed kinetics can allow sufficient time for virus replication and therefore might not deliver any benefit for the host, or selective killing of immune cells results in attenuated immune-response, which is described in the course of human immunodeficiency virus (hiv) infection of t lymphocytes [ ] . furthermore, even a virus propagation-limiting cell death is considered detrimental in tissues with limited renewing potential (e.g. central nervous system) or when the function of the tissue is compromised. if we assume that a given virus infection selectively triggers cell death pathways, then it is expected that the cells exhibit specific "sensors" to recognize this particular event. in fact, detection of viruses in mammalian cells is accomplished by the recognition of molecular elements which are decoded as foreign-to-the-organism, collectively termed as pathogen associated molecular patterns (pamps). two distinct mechanisms can be identified in the context of pamps and cell death. first, pamps are recognized by so called pathogen recognition receptors (prrs) [ ] , which in turn induce pathways leading to immune response and -in certain circumstances-indirectly to cell demise [ ] . second, pamps can also initiate cell-autonomous death signalling pathways. in this article, the machinery of pathogen recognition is reviewed in the aspect of cell death signalling and the major regulated cell death modalities, including apoptosis, necroptosis and pyroptosis are discussed in detail, which can emerge during a viral insult. importantly, regulated cell death forms differ in their inflammatory potential, as apoptosis represent an anti-inflammatory cell death form, whereas necroptosis and pyroptosis may render immune stimulation more efficiently. functionally, caspases are divided into three subgroups: inflammatory, apoptotic-initiator and apoptotic-executioner caspases [ ] . initiator caspases are structurally distinguished by their large pro-domain, which enables them to be recruited into high-molecular-weight protein complexes in response to an apoptotic stimulus [ ] . intrinsic and extrinsic pathways represent the two main pathways of apoptosis initiation. intrinsic pathway is triggered by intracellular stress signals such as genotoxic stress, whereas extrinsic pathway is initiated by the ligation of transmembrane receptors of the tumor necrosis factor (tnf) superfamily, commonly termed death receptors (dr). stimulation of the intrinsic pathway leads to conformational changes of pro-apoptotic members of the b cell lymphoma- (bcl- ) protein family: bcl- -associated x protein (bax) and bcl- -antagonistic killer (bak). consequently, bax and bak translocate, oligomerize and form large pores throughout the mitochondrial membrane [ , ] leading to mitochondrial outer membrane permeabilization (momp) and resulting in the release of cytochrome-c [ ] . the assembly of apoptosome in the cytosol, a complex containing cytochrome c and apoptotic protease activating factor- (apaf- ), culminates in the recruitment of caspase- [ ] . in the extrinsic pathway, the ligation of drs leads to the recruitment of initiator caspases- and - in a high molecular weight complex termed death inducing signalling complex (disc). the recruitment and oligomerization of initiator caspases in both pathways enables proximity-driven self-cleavage and activation [ ] . as a consequence, the activation of initiator caspases results in the cleavage and activation of effector caspases (caspase- , - and - ), and finally activated effector caspases selectively process their substrates. the cleavage of them results in a series of events which finally leads to apoptosis [ ] . immune response-driven cell-killing mechanisms cytotoxic t cells (ctl) play crucial role in the adaptive immune response and natural killer (nk) cells exert similar functions in the innate immune response given to viral infection. in vivo studies by employing reporter viruses demonstrate that one single ctl is capable to kill up to infected cells per day [ ] . it is noteworthy that ctls exhibit distinct cell killing capacity in different tissue types. for instance, herpes simplex and vaccinia virus infected monocytes and macrophages are killed by ctls, whereas epithelial cells are less susceptible to ctl mediated cell death mechanisms [ , ] . ctl and also nk cell driven cell killing is dependent on cell-cell contacts and involves the rapid secretion of cytotoxic perforin and granzymes [ ] and the slower death receptor, fas (also: cd )-driven apoptotic pathway (figure . ). granzymes are a family of serine proteases that are produced and stored in lytic granules inside the ctls and nk cells. a well-studied member of this family, granzyme-b can enter the target cells either via newly formed trans-membrane perforin-pores [ ] or by endocytosis. subsequently, granzyme-b induces mitochondrial apoptosis by performing cleavage of the bcl- homology domain- (bh )-only protein, bh interacting domain death agonist (bid), which then leads to bax/bak-mediated momp and the initiation of the caspase- -driven apoptotic pathway [ ] . ctls express fas ligands (fasl) on their surface and the trimerized ligands are capable to bind the death receptor fas, which is present on the cell surface of the target cells. the ligation of the drs results in the recruitment of the adaptor protein, fas-associated protein with death domain (fadd), leading to recruitment of j o u r n a l p r e -p r o o f initiator caspases with a death domain (caspase- /- ). interestingly the sensitivity towards dr-mediated cell killing is also selectively regulated on the host side, since in response to influenza a virus (iav) and dengue virus infection, the upregulation of the elements of fas pathway is observed. furthermore, iav infection leads to downregulation of the antiapoptotic cellular caspase- and fadd-like apoptosis regulator (cflip) [ ] , which if highly expressed can form inactive hetero-dimers with caspase- . surprisingly, the replication of iav can be amplified by overexpression of pro-apoptotic genes and is blocked by the upregulation of anti-apoptotic genes in human lung epithelial cell line [ ] , which indicates a virus-promoting role for apoptosis in response to iav in lung cells. along these lines, the viral protein hepatitis b virus x (hbx) sensitizes host cells towards tnf driven apoptosis by interfering the anti-apoptotic action of cflip [ ] . increasing body of literature supports the relevance of hiv induced apoptosis. in infected individuals, the progressive depletion of cd t lymphocytes is observed over the period of - years, which leads to opportunistic infections and malignancies. several hiv viral proteins exert pro-apoptotic or apoptosis sensitizing properties by increasing the expression of death ligands and drs, including the envelope protein glycoprotein and the transactivator of transcription (tat) (reviewed in details elsewhere: [ ] ). interestingly, the protease of hiv, encoded by the pol gene is capable to directly process caspase- in order to produce a catalytically active caspase leading to bak activation and mitochondrial apoptosis [ , ] . apart from the significant role of immune response-driven cell death, it has been observed that activation of prrs could lead to direct -cell autonomous-apoptosis initiation. the group of rna viruses (also termed: riboviruses) comprises diverse pathogens that possess either double stranded rna (dsrna) (e.g.: rotaviruses) or single stranded rna (e.g.: filoviruses) as genetic material. since the spatial distribution and also the structure of a viral rna (vrna) can be distinct from that of the host rna, thus vrnas fit well in the definition of pamps. for instance, vrnas are often di or tri-phosphorylated and they lack a -methylguanosine cap [ ] . detection of vrnas takes place by different prrs depending on the localization of the vrna fragments. intracellular dsrna fragments which are produced during virus replication are recognized in the cytosol by two caspase recruitment domain (card) containing rna helicases: retinoic acid inducible gene-i (rig-i) and melanoma differentiation-associated gene- (mda- ) [ ] . rig-i activation by vrnas leads to atp-dependent conformational changes and enables tetramer formation [ ] , which via card-card homotypic interactions results in the recruitment of mitochondrial antiviral-signalling (mavs), an adaptor protein associated with the mitochondrial membrane ( figure . ). at this point mavs forms aggregates providing scaffold for multiple binding partners [ ] and can bifurcate into two main pathways leading to activation of transcription factors interferon regulatory factor- (irf- ) and nuclear factor kappa-light-chain-enhancer of activated b cells (nfᴋb) [ ] . in the irf- branch of the rig-i pathway, tnfr associated factor- (traf- ) interacts with mavs, which results in the recruitment of a protein complex comprising traf family member-associated nfkb activator (tank), nfkb essential modulator (nemo), inhibitor of kappa b kinase-ε (ikkɛ) and tank binding kinase- (tbk ) and subsequent phosphorylation, dimerization and translocation to the nucleus of both irf- and irf- [ ] (figure .). induction of irf- primarily leads to the production of immunomodulatory molecules as type i and iii interferons (ifns), which then renders the activation of ctls, nk cells and initiates targeted cell-killing as discussed earlier ( . ). nonetheless, this process requires the external intervention of immune cells. in contrast, studies on cell lines highlight the existence of direct death-inducing mechanisms. as for instance, iav infected human epithelial cell lines undergo caspase-dependent cell death, which is prevented by genetic depletion of rig-i [ ] . in sendai virus infection the lack of irf- leads to the loss of virus induced apoptosis sensitivity of the cells [ ] . analogously, semliki forest virus infection leads to mavs induction and caspase- activation [ ] . rig-i signalling leads to the expression of ifn-stimulated genes (isgs) and this induces the intrinsic apoptosis pathway [ ] . how exactly this apoptotic process is mediated is still poorly understood, however some studies provide hints about the possible mechanisms concerned, as for example the upregulation or activation of certain pro-apoptotic elements in a rig-i dependent manner. naturally occurring vaccinia virus expresses a bcl- analogue antiapoptotic protein called f l. in contrast, infection with a genetically engineered vaccinia virus strain lacking the f l triggers irf- activation and expression of the pro-apoptotic noxa (from latin: danger) in monocytes and macrophages [ ] . one work highlights the transcriptionally-independent function of irf- in inducing the mitochondrial translocation of pro-apoptotic bax and the initiation of intrinsic apoptosis [ ] , yet the transcriptional function of irf- remains the predominant feature in majority of the studies. along with this, irf- dependent degradation of x-linked inhibitor of apoptosis (xiap) delivers a potential link to virus induced apoptosis. xiap is a potent antagonist of caspase- , - and - dependent apoptotic signalling. sendai and vesicular stomatitis virus (vsv) infection results in tbk /ikkε-mediated phosphorylation of xiap in vivo at ser , leading to autoubiquitination and proteasomal degradation of xiap. given the multiple (~ ) binding partners of aggregated mavs [ ] and a card domain structure highly similar to that of card of caspases it can be speculated that mavs in certain conditions might recruit caspases with a card domain (e.g.: caspase- and - ), however this has not been investigated yet. extracellular vrna detection is accomplished by toll-like receptor- (tlr ), located in cytoplasm membranes and endosomal compartments [ ] . tlr recruits ikk via the adaptor protein toll/interleukin- receptor (tir) and leads to the activation of transcription factor nfkb [ ] and results in the induction of irf- . in addition, tlr stimulation by iav can lead to the assembly of a disc-like complex containing trif, ripk , fadd, cflip and caspase- ( figure .) [ , ] . in this setting the concentration of anti-apoptotic cflip isoforms determine, whether caspase- acts as an apoptotic initiator, since the formation of cflip/caspase- heterodimers works against the apoptotic homodimer formation of caspase- . viral dna in the cytosol triggers also alarm signals, since the dna in its normal condition resides exclusively in the cell nucleus. the cyclic gmp-amp (cgamp) synthase (cgas) exerts dna binding activity, and upon binding produces cgamp leading to the activation of protein stimulator of interferon gene (sting) (figure .) . as a consequence, sting induces irf- activation [ ] [ ] and intrinsic apoptosis [ ] . it is expected that the release of mitochondrial dna (mtdna) into the cytosol have similar consequences and can unleash cgas driven irf- activation. intriguingly, apoptotic caspases suppress the sting mediated interferon production [ ] and studies on mice demonstrate that apoptotic caspases can cleave anti-viral sensors, including cgas, mavs and irf- in order to reduce type i ifn production, indicating that caspase-dependent apoptosis might also contribute to virus propagation [ ] , but also suggests a negative feedback loop in controlling ifn production. the most of the pamp detection mechanisms which lead to apoptosis culminate in irf- activation highlighting the significance of irf- in virus-induced apoptosis. still, the mechanism, by which irf- triggers cell death signalling pathways is only partially understood and the studies indicate a strong cell type specificity in the apoptosis sensitivity in response to viral pamps z-rna and z-dna fragments, which are distinct from the b-structure of eukaryotic rna and dna are recognized by z-dna/rna binding protein- (zbp ; also: dai). in response to iav infection, zbp interacts with receptor interacting serine/threonine kinase (ripk ) [ ] and the recruitment of ripk can initiate necroptosis (described in chapter ), and can serve as a scaffold for caspase- activation-driven apoptosis [ ] . intriguingly, the iav triggered cell death signalling seems to play a unique role in controlling the viral burden, which is discussed in detail in chapter . . human papillomavirus (hpv) is a double stranded dna virus associated with cervical cancer developments in humans [ ] . the early gene-coding region protein (e ) of hpv interferes with apoptosis signalling at the level of p and interacts with the pro-apoptotic bak protein. this example well demonstrates that most of viral anti-apoptotic proteins are encoded by large dna viruses, which replicate with slower kinetics than rna viruses. nevertheless, unlike hpv, a majority of these viruses produce proteins that counteract with the extrinsic, caspase- dependent apoptotic pathway (table . ). for example, the large subunit of the herpes simplex virus (hsv) ribonucleotide reductase icp exerts apoptosis inhibitory potential by directly interacting with caspase- [ , ] . along the same line, viral flip (vflip), the viral counterpart of cflip encoded by molluscum contagiosum and several gamma herpes viruses, such as kaposi´s sarcoma-associated herpes virus (kshv), forms inactive heterodimers with caspase- , which leads to the blockade of extrinsic apoptosis [ ] . the mouse cytomegalovirus (mcmv) is the experimental model of the human cmv, which causes asymptomatic infections in human, yet leads to severe infections in individuals with compromised immune system. along with the previous instances, the m protein encoded by the mcmv also blocks the caspase- -driven extrinsic apoptosis pathway [ , ] . furthermore, an anti-apoptotic product of the cowpox virus, cytokine response modifier a (crma) functions by the inhibition of the activity of caspase- [ ] . a further group of viral proteins can influence the expression of host anti-apoptotic proteins. the adult t-cell leukemia/lymphoma (atl) is a malignancy of mature cd t cells in humans [ ] and infection with the human t-cell leukemia virus type- (htlv- ) is linked to the disease. the transactivator protein tax of htlv- binds to ikk and thereby facilitates the activation of nfkb and subsequently leads to the upregulation of c-flip expression [ ] , which then exhibits caspase- inhibiting activity. necroptosis is a non-apoptotic regulated cell death modality which morphologically resembles necrosis. in general, necroptosis exhibits cytoplasm swelling and early membrane rupture and facilitates the release of immunomodulatory damage associate molecular patterns (damps), for instance atp and high mobility group protein b (hmgb ), into the extracellular space [ ] [ ] . necroptosis initiation takes place upon tnfr ligation, which, however, primarily leads to nfkb activation via the assembly of so called complex-i, including adaptor proteins tnfrsf a associated via death domain (tradd), traf , cellular iap (ciap) and ubiquitinated receptor interacting serine/threonine kinase (ripk ) [ ] . the state of ripk is a decisive factor, since ubiquitination, phosphorylation at ser by protein kinase-a and -c, and phosphorylation at several sites by ikks can all attenuate the cytotoxic potential of this scaffold protein (figure .) [ , ] . once, however, the nfkb wing of the pathway is inhibited tnfr ligation can lead to a pro-apoptotic fadd and caspase- containing complex formation and apoptosis initiation. in the early ´s it has been observed that inhibition of caspases did not result in expected cell death inhibition, instead it led to a necrotic-like cell demise, which is nowadays termed necroptosis [ ] [ ]. ripk and play central role in the regulation of necroptotic signalling pathways [ ] and fully active caspase- can cleave both of them [ , ] , thus inhibition of caspase- is an absolute prerequisite for the assembly of ripk -ripk containing necrosome complex and necroptosis initiation (figure . ). the assembly of necrosome leads to phosphorylation of pseudo-kinase mixed lineage kinase-like (mlkl) [ ] , in turn the phosphorylation-driven conformational changes of mlkl result in oligomerization and subsequent pore formation throughout the cytoplasm membrane accompanied by osmotic imbalances and membrane ruptures. necroptosis represents a lytic and immunogenic cell death modality, therefore it is less surprising that most of the viruses developed countermeasures to evade it. nevertheless, sensitivity to virus-driven necroptosis varies among cell types and viruses. one prominent example in this aspect is the human immunodeficiency virus- (hiv- ) [ ] . as discussed earlier ( . ), hiv infection facilitates the upregulation of drs (as tnfr and fas) and their ligands, which might also promote necroptosis sensitivity. a recent article demonstrates a hiv induced necrotic-like cell death in t cells, which can be inhibited by necrostatin- , an inhibitor of ripk [ ] this in vitro work is supported by an in vivo study, in which ripk knock down and necroptosis inhibition can restore the proliferation potential of cd t cells from hiv-infected patients [ ] . the nucleic acid sensing receptor zbp , analogously to ripks, possess a ripk homotypic interaction motif (rhim), thus zbp is capable to interact with ripk [ ] . ripk then can initiate mlkl dependent necroptosis and can serve as a platform for caspase- dependent apoptosis (figure .) [ ] , thus only blocking of both pathways at the same time can significantly reduce iav induced cell death. the importance of the zbp /ripk axis as antiviral defence is strongly supported by the studies on zbp and ripk deficient mice, which exhibit higher viral burden and are more susceptible to iav induced lethality [ , ] . importantly, it seems, if only one of the two cell death pathways triggered by zbp /ripk functions, is already sufficient to ensure viral clearance, since only a simultaneous depletion of mlkl and caspase- function can lead to an increased iav-driven increased lethality [ ] . taken together, apoptosis and necroptosis represent compensatory pathways upon iav infection, yet necroptosis, as an immune-stimulatory cell death form renders a stronger adaptive immune stimulation [ ] . based on these studies, a limited activity of necroptosis can be considered as a beneficial measure to limit viral burden, however a drastic increase of cell death in the alveolar epithelial cells, which results in compromised lung function in mice and in humans, tightly correlates with lethality [ ] . to date, iav infection represents the only viral infection which results in zbp- -driven necroptosis in its natural state, since all other necroptosis-inducing virus models employ genetically engineered viruses [ , ] . this well demonstrates that several viruses developed strategies to avoid necroptosis (table . ). genetically engineered mcmv lacking the viral m -encoded inhibitor of ripk (vira) initiates premature cell death in mice and depletion of ripk attenuates this cell death [ ] . the mcmv induced necroptosis takes place independent of ripk and tnf, yet it is triggered by zbp in response to infection [ ] . conversely, both the wild type mcmv encoding vira and the human cmv [ ] can efficiently evade ripk -dependent necroptosis initiation. the latent membrane protein- (lmp ) of epstein-barr virus (ebv) elicits the poly-ubiquitination of ripks and thereby inhibits necroptosis in nasopharyngeal epithelial cells [ ] . rhinoviruses, the causative agents of common cold express the viral c protease, which blocks poly(i:c) stimulated necroptosis in airway epithelial cells by the direct cleavage of ripk [ ] . these virally encoded factors have one in common: they all target ripks in order to block necroptosis. a distinct inhibitory mechanism is provided by the vaccinia virus (vacv) encoded innate immune evasion protein- (e ), which interacts with the putative rna binding domain of zbp and the genetic deletion of e leads to an immediate necroptosis in mouse l and human embryonal kidney cell lines [ ] . novel findings demonstrate a virally encoded mlkl homolog (vmlkl) in poxviruses, which exerts antagonistic effect in necroptosis initiation by binding to ripk and thereby blocking the sequestration of human and mouse mlkl [ ] . the species selectivity of viruses and the continuously ongoing evolutionary race between host and virus is well demonstrated by the following example of hsv- . the viral ribonucleotide reductase icp of hsv- possess a rhim-like domain, thus it can engage both ripk and ripk . interestingly, hsv- infection leads to necroptosis in mice via icp and is accompanied by limited propagation of the virus, whereas icp- inhibits the necroptosis in j o u r n a l p r e -p r o o f humans, in the natural host of the virus. furthermore, icp the large subunit of the ribonucleotide reductase of hsv- can inhibit tnfr triggered necroptosis [ ] . a recent study demonstrates that icp blocks the tnfr stimulated mlkl-necrosome assembly into membrane vesicles [ ] . this compartmentalization step represents a critical event towards necroptosis execution in humans, yet can be dispensable in mouse, which might answer why mouse cells are susceptible to hsv- triggered necroptosis. pyroptosis is a regulated cell death form accompanied by osmotic imbalances and early membrane rupture of the cells [ ] . pyroptosis is conducted by inflammatory caspase activation and culminates in the release of pro-inflammatory cytokines. upon stimulation, the inflammatory caspases, as caspase- , caspase- / in humans and caspase- in mice, are recruited in high molecular weight complexes termed inflammasomes [ ] . inflammasomes are considered as pathogen sensors that recognize pamps and based on the chemical nature of the pamps different inflammasomes are assembled. inflammatory caspases, similar to initiator apoptotic caspases possess a large pro-domain structure, which ensures the recruitment either directly or via an adaptor protein to prrs. in turn, caspase- activation in the inflammasome leads to the processing of pro-interleukin- β (il- β) and pro-il- , whereas activation of caspase- and caspase- (and caspase- ) can result in the cleavage of gasdermin-d (gsdmd), a member of a poorly studied protein family [ ] . consequently, the n-terminal region of gsdmd is inserted into the lipid bilayer of the cytoplasm membrane and forms oligomer pores [ ] leading to release of mature ils into the extracellular space (figure .) [ ] . the pore formation triggered osmotic imbalances and membrane ruptures can be interpreted as a collateral event, which takes place in only special circumstances and the primary event of il release can, in fact, occur without cell destruction, yet not in every circumstances [ ] . indeed, inflammasome-dependent pyroptosis is unleashed in response to various viral infections as follows: hiv, dv [ ] , iav [ ] , coxsacivirus-b (cvb ) [ ] , human bocavirus (hbov ) [ ] , hepatitis c virus [ ] and enterovirus (ev ) [ ] . in some cases the presence of pyroptosis seems to be beneficial for the host by limiting the viral spread [ , ] , however increasing number of studies show detrimental effect of pyroptosis in response to infections (table . ). interestingly, unlike in the case of necroptosis and apoptosis, direct anti-pyroptotic measures of viruses are limited, which lets us to speculate that pyroptosis may be positioned later than apo [ ] ptosis and necroptosis in a putative virus-host co-evolutionary time-line. intriguingly, one exemption is provided by ev . the viral protease c of ev cleaves gsdmd into a non-functional form, which then unable to trigger pyroptosis in thp- macrophages and human embryonal kidney cells [ ] . the nucleotide binding oligomerization domain (nod)-like receptors represent a broad group of inflammasome components, among which nod-like receptor family pyrin domain j o u r n a l p r e -p r o o f containing (nlrp ) is considered as the most studied one. in response to a danger signal, nlrp interacts the adaptor protein apoptosis-associated speck-like protein (asc) via its pyrin domain thereby leading to asc oligomerization and the formation of so called asc specks. importantly, asc exhibits a caspase recruitment card domain, thus it is capable to recruit pro-inflammatory caspase- (figure .) [ ] . the conformational changes in nlrp , which ultimately lead to activation, can be efficiently stimulated by uric acid crystals and extracellular atp, however several studies conclude the unified role of osmotic imbalances and especially potassium efflux in activation [ ] . analogously, vsv and encephalomyocarditis virus (emcv) infections trigger nlrp inflammasome formation, which is initiated by a rapidly ongoing lytic cell death accompanied by potassium efflux [ ] . other studies provide some mechanistic clues, how nlrp is activated. for instance, iav induces nlrp dependent pyroptosis in dendritic cells [ , ] and in respiratory epithelial cells [ ] . the depletion of galectin- , a β-galactosidase binding protein results in a strong inhibition of inflammation in lungs of the infected mice and co-immunoprecipitation experiments demonstrate the interaction of galectin- , nlrp and asc in bone marrow derived macrophages in response to avian iav (h n ) infection [ ] . this study indicates that galectin- can serve as an alternative sensor in the nlrp inflammasome-driven pyroptosis. along with this line, gp , an envelope protein of hiv induces inflammation and neuropathy in neurons. interestingly, an atp-dependent cationic ion channel purinergic x (p x ) has been shown to initiate the gp triggered nlrp inflammasome assembly and caspase- activation [ ] . dv has been also reported to stimulate nlrp inflammasome via the cytoplasmic membrane surface c-type lectin a (clec a) [ ] . instead of stimulating nlrp , dv can also directly stimulate caspase- , the human homologue of capase- and leads to caspase- cleavage driven pyroptosis [ ] , thus it seems caspase- can function as an alternative pamp sensor (figure .). the recognition of viral nucleotide patterns play a crucial role in inflammasome induction. activation of absent in melanoma- (aim ) and interferon-γ-inducible protein (ifi ) are two prominent examples for viral nucleotide pattern recognition (figure .). aim in the cytosol binds dna fragments and recruits the adaptor asc leading to asc polymerization and assembly of the asc specks. interestingly, the upregulation of aim has been observed in brain tissues after zika virus infection, collected from fatal human cases [ ]. this is therefore possible that the associated neuronal damage is linked to the over activation of aim- -dependent pyroptosis in neurons. furthermore, ev [ ] , vaccinia virus [ ] and mcmv [ ] infections have been shown to exert aim -driven pyroptosis. in contrast, hsv- inhibits aim activation by the expression of vp , a virally encoded factor, which interacts with the aim protein and prevents the oligomerization of it. conversely, infection of hsv- lacking vp leads to reduced viral replication in vivo, which effect is restored in aim deficient mice [ ] . the dna sensor interferon-γ-inducible protein (ifi ) recognizes aberrant dna transcripts, which are products of an incomplete virus replication cycle in hiv infected non-j o u r n a l p r e -p r o o f permissive cd t lymphocytes. the activation of ifi launches the caspase- -driven pyroptotic pathway consequently attracting more immune cells to the site of infection and leading to an accelerated viral spread [ ] . a neutrophil specific antimicrobial defence mechanism is represented by the release of neutrophil extracellular traps (nets). nets are released from the cells upon neutrophil cell death termed netosis, and contain chromatin and antimicrobial proteins from the cytoplasm and granules of the cell [ ] . the signalling pathways leading to netosis are distinct from necroptosis [ ] and apoptosis [ ] and has been recently discussed in detail by sollberger et al [ ] . importantly, netosis mostly occur in response to bacterial and fungal infections, yet some viruses, including hiv [ ] , hanta virus [ ] and pox virus [ ] can also trigger this specific cell death form exhibiting anti-viral properties. in contrast, a study on human respiratory syncytial virus suggests that, although nets trap viral particles, their exaggerated formation during severe cases plays a role in airway obstruction [ ] . furthermore, h n influenza virus also triggers netosis and this can similarly contribute to disease pathogenesis [ ] . severe acute respiratory syndrome (sars) coronavirus (cov, also: cov- ) has led to a global outbreak of pneumonia resulting in approximately deaths in [ ] . studies on human and mammalian cell lines demonstrate a sars-cov induced cytopathic effect, which exhibits caspase activation and typical apoptotic morphology [ ] . multiple viral proteins of sars-cov have been shown to contribute to apoptosis initiation, including the c terminal domain of sars-cov spike (s) protein [ ] , and the open reading frame (orf) a and a proteins of the virus. the two orf proteins employ distinct mechanisms to activate cell death, since sars-cov-orf a acts as a potassium channel and triggers caspase activation [ ] , whereas sars-cov-orf a directly interacts with bcl-xl [ ] , a pro-survival member of the bcl- family. it is, however, important to note that the above studies employ african green monkey kidney epithelial (vero e ) cells, and not human cell lines. in contrast, if a human lung epithelial cell line is employed, then sars-cov-orf a binds ripk , which however does not lead to necroptotic mlkl activation, instead it facilitates a non-apoptotic lytic cell death [ ] . besides, sars-cov envelope (e) protein acts as an ion channel and is capable to activate nlrp in an in vitro model [ ] . interestingly, the lack of the ion channel activity of sars-cov-e does not influence virus replication in infected mice, yet reduces edema, the major factor of acute respiratory distress syndrome. the reduced edema correlates with preserved lung epithelia integrity and less il- β in the lung airways, which indicates a significant role for inflammasome activation in disease pathogenesis. [ ] . the role of nlrp inflammasome assembly is further demonstrated in human macrophage cell lines, where the orf b protein of sars-cov directly binds nlrp and leads to pyroptotic cell death [ ] . middle east respiratory syndrome coronavirus (mers-cov) is a highly pathogenic betacoronavirus strain with an approximately % case-fatality rate [ ] . mers-cov infection of primary human t lymphocytes [ ] , and kidney and lung cells from human tissue [ ] leads to apoptotic cell death, which might largely contribute to the high pathogenicity of the virus. novel studies on sars-cov- , the pathogenic agent of the newly emerged coronavirus disease (covid- ) already give some hints on the possible role of cell death in the infection caused tissue injury. interestingly, the expression of orf a of sars-cov- is less cytotoxic than that of sars-cov- , and stimulates less apoptosis and caspase activation in human cell lines [ ] . this observation might explain why sars-cov- exhibits lower casefatality rate than sars-cov- (~ %). conversely, direct sars-cov- infection of organotypic airway epithelial cells results in higher cytopathic effect with apoptotic characteristic than infection with human coronavirus nl , a less pathogenic member of the known human coronaviruses [ ] . interesting novel data supports the evidence of nlrp inflammasome activation in response to sars-cov- -s/angiotensin-converting enzyme- receptor binding in hematopoietic and endothelial cells [ ] . in this regard, the role of pyroptosis has not been investigated in detail yet, but one can speculate that the overactivation of nlrp inflammasome and the consequent pyroptotic cell death might be one of the contributing factors in severe cases of covid- . the ion channel property of sars-cov- -e might represent an important factor in this context. this hypothesis will be very likely investigated in the near future. virus infection triggered cell death is a two edged sword, which might serve the purpose of the host by facilitating viral clearance in tissues where the increased cell destruction is not accompanied by function loss, yet in other cases it might assist viral propagation as the example of hiv infected cd t cells shows. the role of the three cell death modalities can be seen as redundant in the context of viral clearance, for instance in iav infection, however they trigger immune stimulation at different extent. this information can be taken into consideration in designing novel therapies aiming at re-activating cell death pathways in dormant infected cells. furthermore, increasing body of literature demonstrates the crosstalk between the discussed cell death pathways, thus manipulating one or the other cell death pathways may unavoidable influence the whole machinery. along with this, the apoptotic effector caspase- [ ] and the initiator caspase- can play a role in the necroptotic zbp- /ripk axis, while, in other settings, caspase- may facilitate pyroptotic pathways [ ] . these and other studies highlight that our knowledge of cell death pathways is still not complete, but it is growing continuously. the understanding of the fine tuning of these pathways might warrant better therapeutic options in treating viral diseases and preventing epidemics in the future. virus infection leads to immune-stimulation and activation of ctls and nk cells. ctls and nk cells release cytotoxic perforin and granzyme-b. granzyme-b enters the cells via the transmembrane perforin pores and results in bid cleavage, leading to bax and bak activation and momp-driven apoptosome formation and caspase- activation. stimulation of death receptors (fas and tnfr) initiate caspase- activation by the formation of death inducing signalling complex (disc). intracellular viral rna (vrna) is recognized by rig-i, leading to atp-dependent conformational changes and recruitment of mavs oligomers, which in turn lead to the recruitment of tbk- and ikkε, resulting in phosphorylation of irf- . extracellular vrna is detected by tlr leading to irf- activation. vdna in the cytosol is recognised by cgas, resulting in cgamp production and activation of sting, which in turn activates irf- . once irf- is activated can translocate into the nucleus and induce transcriptional pathways leading to anti-viral response and apoptosis. moreover, rig-i activation promotes xiap degradation and activation of pro-apoptotic bax. ligation of tnfr recruits the nfkb activating pro-survival complex, containing ubiquitinated ripk . the formation of complex ii, involving de-ubiquitinated ripk , enables activation of caspase- and apoptosis. inactivation of caspase- leads to the assembly of intact ripk and ripk containing necrosome and phosphorylation of mlkl. vrna and dna binding by zbp leads to ripk recruitment and mlkl activation. vrna detection by tlr results in ripk -ripk necrosome formation and mlkl activation. the activation of mlkl in all three cases leads to oligomerization and transmembrane pore formation resulting in osmotic imbalances and the progression of necroptosis. nlrp can be activated by various insults, including potassium efflux, extracellular atp, and binding of virus associated patterns to host proteins (p x , clec a, galectins). activation of nlrp leads to the assembly of asc specks, resulting in caspase- activation. vdna is engaged by aim resulting in asc speck formation. aberrant dna fragments produced by hiv is recognized by ifi- , which initiates caspase- activation. active caspase- accomplishes the processing of gsdmd and pro-il β/ pro-il . cleaved gsdmd oligomers form pores in the cytoplasm membrane. the mature ils are released to the extracellular space throughout the gsdmd pores and the pore formation culminates in membrane ruptures and osmotic imbalances, ultimately leading to a lytic cell death, called pyroptosis. cell death by pyroptosis drives cd t-cell depletion in hiv- infection prrs are watching you: localization of innate sensing and signaling regulators apoptosis: a basic biological phenomenon with wide-ranging implications in tissue kinetics exposure of phosphatidylserine on the surface of apoptotic lymphocytes triggers specific recognition and removal by macrophages caspases: the executioners of apoptosis molecular cell death platforms and assemblies mitochondria and cell death: outer membrane permeabilization and beyond bax assembles into large ring-like structures remodeling the mitochondrial outer membrane in apoptosis the apoptosome: signalling platform of cell death death receptor signal transducers: nodes of coordination in immune signaling networks yama/cpp β, a mammalian homolog of ced- , is a crma-inhibitable protease that cleaves the death substrate poly(adp-ribose) polymerase vivo killing capacity of cytotoxic t cells is limited and involves dynamic interactions and t cell cooperativity subtle cxcr -dependent chemotaxis of ctls within infected tissue allows efficient target localization cxcr chemokine receptor enables local cd (+) t cell migration for the destruction of virusinfected cells rapid and unidirectional perforin pore delivery at the cytotoxic immune synapse granzyme b-mediated apoptosis proceeds predominantly through a bcl- -inhibitable mitochondrial pathway fasl/fas pathway is involved in dengue virus induced apoptosis of the vascular endothelial cells novel pandemic influenza a (h n ) virus infection modulates apoptotic pathways that impact its replication in a cells pro-apoptotic function of hbv x protein is mediated by interaction with c-flip and enhancement of death-inducing signal mechanisms of human immunodeficiency virus-associated lymphocyte regulated cell death, aids research and human retroviruses hiv- protease processes procaspase to cause mitochondrial release of cytochrome c, caspase cleavage and nuclear fragmentation the hiv- -specific protein casp p induces death of infected cells through bax/bak crosstalk between cytoplasmic rig-i and sting sensing pathways differential roles of mda and rig-i helicases in the recognition of rna viruses rig-i family rna helicases: cytoplasmic sensor for antiviral innate immunity mavs forms functional prion-like aggregates to activate and propagate antiviral innate immune response identification and characterization of mavs, a mitochondrial antiviral signaling protein that activates nf-kappab and irf orchestrating the interferon antiviral response through the mitochondrial antiviral signaling (mavs) adapter annexin-a promotes rig-i-dependent signaling and apoptosis via regulation of the irf -ifnar-stat -ifit pathway in a lung epithelial cells irf- activation by sendai virus infection is required for cellular apoptosis and avoidance of persistence a novel mitochondrial mavs/caspase- platform links rna virus-induced innate antiviral signaling to bax/bak-independent apoptosis the interferon stimulated gene promotes apoptosis induction of noxa-mediated apoptosis by modified vaccinia virus ankara depends on viral recognition by cytosolic helicases, leading to irf- /ifnβ-dependent induction of pro-apoptotic noxa the irf- /bax-mediated apoptotic pathway, activated by viral cytoplasmic rna and dna, inhibits virus replication recognition of double-stranded rna and activation of nf-kappab by toll-like receptor proapoptotic signalling through toll-like receptor- involves trif-dependent activation of caspase- and is under the control of inhibitor of apoptosis proteins in melanoma cells leverkus, ciaps block ripoptosome formation, a rip /caspase- containing intracellular cell death complex differentially regulated by cflip isoforms sting is an endoplasmic reticulum adaptor that facilitates innate immune signalling cyclic gmp-amp synthase is a cytosolic dna sensor that activates the type i interferon pathway viral dna sensors ifi and cyclic gmp-amp synthase possess distinct functions in regulating viral gene expression, immune defenses, and apoptotic responses during herpesvirus infection apoptotic caspases suppress mtdnainduced sting-mediated type i ifn production apoptotic caspases suppress type i interferon production via the cleavage of cgas, mavs, and irf dai/zbp /dlm- complexes with rip to mediate virusinduced programmed necrosis that is targeted by murine cytomegalovirus vira ripk activates parallel pathways of mlkl-driven necroptosis and fadd-mediated apoptosis to protect against influenza a virus a papillomavirus dna from a cervical carcinoma and its prevalence in cancer biopsy samples from different geographic regions the ribonucleotide reductase r subunits of herpes simplex virus types and protect cells against tnfα-and fasl-induced apoptosis by interacting with caspase- the r subunit of herpes simplex virus ribonucleotide reductase protects cells against apoptosis at, or upstream of, caspase- activation viral fliceinhibitory proteins (flips) prevent apoptosis induced by death receptors a cytomegalovirus-encoded inhibitor of apoptosis that suppresses caspase- activation block of death-receptor apoptosis protects mouse cytomegalovirus from macrophages and is a determinant of virulence in immunodeficient hosts target protease specificity of the viral serpin crma. analysis of five caspases human t-cell leukaemia virus type (htlv- ) infectivity and cellular transformation human t-cell leukemia virus type-i oncoprotein tax inhibits fas-mediated apoptosis by inducing cellular flip through activation of nf-kappab chemical inhibitor of nonapoptotic cell death with therapeutic potential for ischemic brain injury necroptosis takes place in human immunodeficiency virus type- (hiv- )-infected cd + t lymphocytes serine phosphorylation inhibits ripk kinasedependent cell death in models of infection and inflammation positive and negative phosphorylation regulates rip -and rip -induced programmed necrosis fas triggers an alternative, caspase- -independent cell death pathway using the kinase rip as effector molecule structural basis of rip inhibition by necrostatins the ripoptosome, a signaling platform that assembles in response to genotoxic stress and loss of iaps caspase inhibits programmed necrosis by processing cyld mixed lineage kinase domain-like protein mediates necrosis signaling downstream of rip kinase dysfunctional hiv-specific cd + t cell proliferation is associated with increased caspase- activity and mediated by necroptosis dai senses influenza a virus genomic rna and activates ripk -dependent cell death ripk and nf-κb signaling in dying cells determines crosspriming of cd + t cells pathogenic potential of interferon αβ in acute influenza infection virus inhibition of rip -dependent necrosis murine cytomegalovirus ie -dependent transcription is required for dai/zbp -mediated necroptosis suppression of rip -dependent necroptosis by human cytomegalovirus inhibition of dai-dependent necroptosis by the z-dna binding domain of the vaccinia virus innate immune evasion protein viral mlkl homologs subvert necroptotic cell death by sequestering cellular ripk herpes simplex virus suppresses necroptosis in human cells herpes simplex virus icp impedes tnf receptor -induced necrosome assembly during compartmentalization to detergent-resistant membrane vesicles caspase- -dependent pore formation during pyroptosis leads to osmotic lysis of infected host macrophages caspases in cell death, inflammation, and disease caspase- cleaves gasdermin d for non-canonical inflammasome signalling gsdmd membrane pore formation constitutes the mechanism of pyroptotic cell death gasdermin d is an executor of pyroptosis and required for interleukin- β secretion the pore-forming protein gasdermin d regulates interleukin- secretion from living macrophages dengue virus-infected human monocytes trigger late activation of caspase- , which mediates pro-inflammatory il- β secretion and pyroptosis zbp /dai is an innate sensor of influenza virus triggering the nlrp inflammasome and programmed cell death pathways cathepsin b aggravates coxsackievirus b -induced myocarditis through activating the inflammasome and promoting pyroptosis human parvovirus infection of human airway epithelia induces pyroptotic cell death by inhibiting apoptosis hepatitis c virus infection of cultured human hepatoma cells causes apoptosis and pyroptosis in both infected and bystander cells pyroptosis induced by enterovirus a infection in cultured human neuroblastoma cells caspase- protects against bacteria that escape the vacuole impact of caspase- / , - , - , or il- β/il- deficiency on rabies virusinduced macrophage cell death and onset of disease enterovirus inhibits pyroptosis through cleavage of gasdermin d detection of immune danger signals by nalp gout-associated uric acid crystals activate the nalp inflammasome rna viruses promote activation of the nlrp inflammasome through cytopathogenic effect-induced potassium efflux ion efflux and influenza infection trigger nlrp inflammasome signaling in human dendritic cells influenza a virus infection triggers pyroptosis and apoptosis of respiratory epithelial cells through the type i interferon signaling pathway in a mutually exclusive manner galectin- enhances avian h n influenza a virus-induced pulmonary inflammation by promoting nlrp inflammasome activation involvement of p x receptor in gp -induced pyroptosis in dorsal root ganglia clec a is critical for dengue virus-induced inflammasome activation in human macrophages involvement of caspase- in il- beta production and pyroptosis in human macrophages during dengue virus infection in situ inflammasome activation results in severe damage to the central nervous system in fatal zika virus microcephaly cases aim inflammasome-mediated pyroptosis in enterovirus a -infected neuronal cells restricts viral replication the aim inflammasome is essential for host defense against cytosolic bacteria and dna viruses aim recognizes cytosolic dsdna and forms a caspase- -activating inflammasome with asc herpes simplex virus vp inhibits aim -dependent inflammasome activation to enable efficient viral replication ifi dna sensor is required for death of lymphoid cd t cells abortively infected with hiv neutrophil extracellular traps: the biology of chromatin externalization diverse stimuli engage different neutrophil extracellular trap pathways neutrophil extracellular trap cell death requires both autophagy and superoxide generation neutrophil extracellular traps mediate a host defense response to human immunodeficiency virus- schönrich, β integrin mediates hantavirus-induced release of neutrophil extracellular traps neutrophils recruited to sites of infection protect from virus challenge by releasing neutrophil extracellular traps neutrophil extracellular traps cause airway obstruction during respiratory syncytial virus disease host dna released by netosis in neutrophils exposed to seasonal h n and highly pathogenic h n influenza viruses identification of a novel coronavirus in patients with severe acute respiratory syndrome severe acute respiratory syndrome coronavirus triggers apoptosis via protein kinase r but is resistant to its antiviral activity adenovirus-mediated expression of the c-terminal domain of sars-cov spike protein is sufficient to induce apoptosis in vero e cells the ion channel activity of the sars-coronavirus a protein is linked to its pro-apoptotic function induction of apoptosis by the severe acute respiratory syndrome coronavirus a protein is dependent on its interaction with the bcl-xl protein severe acute respiratory syndrome coronavirus e protein transports calcium ions and activates the nlrp inflammasome severe acute respiratory syndrome coronavirus envelope protein ion channel activity promotes virus fitness and pathogenesis sars-coronavirus open reading frame- b triggers intracellular stress pathways and activates nlrp inflammasomes isolation of a novel coronavirus from a man with pneumonia in saudi arabia middle east respiratory syndrome coronavirus efficiently infects human primary t lymphocytes and activates the extrinsic and intrinsic apoptosis pathways mers coronavirus induces apoptosis in kidney and lung by upregulating smad and fgf the orf a protein of sars-cov- induces apoptosis in cells morphogenesis and cytopathic effect of sars-cov- infection in human airway epithelial cells sars-cov- entry receptor ace is expressed on very small cd (-) precursors of hematopoietic and endothelial cells and in response to virus spike protein activates the nlrp inflammasome caspase- is a key regulator of innate immunity, inflammasome activation, and host defense caspase- is the molecular switch for apoptosis, necroptosis and pyroptosis key: cord- - didq dk authors: sun, ya-jun; feng, yi-jin; chen, jing; li, bo; luo, zhong-cheng; wang, pei-xi title: clinical features of fatalities in patients with covid- date: - - journal: disaster medicine and public health preparedness doi: . /dmp. . sha: doc_id: cord_uid: didq dk objectives: the novel coronavirus disease (covid- ) pandemic has spread to over countries and territories. we sought to describe the clinical features of fatalities in patients with severe covid- . methods: we conducted an internet-based retrospective cohort study through retrieving the clinical information of covid- deaths from nonduplicating incidental reports in chinese provincial and other governmental websites between january and march , . results: approximately of covid- deaths were males ( . %). the average age was . ± . y, and % of patients were elderly (over age y). the mean duration from admission to diagnosis was . ± . d (median: d). the mean duration from diagnosis to death was . ± . d (median: d). approximately of cases ( . %) were complicated by or more chronic diseases, including hypertension ( . %), diabetes ( . %) and coronary heart disease ( . %), respiratory disorders ( . %), and cerebrovascular disease ( . %). fever ( . %), cough ( . %), and shortness of breath ( . %) were the most common first symptoms. multiple organ failure ( . %), circulatory failure ( . %), and respiratory failure ( . %) are the top direct causes of death. conclusions: covid- deaths are mainly elderly and patients with chronic diseases especially cardiovascular disorders and diabetes. multiple organ failure is the most common direct cause of death. i n december , several cases of pneumonia of unknown cause were reported in wuhan, china that were later recognized as a novel coronavirus infection, named coronavirus disease (covid- ) by the world health organization (who). has been included in the laws of the people's republic of china in the prevention and treatment of infectious diseases as a class b infectious disease. all provinces and cities in china have taken first-level public health emergency responses to contain the transmission of the disease and protect vulnerable populations. the epidemic has spread across china as well as into countries and territories. the covid- 's socioeconomic impacts have already far exceeded those of severe acute respiratory syndrome (sars) and the middle east respiratory syndrome (mers), and the pandemic has become a worldwide major public health concern. as of june , , the global number of confirmed cases of covid- exceeded million, leading to over thousand of fatalities. we attempted to describe the clinical characteristics of fatalities in patients with covid- , which may inform the clinical management of patients with severe covid- . this was an internet-based data intelligence study. we constituted a cohort of covid- deaths through retrieving the clinical information on covid- fatalities from nonduplicating incidental reports in chinese provincial and metropolitan city health commission and other governmental official websites between january and march , . the reported clinical characteristics included the patient's age, sex, initial onset symptoms, pre-existing chronic diseases, direct cause of death, date of admission, date of diagnosis, and date of death. the study cohort included cases of covid- fatalities. the study was approved by the research ethics committee of henan university. informed consent was waived, because the study was based on publicly available anonymized incidental fatality reports. spss (version . ) software was used for statistical analysis. mean ± standard deviations (sd) and median (inter-quartile range) were presented for continuous variables, while frequency and percentage were presented for categorical variables. patients were not involved in the study, which is based on anonymized incidental covid- fatality reports from governmental websites. approximately of covid- fatalities ( . %) were males ( table ). the average age was . ± . y (median: . y), and approximately of patients ( . %) were over y of age. the mean duration from admission to diagnosis was . ± . d (median: ). the average duration from the diagnosis to death was . ± . d (median: d). approximately of fatal covid- cases ( . %) had or more pre-existing chronic illnesses. the prevalence rates were . % for hypertension, % for diabetes, . % for coronary heart disease, % for respiratory disorders, % for cerebrovascular disease, % for cancers, % for abnormal renal function, and . % for parkinson's disease. approximately half of patients ( %) had or more chronic diseases. fever ( . %), cough ( . %), shortness of breath ( . %), fatigue and weakness ( . %), sputum ( . %), and dyspnea ( . %) were the common symptoms at onset, whereas palpitations and diarrhea were less frequent. among the covid- fatalities, cases were missing data on direct cause of death. of the covid- cases with known direct cause of death, the top common direct causes of death were multiple organ failure ( . %), circulatory failure ( . %), and respiratory failure ( . %), and were similar (p = . ) for males and females: multiple organ failure ( . % vs . %, respectively), circulatory failure ( . % vs . %, respectively), and respiratory failure ( . % vs . %, respectively). in this internet-based data intelligence study, we observed that the majority of covid- deaths were elderly (approximately of ) and males ( of ), and most fatalities ( of ) occurred in patients with chronic illnesses. the findings were consistent with a recent report in a hospital-based study, and with the who report on covid- in china, and demonstrate the usefulness of an internet-based data intelligence study. previous studies have not clarified the direct causes of death. our data indicate that the most common direct cause of death is multiple organ failure (approximately of ). the initial onset symptoms are not so much saliently worrisome, but the median duration from diagnosis to death was only d, indicating that the disease can worsen rapidly, costing life. the function of innate immunity and neutrophil function may degrade with aging, exposing the elderly to the more deleterious impact of the new coronavirus infection. similar to sars and mers, covid- presents a clear male sex bias. compared with males, the immune response in females may be more vigorous with higher antibody levels following exposure to an infectious agent; thus, females may be less vulnerable to the deleterious consequence of covid- infection. it has been speculated that women's lower susceptibility to viral infections may be related to genetic factors associated with the x chromosome and sex hormones. another possible explanation for the higher incidence and more male covid- fatalities may be due to that males are likely to spend more time outdoors, increasing the chances of exposure to the virus. fatalities occurred in patients with chronic illnesses. , the top were hypertension ( . %), diabetes ( . %), coronary heart disease ( . %), and respiratory disorders ( . %). previous studies indicate that covid- shares the same receptor with sars-cov, and the angiotensin-converting enzyme- (ace ) sensitive cell surface receptors mediate the entry of the virus into the target cells. ace , the functional receptor of sars-cov, is expressed in the islet, through which the virus may invade and destroy the pancreatic islet cells, thus may aggravate diabetes and accelerate the disease progression. the immune system plays a crucial role when the body is confronted with viruses or bacteria. for patients with diabetes, especially those with poor blood glucose control, long-term exposure to hyperglycemia may lead to circulatory failure ( . ) respiratory failure ( . ) a a total of cases were missing data on the time from diagnosis to death. b a total of cases were missing data on direct cause of death. disaster medicine and public health preparedness decreased immune function. other chronic illnesses may also compromise the patient's immune defense system leading to severe consequences. multiple organ failure, respiratory failure, and circulatory failure were the main direct causes of deaths. similar to mers-cov, multiple organ failure appears to be a common direct cause of death in covid- fatalities. the covid- infection may lead to increased blood capillary permeability of the lungs, aggravating inflammation and apoptosis, with lung injuries leading to respiratory distress syndrome. the virus may set off an immune inflammatory response storm, causing tissue damages in multiple organs leading to multiple organ failure. this study has some limitations. first, we did not have the laboratory data in this internet reports-based study. the reported clinical characteristics are relatively limited in internet reports. it is unclear whether there is a selection bias in internet reports of covid- fatalities compared with those fatalities in the general population. however, our data on age and sex distributions of covid- deaths are consistent with the recent report on deaths in a single large hospital-based study in wuhan, china. in conclusion, covid- deaths are mainly elderly and patients with chronic diseases, especially cardiovascular disorders and diabetes. multiple organ failure is the most common direct cause of death. our findings may inform clinical healthcare professionals in better management of severe covid- patients in fighting the emerging pandemic. clinical features of patients infected with novel coronavirus in wuhan covid- coronavirus pandemic clinical characteristics of deceased patients with coronavirus disease : retrospective study who. report of the who-china joint mission on coronavirus disease sex-based differences in susceptibility to severe acute respiratory syndrome coronavirus infection sexual dimorphism in innate immunity case-fatality rate and characteristics of patients dying in relation to covid- in italy genomic characterisation and epidemiology of novel coronavirus: implications for virus origins and receptor binding middle east respiratory syndrome pulmonary vascular endothelialitis, thrombosis, and angiogenesis in covid- we acknowledge zhong-xiang wang (school of nursing and health, zhengzhou university) for his helpful suggestions and spiritual support. all authors contributed to the development of the conceptual framework of this study. p.x.w. and z.c.l. initiated the study, supervised the collection of research data. y.j.s., y.j.f., and c.j. collected the data. b.l. contributed to data interpretation. y.j.s. analyzed the data and drafted the manuscript. all authors contributed to critical revisions of the study, and approved the final version for publication. this work was supported by research grants from the foshan covid- emergency technology project ( ) and the canadian institutes of health research (cihr grant # ). the authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. the study was approved by the research ethics committee of henan university (husom - ), and informed consent was waivered because the study was based on publicly available anonymized incidental fatality reports. the study data are available from the corresponding author upon reasonable request. key: cord- -vfvnjmv authors: carpenito, l.; d'ercole, m.; porta, f.; di blasi, e.; doi, p.; fagara, g. redolfi; rey, r.; bulfamante, g. title: the autopsy at the time of sars-cov- : protocol and lessons date: - - journal: ann diagn pathol doi: . /j.anndiagpath. . sha: doc_id: cord_uid: vfvnjmv a new viral disease named covid- has recently turned into a pandemic. compared to a common viral pneumonia it may evolve in an atypical way, causing the rapid death of the patient. for over two centuries, autopsy has been recognized as a fundamental diagnostic technique, particularly for new or little-known diseases. to date, it is often considered obsolete giving the inadequacy to provide samples of a quality appropriate to the sophisticated diagnostic techniques available today. this is probably one of the reasons why during this pandemic autopsies were often requested only in few cases, late and discouraged, if not prohibited, by more than one nation. this is in contrast with our firm conviction: to understand the unknown we must look at it directly and with our own eyes. this has led us to implement an autopsy procedure that allows the beginning of the autopsy shortly after death (within – h) and its rapid execution, also including sampling for ultrastructural and molecular investigations. in our experience, the tissue sample collected for diagnosis and research were of quality similar to biopsy or surgical resections. this procedure was performed ensuring staff and environmental safety. we want to propose our experience, our main qualitative results and a few general considerations, hoping that they can be an incentive to use autopsy with a new procedure adjusted to match the diagnostic challenges of the third millennium. the autopsy has played an essential role in the classification and definition of the etiology and pathogenesis of diseases for over two centuries, since the pivotal work "de sedibus et causis morborum per anatomen indagatis" of g.b. morgagni (venezia, ). this role is still relevant, given the periodic appearance of new nosological entities and the resulting need to understand their characteristics, consequences on the human body and possible pharmacological treatment. the world health organization (who) reports that new viral diseases and consequent epidemics will continue to appear with relevant consequences for public health. when a new infectious and diffusive disease appears the autopsy provides, other than the usual clinical and scientific data, crucial epidemiological information useful for the proper planning of hygiene and public health programs and contributes also to the correct address of health care expenditure. the autopsy ascertains the cause of death both in hospitalized patients and in people who died without medical assistance, transported to hospital or in a morgue. last, but not least, the autopsy is a valid aid for training and education of clinicians themselves [ ] [ ] [ ] . in recent months we have witnessed the emergency of a new global threat, the sars-cov- , which produced a new disease that quickly turned into a pandemic. its epidemiology, pathogenesis and, therefore, therapeutic possibilities are still little known [ ] . in the current pandemic scenario of sars-cov- , the autopsy appears to be a crucial tool to clarify the virus target cells in human, the frameworks of organ damage and the biological mechanisms that lead to death or allow the patient to heal. the autopsy execution on a patient who died of sars-cov- meets two conflicting needs. first, the high infectivity and dangerousness of the virus that requires the adoption of rigorous but timeconsuming methods to guarantee the safety of the personnel carrying out the investigation and to prevent the spread of the virus outside the autopsy room. on the other hand, the need to carry out the autopsy as soon as possible after death and to perform it quickly, in order to have as little tissue damage as possible from post-mortem degenerative phenomena [ ] . the quality of the samples is j o u r n a l p r e -p r o o f essential for diagnostic and research activities, necessary to improve the standard of health care [ ] [ ] [ ] . the practical challenge with the sars-cov- emergency led us to significantly change our operational autopsy protocol, to obtain qualitative technical results that go beyond the limits of this disease and that can be useful in a much wider range of situations. this article reports our modus operandi and the main qualitative results obtained and discusses the findings and the horizons of autopsy in the third millennium. we performed ten autopsies on sars-cov- positive patients. the major aims that guided us are the following: . minimize the risks for the personnel who performed the autopsy; . reduce the probability of spreading the virus into the environment through leakage of unfiltered air from the anatomical and surrounding rooms or through blood or other biological liquids; . obtain samples for diagnosis and research of quality similar to biopsy or surgical samples; . make a concrete operative contribution for doctors who followed the patients in the hospitalization wards; . give precise informations on causes of death to the relatives of the deceased. the regulatory references framework in which we developed and applied our protocol was issued by the italian government's ministry of health and by the governor of the lombardy region [ ] [ ] . our hospital is equipped with a complex unit (cu) of infectious diseases. the autopsy room and the arrival/sampling/inclusion areas of the biological samples of the c.u. of pathological anatomy are designed with a safety level three (bsl , according to cdc) [ ] . the patient's cardiac death is immediately confirmed by continuous electrocardiographic monitoring that verifies the costant absence of cardiac activity for no less than minutes [ ] . this procedure is almost never done in italy; it is preferred to wait at least hours after the presumed death when the first putrefactive skin spots appear. if the patient died outside the hospital but was still transported to the hospital emergency room, the epa protocol is applied only if it has passed less than one hour since the presumed death. our protocol sets that an epa, complete of samples for electron microscopy and molecular investigations on rna and proteins, must start within three hours from the presumed patient's death. in our experience the effects of postmortem processes that arise after this time prevent a sufficient diagnostic quality for these types of examinations. when it is not possible to do an epa, the autopsy is still carried out but only routine histological, immunohistochemical and molecular investigations on dna are performed. after the instrumental ascertainment of death, the patient is transferred to the autopsy room. while authorization procedures for the autopsy are in progress, the doctor in charge of the deceased patient and the director of pathological anatomy discuss the clinical aspects of the case and any question j o u r n a l p r e -p r o o f that the autopsy must answer. the goal of this type of autopsy is not only the recognition of the cause of death or of pathologies already described in medical literature, but also the understanding of the etiopathogenetic mechanisms of this new disease and the confirmation of the adequacy of the therapies in use. this is made possible by the correlation between organ and cellular pathological findings and clinical aspects (symptoms, signs, laboratory and instrumental data). the epa is a procedure that needs to be meticulously planned before it begins, particularly if it is performed on a patient with a potential high risk of infectivity. nothing should be left to chance and, once the staff has entered the autopsy room, there must be no unforeseen incidents. the staff that performed the autopsy must be trained and tightly-knit to be able to make decisions quickly and to adapt their work to the needs of each specific case. in addition, the organs that need sampling also for special investigations must be planned first since these procedures lengthen the autopsy time. according to the rules of the cdc, "autopsies on decedents known or suspected to be covid- cases should be conducted in airborne infection isolation rooms (aiirs)". these types of rooms must be "at a negative pressure to surrounding areas", with a "a minimum of air changes per hour (ach) for existing structures and ach for renovated or new structures" and "have air exhausted directly outside or through a high efficiency particulate aerosol (hepa) filter" [ ] . the who's guidelines firmly suggest biosafety level (bsl ) for autopsies perdormed on patients died for sars-cov- [ ] [ ] . an autopsy room of bsl must be organized as a surgical room: it must have two differentiated paths, one for entrance ("dirty" one) and one for exit ("clean" one). it must be equipped with adequate instruments and clothing checked and guaranteed. in addition, effective cleaning and sanitization of both the instruments, the sector tables and the areas used must be constantly planned. personnel working in the room must have adequate personal protective equipment (ppe) [ ] and tools that can guarantee a fast surgical act. a modern autopsy cannot be performed with old and blunt instruments. in the room, during the autopsy, there must be abundant availability of alcohol at - volumes to sanitize instruments and surfaces [ ] , % buffered formalin, five containers for formalin fixed samples (three medium-size for organs of the right hemibody, left hemibody and median axis, two big-size for brain and heart taken in full), four numbered containers with glutaraldehyde for electron microscopy samples, numbered containers with % alcohol solution for samples for extraction of dna, numbered containers with rnalater solution for samples for extraction of rna and proteins and a transport bag to convey the containers with the samples outside the autopsy room. any type of freezing in isopentane / nitrogen liquid must be performed inside the autopsy room only if a dedicated freezer at - ° c / - ° c is not available. cell culture should be avoided in the absence of biohazard hoods for high biological risk. the autopsy staff must be composed of at least two subjects, maximum three (table ) : the first operator (the "dirty" one), the second operator (the "clean" one) and, possibly, a technician. in our case the dirty operator has always been a doctor specialized in autopsies. the technician must not replace the first operator in the evisceration and sampling of the organs because these maneuvers are essential for the correct medical diagnosis: the diagnosis is macroscopic and inspective (look, touch, smell), before histological or molecular. the body of the patient must be positioned completely inside the sector table to prevent blood or biological liquids from leaking onto the floor. in particular, it must be taken care that the head is not at the edge of the table (often overdrawn and not aspirated) to prevent the dripping of blood from the incision of the scalp, thus producing small drops that splash in the environment. to minimize the dispersion of blood and biological fluids, it is essential to always operate in the area of the autopsy table: the viscera removed from the body must be placed either on the iron section table, placed above the patient's thighs, or in a large tray with high steel edges resting on the patient's legs, during weighing, macroscopic examination and sampling of the viscera. the autopsy begins with a careful external inspection of the body. the length of the body and its state of nutrition must be detected, as the condition of the skin; any skin lesions and, in particular, those likely to be related to sars-cov- infection, should be sampled [ ] . the scalp incision is made with a single cut from one mastoid process to the other, passing through vertex. the skull opening must be immediately suspended if there is a smell of bone dust in the environment: this means that the suction is not effective and that the risk of environmental contamination is unacceptable. the evisceration of the brain, cerebellum and brainstem has to maintain their anatomical continuity. the block is weighed and samples are immediately taken for electron microscopy and for molecular investigations. after sampling for special surveys ( fig. .a-c) the brain is immediately placed suspended inside the container with formalin, to prevent it from deforming by touching the bottom ( fig. .d-f) . if there are no specific indications, the structures of the inner ear, located inside the petrous rock of the temporal bone, and the eyes are not removed, as the dura mater corresponding to the base and cranial vault. the "y" cut is preferred by us to the longitudinal one conducted from the chin to the pubis for the examination of the viscera of neck and trunk. the examination and evisceration of neck and trunk organs is performed with a mixture of techniques, also on the basis of the specific clinical questions of each individual case ( table ) . after the brain examination, heart and lungs are the organs most frequently evaluated and extensively sampled. a swab for molecular detection (pcr) of sars-cov- is carried out, inserting the swab in large intraparenchymal bronchi or directly into the lung parenchyma. when there is clinical evidence of cardiac arhythmias or sudden death, the heart is opened, dissected only in its lower third and fully placed in formalin to allow the study of its conduction system [ ] . cavities or blood vessels must be inspected for thrombus or clots that must be taken and measured. the next step is the evisceration and examination of liver and gallbladder in a single block; the liver parenchyma should be examined macroscopically with particular attention to the characteristics of the blood vessels contained in it. then spleen, kidneys and adrenal glands are individually eviscerated, mobilizing the segments of the gastrointestinal tract that cover them, incising the retroperitoneal soft tissues that surround some of them and dissecting the hilar structures. these organs should also be examined macroscopically, measured and weighed. the epiploon cavity is opened and the pancreas is inspected: in the absence of focal lesions a full-thickness section of the body, approximately cm in length, is removed, internally examined and sampled. in males, gonads are removed by herniating them in the abdominal cavity through the inguinal canal by traction on the spermatic cord. it is advisable to always carry out samples for electron microscopy and molecular investigations of brain, heart (right and left ventricle, interventricular septum and emerging tract of the aorta and pulmonary artery), lungs, liver (right and left lobe), spleen, kidneys, skeletal muscles ( - different muscles), blood and bone marrow (from the first-second rib). all of this in addition to samples for histology and immunohistochemistry. if possible, examine also gonads and pancreas. once the j o u r n a l p r e -p r o o f samples on the main parenchymatous organs have been obtained, the autopsy is completed by gutting in a single block the tongue in continuity with the viscera of the neck and the posterior mediastinum. also these viscera must be carefully examined macroscopically, opening the hollow ones and sampling them for histological and immunohistochemical examinations. two obstructive ligatures are performed, immediately after the treitz ligament and about cm from the ileocecal valve, to remove the small intestine by cutting it at the base of the mesentery; the intestine must be manually inspected along its entire length and, in the absence of focal lesions (which must be sampled for histological examination), two-three sections of a few centimeters are removed and fixed in formalin; in order to avoid the dispersion of feces, the section of the intestinal segments must be carried out by cutting the bowel between two adjacent ligatures for each section point. the large intestine is then mobilized by detaching it from the abdominal wall but without removing it. the course of ureters, bladder and, in females, uterus are inspected. if there are no focal lesions, they can be sampled in situ for histological and immunohistochemical analysis, but it is advisable not to open the bladder to avoid the spread of urine. if prostate samples must be performed, the bladder has to be opened and emptied by aspirating the content. after these procedures, the abdominal tract of the aorta and the large retroperitoneal vessels are inspected. finally, the stomach is opened in situ to examine, and possibly sample, the mucous surface. in our experience the rigorous application of the protocol described for the execution of complete autopsies did not produce accidents or negative consequences on the staff. to date, there have not been any case of sars-cov- infection in the staff assigned to the autopsy activity (detected with real-time pcr on nasopharingeal swab, serological tests for specific antibodies for sars-cov- and clinical evaluation). the instrumental assessment of cardiac death in patients allowed the rapid execution of the autopsy and the collection of samples for histopathological, ultrastructural and biomolecular tests. we thus obtained visceral samples of the highest quality, comparable to those of biopsy and surgical resections ( fig. , .a-c). the comparison with the autopsy samples collected after or more hours clearly demonstrates how the latter are affected by marked post-mortem artifacts, which can distort the correct interpretation of the morphological pictures observed. in addition, immunohistochemical tests on autopsy samples taken within - hours from death are more reliable, which is particularly important when these tests are used to describe a new pathologic entity such as covid- ( fig. ) . immunohistochemistry, like biomolecular investigations performed on autopsy tissue samples, allows also to detect the presence of the sars-cov- in the body, detailing its distribution in single cell types ( fig. .d) . the rapid collection of samples for ultrastructural examinations gives the possibility to define, with a high degree of certainty, the damaging action of the pathology in progress differentiating it from post-mortem tissue and cellular changes ( fig. ). general guidance on how to perform an autopsy on a patient suspected or infected with sars-cov- have been recently published [ ] ; some autopsy studies on patients with this disease reported the observed histopathological patterns, in particular affecting the lung [ ] [ ] [ ] [ ] . in light of these indications, it seems useful to underline some practical aspects concerning the execution of j o u r n a l p r e -p r o o f autopsies in the countries where this virus is still widely spread and often undiagnosed. it is essential to consider as "potentially positive" also the patients not diagnosed with sars-cov- , because maybe asymptomatic or not included in health surveillance programmes: autopsy on these subjects must be performed with the protocol described for patients definitely infected. all autopsies should always be considered "potentially at risk of contagiousness". in each hospital a specific operating protocol that allows to quickly ascertain the patient's death must be planned out so that the autopsy can be performed rapidly without waiting for the appearance of explicit putrefactive phenomena, which is the normal procedure in italy [ ] . the study of tissues markedly modified by postmortem biological phenomena can allow us to contemplate death but not the causes and the mechanisms that produced it. in our experience the histopathological and ultrastructural frameworks that emerge from an autopsy performed within - hours after death differ significantly from those that we observed in an autopsy performed many hours or days after death and which have been presented in multiple autopsy studies on sars-cov- [ ] [ ] [ ] [ ] . the autopsy must be carried out quickly and planned before executing it, adapting its performance and methodology to the specific items of each case. these must be defined by the previous interview between the doctors and the pathologist. the maneuvers useful for macroscopically examining and sampling the organs identified as the main target in every single autopsy, will be privileged. this type of autopsy is not the occasion for elegant anatomical dissections, nor for the application of traditional teaching methods on evisceration procedures. it is essential to do it early and well, aiming for the planned goals, which may change from case to case. during the collection of material for ultrastructural, molecular or other tests, it is important to be aware that not all viscera can be sampled quickly, because the autopsy has its own procedure time and the more special samples are taken, the more time will increase. before starting the autopsy, the evisceration and sampling sequence must be planned (especially when it is necessary to take tissue for electron microscopy) and be aware that the organs sampled after one hour from the beginning of the autopsy will only be able to provide samples for histology and immunohistochemistry. in our experience the patients who undergo autopsy within few hours of death, bleed much more and the incision of very congested vessels can produce relevant blood splatters than in an autopsy performed after hours or more. in addition, patients hospitalized in intensive care for sars-cov- can be anticoagulated, a condition that increases the leakage of blood. during the autopsy the blood must absolutely not be dispersed in the environment surrounding the sector table. for this reason, the body must be positioned completely inside the sector table and the mobilization of large visceral masses outside the body should be avoided: letulle's technique (evisceration en masse) therefore does not appear adequate. blood, urine and other biological liquids must remain inside the body or aspirated by a vacuum pump in a container that can be sanitized after the autopsy. the commonly recommended use of rags or sponges to collect blood is totally inadequate and dangerous. given the multiple clinical findings of neurological symptoms in patients infected with sars-cov- [ , ] , it appears indispensable to perform the evisceration and examination of the brain and brainstem, for the completeness of the autopsy and for the very few morphological data available today on the central nervous system. our experience shows that the use of a valid circular saw with dust extraction system, combined with adequate ppe, does not put operators at risk and allows the doctor to obtain valid tissue samples. given the importance of time, it seems useless to waste it performing the swabs for the detection of sars-cov- in the usual locations for the living, when abundant material can be collected by inserting the swab directly into the lung parenchyma. to date we are conducting further investigations on the samples obtained from early performed autopsies in order to evaluate the organ damage, particularly on snc, heart and lung. at the beginning of the third millennium, it is anachronistic to engage with the challenges of our discipline with the tools available to morgagni, malpighi, vesalio or virchow, when we have new and powerful weapons at our disposal. the limited possibilities to apply modern investigation technologies to tissues severely damaged by post mortem degenerative phenomena, has led many clinicians and pathologists to believe that autopsies are obsolete although, theoretically, they may be the most complete diagnostic tool. at present the autopsy should be performed quickly after the instrumental assessment of cardiac death, to ensure that the quality of tissues examined is similar to that of biopsy or surgical samples performed on the living. the forensic autopsy is the one with the most problems because of its rapid execution is often hampered by the need to perform it in front of the consultants of the parties involved. since it is in everyone's interest to have well-preserved tissue samples available for valid histopathological, biochemical, molecular or toxicological analyses, a possible solution could be to video-transmit or video-record the stages of the autopsy that take place at the autopsy table and that, once carried out, are not repeatable. in our opinion the sars-cov- pandemic provides two major lessons. the first is that in all the areas where the virus is still circulating, even if not epidemic, autopsies must be considered at high risk of infectivity. this should prompt to increase the number of bsl autopsy rooms in order to satisfy the needs of the territories. the second lesson is that rapid autopsies for health purposes appears mandatory in every day practice, even when it is not necessary to understand the etiology and pathogenesis of a new pathology. rapid autopsies guarantee optimal tissues to apply the most sophisticated diagnostic methodologies. this supports the classic holistic autopsy examinations, which aims not only to define the cause of death but also, and perhaps above all, to reconstruct the pathological history and style of the patient's life. in conclusion even today the autopsy does not lose its etymological meaning, crucial for the correct progress of knowledge: "see with your own eyes" (from the ancient greek "autòs" -same-and "opsis" -sight-). this is the foundamental part of the scientific method (observe -make a hypothesis -verify the hypothesis) masterfully described by galileo galilei, and the first step of every medical thought process. j o u r n a l p r e -p r o o f table autopsy staff -first operator (the "dirty" one): doctor, specialist in pathology, expert in performing autopsies, even in presence of diseases with high infectious risk; performs the autopsy and, if necessary, sews the body at the end. -second operator (the "clean" one): doctor, specialist or resident in pathology, even without particular autopsy experience; takes photographs, transcribes weights, measurements, volumes and observations dictated by the first operator and assists in the collection, cataloging and storage of the samples. -technician: supports the first operator but only if he has been trained for autopsies at high infectious risk; if present, supports the first operator and sews the body at the end. external inspection of the body -body length, state of nutrition (maximum thickness of the adipose panniculus at the level of the chest and abdomen), skin (color, trophy and state of conservation), hair and nails, conjunctivae, sclerae, pupils, evaluation of any material that comes out from the mouth, nostrils or middle ear and, when possible, oral cavity and dentition. -description of any pathological or morphological abnormality detectable on external examination of the body. scalp incision and opening of the skull -bimastoid cut of the scalp, passing through vertex of the head. -the skull opening must be immediately suspended if there is a smell of bone dust in the environment: this fact indicates that the dust extraction system connected to the skull saw does not work. in this case the level of risk for the staff is unacceptable, due to biological microparticles free in the air. the operators must move away from the sector table for at least minutes (time that guarantees at least complete air exchanges in the autopsy room). evisceration of the brain -cerebrum, brainstem and cerebellum must be maintained in their anatomical continuity. -weigh the visceral block. -take samples for electron microscopy and molecular investigations, preferably collecting on a single hemisphere or visceral hemibody, preserving the other for histopathological investigations. -place the residual brain in formalin, suspending it inside the container; the brain suspension is made by passing a thin string under the basilar artery and knotting its ends at the joints of the handle of the container. -remove the pituitary gland. "y" cut of the skin and subcutaneous planes for the examination of neck and trunk cavities and viscera -the "y" cut is conducted from the two acromion processes to the xiphoid process of the sternum, and from this point to the pubis along the anterior median axis of the abdomen. -observe the characteristics of the adipose tissue, the color and trophism of skeletal muscles, the state of congestion of major blood vessels and the possible presence of pathologies or injuries. opening of the rib cage -assess the size of the cardiac area. -assess the size of the visible areas of the lungs. -assess the characteristics of pleural and pleural cavities. -measure of the maximum height of the diaphragmatic dome. opening of the pericardial sac -verify and quantify the presence of effusions or blood in the pericardial cavity. -assess the characteristics of the pericardial and epicardial surfaces -assess the content of the pulmonary veins by incision (any thrombi or clots must be taken and measured). opening in situ of the right chambers of the heart -do not damage the cardiac conduction system, particularly the sinus-atrial node, the atrioventricular node and the bundle of his. -evaluate the content of the right vessels and cardiac chambers (describing and quantifying the characteristics of any observed thrombi or clots). j o u r n a l p r e -p r o o f -evaluate the morphology of the atrium, tricuspid valve, ventricular chamber, ejection cone, pulmonary valve and artery and its branches. opening in situ of the left chambers of the heart -not open the ventricle ejection cone to avoid damaging the cardiac conduction system. -evaluate the content of the left vessels and cardiac chambers (describing and quantifying the characteristics of any observed thrombi or clots). -evaluate the morphology of the atrium, mitral valve, ventricular chamber, ejection cone, pulmonary valve and artery and its branches. eviscerate the heart -weigh it and perform three biventricular sections from the tip towards the plane of the atrioventricular valves, on transverse planes, parallel to each other and about cm apart. -on the most cranial section measure the thickness of the free walls of the two ventricles and that of the interventricular septum. -take samples for electron microscopy and molecular investigations. -place the heart completely in formalin for histological and immunohistochemical studies and, where appropriate, cardiac conduction studies. removal of the lungs by cutting them from the hilum -weigh, macroscopically examine (external and internal parenchymal surfaces) and immediately sample, collecting all the lobes and areas with significant macroscopic differences (color, consistency, content at squeezing). individual evisceration, weighing, sampling and examination of: -liver and gallbladder (in a single block). virchow technique (removal of the viscera one by one) for: lungs, spleen, kidneys, adrenals and gonads. rokitansky's technique (opening and examination in situ of viscera and apparatus) for: heart (before removing it), thorax large blood vessels, stomach, epiploon's cavity and pancreas, duodenum, large bowel, ureters, prostate, bladder and the retroperitoneal large blood vessel. ghon technique (evisceration en bloc) for: brain, tongue, organs of the neck and posterior mediastinum, liver and gallbladder, small bowel. letulle's evisceration technique (evisceration en masse) not performed to minimize the leak of blood, feces and biological fluids from the body special notes: a. to be able to eviscerate the tongue, the sublingual salivary glands and part of the hypopharynx in continuity with the trachea, it is important that the incision of the skin and subcutaneous planes starts bilaterally from the most posterior part of the acromion and not from the anterior surface of the shoulder. b. the tongue is extracted by sectioning the frenulum and the floor below the mandibular arch after lifting the skin and subcutaneous planes of the neck and upper chest, and turning the tip over on the anterior laryngeal surface. the whole block of the tongue and viscera of the neck and the posterior mediastinum is mobilized from the vertebral plane by cutting the main vessels of the neck and dissecting the esophagus after performing an obstructive ligature in correspondence with its overdiaphragmatic tract. to open the thoracic cage, dissect as far as possible the ribs at the level of their cartilage tract, to avoid producing bone spicules that may sting the operator; then disarticulate bilaterally the sternum-clavicular joint and thus remove the sternum and the anterior tract of the ribs. to increase the space of maneuver inside the chest, it is advisable to dissect the muscles of the intercostal spaces. d. removal of liver-gallbladder block cutting the hilar structures, the liver ligaments, freeing the quadrate lobe and by removing the part of the phrenic muscle corresponding to the bare area of the right lobe of the liver. e. when you need to open the bladder to remove the prostate or uterus, you need to turn it; in this case, the bladder wall must be pulled upwards, making a small full-thickness hole in it from which it will be inserted a suction cannula that will empty it. the role of the autopsy in medical malpractice cases, i: a review of appeals court decisions. autopsy committee, college of american pathologists autopsy-a procedure of medical history? post-mortem examination as a quality improvement instrument features, evaluation and treatment coronavirus (covid- ) recommendations to perform autopsies in patients with sars-cov- infection. iss working group on causes of death assessment covid- human postmortem tissue: what quality markers matter effect of fixatives and tissue processing on the content and integrity of nucleic acids assessment of specimen fixation in a surgical pathology service circ. maggio , prot. n. -indicazioni emergenziali connesse ad epidemia covid- riguardanti il settore funebre emergenza covid- . circolare ministero salute n. del . . e ordinanza del capo dipartimento di protezione civile n. del marzo . attività funebre collection and submission of postmortem specimens from deceased persons with known or suspected covid- . interim guidance legge dicembre , n. , contenente: "norme per l'accertamento e la certificazione di morte world health organization. who post-outbreak biosafety guidelines for handling of sars-cov specimens and cultures biosafety level laboratory for autopsies of patients with severe acute respiratory syndrome: principles, practices, and prospects coronavirus disinfection in histopathology cutaneous manifestations of covid- : report of three cases and a review of literature sudden infant death syndrome (sids): a study of cardiac conduction system postmortem examination of patients with covid- pulmonary vascular endothelialitis, thrombosis, and angiogenesis in covid- autopsy findings and venous thromboembolism in patients with covid- : a prospective cohort study neurologic manifestations of hospitalized patients with coronavirus disease rising evidence for neurological involvement in covid- pandemic early performed autopsy (within - hour from death) provides tissue samples for diagnosis and research of quality similar to biopsy or surgical resections early samples collection reduces post-mortem artifacts, thus preventing the wrong interpretation of the morphological pictures observed precise autopsy planning prevents risks for the staff key: cord- -s i qfjn authors: rana, jamal s.; khan, sadiya s.; lloyd-jones, donald m.; sidney, stephen title: changes in mortality in top causes of death from to date: - - journal: j gen intern med doi: . /s - - -z sha: doc_id: cord_uid: s i qfjn nan trends in mortality rates due to leading causes of death reflect the medical, psychosocial, and economic well-being of a society, and a historical snapshot of such trends can help inform policies of the future. therefore, we examined changes in the number of deaths and age-adjusted mortality rates (aamr) attributed to the top causes of death between and , the last year we have data available from the centers for disease control and prevention. we chose as the start date because of earlier work showing a transition in in of the top causes of death (heart disease and stroke) from a long-term decline to increasing numbers of deaths since then. the centers for disease control and prevention wide-ranging online data for epidemiologic research (cdc wonder) dataset was used to identify national changes in the number of deaths and aamr due to the top underlying causes of death from january , , to december , . the population projection was obtained from u.s. census data. as of , the top causes of death were heart disease, cancer, and accidents ( table ). the largest percentage decline for aamr occurred for cancer deaths (− . %), and the greatest increase in aamr occurred for deaths due to alzheimer disease (+ . %). aamr for influenza and pneumonia (− . %) and chronic lower respiratory diseases (− . %) declined. increases in aamr due to accidental deaths (+ . %) and intentional self-harm (suicide) (+ . %) were observed. even though the aamr declined for of the top causes of death, the number of deaths increased for all of the leading causes. this is because the older (age ≥ years) age group grew at a much more rapid rate than that of the younger (age < years) ( . % vs. . %), while % or more of the deaths from of these causes were concentrated in older (age ≥ years) adults ( table ) . important patterns of change in aamr in the past decade have been previously noted, from stalling of the decline in mortality due to heart disease to decrease in life expectancy attributed to drug overdoses and suicides among young and middle-aged adults. while interventions to prevent and treat coronary heart disease (chd) have been successful, with ageadjusted mortality rate decrease . % in last decade, the worrisome plateau in the decline in heart disease mortality seems to be driven by an increase in mortality for heart failure ( . %), with majority of deaths due to heart disease happening in the increasing aging population. the largest percent decline during this time period of the study was noted for cancers. according to a recent report, this progress is driven by long-term declines in death rates for the leading cancers, namely lung, colorectal, breast, and prostate cancers. that report also noted that over - , reductions slowed for female breast and colorectal cancers and stopped for prostate cancer; in contrast, declines accelerated for lung cancer, which remains the biggest contributor of mortality among cancers. it remains to be seen what the final death toll will be due to covid- in . with more than , deaths by may, it has already surpassed the number of deaths attributed to all but the of the leading causes of deaths in including influenza and pneumonia, the th highest cause of mortality in . due to the direct and myriad of indirect consequences of this pandemic, mortality rankings due to top causes noted in the current report may look very different in . as noted, almost three-quarters of the deaths from of these causes were concentrated in older (age ≥ years) adults. further, the ≥ years population is projected to increase by % from . million in to . million in so that the number of deaths from most of the leading causes can be expected to increase unless more effective preventive and therapeutic interventions can be implemented. recent trends in cardiovascular mortality in the united states and public health goals underlying cause of death projected age and sex distribution of the population life expectancy and mortality rates in the united states association between aging of the us population and heart disease mortality from cancer statistics, publisher's note: springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations key: cord- -t hnswsq authors: paul, norbert w.; caplan, arthur; shapiro, michael e.; els, charl; allison, kirk c.; li, huige title: human rights violations in organ procurement practice in china date: - - journal: bmc med ethics doi: . /s - - -x sha: doc_id: cord_uid: t hnswsq background: over % of the organs transplanted in china before were procured from prisoners. although chinese officials announced in december that the country would completely cease using organs harvested from prisoners, no regulatory adjustments or changes in china’s organ donation laws followed. as a result, the use of prisoner organs remains legal in china if consent is obtained. discussion: we have collected and analysed available evidence on human rights violations in the organ procurement practice in china. we demonstrate that the practice not only violates international ethics standards, it is also associated with a large scale neglect of fundamental human rights. this includes organ procurement without consent from prisoners or their families as well as procurement of organs from incompletely executed, still-living prisoners. the human rights critique of these practices will also address the specific situatedness of prisoners, often conditioned and traumatized by a cascade of human rights abuses in judicial structures. conclusion: to end the unethical practice and the abuse associated with it, we suggest to inextricably bind the use of human organs procured in the chinese transplant system to enacting chinese legislation prohibiting the use of organs from executed prisoners and making explicit rules for law enforcement. other than that, the international community must cease to abet the continuation of the present system by demanding an authoritative ban on the use of organs from executed chinese prisoners. electronic supplementary material: the online version of this article (doi: . /s - - -x) contains supplementary material, which is available to authorized users. before a pilot organ donation program was introduced in , at least % of the transplanted organs in china were procured from prisoners [ ] . china is the only country in the world that systematically uses organs from prisoners for transplantation [ ] . after decades of denial, china finally admitted the practice in [ , ] . however, organ procurement from executed prisoners continued. in december , the chair of the china organ donation and transplant committee and former vice-minister of health, huang jiefu, announced that the country would completely cease the use of prisoner organs for transplantation in china after . surprisingly, this announcement has not been followed by any changes to china's organ donation laws or regulations (see website of prc national health and family planning commission: http:// en.nhfpc.gov.cn/regulations.html). the use of prisoner organs remains legal, if so-called consent is obtained from the prisoners [ , ] . such 'consents' , however, even those actually signed by the prisoners, are not accepted by international organisations such as the transplantation society (tts) [ ] . because of the restrictions on liberty in a prison environment, it is unlikely that prisoners are truly free to make independent decisions and thus an autonomous informed consent for donation cannot be obtained [ ] . organ harvesting from executed prisoners violates international principles of medical ethics [ ] . the compromised autonomy of all prisoners as an ethical restriction regarding the process of informed consent had historically been derived from a thorough analysis of crimes against humanity performed by physicians during the german nazi regime mostly on prisoners in german concentration camps. in asia, crimes against humanity performed during world war ii were shamefully not prosecuted in a similar way as the nuremberg military tribunals. the vivisections and experimental atrocities performed by japanese biological warfare unit against occupied chinese and prisoners of war in manchuria were not raised, addressed nor prosecuted during the tokyo trials [ , ] . the unethical practice of organ procurement from executed prisoners in china has lasted for decades [ ] . moreover, this practice is associated with large scale abuse and severe human rights violations. in part, the transgression of ethical boundaries in the case of organ harvesting is triggered by an ever growing local and global demand for transplantable organs associated with the emergence of a global (black) market [ ] [ ] [ ] . this heavily impacts human rights, ethics and justice far beyond ideas calling for regulated markets for organs [ ] . in a prior publication, we explored the historical development and status of organ procurement from death-row prisoners in china [ ] . in the present article, we more closely examine transgressions of human rights (mainly focusing on legal considerations of human rights), contradictory frames offered in justification, conditions claimed for consent of this most vulnerable of populations, issues of brain death criteria in the context of china, and several specific examples. from the s, organs were simply taken from executed prisoners in china without asking for the permission of the prisoners or their families. in , the first "interim rules on using the body or organs of executed criminals" [ ] (also translated as "provisional regulation on the use of dead bodies or organs from condemned criminals" [ ]) officially allowed, for the first time, organ harvesting from executed prisoners, under the condition that either the body is not claimed; or the prisoner volunteers for organ donation; or the family consents. as prescribed by the rules, "the use of the dead bodies or organs from condemned criminals must be kept strictly confidential. […] the operation vehicles from medical institutions can be allowed entry into the execution grounds to remove organs, but vehicles with the logo of medical institutions are not to be used, and white clinic garments are not to be worn. the execution ground should be guarded against before the operation is completed. after the dead bodies are used, the crematory shall assist the [medical] units in timely cremation" [ ] . in many cases, the families receive only the cremains without being informed what has happened with the body of their loved ones. recently, it has been repeatedly admitted by chinese transplant officials that organs from executed prisoners had been procured without consent [ ] [ ] [ ] . for example, a chinese medical official admitted in that "previously, authorities used executed criminals' organs without their consent, while permission has been required in recent years" [ ] . in , when hong kong newspaper mingpao asked the question whether prisoners and their families are informed and consent obtained before organ procurement, huang jiefu admitted that "we have not been able to achieve this, but we will soon" [ ] . in , huang jiefu admitted that the problem with china's organ transplantation system was that the laws were not obeyed: "although the law provides that the use of prisoner organs must be from voluntary donation, but there are still loopholes in the law enforcement" [ ] . all of these statements strongly suggest that organs were procured in china largely without consent of the prisoners or their families. although organ procurement without consent is illegal according to chinese laws, the practice appears to be tolerated by the authorities in china. at a european parliament hearing on january , enver tohti testified that he was ordered in to harvest organs from an incompletely executed, still-living prisoner in china [ ] . more details were revealed in his interviews with abc [ ] , bbc [ ] and journalist ethan gutmann [ , ] . when enver tohti got the body of the 'executed' prisoner, he noticed, however, that the prisoner was not dead. "the gunshot, gun wound was on his right chest. so, i guess that was deliberately to make this prisoner not die immediately to allow some time for us to remove that organ when he is still alive" [ ] . when tohti kept making attempts to follow normal proceduresterilize, minimal exposure, sketch the cut -the chief surgeon told him to hurry up. "no anesthesia, no life support", "we are working against time". finally, tohti extracted the liver and the kidneys from the still-living prisoner [ , ] . unfortunately, this was not the only case reported. an early case has been documented in the book "china's eyes" [ ] . in , zhong haiyuan, a schoolteacher from the jiangxi province, was sentenced to death for her "counterrevolutionary" thoughts. the execution was performed by three officers from the people's armed police forces on april , . two officers fixed zhong while the third officer put the gun against her back on the right side and fired the bullet [ ] . years later, one of the officers told the books' author that the order was not to kill zhong immediately. "the kidneys must be harvested before she dies", because the army doctors wanted high quality kidneys, "kidneys from a living person" [ ] . more recently, mr. wang, who currently lives in canada, revealed that he was once a member of a team extracting organs from a still-living person. the incident happened in the s when wang was an intern doctor at the urology department of the shenyang military general hospital in liaoning province [ ] . in march , jiang yanyong told to hong kong journalists that corruption, illegal transplantation and organ trade were common in military hospitals [ , ] . jiang elaborated on the case of li shiyong, director of the department of general surgery of the beijing military general hospital. with no prior experience in transplantation and without asking for permission of the hospital director, li founded a liver transplantation center in and appointed himself as the director. because li had found ways to obtain donor livers, he could serve as the director (see additional file : table s ). in the same tv interview, jiang also revealed that many of the prisoners used for organ harvesting were shot but not completely killed [ , ] . the purpose was for organ harvesting while the prisoners were still alive in order to keep the warm ischemic time of the sourced organs as short as possible (see additional file : table s ). jiang as a source is, we contend, credible based on his personal story within the chinese medical system. he was a chief physician of the military hospital (people's liberation army general hospital) in beijing where he witnessed the results of the trauma inflicted on the students during the tiananmen square massacre of . jiang was also the person who publicized the cover-up of the severe acute respiratory syndrome (sars) epidemic by the chinese government in . unfortunately, the strength of these examples is constrained by the fact that the number of executions and the detailed techniques used are state secrets in china; empirical data cannot be generated. there are no statistics available on the incidence of the practice of incomplete executions in china. a systematic international investigation into this issue is needed in the interest of (overdue) justice for the victims. in most western cultures, treating a person only as a mere means to an end of another is a challenge to core concepts of human dignity as immanuel kant and other th century philosophers argued [ ] . even from a contemporary perspective, ethicists have always argued that transferring organs to another person either from a living donor or from post mortem procured organs must be contextualized in a way that a) human dignity, b) autonomy and c) social justifiability in the light of shared values are not endangered [ , ] . here, human dignity is understood to be an applied concept. based on the notion of phenomenological dignity it is cognition that enables humans to identify themselves and to perceive others as human beings which are to be treated with the same respect (dignity) that one would expect for oneself. this concept is fostered by a reflexive mode of dignity in which the constant (re-)evaluation and adjustment of actions is put in place by relating both, intentions and consequences to commonly accepted values, like those which constitute our understanding of human rights. finally, the mutual appreciation of humans together with a value based evaluation and adjustment of interaction turn dignity into a relational concept which ought to be the guiding principle of human coexistence. especially the relational dimension of dignity is addressed in discourses on brain death when it comes to evaluations of those values, moral concepts and needs which need to be reflected for both, donors and recipients of organs [ ] . only this clarification could lead to a situation in which the concept of brain death established the medical basis and justification to switch from life-sustaining care setting to considering a deceased person as a cadaver organ donor. ethically, brain death is the fine line on which dignity of a living and acting person changes into the dignity of a deceased person, and respecting autonomy of a living person shifts over into respecting an advanced directive or declared will to donate organs. in this context, the justifiability of organ procurement is derived from the fact that the braindead person can donate organs without suffering from vital consequences caused by organ removal (e.g. pain and death), since he or she is already dead. intensive care is provided only to protect the organs of the deceased. it is this integration of medical and ethical functions of brain death which makes organ procurement a widely accepted practice in western culture. this is especially true because of the potential for other procurement strategies, such as procurement from non-heart-beating donors (donation after circulatory death -dcd), which require careful, detailed and transparent protocols, in order to assure the avoidance of even the potential for conflict of interest regarding treatment, and verification of the death of the donor [ , ] . circulatory death is, if not clearly defined as the irreversible cessation of cardiovascular circulation, potentially reversible. furthermore, it can be prognostically highly dependent on a number of arbitrary, concomitant circumstances (time, temperature, cause), thus requiring strict protocols, including electrocardiography and blood pressure monitoring to assure death has occurred, and is permanent. dcd donation may also lead to decreased quality of recovered organs, because of prolonged ischemia, or reduced number of organs that can be procured. it thus has to be seen as a procedure only applicable in an environment with reliable clinical standard operating procedures (sops) which have to be implemented in an evidence-based mode by well-trained medical personnel [ , ] . however, there is yet no brain death legislation in china and circulatory death is the legal standard despite the absence of evidence-based and reliable sops [ ] . china lacks any state issued official guidelines to diagnose brain death. in , the ministry of health drafted "brain death determination criteria" and "brain death determination technical specifications" (comment drafts). in , the ministry of health revised the two documents. in march , the national health and family planning commission (nhfpc, the former ministry of health) assigned xuanwu hospital of capital medical university as the brain injury quality control evaluation center. in , the center further revised and combined the two documents into "brain death determination criteria and technical specification (adult quality control version)" [ ] . these criteria and technical specifications represent a suggested medical standard; it is not a standard procedure, a mandatory guideline for medical practitioners or an administrative regulation. above all, the standard is not legally binding. as stated in the editor's note accompanying its publication, "the center has revised and improved the above-mentioned documents on the basis of years of clinical practice and research on brain death determination, and hopes that the new document serves as the medical standard to promote the brain death determination in our country to develop orderly and normatively" [ ] . the first documented brain death diagnosis in china was performed on february , in wuhan, hubei province [ ] . the brain death determination was carried out according to the published draft. in november , the kidneys of a boy diagnosed as brain dead were used for transplantation with the consent of the parents [ ] . both events were considered breakthroughs in chinese transplant medicine. both were nominated by major chinese media to be among the top medical stories of [ ] (fig. ). since then, there have been reported increasing numbers of organ donations after brain death, although "it is illegal to take organs from the brain-dead for transplant purposes" in china, as acknowledged by china daily [ ] . even though no brain death legislation has been effectuated in china, the chinese classification of deceased organ donation has been recently formulated as follows: "china category i: organ donation after brain death; china category ii: organ donation after circulatory death; china category iii: organ donation after brain death followed by circulatory death" [ ] . the ambiguity of this classification together with the absence of sops, guidelines, regulations and legislation leads to a situation in which legal, medical and ethical uncertainty continues. the medical journal henan medical research published a research paper titled "the experience of homologuous orthotopic heart transplantation" (fig. ) . the operation was allegedly performed in a hospital of the people's armed police force in and the paper published in [ ] . the -year-old male recipient died from infection days after the heart transplantation [ ] . in the section . of this research paper, the "major points of donor heart removal" included: "systemic heparinization ( mg/kg); delivery of cold cardioplegia to the heart through the aortic root until the heart stopped beating; cut of the superior vena cava at cm above right atrium …" (fig. ) . besides blood type and heart weight, no other information about the donor was provided in the paper. the fact that systemic heparinization was performed and heart beating was stopped by cold cardioplegia implies that the blood was circulating and the heart was functional before the explantation procedure. logically, one of the following two scenarios are applicable to describe the heart explantation: (i) the donor was a brain death patient and a brain death diagnosis was performed as it would have in most confirmatory brain death diagnoses based on neurological standard operating procedures (e.g. in the eurotransplant region); (ii) the donor was not a brain death patient and the cardiac death was induced by the cold cardioplegia delivered by the medical professionals. in this context, it is important to re-emphasize that there is no brain death legislation in china and circulatory death is still the legal standard in china [ ] . as mentioned above, the first brain death determination in china was performed in ( fig. ) [ ] . the heart transplantation in this publication, however, was performed in and the concept of "brain death" is not mentioned. this therefore raises the question as to whether the donor was brain dead, especially given that the paper was published at a time where the diagnostic procedure alone would have captured the attention of the chinese medical community and would have contributed positively to the scientific impact of the paper. it is thus reasonable to assume that the delivery of cold cardioplegia served the purpose of execution and explantation at the same time. on june in san francisco, the republic of china (roc), first in suffering axis aggression, was also first to sign the united nations charter and statute for the international court of justice, which entered into force on october, establishing the united nations. the latter date followed the double tenth agreement ( / / ) [ ] . the united nations office of high commissioner for human rights characterizes universal human rights as "interrelated, interdependent and indivisible" often expressed [ ] . text in left box: 'at the beginning of , the drafting group of the ministry of health completed the "brain death determination criteria (for adults)". according to these criteria, the brain death coordinating group of the tongji hospital at the tongji medical college, huazhong university of science and technology, completed the first brain death determination and treatment cessation in china on february '. text in right box: 'the animal model of brain death, the first brain death determination in adults, and the first brain death in children followed by unpaid organ donationthese three pioneering works were nominated by major chinese media to be among the top ten medical news of , ranking from to , respectively' in treaties, customary international law (i.e. implicit norms) or general principles. documents comprise moral declarations, which implicitly presuppose a concept of human flourishing, and derivative binding treaties (covenants, conventions, protocols). the troika of the udhr and the complementary international covenants on economic, social and cultural rights (icescr) and civil and political rights (iccpr), respectively opened for signature in parallel in , comprise the so-called 'bill of human rights' , with others building upon them. china is a full party to the icescr (signed , ratified ); the convention against torture and other cruel, inhuman or degrading treatment or punishment (cat, ratified ); the convention on the prevention and punishment of the crime of genocide (signed , ratified ); and the united nations convention against transnational crime (ratified ) whose concerns include organ trafficking (albeit patients rather than prisoner organs have comprised the transnational element in china). [ ] china also signed, but has not ratified, the iccpr in october . while formally not bound by iccpr provisions until ratification or accession, china is obliged not to defeat its general purpose. per the foreign ministry at the time, china's signature "demonstrates its firm determination to promote and protect human rights" as well as concretely commemorating the udhr's th anniversary, acknowledging a coherence bridging moral and legal rights. it also emphasized flexibility in prioritization: "the principle of the universality of human rights must be respected, but the specific conditions of each country must also be taken into consideration in observing this principle," highlighting economic development successes against poverty. the tone markedly contrasts more recent ideological tightening contra "western ideas", expressly against universal values of human rights, parallel to an anticorruption campaign short on procedural protections [ ] . china domestically has circulated "alternatively phrased" translations of signed human rights conventions rather than the un's official chinese text, according to one source providing an interpretive buffer while also imprisoning those who ask for the iccpr to be ratified [ ] . legally binding, however, are the official un language versions, not ccp state paraphrases. human rights violations ending in organ harvesting from prisoners are, like human rights, also interrelated and interdependent. violation cascadesone violation increasing probability of another given the substantive interrelationship of rights -increase vulnerability cumulatively whether in a specific person or in a targeted population. rights violations via detention schemes and a death penalty regime subject to manipulation with lack of sufficient representation or appeal [ ] precondition the subsequent act of forced organ extraction: risk accumulates to a specific subject and identifiable population. in this sense, the implication of the title of this paper is extensive and systemic, not simply perioperative. de jure and de facto factors, frequently in violation of international human rights standards, lead to and condition the supply of prisoners being solicited before execution or simply exploited in the first instance. specifically concerning the death penalty, variability includes bringing charges across potentially capital offenses; death sentences and executions waxing and waning during 'strike hard' campaigns; and executions increasing before the new year. furthermore, judges in china are not unaware that executees are the mainstay of transplantation. beyond the design of preserving public order and social control through execution is added the exogenous influence of medical demand for execution as the chief gateway to a social good (organs for transplantation), at times indirectly abetted by the international community (e.g. the transplantation society's policy supporting providing training to chinese transplant surgeons under the banner of influencing eventual reform while simultaneously expanding capacity, as in the tts ethics committee letter of november to tts members [ ] ). it is notable that the initial rules officially allowing organ harvesting from executed prisoners cited the initiative of medical personnel, not the state, as seeking to exploit this context for organ sourcing [ ] . the practice of organ procurement in china that we have described violates freedom from torture and other cruel, inhuman or degrading treatment or punishment (udhr, cat, iccpr art. ). nowhere is an individual as subject to state power than in prison, and nowhere in prison than when awaiting execution. however, it is the contiguous context, not merely local 'choice' , that is proper object of the human rights critique, exposing the step-wise cumulative vulnerability of prisoners at risk of being exploited in and through execution. this is aided and abetted by medical demand by an occupation that first pushed for exploiting execution for organ harvesting, and by a citizen population willing to rely upon, benefit from, and exploit the bound population -evident from ongoing reticence to participate in voluntary donor registration, yet seeking transplantation surgeries. in general, an alibi of system reform has been countenanced too long. after over a decade of reform claims and the redefinition of the status of prisoners, the ethical outcome respecting human rights in china, here conforming with international medical ethical standards, is categorical cessation of organ sourcing from prisoners. practically, this would remove the perverse incentives shared and relied upon by medical and judicial establishments, and by the general population. given the nature of the violations and delay, justification of gradually realizing an ethical practice fails. this recognition should hold for actors and institutions of influence outside of china now in possession of over a decade of knowledge. while human rights violations in china span systemic structural preconditions, augmented by political whim, the most proximate point of intervention still lies with the medical community and professional societies. admitting the failure of gradualism, and increasing, rather than decreasing, professional sanctions, may more quickly realize the intent: cessation of prisoner organ sourcing generally; reducing perverse incentives in death penalty demand (moral hazard); and confronting the general population with two ethical alternatives: supporting or declining voluntary organ transplantation as a system, while bearing the cost of either choice. the unethical practice of organ procurement from executed prisoners in china is associated with a large scale of abuse and a cascade of severe human rights violations, including, we contend, organ explantation from stillalive human beings, and, upstream, conditioning the supply of prisoners exploited per se or then solicited to 'freely' offer organs as atonement for real or supposed crimes. those involved in organ harvesting from stillalive prisoners must be prosecuted. the unethical practice of lethally procuring vital organs from the living must be prevented by a law prohibiting use of prisoner organs generally, supporting change in the practical legal, medical and popular culture surrounding transplantation in china. finally, greater influence may be exerted by international institutions through change of strategy. additional file : table s . government policy and organ transplantation in china open letter to xi jinping, president of the people's republic of china: china's fight against corruption in organ transplantation china to 'tidy up' trade in executed prisoners' organs. the times organ procurement from executed prisoners in china historical development and current status of organ procurement from death-row prisoners in china transplant medicine in china: need for transparency and international scrutiny remains the use of prisoners as sources of organs-an ethically dubious practice victims of japan's notorious unit sue unit and moral repair organ trafficking: global solutions for a global problem human organ markets and inherent human dignity transplant tourism and unregulated black-market trafficking of organs regulating the sale of human organs: a discussion in context with the global market provisional regulations on the use of dead bodies or organs from condemned criminals ( ). in: organs for sale: china's growing trade and ultimate violation of prisoners' rights: hearing before the subcommittee on international operations and human rights of the committee on international relations, house of representatives, one hundred seventh congress, first session chinese medical official admits that organs were extracted without consent. epoch times high official admitting that prisoner organs are still being used -without consent of the prisoners or their families (in chinese) former vice-minister of health: foreigners have demonized chinese doctors because of use of prisoner organs (in chinese) a horror story. unpo. china promises prisoner organ donation phasing out. abc.net.au. may the xinjiang procedure. weekly standard the slaughter : mass killings, organ harvesting, and china's secret solution to its dissident problem chinas eyes (chinese edition isbn: ). nanchang city: th century publishing house former chinese hospital intern recounts a live organ harvest. epoch times retired military doctor jiang yanyong: trading of organs from executed prisoners common in militaryhospitals military doctor of conscience jiang yanyong denounced serious corruption in the army (in chinese) groundwork of the metaphysics of morals : a german-english edition. cambridge replacement parts : the ethics of procuring and replacing organs in humans the ethics of organ donation by living donors transplantation ethics presumed consent for organ preservation in uncontrolled donation after cardiac death in the united states: a public policy with serious consequences non-heart-beating organ transplantation: medical and ethical issues in procurement donation after cardiac death: respecting patient autonomy and guaranteeing donation with guidance from oregon's death with dignity act voluntary organ donation system adapted to chinese cultural values and social reality brain death determination criteria and technical specification (for adults; quality control version) (in chinese) china's first brain death determination in children and free organ donation and transplantation brain death and organ donation: the ultimate combination of modern science and humanism health authorities to define legal brain death. china daily the experience of homologuous orthotopic heart transplantation (in chinese with english abstract) ministry of foreign affairs of the peoples republic of china. the signing of the international convention on civil and political rights by the chinese government china takes aim at western ideas suppressed in translation: how chinese versions of un covenants gloss over human rights false promises of fair trails we thank the bmc medical ethics reviewers and editors for their comments and suggestions that helped us substantially to revise this article. this work was partly supported by the grant graduiertenkolleg (grk) / "life writing -life sciences" from the deutsche forschungsgemeinschaft (dfg), bonn, germany. the datasets supporting the conclusions of this article are included within the article and its additional file. some reports (e.g. references [ ] and [ ] ) are additionally archived on a publicly accessible server (https://archive.is). the web addresses of the archived reports are provided at the end of the references.authors' contributions nwp drafted the manuscript and critically revised the manuscript. ac helped to draft the manuscript and critically revised the manuscript. mes helped to draft the manuscript and critically revised the manuscript. ce helped to draft the manuscript and critically revised the manuscript. kca drafted the manuscript and critically revised the manuscript. hl drafted the manuscript and critically revised the manuscript. all authors read and approved the final manuscript. the authors declare that they have no competing interests. ethics approval and consent to participate not applicable. submit your next manuscript to biomed central and we will help you at every step: key: cord- - cs z x authors: baraitser, lisa title: the maternal death drive: greta thunberg and the question of the future date: - - journal: psychoanal cult soc doi: . /s - - -y sha: doc_id: cord_uid: cs z x the centenary of freud’s beyond the pleasure principle (freud, a/ ) falls in , a year dominated globally by the covid- pandemic. one of the effects of the pandemic has been to reveal the increasingly fragile interconnectedness of human and non-human life, as well as the ongoing effects of social inequalities, particularly racism, on the valuing of life and its flourishing. drawing on earlier work, this paper develops the notion of a ‘maternal death drive’ that supplements freud’s death drive by accounting for repetition that retains a relation to the developmental time of ‘life’ but remains ‘otherwise’ to a life drive. the temporal form of this ‘life in death’ is that of ‘dynamic chronicity’, analogous to late modern narratives that describe the present as ‘thin’ and the time of human futurity as running out. i argue that the urgency to act on the present in the name of the future is simultaneously ‘suspended’ by the repetitions of late capitalism, leading to a temporal hiatus that must be embraced rather than simply lamented. the maternal (death drive) alerts us to a new figure of a child whose task is to carry expectations and anxieties about the future and bind them into a reproductive present. rather than seeing the child as a figure of normativity, i turn to greta thunberg to signal a way to go on in suspended ‘grey’ time. and why should i be studying for a future that soon will be no more, when no one is doing anything whatsoever to save that future? (greta thunberg) this paper is late. not just a little late but seriously forestalled. there is some pressure -an urgency produced by the centenary of freud's beyond the pleasure principle falling in -and the desire and pleasure in partaking in a collaborative, timely celebration of the work. there are the ordinary repetitions that are holding this up: a chronic relation to my own thoughts, veering towards and away from the satisfactions and disturbances of ideas connecting or linking; the chronic overwhelm produced by the difficulty of saying 'no' and resisting the temptations of an overloaded life; and the realities of overload brought on not by a chronic relation to limits but by their obliteration by the institutions and systems that govern our lives. then, of course, as has deepened, there have been the temporalities of illness, care and grief; of the suspension of time under conditions of lockdown; the stop-start of uncertainty and helplessness. for some, it has been a time of permanent and dangerous work; of intolerable waiting for others; and of the fault-lines of inequality and racial injustice urgently rupturing the otherwise monotonous rhythm of a global pandemic. in , everything and nothing went on hold. during this time, i continued to work with patients, albeit 'remotely', in the strange temporality of a five times per week psychoanalysis. even with so much time, the wait between sessions can be felt to be intolerable. to be in an analysis is to be held in suspension from one session to the next. one of my patients describes the wait as an agonizing 'blank time', like the crackling of an oldfashioned tv. it is not dead time as such but the incessant noise of nothing happening. to be in the session, however, produces a different kind of disturbance: an utterly absorbing kind of time that they liken to the colour blue. we move between the absorbing blue time of the sessions to the blank, crackling, maddening time between them. there is a 'session-time' analyst, who is blue, and a 'between-session-time' analyst, who maddens with a blank, crackling absence. time is both interminable -a wait between the sessions that feels like it goes on forever -and chronic: the repetition of blue, blank, blue, blank, blue, blank… beyond the pleasure principle is freud's meditation on the temporalities of repetition and return as species-time articulates with the time of the subject. in many ways, the death drive is a temporal concept, holding together the paradoxical time in which repetition contains within it a backwards pull towards the no-time of the living organism, even as the shape of this relation describes 'a life'. years later, time in the early decades of the st century Ó springer nature limited. - psychoanalysis, culture & society appears oddly analogous: it seems to loop or repeat but is undercut by a pull towards no-time, since the human and planetary future is not just foreshortened but now 'foreclosed' by the immanent twin disasters of capitalist and (neo)colonial expansion (baraitser, a, p. ) . franco 'bifo' berardi has long argued that our collective human future has come and gone and that the future has outlived its usefulness as a concept (berardi, ) . time after the present will come, but it will not bring the promises of bettering the conditions of the now for most, this having been a central aspect of european and north american future narratives in the post-war period (toffler, ; lee, ; luhmann, ) . in fact, as naomi klein ( ) argues, the very folding of disaster into capitalist discourses, governmental policies and institutional practices does not stave off disaster but profits further from it, pushing the relations between the human and non-human world to the brink of sustainability. what this implies is that disaster is not a future horizon we must urgently draw back from but a condition we have already incorporated, profited from and continue to sustain in the present. in these conditions of 'crisis capitalism', whole populations are kept in a 'chronic state of near-collapse' (invisible committee, , p. ) , a kind of temporal hiatus in which one goes on but without a future. amy elias ( ) has noted the intensive discussions about the 'presentism' of post-wwii globalized societies that have revolved around the idea of the loss of history (p. ). in these narratives, a sense of a saturated, elongated, thin present is a product of a traumatized western collective consciousness confronting the unprecedented 'event' of wwii. however, these narratives, she argues, have given way in the st century, as humankind 'has created its own version of durational time inside (rather than outside) the box of historicity' (p. ). this durational time is not bergson's duration that teams with experience (bergson, (bergson, / (bergson, , (bergson, / but the empty, timeless time of a 'marketplace duration' (elias, , p. ) , closer to the maddening crackling of nothing happening that my patient describes. in addition, as time is increasingly synchronized in the post-war period in terms of economic, cultural, technological, ecological and planetary registers, the 'present' itself becomes the management of a tension between time that is felt to be synced or simultaneous and time that is multiple or heterogeneous to simultaneity (burges and elias, , p. ) . we could think of this tension as produced by the dominating effects of european models of time (mills, (mills, , . european time is constantly imposed by the west on 'the rest' through the temporal structures of empire and enacted through colonization, exploitation, extraction and enslavement. european time comes to mediate representations of the world through the imposition of a particular account of the world-historical present on other temporal organizations -cosmic time, geological time, earth time, soil time, indigenous time, women's time, queer time, to name a few (chakrabarty, ; freeman, ; kristeva, kristeva, / nanni, ; puig de la bellacasa, ) . another way to put this is that, although freud proposes that repetition leads to the ultimate suspension of time -the return to non-being -the state of nonbeing produced by temporal suspension in the early st century is radically unequally distributed. writing under conditions of lockdown during the covid- pandemic, achille mbembe ( ) states: for we have never learned to live with all living species, have never really worried about the damage we as humans wreak on the lungs of the earth and on its body. thus, we have never learned how to die. with the advent of the new world and, several centuries later, the appearance of the 'industrialized races,' we essentially chose to delegate our death to others, to make a great sacrificial repast of existence itself via a kind of ontological vicariate. non-being, or death, is a luxury that hasn't yet been learnt by the 'human', non-being having been delegated to slaves -those humans who are denied status as humans against which the category of 'human' is both founded and flounders -as well as to non-human others. unless we recognize the 'universal right to breath' (emphasis added) for all organic matter, mbembe argues, we will continue to fail to die for ourselves, the death drive being projected, that is, into the body of that which is deemed non-human. if we go on collectively refusing to die for ourselves, we could say that the temporality of the current human predicament is closer to what martin o'brien calls 'zombie time' (o'brien, ) . as an artist and writer living with cystic fibrosis, which gives rise to symptoms very similar to covid- (coughing, shortness of breath, exhaustion), o'brien has now outlived his own life expectancy. he writes: zombie time insists on a different temporal proximity to death. like the hollywood zombie which holds within it a paradox, in that it is both dead and alive, those of us living in zombie time experience death as embodied in life […] .we had come to terms with the fact that we are about to die, and then we didn't. freud's movement towards death is circular: a repetitive arc that leads us back to the inorganic, so that in some sense it too describes zombie time, the fact we have always already surpassed our death date, whereby a life is an act of return. each organism follows its own path, he tells us, to death, and that deviation is a life. a path, however, is not quite what o'brien is suggesting. here the presence of death is sutured to every aspect of life, closer perhaps to melanie klein's insistence on the death drive as a permanent unconscious phantasy that must be managed as a life-long psychic struggle (klein, (klein, / . two questions arise from this. firstly, does recognizing 'death as embodied in life' lead us to begin to die for ourselves? in this 'hour of autophagy', as mbembe ( ) puts it, we will no longer be able to delegate death to an other. we do, indeed, have to die not just in our own fashion but on our own behalf. in one reading of freud's death drive, it is associated with the freedom to do one's own thing, follow one's own path and stands as a marker of an independent life in many ways free from others -even if, as lacan would have it, not free from the big other. but, as so many feminist, queer, disability, and black studies scholars have attested, living an independent life is a fantasy; it is always premised on dependency or interdependency, which so often requires the temporary or permanent tethering of the life of an other, or, more profoundly, the harnessing of 'life' itself. judith butler ( ) writes in the force of non-violence that we are all born into a condition of 'radical dependency' (p. ), that no-one stands on their own, that we are all at some level propped up by others. freud's suggestion of 'eternal return' requires practices of maintenance that have largely been accorded to women, people of colour, animals, and other non-human others. these practices of maintenance entail the temporalities of often mind-numbing repetition: reproductive and other forms of labour that support, sustain, and maintain all living systems. in order to 'deviate', someone or something else needs to preserve, maintain, protect, sustain, and repeat. those 'others' stay on the side of life, not as progression or even deviation towards death but as a permanent sustaining of life-processes. death in life requires a simultaneous articulation, in other words, of life in death, in which the temporalities of progression, regression, and repetition can be understood as supported and supplemented by another temporal element within the death drive that operates through 'dynamic chronicity': an element that animates 'life' in such a way as to allow the subject to die in its own fashion. i call this life in death the 'maternal death drive' (baraitser, a) to distinguish it from the pleasure principle or the 'life' drive. secondly, if the time of the 'now', as i've elaborated above, takes the form of dynamic chronicity, a suspended yet chronically animated time that pushes out temporal multiplicity, what work needs to be done in order that this form of time retains some connection to a futurity for all? do the repetitions of 'blue blank' in their own circular fashion retain within them a relation to futurity, even if they don't exactly lead us somewhere else? i would hope, after all, that my patient may eventually, with time, come to experience the 'blue-session' analyst and the 'blank-absent' analyst as one and the same analyst, even as the agonies of having and losing may continue to be difficult. from a kleinian perspective, the time that this requires is the time in which what is hated and what is loved come to have a relation to one another, which klein calls 'depression' (klein (klein / and which may entail 'depressing time'. we could say that it is the time in which we come to be concerned about the damage done to what is loved, the time whereby what is loved and what is hated can come to matter to one another, making the time of working through that of 'mattering' itself. furthermore, mbembe ( ) writes: community -or rather the in-common -is not based solely on the possibility of saying goodbye, that is, of having a unique encounter with others and honoring this meeting time and again. the in-common is based also on the possibility of sharing unconditionally, each time drawing from it something absolutely intrinsic, a thing uncountable, incalculable, priceless. (emphases in original) this would suggest that, supplementary to the time of blue-blank (saying goodbye again and again), there is another time: that of the 'in-common'. this is a time of permanent mattering, which also takes time to recognize. it is, if you like, the time in which depressive guilt survives and hence the time it takes for a future to be recognized within the present, rather than being the outward edge, the longed-for time that is yet to come. in what follows, and taking my cue from beyond the pleasure principle itself, i attempt to rework freud's death drive by drawing attention to a particular form of developmental time that lies inside the time of repetition, which i link to 'life in death'. in chapter ii of freud's essay, in the midst of his struggle with the meaning of repetition, pleasure and unpleasure, he turns to a child. the function of the child at this point in the text is to provide the case of 'normalcy' -the play of children -in order to help him understand the 'dark and dismal topic of traumatic neurosis' (freud, b (freud, / . the child will be 'light' (read white) and playful but turns out to be deeply troubled. instead of dragging the cotton reel along the floor as the adults intended, so it could turn and check its existence at any point, the child, standing outside the cot, throws the reel into the cot, accompanied by an o-o-o-o sound, so it cannot be seen, and then pulls it out with a 'da!' that freud describes as 'joyful' (p. ). the pleasure of refinding, however, is postponed -in the time between 'gone' and 'found', the child plays at waiting, as it attempts to remaster the experience, freud tells us, of its 'gone' mother. this is of course also an attempt to deal with its own goneness from the imagined place of the mother; the child is standing outside the cot, after all. the passivity of being left is repeated but transformed through an act of 'revenge', a repetitive act of aggression in which, through psychic substitution, something essentially unpleasurable is turned into something 'to be remembered and to be processed in the psyche' (p. ). the child does this by identifying with the mother, waiting in her place. my aim is to repeat freud's impulse, re-inserting a mother and child into the scene of the death drive 'proper' as a way to signal how to die on our own behalf and therefore how to go on in the suspended hiatus we appear to be living through. the maternal, as i will elaborate, appears as a non-normative developmental temporality within the death drive. in my account, the child reappears, however, in the figure of the child-activist greta thunberg. she is the child who has been invested in symbolically to carry hope for the future, a hope that she is decidedly pushing back towards those of the generation who came before her, calling on them to take action now, before it is too late. although thunberg names her vision of the world in terms of 'black and white' thinking, i draw on laura salisbury's notion of 'grey time' (salisbury, in press) in order to understand what to do with the time that remains in which action can still take place. it is always an uncomfortable thing to do, to insert a mother and child into a scene where they are ostensibly not wanted. it carries the sour smells of heteronormativity and essentialism that still cling to discussions of the maternal and relegate mother-child configurations as the counterpoint to those who are 'not fighting for the children', as lee edelman ( ) suggested in his famous polemic no future. for edelman, the death drive is a queer refusal of futurity that allows negativity to operate as a 'pulsive force' that would otherwise trap queer as a determinate stable position (p. ). the child and mother come to represent the ultimate trap, that of development itself -the unfolding of the normative temporalities of birth, growth, development, maturation, reproduction, wealth generation and death. in some ways, this is what makes the insertion of mother-child back into discourses about the death drive rather 'queer'. in doing so, i deliberately refuse the association between motherhood and normativity and suggest that motherhood is the name for any temporal relation of 'unfurling' whereby the unfurling of one life occurs in relation to the unfurling of another, albeit out of sync. in fact, as i will elaborate below, for a life to unfurl there needs to be the presence of another life that is prepared to wait whilst life and death can come to have a relation to one another. this suspended time of waiting for life to unfurl is a non-teleological, crystalline form of developmental time based on the principle of life in death (baraitser, a, p. ) . whilst motherhood is always in danger of being squeezed out of this kind of queer theory, it is also in danger of being squeezed out of feminist theories that purport to make space for the maternal. julia kristeva's essay 'women's time' ( / ), for instance, conceptualized female subjectivity as occupying two forms of time: cyclical time (repetition) and monumental time (eternity without cleavage or escape). these two 'feminine' forms of time, she argued, work to conceal the inherent logic of teleological, historical, 'masculine' time, which is linear, progressive, unfolding and yet constantly rupturing, an 'anguished' time (p. ) . masculine time rests on its own stumbling block, which is death. cyclical time and 'monumental' or eternal time, kristeva argued, are both accessed through the feminine, so that the feminine signifies a less 'anguished' time because it is uncoupled from the death of the subject and more concerned with suturing the subject to extrasubjective time. although this has been rightly critiqued for essentializing 'the feminine' through the normative positioning of the female subject on the side of the biological, as well as mobilizing a nonpolitical appeal to 'nature', i have argued elsewhere that, in attempting to separate the feminine from cyclical and monumental time, feminist theory designates the maternal as the keeper of species-time, in which the mother becomes a biologistic and romanticized subject attached to the rhythms of nature (baraitser, , p. ) . toril moi ( ) writes of kristeva's essay that the question for kristeva was not so much how to valorize the feminine but how to reconcile maternal time with linear (political and historical) time (p. ). without a theory of the desire to have children (a desire that can permeate any gender configuration and that i name as maternal regardless of the gendered body that desires it), we leave the door open to the consequence of a failure to theorize and the maternal falls out of signification, time and history. moreover, motherhood is not just the desire for children but a particular form of repetitive labour relegated largely to women and particularly, in the global north, to women of colour and women from the south. although the concept of 'social reproduction' has been expanded to incorporate a much broader array of activities than caring for children, maternal labour remains distinct from other forms of domestic labour. joy james ( ) argues that the ongoing trauma and theft involved in slavery, for instance, produces not only western democracy but a repudiated 'twin' within western theory that she names 'the black matrix' (p. ). where mothers in captivity and slavery have always provided the reproductive and productive labour that underpin wealth and culture, they are systematically erased -not just in culture but in what she calls 'womb theory' (theory, for instance, that accommodates feminism, intersectionality and antiracism, whilst still denying the maternal captive). despite this, she claims, the black matrix can act as a 'fulcrum' that leverages power against captivity (p. ). i would argue that this power comes, in part, from the impossibility of the maternal captive remaining indifferent to her labour. subsistence farming, cooking, cleaning, household maintenance, support work and the production of status are forms of repetition from which it remains possible to emotionally disattach. but the 'labour' of maternity is 'affective, invested, intersubjective' (sandford, , p. ) and retains an ethical dimension that is distinct. here the maternal emerges as a figuration of the subject that is deeply attached to its labouring, whose labouring is a matter of attachment to that labour, as well as providing the general conditions for attachment (the infant's psychic struggle to become connected to the world) to take place. we could say, then, that the time of repetition under the condition that is maternity becomes the time of mattering, as opposed to the 'meaningless' time of reproduction: the time, that is, in which repetition may come to matter. this time can be felt as obdurate, distinctively uncertain in its outcome, both intensive and 'empty', and bound to the pace of the unfurling other. what is at play is a kind of crystalline developmental time within the time of history. it takes the form of repetition, but this repetition holds open the possibility of something coming to matter, rather than the death drive understood only as a return to non-being. a maternal death drive? what might this conjunction mean? freud always maintained that the two elements of psychic life that couldn't be worked through were the repudiation of femininity in both men and women, by which he meant the repudiation of passivity; and the death drive, the repetitive return again and again to our psychic dissolution or unbinding. in 'analysis terminable and interminable', written in the last years of his life, freud ( freud ( / ) named these the 'bedrocks' of psychic life, evoking an immoveable geological time. the permanent fixtures of psychic life that an analysis cannot shift are the hatred of passivity and the simultaneous impulse to return to an ultimate passive state, suturing the feminine to death in psychoanalysis. earlier, in beyond the pleasure principle, freud had offered an hypothesis in which, despite his conception of drives as exerting the pressure that presses for change, they are constrained by a conservatism, meaning they do not operate according to one singular temporality. this double temporality within the death drive is drawn out by adrian johnston ( ) , who has noted freud's ( freud's ( / ) developmental account of the drive in three essays on the theory of sexuality and later in 'instincts and their vicissitudes ' ( / ) , where the drive is articulated as maturing over time. johnston ( ) maintains that freud's drive is simultaneously timeless and temporal, both interminable (it repeats) and containing an internal tendency to deviate, to change its object and its aim (it develops or alters) (p. ). after all, something happens, according to freud, that shifts the human organism from one that dies easily to one that diverges ever more widely from the original course of life (that is, death) and therefore makes ever more complicated detours before reaching death. for johnston, alteration can be understood as an intra-temporal resistance to the time of iteration, a negation of time transpiring within time. this means that the death drive therefore includes rather than negates developmental time. this is not a developmental tendency separated off and located within the selfpreservative drives or a 'life' drive but a death drive that contains within it its own resistance to negation. i would want to reclaim this doubled death drive as 'maternal', the drive that includes within it the capacity for development, for what johnston calls 'alteration', which always mediates the axis of repetition or 'iteration' (p. ). the maternal death drive would describe the unfolding of another life in relation to one's own path towards death and marks the point that alteration and iteration cross one another. if we move from freud to klein, we see how this double temporality plays out between the maternal and child subject. i have described elsewhere how, in love, guilt and reparation, klein ( klein ( / tells us that anxiety about maternal care and dependency on the maternal body in very early life -the relationship, that is, with a feeding-object of some kind that could be loosely termed 'breast' -is a result of both the frustrations of that breast (its capacities to feed but also to withhold or disappear at whim) and what the infant does with the hatred and aggressive feelings stirred up by those experiences of frustration that rebound on it in the form of terrifying persecutory fantasies of being attacked by the breast itself (pp. - ; see also baraitser, b, p. ). klein's conceptual infant swings in and out of psychic states that are full of envious rage and makes phantasized aggressive raids on the maternal body in an attempt to manage the treacherous initial experiences of psychical and physical survival. klein ( klein ( / moves us closer to a more thing-like internal world permeated less with representations and more with dynamic aggressive phantasies of biting, hacking at and tearing the mother and her breasts into bits, and attempts to destroy her body and everything it might be phantasized to contain (p. ). in klein's thinking, libido gives way to aggression, so that the defences themselves are violent in their redoubling on the infant in the form of persecutory anxiety. one's own greed and aggressiveness themselves become threatening, along with the maternal object that evokes them, and have to be split off from conscious thought. coupled with this are feelings of temporary relief from these painful states of mind (p. ) and these 'good' experiences form the basis for what we could think of as love. it is only as the infant moves towards a tolerance of knowing that good and bad 'things' and experiences are bound up in the same person (that is, both (m)other and self) that guilt arises as an awareness that we have tried to destroy what we also love. whilst this can overwhelm the infant with depressive anxiety that also needs to be warded off, there is a chance that this guilt can be borne and a temporary state of ambivalence can be achieved that includes the desire to make good the damage done. 'unfurling', then, arises out of the capacity to tolerate the proximity of love and hate towards the mother, but the mother also needs to tolerate the time this takes -to be prepared to go back 'again and again' to the site of mattering without becoming too overwhelmed or rejecting. it is here that futurity emerges, not as that which is carried forward by the child but as this element within the death drive that i am naming as maternal, which is a capacity to tolerate repetition within the present. to return to a lacanian formulation, chenyang wang ( ) , in his work on differentiating real, imaginary and symbolic time in lacan, shows how lacan's death drive is not so much the reinsertion of the bodily or biological into the human subject but the traumatic intrusion of the symbolic into the organism at the expense of the imaginary, which evokes the real body. wang describes how what he calls the 'real future' (p. ) does not involve the human subject. where the ego may continue to imagine a future of fulfilled wishes, hopes and expectations, in which the present is characterized as a mode of 'waiting' until the future unfolds, the death drive in fact interrupts the fantasy of the future as something unreachable or unattainable and instead returns the future to the subject as something that has already structured it. for wang, real time opens the subject to the real present that is neither instantaneous or immediate but the freedom of returning to the same place in one's own way. he sees this as the offer of the possibility of freedom that transcends the isolated, egoic individual, otherwise trapped in its established temporal order (p. ). we could say, then, that the death drive includes rather than negates developmental time and holds out the possibility of a time that breaks free of the ego's imaginary sense of past, present and future. developmental time, from this perspective, is precisely a suspension of the flow of time, a capacity to wait for the other to unfold. maternity, in its failure to be indifferent to the specificity of its labour, implies a return, again and again, to a scene that matters, a kind of repetition that is not quite captured by the death drive as excessive access to jouissance, nor to the death drive as a deviation towards a unique form of death, but that might after all have something to do with generativity, indeed with freedom, not of the self, but of the other. the return to a scene that matters is not a kind of flowing time (anyone who has spent time with small children will know this) nor the stultifying time of indifferent labour, but living in a suspended or crystalline time, which is the time it takes for mattering to take place. finally, we can link the maternal death drive to elizabeth freeman's ( ) concept of 'chronothanatopolitics' (p. ) that extends mattering beyond the mother-child relation to the politics of mattering in the contemporary moment. in her discussion of 'playing dead' in th century african-american literature, freeman notes that many african-american stories involve 'fictive rebirths'(p. ). these are stagings of death and rebirth, not just once but multiple times, so that in these stories slaves and their descendants are constantly moving towards and away from death. feigning death, she argues, does not solve the problem of having not been 'born' as human -a position well established within afropessimist thought -but allows an engagement through repetitive staged dying with what jared sexton ( ) has called 'the social life of social death' (quoted in freeman, , p. ). freeman therefore builds on freud's death drive to develop a concept of 'chronothanatopolitics' in which life is not simply the opposite of death but the opposite of the 'presence' of death (p. ), a temporary 'disappearing' of death within life, the counterpart to the maternal death drive as life in death. staging one's death again and again, she states, is a way of managing the life/death binary, rather than simply a commitment to life or an acceptance of unchanging black deathliness. where freud's death drive does refuse any simple opposition between life and death, freeman notes, it nevertheless proposes a universal and purely psychic drive. she calls instead for recognition of a socio-political death drive enacted by white supremacy: chronothanatopolitics is the 'production of deathliness and nonbeing by historical forces external to the subjectivity it creates for nonblack people, and forecloses for people of african descent' (p. ). in the st century, we see 'playing dead' resurfacing in the 'die-ins' revived by the protest movement black lives matter. time becomes central, creating what freeman terms 'temporal conjoinments' with death (p. ) through counting 'i can't breathe' times, as eric garner did. we have seen this repeated in , when protesters hold a silence or take the knee for minutes and seconds, the time that george floyd had his neck knelt on by the police officer who killed him on may. 'mattering', in the sense of black life coming to matter, freeman notes, captures the double meaning of coming to importance and becoming-inert substance or matter, giving the phrase an ambivalent valence. mattering refuses the afropessimist insight that black life is structurally foreclosed and instead implies a more open stance towards non-being. by miming death rather than life, black lives matter activists 'commit to an (a)social life within death even as they fight for an end to the annihilation of blackness' (p. ). here, life in death is the 'social' work of activism that counts the time that is left within black life even as it is extinguished, just as it is the social work of mothering that waits for life to unfurl towards its death without knowing when or how this will take place. miming death, again and again, is analogous to returning to the scene of mattering again and again, the hiatus within the path towards death that i have described as the maternal death drive. however, freeman's work provides the corrective to an easy universalizing of the drive, pointing us towards the way that black lives matter politicizes repetition in the name of life in death. recently i've seen many rumours circulating about me and enormous amounts of hate. (greta thunberg) in the child to come: life after the human catastrophe, rebekah sheldon ( ) charts a recent shift in the use of the child to suture the image of the future. the child, metonymic with the fragility of the planetary system and therefore in need of protection, has become 'the child as resource' (p. ). as resource, the child is used to carry both expectations and anxieties about the future. unlike earlier iterations, the child as resource is premised on a future that cannot be taken for granted. much of the affect around ecological disasteranxiety, fear, terror, hopelessness, despair, guilt, determination, protectivenesscomes not so much from an awareness of the current effects of global climate change as they play out in the present but from the projected harm to the future that it portends. and the future, sheldon reminds us, is the provenance of the child. sheldon describes the history of this relationship between child and future as emanating from the th century at the same point as modern theories of 'life' begin to proliferate in darwin and of course in freud. 'the link forged between the child and the species', she writes, 'helped to shape eugenic historiography, focalized reproduction as a matter of concern for racial nationalism, and made the child a mode of time-keeping' (p. ). in the face of anxious concerns about the deep biological past of the human species, the child held open a future through a coordination of the trio 'life, reproduction and species' with that of 'race, history and nation'. freud's child, for instance, caught both in the relentless unfolding of developmental time and the timelessness of unconscious life, is also the site of the regulation of 'life' itself. whilst these two axes of temporality (development and timelessness), as we saw above, cross one another, the figure of the child is nevertheless a 'retronaut, a bit of the future lodged in the present ' (p. ) . yet, at the same time, sheldon's child is already melancholic. it knows its childness can't be preserved; it will be lost; just as the future is felt also to be something constantly slipping away. as a melancholic figure, sheldon suggests that the child as resource has a very specific task right now: to cover over the complex systems at work in biological materiality. as non-human animacy becomes more visible in conditions of planetary crisis, with it comes the terrifying potential (at least for the human world) of nature to slip its bonds. the child stands in for life itself at a time of vibrant and virulent reassertion of materialisms in all their forms. the child's new task, according to sheldon, becomes one of binding nonhuman vibrancy back into the human, into something safer, and into the frame of human reproduction. this perhaps helps us modulate how we might respond to the figure of greta thunberg, the climate activist who describes herself as both 'autistic' and living with asperger's, and to her work as a 'cry for help' (thunberg, , p. ) . during , when she was years old, thunberg started to skip school to sit outside the swedish parliament with a sign reading 'skolstrejk fö r klimatet' [school strike for climate]. as a result of the school climate change movement that grew around thunberg's 'fridays for future' actions during , there has been an intensive, rapid sanctification of the plain-speaking, white, plaitedhaired child now simply known as 'greta'. although she herself acknowledges that she is not unique and is part of a network of youth movements in the global south who bear the brunt in the present for the effects of climate disaster largely produced by the global north, she has nevertheless become an enormously influential figure through whom climate discussions now pass. some describe her influence as simply the 'greta effect' (watts, ) . there is a specific and careful simplicity to the way thunberg talks. in a speech entitled 'almost everything is black and white', she states, 'i have asperger's syndrome, and to me, almost everything is black or white' (thunberg, , p. ) . utilizing what others may see as a disability, a difficulty in seeing shades of grey, she speaks against the need for more complexity, more reflection, more science; in short, a more 'grown up' approach to climate chaos: 'we already have all the facts and solutions. all we have to do is to wake up and change […] everything needs to change. and it has to start today' (p. ). it is this rhetorical insistence that there is no more time and that the future of her generation has been stolen by the inaction of the generation that has come before that positions her as not so much future-orientated but backed up against a closing future, looking back towards those who came before her as they continue to gaze ahead towards what they imagine is her future. as she states, 'we children are doing this to wake the adults up. we children are doing this for you to put your differences aside and start acting as you would in a crisis. we children are doing this because we want our hopes and dreams back' (p. ). in many ways, we could see thunberg as performing a call, in the name of a human reproductive future, for the binding of nonhuman vibrancy back into the human, into something safe and stable, the child's new task that sheldon describes. we could also make a critical reading of the ways thunberg -as a contemporary incarnation of maisie in henry james ' ( / ) what maisie knew, where the child-protagonist is sacrificed to save a negligent and damaged society -re-mobilizes a discourse that re-stabilizes the differences between the generations in the name of the reproduction of the white heteronormative social bond. however, i want to read thunberg's 'black and white' thinking as metonymic with my patient's blank and blue: the oscillation between the absorbing blue of the analytic session and the suspended time of nothing happening between the sessions; the time of no-analyst and the agonies of waiting. thunberg ( ) states: 'there are no grey areas when it comes to survival. either we go on as a civilization or we don't. we have to change' (p. ). in many ways, she refuses 'development' in the sense of klein's depressive position functioning, where blue and blank come to be understood as having a relation to one another, and insists instead on their separation, on what klein would call 'paranoid-schizoid' thinking, in which blue and blank are radically split apart, as a viable place to speak from. indeed, she goes on insisting she is a child and that development is precisely what has got us into so much trouble. she warns us that, from the perspective of blank time (the time of nothing happening), blue time is absorbing for sure, but it is short, cannot last, and time itself needs to urgently come to matter if we are to find a way out of the current predicament. if we want to repair a relationship with monumental time, there is only action or no action, blue or blank, as we have now run out of time. despite the obvious occlusion of the many brown and black children who have protested, spoken out, organized school strikes and presented to the un over the years and gained no coverage, what is striking is that the white child claims that it is her unusual perspective, in which black and white remain separate, that is our only way out. in describing what she calls 'grey time', laura salisbury (in press) reminds us that grey is not, strictly speaking, a colour at all; rather, it is a shade. as such, it is achromatic, composed of black and white in various shades of intensity, rather than hues. moving from colour to time, salisbury claims that grey time can be thought of as similarly a time that contains intensities of affect, naming grey time as 'anachromistic', a form of intensive temporality that belongs to and traverses the perceiving subject and the aesthetic object. to speak of grey time as anachromistic is to evoke an aesthetic experience that is against colour or hue, but, with its echo of anachronism, also produces a slub in the fabric of time as it is usually thought. the double gesture of the term anachromism is the attempt to speak to time's intensity rather than, as is more usual, concentrating on its flow or movement, while trying to capture an atmosphere where there is a weaving or binding in of blank, uncertain, colourless 'colour', and affect into what is felt of time. (emphasis in original) grey time, then, is an intensity of time that moves us beyond the impasse of action and no action, or blue and blank, by acting as a slub or thickening in the oscillation between the two. this thickening, if we follow salisbury, both reveals time's stuck oscillation between black and white at the same point as it acts to bind greyness into what is felt of time. grey inhabits black and white without resolving the oscillation, both intensifying the sense of time's stuckness but also drawing attention to the affect of greyness, of uncertainty. whilst the time for grey thinking, as thunberg states, may have passed, perhaps salisbury's attention to grey time is important. as the existential dangers facing humanity deepen -by mbembe's description, the destruction of the biosphere, the criminalization of resistance and the rise of determinisms, whether genetic, neuronal, biological or environmental -so perhaps greta thunberg's urgency cannot be heard until we bind the blank, uncertain, colourless affect of the grey 'now' into what is felt of time. mbembe ( ) writes of the covid- virus: of all these dangers, the greatest is that all forms of life will be rendered impossible. […] at this juncture, this sudden arrest arrives, an interruption not of history but of something that still eludes our grasp. since it was imposed upon us, this cessation derives not from our will. in many respects, it is simultaneously unforeseen and unpredictable. yet what we need is a voluntary cessation, a conscious and fully consensual interruption. without which there will be no tomorrow. without which nothing will exist but an endless series of unforeseen events. (emphasis in original) this is, indeed, grey time -a voluntary cessation, a conscious and fully consensual interruption to business as usual as a response to the profound Ó springer nature limited. - psychoanalysis, culture & society uncertainty that is the reality of the interdependencies of all forms of life. although i know that there is no way for 'couch time' to have an effect without a 'session-time' analyst and a 'between-session-time' analyst eventually coming together in the time that is an analysis, it may be that we have simply run out of time. then a new psychoanalytic temporality may be needed, one that understands the simultaneous need for and suspension of development in the name of really knowing about the death drive; one in which action would no longer be simply understood as acting out but in which the mutative interpretation, the one that brings about change, can be grey, ill-timed, coming too soon and too late, before it is too late. maternal encounters: the ethics of interruption Ó springer nature limited. - psychoanalysis postmaternal, postwork and the maternal death drive. special issue: the postmaternal after the future / ) time and free will: an essay on the immediate data of consciousness / ) matter and memory the schoolchildren strikes when the kids are united introduction: time studies today the force of non-violence the climate of history: four theses no future: queer theory and the death drive past/future time binds: queer temporalities, queer histories. durham and london beside you in time: sense methods and queer sociabilities in the american nineteenth century / ) three essays on the theory of sexuality / ) instincts and their vicissitudes / ) beyond the pleasure principle beyond the pleasure principle / ) analysis terminable and interminable a queer place and time lose your mother: a journey along the atlantic slave route semiotext(e). invisible committee, the ( ) to our friends the ( ) now. new york: semiotext(e) / ) what maisie knew the womb of western theory: trauma, time theft and the captive maternal time driven: metapsychology and the splitting of the drive / ) notes on some schizoid mechanisms / ) love, guilt and reparation the shock doctrine: the rise of disaster capitalism / ) women's time chronophobia: on time in the art of the s the future cannot begin: temporal structures in modern society the universal right to breathe. translated by c. shread. critical inquiry white time: the chronic injustice of ideal theory the chronopolitics of racial time introduction to women's time the colonisation of time: ritual, routine and resistance in the british empire you are my death: the shattered temporalities of zombie time matters of care: speculative ethics in more than human worlds grey time: anachromism and waiting for beckett what is maternal labour? the social life of social death: on afro-pessimism and black optimism the child to come: life after the human catastrophe no one is to small to make a difference. london: penguin, random house future shock subjectivity in-between times: exploring the notion of time in lacan's work the greta thunberg effect: at last, mps focus on climate change. the guardian publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Ó springer nature limited. - psychoanalysis the research in this paper was funded by a wellcome trust collaborative award, 'waiting times', grant number [ /a/ /z] (see waitingtimes.exeter.ac.uk). data sharing is not applicable as no datasets were generated and/or analysed for this study. ( ) and hedva ( ) . . in doing so, she was perhaps unwittingly building on a long history of school strikes, certainly dating back at least years in the uk, in which schoolchildren mobilised against caning in and later came out on strike as part of a number of localised general strikes in . see bloom ( bloom ( , . key: cord- -zj oixvk authors: clift, ash k; coupland, carol a c; keogh, ruth h; diaz-ordaz, karla; williamson, elizabeth; harrison, ewen m; hayward, andrew; hemingway, harry; horby, peter; mehta, nisha; benger, jonathan; khunti, kamlesh; spiegelhalter, david; sheikh, aziz; valabhji, jonathan; lyons, ronan a; robson, john; semple, malcolm g; kee, frank; johnson, peter; jebb, susan; williams, tony; hippisley-cox, julia title: living risk prediction algorithm (qcovid) for risk of hospital admission and mortality from coronavirus in adults: national derivation and validation cohort study date: - - journal: bmj doi: . /bmj.m sha: doc_id: cord_uid: zj oixvk objective: to derive and validate a risk prediction algorithm to estimate hospital admission and mortality outcomes from coronavirus disease (covid- ) in adults. design: population based cohort study. setting and participants: qresearch database, comprising general practices in england with linkage to covid- test results, hospital episode statistics, and death registry data. . million adults aged - years were included in the derivation dataset and . million in the validation dataset. the derivation and first validation cohort period was january to april . the second temporal validation cohort covered the period may to june . main outcome measures: the primary outcome was time to death from covid- , defined as death due to confirmed or suspected covid- as per the death certification or death occurring in a person with confirmed severe acute respiratory syndrome coronavirus (sars-cov- ) infection in the period january to april . the secondary outcome was time to hospital admission with confirmed sars-cov- infection. models were fitted in the derivation cohort to derive risk equations using a range of predictor variables. performance, including measures of discrimination and calibration, was evaluated in each validation time period. results: deaths from covid- occurred in the derivation cohort during follow-up and in the first validation cohort period and in the second validation cohort period. the final risk algorithms included age, ethnicity, deprivation, body mass index, and a range of comorbidities. the algorithm had good calibration in the first validation cohort. for deaths from covid- in men, it explained . % ( % confidence interval . % to . %) of the variation in time to death (r( )); the d statistic was . ( % confidence interval . to . ), and harrell’s c was . ( . to . ). similar results were obtained for women, for both outcomes, and in both time periods. in the top % of patients with the highest predicted risks of death, the sensitivity for identifying deaths within days was . %. people in the top % of predicted risk of death accounted for % of all deaths from covid- . conclusion: the qcovid population based risk algorithm performed well, showing very high levels of discrimination for deaths and hospital admissions due to covid- . the absolute risks presented, however, will change over time in line with the prevailing sars-c v- infection rate and the extent of social distancing measures in place, so they should be interpreted with caution. the model can be recalibrated for different time periods, however, and has the potential to be dynamically updated as the pandemic evolves. objective to derive and validate a risk prediction algorithm to estimate hospital admission and mortality outcomes from coronavirus disease (covid- ) in adults. population based cohort study. setting and participants qresearch database, comprising general practices in england with linkage to covid- test results, hospital episode statistics, and death registry data. . million adults aged - years were included in the derivation dataset and . million in the validation dataset. the derivation and first validation cohort period was january to april . the second temporal validation cohort covered the period may to june . the primary outcome was time to death from covid- , defined as death due to confirmed or suspected covid- as per the death certification or death occurring in a person with confirmed severe acute respiratory syndrome coronavirus (sars-cov- ) infection in the period january to april . the secondary outcome was time to hospital admission with confirmed sars-cov- infection. models were fitted in the derivation cohort to derive risk equations using a range of predictor variables. performance, including measures of discrimination and calibration, was evaluated in each validation time period. results deaths from covid- occurred in the derivation cohort during follow-up and in the first validation cohort period and in the second validation cohort period. the final risk algorithms included age, ethnicity, deprivation, body mass index, and a range of comorbidities. the algorithm had good calibration in the first validation cohort. for deaths from covid- in men, it explained . % ( % confidence interval . % to . %) of the variation in time to death (r ); the d statistic was . ( % confidence interval . to . ), and harrell's c was . ( . to . ). similar results were obtained for women, for both outcomes, and in both time periods. in the top % of patients with the highest predicted risks of death, the sensitivity for identifying deaths within days was . %. people in the top % of predicted risk of death accounted for % of all deaths from covid- . the first cases of severe acute respiratory syndrome coronavirus (sars-cov- ) infection were reported in the uk on january , with the first death from coronavirus disease (covid- ) on february . as of august , more than deaths from covid- had occurred in the uk and more than deaths globally. in the initial absence of any vaccination or prophylactic or curative treatments, the uk government implemented social distancing and shielding measures to suppress the rate of infection and protect vulnerable people, thereby trying to minimise the risk of serious adverse outcomes. emerging evidence throughout the course of the pandemic, initially from case series and then from cohorts of patients with confirmed sars-cov- doi: . /bmj.m | bmj ; :m | the bmj infection, has shown associations of age, sex, certain comorbidities, ethnicity, and obesity with adverse covid- outcomes such as hospital admission or death. [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] the knowledge base regarding risk factors for severe covid- is growing. as many countries are cautiously attempting to ease "lockdown" measures or reintroduce measures if rates are rising, an opportunity exists to develop more nuanced guidance based on predictive algorithms to inform risk management decisions. better knowledge of individuals' risks could also help to guide decisions on mitigating occupational exposure and in targeting of vaccines to those most at risk. although some prediction models have been developed, a recent systematic review found that they all have a high risk of bias and that their reported performance is optimistic. the use of primary care datasets with linkage to registries such as death records, hospital admissions data, and covid- testing results represents a novel approach to clinical risk prediction modelling for covid- . it provides accurately coded, individual level data for very large numbers of people representative of the national population. this approach draws on the rich phenotyping of individuals with demographic, medical, and pharmacological predictors to allow robust statistical modelling and evaluation. such linked datasets have an established track record for the development and evaluation of established clinical risk models, including those for cardiovascular disease, diabetes, and mortality. [ ] [ ] [ ] we aimed to develop and validate population based prediction models to estimate the risks of becoming infected with and subsequently dying from covid- and of becoming infected and subsequently admitted to hospital with covid- . the model we have developed is designed to be applied across the adult population so that it can be used to enable risk stratification for public health purposes in the event of a "second wave" of the pandemic, to support shared management of risk and occupational exposure, and in early targeting of vaccines to people most at risk. an ongoing companion study will externally validate the models, using datasets across all four nations of the uk, and will be reported separately. this study was commissioned by the chief medical officer for england on behalf of the uk government, who asked the new and emerging respiratory virus threats advisory group (nervtag) to establish whether a clinical risk prediction model for covid- could be developed in line with the emerging evidence. the protocol has been published. the study was conducted in adherence with tripod and record guidelines and with input from our patient advisory group. study design and data sources we did a cohort study of primary care patients using the qresearch database (version ). qresearch was established in and has been extensively used for the development of risk prediction algorithms across the national health service (nhs) and for epidemiological research. by april , practices in england were contributing to qresearch, covering a population of . million patients. the database is linked at individual patient level, using a project specific pseudonymised nhs number, to hospital admissions data (including intensive care unit data), positive results from covid- real time reverse transcriptase polymerase chain reaction tests held by public health england, cancer registrations (including detailed radiotherapy and systemic chemotherapy records), the national covid- shielded patient list in england, and mortality records held by nhs digital. we identified a cohort of people aged - years registered with participating general practices in england on january . we excluded patients (approximately . %) who did not have a valid nhs number. patients entered the cohort on january (date of first confirmed case of covid- in the uk) and were followed up until they had the outcome of interest or the end of the first study period ( april ), which was the date up to which linked data were available at the time of the derivation of the model, or the second time period ( may until june ) for the temporal cohort validation. the primary outcome was time to death from covid- (either in hospital or outside hospital), defined as confirmed or suspected death from covid- as per the death certification or death occurring in an individual with confirmed sars-cov- infection at any time in the period january to april . the secondary outcome was time to hospital admission with covid- , defined as an icd- (international classification of diseases, th revision) code for either confirmed or suspected covid- or new hospital admission associated with a confirmed sars-cov- infection in the study period. we selected candidate predictor variables on the basis of the presence of existing clinical vulnerability group criteria (table ) , associations with outcomes in other respiratory diseases, or hypothesised to be linked to adverse outcomes on clinical/biological plausibility and likely to be available for implementation. they are summarised in box and supplementary box a. we defined variables according to information recorded using read codes in general practices' electronic health records at the start of the study period. the exception to this was information on chemotherapy, radiotherapy, and transplants, which was based on linked hospital records. we randomly allocated % of practices to the derivation dataset, which we used to develop the models. we evaluated the models' performance in the remaining % of practices (the validation set). all models were fitted separately in men and women. the outcomes of interest are subject to competing risks. for the primary outcome of death from covid- , the competing risk is death due to other causes. for the secondary outcome of hospital admission, the competing risk is death from any cause before admission. we fitted a sub-distribution hazard (fine and gray ) model for each outcome to account for competing risks. individuals who did not have the outcome of interest were censored at the study end date, including those who had a competing event. for all predictor variables, we used the most recently available value at the entry date ( january ). we used second degree fractional polynomials to model non-linear relations for continuous variables (age, body mass index, and townsend material deprivation score, an area level score based on postcode ). initially, we fitted a complete case analysis by using a model within the derivation data to derive the fractional polynomial terms. for indicators of comorbidities and medication use, we assumed the absence of recorded information to mean absence of the factor in question. data were missing in four variables: ethnicity, townsend score, body mass index, and smoking status. we used multiple imputation with chained equations under the missing at random assumption to replace missing values for these variables. for computational efficiency, we used a combined imputation model for both outcomes. the imputation model was fitted in the derivation data and included predictor variables, the nelson-aalen estimators of the baseline cumulative sub-distribution hazard, and the outcome indicators (death from covid- and hospital admission with covid- ). we carried out five imputations. each analysis model was fitted in each of the five imputed datasets. we used rubin's rules to combine the model parameter estimates and the baseline cumulative incidence estimates across the imputed datasets. we initially sought to fit models using all predictor variables. owing to sparse cells, some conditions were combined if clinically similar in nature (such as rare neurological disorders). we examined interactions between body mass index and ethnicity and interactions between predictor variables and age, focusing on predictor variables that apply across the age range (asthma, epilepsy, diabetes, severe mental illness). we explored the use of penalised models (lasso) to screen variables for inclusion, but this retained all the predictor variables and most interaction terms. in line with the protocol, we subsequently removed a small number of variables with low numbers of events and adjusted (sub-distribution) hazard ratios close to (as these will have minimal effect on predicted risks) or with uncertain clinical credibility, defined as counterintuitive results in light of the emerging literature. lastly, we combined regression coefficients from the final models with estimates of the baseline cumulative incidence function evaluated at days to derive risk equations for each outcome. we used all the available data in the database. we did all model evaluation using the validation data with two separate periods of follow-up. the first validation study period was the same as for the derivation cohort: january to april . the second temporal validation covered the subsequent period of may to june . this was carried out with the same validation cohort except for exclusion of patients who died during january to april . in the validation cohort, we fitted an imputation model to replace missing values for ethnicity, body mass index, townsend score, and smoking status. this excluded the outcome indicators and nelson-aalen terms, as the aim was to use covariate data to obtain a prediction as if the outcome had not been observed to reflect intended use. we applied the final risk equations developed from the derivation dataset to men and women in the validation dataset and evaluated r values, brier scores, and measures of discrimination and calibration for the two time periods. where lower values indicate better accuracy. d statistics (a discrimination measure that quantifies the separation in survival between patients with different levels of predicted risks) and harrell's c statistics (a discrimination metric that quantifies the extent to which people with higher risk scores have earlier events) were evaluated at days (the maximum followup period available at the time of the derivation of the model) and days for the second temporal validation, with corresponding % confidence intervals. we assessed model calibration by comparing mean predicted risks with observed risks by twentieths of predicted risk for each of the validation cohorts. observed risks were derived in each of the groups by using non-parametric estimates of the cumulative incidences. additionally, we did a recalibration for the mortality outcome, using the method proposed by booth et al by updating the baseline survivor function based on the temporal validation cohort with the prognostic index as an offset term. we also applied the algorithms to the validation cohort for the first time period to define the centile thresholds based on absolute risk. we also defined centiles of relative risk (defined as the ratio of the individual's predicted absolute risk to the predicted absolute risk for a person of the same age and sex with a white ethnicity, body mass index of , and mean deprivation score with no other risk factors). we calculated the performance metrics in the whole validation cohort and in the following pre-specified we evaluated performance by calculating harrell's c statistics in individual general practices and combining the results using a random effects meta-analysis. patient and public involvement patients were involved in setting the research question and in developing plans for design and implementation of the study. patients were asked to aid in interpreting and disseminating the results. overall study population overall, practices in england met our inclusion criteria. of these, practices were randomly assigned to the derivation dataset and to the validation cohort. the practices had registered patients aged - years on january . we included of these in the derivation cohort, and the validation dataset comprised people. table shows the baseline characteristics of patients in the derivation cohort. of these patients, ( . %) were men and ( . %) were of black, asian, or other minority ethnic (bame) background. in the derivation cohort, ( . %) patients had a covid- related hospital admission and ( . %) had a covid- related death during the days' follow-up, of which ( . %) were recorded on the death certificate and ( . %) were based only on a positive test (and of these < were based on a test more than days before death). admissions and deaths due to covid- occurred across all regions, with the greatest numbers in london, which accounted for ( . %) of admissions and ( . %) of deaths. of those who died, ( . %) were male, ( . %) were bame, ( . %) were aged and over, ( . %) had type diabetes, ( . %) had dementia, and ( . %) were identified as living in a care home. the characteristics of the validation cohort were similar to those of the derivation cohort, as shown in supplementary tables a and b. in the first validation period ( january to april ), deaths and hospital admissions due to covid- occurred. in the second validation period ( may to june ), deaths and admissions due to covid- occurred. the variables included in the final models were fractional polynomial terms for age and body mass index, townsend score (linear), ethnic group, domicile (residential care, homeless, neither), and a range of conditions and treatments as shown in figure , figure , figure , and figure . these conditions and treatments were cardiovascular conditions (atrial fibrillation, heart failure, stroke, peripheral vascular disease, coronary heart disease, congenital heart disease), diabetes (type and type and interaction terms for type diabetes with age), respiratory conditions (asthma, rare respiratory conditions (cystic fibrosis, bronchiectasis, or alveolitis), chronic obstructive pulmonary disease, pulmonary hypertension or pulmonary fibrosis), cancer (blood cancer, chemotherapy, lung or oral cancer, marrow transplant, radiotherapy), neurological conditions (cerebral palsy, parkinson's disease, rare neurological conditions (motor neurone disease, multiple sclerosis, myasthenia, huntington's chorea), epilepsy, dementia, learning disability, severe mental illness), other conditions (liver cirrhosis, osteoporotic fracture, rheumatoid arthritis or systemic lupus erythematosus, sickle cell disease, venous thromboembolism, solid organ transplant, renal failure (ckd , ckd , ckd , with or without dialysis or transplant)), and medications (≥ prescriptions from general practitioner in previous six months for oral steroids, long acting β agonists or leukotrienes, immunosuppressants). figure and figure show the adjusted hazard ratios in the final models for covid- related death in the derivation cohort in women and men. figure and figure show the adjusted hazard ratios for the final models for covid- related hospital admission in the derivation cohort. supplementary figures a and b show graphs of the adjusted hazard ratios for body mass index, age, and the interaction between age and type diabetes for deaths and hospital admissions due to covid- (which showed higher risks associated with younger ages). supplementary figures c and d show fully adjusted hazard ratios for variables for the full model, including variables that were not retained in the final model (for example, adjusted hazard ratios close to one or those which lacked clinical credibility). other variables with too few events for inclusion were hiv, sphingolipidoises, short bowel syndrome, polymyositis, dermatomyositis, ehlers-danlos syndrome, biliary cirrhosis, hepatitis b and c, haemochromatosis, non-alcoholic fatty liver disease, chronic pancreatitis, drug misuse, asplenia, cholangitis, scleroderma, sjogren's syndrome, and pregnancy. supplementary figures e and f show fully adjusted hazard ratios for a combined outcome of either covid- related death or hospital admission. this gave very similar absolute risks to the hospital admission outcome. table shows the performance of the risk equations in the validation cohort for women and men over days for the main study period and for the temporal validation cohort evaluated from may to june . overall, the values for the r , d, and c statistics were similar in women and men. values for the mortality outcome tended to be higher than those for the hospital admission outcome. for example, in the first validation period, the equation explained % of the variation in time to death from covid- in women; the d statistic was . , and harrell's c statistic was . . the corresponding values in men were . %, . , and . . the results for the second validation period were similar except for covid- related admissions in women, for which the explained variation and discrimination were lower than for the first period (explained variation . %, d statistic . , and harrell's c statistic . ). supplementary tables c-f show the corresponding results by region, age band, and fifth of deprivation and within each ethnic group in men and women in both validation periods. performance was generally similar to the overall results except for age, for which the values were lower within individual age bands. figure shows funnel plots of harrell's c statistic for each general practice in the validation cohort versus the number of deaths in each practice in men and women in the first validation period. the summary (average) c statistic for women was . ( % confidence interval . to . ) from a random effects meta-analysis. the corresponding summary c statistic for men was . ( . to . ). we have developed and evaluated a novel clinical risk prediction model (qcovid) to estimate risks of hospital admission and mortality due to covid- . we have used national linked datasets from general practice and national sars-cov- testing, death registry, and hospital episode data for a sample of more than million adults representative of the population of england. the risk models have excellent discrimination (harrell's c statistics > . for the primary outcome). although the calibration for the hospital admission outcome was good in both time periods, some under-prediction existed for the mortality outcome in the second validation cohort, which improved after recalibration. the recalibration method could be used to transport the risk models to other settings or time periods with different absolute risks of covid- . qcovid represents a new approach for risk stratification in the population. it could also be deployed in several health and care applications, either during the current phase of the pandemic or in subsequent "waves" of infection (with recalibration as needed). these could include supporting targeted recruitment for clinical trials, prioritisation for vaccination, and discussions between patients and clinicians on workplace or health risk mitigation-for example, through weight reduction as obesity may be an important modifiable risk factor for serious complications of covid- if a causal association is established. although qcovid has been specifically designed to inform uk health policy and interventions to manage covid- related risks, it also has international potential, subject to local validation. one of the variables in our model (the townsend measure of deprivation) may need to be replaced with locally available equivalent measures, or some recalibration may be needed. previous risk prediction models based doi: comparison with other studies although similarities exist between our study and the recently reported analysis of risk factors from another english general practice database using a different clinical computer system, our project had a different aim-namely, to develop and evaluate a risk prediction model. we used a more comprehensive outcome (including deaths in patients with positive tests for sars-cov- ), a much wider range of predictors, and a more granular assessment of ethnicity and body mass index. our c statistic for mortality (> . ) is substantially higher than the previous study's reported value of . . other prediction models have been reported, although these focus on other outcomes of covid- , including risk of admission to intensive care or death following a positive test, or clinical decision tools that integrate biochemical and imaging parameters to aid diagnostis. however, most such studies are at high risk of bias, as they have been developed in highly selected cohorts, have limited transparency, are likely to have optimistic reported performance, or did not use covid- specific data. this study represents a substantial improvement on previously developed risk algorithms in terms of the size and representativeness of the study population, the richness of data linkages enabling accurate ascertainment of cases (including both in-hospital and out of hospital deaths) across the health network, and the breadth of candidate predictor variables considered. importantly, it analyses risks at the population level, rather than risks in people with confirmed or suspected infection, and may have relevance for shielding or other policies that seek to mitigate risk of viral exposure. complexities of modelling several complexities of modelling adverse risks from covid- in the general population warrant discussion. we used a general population approach which, although not able to incorporate all determinants of being infected, offers an overall estimate of risk of adverse outcomes from covid- that could be used in discussions between clinicians and patients about adjustment of lifestyle or occupational and behavioural factors that could limit viral exposure. our model predicts risks of "catching covid- and then having a severe outcome," on the basis of data collected during the first peak of the pandemic. the endpoint in this study examines a risk trajectory that comprises two elements: becoming infected, which is predominantly a function of behavioural/environmental factors including occupation, local infection rate, and numbers of social interactions; and risk of hospital admission and death due to the infection, which is arguably primarily driven by "vulnerability" (that is, biological/ physiological factors including age, sex, body mass index, comorbidities, and medications). although producing a prediction model for risk of "death if infected" is feasible in principle, this approach is not yet possible owing to the approach to testing in the uk and the context of an as yet incompletely quantified degree of asymptomatic background transmission. limited covid- testing data are available, but the difficulty is that no systematic community testing was done in the uk during the study period, so only patients unwell enough to attend hospital were tested. this means that a risk score developed in those who tested positive would overestimate risks of severe outcomes. as more widespread testing is done and those data become available, we will be able to update the model to take background infection rates into account and also model regional differences. although the absolute risk levels will of course change over time, depending on the incidence of the disease, our analysis over two validation time periods indicates that the relative risk measures and discrimination are likely to remain stable. secondly, the model estimates the absolute risk for a non-infected individual in the general population of becoming infected and then dying (or needing to be admitted to hospital) from the virus over a day period. although many more than people have died from covid- in the uk to date, when the denominator is a population of multi-millions, the absolute risk for most people may be low. therefore, when conveying this type of risk score to an individual, due emphasis is needed on the different meanings of absolute and relative risk. thirdly, the absolute risk of catching covid- depends not only on the incidence of the infection but also on the number of people one gets close to. for this reason, non-pharmacological interventions such as social distancing and shielding were introduced in the uk during the study period. we have included some measures of multi-occupancy, as we have factored care homes into the analysis. the data generated during the study period will therefore be affected by the uptake of harrell's c statistic twentieth of predicted risk at days twentieth of predicted risk at days interventions such as social distancing and shielding, intended to mitigate the risks of sars-cov- infection. this could result in underestimation of some model coefficients and hence underestimation of absolute risk in people who were shielded. also, as this is a prediction model derived from an observational study, the associations estimated for individual predictor variables should not be interpreted as causal effects. however, ethical questions must be considered regarding how the tools may be used. we have presented two ways of stratifying risk based on either absolute or relative risk measures with associated centile values, but the choice of whether to have a threshold (given that risk is a continuous measure), and if so what threshold, will depend on the purpose for which the risk assessment tool is to be used, the available resources, and the ethical framework for decision making. we have analysed this within the "four ethical principles" framework that is widely used in medical decision making. the four principles are autonomy, beneficence, justice, and non-maleficence. the new risk equations, when implemented in clinical software, are designed to provide more accurate information for patients and clinicians on which to base decisions, thereby promoting shared decision making and patient autonomy. they are intended to result in clinical benefit by identifying where changes in management are likely to benefit patients, thereby promoting the principle of beneficence. justice can be achieved by ensuring that the use of the risk equations results in fair and equitable access to health services that is commensurate with patients' level of risk. lastly, the risk assessment must not be used in a way that causes harm either to the individual patient or to others (for example, by introducing or withdrawing treatments where this is not in the patient's best interest), thereby supporting the non-maleficence principle. how this applies in clinical practice will naturally depend on many factors, especially the patient's wishes, the evidence base for any interventions, the clinician's experience, national priorities, and the available resources. the risk assessment equations therefore supplement clinical decision making and do not replace it. with these caveats, the predicted risk estimates can be used to identify people at higher risk, to inform shared decision making between healthcare professionals and service users, or for population level stratification. strengths and limitations of study our study has some major strengths, but some important limitations, which include the specific factors related to covid- along with others that are similar to those for a range of other widely used clinical risk prediction algorithms developed using the qresearch database. [ ] [ ] [ ] key strengths include the use of a very large validated data source that has been used to develop other risk prediction tools; the wealth of candidate risk predictors; the prospective recording of outcomes and their ascertainment using multiple national level database linkage; lack of selection, recall and respondent biases; and robust statistical analysis. we have used non-linear terms for body mass index and age. we examined interaction terms, which show increased risks at younger ages for adults with type diabetes. we also established a new linkage to the systemic anti-cancer therapy (sact) database for chemotherapy prescribed and administered in secondary care (which may not be recorded well in general practice software) to circumvent possible missing data for this important variable. specific limitations include the occurrence of shielding during the study period and that the study was conducted during the first phase of the uk epidemic. we have accounted for many risk factors for covid- mortality, but risks may be conferred by some rare medical conditions or other factors such as occupation that have not yet been observed or are poorly recorded in general practice or hospital data. in particular, the model does not include two important predictorsnamely, prevailing infection rate and personal social distancing measures. a lack of comprehensive testing has led to some missing data on covid- admissions and/or deaths, which means that development of a valid model for predicting death in people infected with sars-cov- is not yet possible. we acknowledge that absolute risks are changing during the course of the pandemic, so these should be interpreted with caution. however, we would expect predictors of risk, relative risk measures, and discrimination to be more stable over time, which is consistent with the results from our temporal validation. although this tool was modelled on the best available data from the first wave of the pandemic, it will be updated as further testing and outcome data accrue, immunity levels change, and (potentially) a vaccine becomes available. nevertheless, having a risk score available at this stage of the pandemic may be useful to identify people at high risk before a vaccine or treatment is available. we have reported a validation in each of two time periods using practices from qresearch, but these practices were completely separate from those used to develop the model. we have used this approach previously to develop and validate other widely used prediction models. when these have been further externally validated on completely different clinical databases, by ourselves and others, the results have been very similar. [ ] [ ] [ ] work is already under way to evaluate the models in external datasets across all four nations of the uk and to integrate the algorithms within nhs clinical software systems. this study presents robust risk prediction models that could be used to stratify risk in populations for public health purposes in the event of a "second wave" of the pandemic and support shared management of risk. we anticipate that the algorithms will be updated regularly as understanding of covid- increases, as more data become available, as behaviour in the population changes, or in response to new policy interventions. it is important for patients/carers and clinicians that a common, appropriately developed, evidence based model exists that is consistently implemented and is supported by the academic, clinical, and patient communities. this will then help to ensure consistent policy and clear national communication between policy makers, professionals, employers, and the public. impact assessment of nonpharmaceutical interventions against coronavirus disease and influenza in hong kong: an observational study effects of non-pharmaceutical interventions on covid- cases, deaths, and demand for hospital services in the uk: a modelling study clinical course and risk factors for mortality of adult inpatients with covid- in wuhan, china: a retrospective cohort study covid- and african americans clinical characteristics of deceased patients with coronavirus disease : retrospective study clinical course and mortality risk of severe covid- no commercial reuse: see rights and reprints variation in covid- hospitalizations and deaths across new york city boroughs writing group form obesity uk, obesity empowerment network, uk association for the study of obesity. obesity and covid- : a call for action from people living with obesity obesity is a risk factor for severe covid- infection: multiple potential mechanisms prevalence of co-morbidities and their association with mortality in patients with covid- : a systematic review and meta-analysis shielding from covid- should be stratified by risk prediction models for diagnosis and prognosis of covid- infection: systematic review and critical appraisal development and validation of qrisk risk prediction algorithms to estimate future risk of cardiovascular disease: prospective cohort study development and validation of qdiabetes- risk prediction algorithm to estimate future risk of type diabetes: cohort study development and validation of qmortality risk prediction algorithm to estimate short term risk of death and assess frailty: cohort study protocol for the development and evaluation of a tool for predicting risk of short-term adverse outcomes due to covid- in the general uk population transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (tripod): the tripod statement record working committee. the reporting of studies conducted using observational routinely-collected health data (record) statement the black report. penguin a proportional hazards model for the subdistribution of a competing risk explained variation for survival models a new measure of prognostic separation in survival data multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors assessing the performance of prediction models: a framework for traditional and novel measures comparing the predictive powers of survival models using harrell's c or somers' d temporal recalibration for improving prognostic model development and risk predictions in settings where survival is improving over time external validation of clinical prediction models using big datasets from e-health records or ipd meta-analysis: opportunities and challenges improvement in cardiovascular risk prediction with electronic health records non-invasive risk scores for prediction of type diabetes (epic-interact): a validation of existing models factors associated with covid- -related death using opensafely medical ethics: four principles plus attention to scope an independent external validation and evaluation of qrisk cardiovascular risk prediction: a prospective open cohort study an independent and external validation of qrisk cardiovascular disease risk score: a prospective open cohort study predicting the year risk of cardiovascular disease in the united kingdom: independent and external validation of an updated version of qrisk web appendix: supplementary materials imperial college london, london, uk we acknowledge the contribution of emis practices who contribute to qresearch and emis health and the universities of nottingham and oxford for expertise in establishing, developing, or supporting the qresearch database. this project involves data derived from patient level information collected by the nhs, as part of the care and support of cancer patients. the data are collated, maintained, and quality assured by the national cancer registration and analysis service, which is part of public health england (phe). access to the data was facilitated by the phe office for data release. the hospital episode statistics data used in this analysis are reused by permission from nhs digital, which retains the copyright in that data. we thank the office for national statistics (ons) for providing the mortality data. nhs digital, phe, and the ons bear no responsibility for the analysis or interpretation of the data. we express our gratitude to anne rigg, nisha shaunak, tom charlton, ana montes, claire harrison, susan robinson, david wrench, matthew streetly, omer bengal, doraid alrifai, and rajjinder nijjar for aiding the authors (notably pj and jhc) with the classification of agents on the sact dataset linkage used in this study and to david coggon for general comments on the study protocol and interpretation.contributors: jhc, cc, akc, rk, kdo, ph, and nm led study conceptualisation. all authors contributed to the development of the research question and study design, with development of advanced statistical aspects led by jhc, cc, rk, kdo, and akc. jhc, akc, cc, jb, and pj were involved in data specification, curation, and collection. jhc and akc developed, checked, or updated clinical code groups. jhc did the statistical analyses, which were checked by cc. jhc developed the software for the web calculator. all authors contributed to the interpretation of the results. akc and jhc wrote the first draft of the paper. all authors contributed to the critical revision of the manuscript for important intellectual content and approved the final version of the manuscript. the corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. jhc is the guarantor. funding: this study is funded by a grant from the national institute for health research (nihr) following a commission by the chief medical officer for england, whose office contributed to the development of the study question, facilitated access to relevant national datasets, and contributed to interpretation of data and drafting of the report. produces open and closed source software to implement clinical risk algorithms (outside this work) into clinical computer systems; cc reports receiving personal fees from clinrisk, outside this work; ah is a member of the new and emerging respiratory virus threats advisory group; pj was employed by nhs england during the conduct of the study and has received grants from epizyme and janssen and personal fees from takeda, bristol-myers-squibb, novartis, celgene, boehringer ingelheim, kite therapeutics, genmab, and incyte, all outside the submitted work; akc has previously received personal fees from huma therapeutics, outside of the scope of the submitted work; rl has received grants from health data research uk outside the submitted work; as has received grants from the medical research council (mrc) and health data research uk during the conduct of the study; cs has received grants from the dhsc national institute of health research uk, mrc uk, and the health protection unit in emerging and zoonotic infections (university of liverpool) during the conduct of the study and is a minority owner in integrum scientific llc (greensboro, nc, usa) outside of the submitted work; kk has received grants from nihr, is the national lead for ethnicity and diversity for the national institute for health applied research collaborations, is director of the university of leicester centre for black minority ethnic health, was a steering group member of the risk reduction framework for nhs staff (chair) and for adult care staff, is a member of independent sage, and is supported by the nihr applied research collaboration east midlands (arc em) and the nihr leicester biomedical research centre (brc); rhk was supported by a ukri future leaders fellowship (mr/s / ); kdo was supported by a grant from the alan turing institute health programme (ep/t / ); no other relationships or activities that could appear to have influenced the submitted work. the views expressed are those of the author(s) and not necessarily those of the nihr, the nhs, or the department of health and social care. data sharing: to guarantee the confidentiality of personal and health information, only the authors have had access to the data during the study in accordance with the relevant licence agreements. access to the qresearch data is according to the information on the qresearch website (www.qresearch.org).the lead author affirms that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.dissemination to participants and related patient and public communities: patient representatives from the qresearch advisory board have advised on dissemination of studies using qresearch data, including the use of lay summaries describing the research and its findings.provenance and peer review: not commissioned; externally peer reviewed. this is an open access article distributed in accordance with the terms of the creative commons attribution (cc by . ) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. see: http://creativecommons.org/licenses/by/ . /. key: cord- -p htdvrp authors: haldon, john; eisenberg, merle; mordechai, lee; izdebski, adam; white, sam title: lessons from the past, policies for the future: resilience and sustainability in past crises date: - - journal: environ syst decis doi: . /s - - - sha: doc_id: cord_uid: p htdvrp this article surveys some examples of the ways past societies have responded to environmental stressors such as famine, war, and pandemic. we show that people in the past did think about system recovery, but only on a sectoral scale. they did perceive challenges and respond appropriately, but within cultural constraints and resource limitations. risk mitigation was generally limited in scope, localized, and again determined by cultural logic that may not necessarily have been aware of more than symptoms, rather than actual causes. we also show that risk-managing and risk-mitigating arrangements often favored the vested interests of elites rather than the population more widely, an issue policy makers today still face. effective risk management and assessment require knowledge of past events to generate comparative risk scenarios. yet understanding the impacts of environmental stress on historical societies is an underdeveloped and fragmented field of study, with substantial disagreement among specialists. as a result, we cannot say with precision what constitutes an existential risk to a given historical society, i.e., a risk that could trigger the collapse of a political or cultural system. past human societies as a whole have been extraordinarily resilient in the face of severe challenges, but the configuration of social and political structures was always impacted in a number of ways, with substantial implications for development pathways (e.g., the different medium-term outcomes of the black death in england and france) (borsch , pp. - ; herlihy ) . historical case studies, therefore, can offer valuable guidance on present day issues in designing risk management strategies and sustainable policies (haldon and rosen ; rosen ) . detailed research into what, if any, role environmental challenges have played in the transformation of previous societies, including in conflict, migration, critical systems failure, and politics, is an essential requirement, along with grounded inquiry into socio-economic feedback loops. the study of complex historical societies can reveal how such challenges worked to transform structural relationships and daily life. but it can also tell us about what happened when the dust settled and how both leaders and governments and the members of society more broadly re-evaluated their situations. there are several key questions that historical case studies reveal. did people in the more distant past think about system recovery? we show that this depends significantly on system structure and capacity, nuanced and constrained by ideological assumptions. contemporaries perceived the scope of recovery within a state as a goal, but were focused on their own institutions, while considering recovery at a global scale was rare and conceptualized and promoted generally within the purview of religion. were people aware and did they respond, negatively or positively, to changes in their (perceived) social situation, impacts on the economy and finance, on beliefs and attitudes? the answers reflect different historically specific understandings * john haldon jhaldon@princeton.edu of a situation, both the type of society as well as their social status. moreover, in the pre-modern/pre-scientific world, moral and religious responses were as important as practical responses. did rulers and elites, or farmers and producers, implement policies that would mitigate risk and absorb future shocks? we suggest how some past societies, ranging from the later roman period down to the early modern world, responded, consciously or less so, to transformative and unpredictable environmental pressures. one common thread of these historical examples is a tendency of system recoveries to focus on elite-level actors, rather than all levels of society that were more numerous. however, as some of these cases demonstrate, a more just and equitable resiliency can lead to longer-term stability for the state and its institutions. how societies in the past responded to stress depends on three key sets of conditions: their complexity (the degree of interdependency across social relationships and structures), their institutional and ideological flexibility, and their systemic redundancy, all of which together determine the resilience of the system. these three conditions do not exist in isolation, but combine and recombine in innumerable historical configurations. historians must reduce this to ideal-typical models, since it is practically impossible to analyze them all. moreover, we must research particular historical case studies to illustrate these general patterns and to show how each case is subtly different from the next. 'resilience' is invoked in different ways within different disciplines. in historical research, it has largely played a role in work on collapse and adaptation, where societies are understood as complex adaptive systems and in which ecological models have been influential. since the basic structural dynamics of a societal system contribute to the types of collapse to which it may be subject, approaches to collapse, and resilience that unites structure and process are the best way forward in applying historical examples to contemporary planning initiatives with respect to environmental problems. this is a helpful approach, especially when allowances are made for individual human agency and belief systems (cumming and petersen ; haldon , building on ecological theory and formal resilience theory; also anderies ; berkes and ross ; gunderson and holling ) . resilience and the potential for a society to maintain cohesion and cultural continuity through periods of systemchallenging stress has costs. the question of how to distribute the costs of resilience, and the degree to which this might be built into any system, varies across time and cultural milieu. in the following, we examine several cases in past societies where we can observe ( ) both top-down and bottom-up responses to significant environmental challenges, how different sectors of society responded or reacted, and where we can detect positive as well as negative outcomes; ( ) the differential costs of resilience when states are faced with substantial economic and political challenges; and ( ) state-and society-level responses to pandemics and both planned and unintended consequences. pandemics: responses, resilience, and consequences by the year , the western roman empire had disappeared as a cohesive political state across western europe, but the eastern roman empire simultaneously flourished. the western half had divided into successor "barbarian" kingdoms, while the eastern roman empire was centered around the balkans, anatolia (mod. turkey), and the middle east. one of its main initiatives, the re-conquest of formerly roman regions in the west, such as north africa and italy, had begun well-with quick and cheap conquests. at the height of this re-conquest in the early s, the justinianic plague erupted across the mediterreanean world and europe. the justinianic plague was a pandemic of the bacterium yersinia pestis that remained active for over two centuries, c. - c.e. its impact on the eurasian population has been suggested as significant (i.e., causing the deaths of tens of millions), while its effects on human behavior from economics to culture and religion have also been described as pivotal (meier ; harper ; sarris ) . some have described it as the watershed moment separating a flourishing ancient world from the darker medieval world. while this master narrative is deeply problematic and has been challenged in recent years, individual outbreaks-the local impact of the plague during that two-century frameworkhave been entirely ignored ). the problematic 'catastrophe' narrative is simply copied and pasted across all of its outbreaks (sarris (sarris , ). yet if we examine particular outbreaks, even the destructive demographic narrative demonstrates the ability of the eastern roman state to react both immediately to the increased numbers of deaths, maintain vital administrative efforts, and continue its long-term political goals. the contemporary writers procopius of caesarea and john of ephesus recorded intricate details about the first outbreak in the city of constantinople in c.e. the number of deaths was significant, especially given the city's urban density, but hard numbers or even percentages are completely uncertain. procopius wrote that the outbreak lasted for three particularly virulent months and claimed that and then , people died per day. using these numbers, the number of deaths would have reached , by the end of the three months, while the population of constantinople was perhaps around , , so these numbers are impossible (procopius, wars , ii. xxii-xxxiii) . john of ephesus included even higher numbers, up to , deaths per day (john of ephesus). some modern scholars have suggested that % of the city's population died and more recently that % died, although these percentages are conjectures based on dubious assumptions (bratton ; stathakopoulos ; harper ) . recent attempts to use epidemiological modeling could not resolve this problematic question (white and mordechai ) . while sixth century romans would not have understood modern epidemiology, there are examples of self-isolation to stop the spread of plague. procopius recounts that many people who had the economic means locked themselves into their houses as a way to avoid infection. sometimes this attempt backfired, since entire houses would perish from plague and the dead would remain unburied for days (procopius, wars , ii. xxii-xxxiii) . during later outbreaks of plague in constantinople, elites, including the emperor himself, fled to nearby urban centers, while in other cases an entire city fled the arrival of the plague, decamping into the mountains (theophanes confessor, chronographia, am ; paul the deacon, history of the lombards , . ). in total, romans understood that the plague was an unexpected new phenomenon and adapted accordingly by trying to mitigate its spread. these lessons of how to prevent the spread of plague also suggest that people around the mediterranean world not only tolerated disruptions to their socio-economic activity, but even embraced them in some cases-preferring flight over simple isolation. despite the limited information at their disposal, communities and individuals attempted to weather the storm using whatever means were at their disposal. as twentieth century disease outbreaks suggest, even a much smaller percentage increase in deaths per day can quickly overwhelm the existing capacity of a modern state to bury its dead (crosby ; kilgannon ) . the imperial administration and the emperor justinian (r. - ), who took a personal hand in directing the response, recognized the overwhelming burial problem and responded quickly by appointing an administrator named theodorus to solve it. theodorus received money to hire people to help bury the poorer members of society, whose families could not afford or arrange for their burial. initially, they were placed in existing cemeteries, but as these filled up, theodorus' task force dug new larger trenches and pits to place the large numbers of bodies, a scene reminiscent of current day burials. theodorus, and by extension the state itself, could clearly adapt to different burial requirements both culturally and logistically as the situation required it. unlike in many cases today, the various political factions stopped their fighting and helped provide the manpower for theodorus to meet his grim new task (procopius). despite the significant mortality, these burial measures resolved the most pressing issue for constantinople: removing dead bodies to ensure other infections did not spread. the pandemic appears to have reached the rest of the mediterranean from egypt, where it arrived perhaps via the red sea and the indian ocean trade from central asia. from egypt, it reached constantinople through the regular grain shipments the city required to sustain its population. other port cities around the eastern mediterranean were struck soon afterwards (procopius). in the first few years after the outbreak in , the grain shipments from egypt appear to have been reduced due to some combination of a smaller population that required less food and the disruption of trade routes due to the pandemic. yet within five years the grain shipments had returned to their original quantity demonstrating that the state had bounced back and was able to provide resources as it had before the outbreak (zuckerman ) . in both of these short-term solutions, the state was remarkably flexible in its approaches to quickly meet an entirely new problem with limited resources at hand. most of the eastern roman budget was used to pay the army, with a smaller amount for the salaries of administrators (hendy , ff.). the roman government was able to quickly shift resources, establish an entirely new ad hoc administrative section, and ensure the continued flow of food to constantinople to prevent large-scale famine despite a significant epidemic outbreak. the short-term flexibility depended on systemic capabilities developed and maintained over centuries, which had long incorporated significant fluctuations to the supply of goods (rickman ). for instance, procurement of grain in egypt was likely a highly planned and regulated practice that had been operating for almost six centuries. both the government in constantinople and local officials in egypt prioritized grain procurement as a core systemic function. the grain shipments had significant built-in redundancies that would guarantee that enough grain reached constantinople despite unexpected (but known and relatively common) dangers such as shipwrecks, spoiled food, and unfavorable sailing weather slowing shipments. administrators often had significant experience in their positions adapting to these known problems and likely passed much of this to their replacements. this approach allowed the system to mitigate disruptions such as the outbreak of plague. however, major political disruptions such as the conquest of egypt by the persians several decades later had a far more significant result-the abrupt ending of all shipments from egyptwhich forced the government to scramble to search for new grain-supplying regions (haldon ) . despite the immediate disruption of plague, the eastern roman state was able to continue its existing military, political, and administrative goals. justinian's re-conquest of italy proceeded in slow, halting steps and was completed within a little over a decade; italy was then integrated back into the eastern roman state (heather ) . in the east, the status quo in the conflict with the persian empire was largely maintained and the roman state continued spending vast sums on the military for the next half century. taxes and resources continued to flow freely to the state with no evidence of plague-related problems (sarris ) . part of this was due, in all probability, to the redundancies built into the imperial system itself as in the case of the grain supply noted above. the late roman system was, by modern standards, inefficient and did not prioritize the market gains at every level or create massive, intricate supply lines. instead, it prioritized flexibility in using governmental resources, while assuming most needs would be met locally, a system that also generated duplication of activity in various administrative departments both centrally and in the provinces. within these constraints, and in part as a direct result of them, the roman state appears to have had little trouble returning to its baseline requirements even amidst the most significant epidemic to yet strike the known world. it is impossible to discern what the immediate demographic (let alone cultural, economic, and social) impact of this plague outbreak was. what is clear, however, is that the state weathered the storm through short-term efforts to protect the population of constantinople, while the populace itself created their own measures to prevent the spread of plague. while commerce and food supplies might have been stressed during the immediate outbreak, contemporary historians do not report on foot shortages; regardless, both of these returned to their pre-outbreak levels within just a few years. the outbreak did not change long-term political goals, the state's ability to fight wars, or raise taxes, and move resources around, and unlike other examples in this paper, there was no discernible change in power structures and elite-lower class labor relations. in fact, the plague outbreak seems not to have catalyzed any significant changes in the way the east roman state was managed. its demographic impact may have produced short-term shock waves across constantinople (and probably other parts of the empire about which we are less well-informed), but it hardly transformed the shape of the world or even the empire. a much better-known case of plague is the notorious black death of - , the second of the three pandemics of yersinia pestis that have impacted the world since the late ancient period (the third was a global pandemic from c. . claims of mortality rates of as much as % give the impression that this event must have been devastating for the societies affected. yet when we examine how different states and societies responded, we find that-without minimizing the terrible impact on people and communities-the medieval world did not grind to a halt, still less did a series of revolutionary transformations occur. indeed, the black death struck at the beginning of the hundred years' war, and in spite of its demographic impact both the kingdoms of england and france continued to field effective armies, even if there was a brief pause in hostilities (similar to contemporary calls for ceasefires in ongoing international conflicts in the context of covid- ). instead, some societal developments that were already under way accelerated while various groups within society responded by exploiting their situation and attempting to slow down, stop or otherwise control changes which they perceived as disadvantageous. the arrival of the black death in is sometimes associated with the eventual breakdown of 'feudalism' and the rise of capitalism, as the plague's high mortality and consequent shortage of labor was supposed to have dramatically changed the way the labor market was structured. in short, it is said to have challenged existing landlord-tenant relationships and the whole basis of serfdom (the legal binding of peasants to the lands of elites). yet in england serfdom was already declining by the time of the black death (hilton , pp. - ; campbell ) . moreover, landlords can respond to labor shortages in various ways-in eastern europe, for example, they increasingly tied labor to the land and imposed a more oppressive serfdom on the peasantry. from the later twelfth century on, reflecting elite demand in respect of consumption and increased expenditures, and backed by royal legislation, landlords in england demanded heavier labor services, reducing peasant holdings and thus the ability of the serfs to do anything more than maintain a bare subsistence. this last tendency was intensified, again with the backing of the state, in further repression of peasants' rights and freedoms following the demographic collapse caused by the pandemic. yet a trend toward labor shortages and demographic changes had set in well before this, following the so-called great famine ( - ). in this context, the black death was less a prime cause than a further exacerbating and intensifying factor (britnell , pp. - ; cohn ; whittle ; postan ; hilton hilton / hilton , . the black death and the accompanying increased pace of demographic contraction did not in themselves, therefore, lead either to the end or to an intensification of servile relationships. rather, a number of regionally nuanced factors played a role, among which one of the most important was the degree of class difference. as wealthier peasants resisted labor services, the poorer members of their communities could be employed-and exploited-as wage earners. indeed, one answer to the lords' demands for greater cash liquidity was the leasing of their estate land to wealthier peasant tenants. where this occurred the demand for wagelabor among the lessees of lords' estate lands rose, thus stimulating a clearer rural social hierarchy (britnell , pp. - , esp. - ) . the internal social structure of peasant communities was the key determining element. the contours of rural and urban society in england shifted considerably between the later twelfth and later fifteenth centuries, but no single factor was the cause. a reassessment of the black death reveals that its diverse indirect consequences were perhaps more important than the immediate, felt and visible impacts as reported by eyewitnesses, that tend to draw our attention. as such, the black death played an important role in accelerating existing trends, pushing some-but over several decades-beyond a threshold that then led to substantial change. when thinking about similar moments in the past or the present, it is the impact on the underlying structures of social and economic organization to which we need to pay attention. government intervention to address problems perceived from the center often fail to adequately take into account the range and complexity of causes underlying the issue addressed in state action. the ottoman empire from the late sixteenth to early seventeenth centuries ce provides a good illustration of the limits to resilience in a pre-industrial society. beginning from a small emirate in northwest anatolia ca. ce, ottoman rulers had by the s expanded their territory to three continents covering present day countries and built an empire that drew on administratively and geographically diverse sources of income. a key factor in the empire's resilience was thus its size. it developed systems to mobilize crucial resources from distant locations to provision its cities and military and to balance regional surpluses and deficits, including food, labor, timber, and strategic materials (e.g., gunpowder). the security provided by ottoman soldiers as well as legal and tax provisions encouraged the expansion of agriculture and the containment of mobile pastoralism. the empire seemed resilient to socio-environmental stress: when tested by a series of local droughts, shortages and famines during the s- s, ottoman officials were able to contain the damage by shifting tax burdens from the affected areas, ordering fixedprice sales of grain from other provinces, and in some cases arranging direct shipments from local or imperial granaries (white ; mikhail ; agoston ). the ottoman system of resource management could recover from small impacts, but multiple, continuous or repeated shocks pushed it towards breakdown, a situation that underlies the scale of crisis in the empire during the s- s. this was a period of major crisis triggered by environmental and human stressors followed by a protracted and intermittent recovery, in terms of population, agricultural production, political stability and military power. extended drought in central anatolia in - severely curtailed food output causing prices to double. near-famine conditions developed in some regions. this coincided with a series of extraordinarily cold winters, a combination that caused a major epizootic outbreak affecting sheep and cattle across anatolia, the crimea, and the balkans, eventually reaching hungary and central europe. this massive death of livestock deprived rural producers of a major source of wealth and subsistence, and deprived ottoman armies of a key source of protein (white with detail and sources). this was not, however, the only set of stress factors the empire faced, since it was at this time deeply enmeshed in the so-called long war ( - ) with the habsburg empire. therefore, instead of reducing taxation or providing relief supplies-the usual state response to droughts and famines-the state had to increase requisitions from the balkan and anatolian provinces that were the worst hit by escalating shortages and famines. this led to a major rural uprising, the so-called celâlî rebellion ( - ) (white ) . the combination of famine, violence, population displacement and disease generated a significant mortality crisis in parts of the empire-tax records from the s- s suggest up to % mortality in many parts of anatolia after the s (Özel , )-all of which produced a situation that induced a long-term shift in ottoman population and land use (Özel ; white and sources therein; ocakoğlu et al. ) . the history of the late s-early s is a good illustration of how political complexity could constrain resilience in a situation where a combination of factors amplified the negative consequences of state activities, in this case a focus on revenue, provisioning, and military mobilization at the expense of diversification and risk reduction in during environmental stress. lack of agricultural diversification in semi-arid regions, dependence on provinces near the imperial capital for extraordinary taxes and requisitions, lack of spare capacity in dealing with both simultaneous military and infrastructural emergencies, all stressed the system to capacity. these factors combined with difficulties of supplying and pacifying inland regions, poor overland communications and the interaction of famine, flight, insecurity, and disease. together with inadequate public health systems that might mitigate epidemic disease impacts, the result was a severe and sustained population loss leading to an unstable balance between village agriculture and mobile pastoralism-and ultimately a fracturing of state management and control over provincial economies. while the imperial system as a whole held together, the cost of the vulnerabilities inherent in the ottoman system were borne disproportionately by the least privileged social groups. just as in a markedly different context with the eastern roman empire (the case discussed below), this potentially undermined the resilience of the entire socioeconomic and political system, since these groups formed the backbone of pre-modern economies. while these groups possessed a remarkable degree of resilience within the limits imposed by environmental and political conditions, when both acute social and environmental problems combined they could neither sustain their own livelihoods nor shoulder the burdens of imperial economies and ecologies. the role of elites and particular groups in social classes are central to the resilience of 'states' themselves. historically 'states' have tended to be dominated-managed and administered-by members of a power-elite drawn from a socially privileged sector of society. members of such groups are generally concerned as much with their own interests as they are with those of the state or ruler they serve, although some pre-modern states have been able to maintain, for a while, an establishment entirely divorced from the vested interests of their society. the eastern roman empire was undoubtedly one of the most sophisticated states in western eurasia, with a complex and effective fiscal and administrative structure that maximized resource extraction and maintained a balance of power between the state, elites and provincial society. by the early tenth century, after two centuries of rebuilding following the shock of the early islamic conquests, it was entering a period of expansion in both the balkans and the middle east. in parallel, there had evolved a social elite of office holders and landowners who gradually achieved a near monopoly on the senior and middling posts in the military and civil administration. it was their task to implement government policy in the provinces, but their increasing wealth and status meant that by the tenth century they were also a potential source of opposition to the central government. the tension between these two aspects of the east roman state revealed itself in the efforts of the elite to expand its wealth in land, generally at the expense of village communities who were a key element in the state's finances and provided the core of the provincial armies, thus jeopardizing the effectiveness of the central state administration itself. it should be noted that this is a structural problem common to all pre-modern/pre-capitalist systems: states must rely on elites to maintain themselves, yet those elites, whatever their origins, also develop vested interests that compromise or jeopardize those of the state. the ways this relationship has worked itself out historically varied enormously. the problem remains today, of course, although 'elites' are generally both more complexly structured and sectorized (national, international and multinational), and state autonomy-and thus state economies-compromised by global economic factors: the interests of international finance and investment capital rarely overlap neatly with those of nation states, as variations in the markets, particularly during moments of global crisis, daily illustrate. in the s a series of natural disasters disastrously impacted the agriculture of the western anatolian provinces, giving the wealthy or powerful opportunities to absorb further properties into their estates (kaplan ; svoronos ; mcgeer ; morris ). in - ce there occurred a particularly severe winter in the balkans and anatolia, combined with a series of extremely poor reduced harvests. the result was later remembered (just as was the similarly disastrous famine that preceded the black death in europe) as the 'great famine'. in their description of the resulting social crisis, legal sources distinguish between the 'powerful' (military and civil officials paid in gold coin by the central government and possessing liquid assets) and the 'weak' (peasant farmers and laborers whose livelihood and ability to pay their taxes depended on their harvest). the latter were forced into selling their land for food or money to survive. it is not surprising, therefore, that a subsistence crisis provided those with the necessary resources an opportunity to exchange liquid assets for large tracts of land. to protect its own interests the state had to intervene through legislation that attempted to stop this increased inequality. however, the legislation the state promulgated to try to deal with the issue ultimately had only short-term success, chiefly because the people it depended on to implement these laws were themselves the people against whom the legislation was directed. instead, the government was eventually forced to adopt the tactics of the elite, converting public land into imperial estates in order to secure the income derived from them. the great famine of - did not create social change but did accelerate it. its impact was twofold. it presented the better-off with an opportunity to exploit peasants whose livelihood had been destabilized by the severity of the winter. the state reacted by guaranteeing its survival by effectively seizing control itself of the private land of the free peasantry, who thus found themselves reduced to dependency either by the state that should have protected them or by those who sought to dispossess them. in the state's attempt to restrain its own elites, it destroyed the fortunes of the more vulnerable members of society. antioch (today antakya, turkey) in the northeastern corner of the mediterranean, was perhaps the third largest city in the flourishing sixth century eastern roman empire. as an administrative center, it connected the empire's heartlands in the s. balkans and nw. anatolia to its wealthy province of syria. antioch was also an economic and cultural powerhouse, serving as a major node in the network of eastern mediterranean cities and boasting rare amenities for the time (e.g., night lighting), a religious hub (e.g., it was one of the five key centers of christianity), as well as a major economic node in the dense network of eastern mediterranean cities. over the sixth century, antioch suffered a series of major disasters that included at least six destructive earthquakes, a major fire, a foreign sack and several smaller raids, the deportation of many of its citizens to a foreign country, and at least four outbreaks of plague. yet, antioch survived these disasters and remained a major regional center. since antioch could not be allowed to collapse for the political, ideological and administrative reasons summarized above, the central government at constantinople ensured its survival by providing it with constant infusions of resources. these included tax remittances, outside workers and resources for reconstruction after each major disaster, and even an extensive imperial initiative to remodel the city (procopius, buildings , . ). the government complemented its material efforts with public relations campaigns to lure immigrants to antioch. it employed propaganda, such as changing the city's name to theoupolis, literally "the city of god", used in official communications and on newly minted coins. the government instituted additional popular measures such as free food rations, bringing in grain from egypt to the city where it was baked and handed out (part of the same institution that supplied constantinople with grain, discussed above). such practices served to maintain antioch's population by attracting poor and refugees from the greater region and keeping those destitute from the recent disasters in the city. from the perspective of the central government, these actions were successful in maintaining antioch's position, its primary objective in the region. its actions, however, had indirect consequences. although specific economic costs are unknown, the governmental support of antioch was a decades-long process that strained the central government, forcing it to reduce the amount of resources available for other central initiatives such as its foreign policy (e.g., wars in italy). at the same time, drawing immigrants to antioch weakened social cohesion in the city. frequent riots among different groups within the city occurred, frequently based on religious or other markers of difference. the government's pumping of money into antioch had other unintended consequences. the city was part of several trade networks, importing goods from across the mediterranean, such as north african pottery, and exporting them to the inland syrian cities. at the same time, it exported its own local goods, such as olive oil, across the mediterranean. this oil was produced in antioch's hinterland. the disruption of these economic networks damaged antioch's commercial relationships and crashed the local economy, worsening the economic condition of the residents of antioch's hinterland and likely establishing a feedback loop that further encouraged migration to the city. the former trend of rural economic expansion stopped, and although antioch slowly re-established its inter-regional connections, its local olive oil industry never recovered (mordechai ) . whether governmental policy on antioch could resolve these more complex stresses remains unknown. a decade into the seventh century, antioch was lost temporarily to a persian attack and then permanently to islamic conquest. such regional disruptions prevent us from assessing the longer-term results of the imperial government's policies on antioch. although the causes and even the process of the islamic conquest remain unclear and debated, the disruption wrought by the sixth century disasters and the imperial response likely contributed to the eastern provinces' vulnerability over the longer term. a second case study from the same period reveals a different story with the participation of the central government. the history of the eastern regions of the later roman empire in the fifth-seventh centuries ce offers a good example of how one sector within a social system lost even as society as a whole benefitted. between ce and lasting until ca. / ce (izdebski et al. ) , climate change led to increased winter precipitation that enabled a profound transformation in late antique society. the increased winter rainfall, crucial for cereal cultivation in most of the eastern mediterranean, shifted the ecological frontier between arid areas and the land suitable for cereal cultivation. although the introduction of farming on these marginal lands often required sophisticated irrigation and water-harvesting techniques (in the negev desert, for example), the changing climate permitted an extension of agricultural production into regions previously left uncultivated or used only as pastures. eastern roman society was remarkably efficient at turning the environmental challenge of increased precipitation into an economic opportunity, and several regions saw an unprecedented expansion of agriculture and rural settlement (izdebski et al. ) . this increase in agricultural production in turn encouraged contemporaries to specialize in cash crops, such as olives and vine, visible in a number of microregions across the eastern mediterranean (see for instance england et al. ; varinlioğlu ; izdebski a, b) . however, this climate-related economic growth also contributed to a shift in the balance of control over resources between urban and rural populations in anatolia and the levant. in the late fifth and sixth centuries, new agricultural regions had fewer cities, while large villages dominated as opposed to densely populated urban centers. although the dramatic expansion into marginal lands occurred on the fringes of urbanized regions, new settlements encroached on lands that were located beyond the traditional hinterlands of existing urban communities, and escaped their political and economic control (e.g., niewöhner ) . this inaugurated a decline in the political and cultural role of many cities, contributing to an ongoing transformation of economic relationships between town and country, and elites and the producing population (brandes and haldon ) . climate changes allowed rural populations to accelerate this process and become more independent from cities while at the same time leading to a more general urban decline. here we see a situation in which different social groups depend on various ecological niches. environmental stress led to varying consequences for each of them, leading to a renegotiation of economic relationships and to changes in the balance of power. in the late roman world, urban populations were relatively privileged, both in their access to local resources and the state support they often received. in theory, therefore, they should have fared better through greater access to resources. but their position was actually undermined by what would generally be understood as a beneficial environmental change: their control and dominance over rural populations diminished as a result of the expansion of rural settlement onto new agricultural land, made possible to a large extent by a shift in precipitation patterns. a number of conclusions or lessons can be drawn from these examples, all of which involved states or societies that were complex, possessed institutional and ideological flexibility, and a degree of systemic redundancy, which is to say, overlapping institutional arrangements that in many instances could permit elements of one facet of social organization or state structure to fail without jeopardizing the system as a whole. first, the costs of resilience have never been shared evenly among the different components of complex social systems. political complexity has always had advantages and disadvantages in maintaining resilience in the face of environmental stress. resilience in one social group or institution-the rapid return of its baseline function, lifestyle and living conditions-influenced other groups within the same society. but to understand the potential for all groups to receive just returns we need to understand the connections between different social groups and their environments. the underprivileged or less powerful have always been the most likely to bear the costs of societal resilience to environmental stress. the examples also illustrate the point that even where efforts were made consciously to assist a whole community-for example, in the case of sixth century antioch-the outcome could often generate unpredictable changes that could impact that community negatively. not only do shortterm strategies that sustain a state or a specific bundle of vested interests not necessarily promote longer-term societal resilience, they can also increase longer-term structural pressures leading to systemic crisis. to a degree this applies also in the case of the black death in england, where the unforeseen longer-term result of the short-term responses was an increase in peasant social mobility and rural social diversity, facilitating a transformation of the labor market and social class relations. this forced the social elite to realign itself with new fiscal and market circumstances in order to protect its socio-economic dominance and at the same time inaugurated a longer-term challenge to its monopoly on local and central political office. nevertheless, there are examples where states possess sufficient systemic resilience to overcome very serious short-term challenges without further exacerbating existing inequalities, as the example of the justinianic plague suggests. while no modern public health infrastructure existed, the imperial government readjusted quickly to meet two immediate needs: burial of the dead and the provision of food supplies for the population. by fulfilling these two duties and when combined with individual self-isolation, constantinople sustained itself and rebounded within just a few years. flexibility built into the imperial administration provided the empire with the capacity to manage a number of (expected) short-term problems, from lower crop yields to catastrophic weather, that allowed the state to mitigate sudden changes to market supply. these short-term strategies secured the stability to meet the state's longer-term strategic aims such as the re-conquest of italy, as well as continuity of administration and government. secondly, our case studies illustrate that states, even where ideologically predisposed to assist the poorest or weakest in society (e.g., the christian eastern roman empire and the islamic ottoman empire) often resolved challenges, both short-term and longer-term, by pushing increased costs for state survival onto those sectors of the society least able to resist-an inevitable consequence of pre-existing systemic inequalities. in the process, however, the state in the past also unconsciously transformed the relationships between the central power and central and regional elites. the great famine in the eastern roman empire that followed the severe winter of / ce tells exactly this story. this case, along with the example of the catastrophic droughts suffered by the ottomans in the s ce, illustrates how a central government damaged its own economic base when the largest social group, the lower strata of the society, lost its resilience to environmental stress. their vulnerability undermined the foundations of the state's fiscal economy: lack of resilience on the part of the less privileged directly impacted the resources available to the state through taxes and requisitions, thus weakening the core functions of the central government, including the military. a third conclusion is, therefore, that the greater the degree of baseline inequality at the outset of a crisis, the less resilience there is in the system as a whole, the more uneven the distribution of the resilience burden, and the greater the potential for post-solution breakdown of a given social order. social elites, as identifiable groups, generally survive societal crises and transformations because they have a vested interest in preserving their position and generally retain the resources to do so. while individual members or sectors of elites may die or lose their positions of wealth and power, as a visible societal group, they are often still around and at the top of the heap when the dust settles. naturally, there are exceptions: rapid revolutionary events such as in france between and or russia in - can result in the effective removal of much of an established super-elite even if individual members of the old establishment changed sides and joined the revolution. but it is not uncommon, even where a major shift in political and ideological control takes place, for substantial elements of an established elite to adapt to radically changed circumstances and retain their basic socio-economic advantage, even where they are no longer the ruling element. this occurred, for example, with the middling elite of sasanian iran after the islamic conquest in the s- s (pourshariati ; morony ) ; it was true of the middling elites of the western roman empire (halsall ) ; it was true of local balkan elites after the ottoman conquests in the fourteenth-fifteenth centuries ce (kunt ; inalcık ) ; it was just as true of traditional elites throughout the middle east, iran and as far as afghanistan after alexander's conquests in the fourth century bce (adams ; erskine ) . did people in the past think about system recovery? on a global scale, no. but sectorally, as in the case of central governments with the means at their disposal, the answer varied: the degree of the problems they faced, the nature of economic and social class relationships, and, to some extent, the overall ideology and its key motifs were central factors. did people understand the challenges and respond appropriately? here we have to say, it depends: representatives of religions tended to have a more global outlook (i.e., for people of the same faith), but their responses tended to be moral rather than practical (i.e., prayers to stop the calamity), and when they were practical they were inevitably local and shortterm (i.e., famine relief, for example). ruling elites could respond, but they tended to react primarily to perceived threats to their own survival. this might well embrace the entire state, but as we have seen in some of the examples above, such responses were generally compromised by elite interests, as in the tenth century medieval eastern roman empire. furthermore, they usually were able to respond only to the immediately perceived problem-which may have been just a symptom of deeper issues. and were they able to implement policies to mitigate future risk? yes, but again, for example in the case of antioch, often with unforeseen consequences for the longer term. these historical examples illustrate clearly that policy makers and political leaders today generally have a much greater appreciation of threats and risks, potential and actual-which puts them in a far better position to plan for system recovery. their ability to respond appropriately, however, continues to be determined by a range of cultural/ideological, political/structural, and economic factors, including elite interests, many of which work to constrain or even discourage the implementation of potentially effective policies that could address both short-term challenges and mitigate future risks. this becomes particularly acute when these elite interests do not align with those of the far more numerous non-elites, who are significantly more likely to be affected, as we have seen. in this context, we would suggest that the tendency towards structural socio-economic imbalance in responses to environmental challenges must be a question that future policy planners place at the heart of their calculations. because this sort of imbalance has generally been the case until now does not mean it has to be the case in the future-but in what circumstances this would not occur is an important, largely unanswered, and generally avoided question, except as a statement of general rhetoric. ensuring a more equal and just distribution of the costs and thus extending resilience more evenly across all social-economic sectors would appear to be the obvious solution towards a more sustainable future for any complex socio-political system. the hellenistic kingdoms guns for the sultan: technology, industry, and military power in the ottoman empire robustness, institutions, and large-scale change in social-ecological systems: the hohokam of the phoenix basin panarchy and community resilience: sustainability science and policy implications the black death in egypt and england: a comparative study towns, tax and transformation: state, cities and their hinterlands in the east roman world britain and ireland - : economy and society the identity of the plague of justinian the agrarian problem in the early fourteenth century rodney hilton's middle ages. an exploration of historical themes america's forgotten pandemic the influenza of unifying research on social-ecological resilience and collapse historical landscape change in cappadocia (central turkey): a palaeoecological investigation of annually-laminated sediments from nar lake a companion to the hellenistic world society and environment in the east mediterranean ca - ce. problems of resilience, adaptation and transformation demystifying collapse: climate, environment, and social agency in pre-modern societies barbarian migrations and the roman west, - , cambridge medieval textbooks the fate of rome. climate, disease, and the end of an empire rome resurgent: war and empire in the age of justinian the black death and the transformation of the west bond men made free. medieval peasant movements and the english uprising of english peasantry in the later middle ages the brenner debate: agrarian class structure and economic development in pre-industrial the economic expansion of the anatolian countryside in late antiquity: the coast versus inland regions a rural economy in transition. asia minor from late antiquity into the early middle ages. taubenschlag foundation the environmental, archaeological and historical evidence for regional climatic changes and their societal impacts in the eastern mediterranean in late antiquity les hommes et la terre à byzance du vie au xie siècle: propriété et exploitation du sol. publications de la sorbonne as morgues fill, n.y.c. to bury some virus victims in potter's field. the new york times the sultan's servants: the transformation of ottoman provincial government - the "justinianic plague": the economic consequences of the pandemic in the eastern roman empire and its cultural and religious effects nature and empire in ottoman egypt environment and society during the long late antiquity late antique archaeology rejecting catastrophe: the case of the justinianic plague the justinianic plague: an inconsequential pandemic iraq after the muslim conquest the powerful and the poor in tenth-century byzantium: law and reality aizanoi and anatolia. town and countryside in late late antiquity', millenium a -year multi-proxy sedimentary record of climate change from lake Çubuk (göynük population changes in ottoman anatolia during the th and th centuries: the 'demographic crisis' reconsidered mgh srl . paul the deacon, hanover postan mm ( ) essays on medieval agriculture and general problems of the medieval economy decline and fall of the sasanian empire. the sasanian-parthian confederacy and the islamic conquest of iran. i.b. tauris, new york procopius ( a) buildings. in: dewing hb (ed) loeb classical library. the macmillan co civilizing climate: social responses to climate change in the ancient near east the justinianic plague: origins and effects oxford stathakopoulos dc ( ) famine and pestilence in the late roman and early byzantine empire: a systematic survey of subsistence crises and epidemics les novelles des empereurs macédoniens concernant la terre et les stratiotes. centre de recherches byzantines, athens theophanes the confessor, chronographia, c. de boor trade, craft and agricultural production in town and countryside in southeastern isauria the climate of rebellion in the early modern ottoman empire a model disaster: from the great ottoman panzootic to the cattle plagues of early modern europe. in: varlık n (ed) plague and contagion in the islamic mediterranean modeling the justinianic plague: comparing hypothesized transmission routes peasant politics and class consciousness: the norfolk rebellions of and compared centre de recherche d'histoire et civilisation de byzance. monographies key: cord- -kusqyumn authors: alves, t. h. e.; souza, t. a. d.; silva, s. d. a.; ramos, n. a.; oliveira, s. v. d. title: underreporting of death by covid- in brazil's second most populous state date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: kusqyumn the covid- pandemic brings to light the reality of the brazilian health system. the underreporting of covid- deaths in the state of minas gerais (mg), where is concentrated the second largest population of the country, reveals government unpreparedness, as there is a low capacity of testing in the population, which prevents the real understanding of the general panorama of sars-cov- dissemination. the goals of this research are to analyze the causes of deaths in the different brazilian government databases (arpen and sinan) and to assess whether there are sub-records shown by the unexpected increase in the frequency of deaths from causes clinically similar to covid- . a descriptive and quantitative analysis of the number of covid- deaths and similar causes was made in different databases. ours results demonstrate that the different official sources had a discrepancy of . % between these data referring to the same period. there was also a . % increase in sars deaths in , when compared to the average of previous years. finally, it was shown that there was an increase in the rate of pneumonia and respiratory insufficiency (ri) by . % and . %, respectively. in conclusion, there is an underreporting of covid- deaths in mg due to the unexplained excess of sars deaths, respiratory insufficiency and pneumonia compared to previous years. respiratory syndrome (sars) that emerged in december in china and became a pandemic in march due to its high infection and mortality rates [ ] [ ] [ ] . covid- was the official name given by the world health organization (who) to the disease caused by the new coronavirus of (sars-cov- ) [ ] . the first epicenter of covid- was observed in wuhan, the capital of hubei, china, in december based on the several pneumonia cases notifications [ ] . since then, covid- has rapidly spread around the world and, as of may th , more than . million cases of the disease have been confirmed, causing over , deaths worldwide. [ ] . of this total, brazil has reported more than , cases and over , deaths, according of coronavírus brasil database [ ] . covid- is classified according to the symptoms' severity. patients with the mild form ( % of the cases) present fever, dry cough, chills, malaise, muscle pain,and sore throat. patients with moderate form present fever, respiratory symptoms, and radiographic characteristics. severe patients ( % of the cases) manifest dyspnea (> bpm), low oxygen saturation (< %) and low pao /fio ratio (< mmhg), and may evolve to a respiratory failure, septic shock, and multiple organ failure [ ] [ ] [ ]. furthermore, increased age and the presence of comorbidities, such as hypertension, diabetes, and coronary disease, are associated with mortality in covid- patients [ ] [ ] . the accurate diagnosis of covid- is carried out by searching the genetic material of the virus and, in a complementary way, by imaging methods. computed tomography and radiographs can identify lesions in the lungs due to viral multiplication [ ] [ ] . . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint laboratory confirmation is essential for the timely management of cases to avoid the spread of transmission. however, brazil is far below the ideal number of tests for covid- , as there are not enough laboratory inputs to understand the overall panorama of the virus' spread. furthermore, confirmatory molecular tests depend on the availability of imported reagents,whichare globally scarce, and on government investments that prioritize this strategy. this scenario leads brazil to have a delay in the number of covid- cases and deaths confirmations. these aspects become more aggravated when the patient evolves to death, because the effectiveness of tests, for these cases, is even more difficult. in addition, the recommendation is to collect blood and sputum to perform the culture, since these samples have a higher viral loadconsidering the studies done to date [ ] . the difficulty regarding death registration have been also presented in the state of minas gerais, which, by the end of april , had suspected deaths notifications, of which ( %) had not yet been confirmed or discarded [ ] . thus, it is possible to state that there is a disparity between the real number of covid- deaths and the numbers that are reported in different brazilian sources of information, since not all deaths have been tested for confirmation or exclusion and are potentially being confirmed by others causes than covid- . the present study aims to analyze the death causes in the notary records and in the brazilian national disease notification system records, and thus evaluate the subregistries and the possible increase in the frequency of deaths with clinically compatible causes to covid- in the minas gerais territory. . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . [ ] . for this study, the notary offices records were analyzed from january to april of in the state of mg. additionally, to assess the deaths excess in this period according to their causes, information from the sinan was accessed referring to the range of years to . the notary data were obtained from the civil registry transparency portal, which is a free access platform developed to provide information about births, marriages, and deaths. due to the covid- pandemic, these data are being grouped in the special sections covid- and the covid registral panel made available on arpen database [ ] . the information presented here (accessed on / / ) is based on death certificates (dd), presenting only one cause for each death certificate [ ] . to evaluate sub-registrations in the different information systems in brazil, sinan data were collected through the infogripe platform of the oswaldo cruz foundation (fiocruz) infogripe database [ ] is an initiative that aims to monitor and present alert levels for reported cases of severe acute respiratory syndrome (sars) in sinan [ ] . the data in this system were compared with the notary data. on this platform, the records of sars and covid- were selected on / / according to . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . we also evaluated the death excess from causes that present clinical compatibility with covid- , according to the following etiology: severe acute respiratory syndrome (sars), pneumonia, respiratory insufficiency (ri), sepsis (sepsis/septic shock), indeterminate causes (deaths related to respiratory diseases, but not conclusive), and other deaths (all other types of deaths that are not listed above) [ ] . the data were collected and analyzed in spreadsheet by descriptive statistics and presented in raw numbers, relative frequency, and central tendency measures. to assess the death excess per epi week, the average, minimum and maximum values of deaths from the years to were calculated and confronted with diseases that presented changes in the pattern of distribution in the fourth quarter of . all graphs were prepared using graphpadprism software (graphpad software, inc. san diego, ca). a total of covid- deaths were identified in the notary records and this number differs in . % of the deaths registered in the sinan at the same period (table ) . . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. regarding to the elevation in the rates of pneumonia and respiratory insufficiency from january to march compared to the same period in , it was around . % and . %, respectively. . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint the study points out divergences of information between different death registrations systems. the important increase in sars deaths that started earlier than those from covid- , in epidemiological week , is also highlighted, suggesting the underreporting of covid- deaths in the state of minas gerais. the covid- situation is particularly challenging because, besides being a new and unprecedented disease, it is also capable of triggering other conditions, such as pneumonia and sars, which can be characterized as the main cause of death. in other words, the covid- may be the underlying cause, that is, it may not be the direct cause of death that has been registered. in this perspective, there is a subjectivity bias, since the physician can attest or not the death from covid- according to his clinical knowledge without the need of laboratory tests [ ] . this finding corroborates with data from hubei, china and northern italy, where mortality calculations were adjusted for the biases of preferential verification, symptomatic and severe cases, and delay in death records. an increase in the mortality rate was found, which confirms the existence of underreporting covid- deaths in those regions [ ] . in relation to underreporting in brazil, the ministry of health (mh) reports that the number of under-reported deaths is low according to the mortality information system (mis), because states and municipalities are advised to include deaths from covid- , either confirmed cases or only suspects, in the system as a priority, in order to advance analysis of these cases [ ] . however, our results show that there is a significant underreporting of the occurrences by covid- , given the excess of sars deaths. . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org / . / . / another issue that should be analyzed is that although the civil registry information center takes into consideration both confirmed deaths and suspects, the mh discloses in its reports only the laboratory proven covid- deaths [ ]. however, suspect deaths need to be considered in the count, even though it is noted that they have not been confirmed. this is stated because it is known that many of these deaths will not be able to be analyzed, given the difficulties in collecting, transporting, and wrapping the post-mortem samples. thus, if they are not mentioned, there will be a relaxation of the real situation in brazil and, consequently, in the state of mg. the brazilian mh also points out that in the same death certificate more than one cause of death can be described, so that the record of covid- can be associated with other diseases. however, the civil registry transparency portal presents these causes separately, even those included or registered in the same death certificate. thus, one cannot only add up the deaths made available on the portal by the different diseases, because they would generate false over-notification. a thorough investigation must be made when considering each death and the causes that were cited in the death certificate [ ] . however, according to the hierarchical criteria exposed in the civil registry transparency portal, only one cause of death is selected to make the count, and not all the causes present in the same death certificate [ ] , which validates the data exposed in this platform and the information presented here. it is worth noting that the different systems of deaths registration of the government, such as the municipalities and states, are not fully connected and that several of them depend on manual labor to be registered. this is capable of causing discrepancies and delays in data traffic and, consequently, in the production of timely and reliable information. . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint who has been advising countries on the need to expand laboratory testing capacity as a strategy to overcome the pandemic [ ] . this action will enable the real knowledge of a population's immunity, providing reliable statistics for a better understanding of the circulation of the disease. consequently, strategies to control the pandemic and even the relaxation of non-pharmacological measures, such as social isolation and quarantines, may be proposed. in brazil, a network formed by referenced laboratories was established to help fight covid- [ ] . however, the country is far below the optimal number of tests for covid- , as there are not enough tests to have a reliable panorama of the real number of cases and deaths. this scenario leads brazil to have a delay in accounting the records of covid- . in conclusion, our results reveal that covid- deaths in the state of minas gerais are higher than the official statistics presented. in view of these aspects, it is necessary to expand brazil's diagnostic capacity, which will allow us to recognize the real number of covid- deaths and cases in minas gerais. thanks to gabriela geraldo mendes and adélio tiago da mota for the collaborations. thanks to the department of collective health of the faculty of medicine of the federal university of uberlândia for the encouragement. . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint who director-general's opening remarks at the media briefing on covid- clinical characteristics of coronavirus disease in china examining the effect of social distancing on the compound growth rate of sars-cov- at the county level (united states) using statistical analyses and a random forest machine learning model. public health the covid- epidemic. tropical medicine & international health covid- coronavirus pandemic painel de casos de doença pelo coronavírus (covid- ) no brasil pelo ministério da saúde unique epidemiological and clinical features of the emerging novel coronavirus pneumonia (covid- ) implicate special control measures clinical, laboratory and imaging features of covid- : a systematic review and meta-analysis. travel medicine and infectious disease sobre a doença clinical features of patients infected with novel coronavirus in wuhan, china. the lancet clinical course and risk factors for mortality of adult inpatients with covid- in wuhan, china: a retrospective cohort study. the lancet recomendações de uso de métodos de imagem para pacientes suspeitos de infecção pelo covid- essentials for radiologists on covid- : an update radiology scientific expert panel diretrizes para diagnóstico e tratamento da covid- informe epidemiológico coronavírus international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted portal da transparência: registro civil do brasil em associação dos registradores de pessoas naturais (arpen) do brasil database monitoramento de casos reportados de síndrome respiratória aguda grave (srag) hospitalizados data (owd) database statistics and research coronavirus pandemic estimation of sars-cov- mortality during the early stages of an epidemic: a modelling study in hubei boletim epidemiológico especial it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review)the copyright holder for this preprint this version posted may , . is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint key: cord- - qqyb sd authors: pane, masdalina; imari, sholah; alwi, qomariah; nyoman kandun, i; cook, alex r.; samaan, gina title: causes of mortality for indonesian hajj pilgrims: comparison between routine death certificate and verbal autopsy findings date: - - journal: plos one doi: . /journal.pone. sha: doc_id: cord_uid: qqyb sd background: indonesia provides the largest single source of pilgrims for the hajj ( %). in the last two decades, mortality rates for indonesian pilgrims ranged between – deaths per , pilgrims over the -week hajj period. reasons for high mortality are not well understood. in , verbal autopsy was introduced to complement routine death certificates to explore cause of death diagnoses. this study presents the patterns and causes of death for indonesian pilgrims, and compares routine death certificates to verbal autopsy findings. methods: public health surveillance was conducted by indonesian public health authorities accompanying pilgrims to saudi arabia, with daily reporting of hospitalizations and deaths. surveillance data from were analyzed for timing, geographic location and site of death. percentages for each cause of death category from death certificates were compared to that from verbal autopsy. results: in , , indonesian undertook the hajj. there were deaths, equivalent to , deaths per , pilgrim years. most pilgrims died in mecca ( %) and medinah ( %). there was no statistically discernible difference in the total mortality risk for the two pilgrimage routes (mecca or medinah first), but the number of deaths peaked earlier for those traveling to mecca first (p= . ). most deaths were due to cardiovascular ( %) and respiratory ( %) diseases. a greater proportion of deaths were attributed to cardiovascular disease by death certificate compared to the verbal autopsy method (p< . ). significantly more deaths had ill-defined cause based on verbal autopsy method (p< . ). conclusions: despite pre-departure health screening and other medical services, indonesian pilgrim mortality rates were very high. correct classification of cause of death is critical for the development of risk mitigation strategies. since verbal autopsy classified causes of death differently to death certificates, further studies are needed to assess the method’s utility in this setting. each year, muslims from all over the world undertake the hajj pilgrimage to and in saudi arabia [ ] . in recent years, over million people from countries undertook the hajj annually, including over , people from indonesia, the world's most populous muslim majority country. performance of the hajj and its rites is physically demanding [ ] . extreme physical stressors such as sun exposure, heat ( °c during the day and °c at night), thirst, crowding, steep inclines and traffic congestions over a prolonged period of time ( days per pilgrim over the -day hajj season) increase health risks. since pilgrims tend to be older and many have medical comorbidities [ ] , these factors exacerbate existing risk for ischemic and congestive cardiovascular disease, fluid and electrolyte abnormalities, and respiratory and other infectious diseases including emerging diseases such as middle east respiratory syndrome coronavirus [ ] . in the last two decades, the mortality rate of indonesian pilgrims, excluding years in which disasters such as stampedes occurred, fluctuated between - deaths per , persons during the ten-week hajj period [ ] . few countries have published pilgrim mortality rates, but compared to where they are available, the indonesian rate is much higher [ ] . for example, in , the hajj mortality rate amongst isfahani pilgrims from iran was per , pilgrimages [ ] . in , the mortality rate for all iranian pilgrims was per , pilgrimages, and in , per , [ ] . even compared to the yearly mortality rate in indonesia, the mortality rate for hajj pilgrims ranged between , and , per , per year; by comparison, the indonesian estimated national crude death rate was per , in [ ] . since , indonesian authorities have provided medical services to hajj pilgrims. pre-departure health screening and vaccination as well as temporary medical clinics staffed by indonesian doctors in saudi arabia were introduced progressively. mortality surveillance was also established to reduce mortality and aid in administrative processes for life insurance claims. prior to , the patterns, causes of death and factors associated with mortality were not well understood for indonesian pilgrims. data collected were limited to basic demographic characteristics and general cause of death. cause of death could only be obtained from the hospital death certificate or the flight doctor's records if a death occurred in the community. these were based on clinical examination and any available laboratory tests as documented in the patient medical record, but they lacked detailed information about the cause of death or the patient's pre-existing conditions. given the hugely elevated mortality risk relative to the general indonesian population, information on cause of death and underlying health conditions is critical to enhance pilgrim health management and to prevent excess mortality. in , the ministry of health introduced an additional surveillance tool -verbal autopsy -to better understand the causes of pilgrim mortality. verbal autopsy allows for the systematic investigation of probable causes of death through structured questionnaires [ ] . the world health organization (who) advocates the use of verbal autopsy in situations where only a fraction of deaths occur in hospitals or in absence of vital registration systems [ ] . previous studies have shown that verbal autopsy improves diagnosis of cause of mortality and the method continues to be used in various countries [ ] [ ] [ ] . this study describes the findings from the mortality surveillance conducted during the hajj in ; the year in which verbal autopsy was introduced. we explore both personspecific and site-specific factors associated with mortality, and compare the cause of death from the routine death certificate to the newly introduced verbal autopsy method. ethics approval for the study was obtained from the ministry of health national institute of health research and development ethics review committee with approval number lb. . /ke/ / . written consent was not obtained from the patients involved, but this was waived by the ethics review committee as mortality data were analysed anonymously. in , , indonesians undertook pilgrimage during hajj. the majority were from java ( %) and sumatra ( %) islands, with small proportions from kalimantan ( %), sulawesi ( %) and other ( %) islands. the majority of pilgrims, anecdotally reported to be %, joined one of the governmentsponsored hajj pilgrimage travel services that include days of travel. the remainder joined private more expensive pilgrimage travel services that are of shorter duration ( days) and provide better accommodation and services. a number of preventive and curative healthcare services were available to all hajj pilgrims prior to and during their travels in saudi arabia. pre-departure, each indonesian pilgrim was required to visit a government healthcare facility for a medical check-up and to receive a pocketbook outlining their health conditions, medications and vaccination status. pilgrims were mandated to receive meningococcal vaccine and were advised to receive influenza vaccine before their departure. flights were chartered by the indonesian government to accommodate - pilgrims. each flight had one doctor and two nurses to accompany the pilgrims. these healthcare workers treated, triaged and conducted health surveillance for pilgrims. hospitalizations and incidental reporting of any deaths occurring outside a healthcare facility were notified daily to the indonesian public health team based in saudi arabia during the hajj. saudi authorities made first aid posts, health centres and hospitals available. in mecca, six hospitals were set up during pilgrimage, while in medinah, pilgrims could access established hospitals. in addition to the saudi facilities, indonesian health authorities set up field hospitals in mecca and medinah specifically for indonesian pilgrims. indonesia also sent specialist doctors including internists, pulmonologists, cardiologists and psychiatrists, public health workers, nurses, pharmacists and sanitarians to support the pilgrimage. all deaths were reported from the flight doctor or the saudi hospital to the indonesian public health team located in saudi arabia during the hajj. mortality data collected by the indonesian public health team were entered into a database for data analysis. standard variables in the database included name, age, sex, home address, employment, flight group, time and place of death, date of arrival into saudi arabia, cause of death as per the hospital or flight doctor death certificate. in , the indonesian ministry of health mandated a verbal autopsy form to be filled out by the flight doctors accompanying pilgrims. the verbal autopsy form was developed based on the standards established by who and adapted to hajj pilgrimage needs [ ] . the verbal autopsy form obtained detailed information about medical history, signs and symptoms, and other circumstances regarding the death from family or friends who travelled with the deceased. in most cases, interviews were conducted with a combination of the treating physician, the deceased person's spouse or pilgrims in the same flight group. the form aimed to increase the specificity of the cause of death by elucidating medical history and recent events. the flight doctors were trained in administering the verbal autopsy form prior to departure from indonesia and were required to complete it within the week of a pilgrim's death. once the form was completed, the indonesian public health team stationed in saudi arabia sent it to the ministry of health in indonesia for analysis and determination of cause of death. the form was analysed separately by two trained staff at the ministry of health to determine the cause of death. if discordant, the staff compared their analyses to achieve consensus. the cause of death based on this verbal autopsy method was then recorded in the database and compared to that reported by the hospital or flight doctor death certificate. we used counts and proportions to describe demographic characteristics of indonesian pilgrims and fatalities. the chi-square test and chi-square test for trend were used for the analysis of categorical data. to compare the two methods for establishing cause of death (verbal autopsy and death certificate), we compared the percentages for each cause of death category using mcnemar's test. all , indonesian pilgrims were aged years or more, where the majority ( %) were aged between and years ( table ) . most pilgrims were female ( %, table ). according to the pre-embarkation medical assessment, % of pilgrims were classified as high risk due to underlying health conditions such as diabetes, hypertension, other chronic diseases or if they were years or older. pilgrims in , indonesian pilgrims died during the hajj in saudi arabia. the overall mortality rate was deaths per , pilgrimages. mortality rates were highest in those ≥ years of age ( per , pilgrims), and rates significantly increased with increasing age (p< . , table ). mortality rates were higher in males ( deaths per , pilgrims) compared to females ( deaths per , pilgrims, p< . , table ). for the deaths in - year old age-group, were male and were female. according to the death certificates, three died due to cardiac arrest and one due to asphyxia in a patient with active tuberculosis. most deaths occurred in mecca (n= , %), followed by medinah (n= , %) and jeddah (n= , %). the majority ( %) of deaths occurred in hospital, but a large proportion of deaths also occurred in pilgrims' apartments/sleeping areas ( %). most ( %) deaths occurred during pilgrims' active hours between am and pm: mortality rates were higher in the afternoon (possibly due to the heat) and early morning ( figure a ). weekly mortality rates increased in week , exceeding the expected crude mortality rate (cmr) of per , per day [ ] , and remained high until the end of the -week hajj period (figure b ). in week , the number of deaths started to decrease. however, since the overall number of pilgrims also decreased, the mortality rates remained very high since some of the indonesian pilgrims who were hospitalized died in the later weeks of the hajj. for pilgrims on route , the number of deaths increased sharply four weeks after arrival into saudi arabia by which time the pilgrims had already reached mecca (figure b ). for pilgrims on route , the number of deaths peaked in the third week after their arrival into saudi arabia, at which stage the pilgrims were in mecca (figure c ). the number of deaths remained large thereafter until the end of the hajj for those undertaking route . there was no statistical difference in the number of deaths occurring for pilgrims undertaking route ( out of , pilgrims) compared to those on route ( out of , , p= . ). however, the trends from week to week during the hajj period were significantly different between pilgrims on route compared to those on route (χ for trend= . , p= . ). most deaths were due to cardiovascular diseases and respiratory diseases (table ) . a greater proportion of deaths were attributed to cardiovascular disease by the flight doctor or hospital death certificate ( %) compared to the cause of death ascertained using the verbal autopsy method ( %, p< . ). significantly more deaths had unspecified cause based on the verbal autopsy method ( %, versus , p< . ). as part of the demographic data collected for each pilgrim, height and body mass were recorded. based on the hospital/flight doctor death certificate, of the ( %) pilgrims who died had body mass index ≥ . indicating obesity for asian body types [ ] . of these, died due to cardiovascular disturbances, due to respiratory illness, due to metabolic disturbances, had undefined sudden death and due to trauma (neck fracture). this study describes the demographics, patterns and causes of mortality in indonesian pilgrims in . nearly one third of pilgrims undertaking the hajj in were considered high risk due to underlying health conditions or their age. mortality rates were found to be greatest in males and in those aged ≥ years, in whom most deaths were attributed to cardiovascular and respiratory diseases. studies from other countries with pilgrims found similar trends, including a preponderance in male mortality [ , ] . these findings highlight the special characteristics of the hajj compared to other mass gathering events. many muslims wait decades for the opportunity to perform the hajj, and by the time they receive the chance, they may have a multitude of age-related health concerns [ ] . correct classification of deaths is critical to target preventive interventions and provide health services [ ] . this study compared cause of death according to the flight doctor and death certificate records to the newly introduced verbal autopsy method. fewer deaths were attributed to cardiovascular diseases using verbal autopsy but this method resulted in a greater number of deaths having ill-defined cause of death. verbal autopsy method may have reduced misclassification by removing pressure from clinicians having to extrapolate cause of death in situations where it may have been ill-defined or unclear. however, this hypothesis warrants further investigation. since the verbal autopsy method is dependent on the skills of the field personnel collecting the data, the timing to limit recall bias and the method is most suited to diseases with specific symptoms and presentation [ , ] , the use of verbal autopsy for hajj mortality surveillance should be further evaluated. based on both the death certificates and verbal autopsy categories, cardiovascular disease was the leading cause of indonesian pilgrim mortality in . performance of obligatory rites during the hajj constitutes stressful exercise which is not generally recommended by doctors for those with ischemic heart disease, hypertension or heart failure as such exercise may increase the risk of heart attacks [ ] . this risk may be further elevated in the heat, where dehydration may lead to increase in body temperature and heart rate, and a decrease in cardiac output [ ] . an iranian study showed that when patients with severe cardiovascular disease were prohibited from attending the hajj and other patients with cardiovascular disease were provided with appropriate medications and monitoring during pilgrimage, mortality rates were significantly lower than those for other pilgrims [ ] . this supports the need for careful pre-departure health screening, exclusion of the severely ill, provision of appropriate drug therapies or increased physical exercise prior to departure, and monitoring during pilgrimage to reduce mortality. one-third of indonesian pilgrim mortality was attributed to respiratory diseases. pneumonia is a common illness that is life-threatening to the elderly, especially those with comorbidities such as diabetes or hypertension [ ] . a number of studies have shown that pneumonia is the primary cause of critical illness during the hajj and that etiologies include gramnegative organisms, streptococcus pneumoniae, and mycobacterium tuberculosis [ , , ] . for tuberculosis patients, the physical stressors of the hajj may increase the risk of severe illness and mortality, and the intense crowding during the pilgrimage may increase the risk of disease transmission. a recent study found that % of malaysian pilgrims had a significant increase in immune response to quantiferon tuberculosis assay antigen post-hajj compared to pre-hajj [ ] . since indonesia is a high-burden country for tuberculosis [ ] , pre-departure screening should continue to exclude those with active disease. other potential public health measures to reduce mortality due to respiratory diseases include increasing coverage of influenza vaccine and pneumococcal vaccine. such measures were applied by iranian public health authorities in , which halved the incidence of respiratory diseases and decreased the mortality rates from per , in to per , in [ ] . deaths amongst indonesian pilgrims traveling on route who went to mecca first peaked earlier than those traveling on route . most deaths among indonesian pilgrims occurred in the middle-latter weeks of the hajj period during the stay in mecca and afterwards in arafah-mina. obligatory rites conducted at holy sites in mecca and arafah-mina are known to involve intense physical activity [ ] . these may have been too strenuous for some pilgrims, especially older or relatively sedentary pilgrims. surprisingly, % of deaths occurred in the accommodation provided to pilgrims during the hajj. this highlights that despite the presence of an accompanying health team, not all patients were referred to hospital prior to critical stages of illness. one limitation of this study is that further details about deaths occurring in accommodation were not available for analysis. lack of data limited other important analyses including deaths by health risk status and type of travel service (government or private) used. the rates of mortality as well as causes of death may differ based on these categories and may impact recommendations for intervention. pattern of diseases among visitors to mina health centers during the hajj season, h ( g) hajj: health lessons for mass gatherings common health hazards in french pilgrims during the hajj of : a prospective cohort study the epidemiology of hajj-related critical illness: lessons for deployment of temporary critical care services* comparison of mortality and morbidity rates among iranian pilgrims in hajj how to reduce cardiovascular mortality and morbidity among hajj pilgrims: a multiphasic screening, intervention and assessment verbal autopsy: current practices and challenges potential and limits of verbal autopsies factors associated with place of death in addis ababa applying verbal autopsy to determine cause of death in rural vietnam accuracy of who verbal autopsy tool in determining major causes of neonatal deaths in india verbal autopsy standards: ascertaining and attributing causes of death famine-affected, refugee, and displaced populations: recommendations for public health issues appropriate body-mass index for asian populations and its implications for policy and intervention strategies causes of admission to intensive care units in the hajj period of the islamic year clinical and temporal patterns of severe pneumonia causing critical illness during hajj a review of data-derived methods for assigning causes of death from verbal autopsy data physical activity and stroke in british middle aged men influence of graded dehydration on hyperthermia and cardiovascular drift during exercise tuberculosis is the commonest cause of pneumonia requiring hospitalization during hajj (pilgrimage to makkah) high risk of mycobacterium tuberculosis infection during the hajj pilgrimage country profile -indonesia indonesian pilgrims suffer high mortality rates despite predeparture screenings, accompanying medical teams and the availability of specialized health services during the hajj. this study highlights the importance of surveillance during the hajj to understand the health risks and strengthen the evidencebase on which policy can be developed [ ] . further studies are needed to assess verbal autopsy's utility in this setting. the role of the accompanying health teams as first responders needs to be reviewed to determine how they can best reduce indonesian pilgrim mortality. an evaluation of the current mortality surveillance system is also warranted to ensure that the data collected appropriately serves the public health purpose of reducing pilgrim mortality. lastly, lessons need to be learnt from other countries including their hajj mortality patterns and risk mitigation strategies. key: cord- -fc tmx authors: ciminelli, gabriele; garcia-mandicó, sílvia title: covid- in italy: an analysis of death registry data date: - - journal: j public health (oxf) doi: . /pubmed/fdaa sha: doc_id: cord_uid: fc tmx background: there are still many unknowns about covid- . we do not know its exact mortality rate nor the speed through which it spreads across communities. this lack of evidence complicates the design of appropriate response policies. methods: we source daily death registry data for municipalities in italy’s north and match them to census data. we augment the dataset with municipality-level data on a host of co-factors of covid- mortality, which we exploit in a differences-in-differences regression model to analyze covid- -induced mortality. results: we find that covid- killed more than . % of the local population during the first wave of the epidemic. we also show that official statistics vastly underreport this death toll, by about %. next, we uncover the dramatic effects of the epidemic on nursing home residents in the outbreak epicenter: in municipalities with a high share of the elderly living in nursing homes, covid- mortality was about twice as high as in those with no nursing home intown. conclusions: a pro-active approach in managing the epidemic is key to reduce covid- mortality. authorities should ramp-up testing capacity and increase contact-tracing abilities. adequate protective equipment should be provided to nursing home residents and staff. there are still many unknowns about covid- . we do not know its true mortality rate nor the speed through which it spreads across communities. this lack of evidence complicates the design of appropriate response policies. the case of the uk is illustrative. the government first opted for bare minimum mitigation. it then drastically reversed course after microsimulations by the imperial college covid- response team showed that that strategy could have resulted in hundreds of thousands of deaths. this paper provides first-hand evidence on some of these questions. we focus on italy's north, for long the world's worst affected area by covid- . using daily death registry data from the italian statistical agency, we first quantify the death toll from and by covid- and show that many deaths went unrecorded in official statistics. covid- was responsible for the death of about people-more than . % of the local population-during the first wave of the epidemic (from mid-february to mid-may ). its mortality is vastly undercounted in official statistics. a plausible estimate suggests that true covid- deaths might have been % higher than what is officially reported. at the peak of the epidemic, one additional death went undetected for each officially recorded covid- fatality. we then uncover large regional differences in mortality, which can be partly explained through the different approaches taken to deal with the epidemic. in the veneto region, which embraced mass testing, contact tracing, and at-home care provision, covid- -induced mortality was, respectively, three and six times smaller than in neighboring emilia-romagna and lombardy. we close the paper by discussing the management of nursing homes in lombardy, italy's covid- outbreak epicenter. nursing homes have been in the spotlight across europe and beyond for being possible hotspots of contagion and deaths. our results suggest that living in a nursing home may have significantly increased the probability of dying during the covid- epidemic. we find that in municipalities with a percentage point higher share of people living in a nursing home, mortality among the elderly was about twice as high as in those with no nursing home in town. this result helps to rationalize the serious undercounting of covid- fatalities in official statistics, which do not include deaths in nursing homes. we source daily death registry data from istat, the italian statistical agency. (accessed th may ). the data provide information on daily deaths by age and gender for the vast majority of italian municipalities, for the period january to may, for the years - . our focus is on the eight regions of italy's north (emilia-romagna, friuli-venezia giulia, liguria, lombardy, piedmont, trentino-south tyrol, valle d'aosta and veneto), which together account for roughly half of italy's population and about % of all covid- fatalities (as of may ). our data cover almost % of all municipalities in these eight regions and about % of their population. table a in the appendix reports descriptive statistics on the coverage of the dataset. figure below takes a first look at the data by comparing deaths in (red line) against those in each of the five preceding years. soon after the detection of the first covid- community case (vertical line), deaths increased rapidly. they jumped from about at the end of february to over at the end of march. from the beginning of april, as the government social distancing policies began to bear an effect, deaths slowly started to decrease and returned to normal levels around mid-may. our sample thus spans the entire first wave of the covid- epidemic in italy. we then source census population data from istat and match them to death registry data to analyze covid- induced mortality. census data provide information on the resident population by age and gender, as of january, for the years - . (accessed th april ). we impute values by using growth rates. we also source official covid- fatality data from protezione civile, to compare officially reported mortality to the mortality calculated from death registry data. lastly, we source municipality-level data on a host of cofactors of covid- mortality, such as population density, commuting, digital employment, air pollution and several demographic characteristics. we also source data on health care management, in particular nursing homes, for the lombardy region. deaths before the covid- outbreak were following fairly closely deaths in . assuming that this trend would have continued in the absence of covid- , we calculate a measure of excess deaths by subtracting deaths in to deaths in . to empirically evaluate the impact of covid- on mortality, we embed the concept of excess deaths in a differences-in-differences regression model, which allows us to control for a host of potentially confounding factors. more detail on the methodology is provided in the online appendix. using the concept of excess deaths, we calculate that the virus might have caused the death of about people-more than . % of the local population-during the first wave of the epidemic. next, we compare excess deaths in to official covid- fatality data. figure below plots the official daily number of covid- fatalities (dashed blue line) and excess deaths in relative , as recorded in municipal death registry data (solid red line). strikingly, excess deaths were higher than official fatalities throughout the end of april, suggesting that covid- deaths might have been vastly underreported in official statistics. what is the scale of underreporting? excess deaths include both indirect and direct covid- deaths, while official data only cover direct deaths. to gauge the importance of indirect deaths, we consider the regions of italy, which are not included in our sample because they were relatively unaffected by covid- . these display ratios of excess deaths to official covid- fatalities that are similar to the regions in our sample. assuming that these regions correctly detected all covid- deaths, we obtain a measure of indirect deaths as a share of the population, which we use to infer the number of indirect deaths in the eight regions in our sample. using this approach, we estimate that indirect deaths might account for - % of all deaths attributable to covid- . discounting indirect deaths, we calculate that covid- deaths may have been % higher than what was officially reported. at the peak of the epidemic (around end-march), deaths may have been underreported by about a factor of two. that is, for each officially recorded death, an additional one went undetected. what can explain underreporting? in italy, guidelines for the classification of covid- fatalities vary by region, but in most cases, deaths outside hospitals are not counted in official statistics. anecdotal evidence suggests that, as the health system struggled with a surge in demand, many old patients died of covid- at home or in elderly care facilities, without being tested. , this hypothesis seems plausible since the extent of underreporting was higher at the peak of the epidemic. limited testing capacity may be a complementary reason. while excess deaths from registry data decreased sharply after peaking at the end of march, official covid- fatalities peaked later. they also did not fall as sharply, suggesting that the authorities may have progressively ramped up testing capability. in may, the number of official fatalities was even higher than that of excess deaths. this may be due to lags in the official reporting of fatalities relative to the fatality date. data presented so far were aggregate for the entire sample of the municipalities covered in the dataset. in fig. below, we uncover variation in the extent of the epidemic both across different municipalities and over time. we assign different colors to municipalities, based on the mortality rate. municipalities with a mortality rate within pre-epidemic 'normal' levels ( - daily deaths per inhabitants) are denoted with blue. those with a mortality rate between and daily deaths per inhabitants are marked by different shades of orange, while red colors indicate those with very high rates (above per inhabitants). at the start of the epidemic, the mortality rate was generally within normal levels, except for a few municipalities. by march, two large outbreaks in the lombardy region had become apparent: one in the south, around the town of codogno (the epicenter), and another in the north, around alzano lombardo (a town close to bergamo). while deaths increased across most municipalities during the unfolding of the outbreak, the increase around these two clusters was much higher. along the border between lombardy and neighboring regions, death rates were high in both emilia-romagna and piedmont, while they stayed low in veneto. next, we use a differences-in-differences method to confirm the patterns observed in fig. and quantify the effects of covid- on the mortality rate by region. we estimate that the pandemic resulted in almost three deaths per inhabitants per day in lombardy, by far the region in which covid- had the highest effect on mortality. across the four regions bordering lombardy, we estimate the lowest impact in veneto, at around . daily death for every inhabitants. that is about three times less than trentino-south tyrol and piedmont and four times less than emilia-romagna-all sharing a long border with lombardy-and suggests that the bold approach adopted by the veneto region in managing the epidemic may have sensibly reduced covid- mortality there. details on the estimation and additional results are reported in the online appendix. the veneto region has been widely praised for its timely and proactive response to the pandemic. much has been written about the 'veneto model', , , and we only summarize its main characteristics: (i) mass testing, including testing of asymptomatic cases; (ii) at-home testing and care provision and (iii) tracing and quarantining contacts. both ramping up the testing capacity and developing a comprehensive tracking system have been recognized as essential elements to coping with the epidemic. , testing and providing care at home works towards preventing the spread of the virus. in a testament to the success of its strategy, on may, the veneto region registered zero new infections, while lombardy had more than . the veneto model provides useful lessons for policy makers around the world on how to manage new outbreaks of covid- in the future. living in a nursing home may have increased the chances of dying for the elderly next, we explore gender and age differences in covid- mortality. details on the estimation are provided in the online appendix, while here we only summarize the main findings. we first show that mortality increases exponentially with age, but at much higher levels for men, confirming that gender differences play a crucial role in understanding the distribution of risk from the epidemic. at the same time, underreporting greatly increases with age and is particularly high among older women. why do we observe so many undetected deaths among the elderly, and particularly so among women? all available anecdotal evidence points to nursing homes. , with a large number of residents sharing common spaces and having close contacts with multiple staff members, nursing homes may have acted as hotbeds of contagion. moreover, as in italy nursing homes do not qualify as medical centers, they were heavily understaffed and unprepared to deal with the crisis, lacking protective equipment for staff and emergency care equipment for infected patients. in lombardy, these inherent characteristics may have been particularly aggravating, as the regional authority decided to relocate covid- positive patients with mild symptoms from hospitals to nursing homes. since italy does not include fatalities in nursing homes in its covid- statistics, they likely account for a big chunk of the undetected deaths. we zoom in on lombardy, the worst affected region, to test whether covid- had an additional effect on mortality in municipalities with a higher share of people living in nursing homes (see the online appendix for details on the estimation). in fig. below, we compare the mortality effect of covid- in municipalities where % of the elderly population lives in nursing homes with no nursing home in town. the results suggest that living in a nursing home may have significantly increased the chance of dying during the covid- epidemic for both men and women. strikingly, we find that covid- mortality was twice as high in municipalities with a high proportion ( %) of the elderly living in a nursing home as in those municipalities with no nursing home in town. this paper provides first-hand evidence on the true death toll from and by covid- using italy's north-for long the world's worst affected area by covid- -as a case study. we show that covid- caused the death of more than . % of the local population during the first wave of the epidemic. we also show that many deaths go unrecorded in official statistics. a plausible estimate suggests that true deaths were about % higher than what was officially reported during the first wave of the covid- epidemic in italy. this undercounting is much more severe for the elderly, and in particular, for women. since italy does not include fatalities in nursing homes in its covid- statistics, they likely account for a big chunk of the undetected deaths. our analysis quantifies the catastrophic effects of covid- in nursing homes. in municipalities with a percentage point higher share of the elderly population living in a nursing home, mortality was about twice as high as those with no nursing home in town. our analysis shows that many of them could have been prevented through better preparedness. providing adequate protective equipment is key to protecting nursing homes' residents and staff. even more essential is the need to identify and isolate positive cases, and prevent staff from going to work if they are affected. the concept of excess deaths has been used by several statistical offices around the globe to quantify covid- mortality. a notable example is the uk statistical office (ons), which has produced relevant work comparing excess mortality across european countries. some media outlets have already relied on the concept of excess deaths to suggest that covid- fatalities are undercounted in official statistics. , others have provided some anecdotal evidence suggesting that nursing homes around the globe may have been hotbeds for contagion and mortality. for instance, the economist has analyzed data from european countries that count deaths in nursing homes in official statistics and found that these make up for percent of all official fatalities, on average. this ratio increases to percent in belgium. we use highly granular daily death registry data for thousands of municipalities in italy's north to conduct a precise estimation of the true effect of covid- on the mortality rate and compare the real death toll with what is reported in official statistics. in the next step, we extend the analysis by using detailed health care data as well as data on a host of sociodemographic, labor market and territorial characteristics to credibly estimate the 'nursing homes' effect on covid- mortality. this study relies on the definition of excess deaths to uncover the effect of covid- and draw some policy lessons. this approach does not allow us to calculate the case fatality rate of covid- , which is essential to understand how prevalent the virus is in the population and thus inform social distancing policies. moreover, our definition of excess deaths attributable to covid- includes both direct deaths-people dying of covid- -and indirect deathspeople dying for causes related to covid- , such as overcrowded hospitals. supplementary data are available at the journal of public health online. impact of non-pharmaceutical interventions (npis) to reduce covid mortality and healthcare demand dipartimento di protezione civile death at home: the unseen toll of italy's coronavirus crisis. united states: reuters world news the economist. many covid deaths in care homes are unrecorded , the economist group scovare i positivi casa per casa: così abbiamo sconfitto il virus a vo'. il virologo crisanti racconta il modello veneto managing the pandemic: lessons from italy's veneto region why mass testing is crucial: the u.s. should study the veneto model to fight covid- covid- testing for testing times: fostering economic recovery and preparing for the second wave mitigating the covid economic crisis: act fast and do whatever it takes covid- : the gendered impacts of the outbreak the economist. the impact of covid- on care homes, . london: the economist group liste di attesa nelle rsa. sistema socio sanitario regione lombardi fontana su anziani morti nelle rsa lombarde: 'l'utilizzo di quelle case di riposo per i malati covid è responsabilità dell'ats comparisons of all-cause mortality between european countries and regions global coronavirus death toll could be % higher than reported , london: the financial times group the economist. tracking covid- excess deaths across countries data gaps and the policy response to the novel coronavirus deaths from cardiac arrest have surged in new york city. london: the economist graphic detail we are deeply grateful to travers barclay child, hans genberg, coen van der kraats and eli remolona for extensive discussions in the early stages of this project. we also thank richard baldwin, xu bin, emanuele ciani, jacopo cimadomo, owen o'donnell, viktar fedaseyeu, pilar garcia-gómez, massimo giuliodori, sergi jiménez-martín, luca marcolin, magdalena rola-janicka, jonathan stöterau, stefan thewissen and seminar participants at ceibs and the oecd for helpful feedback. the views expressed in this paper are those of the authors and do not represent those of the oecd or its member countries. notes: 'municipalities' report the number of municipalities covered in the daily death registry data. '% of all municipalities' reports the share of municipalities covered over the region's total. 'population (in s)' indicates the number of residents in the municipalities covered by the death registry data, imputed from census population data, as of january . '% of all population' reports the share of the population covered over the region's total 'full sample' report statistics for the full sample of municipalities.sources: istat. , key: cord- - hatbzra authors: bone, anna e; finucane, anne m; leniz, javiera; higginson, irene j; sleeman, katherine e title: changing patterns of mortality during the covid- pandemic: population-based modelling to understand palliative care implications date: - - journal: palliat med doi: . / sha: doc_id: cord_uid: hatbzra background: covid- has directly and indirectly caused high mortality worldwide. aim: to explore patterns of mortality during the covid- pandemic and implications for palliative care, service planning and research. design: descriptive analysis and population-based modelling of routine data. participants and setting: all deaths registered in england and wales between march and may . we described the following mortality categories by age, gender and place of death: ( ) baseline deaths (deaths that would typically occur in a given period); ( ) covid- deaths and ( ) additional deaths not directly attributed to covid- . we estimated the proportion of people who died from covid- who might have been in their last year of life in the absence of the pandemic using simple modelling with explicit assumptions. results: during the first weeks of the pandemic, there were , baseline deaths, , covid- deaths and , additional deaths. deaths in care homes increased by %, while home and hospital deaths increased by % and %, respectively. hospice deaths fell by %. additional deaths were among older people ( % aged ⩾ years), and most occurred in care homes ( %) and at home ( %). we estimate that % ( %– %) of covid- deaths occurred among people who might have been in their last year of life in the absence of the pandemic. conclusion: the covid- pandemic has led to a surge in palliative care needs. health and social care systems must ensure availability of palliative care to support people with severe covid- , particularly in care homes. • • we estimate that just over one in five of all covid- deaths occurred among people who might have been in their last year of life in the absence of the pandemic. • • health and social care systems must ensure availability of palliative care to support people with severe covid- , particularly in care home settings. • • the need for integrated models of palliative care in care home settings is key, and research to underpin these models is needed. approaches to death and dying reveal much of the attitude of society as a whole to the individuals who compose it (cicely saunders, ) . the covid- pandemic has led to an unprecedented surge in mortality around the world, with some countries such as united kingdom, the united states and italy particularly affected. in the united kingdom and elsewhere, the government response has been focused on prevention and curative treatments. there has been very little emphasis internationally on care needs, including palliative care needs, for people dying during the covid- pandemic. clinical research and observation has found that people dying with covid- are symptomatic, frequently experiencing breathlessness and agitation near the end of life. clinical uncertainty around the illness trajectories of those infected increases the complexity of clinical decision-making and communication with patients and family. deaths occurring during the covid- pandemic are likely to be associated with poorer bereavement outcomes for family and friends and higher distress and support needs among frontline health and social care staff. there is an important role for palliative care, but the palliative care needs of the population have not been described. large-scale population-based studies exploring patterns of mortality are important for anticipating numbers of people in need of care and for informing governments about the need to plan and reorganise services. , previous research suggests that up to % of people dying in a typical year have palliative care needs. however, the emergence of covid- has changed patterns of mortality. we currently lack an understanding of the palliative care needs for people dying during the covid- pandemic at a population level. in england and wales, official mortality data from the office for national statistics (ons) reveal that during the first weeks of the pandemic ( march to may ), there were over , covid- deaths, most occurring in hospitals ( %), with % in care homes and few elsewhere ( %). the ons estimated that there were also over , 'additional' deaths, which were deaths over and above the number of deaths expected based on the -year average and which were not certified as caused by most people dying with covid- are older with underlying health conditions. the extent to which people dying with covid- may have been in their last year of life in the absence of the covid- pandemic has stimulated considerable debate. estimating this could help inform resource allocation and service planning, since the goals of care for those with a pre-existing advanced progressive or terminal illness may differ from those with no serious health concerns prior to a covid- infection. this study aims to explore patterns of mortality during the first weeks of the covid- pandemic in england and wales ( march to may ) to understand implications for palliative care, service planning and research. the objectives are ( ) to explore trends in place of death; ( ) to explore the age and gender distribution of baseline deaths, covid- deaths and additional deaths; ( ) to estimate the proportion of people who died from covid- who would have been in their last year of life, and differences by age; and ( ) to use this information to discuss implications for palliative care provision, service planning and research. descriptive analysis and population-based modelling using routinely collected mortality data for england and wales. we used publicly available data from the ons including the following: death registration data have been used in previous studies examining national trends in place of death. in response to the pandemic, the ons has released weekly death registration data, including deaths involving covid- and place of occurrence for all deaths. guidance for professionals completing death certificates states that covid- should be included as a contributing or underlying cause of death, when this is suspected even in the absence of laboratory confirmation. we first defined the following mortality categories: baseline deaths, covid- deaths and additional deaths, as detailed in box . the numbers of people in each mortality category were calculated for england and wales between march (week ) and may (week ) as follows: . baseline deaths. we estimated the baseline deaths according to the average deaths in the -week period of interest (weeks to ) over the years to , by age and gender. . covid- deaths. data on covid- deaths were obtained from the ons weekly registered deaths, which include deaths by age, gender and place of occurrence. the ons produces figures for deaths involving covid- based on any mention of covid- on the death certificate, including when the physician suspects covid- based on symptoms but where no test was available. these figures include deaths outside of hospital. a) people who died from covid- who might have been in their last year of life in the absence of the pandemic. first, we derived an estimate of the proportion of the population infected in each age group by re-arranging the following equation: covid- deaths = total population x proportion infected x infection fatality ratio. the information we used in this equation included the number of registered covid- deaths reported by ons; the total population from the ons principal projections and infection fatality ratios by -year age groupings obtained from the work by ferguson et al. , (supplementary table ). second, we hypothesised that someone who is in their last year of life will have an increased risk of dying from covid- compared to someone in the same age group who is not in their last year of life. to account for this, we applied a risk of dying multiplier to this group based on evidence of increased risk of death among people with serious conditions (supplementary table ). we applied a range of multipliers, with our primary estimate based on the median risk of death across different conditions. therefore, the number of people who die with covid- who would be expected to die within a year was calculated using the following equation: covid- deaths among those in their last year of life = baseline deaths x proportion infected x infection fatality ratio x risk of dying multiplier . additional deaths. we calculated additional deaths by subtracting the estimated baseline deaths and covid- deaths from all registered deaths on a per week basis over the -week period of interest (weeks [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] . this was calculated for each age and gender group separately to identify additional deaths within different demographic categories. where this resulted in a negative value, the estimated baseline figure was reduced by this amount and the value for additional deaths was zero. age, gender and place of death. to understand the age and gender distribution of the covid- and additional deaths, we grouped data into four age groups (< , - , - and ⩾ years) and compared them with baseline deaths over the period of interest (weeks [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] . we described trends in place of occurrence of deaths using the ons defined categories of 'hospital' (acute and community, not psychiatric), 'care home' (nursing and residential), 'home', 'hospice' and combined 'other communal' and 'elsewhere'. to estimate additional deaths by place of death over the period, we used place of death data for week to estimate baseline place of death, as weekly place of death information was unavailable for previous years. we sought the views and experiences of members of our patient and public network regarding their concerns around covid- pandemic. these consultations informed the focus and design of this study. patient and public involvement (ppi) representatives provided feedback on the article. this study uses routinely collected anonymised and publicly available data. no ethical approvals were necessary. between march and may (weeks [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] in england and wales, there were , deaths registered, of which , were from covid- . we estimate that this included , baseline deaths (as estimated by average weekly deaths over the same period during - ) and , additional deaths ( figure ). the lower number of deaths observed in week is likely due to a public holiday in that week which delayed the registration of deaths. where people die in england and wales has changed ( figure ). taking all deaths together, weekly hospital deaths increased by % from week (n = ) to a peak in week (n = ). we observe a % increase in home deaths between weeks (n = ) and (n = ), while care home deaths increased by % during the same period (week n = and week n = ). in week , care homes temporarily became the most common place to die. hospice deaths fell by % between week (n = ) and week (n = ), representing % of deaths during the -week period. most covid- deaths occurred in hospitals ( %) and care homes ( %), with few occurring at home and hospice ( % and %, respectively). the location of the additional deaths has not been described. comparing place of death for additional deaths occurring in weeks - with week , we found the majority of additional deaths occurred in care homes ( %) and at home ( %) (figure ). during weeks - , there were fewer additional deaths in hospitals and hospices compared to week , indicating that there were fewer non-covid- deaths in these settings than usual. we found the age distribution of covid- deaths to be overall older compared to baseline deaths, with % of covid- deaths aged ⩾ years compared to % of baseline deaths (figure ). additional deaths are older still, with % aged ⩾ years and % aged ⩾ years. a greater proportion of covid- deaths are male ( %), compared to % of baseline deaths and % of additional deaths. among the people who are registered as dying with covid- , we have estimated a group who would be in their last year of life in the absence of the pandemic. using our primary (and lower and upper) estimates for the risk of dying multiplier, we found that % ( %- %) of people who died with covid- over the period of interest were likely to be in their last year of life (table and supplementary figure ). this increases with age, such that by ≥ years, % ( %- %) of people who die with covid are likely in their last year of life in the absence of the pandemic. using routine data and modelling scenarios to understand mortality patterns during the covid- pandemic, we highlight that care homes temporarily became the most common place to die in england and wales, and that hospital and home deaths increased by over %, while deaths in hospices fell by %. we show that people in the additional deaths category are proportionately more likely to be older, and to die in care homes and at home, than those who are certified as dying with covid- . finally, we estimate for the first time that just over one in five covid- deaths occur among people who might be in their last year of life in the absence of the pandemic, and that this increases to a third of those aged over years. deaths in hospitals and care homes and at home increased during the first weeks of the covid- pandemic. rapid adaptation of models of palliative care is needed at the first sign of subsequent waves of the covid- pandemic or future pandemics to support the increased number of people dying in these settings. guidance on remote consultations in primary care and on managing distressing symptoms in the community is available to help support care for people with suspected covid- , but challenges remain in terms of workforce and equipment. a pilot study of palliative care delivered in care homes via teleconference supported the feasibility of this approach where required. specialist palliative care provision in care homes is currently variable, [ ] [ ] [ ] and guidance to support palliative care for people with covid- in nursing homes has been found to be lacking and limited in focus. in contrast, deaths in hospices fell during the time period. a similar phenomenon was found during the sars pandemic and highlighted the need for seamless continuity of care between settings. systems that enable hospice resources to rapidly shift from inpatient to community settings are likely to be important in future pandemic peaks. little has been described about the characteristics of the additional deaths that have occurred during the covid- pandemic. the aetiology of these deaths is not known, in particular whether they are undiagnosed covid- deaths or deaths associated with the pandemic without being a result of the disease (e.g. due to avoidance of urgent healthcare). there is emerging evidence that older people with covid- may present with non-specific and atypical symptoms such as delirium or low-grade fever. we found that the additional deaths group comprises predominantly older people in care homes, suggesting that these may be undiagnosed covid- deaths. irrespective of aetiology, this group are likely to have significant palliative care needs. the characteristics of the additional deaths population is likely to change as the pandemic progresses, and deferred harm from missed routine check-ups or screening becomes more likely. at a population level, we estimate that % of covid- deaths occur among people who might have been in their last year of life in the absence of the pandemic, and that among those aged > years, % might be in their last year of life. for those with advanced conditions living at home and in care homes, guidelines recommend that general practitioners (gps), community healthcare staff and community geriatricians urgently review advance care plans with patients and their families and discuss their care preferences in the case of contracting covid- . , these discussions are challenging, particularly when conducted over the telephone. specialist palliative care services may play a role in offering advice and support for conducting difficult conversations in difficult situations. while the focus of this article is covid- , it should be noted that baseline deaths outnumber covid- and additional deaths. previous modelling has shown that around %- % of the baseline group are likely to have palliative care needs. covid- will impact this group, for example, through fewer face-to-face contacts and potentially less availability of specialist palliative care support, as services are suddenly stretched beyond their usual bounds. informal caregivers are likely to play an important role in caring for those in home settings and may need additional support. across all mortality categories, there is the need for bereavement support as a result of higher prevalence of complicated grief. there will be the typical grief associated with death and dying, but on a much larger scale. in addition, there will also be novel grief processes, linked to social isolation and distancing, self-blame around infection and not being able to attend funerals. palliative care specialists with particular expertise in bereavement support have a role in supporting families in coping with bereavement and supporting colleagues to deliver this care alongside other relevant organisations. internationally, the covid- pandemic has affected countries at different times and to varying degrees. the united kingdom has experienced particularly high mortality during the covid- pandemic. a common feature across nations affected by covid- is the high proportion of covid- deaths among care home residents, for example, % in south korea and % in france. all countries severely affected by covid- will have a surge in need for palliative care, though country-level factors, such as socio-demographic characteristics, will influence the magnitude of this. national guidance on providing palliative care during pandemics, taking into account the structure of the health and social care system and level of integration of palliative care, is essential. this is the first article to estimate and characterise population-level patterns of mortality during the covid- pandemic in relation to palliative care. we have produced simple models with explicit assumptions to raise questions and stimulate discussion. we calculated additional this study focused on mortality only. other aspects of palliative care are important to be considered, for example, symptom control, support with difficult conversations, advance care planning, and complex clinical decision-making. there are also longer-term needs among those surviving covid- , such as rehabilitation and palliative care, which are likely to be significant and important for service planning but were not modelled here. we have produced population-based models that are useful for strategic health service planning but are not directly applicable to clinical decision-making at an individual level. for example, while we have estimated the number of people dying from covid- who might have been in their last year of life in the absence of the pandemic at the population level, we acknowledge the challenges of prognostication at the individual level. our estimates used the currently available data on the risk of death from covid- across different conditions, and this multiplier should be revised as more data become available. a population-level understanding of the palliative care needs of people with severe covid- is essential to guide efficient health service planning, but has been missing from political rhetoric around the world, much of which has been focused on intensive care capacity. greater support for care homes through closer collaboration with primary care, specialist palliative care services and community geriatricians is needed for now and for future pandemics. investment in palliative care training and education for social care staff so that they have the skills, competence and confidence to provide care to those approaching the end of life. given the steep rise in deaths that have occurred in challenging circumstances, ensuring provision of bereavement support should be a priority. covid- is now the focus of much research, but very few studies are focused on palliative and end of life care. , data and research are urgently required to answer the following questions: what are the best integrated models of palliative care in care homes that can be adapted for use during the covid- pandemic? what is the impact of covid- on the hospice sector and how might hospice services best use their available resources? what is the impact of covid- on complex grief, and what resources are needed to support those experiencing this? what is the impact of covid- on carers supporting people with palliative care needs in community settings? our study has highlighted the high level of need for palliative care during the covid- pandemic. healthcare systems responding to covid- must consider the need for palliative care surge capacity, including the workforce and equipment, to enable people to be cared for appropriately in all settings. the need for integrated models of palliative care in care home settings is imperative. the rapid changes to practice during the pandemic provide important opportunities for research, evaluation and learning, and it is essential that new models are underpinned by evidence. draft of the manuscript. all authors critically revised the manuscript for intellectual content and approved the final version of the submitted manuscript. the corresponding author attests that all listed authors meet authorship criteria and that no one meeting the criteria have been omitted. all data are publicly available the author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: all authors have completed the icmje uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous years; no other relationships or activities that could appear to have influenced the submitted work. world health organisation. coronavirus disease (covid- ) pandemic characteristics, symptom management and outcomes of patients with covid- referred for hospital palliative care the importance of addressing advance care planning and decisions about do-not-resuscitate orders during novel coronavirus (covid- ) bereavement support on the frontline of covid- : recommendations for hospital clinicians what is the impact of population ageing on the future provision of end-oflife care? population-based projections of place of death how many people will need palliative care in ? past trends, future projections and implications for services how many people need palliative care? a study developing and comparing methods for population-based estimates how much 'normal' risk does covid represent principal projection -england and wales population in age groups guidance for doctors completing medical certificates of cause of death in england and wales for use during the emergency period only impact of non-pharmaceutical interventions (npis) to reduce covid- mortality and healthcare demand global covid- case fatality rates risk factors for mortality in patients with coronavirus disease (covid- ) infection: a systematic review and meta-analysis of observational studies covid- : a remote assessment in primary care covid- rapid guideline: managing symptoms (including at the end of life) in the community early integration of palliative care in a long-term care home: a telemedicine feasibility pilot study a survey of end-of-life care in care homes: issues of definition and practice palliative care provision in long-term care facilities differs across europe: results of a cross-sectional study in six european countries (pace) strategies for the implementation of palliative care education and organizational interventions in long-term care facilities: a scoping review international covid- palliative care guidance for nursing homes leaves key themes unaddressed hospice utilization during the sars outbreak in taiwan the role and response of palliative care and hospice services in epidemics and pandemics: a rapid review to inform practice during the covid- pandemic what is happening to non-covid deaths? atypical presentation of covid- in a frail older person collateral damage: the impact on cancer outcomes of the covid- pandemic. medrxiv. epub ahead of print covid- : managing the covid- pandemic in care homes for older people administration of endof-life drugs by family caregivers during covid- pandemic complicated grief after death of a relative in the intensive care unit grief during the covid- pandemic: considerations for palliative care providers coronavirus pandemic -country comparisons mortality associated with covid- outbreaks in care homes: early international evidence. international long term care policy network global atlas of palliative care at the end of life crisis symptom management and patient communication protocols are important tools for all clinicians responding to covid- prognostication in palliative care commentary: covid in care homes -challenges and dilemmas in healthcare delivery urgent public health covid- studies uk research innovation. covid- research and innovation supported by ukri we are grateful for the critical review from janice tausig, richard meade, scott murray and helen findlay. supplemental material for this article is available online. key: cord- - w qsld authors: valev, d. title: relationships of total covid- cases and deaths with ten demographic, economic and social indicators date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: w qsld we have used data for countries with a population of over millions in which . % of the world's population lives. the statistical relationships of total covid- cases and deaths per million populations in these countries with demographic, economic and social indicators (indices) were studied. these indicators are life expectancy, median age, growth rate, population density, gdp ppp per capita, human development index (hdi), gini index of income equality, intelligence quotient (iq), corruption perceptions index (cpi) and democracy index. statistically significant relationships were found with all indicators excluding gini index and population density. we have found that the closest is the relationship of deaths per million population and total cases per million population with correlation coefficient r = . . therefore, it is clear statistically that the more are cases per million in a country the more are deaths per million. this confirms the correctness of the timely and effective introduction of the necessary pandemic restrictions in the countries. it is interesting that the close correlations were found of cases and deaths per million with a purely economic index like gdp ppp per capita, where r = . and r = . , respectively. even more close correlations were found of cases and deaths per million with a composite index hdi, where the correlation coefficients reach . and . , respectively. this paradoxical results show that the richest and well-being countries are most seriously affected by the covid- pandemic. the most probable reason for this is the large percentage of aging population, comorbidity of population with severe chronic diseases and obesity in countries with high gdp and hdi. no less important reason appears the delayed and/or insufficiently effective pandemic restrictions in these countries, which have underestimated the danger of a pandemic in early stage. other indicators (excluding gini index and population density) also show statistically significant correlations with cases and deaths per million with correlation coefficients from . to . . the countries that deviates the most from the regression lines were shown. surprisingly, there was no statistically significant correlation between cases and deaths with population density. the statistical significance of the found correlations determined using student's t-test was p < . . the countries that deviate the most from both sides of the regression line were shown. it has been shown that the correlations of covid- cases and deaths with the studied indicators decrease with time. key words: covid- pandemic; statistics; cases and deaths per million; demographic, economic and social indicators coronavirus disease (covid- ) is an infectious disease caused by severe acute respiratory syndrome coronavirus (sars-cov- ) [ ] . it was first identified in december in wuhan, china, and has resulted in an ongoing pandemic [ ] . as of september , more than . million cases have been reported across countries and territories, resulting in more than thousand deaths. more than . million people have recovered [ ] . common symptoms include fever, cough, fatigue, shortness of breath, and loss of smell and taste. while the majority of cases result in mild symptoms, some progress to acute respiratory distress syndrome (ards) possibly precipitated by cytokine storm, multi-organ failure, septic shock, and blood clots. the virus is primarily spread between people during close contact, most often via small droplets produced by coughing, sneezing, and talking. according to the world health organization (who), there are neither vaccines nor specific antiviral treatments for covid- [ ] . management involves the treatment of symptoms, supportive care, isolation, and experimental measures. the who recommends - meters of social distance. severe acute respiratory syndrome coronavirus (sars-cov- ) is a novel severe acute respiratory syndrome coronavirus closely related to the original sars-cov [ ] . it is thought to have an animal (zoonotic) origin. it is % identical at the whole genome level to other bat coronavirus samples (batcov ratg ) [ ] . it is the seventh known coronavirus to infect people [ ] . the coronaviruses are capable of causing illnesses ranging from the common cold to more severe diseases such as middle east respiratory syndrome (mers, fatality rate ~ %). the standard method of testing is polymerase chain reaction (pcr) [ ] . the test is typically done on respiratory samples obtained by a nasopharyngeal swab, however, a nasal swab may also be used. based on johns hopkins university statistics, the global case fatality rate (cfr) is . % as of june . [ ] the number strongly varies by regionfrom less than % in saudi arabia, ethiopia, and uzbekistan to - % in italy, united kingdom and france [ ] . several factors have been found statistically that increase severe course and mortality of covid- : . one of the most important factors in covid- mortality is the age of infected. the case fatality rate (cfr) is the number of deaths divided by the number of diagnosed cases. according to data for china, cfr remains within . % for children and young people in the age groups - , - , - and - years. with age, cfr increases rapidly and reaches . % for the age group + years [ ] . dependence of cfr from age of infected is shown on fig. . is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . . https://doi.org/ . / . . . doi: medrxiv preprint the exponential type of cfr dependence on the age of the infected is similar in other countries, but in some countries with increasing age the cfr increases more sharp. for example, in the netherlands, sweden, finland, israel, canada, mexico, and the philippines, the cfr for the - age group exceeds %. in these countries, the cfr is significantly higher in the age groups over years too. [ ] . . the next crucial factor of mortality from covid- appears comorbidity of patients with severe diseases. most of those who die of covid- have pre-existing (underlying) conditions, including hypertension, diabetes mellitus, and cardiovascular disease [ , ] . . vitamin d deficiency in populations. strong correlation between prevalence of severe vitamin d deficiency and population mortality rate from covid- in europe has been found in [ ] . very high body mass index (obesity). johns hopkins cardiologist david kass discusses a recent study he co-led that links higher body mass index to more severe cases of covid- and points to obesity as a significant pre-existing condition in younger patients in particular [ ] . . the role of the bcg vaccine for the prevention of covid remains controversial yet [ , ] . the infection fatality rate (ifr) reflects the percent of infected individuals (diagnosed and undiagnosed) who die from a disease. ifr at least is several times less than cfr because of many asymptomatic and mild symptoms infected persons. it is hardly be accurately calculated yet the global value is of the order of . % [ ] . ifr strongly varies across regions and countriesfrom . % in sub-saharan africa to slightly above % in western europe and high-income asia pacific [ ] . the basic reproduction number (r ) of the virus has been estimated to be . [ ] . therefore each infection from the virus is expected to result in . new infections when no members of the community are immune and no preventive measures are taken. this shows that covid- is much more contagious than seasonal flu, which has r between and . preventive measures to reduce the chances of infection include staying at home, avoiding crowded places, keeping distance from others, washing hands with soap and water often and for at least seconds, avoiding touching the eyes, nose, or mouth with unwashed hands, and monitoring and self-isolation for people who suspect they are infected. authorities worldwide have responded by implementing travel restrictions, lockdowns, workplace hazard controls, and facility closures. many places have also worked to increase testing capacity and trace contacts of infected persons. many countries have recommended that healthy individuals wear face masks at least in certain public settings. this recommendation is meant to reduce the spread of the disease by asymptomatic and pre-symptomatic individuals and is a complementary measure to established preventive measures such as social distancing. the severity of covid- varies. the disease may take a mild course with few or no symptoms. social distancing includes infection control actions intended to slow the spread of disease by minimizing close contact between individuals. methods include quarantines; travel restrictions; and the closing of schools, workplaces, stadiums, theatres, or shopping centers. self-isolation at home has been recommended for those diagnosed with covid- and those who suspect they have been infected. many governments have mandated or recommended self-quarantine for entire populations. those who may have been exposed to someone with covid- and those who have recently traveled to a country or region with the widespread transmission have been advised to self-quarantine for days from the time of last possible exposure [ ] . containment is undertaken in the early stages of the outbreak and aims to trace is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . . https://doi.org/ . / . . . doi: medrxiv preprint and isolate those infected as well as introduce other measures to stop the disease from spreading. suppression requires more extreme measures so as to reverse the pandemic by reducing the basic reproduction number to less than [ ] . more drastic actions aimed at containing the outbreak were taken in china once the severity of the outbreak became apparent, such as quarantining entire cities and imposing strict travel bans [ ] . contact tracing is an important method for health authorities to determine the source of infection and to prevent further transmission [ is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . . https://doi.org/ . / . . . doi: medrxiv preprint we used data for countries with a population of over million. there are countries from china with million inhabitants to angola with . million inhabitants [ ] . in our work, we have used the limit of more than million populations, because these countries are less than a quarter of the countries in the world, and . % of the world's population lives in them. moreover, the difference in population of the countries used for statistical analysis does not exceed times. if all countries were used in the study, taking into account that the difference in the population between the largest country and smallest one (china and the vatican, respectively) is over a million times, an unacceptably large difference in the statistical weight of the countries would be obtained. data for total covid- cases per million population (cases/ m) and deaths per million populations (deaths/ m) were used for all countries studied. the data were taken from [ ] . data for may were used in the main calculations. to check whether there is a change in the correlations with the development of the pandemic, a statistical analysis was made for three different dates roughly two months apart - april, may and august . the following demographic, economic and social indicators (indices) for the surveyed countries were used, namely: is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . in addition to linear regression, power, logarithmic and exponential functions were used for statistical data processing. the statistical significance of the found correlations was assessed using student's t-test. the statistical package statistica has been used for calculations. below the results of statistical studies on the relationship of total covid- cases per million population and deaths per million populations at may with demographic, economic and social indicators (indices) are shown. we have found that the log deaths per million is the most closely connected with log cases per million and the coefficient of correlation reaches r = . . the dependence of log deaths per million from log cases per million is shown in fig. . the names of the most deviated countries from the regression line are shown in the figure. therefore, it is statistically clear that how the more are cases per million in a country, the more are deaths per million. from fig. it is seen that saudi arabia, russia, vietnam and uganda lie below the regression line, i.e. mortality in these countries is lower. a case fatality rate (cfr) from cases per million was then sought and it was found that cfr increased logarithmically with increasing of cases. the relationship is statistically significant with a correlation coefficient r = . and is shown in fig. . this confirms that the higher the cases per million, the higher the case fatality rate. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . a statistical analysis of the relationship of total covid- cases and deaths with demographic, economic and social indices has been implemented below. a statistical link between total cases and deaths with life expectancy was first sought. a positive correlation was found with correlation coefficients r = . and r = . , respectively. as total cases and deaths per million differ by several orders of magnitude in different countries, we hypothesized that they increase exponentially with life expectancy. indeed, the correlation coefficients of the exponential relationship increased and reached . and . , respectively. the dependence of log cases/ m from life expectancy is shown in fig . the most deviating countries from exponential relationship were shown. vietnam, japan, south korea, thailand, myanmar and angola lie below the regression line, whereas spain, usa, peru, russia, south africa, nigeria and dr congo lie above the regression line. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . the statistical significance of the found correlations determined using student's t-test was p < . . the trend of exponential growth of total cases and deaths with life expectancy is unambiguous. this is not a surprising result as it is well known that covid- disease affects people over the age of much more severely [ ] . due to the high proportion of adults over the age of living in countries with a high life expectancy, their cases and deaths are significantly higher than those with low life expectancy. the median age of a population is the point at which half the population is older than that age and half is younger. the median age closely correlated with life expectancy (r = . ). by reason of that the correlations of log cases and log deaths with median age were similar to log cases and log deaths with the life expectancy, respectively r = . and r = . . the relationship of log deaths per million from median age is shown in fig. . the countries that deviate significantly from the exponential curve are also shown in the figure. the most of them coincide with the countries that deviate extremely from the relationship between log cases per million from life expectancy presented in fig. . this is result from the close relationship of median age and life expectancy. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . the last studied demographic indicator is the population density. the usual expectations are that the countries with higher population density will have higher total cases and deaths per million. the statistical analysis has shown a surprising result -absence of significant correlation of cases and deaths with population density. the found negative coefficients of correlation of cases and deaths are: r = - . and r = - . , respectively. these correlations are low and not statistically significant, therefore cases and deaths don't depend from the population density. the correlations of cases and deaths per million was then sought with a purely economic index, namely gross domestic product at purchasing power parity per capita (gdp). close correlation of log cases and log deaths with log gdp was established having correlation coefficients r = . and r = . , respectively. the results for cases are presented in fig. . is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . therefore, the total cases and deaths per million increase by degree law with gross domestic product per capita. this result is astounding because it shows that the higher the gross domestic product per capita, the higher the rate of infection and mortality from covid- , i.e. the richer the countries, the harder they are hit by the covid- pandemic! the next index included in statistical calculations is human development index (hdi). the hdi is a statistic composite index of life expectancy, education, and per capita income indicators, which are used to rank countries by tiers of human development. it is accepted that hdi evaluates development not only by economic advances but also improvements in human well-being. the dependence of log cases and deaths per million from hdi is shown in fig. . the cases per million are shown in blue and the deaths per million are in red. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . the correlation coefficients reach high values of r = . and r = . , respectively. this result confirms above mention observation that the richest and well-being countries are most severe affected from the covid- pandemic. this paradoxical result cannot be simply explained, because countries with high per capita incomes, well-developed health systems and a high level of education should be expected to be better protected from a pandemic. apparently, in the case of the covid- pandemic, part of the population of these countries is significantly more vulnerable than countries with a lower human development index. several factors can be suggested that are probably the cause of this paradox: . the countries with high hdi have a significant part of older population (high life expectancy and median age), that is highly vulnerable to covid- . our statistical calculations have shown that hdi is closely correlated with life expectancy (r = . ), and as we have seen above cases and deaths are closely correlated with life expectancy (r = . and r = . , respectively). . the health systems in these countries support the lives of a significant number of people with severe chronic diseases such as diabetes, cancer, cardiovascular disease and others. infection of such patients with coronavirus overloads their weakened body and they become seriously and often fatally ill. . a significant percentage of people in countries with high hdi suffer from overweight and obesity due to immobilization and consumption of high-calorie foods, and obesity is known to be a risk factor for severe course and death from covid- . . the population in countries with high hdi is not inclined to enough strictly pandemic restrictions, excluding some east asian countries. . other factors -the population in countries with high hdi is exposed to higher stress, uses more drugs and canned foods that weaken the immune system and more. fig. . it can be seen that this distribution is very similar to the distribution of countries by covid- deaths per million people presented in fig. . is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . . https://doi.org/ . / . . . doi: medrxiv preprint it has been found that the total cases and deaths per million showed no correlations with gini index of income equality. therefore, cases and deaths don't depend from income equality. the statistical relationship of log cases and deaths per million with corruption perceptions index (cpi) has been analyzed, too. the found correlation coefficients are r = . and r = . , respectively. the corruption perceptions index is closely related with hdi (r = . ) and the last index could explain the above correlation. the situation is similar with the connection of log cases and deaths per million with intelligence quotient (iq) of countries. the respective correlation coefficients are r = . and r = . , respectively. the iq of countries is also closely correlated with hdi (r = . ). therefore, the defining parameter is again the human development index. finally, we found correlation of log cases and deaths per million with democracy index r = . and r = . , respectively. the democracy index is also well correlated with hdi (r = . ). the found statistically significant relationships of covid- cases and deaths with most of studied demographic, economic and social indicators are the result of their connections with log gdp ppp. the correlation coefficients r and significance levels p of the connection of log gdp ppp with the rest indicators is presented in table . is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . . https://doi.org/ . / . . . doi: medrxiv preprint gini index of income equality -. n. s. population density - . n. s. table shows that the studied indicators excluding density of population and gini index have statistically significant correlations with log gdp ppp. to check whether there is a change in the correlations with the development of the pandemic, a statistical analysis was made for three different dates roughly two months apart. the results of this analysis are presented in table . table shows that the correlation of deaths per million from covid- cases per million slow increase during the pandemic and reach r = . . the correlation coefficients of covid- cases and deaths with the rest indicators decrease during the pandemic. furthermore, at the beginning of august the relationships of covid- cases and deaths with iq index ceases to be statistically significant, and the relationships of covid- cases with cpi and growth rate has significance level p < . . a probable reason for this time course of correlation coefficients is that most west european countries with high gdp, hdi, and life expectancy (italy, spain, uk, france) introduced significantly stricter pandemic restrictions than at the beginning of the pandemic and they succeeded significantly slow the spread of the infection. on the other hand, in recent months, the pandemic has spread rapidly to a number of countries with medium or low gdp, hdi, and life expectancy in the latin america (mexico, brazil, argentina, peru, colombia), india and south africa. in these countries, the pandemic began later, and when the number of covid- cases and deaths in western europe and the united states increased alarmingly, the cases and deaths in the indicated countries remained low. this created illusions among the people and institutions in these countries that the pandemic would not affect them severely, and therefore . cc-by-nd . international license it is made available under a perpetuity. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . . https://doi.org/ . / . . . doi: medrxiv preprint pandemic restrictions were significantly weakened. as a result, covid- cases and deaths have risen sharply in these countries. we have used data for countries with a population of over millions in which . % of the world's population lives. the statistical relationships of total covid- cases and deaths per million populations in these countries with demographic, economic and social indicators (indices) were studied. these indicators are life expectancy, median age, growth rate, population density, gdp ppp per capita, human development index (hdi), gini index of income equality, intelligence quotient (iq), corruption perceptions index (cpi) and democracy index. statistically significant relationships were found with all indicators excluding gini index and population density. we have found that the closest is the relationship of deaths per million population and total cases per million population with correlation coefficient r = . . therefore, it is clear statistically that the more are cases per million in a country the more are deaths per million. this confirms the correctness of the timely and effective introduction of the necessary pandemic restrictions in the countries. it is interesting that the close correlations were found of cases and deaths per million with a purely economic index like gdp ppp per capita, where r = . and r = . , respectively. even more close correlations were found of cases and deaths per million with a composite index hdi, where the correlation coefficients reach . and . , respectively. this paradoxical results show that the richest and well-being countries are most seriously affected by the covid- pandemic. the most probable reason for this is the large percentage of aging population, comorbidity of population with severe chronic diseases and obesity in countries with high gdp and hdi. no less important reason appears the delayed and/or insufficiently effective pandemic restrictions in these countries, which have underestimated the danger of a pandemic in early stage. other indicators (excluding gini index and population density) also show statistically significant correlations with cases and deaths per million with correlation coefficients from . to . . the countries that deviates the most from the regression lines were shown. surprisingly, there was no statistically significant correlation between cases and deaths with population density. the statistical significance of the found correlations determined using student's t-test was p < . . the countries that deviate the most from both sides of the regression line were shown. it has been shown that the correlations of covid- cases and deaths with the studied indicators decrease with time. the established close positive correlation between cases and deaths per million with gdp (ppp) per capita and hdi seems paradoxical at first glance. because countries with high gdp (ppp) per capita and hdi have well-developed health systems that are provided with sufficient high-tech medical equipment and highly qualified specialists. what, then, is the reason why most of these countries allow the highest rates of covid- infection and mortality? these countries do have well-off and well-functioning health systems in a normal health situation. in this case, however, the situation is not normal, but a crisis one, because we have a pandemic due to a new coronavirus sars-cov- against which there is no vaccine or specific drug. this virus causes airborne infection and has a reproductive number r ~ . , which is significantly higher than that of seasonal flu (r ~ . ), i.e. coronavirus is significantly more contagious than the flu. in addition, mortality from covid- is of the order of a few percent, therefore many times greater than that of seasonal influenza, in which the mortality rate is about . %. the situation is further complicated by the fact that the latency period of covid- is much longer than that of seasonal flu and reaches - weeks, compared to - days for seasonal flu. in addition, most cases of covid- infection have mild flu-like . cc-by-nd . international license it is made available under a perpetuity. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted september , . . https://doi.org/ . / . . . doi: medrxiv preprint symptoms or asymptomatic. therefore, it is not possible to distinguish well between the sick and to isolate or quarantine them, and for the healthy to continue to work and live unrestricted by quarantine measures. in case of dangerous infectious diseases against which there is no vaccine, the only effective means of limiting the percentage of infected is quarantine, which has been known and applied since the middle ages. for this reason, the precautionary and quarantine measures of the entire population play a crucial role in limiting the infection rate and the associated mortality of the population. developing countries generally have lower rates of relevant comorbidities compared to high-income countries (where the best measures of infection fatality rates come from). therefore, the material wellbeing of the citizens and the provision of the health systems of the countries with equipment, consumables and qualified specialists play a secondary role. the main thing is the discipline and self-discipline of the population, its readiness to comply with pandemic restrictions at a distance, isolation and disinfection and especially the timely and decisive introduction and implementation of pandemic restrictions by the executive authorities -government, police, health services and others. this is why many countries with relatively low gdp per capita and hdi (vietnam, philippines, ethiopia, nigeria) have been able to achieve much lower covid- infection and mortality than countries with high gdp per capita and hdi (italy, spain, usa, united kingdom, france). most countries with high gdp per capita and hdi are developed democracies with high individual freedom from the state and its organs, and their governments hesitated too long before introducing pandemic restrictions restricting the freedom of movement, assembly and life of citizens. too much time was wasted (of the order of - weeks) during which the epidemic spread uncontrollably and the number of infected reached high values that made it very difficult to limit the pandemic, despite the strict restrictive pandemic restrictions. of course, there are countries with high gdp per capita and hdi, such as japan, south korea and malaysia, which have managed to effectively limit the spread of covid- and prevent high mortality. this is due to the traditional culture of these countries, in which the trust of the citizens in the ruling elite and their tendency to obey the reasonably justified decisions of these elite is deeply rooted. the continuing -ncov epidemic threat of novel coronaviruses to global health-the latest novel coronavirus outbreak in wuhan covid- dashboard by the center for systems science and engineering quarantine alone or in combination with other public health measures to control covid- : a rapid review mystery deepens over animal source of coronavirus real-time rt-pcr panel for detection -ncov age comparison of covid- fatality rate in china hospitalization rates and characteristics of patients hospitalized with laboratory-confirmed coronavirus disease -covid-net predicted covid- fatality rates based on age, sex, comorbidities, and health system capacity more evidence on vitamin d deficiency and death rates from covid- obesity could shift severe covid- disease to younger ages a shred of evidence that bcg vaccine may protect against covid- : comparing cohorts in spain and italy systematic review and meta-analysis of published research data on covid- infection-fatality rates the potential effects of widespread community transmission of sarscov- infection in the world health organization african region: a predictive model impact of non-pharmaceutical interventions (npis) to reduce covid mortality and healthcare demand. imperial college covid- response team china may be beating the coronavirus nick phin, coronavirus (covid- ) expert interview: what is contact tracing? blog: public health matters. public health england, government of the united kingdom coronavirus: first two cases in italy. corriere della sera who says europe is new epicenter of coronavirus pandemic coronavirus cases have dropped sharply in south korea. what's the secret to its success? science the who ignores taiwan. the world pays the price. the nation total confirmed deaths due to covid- per million people the authors have declared no competing interest. no funding. i confirm all relevant ethical guidelines have been followed. no human subjects research as i accessed only publicly available data from the web. key: cord- -qbvgnr r authors: ioannidis, john p.a.; cripps, sally; tanner, martin a. title: forecasting for covid- has failed date: - - journal: int j forecast doi: . /j.ijforecast. . . sha: doc_id: cord_uid: qbvgnr r epidemic forecasting has a dubious track-record, and its failures became more prominent with covid- . poor data input, wrong modeling assumptions, high sensitivity of estimates, lack of incorporation of epidemiological features, poor past evidence on effects of available interventions, lack of transparency, errors, lack of determinacy, looking at only one or a few dimensions of the problem at hand, lack of expertise in crucial disciplines, groupthink and bandwagon effects and selective reporting are some of the causes of these failures. nevertheless, epidemic forecasting is unlikely to be abandoned. some (but not all) of these problems can be fixed. careful modeling of predictive distributions rather than focusing on point estimates, considering multiple dimensions of impact, and continuously reappraising models based on their validated performance may help. if extreme values are considered, extremes should be considered for the consequences of multiple dimensions of impact so as to continuously calibrate predictive insights and decision-making. when major decisions (e.g. draconian lockdowns) are based on forecasts, the harms (in terms of health, economy, and society at large) and the asymmetry of risks need to be approached in a holistic fashion, considering the totality of the evidence. modeling resurgence after reopening also failed (table ). e.g. a massachusetts general hospital model predicted over , deaths within a month of georgia reopening -actual deaths were . table lists some main reasons underlying this forecasting failure. unsurprisingly, models failed when they used more speculation and theoretical assumptions and tried to predict long-term outcomes, e.g. using early sir-based models to predict what would happen in the entire season. however, even forecasting built directly on data alone fared badly, , failing not only in icu bed predictions ( figure ) but even in next day death predictions when issues of long-term chaotic behavior do not come into play (figures and ). even for short-term forecasting when the epidemic wave waned, models presented confusingly diverse predictions with huge uncertainty (figure ). failure in epidemic forecasting is an old problem. in fact, it is surprising that epidemic forecasting has retained much credibility among decision-makers, given its dubious track record. modeling for swine flu predicted , - , deaths in the uk. eventually deaths occurred. models on foot-and-mouth disease by top scientists in top journals , were subsequently questioned by other scientists challenging why up to million animals had to be slaughtered. predictions for bovine spongiform encephalopathy expected up to , deaths in the uk. however, the lower bound predicted as low as deaths, let's be clear: even if millions of deaths did not happen this season, they may happen with the next wave, next season, or some new virus in the future. a doomsday forecast may come handy to protect civilization, when and if calamity hits. however, even then, we have little evidence that aggressive measures focusing only on few dimensions of impact actually reduce death toll and do more good than harm. we need models which incorporate multicriteria objective functions. isolating infectious impact, from all other health, economic and social impacts is dangerously narrow-minded. more importantly, with epidemics becoming easier to detect, opportunities for declaring global emergencies will escalate. erroneous models can become powerful, recurrent disruptors of life on this planet. civilization is threatened from epidemic incidentalomas. cirillo and taleb thoughtfully argue that when it comes to contagious risk, we should take doomsday predictions seriously: major epidemics follow a fat-tail pattern and extreme value theory becomes relevant. examining major epidemics recorded through history, they demonstrate a fat-tailed mortality impact. however, they analyze only the most-noticed outbreaks, a sample with astounding selection bias. for example, according to their dataset, the first epidemic originating from sub-saharan africa did not occur until ad, namely , one of them, oc seems to have been introduced in humans as recently as , probably causing a "bad influenza year" with over a million deaths. based on what we know now, sars-cov- may be closer to oc than sars-cov- . this does not mean it is not serious: its initial human introduction can be highly lethal, unless we protect those at risk. a heavy tail distribution ceases to be as heavy as taleb imagines when the middle of the distribution becomes much larger. one may also argue that pandemics, as opposed to epidemics without worldwide distribution, are more likely to be heavy-tailed. however, the vast majority of the contagious events listed by taleb were not pandemics, but localized epidemics with circumscribed geographic activity. overall, when a new epidemic is detected, it is even difficult to pinpoint which distribution of which known events it should be mapped against. blindly acting based on extreme value theory alone would be sensible if we lived in the times of the antonine plague or even in , with no science to identify the pathogen, elucidate its true prevalence, estimate accurately its lethality, and carry out good epidemiology to identify which people and settings are at risk. until we accrue this information, immediate better-safethan-sorry responses are legitimate, trusting extreme forecasts as possible (not necessarily likely) scenarios. however, caveats of these forecasts should not be ignored , and new evidence on the ground truth needs continuous reassessment. upon acquiring solid evidence about the j o u r n a l p r e -p r o o f journal pre-proof epidemiological features of new outbreaks, implausible, exaggerated forecasts should be abandoned. otherwise, they may cause more harm than the virus itself. the insightful recent essay of taleb offers additional opportunities for fruitful discussion. taleb ruminates on the point of making point predictions. serious modelers (whether frequentist or bayesian) would never rely on point estimates to summarize skewed distributions. even an early popular presentation from has a figure (see page ) with striking resemblance to taleb's figure . in a bayesian framework, we rely on the full posterior predictive distribution, not single points. moreover, taleb's choice of a three-parameter pareto distribution is peculiar. it is unclear this model provides a measurably better fit to his (hopelessly biased) pandemic data than, say, a two parameter gamma distribution fitted to log counts. regardless, either skewed distribution would then have to be modified to allow for the use of all available sources of information in a logically consistent fully probabilistic model, e.g. via a bayesian hierarchical model (which can certainly be formulated to accommodate fat tails if needed). in this regard, we note that examining the ny daily death count data studied in ref. , these data are found to be characterized as stochastic rather than chaotic. taleb seems to fit an unorthodox model, and then abandons all effort to predict anything. he simply assumes doomsday has come, much like a panic-driven roman would have done in the antonine plague, lacking statistical, biological, and epidemiological insights. j o u r n a l p r e -p r o o f journal pre-proof taleb caricatures the position of a hotly debated mid-march op-ed by one of us, alluring it "made statements to the effect that one should wait for "more evidence" before acting with respect to the pandemic", an obvious distortion of the op-ed. anyone who reads the op-ed unbiasedly realizes that it says exactly the opposite. it starts with the clear, unquestionable premise that the pandemic is taking hold and is a serious threat. immediate lockdown certainly makes sense when an estimated million deaths are possible. this was stated emphatically in multiple occasions these days in interviews in multiple languages -for examples see refs. - . certainly, adverse consequences of short-term lockdown cannot match million lives. however, better data can recalibrate estimates, re-assessing downstream the relative balance of benefits and harms of longer-term prolongation of lockdown. that re-appraised balance changed markedly over time. another gross distortion propagated in social media is that supposedly the op-ed had predicted that only , deaths in the usa. the key message of the op-ed was that we lack reliable data, i.e. we don't know. the self-contradicting misinterpretation as "we don't know, but actually we do know that , deaths will happen" is impossible. the op-ed discussed two extreme scenarios to highlight the tremendous uncertainty absent reliable data: , deaths in the us and , , deaths. we needed reliable data, quickly, to narrow this vast uncertainty. we did get data and did narrow uncertainty. science did work eventually, even if forecasts, including those made by one of us (confessed and discussed in box ), failed. taleb offers several analogies to assert that all precautionary actions are justified in pandemics, deriding "waiting for the accident before putting the seat belt on, or evidence of fire before buying insurance". the analogies assume that the cost of precautionary actions are small j o u r n a l p r e -p r o o f journal pre-proof in comparison to the cost of the pandemic, and that the consequences of the action have little impact on it. however, precautionary actions can backfire badly when they are misinformed. in march, modelers were forecasting collapsed health systems, e.g. , beds would be needed in new york, when only a small fraction were available. precautionary actions damaged the health system, increased covid- deaths, and exacerbated other health problems (table ). seat belts cost next to nothing to produce in cars and have unquestionable benefits. despite some risk compensation and some excess injury with improper use, eventually seat belts prevent ~ % of serious injuries and deaths. measures for pandemic prevention equivalent to seat belts in terms of benefit-harm profile are simple interventions like hand washing, respiratory etiquette and mask use in appropriate settings: large proven benefit, no/little harm/cost. , even before the covid- pandemic, we had randomized trials showing % reduced odds of influenza infection with hand washing and (non-statistically significant, but possible) % reduced odds with proper mask wearing. despite lack of trials, it is sensible and minimally disruptive to avoid mass gatherings and decrease unnecessary travel. prolonged draconian lockdown is not equivalent to seat belts. it resembles forbidding all commute. similarly fire insurance offers a misleading analogy. fire insurance makes sense only at reasonable price. draconian prolonged lockdown may be equivalent to paying fire insurance at a price higher than the value of the house. taleb refers to the netherlands where maximum values for flooding, not the mean, are considered. anti-flooding engineering has substantial cost, but a favorable decision-analysis profile after considering multiple types of impact. lockdown measures were decided based on examining only one type of impact, covid- . moreover, the observed flooding maximum to-j o u r n a l p r e -p r o o f journal pre-proof date does not preclude even higher future values. netherlands aims to avoid devastation from floods occurring once every , years in densely populated areas. a more serious flooding event (e.g. one that occurs every , years) may still submerge the netherlands next week. however, prolonged total lockdown is not equivalent to building higher sea walls. it is more like abandoning the country -asking the dutch to immigrate, because their land is quite unsafe. other natural phenomena also exist where high maximum risks are difficult to pinpoint and where new maxima may be reached. e.g., following taleb's argumentation, one should forbid living near active volcanoes. living at the santorini caldera is not exciting, but foolish: that dreadful island should be summarily evacuated. same applies to california: earthquake devastation may strike any moment. prolonged lockdown zealots might barely accept a compromise: whenever substantial seismic activity occurs, california should be temporarily evacuated until all seismic activity ceases. furthermore, fat-tailed uncertainty and approaches based on extreme value theory may be useful before a potentially high-risk phenomenon starts and during its early stages. however, as more data accumulate and the high-risk phenomenon can be understood more precisely with plenty of data, the laws of large numbers may apply and stochastic rather than chaotic approaches may become more relevant and useful than continuing to assume unlikely extremes. further responses to taleb appear in table . the short answer is: using science and more reliable data. we can choose measures with favorable benefit-risk ratio, when we consider together multiple types of impact, not only on covid- , but on health as a whole, as well as society and economy. j o u r n a l p r e -p r o o f currently we know that approximately half of the covid- deaths in europe and the usa affected nursing home residents. , another sizeable proportion were nosocomial infections. if we protect these locations with draconian hygiene measures and intensive testing, we may avert % of the fatalities without large-scale societal disruption and without adverse consequences on health. other high-risk settings, e.g. prisons, homeless shelters, meatprocessing plants also need aggressive protection. for the rest of the population, we have strong evidence on a very steep age gradient with ~ -fold differences in death risk for people > years old versus children. we have also detailed insights on how different background diseases modify covid- risk for death or other serious outcome. we can use hygiene and some least disruptive distancing measures to protect people. we can use intensive testing (i.e. again, use science) to detect resurgence of epidemic activity and extinguish it early -the countries that faced most successfully the first wave, e.g. singapore and taiwan, did exactly that highly successfully. we can use data to track how the epidemic and its impact evolves. data can help inform more granular models and titrate decisions considering distributions of risk ( figure ). poorly performing models and models that perform well for only one dimension of impact can cause harm. it is not just an issue of academic debate, it is an issue of potentially devastating, wrong decisions. taleb seems self-contradicting: does he espouse abandoning all models (since they are so wrong) or using models but always assuming the worst? however, there is no single worst scenario, but a centile of the distribution: should we prepare for an event that has . %, . %, or . % chance of happening? paying what price in harms? abandoning all epidemic modeling appears too unrealistic. besides identifying the problems of epidemic modeling, table also offers suggestions on addressing some of them. to summarize here some necessary (although not always sufficient) targets for amendments: • invest more on collecting, cleaning, and curating real, unbiased data, not just theoretical speculations unfortunately flirting with this slippery, defensive path. total lockdown is a bundle of dozens of measures. some may be very beneficial, but some others may be harmful. hiding uncertainty can cause major harm downstream and leaves us unprepared for the future. for papers that fuel policy decisions with major consequences, transparent availability of data, code, and named peer-review comments is also a minimum requirement. the possibility of calibrating model predictions for looking at extremes rather than just means is sensible, especially in early days of pandemics, when much is unknown about the virus and its epidemiological footprint. however, when calibration/communication on extremes is adopted, one should also consider similar calibration for the potential harms of adopted measures. for example, tuberculosis has killed billion people in the last years, it still kills . million people (mostly young and middle age ones) annually, and prolonged lockdown may use of extreme case predictions for covid- deaths should be co-examined with extreme case predictions for deaths and impact from many other lockdown-induced harms. models should provide the big picture of multiple dimensions. similar to covid- , as more reliable data accrue, predictions on these other dimensions should also be corrected accordingly. eventually, it is probably impossible (and even undesirable) to ostracize epidemic forecasting, despite its failures. arguing that forecasting for covid- has failed should not be misconstrued to mean that science has failed. developing models in real time for a novel virus, with poor quality data, is a formidable task and the groups who attempted this and made public their predictions and data in a transparent manner should be commended. we readily admit that it is far easier to criticize a model than to build one. it would be horrifically retrograde if this debate ushers in a return to an era where predictions, on which huge decisions are made, are kept under lock and key (e.g. by the government -as is the case in australia). we wish to end on a more positive note, namely where we feel forecasting has been helpful. perhaps the biggest contribution of these models is that they serve as a springboard for discussions and debates. dissecting variation in performance of various models (e.g. casting a sharp eye to circumstances where a particular model excelled) can be highly informative and a systematic approach to the development and evaluation of such models is needed. this demands a coherent approach to collecting, cleaning and curating data, as well as a transparent approach to evaluating the suitability of models with regard to predictions and forecast uncertainty. what we have learned from the covid- pandemic can be passed to future generations that hopefully should be better prepared to deal with a new, different pandemic, learning from our failures. there is no doubt that, again, an explosive literature of models and forecasting will emerge again as soon as a new pandemic is suspected. however, we can learn from our current mistakes to be more cautious with interpreting, using, and optimizing these models. being more cautious does not mean not to act decisively, but it requires looking at the totality of the data; considering multiple types of impact; having scientists from very different disciplines involved; replacing speculations, theories and assumptions with real, empirical data as quickly as possible; and modifying and aligning decisions to the evolving best evidence. in the current pandemic, we largely failed to protect people and settings at risk. we could have done much better in this regard. it is difficult to correct mistakes that have already led to people dying, but we can avoid making the same mistakes in future pandemics from different pathogens. we can avoid making the same mistakes even for covid- going forward, since this specific pandemic has not ended as we write. in fact, its exact eventual impact is still unknown. for example, the leader of the us task force, dr. anthony fauci, recently warned of reaching , covid- us cases per day. maybe this prediction is already an underestimate, because with over , cases diagnosed per day in early july , the true number of infections may be many times larger. there is currently wide agreement that the number of infections in many parts of the united states is more than times higher than the reported rates. "according to the penn wharton budget model (pwbm), reopening states will result in an additional , deaths from the virus -even if states don't reopen at all and with social distancing rules in place. this means that if the states were to reopen, , people in total would die from coronavirus by the end of june, the study found." yahoo, may , (https://www.yahoo.com/now/reopeningstates-will-cause- -more-people-to-die-from-coronavirusaccording-to-wharton-model- .html) based on jhu dashboard death count, number of additional deaths as of june was , instead of , , i.e. total deaths was . thousand instead of thousand. it is unclear also whether any of the , deaths were due to reopening rather than error in the original model calibration of the number of deaths without reopening. "dr. ashish jha, the director of the harvard global health institute, told cnn's wolf blitzer that the current data shows within less than weeks of this quote the number of j o u r n a l p r e -p r o o f that somewhere between to , americans are dying from the virus daily, and even if that does not increase, the us is poised to cross , deaths sometime in september. "i think that is catastrophic. i think that is not something we have to be fated to live with," jha told cnn. "we can change the course. we can change course today." "we're really the only major country in the world that opened back up without really getting our cases as down low as we really needed to," jha told cnn." business insider, june , (https://www.businessinsider.com/harvard-expert-predictscoronavirus-deaths-in-us-by-september- - ) daily deaths was much less than the - quote ( daily average for the week ending july ). then it increased again. the number of actual total deaths as of september will be added here when available. lack of consensus as to what is the 'ground truth" even for seemingly hard-core data such as daily the number of deaths. they may vary because of reporting delays, changing definitions, data errors, and more reasons. different models were trained on different and possibly highly inconsistent versions of the data. as above: investment should be made in the collection, cleaning and curation of data. wrong assumptions in the modeling many models assume homogeneity, i.e. all people having equal chances of mixing with each other and infecting each other. this is an untenable assumption and in reality, tremendous heterogeneity of exposures and mixing is likely to be the norm. unless this heterogeneity is recognized, estimated of the proportion of people eventually infected before reaching herd immunity can be markedly inflated need to build probabilistic models that allow for more realistic assumptions; quantify uncertainty and continuously readjust models based on accruing evidence high sensitivity of estimates for models that use exponentiated variables, small errors may result in major deviations from reality inherently impossible to fix; can only acknowledge that uncertainty in calculations may be much larger than it seems lack of incorporation of epidemiological features almost all covid- mortality models focused on number of deaths, without considering age structure and comorbidities. this can give very misleading inferences about the burden of disease in terms of quality-adjusted life-years lost, which is far more important than simple death count. for example, the spanish flu killed young people with average age of and its burden in terms of number of quality-adjusted person-years lost was about fold higher than the covid- (at least as of june , ). incorporate best epidemiological estimates on age structure and comorbidities in the modeling; focus on quality-adjusted lifeyears rather than deaths the core evidence to support "flatten-the-curve" efforts was based on observational data from the spanish flu pandemic on us cites. these data are > -years old, of questionable quality, unadjusted for confounders, based on ecological reasoning, and pertaining to an entirely different while some interventions in the broader package of lockdown measures are likely to have beneficial effects, assuming huge benefits is incongruent with the past (weak) evidence and should be avoided. large benefits may be feasible from j o u r n a l p r e -p r o o f (influenza) pathogen that had ~ -fold higher infection fatality rate than sars-cov- . even thus, the impact on reduction on total deaths was of borderline significance and very small ( - % relative risk reduction); conversely many models have assumed -fold reduction in deaths (e.g. from , deaths to , deaths in the imperial college model) with adopted measures precise, focused measures (e.g. early, intensive testing with through contact tracing for the early detected cases, so as not to allow the epidemic wave to escalate [e.g. taiwan or singapore]; or draconian hygiene measures and thorough testing in nursing homes) rather than from blind lockdown of whole populations. lack of transparency many models used by policy makers were not disclosed as to their methods; most models were never formally peer-reviewed and the vast majority have not appeared in the peer-reviewed literature even many months after they shaped major policy actions while formal peer-review and publication may take more time unavoidably, full transparency about the methods, and sharing of the code and data that inform these models is indispensable. even with peer-review, many papers may still be glaringly wrong, even in the best journals. errors complex code can be error-prone and errors can happen even by experienced modelers; using oldfashioned software or languages can make things worse; lack of sharing code and data (or sharing them late) does not allow detecting and correcting errors promote data and code sharing; use up-todate and well-vetted tools and processes that minimize the potential for error through auditing loops in the software and code lack of determinacy many models are stochastic and need to have a large number of iterations run, perhaps also with appropriate burn-in periods; superficial use may lead to different estimates promote data and code sharing to allow checking the use of stochastic processes and their stability looking at only one or a few dimensions of the problem at hand almost all models that had a prominent role in decision-making focused on covid- outcomes, often just a single outcome or a few outcomes (e.g. deaths, or hospital needs). models prime for decision-making need to take into account the impact on multiple fronts (e.g. other aspects of health care, other diseases, dimensions of the economy, etc.) interdisciplinarity is desperately needed; since it is unlikely that single scientists or even teams can cover all this space, it is important for modelers from diverse ways of life to sit on the same table. major pandemics happen rarely and what is needed are models which fuse information from a variety of sources. information from data, from experts in the field, from past pandemics, need to fused in a logically consistent fashion if we wish to get any sensible predictions. lack of expertise in crucial disciplines the credentials of modelers are sometimes undisclosed; when they have been disclosed, these teams are led by scientists who may have strengths in some quantitative fields, but these fields may be remote from infectious diseases and clinical epidemiology; modelers may operate in subject matter vacuum make sure that the modelers' team is diversified and solidly grounded in terms of subject matter expertise groupthink and bandwagon effects models can be tuned to get desirable results and predictions, e.g. by changing the input of what are deemed to be plausible values for key variables. this is especially true for models that depend on theory and speculation, but even data-driven forecasting can do the same, depending on how the modeling is performed. in the presence of strong groupthink and bandwagon effects, modelers may consciously fit their predictions to what is the dominant thinking and expectations -or they may be forced to do so. maintain an open-minded approach; unfortunately models are very difficult, if not impossible, to pre-register, so subjectivity is largely unavoidable and should be taken into account in deciding how much forecasting predictions can be trusted j o u r n a l p r e -p r o o f forecasts may be more likely to be published or disseminated, if they are more extreme very difficult to diminish, especially in charged environments; needs to be taken into account in appraising the credibility of extreme forecasts j o u r n a l p r e -p r o o f inform the public that we are doing our best, but it is likely that hospitals will be overwhelmed by covid- honest communication with the general public patients with major problems like heart attacks did not come to the hospital to be treated, while these are diseases that are effectively treatable only in the hospital; an unknown, but probably large share of excess deaths in the covid- weeks were due to these causes rather than covid- itself re-orient all hospital operations to focus on covid- be prepared for the covid- wave, strengthen the response to crisis most hospitals saw no major covid- wave and also saw a massive reduction in overall operations with major financial cost, leading to furloughs and lay-off of personnel; this makes hospitals less prepared for any major crisis in the future j o u r n a l p r e -p r o o f table . taleb's main statements and our responses forecasting single variables in fat tailed domains is in violation of both common sense and probability theory. serious statistical modelers (whether frequentist or bayesian) would never rely on point estimates to summarize a skewed distribution. using data as part of a decision process is not a violation of common sense, irrespective of the distribution of the random variable. possibly using only data and ignoring what is previously known (or expert opinion or physical models) may be unwise in small data problems. we advocate a bayesian approach, incorporating different sources of information into a logically consistent fully probabilistic model. we agree that higher order moments (or even the first moment in the case of the cauchy distribution) do not exist for certain distributions. this does not preclude making probabilistic statements such as p(a<x<b). pandemics are extremely fat tailed. yes, and so are many other phenomena. the distribution of financial rewards from mining activity, for example, is incredibly fat tailed and very asymmetric. as such, it is important to accurately quantify the entire distribution of forecasts. from a bayesian perspective, we can rely on the posterior distribution (as well as the posterior predictive distribution) as the basis of statistical inference and prediction. science is not about making single points predictions but understanding properties (which can sometimes be tested by single points). we agree and that is why the focus should be on the entire predictive distribution, and why we should be flexible in the way in which we model and estimate this distribution. bayesian hierarchical models (which can be formulated to account for fat tails, if need be) may allow using all available sources of information in a logically consistent fully probabilistic model. j o u r n a l p r e -p r o o f "if i were to make an informed estimate based on the limited testing data we have, i would say that covid- will result in fewer than , deaths this season in the usa" -my quote appeared on april in cnn and washington post based on a discussion with fareed zakaria a few days earlier. fareed is an amazingly charismatic person and our discussion covered a broad space. while we had focused more on the need for better data, when he sent me the quote that he planned to use, i sadly behaved like an expert and endorsed it. journalists and the public want certainty, even when there is no certainty. here is an effort to dissect why i was so wrong. there is more to this: since mid-march, i wrote an article alerting that there are two settings where the new virus can be devastating and that we need to protect with draconian measures: nursing homes and hospitals. over several weeks, i tried unsuccessfully to publish this in three medical journals and in top news venues that i respect. among top news venues, one invited an op-ed, then turned it down after one week without any feedback. conversely, the new york times, offered multiple rounds of feedback over days. eventually, they rewrote entirely the first half, stated it will appear next day, then said they were sorry. stat kept it for days and sent extensive, helpful comments. i made extensive revisions, then they rejected it apparently because an expert reviewer told them "no infectious disease expert thinks this way" -paradoxically, i am trained and certified in infectious diseases. - % of deaths in the us (and as many or more in several european countries) eventually were in nursing homes and related facilities and probably another large share were nosocomial, hospital-acquired infections. an editor/reviewer at a top medical journal dismissed the possibility that many hospital staff are infected. seroprevalence and pcr studies, however, have found very high infection rates in health care workers - and in nursing homes. , had we dealt with this coronavirus thinking what other widely-circulating coronaviruses do based on medical or infectious disease (not modeling) textbooks (=they cause mostly mild infections, but they can devastate specifically nursing homes and hospitals), , , my foolish prediction might have been less ridiculous. why was my article never accepted? perhaps editors were influenced by some social media who painted me and my views as rather despicable and/or a product of "conservative ideology" (a stupendously weird classification, given my track record). as i say in my stanford webpage: "i have no personal social media accounts -i admire people who can outpour their error-free wisdom in them, but i make a lot of errors, i need to revisit my writings multiple times before publishing, and i see no reason to make a fool of myself more frequently than it is sadly unavoidable." so, here i stand corrected. predictions for number of us deaths during week (only ~ weeks downstream) with these models ranged from to , a . -fold difference, and the spectrum of % confidence intervals ranged from fewer than deaths to over , deaths, almost a -fold difference. predictions for number of us deaths during week (only ~ weeks downstream) with these models ranged from to , a . -fold difference, and the spectrum of % confidence intervals ranged from fewer than deaths to over , deaths, almost a -fold difference. the actual number of deaths during week will be added when it is complete. classification should be readily attainable. as shown, < % of the population is at high risk (shown with dense-colors and thus worth of special protection and more aggressive measures), and these people account for > % of the potential deaths. > % of the population could possibly continue with non-disruptive measures, since they account for only < % of the total potential deaths. wrong but useful -what covid- epidemiologic models can and cannot tell us will have million cases of covid- in four weeks, doubling every four days forecasting covid- impact on hospital bed-days, icu-days, ventilator-days and deaths by us state in the next months reduced rate of hospital admissions for acs during covid- outbreak in northern italy collateral damage: the impact on outcomes from cancer surgery of the covid- pandemic years of life lost due to the psychosocial consequences of covid mitigation strategies based on swiss data should governments continue lockdown to slow the spread of covid- ? the totality of the evidence learning as we go: an examination of the statistical accuracy of covid daily death count predictions a case study in model failure? covid- daily deaths and icu bed utilisation predictions the influenza pandemic review transmission intensity and impact of control policies on the foot and mouth epidemic in great britain the foot-and-mouth epidemic in great britain: pattern of spread and impact of interventions use and abuse of mathematical models: an illustration from the foot and mouth disease epidemic in the united kingdom estimating the human health risk from possible bse infection of the british sheep flock tail risk of contagious diseases clinical impact of human coronaviruses e and oc infection in diverse adult populations an outbreak of human coronavirus oc infection and serological cross-reactivity with sars coronavirus complete genomic sequence of human coronavirus oc : molecular clock analysis suggests a relatively recent zoonotic coronavirus transmission event five ways to ensure that models serve society: a manifesto estimating the effects of non-pharmaceutical interventions on covid- in europe the illusory effects of non-pharmaceutical interventions on covid- in europe us could see , new covid- cases per day key: cord- - g zzj o authors: farooq, junaid; bazaz, muhammad abid title: a novel adaptive deep learning model of covid- with focus on mortality reduction strategies date: - - journal: chaos solitons fractals doi: . /j.chaos. . sha: doc_id: cord_uid: g zzj o we employ deep learning to propose an artificial neural network (ann) based and data stream guided real-time incremental learning algorithm for parameter estimation of a non-intrusive, intelligent, adaptive and online analytical model of covid- disease. modeling and simulation of such problems poses an additional challenge of continuously evolving training data in which the model parameters change over time depending upon external factors. our main contribution is that in a scenario of continuously evolving training data, unlike typical deep learning techniques, this non-intrusive model eliminates the need to retrain or rebuild the model from scratch every time a new training data set is received. after validating the model, we use it to study the impact of different strategies for epidemic control. finally, we propose and simulate a strategy of controlled natural immunization through risk based population compartmentalization (pc) wherein the population is divided in low risk (lr) and high risk (hr) compartments based on risk factors (like comorbidities and age) and subjected to different disease transmission dynamics by isolating the hr compartment while allowing the lr compartment to develop natural immunity. upon release from the preventive isolation, the hr compartment finds itself surrounded by enough number of immunized individuals to prevent spread of infection and thus most of the deaths occurring in this group are avoided. o an effective strategy to minimize the number of deaths through controlled natural immunization in absence of availability of vaccination at mass level. covid- is a highly contagious epidemic disease caused by novel coronavirus (sars-cov- ) that originated in wuhan, hubei province of china in late december . world health organization (who) declared covid- as a pandemic on th march [ ] . researchers are policy makers are working round the clock to find solutions and design strategies to control the pandemic and minimize its impact on human health and economy. the transmission of sars-cov- in humans is mostly through respiratory droplets (sneezing, coughing and while talking) and through contaminated surfaces [ ] . the most significant property of sars-cov- is that it can persist on a variety of surfaces from hours to days at room temperature which makes its transmission more rapid [ ] . this virus can cause acute respiratory distress syndrome (ards) or multiple organ dysfunction, which may lead to physiological deterioration and death of an infected individual [ ] . mathematical modeling of infectious diseases and epidemics has been employed as an important tool for analysis of disease characteristics and investigation of disease spread ever since the ground breaking work of kermack and mckendrick in [ ] . it plays a useful role in efficient decision making and optimal policy framing. different models have been developed to analyse the transmission dynamics of many infectious diseases like malaria (ronald ross model) [ ] , cholera (capasso and pareri-fontana model, ) [ ] , gonorrhea (hethcote and yorke model, ) [ ] , ebola [ ] , h n [ ] etc. in this work, we employ deep learning to propose an artificial neural network (ann) based real-time online incremental learning technique to estimate parameters of a data stream guided analytical model of covid- to study the transmission dynamics and prevention mechanism for sars-cov- novel coronavirus in order to aid in optimal policy formulation, efficient decision making, forecasting and simulation. modeling and simulation of such problems poses an additional challenge of continuously evolving training data in which the model parameters change over time depending upon external factors. our main contribution is that in a scenario of continuously evolving training data, unlike typical deep learning techniques, this model eliminates the need to retrain or rebuild the model from scratch every time a new training data set is received. using a data science approach, model parameters are intelligently adapted to the new ground realities. to the best of our knowledge, this paper develops for the first time a deep learning model of epidemic diseases with data science approach in which parameters are intelligently adapted to the new ground realities with fast evolving infection dynamics. the covid- data from india has been taken as the case study. the first case of covid- in india was reported on th january originating from wuhan, china [ ] . as on june , the total number of cases reported in india is , with , recoveries and , deaths [ ] . hence the number of active cases is , . the government of india imposed a country wide complete lockdown on th march with strict restrictions on the movement of people while allowing only the essential services to operate under the supervision of administration and health officials. the lockdown was renewed thrice on april, may and may and has been relaxed since june [ ] . india is the second largest populated country in the world with a total population of around . billion. the health care facilities in india are considered poor with . hospital beds per thousand people of the population [ ] . therefore, the covid- pandemic has emerged as a major challenge for the people, health workers and policy makers of the country. using a control theory approach, we analyze the stability of different disease prevention strategies. finally, we propose a strategy of controlled natural immunization of the population by dividing it in low risk and high risk compartments based on various risk factors like age and comorbidities. the two groups are treated separately and subjected to different disease mechanics with the aim to minimize the total number of deaths given the fact that the probability of death is very high in the high risk group as compared to the low risk group. the two compartments are isolated from each other for a certain period of time. the low risk compartment is allowed to fully brace the infection with maximum speed and develop immunity owing to its very low death rate while as the high risk compartment is put under preventive isolation till the infection growth curves fatten for the low risk group. upon release from the preventive isolation, the high risk group finds itself surrounded by enough number of immunized individuals to prevent spread of infection and thus most of the deaths occurring in this group are avoided. we simulate this strategy in matlab environment to establish its effectiveness in significant reduction in the number of deaths while demonstrating the usefulness of the deep learning based mathematical model. to the best of our knowledge, such an approach to reduce mortality has not been modeled and simulated earlier by the scientific community. sirvd refers to susceptible, infected, recovered, vaccinated and deceased states of individuals in a population going through an epidemic. the pioneer work in development of mathematical models for infectious diseases was carried out by by [ ] known as the susceptible-infectious-recovered (sir) model. as one of the most classical models, it has been used by many researchers to study and analyse many infectious diseases like seasonal flu [ , ] , pandemic flu [ , ] , hiv/aids [ ] , sars [ , ] etc. these studies have shown that sir models are reliable for analysis of the infectious disease spread and evaluation of the impact of prevention schemes in different scenarios. the basic sir model is described by the following differential equations: where s , i and r are functions of time representing the number of susceptible, infected and recovered individuals in a population of size n at time t. β is the rate of transmission and γ is the rate of recovery of infected individuals. it is assumed that those recovered develop immunity and do not catch the infection again in the time span of interest. the basic sir model can be modified in various ways to accommodate different scenarios. a modified sir model known as sird (susceptible-infected-recovered-deceased) model is of our interest here and is based on the following assumptions: (i) this model is fatal unlike a typical non-lethal sir model which means that there is a positive probability of an infected person dying, p(death) = δ and δ ≥ . (ii) a typical sir model assumes that the recovered group gains full immunity from reinfection. however, this model accommodates the possibility of a recovered person being reinfected with probability of reinfection, p(rein f ection) = σ and σ ≥ . (iii) the impact of new births and unrelated deaths in ignored and the total population remains constant, n = constant ∀ t. (iv) the population is distributed randomly over the area. therefore, there are four classes of individuals: susceptible (s), infected (i), recovered (r) and deceased (d) as described by the following equation: where β = rate of infection, σ = rate of susceptibility, γ = rate of recovery and δ = rate of death. since the final cure for covid- pandemic is the successful discovery and optimal administration of the vaccine in the population, therefore we introduce the effect of vaccination with a given rate of vaccination under resource limited settings. this is achieved by adding a new class of individuals called vaccinated (v) in the population. it can be fairly assumed that there is no limit on the total number of vaccines produced as all the available resources for vaccine production are employed to eliminate the epidemic. however, the vaccine production capacity will have some limit based on availability of resources and facilities. therefore, there will be a limited number of vaccines available at a particular point of time. thus, it is assumed that the per capita rate of vaccination, α < where is a constant. it is assumed that the vaccination imparts long term immunity against the disease in the vaccination individuals. based on these assumption, we propose a final model as described by the following first order ordinary differential equations (ode's) and it illustrated by figure : it must be noted here that the process of mass action transmission is described by the non-linear term βs i/n where β = number of contacts per unit time by a person in group i required to transmit the disease to a person in group s, n − ≈ n = total number of possible contacts of a person, s /n = the fraction of possible contacts of a person that are from group s, i = the number of infected persons at time t, therefore βs i/n = number of people transmitted from group s to group i per unit of time. the next task is to to learn the model parameters which can be quite challenging in an epidemic scenario like covid- as the the model parameters are supposed to change with time. this section proposes an artificial neural network (ann) based adaptive incremental learning technique (annail) for online learning of the sirvd model parameters with the following assumptions: (i) the rate of vaccination α as a function of time (t) is set by the vaccine production capacity decided by availability of skilled labour, resources and facilities and by vaccination policy as well. maximum vaccine production capacity α max has been assumed to be constant indicating that there is no change in vaccine production infrastructure, technology or facilities. (ii) the rate of infection β as a function of time (t) is the major challenge for parameter learning. it is affected by external factors like degree of social distancing, lockdown etc. in case of a lockdown β decreases exponentially. therefore, in order to take into account both the lockdown and no lockdown scenarios, β has been modelled as: where t l is the time when the lockdown begins. therefore, the learning algorithm has to learn parameters (β , β , τ β ) to find β. (iii) the rate of reinfection σ has been assumed to be zero for covid- disease as the human body develops antibodies to prevent re-infections in future against such a virus [ ] . (iv) rate of recovery γ and rate of death δ are affected by factors like change in health care facilities, possible overcrowding of hospitals, development of new drugs to manage or treat the disease etc. both of these have been assumed to be constant in this paper. for a typical neural network or any other technique of model parameter estimation, the training data is required first to train the model before applying it on future scenarios. however, in case of an epidemic like covid- , the training data is continuously evolving with time and the model needs to be trained and executed at the same time as the model parameters may change over time based on different external factors like government policies, social distancing etc which can be known only from newly arriving data sets. therefore, we propose a technique for the model to learn these parameters from new data sets in an adaptive manner while continuously updating the old models without the need to build the model from scratch every time a new training data set is received. deep learning and other machine learning techniques stand out in solving problems of data based model parameter estimation due to their state-of-the-art results. however, they face the problem of catastrophic forgetting which reduces their performance as new training data becomes available with time. this is because the typical neural networks require the entire dataset to update the model each time a new training data set becomes available as in case of an epidemic modelling problem where the training data becomes available incrementally with time. to address such issues, different incremental learning algorithms have been suggested in the literature [ , , ] . incremental learning refers to an online learning technique of continuous model adaptation under a scenarios of continuously evolving training data. therefore, storage or access to the previously observed data is not required each time a new data set is received, as in case of an epidemic like covid- . in order to adapt the model parameters in light of new data, it is not needed to use all the previously accumulated data for developing the model from scratch. rather, the learning network modifies the previous hypothesis to adapt to the new data chunk. in this paper, we propose hypothesis generation via an artificial neural network (ann). let d j− be the data set received between time t j− and t j , and h j− be the hypothesis generated on this data set. the hypothesis h j for a new data set d j received between time t j and t j+ is a function of d j and h j− only as under: the experience gained from this step is stored and integrated to support in future adaptation process. thus the objective here is to integrate the previously learned knowledge into the new raw data set to adapt the model parameters accordingly; and to accumulate this experience over time to increase the model efficiency, accuracy and flexibility. the proposed framework for the above problem is shown in figure . the ann is based on a non-linear activation function for successful regression analysis. the hidden layers are represented by the function f nn . with the continuous data stream, the weight distribution functions are generated to describe the learning capability of the ann where the decision boundary is adjusted to focus especially on the hard to learn data examples. the algorithm for this framework is given in algorithm . , where x i represents the input vector and y i represents the output. -associated mapping function output: where f nn is the ann defined mapping function. (ii) apply hypothesis h t− to d t , and find the pseudo-error (vi) repeat the above procedure for d t+ . the final hypothesis is: where t is the set of incrementally developed hypotheses in the learning life. this algorithm is run in a top-down and horizontal signal flow, as shown in figure . the adaptive nature of this algorithm is due to the mapping function based on ann which estimates the initial distribution functionΦ t− for d t while providing a quantitative approach to indicate the learning power of the new data set based on previously trained model.Φ t− is applied to the new data set to find pseudo-error, Υ. thus a hard to learn example will have higher Υ in step (iii) of the learning procedure, and will in turn receive higher weight in step (iv). this ensures the adaptive nature of the algorithm. mapping function connects the past experiences to the new data in an adaptive fashion. there can be many ways to design the mapping function. however, we implemented the nonlinear regression by an ann based approximation of mapping function owing to its flexibility. any such neural network based function approximation technique can be used. as in illustration we take the multilayer perceptron (mlp) in this paper. this is shown in figure . the input is an n-dimensional vector (for example, number of infections, deaths, recoveries in an epidemic) of example i. the distribution function is currently estimated as j t− , w represents the weights of a layer. backpropagation is used to tune the weights w of different layers, where error function is defined as where k is the training epoch of the backpropagation. the neural network gives the following output: where h f represents the input to fth hidden node while as g f represents its output, υ is the input to the final node, n h is number of hidden neurons, and n is the total number of inputs. weights on the ann are updated by applying the above defined backpropagation strategy as explained below. backpropagation: weight adjustment for the hidden to out- similarly, weight adjustments for the input to hidden layer is described as: where α(k) describes the learning rate. estimation of initial distribution functionΦ t for d t required only the feedforward path of the mlp. this model was validated for covid- in india where the first % of data was used for training and the remaining % was used for testing as shown in figure . it is clearly evident from the plots shown in the figure that the results given by the model during testing are very close to the actual data. the inputs and outputs in this algorithm were: inputs : number of new infections, deaths and recoveries; rate of vaccination (α). outputs : β , β , τ β , δ, γ these outputs are fed to the analytical sirvd model at every time instant when a new set of input data is received to simulate and forecast different scenarios. in this section, we analyse three possible strategies to combat an epidemic like covid- : (i) herd immunity, (ii) complete vaccination, (iii) complete lockdown and finally our proposed (iv) controlled natural immunization through risk based population compartmentalization is discussed in the next section. however, following definitions are needed beforehand: definition . (stability) if the jacobian matrix j for a system of n differential equations has eigenvalues λ , λ , ..., λ n , for a trivial steady state equilibrium at ( , ,..., ), then the stability of the solution is determined as following: .., n then the system has uniform and asymptotic stability (uas). .., n and the algebraic multiplicity equals the geometric multiplicity whenever λ i = for any i, then the system has uniform stability (us). (iii) if re(λ i ) > for any i and the algebraic multiplicity is greater then the geometric multiplicity whenever λ i = for any i, then the system has instability. since the impact of new births and unrelated deaths has been fairly ignored for the purpose of this study and it has been assumed that the total population n stays constant throughout the epidemic, therefore the system of odes ( - ) describing the model satisfies the following two conditions: this shows that a state of equilibrium always exists in the system. definition . (dfe) a disease free equilibrium (dfe) is defined as a state of equilibrium according to ( , ) in which the number of infected and recovered individuals equal to zero such that there are no further deaths (i → d), infections (s → i) or reinfections (r → i): which gives us the following from the system of odes ( - ) : we take the jacobian matrix of the above system to evaluate its stability. however, the fifth differential equation (representing d) is uncoupled from the first four differential equations and it can be derived from the first four equations using ( , ) , therefore we consider the jacobian for the first four equations only which is given as: given below is the investigation of dfe stability for different epidemic control strategies. herd immunity is the idea that a virus cannot spread easily after enough people develop immunity against it. this reduces the chances of the virus transmitting from person to person and infecting those who haven't been infected yet [ ] . such an immunity can be induced artificially by a vaccine, however it can also be developed naturally by the infection itself as the immune system of the body develops antibodies against the virus which prevent reinfection in future. this is based on the fact that the human body produces a non-specific innate response to a viral infection initially using neutrophils, macrophages, and dendritic cells. however, this is followed by a more specific adaptive response in the form of development of proteins called immunoglobulins which act as antibodies specifically binding to the virus. this is coupled with the formation of t-cells which generate cellular immunity by identifying and eliminating the cells that are infected with the virus. generally, sufficient presence of such antibodies in collaboration with cellular immunity prevents reinfection after recovery. although every recovered patient may not develop complete immunity, but that is the case for the most of them [ ] . in case of covid- , although research is till going on to reach a conclusive opinion, promising studies suggest that nearly all the recovered patients develop such antiviral immunoglobulin-g (igg) antibodies and are immune to reinfection [ ] . this lays the basis for mass serological testing for covid- being practised by many governments across the world in which the blood samples of people are tested for the presence of these antibodies indicating present or past covid- infection. further, the data on reinfection, even if rare, is not available. therefore, the idea of herd immunity is to let the infectious disease take its natural course of action and let the population naturally develop immunity against the disease after most of the population gets infected. the dfe after herd immunity has the following form: where d is the total number of deaths at the time of dfe and thus r = n − d . since there is no vaccination, therefore α = and v = . further, the rate of reinfection σ is zero as well, meaning recovered people cannot catch the infection again. using these values in ( ), the jacobian at the dfe after herd immunity is given as: equating characteristic equation for this jacobian to zero gives: therefore, the eigenvalues of this system are: thus, all the eigenvalues have real parts less or equal to zero: re(λ i ) ≤ ∀ i = , , ..., n and d ∈ r ∩ [ , n]. hence, according to definition , the dfe for herd immunity possesses uniform stability (us). in this case, there are no chances of retriggering of the disease after the dfe has been reached. however, this approach has been criticised as dangerous as it will result in a large number of deaths [ ] . to develop herd immunity for covid- in the population, roughly % or more of the population needs to have gone through the infection [ ] . the current death rate in india due covid- is around % which means that if this death rate is maintained, nearly . % of the population will die till % gets infected. this leads to nearly million deaths. the actual number of deaths would be more than this, because with rapid growth of the infectious disease, hospitals would be over-flooded and health care infrastructure will not be able to cater to the demands of high number of patients leading to increase in the death rate. the growth of disease in india is shown in figure if the disease is allowed to take its natural course without any intervention and control. the strategy of complete lockdown focuses on minimization of mobility and contact among the population with maxi-mum possible social distancing. therefore, it minimizes β (the rate of infection). the dfe under complete lockdown has the following form: where d is the total number of deaths at the time of dfe and thus s = n − d . since there is no vaccination, therefore α = and v = . using these values in ( ), the jacobian at the dfe under complete lockdown is given as: furthermore, equating characteristic equation for this jacobian to zero gives: therefore, the eigenvalues of this system are: thus, real part of one of the eigenvalues is always grater than zero: re(λ ) > ∀ d ∈ r ∩ [ , n]. hence, according to definition , the dfe under complete lockdown is always unstable. thus, whenever a dfe is reached in this case, the tendency for the disease to be triggered again is always there. even a single case of infection can restart the disease. apart from the chances of reemergence of the disease, the time taken by the system to reduce the number of active infections to zero may be large enough to make the total lockdown unsustainable and practically impossible. the impact of lockdown on the disease growth in india is shown in figure . it has been assumed that the lockdown ends on day of the disease after it starts on day . it can be compared with figure which shows the disease growth in case no preventive measures are taken. these results confirm that the infection plots shoot up as soon as the lockdown is lifted and there is no significant difference in terms of total number of infections or deaths. the only benefit of lockdown, as can be seen from these plots, is that the peak of infection is delayed which can be useful for the administration to buy some time to prepare the healthcare infrastructure of the country to brace for the full blown impact of the disease. researchers and health industries worldwide are working around the clock to discover a vaccine against sars-cov- [ ]. moderna, a pharmacological company had started clinical testing of its mrna-based vaccine (mrna- ) just months after the complete sequencing of covid was done and published by different research groups [ ] . the most potential candidate for the vaccine development would be to induce our immune system to synthesize neutralizing antibodies against the viral spike protein which block its entry via ace receptors [ ] . the dfe with vaccination of all living members of the population n has the following form: where d is the total number of deaths at the time of dfe and thus v = n − d as the rest of population has been vaccinated and is now immune to the disease. using these values in ( ), the jacobian at the dfe with full vaccination is given as: furthermore, equating characteristic equation for this jacobian to zero gives: therefore, the eigenvalues of this system are: thus, all the eigenvalues have real parts less or equal to zero: members of the population is the way to completely eliminate the disease and its chances of re-triggering as well. however, successful development of a vaccine for sars-cov- has still a long way to go. further, a vaccination rate of million per day is highly ambitious for a country like india. therefore, elimination of the current wave of covid- epidemic and minimization of number of deaths by vaccination is practically impossible as the successful development and mass administration of a vaccine is expected to take more than a year, at least. however, vaccination may still be necessary to prevent new waves of the disease in future. the strategy of herd immunity discussed in the previous section aims at minimizing the number of s (susceptible) and maximizing the number of r (recovered) people who are supposed to have developed the immunity. this will minimize the the factor β s i n in the sirvd model ode's discussed in section . however, this strategy results in the maximum number of deaths. the strategy of complete lockdown aims to minimize the rate of infection (β) by reducing the mobility of people. however, this strategy is not sustainable in the long term as discussed. the strategy of complete vaccination is an ideal solution to the problem. it aims at maximizing the number of people falling in v (vaccinated group) through artificial immunization. however, as discussed, the successful development of a vaccination and its administration in the population is expected to take enough time to cause a large number of deaths. in this section we propose a strategy of minimizing the number of deaths by controlled natural immunization by compartmentalization of the population in two groups: low risk and high risk. high risk group comprises of the people having co-morbidities or are aged above years and as a result have high probability of death under covid- infection. the aim of this strategy is to minimize the number of deaths caused by covid- disease: the death rate (δ) is very high in the high risk group while as it is very low in the low risk group. the most prevalent comorbidity for covid- is hypertension, followed by diabetes with mean age of around . years [ ]. in india, the percentage of population above years of age is . % while as the percentage above is nearly % [ ] . most of the people having comorbidities are expected to fall in above age group. in this study, the high risk group has been assumed to be % of the population. it is proposed that these two population groups be subjected to different disease mechanics. the high risk group is subjected to a preventive quarantine or isolation wherein they are isolated from the low risk group by placing them in separate rooms or sections in homes with minimum contact with the low risk group. whatever necessary contact is required, it should be done with maximum preventive measures like wearing of face masks, sanitization etc. very high risk individuals may be placed in designated care centers where their health needs are met by medical professionals. therefore, for the high risk compartment, β is reduced to minimum as in the case of a lockdown. meanwhile, the low risk group is subjected to maximum mobility and contact among its members effectively increasing β to the maximum possible value. there should be no social distancing and other preventive measures for this group. as a result, the infection will spread very quickly in the low risk group and its members shall develop the immunity and transfer from s to r having gained the immunity naturally with a very low death rate. once this is achieved or in other words once the disease curves are flattened for the low risk group, the high risk group is released from the preventive isolation and allowed to mix with the low risk group. since most of the population ( - %) is already immune and the value of s and i is low, the chances of the high risk group receiving infection from the low risk is very low because the factor β s i n has already been reduced to minimum. during pc, β is minimum for high risk group while as after pc, s i is minimum. after pc, the mobility and the contact in the population, represented by β, should be kept moderate. since, α and σ are zero, therefore: here, β represents the rate of infection which depends upon the mobility and contact among the population. due to the lockdown, this can be considered the lowest possible β in india. we have simulated the pc strategy with different values of β l ranging from to times of β given above to account for high mobility and contact during pc. β h = . β to assume that the isolation of high risk individual is at least times stronger than than the social distancing practised by the whole population during the nation wide lockdown. δ and γ are complementary in the sense that an individual either recovers or dies from the infection. to distribute the current death and recovery rate in low and high risk groups: integrating the equations of sirvd model for this case: where s (t ) and i(t ) are initial states. this is true for both low and high risk groups. however, for high risk group, s h (t ) = i h (t ) = while as for low risk group; where t f is the time when pc ends and the two groups are allowed to remix. after pc: δ and γ have been restored to original values while as β has been multiplied by a factor of to signify higher level of mobility and contact than lockdown albeit with social distancing. the stability analysis and eigenvalues for this case are same as that of herd immunity i.e, it posses uniform stability. this strategy was simulated using the model proposed in section and . the results of the simulation are shown in figure ( - ) . in all these figures, black line represents the number of deaths, blue is the total number of cases, red is the active number of infections while as green line represents the number of recovered people. in figure , the preventive quarantine or compartmentalization of the population is done from day after the start of the population till day while as in in figure the same is continued till day . the rate of infection β is shown as a multiple of the average rate of infection during the nation wide lockdown from march to june, which was the minimum possible. increase of β is achieved by increase of contact and mobility among the population, for example β = x means five times mobility and contact as compared to the days of nation wide lockdown. if no preventive measures are taken, then as per the current trend in the disease growth we may expect more than million deaths in the country if all the population gets infected. this is shown in figure . million deaths is not surprising even if linear growth of the disease is assumed with . % death rate for a population of . billion. different hypothetical experiments were simulated and their results are given in table . as discussed earlier, β is supposed to be kept on lower side after the end of pc and social distancing is advised. these results show that the total number of deaths can be reduced to . million from million, if mobility and contact is made times to that of the lockdown period and pc is ended on day of the pandemic while as after the end of pc, the mobility is reduced to two times. the growth of disease in such a scenario is shown in figure . in this work, we developed an analytical epidemiological model for covid- pandemic where model parameters are continuously updated to intelligently adapt to new data sets using an ann based adaptive online incremental learning technique. in a scenario of continuously evolving training data, unlike typical deep learning techniques, the model eliminates the need to retrain or rebuild the model from scratch every time a new training data set is received. the model was validated and different scenarios were simulated to demonstrate its usefulness and significance. india was taken as a case study. however, this model can be applied to any population in the world and would be a useful tool for policy makers, health officials and researchers in improving decision making efficiency, policy formulation and forecasting. the simulation work was carried out in matlab environment. using this model, we simulated preventive measures like lockdown, vaccination and herd immunity to study their impact on the evolution of covid- disease. finally we proposed an effective method to significantly reduce the number of deaths caused by the pandemic in case a vaccine is not available at the mass level. this technique aims to develop natural immunity in the low risk group of the population by subjecting them to the full blown impact of sars-cov- virus while as subjecting the high risk group to preventive isolation during this time period. once the low risk group develops natural immunity and its disease curves are flattened, the high risk group is released from the preventive isolation. upon release, the high risk group doesn't find enough infected or susceptible people in the environment to catch the infection at a high rate and in this way the maximum number of deaths are avoided in the high risk group. the impact of this strategy has been simulated and it has been shown the the number of deaths can be reduced from million to . million if the population compartmentalization starts tomorrow and ends on day of the pandemic in india. during this period, the mobility and contact in low risk group has to be made five times as compared to the lockdown period and upon remixing of the two groups the mobility and contact should be reduced to times from . the novelty of this paper lies in the use of real-time online incremental learning technique in epidemic disease modeling. many machine learning techniques have been used in epidemic disease modeling [ ], however this paper is the first instance of development of an incremental learning algorithm as a real-time adaptive deep learning technique for parameter estimation of an epidemiological model thus providing the model with the capability to work online i.e, unlike typical machine learning techniques, it doesn't require to rebuild or retrain the model from scratch every time a new data set is received but intelligently adapts the model to ever changing infection dynamics. since the model is non-intrusive, adaptive, intelligent, real-time and online in nature, therefore it can be employed to monitor, forecast and simulate the growth of any infectious disease over a large sized population without losing accuracy, fidelity or com-putational performance due to limitations like run-time duration, size of training data, computational complexity, change in transmission dynamics due to mutations in virus or bacteria, change in prevention mechanisms or government policies. even if the epidemic continues for decades in the whole world, the model will keep working efficiently on daily basis without any decay in performance or rte (run-time environment). further, to the best of our knowledge, population compartmentalization to achieve natural immunity against an infectious disease while significantly reducing the mortality has been modeled and simulated for the first time in this paper. these findings could be highly useful to policy makers around the world to reduce the number of deaths in any country in case a vaccine is not readily available and lockdown is not sustainable economically. further, this is a demonstration of the usefulness and efficiency of deep learning based incremental learning algorithm in model parameter estimation and simulation of different epidemic scenarios. the doctoral research funding from ministry of human resource development, government of india, in favor of the first author is duly acknowledged. the authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. detail/who-director-general-sopening-remarks-at-the-mission-briefing-on-covid aerosol and surface stability of sars-cov- as compared with sars-cov- covid- ards: clinical features and differences to "usual" pre-covid ards contributions to the mathematical theory of epidemics mathematical biology: . an introduction modeling influenza epidemics and pandemics: insights into the future of swine flu (h n ) covid- covid- pandemic lockdown in india transmission of influenza: implications for control in health care settings a bayesian mcmc approach to study transmission of influenza: application to household longitudinal data a 'small-world-like' model for comparing interventions aimed at preventing and controlling influenza pandemics transmissibility of pandemic influenza analysis of recruitment and industrial human resources management for optimal productivity in the presence of the hiv/aids epidemic murray, transmission dynamics and control of severe acute respiratory syndrome transmission dynamics of the etiological agent of sars in hong kong: impact of public health intervention overview of some incremental learning algorithms endto-end incremental learning incremental learning algorithms and applications herd immunity": a rough guide antibody responses to sars-cov- in patients with covid- early herd immunity against covid- : a dangerous misconception estimating the number of infections and the impact of non-pharmaceutical interventions on covid- in european countries covid- vaccine tracker draft landscape of covid- candidate vaccines immune responses in covid- and potential vaccines: lessons learned from sars and mers epidemic ☒ the authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☒the authors declare the following financial interests/personal relationships which may be considered as potential competing interests: key: cord- -fgtatt authors: ghaffarzadegan, navid; rahmandad, hazhir title: simulation‐based estimation of the early spread of covid‐ in iran: actual versus confirmed cases date: - - journal: syst dyn rev doi: . /sdr. sha: doc_id: cord_uid: fgtatt understanding the state of the covid‐ pandemic relies on infection and mortality data. yet official data may underestimate the actual cases due to limited symptoms and testing capacity. we offer a simulation‐based approach which combines various sources of data to estimate the magnitude of outbreak. early in the epidemic we applied the method to iran's case, an epicenter of the pandemic in winter . estimates using data up to march th, , point to , ( % ui: k, . m) cumulative cases and , ( % ui: . k, . k) total deaths, numbers an order of magnitude higher than official statistics. our projections suggest that absent strong sustaining of contact reductions the epidemic may resurface. we also use data and studies from the succeeding months to reflect on the quality of original estimates. our proposed approach can be used for similar cases elsewhere to provide a more accurate, early, estimate of outbreak state. © system dynamics society the novel coronavirus (sars-cov- ), the pathogen that causes covid- infection, is exposing the world to one of its largest global health challenges of recent decades. from public understanding to policy choices, much depends on the data about the epidemics spread and models that integrate such data into actionable policies (kaplan et al., ; thompson et al., ) . yet the official data are highly uncertain, with large variations in quality depending on the country reporting it, and regularly offering lower bounds that have unknown error bounds compared to the reality on the ground. other sources of data, often based on smaller samples, travel screening, and anecdotal evidence may offer relevant hints, but they are not easily generalized or combined with official data. the current paper focuses on using a standard dynamic epidemiological model as a tool for incorporating various sources of data into a unified estimation of the actual trajectory of disease, applying the method to covid- outbreak in iran. our focus is on how well one can realistically estimate the magnitude of the epidemic in early stages when disease parameters are uncertain, data are limited, and projections are highly volatile. current information (as of may , ) point to the spread of covid- starting from a food market in wuhan china in mid-november . the epidemic initially spread mostly in china but has become a global pandemic by may with over million official cases and thousand deaths (worldmeters, ) . the situation may also be worse than official statistics portray. for example, despite a massive screening and response campaign in china, li and colleagues estimate that % of early cases in china were undocumented . the situation in iran, another early epicenter, is instructive. the first official cases were reported on february th, , in the city of qom (wikipedia, ). yet, later reports suggested the disease was likely circulating in iran as early as january (iran international, ) . thus from february th on the spread of disease was very rapid according to official statistics. by march th some , cases and , deaths were recorded. the question in the minds of policy makers and public was: how far will this epidemic go and what policies should be put in place to control and mitigate the risks? the answers heavily depend on understanding the true magnitude of the epidemic. any under-estimation is worrisome, but early in the exponential growth phase of an epidemic such errors could be extremely costly as those estimates drive the mobilization of public health resources and government responses. the risk of under-estimation is partly driven by the characteristics of the disease such as a potentially large population of unrecognized patients with mild symptoms (at least % of the cases have symptoms not very different from common cold or flu) (novel coronavirus pneumonia emergency response epidemiology, ). as a result no consensus also exists on other parameters of the disease, such as fatality rate (fauci et al., ; wu and mcgoogan, ) . moreover, the country-specific variations in measurement and reporting may exceed those due to the nature of the disease. to illustrate, in the next section we provide a quick survey of various early clues related to the magnitude of epidemic in iran, which we then build on in our analysis. there have been a few clues from iran which may inform efforts to estimate the true cases. first is the number of cases identified among travelers arriving in other countries from iran. screening of passengers from high-risk countries in airports is more reliable than most country-level screening statistics. one study in pre-print estimated , total cases of infected individuals in iran by february (tuite et al., ) . the method relied on estimation of cases in the whole country based on three diagnosed cases of infection upon individuals' arrival from iran in various international airports and the likelihood of such an incident given approximately daily outbound passengers from iran during those early days (fraser et al., ). this number was two orders of magnitude larger than official statistics at the time. later reports on the number of infected travelers from iran rose rapidly, to cases by february (radiofarda, ). an article in the atlantic, offered a series of back-of-the-envelope calculations which, with strong simplifying assumptions, estimated million accumulated cases of infected individuals by march th (wood, ) . one may expect the death statistics to be more reliable. but test kits for identifying have been in short supply, and post-mortem testing may not have been a priority of officials in iran. on february th a member of iran's parliament reported deaths only in qom, a city of . million (wikipedia, ) . several reports of government officials contracting the disease have also been released, including reports on infection of member of the iranian parliament (a body of members), as well as several deaths among officials (bbcpersian, b) . a bbc persian report on february th used interviews with an unspecified number of hospitals in iran to put the death from the disease at , an order of magnitude larger than official numbers at the time (bbcpersian, a). another news agency quoted similar sources for a total of deaths by march st (iran international, ) and by march th. we develop a dynamic simulation model of the spread of the disease in iran to estimate the likely trajectory of the disease that is consistent with the evidence summarized above. we start with the traditional seir (for susceptible, exposed, infectious, and recovered stocks representing population groups) model and incorporate feedbacks regulating endogenous changes in contact rate, screening, diagnosis, and reporting in response to risk perception and other relevant factors. thus not only reported statistics, but also the effective reproduction number (r e ), are endogenously generated and can change as people respond to the epidemic. we use this model, along with various strands of data, to weave together an estimate of disease trajectory early on, simulation-based estimation of the early spread of covid- in iran and offer projections for medium term future. keeping in mind the various uncertainties involved in this analysis, reflected in our wide uncertainty intervals (ui), we estimate over , ( % ui: k, . m) cumulative cases of the disease in iran as of march th, with over , ( % ui: . k, . k) deaths. we thus estimate that only . % of cases and . % of deaths are officially attributed to covid- , with the rest going undetected. these results point to extreme gaps between official data and actual trajectory of disease, which may lead to slow response and under-appreciation of risks of the diseases. the rest of the paper is largely focused on documenting the original estimates developed on march th. as a secondary goal we also consider their accuracy in light of the data and findings in the two months following original estimates. this latter exercise provides a window into the actual challenges and promises of conducting policy-relevant projections early in a crisis. figure offers a simple representation of the model's structure. model equations, and parameter values are documented in appendix a. the model builds on the well-known seir (susceptible, exposed, infectious, and recovered) framework. the model includes the 'infection' reinforcing loop (r ) which creates the initial exponential growth in the number of cases. we divide the population of infected to early infected and late infected, and assume that early infected, while largely asymptomatic, might also be infective though at a lower rate (bai et al., ) . the infected population will follow two different paths of recovery or death. a fraction of late infected are symptomatic (in the figure, the variable "symptomatic late infected"). to avoid proliferation of free parameters we use general population averages rather than disaggregating into population groups. to this basic epidemiological model we add two endogenous mechanisms. first, we formulate contact rate to be endogenously changing in response to perceived risk of infection (a function of death statistics). this feedback captures changes in social interactions, gatherings, self-isolation of suspected cases, and hygiene as well as government-mandated closure of events, schools, and businesses in response to perceived risk. we assume public risk perception depends on the number of recent reported cases of death. this balancing feedback (loop b) can bring down contact rate, potentially enough to slow down the epidemic. second, we explicitly model the endogenous changes in screening and reporting of cases over time. here, the increased understanding of risks leads to mobilization of screening resources that expand the fraction of cases that are tested and thus will show up in the official statistics. we note that even with good screening many mild cases will go undiagnosed and will not be in official statistics. this analysis is focused on understanding the spread of the disease in its early phases, about one month after first diagnosed cases, and the core estimations in this paper are based on data only from that early period, until march th, . we use time series data for official reports of death, recovered, and cumulative number of infection over time. we also use unofficial data points including four observations about the number of iranian passengers diagnosed with covid- upon arrival in international airports, and three estimates aggregated by healthcare providers in iran and reported by bbc and iran international news agencies about total cases of death from covid- . the model includes three biological parameters (early-stage exposed period, average duration of illness, and fractional infectivity of early-stage exposed group) which we specify based on prior literature novel coronavirus pneumonia emergency response epidemiology, ; wu and mcgoogan, ) . population size and travel scope are also input using existing data. nine uncertain parameters remain that are estimated using the above data. three of the parameters are used to specify how official measurement and reporting relates to "true" values of infection and death, three parameters estimate public reaction to the reports, and two parameters are for mortality rate among two different groups of patients. finally, the arrival time of first cases of the virus is estimated as a separate parameter. simulation-based estimation of the early spread of covid- in iran our calibration method is based on forming a likelihood function for observing the actual time series data conditional on model parameters. we then conduct a markov chain monte carlo (mcmc) simulation to estimate the joint posterior distribution of the model parameters subject to observed data. we define a likelihood function for change over time (net-inflow) of official reports on cumulative death, recovered and infection assuming they are count events drawn from model-predicted rates (poisson distribution). we use a similar poisson distribution assumption for number of infected passengers and unofficial reports of death as well, since they both fit well into a count measure framework. the mcmc method searches over the feasible ranges for the uncertain parameters and accepts combinations that are consistent with the observed data. similar methods are used frequently in estimating dynamic models in this domain . our prior experience with the use of mcmc methods in nonlinear dynamic models highlights the risk that uncertainty intervals do not capture model mis-specifications and thus may be too tight. we therefore downscale the likelihood function to expand the uncertainty intervals (details and sensitivity in the online supplement) to err on the side of caution in assessing uncertainties and structural nuances not explicitly modeled. more details are reported in appendix a. we follow the procedure described in the modeling section and calibrate our model using multiple data sources and find uncertainty ranges using mcmc. figure a compares best-fitting simulation results for official confirmed death, recovered, and cumulative cases of infection against data. root mean squared percentage error values of reported recovered, death, and cumulative infected are %, %, and %, and root mean squared error values are , , and . simulation runs almost perfectly correlate with data. table a in the appendix a reports estimated parameters along with those adopted from the literature. figure b shows how simulation runs replicate unofficial statistics including the number of infected travelers diagnosed before air travel patterns changed substantially, and medical community's counts of cumulative death reported by media sources. we use the calibrated simulation model to estimate cumulative number of infected (panel c in figure ), cumulative number of death (panel d), and current number of infected (panel e) until march , , starting from december st, (day in simulation). we find that total infected cases may have been closer to , , than the reported , on march th. cumulative deaths, over , , may also be an order of magnitude higher than official statistics and might have almost tripled in the last days of the analysis. there is also much uncertainty in these numbers: our baseline estimate for cumulative infected may be as low as , and as high as . million ( % uncertainty interval). similarly cumulative deaths is between and , ( % confidence interval). the wide uncertainty intervals are partly due to the nature of the data: absent data on testing coverage in iran it is very hard to fully know the underlying diffusion patterns based only on the formal statistics. the travel and informal estimates of death partially address these limitations, but much uncertainty remains. our conservative assumption on downscaling the likelihood function may also have contributed to this wide interval. nevertheless, even our lower bound are times the official statistics for total infections and times larger than the death statistics. there are some glimmers of hope in our estimates as well. the drop in contact rate in response to perceived risk, which we have captured in our model, might have reduced r e significantly, to just below as of late march (panel f). that reduction (which was confirmed by later data), if sustained, would help bring down the number of new cases. thus, figure e shows the number of currently infected cases may well be reaching its peak at the end of the simulation. the epidemic seems to have raged with a reproduction number close to . ( % ui: . - . ui) for more than six weeks before the behavioral and policy interventions have slowed down the diffusion rate starting in late february. some of this drop may also be attributed to weather, but current research is mixed on the impact of weather wang et al., ; xu et al. , ) so we did not include it in our estimation (but would explore it in the forward projection in the next sections). those factors, we estimate, brought down the aggregate contact rate to % ( % ui: %- %) of pre-epidemic levels by the date of analysis. this number may just succeed in bringing down the reproduction number below , which is the necessary condition for containment of the epidemic. whereas the case fatality rate in iran, based on official statistics until march th, is . %, our estimated infection fatality rate is . ( . )% (standard errors for estimates are in parentheses). this estimate is consistent with, and somewhat higher than, the . % reported by wu and mcgoogan ( ) based on tracing confirmed cases in china. if correct, the toll of epidemic in iran would have been larger than that in most other countries by the estimation date, and limits to healthcare services may explain the increased death rate. this finding is reinforced by the limited coverage of testing in iran that is indicated in our estimation. many cases go un-confirmed, and at its maximum, formal testing is covering only . ( . )% of infections. these estimates are consistent with qualitative media reports from iran on very limited availability of testing, the doctors' need for getting authorization before conducting tests, and the multi-day delays in receiving test results that render them ineffective in the clinical decision making processes. we estimate that hospital sources of news organizations who have offered alternative versions of actual death statistics in iran have had a much wider coverage of true death statistics ( ( )% of estimated true values). their count offers a lower bound for true deaths, is also close to our lower confidence bound, and suggests that the true number of cumulative cases, even with an infection fatality rate as high as . %, would not be lower than , . nevertheless, we estimate that even those medical community reports offer coverage significantly below %, consistent with the lessthan-perfect coverage of hospital reports by media sources, and anecdotal evidence that hospital system has been overwhelmed in many hot spots of the disease and many patients have died at home and been buried with no testing or proper association of cause of death to covid- . the model used prior estimates from the literature for three important parameters: the total duration of illness (d = days), the exposure (early infection) period (τ = days), and the fractional infectivity of the exposure period (θ = . ). noting significant uncertainty in such estimates early in an epidemic, we assess our projections' sensitivity to these structural assumptions by re-estimating the model for different values of each parameter. in this analysis we find a new set of best-fitting parameters and estimates for true cases given each different assumption on disease parameters. in table results are summarized as percentage changes in key estimates (cumulative infections and cumulative death until march th, ) compared to baseline best-fitting projections. results suggest these structural parameters are important in identifying the true magnitude of the epidemic and may propagate additional uncertainty in our projections due to limited early data on their exact values. specifically, a smaller fractional infectivity during exposure period (θ), a longer exposure period (τ), or a shorter duration of illness all significantly increase the projections for total cases and deaths. reductions in the estimates are less significant when parameters move in the opposite direction, though a very high level of θ would bring down the estimated cumulative cases notably. in lower values of θ, model calibration suggests earlier arrival of first cases of infection, which, in a few days, can make a difference in the number of late-infected, and increase exposure rate. current evidence for transmission during early infection (exposure) period is rather limited, and thus our baseline projections for total cases may be somewhat conservative. nevertheless, given our conservative treatment of the uncertainty intervals all the best-fitting projections with different disease parameters remain within the % bounds of our baseline projections. we run the model until july st of under six scenarios (three alternative assumptions about the impact of seasonality times two policy measures). specifically, our scenarios about seasonality include no effect (status quo), moderate effect (infectivity of the virus decreases linearly from may st and halves by june st, then stays the same for the rest of the simulation), and very strong mitigating effect (infectivity of the virus decreases from may st to a quarter of its base value by june st, then stays the same for the rest of the simulation). for policy, our focus is on contact rate, and we include two conditions of status quo (i.e. only behavioral responses) and aggressive efforts to decrease contact rate by half what it would be otherwise. these six conditions provide intuitions for a potentially wide range of cases. both intervention and weather impacts are assumed to include rather strong options to inform the range of possibilities, and should not be interpreted as predictions. moreover, we also note that the wide cone of uncertainty in the baseline simulations will continue and expand in future projections. keeping the uncertainty considerations in mind (but not graphed due to clutter), figure shows results based on best-fit parameters. if our best case scenario on the reduction of contacts and benefits of seasonality for containment materialize, the number of infected cases would be expected to peak soon after the analysis date at approximately , ( % ui: k- k), and will go down later on. this optimistic scenario will still lead to over a million infections and some , deaths ( % ui: k- k) by the end of june. the less aggressive scenarios point to continued spread of the epidemic at large rates. a reduction in contagion is realized only when reproduction number remains below one as a result of reduced physical contact, government interventions, and seasonality. among these three, the first (behavioral response) is endogenous to recent death rate. when death rate comes under control (due to the combination of all three factors), our model assumes the contract rate rebounds, weakening the first channel and increasing the transmission. economic pressures, normalization of death, and behavioral modeling after those not practicing distancing (or those already recovered and thus presumably immune) could all weaken the behavioral response when the perceived risks fade. the increased transmission would then, with a delay, bring up the death rate close to levels sustaining a reproduction number around one. this creates a strong attractor in the dynamics where in steady state contact rate is high enough to sustain the contagion but not so high to lead to the rapid infection of all the population. it may also lead to oscillations in peak transmission. therefore it is only with strong weather and/or government interventions that reduced reproduction number brings the epidemic under control. this dynamic offers a cautionary tale against declaring victory early in the fight against the epidemic. however the actual magnitude of such rebound effect is not known in this setting and our data offers no guidance on the relevant parameters. therefore these results are only qualitatively suggestive but not quantitatively reliable. this research was conducted under extreme urgency to inform decisions amidst crisis. the primary objective of the analysis was to understand the magnitude of the outbreak during early stages of infection when uncertainties are large and major decisions at stake. the main analysis of this paper was completed on march th and first reported publicly on march th on medrxiv pre-print server. thus we intentionally leave those projections untouched to avoid post-hoc justification and corrections that would cover the true uncertainties involved in real world modeling projects targeting fast evolving situations. furthermore, given that march th marks the end of the winter and beginning of persian new year, emerging annual reports can be used to assess some of our findings. in this section we summarize a few such tests and observations made based on findings post initial analysis. first, our estimates can be partially tested against official all-cause mortality data which are reported seasonally by the iranian government. the data for winter were released on may th, during the final preparations of this manuscript. based on a regression analysis on the all-cause mortality data we find that covid- related deaths are likely . ( % ci: . , . ) times more than official death counts for the disease (see appendix c). our simulation-based estimates, suggesting . ( % ci: . , . ) times more deaths than official data, are consistent with this new information. a second study informed the true magnitude of epidemic in one of the provinces of iran using anti-body testing. specifically, testing for covid- antibodies informs whether a person has been previously infected by the disease, offering estimates for cumulative infection when randomly conducted on a population. we have found one such study conducted on a random population of iran's guilan province (population of . million) (shakiba et al., ) . these researchers estimate that - % of guilan residents were infected by late april. we did not simulate our model at the province level, but our march th estimate of times undercounting, combined with confirmed cases in guilan ( . % of the residents), gives an estimate of % prevalence at the end of winter. our model estimates about %- % growth in cases over the following month (see figure b ), so our estimate of cumulative infections in guilan, at %- % for late april, are still five to ten times lower than this new study. these findings would suggest that a rather high fraction of cases were asymptomatic and/or not included in any of the data sources for infection that we used (official data and infected traveler counts). incidentally, if we assume that our estimates for infection were times below true numbers, the infection fatality rate for covid- in iran goes down to close to . %, a number much closer to some recent estimates for this key parameter (basu, ) . third, we can test model's capability in projecting confirmed cases. although prediction of official cases was not the purpose of the model, we present this out-of-sample test as an indirect way of assessing the model's predictive power. the model was calibrated with days of data, and since the pre-print, days of more data on confirmed cases and deaths have become available, which we use to compare the model's predictive performance for official, reported cases. these results are provided in appendix b, and show that while qualitative trends are consistent, the model misses a major increase in the number of cases right after the first days. confirmed death projections are closer to data. two mechanisms may explain this gap. first, our simple formulation for testing only tracks the fraction of cases tested, saturating that fraction at a maximum. that formulation is fine in capturing growth in test capacity during the rise of epidemic, but would underestimate cases when the growth slows down and milder cases can be found with the already enhanced test capacity. indeed since estimation we have found that initial test capacity in iran grew to about tests per day, stayed at that level for the later part of our estimates, and then jumped to , tests per day; similar increasing patterns of tests are seen across nations globally. thus, the testing formulation likely introduces under-counting in simulating official cases later in the epidemic, but this does not mean the model is under-predicting the actual infections and deaths. a second factor may explain why our model predicted an earlier peak, by about two weeks. iran's new year holiday season, nowruz, which starts on march th, is a major time of travel, family visits, and social interaction. those activities may well have increased contact rates in later march, but we decided against post hoc introduction of a nowruz parameter in the model. finally, we note that our model predicted a resurgence of the epidemic in early may due to reduced risk perception, which is consistent with a rise in cases in the real world in may. it is harder to fully assess the impact of these discrepancies on future estimation of true cases, had we decided to extend those estimates beyond the original timeline. nevertheless, it seems our model is somewhat conservative in predicting the 'reported' infections and deaths after the estimation period, and a more explicit account of physical test capacity would add value in longer term projections. finally, upon the release of the first draft of the paper we have received reports from a few hospitals and officials in iran that confirm: a) unofficial death counts by hospital sources that continue to be about five times the official death statistics, and in some states of iran are close to our estimates; and b) the low sensitivity (false negative risk) of tests, at about %, when conducted by hospital staff. moreover, availability of tests even in major hospitals is limited, covering about % of those clinically diagnosed using other methods (often ct scans). it is important to note that some evidence from random testing in other countries (including iceland and germany) point to a very high asymptomatic fraction of patients, which would not be captured in any of our data sources and is consistent with our comparisons with the guilan study. furthermore, new evidence suggests we might have used under-estimates for the delay between exposure to onset of symptoms (about days), from onset to testing (another days), and from testing to resolution (about days) (lauer et al., ; linton et al., ) . adjusting for these parametric assumptions would increase our estimated case count as discussed in the sensitivity analysis and may bring our overall estimates closer to the new evidence on the cumulative infections. in this paper we provide a more sobering picture of the covid- outbreak in iran using a dynamic model that goes beyond official statistics. integrating data from various sources we suggest that the official statistics are at least an order of magnitude below actual spread of the epidemic, and even in optimistic scenarios the burden of disease will be large and lasting for many months. implementing and sustaining strong policies that target physical distancing offers the main hope for containing the epidemic until a vaccine becomes available. the gap between official statistics and our estimates may be due to various complications in measurement as well as policy choices. the availability of testing infrastructure has been a major bottleneck for detecting cases in iran. citizens with suspicious symptoms have had no easy way for getting tested; most tests are conducted on hospitalized patients, require specific authorization, and are processed in a few centralized labs adding significant delays to the process. not only mild cases are missed, but also many critical patients have been unable to access the care they need due to hospital congestion. the fact that majority of cases have mild symptoms similar to flu (wu and mcgoogan, ) adds to the risk of under-counting even if the testing capacity was ample. in fact many mild or moderately sick patients may have preferred to stay home than risk being infected upon visiting a congested hospital, further reducing demand for testing. finally, false negative test results seem to be common among covid- patients in iran, with healthcare providers indicating sensitivity levels as low as % due to challenges in sample extraction and varying concentration of viral load over the course of disease. the problem of undercounting cases of infection and death is not limited to iran. due to the relatively long period of incubation, a considerable proportion of asymptomatic cases, significant fraction of false negatives in existing tests, and especially the limits on test capacity, the undercounting problem is common, but its magnitude would vary across different countries. even with ample test capacity, the undercounting of infection is likely large due to both the asymptomatic patients (estimated to be more than % (mizumoto et al., ; nishiura et al., ) ) and those cases with false negative results (about % of positives in case of pcr based covid- tests (fang et al., ) ). for example a pre-print paper estimates that in the us less than % of cases of infection were diagnosed until march (pei and shaman, ). others have made similar claims about official data of china . the undercounting of deaths may be less in magnitude, but still very common. under significant strain, limited testing capacity may not be allocated to post-mortem testing of suspected cases, and thus country-specific routines for categorizing death in those cases may lead to large variations. we suspect this is a major reason why iran's official death counts may be significantly below the actual numbers. but other countries may have similar issues. for example a recent article estimates much higher number of deaths in italy than official numbers (stancati and sylvers, ) . moreover, recent investigative reporting by financial times, new york times, wall street journal, and economist has used all-cause death data of different countries and cities and estimated that globally, as much as % of covid- death cases are missed (burn-murdoch et al., ) . for example italy's excess death in march is , , out of which only , is attributed to covid. jakarta (indonesia) had excess deaths in march and april, with only attributed to covid. absent extended testing capacity, expanded post-mortem testing, and random testing in the population, policymakers may face a significant undercount of existing cases and thus miscalculate their response. we hope that our proposed approach can be used for similar cases elsewhere to de-bias such estimates. methodologically, we faced important challenges to estimation and projection of s-shaped behavior based on early data that are worth highlighting here. first, estimating s-shaped curves before observing their inflection point can be unreliable. minor changes in the slope of the cumulative curve can significantly change the estimated future trajectory when simple curve-fitting is pursued. overcoming this risk requires a more mechanistic model that is fitted to the flow (infections in this case) rather than cumulative stock, so that the slope is constrained by physical mechanisms determining the flow variable (e.g. the size of the susceptible population). the calibration should also properly weigh different data points so that the larger numbers during peak period do not overwhelm the signal from other periods. in our case we achieved this using poisson log-likelihood error. other promising alternatives may include negative binomial log-likelihood, among others. a second challenge is teasing out the various potential reasons for reduction in new infections. for example those may include a reduction in susceptible population, a change in contact rate due to social distancing, or exogenous changes such as weather. while mechanistically proper model specification can partially address this challenge, one needs to be cognizant of the uncertainty in estimated parameters, e.g. by using mcmc methods to identify the various combinations of those effects that are broadly consistent with the data. the problem is rapidly changing, and timely estimation is critical. therefore we have made several simplifying assumptions that readers should be aware of in interpreting our results. first, consistent with prior findings (rahmandad and sterman, ) we have focused on capturing behavioral feedbacks in contacts and testing and have adopted a formal estimation process, but have done so at the cost of abstracting away from much detail complexity and heterogeneity in populations and risks. we converged on the current model structure in an iterative process, attempting to keep the model simple and feasible to estimate, but not missing components that would qualitatively change the results. for example we tested the inclusion of healthcare capacity, test accuracy, and physical test capacity (rather than as a percentage of cases) and decided against including them explicitly because they did not change the conclusions qualitatively and our data did not offer reliable ways to estimate those effects beyond what is informed by our current structures. generally, a more detailed model may be appealing, but if the purpose is to offer quantitative estimates and projections, constraining additional model parameters during calibration would be a major constraint that should be weighed against the value of additional structure. second, during the early periods of infection, there is considerable uncertainty in most data sources. for example, infected passenger data, while useful, is a not a perfect data source for modeling. it can be argued that exposed to more travel, this population may be exposed earlier than a random sample. on the other hand, this is typically a higher income and more educated population that may be more aware of, and responsive to, infection risks. similar issues can be raised for informal death reports in the media, which are based on potentially limited samples of hospitals and may include subjective assessments of medical cases. such uncertainties are part of any emerging, fastmoving situation, and one should be cognizant of these factors in interpreting modeling results. third the model is scoped only around iran --we ignore potential effects of global spread of the disease on iran, including the risk of reintroduction of cases in future. forth, we assumed that the recovered population are immune during our simulation horizon. fifth, we ignore mutations in the virus which may change its contagiousness and case fatality rate. despite these limits we hope the paper offers an accurate assessment of the risks and scope of the epidemic for iran and beyond. we also hope other researchers build on this work using the publicly available data, models, and replication instructions provided in the online supplement. finally, we hope that the study contributes to the system dynamics modeling literature. there are a wide range of epidemic models with different levels of dynamic and detailed complexity (darabi and hosseinichimeh, ) and a large number of models have been built in response to covid- . while epidemic models with interesting and relevant feedback mechanisms could be built rather quickly, their results depend heavily on the parameters regulating the strength of contagion, behavioral response, testing, and other mechanisms. for example reputable research groups have offered models with forecasted cumulative deaths that vary significantly (ferguson et al., ; murray and ihme, ) . without quantitative estimation methods, different models may point to starkly different conclusions, with little guidance on how to select the appropriate analysis. our contribution is to offer a model-based estimation approach during an on-going pandemic where data are uncertain and limited. this requires combining all sources of available data with different levels of uncertainty, each one, potentially an imperfect indicator of the reality, yet collectively providing a better picture. our data-driven approach is consistent with recent calls for detailed data-informed modeling approaches (sterman, ) , and combining different methods and techniques to enrich dynamic modeling (duintjer tebbens and thompson, ; ghaffarzadegan and larson, ) . we hope the methods used in this paper can be extended to a variety of other system dynamics applications where actionable insights rely on rigorous combination of feedback rich models and quantitative data. data sources to study dynamic problems in organizational and health domains. as the number of infected cases increase, the probability of infection increases given the increased number of contacts between infected and susceptible populations. some of these contacts lead to new cases of infection (loops r and r ). infections may be due to a fraction of early infected cases who can infect others (bai et al., ) , and the late infected population. exposure rate (i e ) is formulated as following: we set e at (no seasonal effect) for historical calibration, and later for scenario analysis, check different functions that give lower values during summer. the equation has two degrees of freedom, and since the actual value of c and i are not independently knowable, we assume i = . and vary c to fit the model output against data (details below). death and recovery rates are estimated using fractional death rate. we have, where the disease duration is assumed to be total of τ e (early-stage asymp- contact rate can change in response to the progression of the disease and consequently change infection rate (see figure a ). the impact of perceived risk on contact rate captures not only the endogenous changes in social interactions, gatherings, self-isolation of suspected cases, and hygiene, but also government mandated closure of events, schools, and businesses to name a few. we assume public risk perception depends on reported cases of death. finding infected cases is challenging given that many cases remain mild and unknown, yet potentially infectious. furthermore, especially early on, many cases were misdiagnosed as healthcare providers were unaware of the existence of the virus in the country. as the first cases are revealed and reported, a higher fraction of cases are found (loop r ). there are three death-rate-related variables: actual death rate ( c is inversely related to o d . we bound c between and c max , and use the following sigmoid function to represent c: ρ represents the impact of other policy measures beyond the perception of death that may impact c (e.g. various government interventions) and is set to for base run simulations but experimented with for policy analysis. o à d and s represent the relation between o d and c. conceptually o à d is a threshold for public report of death, at which contact rate endogenously declines to half of c max and s is the sensitivity of public behavior to the reported death rate. the free parameters, o à d ,s, and c max , are estimated. we assume only symptomatic cases of patients can be diagnosed. let f symp < be the fraction of late-infected individuals that develop symptoms. let α be the fraction of symptomatic who are tested, then αf symp is the fraction of i l who will be tested. more severe cases are more likely to be diagnosed, so mortality rate of diagnosed cases(f ), is likely to be higher than average fatality rate (f). death rate of diagnosed individuals is f (αf symp i l )/τ l , with the constraint αf symp f ≤ f. we assume no post-mortem diagnosis. including first cases of death discovery diagnosed early on (x): where d is the cumulative reported cases of death. x is equal to on t =february th and for the rest of the simulation it is zero. we set α = min ad, À Á Á α max , and let the model estimate α max , and a through calibration (explained below). for reported cases of recovered (r), we use the following equations: is reported cases of recovery per day. this formulation assumes the reporting of recovered cases happens with the same rate (α) as the reporting of infections and death. while iran's data are supportive of this assumption, exploration of statistics from other countries suggests many do not track the recovered cases. for total infected (i), cumulative cases (i cum ) and reported cumulative cases of infection (i cum ) we have: we set f symp = . , but the model is not sensitive to the value of f symp , as any change in f symp will be compensated by α to match the simulation with simulation-based estimation of the early spread of covid- in iran the data, and calculate total fraction of infected who are diagnosed. the reproduction number is estimated asr = ( τ e + τ l )(i e /i). besides the official data, we use a few unofficial data points: four observations about number of iranian passengers diagnosed with covid- upon arrival in international airports (p ), and three unofficial estimations from bbc and iran international news sources about cumulative death from covid- (d u ) early in the epidemic. in our model, we estimate them as following: where d u, = , and μ is the ratio of number of passengers going from iran abroad in early days of the outbreak. the term (t + − t ) represents days since the first cases of infection were diagnosed. it is assumed that the daily number of passengers is fixed and death is minimum, thus this estimation only works during the first few days of the outbreak. for d u , we assume γ represents the fraction of actual deaths identified by these sources. note that the unofficial reports for death cases in media are based on unofficial reports from the medical community with limited samples, thus γ is likely below one. γ will be estimated through model calibration. we initialize the model as i e, = i l, = r = d = , and s = million people. we set r = d = . new infected cases are injected to the model through i f , at time t , to be estimated through model calibration. although we are not aware of number of foreign patients entering iran, since t is also unknown, we have two degrees of freedom, and one can be assumed and the other estimated through model calibration. we set i f equal to patients per day at t , and throughout the rest of the simulation. we run the model from t = , december , , and our time unit is a day. t , the time first cases are diagnosed, is february , . we have time series data for d, r, i cum , and a few data points for p and d u . our method is mainly based on forming a likelihood function for observing the actual time series data conditional on model parameters. we then conduct a markov chain monte carlo (mcmc) simulation to estimate the joint posterior distribution of the model parameters subject to observed data. we define a likelihood function for change over time (net-flow) of d , r , i cum assuming they are count events drawn from model-predicted rates (poisson distribution). we use a similar poisson distribution assumption for p and d u as well, since they both fit well into a count measure framework. the mcmc method searches over the feasible ranges for nine uncertain parameters in our model. these include: three data-related parameters of α max , a, and γ; three public-behavior-related parameters of c max , o à d , and s; and three disease-related parameters of fractional death rate (f and f ) and time for the injection of first cases of infection into the population (t ). prior experience and this exercise point to rather tight confidence intervals coming from mcmc methods directly applied to large nonlinear models. the problem could be addressed using filtering (e.g. extended kalman filtering or particle filter) methods or heuristics that scale the likelihood function in mcmc method to be consistent with model's predictive performance. in light of the computational costs of the former and time sensitivity of the research topic, we chose the latter option. to be conservative, we therefore downgrade our confidence in the data, scaling the likelihood function by a factor of . , which allows the algorithm to explore the parameter space much more freely. this approach is a heuristic we use to avoid putting too much confidence in our estimation results; however, without strong theoretical reasons to support the specific scaling factor, we caution against a strong interpretation of these results. with this caveat in mind, the mcmc method offers % confidence intervals for joint-distribution of the parameters. we conducted additional sensitivity analysis to this downscaling parameter choice. the rest of the parameters are specified based on the literature. the counter factual of "what is expected absent covid- " is. a simple counter-factual would compare deaths in previous winters with that in winter . yet, depending on which previous years one includes in such counter-factual, excess deaths between to thousands could be estimated. however, the released death data is disaggregate at the province level ( provinces). this provides a unique opportunity to more accurately estimate the excess deaths in iran by comparing the variations in death in different provinces and how those variations could be explained by the official death toll in each providence. formally, we run the following regression to estimate the undercounting of covid- related deaths: excessdeath i = β + β :pastdeath i + β :confcoviddeath i: where excessdeath i is the difference between death in winter and winter in province i, and pastdeath i is a vector of death in past seasons (e.g., winter , , etc.) in province i. confcoviddeath i is an estimation for confirmed death cases based on confirmed total cases for each province which were available for the analysis (iran does not provide province level death statistics for covid, but did provide province level cumulative infection statistics; scaling total death based on province level infections we calculate "confcoviddeath"). β will represent death undercounting factor. our estimate of , ( % ui: . k, . k) death implies that this ratio should be around . ( % ci: . , . ). our regression models are reported in table a . based on the best regression model (in terms of adjusted r-squared), we find that β = . ( % ci: . , . ). this ***p < . , **p < . , *p < . . presumed asymptomatic carrier transmission of covid- estimating the infection fatality rate among symptomatic covid- cases in the united states: study estimates the covid- infection fatality rate at the us county level corona virus has at least claimed lives in iran who, among political and government figures, have contracted the corona virus? global coronavirus death toll could be % higher than reported. financial times system dynamics modeling in health and medicine: a systematic literature review using integrated modeling to support the global eradication of vaccine-preventable diseases sensitivity of chest ct for covid- : comparison to rt-pcr covid- -navigating the uncharted impact of non-pharmaceutical interventions (npis) to reduce covid- mortality and healthcare demand pandemic potential of a strain of influenza a (h n ): early findings sd meets or: a new synergy to address policy problems the number of deaths from corona virus in iran reach db% c%d %b %d %a %d % -%d %a %db% c%d % %d %aa% d %b %d % %d %b %d % %d %a %d % -%d %b %d % % d %a %d %b -%d %ac%d %a %d % %e % % c%d %a %d % a %d %ae%d %aa%da%af%d %a %d % -%da%a %d %b %d % %d % %d %a -%d %af%d %b -%d %a %db% c%d %b %d % a %d % -%d %a %d % -%db%b %db%b %db%b -%d % %d % %d %b -%d %b %d %b %db% c%d %af emergency response to a smallpox attack: the case for mass vaccination the incubation period of coronavirus disease (covid- ) from publicly reported confirmed cases: estimation and application substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (sars-cov- ) incubation period and other epidemiological characteristics of novel coronavirus infections with right truncation: a statistical analysis of publicly available case data the role of absolute humidity on transmission rates of the covid- outbreak estimating the asymptomatic proportion of coronavirus disease (covid- ) cases on board the diamond princess cruise ship forecasting covid- impact on hospital bed-days, icudays, ventilator-days and deaths by us state in the next months novel coronavirus pneumonia emergency response epidemiology team. . the epidemiological characteristics of an outbreak of novel coronavirus diseases (covid- ) in china initial simulation of sars-cov spread and intervention effects in the continental us who chief says cases of coronavirus in countries originated from iran heterogeneity and network structure in the dynamics of diffusion: comparing agent-based and differential equation models seroprevalence of covid- virus infection in guilan province italy's coronavirus death toll is far higher than reported system dynamics at sixty: the path forward the risks, costs, and benefits of possible future global policies for managing polioviruses estimation of covid- burden and potential for international dissemination of infection from iran temperature significant change covid- transmission in cities wikipedia. . coronavirus outbreak in iran iran has far more coronavirus cases than it is letting on. the atlantics nowcasting and forecasting the potential domestic and international spread of the -ncov outbreak originating in wuhan, china: a modelling study characteristics of and important lessons from the coronavirus disease (covid- ) outbreak in china: summary of a report of cases from the chinese center for disease control and prevention weather conditions and covid- transmission: estimates and projections. medrxiv we are thankful to mohammad akbarpour, narges doratoltaj, babak heydari, hamed ghoddusi, and tse yang lim for their thoughtful feedback on various drafts of this paper. navid ghaffarzadegan is an associate professor in the department of industrial and systems engineering at virginia tech. he develops system dynamics simulation models to study complex social systems and policy problems. the main application areas of his research include science policy and health policy.hazhir rahmandad is an associate professor of system dynamics at mit sloan school of management. he combines simulation models and various simulation-based estimation of the early spread of covid- in iran basic model architecturethe model uses a country-level system of differential equations to capture the dynamics of contagion ( figure a ). the infected population is first asymptomatic and is represented as "early infected" and later becomes "late infected" with the majority being symptomatic.the following equations represent the aging chain:appendix b: out of sample prediction testour main analysis in this paper is done based on data up to march th, (total of days). since then days of new data have become available which we use to conduct an out-of-sample test of our model's projection for confirmed cases. figure a shows the results. the model is better at replicating reported cases of death (panels c and d) than reported infection (panels a and b) and recovery (panels e and f). it appears that by increased testing, milder cases of infection are diagnosed in iran than we predicted, while confirmed death rate is still in the same range as our model's prediction. in recovery data we see a peak on a specific day (panel f) that might relate to some follow-up or closing active cases. the data for all-cause death, which in iran is reported seasonally, were released on may th, , during the final preparation of this manuscript. winter ends on march th, thus seasonal reports on all-cause number is lower than our estimate for death, but of comparable magnitude, and falls within our confidence intervals. it is worth noting that the covid-related excess death may not be all attributed to covid- alone. for example social distancing policies in response to covid may reduce traffic-accidents and death due to air pollution (both major causes of mortality in iran), leading to an under-estimate of covid- deaths in this approach. on the other hand hospital congestion may increase death due to other conditions, leading to over-estimation of covid- deaths using this method. we do not have data to correct for these potential biases, but overall we suspect that they are a second order effect and the true magnitude of covid- death is captured by the excess mortality data. additional supporting information may be found in the online version of this article at the publisher's website.appendix s : supporting information appendix s : supporting information simulation-based estimation of the early spread of covid- in iran key: cord- -o qmp x authors: bayraktar, e.; cohen, a.; nellis, a. title: a macroeconomic sir model for covid- date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: o qmp x the current covid- pandemic and subsequent lockdowns have highlighted the close and delicate relationship between a country's public health and economic health. macroeconomic models which use preexisting epidemic models to calculate the impacts of a disease outbreak are therefore extremely useful for policymakers seeking to evaluate the best course of action in such a crisis. we develop an sir model of the covid- pandemic which explicitly considers herd immunity, behavior-dependent transmission rates, remote workers, and indirect externalities of lockdown. this model is presented as an exit time control problem where the lockdown ends when the population achieves herd immunity, either naturally or via a vaccine. a social planner prescribes separate levels of lockdown for two separate sections of the adult population - those who are low-risk (ages - ) and those who are high-risk (ages and over). these levels are determined via optimization of an objective function which assigns a macroeconomic cost to the level of lockdown and the number of deaths. we find that, by ending lockdowns once herd immunity is reached, high-risk individuals are able to leave lockdown significantly before the arrival of a vaccine without causing large increases in mortality. additionally, if we incorporate a behavior-dependent transmission rate which represents increased personal caution in response to increased infection levels, both output loss and total mortality are lowered. lockdown efficacy is further increased when there is less interaction between low- and high-risk individuals, and increased remote work decreases output losses. overall, our model predicts that a lockdown which ends at the arrival of herd immunity, combined with individual actions to slow virus transmission, can reduce total mortality to one-third of the no-lockdown level, while allowing high-risk individuals to leave lockdown well before vaccine arrival. the current covid- global pandemic has led to massive lockdowns to slow the spread of the virus. now, policymakers face a dilemma -extended periods of lockdown have put strain on the economy, but returning to "normal" too quickly could result in an equally troubling second wave of infections. the task is therefore to find the optimal balance between public health and economic growth. models such as those proposed by alvarez et al. [aal ] and acemoglu et al. [ace+ ] have used a macroeconomic approach and variations on the susceptible-infectious-recovered (sir) epidemic model proposed by kermack et al. [kmw ] to solve an optimization problem determining the lockdown policy that minimizes both loss of life and effects on output. we consider a variation on these models which incorporates several new concepts and gives a wider picture of the overall situation. • we formulate an exit time control problem where lockdown measures are lifted when the population reaches herd immunity, even if this occurs before a vaccine is developed. • we incorporate a transmission rate that captures how individuals reacts to current infection levels, as discussed in [coc] . this "behavior-dependent" transmission rate seeks to model individual behaviors that occur independently of lockdown. for example, individuals might wear masks, practice social distancing, and take other precautions to reduce their risk as infection numbers go up, even in the absence of official lockdown measures. for the full . years until the vaccine). additionally, lockdowns for the low-risk group are weeks shorter. • the addition of a behavior-dependent virus transmission rate contributes to these shorter lockdowns and decreases mortality. in an extreme situation where individuals can take measures that decrease transmission by % when infections reach %, less than a month of lockdown is prescribed for the low-risk group. in the more moderate benchmark case, where individuals are able to reduce their transmission by % when infections reach % of the population, herd immunity arrives a month earlier than in a situation with a constant disease transmission rate. in both cases, we also observe lower output loss due to shorter lockdown and fewer deaths due to slower transmission. • increasing the level of remote work reduces the impact of covid- by decreasing both mortality and output loss, even though a longer lockdown is imposed. this supports the intuitive idea that increased remote work reduces infection risk without sacrificing economic activity. • increasing the predicted length of future unemployment and the predicted rate of lockdownrelated deaths both decrease lockdown length in a similar manner, but also have negative impacts on outcomes. adjusting the length of future unemployment and the predicted number of indirect deaths due to lockdown lead to trade-offs between output and mortality. running the model with different initial conditions shows that higher pre-lockdown infection levels lead to earlier onset of herd immunity but higher death tolls, highlighting the risks of infection spikes. future impacts of current missed health screenings and a penalty for overfull icus are revealed to have little impact on the optimal lockdown policy, at least in our formulation. the main body of the paper presents our model and its numerical results. in section , we lay out the sir dynamics used to model the transmission of the virus and discuss certain model additions, especially the addition of deaths indirectly caused by lockdown and a behavior-depending transmission rate. in section , we introduce the exit time control problem that ends when the population reaches herd immunity and discuss the terms in the objective function. in section , we discuss our numerical model, which discretizes the problem and is solved through value iterations. we calibrate it with the results of [ace+ ] and [aal ] and compare these results to our augmented model using death rates on the same scale. then, we update death rates to match more recent data from [cdc] and adjust the non-pecuniary value of life. we present and discuss our results and perform some parameter robustness analysis. these experiments serve to illustrate the general mechanisms of the model and to present planners with an idea of our model's potential. if a planner wishes to use our model, parameter values can be changed in our code, found at , to accurately reflect a specific planner's current situation. as in [ace+ ], we consider policies which assign different lockdown strategies to population groups with different responses to infection and lockdown. influenced by their results, we divide adults into one group aged - , called"low-risk" and indexed by j = , and one group aged and over, called "high-risk" and indexed by j = . we will only consider adults older than , so the low-risk group makes up % of the population of interest, while the high-risk group makes up % [hm] . the second group can also include individuals of any age who are more likely to contract severe cases of covid- and experience complications due to immunodeficiency, respiratory weakness, or other preexisting conditions. these individuals are considered separately from the general working population. we denote the population of group j as a proportion, n j , of https://github.com/april-nellis/covid -bsir all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint the total population and lay out the following relationship: individuals move from susceptible (s j (t)) to infected (i j (t)) to recovered (r j (t)), with additional flows from all three to death (d j (t)). the dynamics are given bẏ ( . ) the number of new infections depends on the size of the susceptible and infected populations, as well as the lockdown levels, l j (t), and transmission rates between groups, β ij . as others have suggested ([ace+ ], [aal ] ), a portion of the population will disregard lockdown orders. this level of obedience is represented by θ. as in [ace+ ] , we set θ = . , though this parameter is difficult to quantify exactly. patients move out of the infected category with rate γ in accordance with the expected recovery time of days. deaths due to covid- occur at rate φ(i j (t)) and other deaths occur at rate ξ(l j (t)). for convenience, the parameters that appear in ( . ) and in the objective function ( . ), along with their levels, are listed in table of the appendix. . . deaths. one unique element of covid- is its significantly different death rates for different groups. therefore, the base death rate δ j for each group is set individually. in addition, as the number of infected individuals increases and hospitals become more crowded, death rates increase as a function of the total number of infected patients as in [aal ] . we represent this increase in the death rate as δ j and follow [ace+ ] in assuming that an infection level of % increases the death rate by a factor of five. therefore, the death rate due to viral infection is additionally, as lockdowns stretch on concerns have been raised regarding "deaths of despair" due to the impacts of lockdowns on mental health ([zha+ ], [ems ] ). hospitals have also shut down many departments to accommodate the increased need for icu units for covid- patients. many non-elective surgeries and routine health checks have also been cancelled or rescheduled ( [sor+ ] ). this neglect of health maintenance is also very likely to have repercussions on public health. to encompass all this, we add the term ξ(l j (t)), which is written as this represents the number of deaths indirectly caused by the lockdown, and scales with l j (t). we argue that, in the absence of any way to verify immunity, indirect deaths occur in both the susceptible and recovered populations. so, the total number of deaths is given by j φ(i j (t))i j (t) + ξ(l j (t))(s j (t) + r j (t)). behavior-dependent disease transmission. the basic transmission rate of covid- is approximately . ([aal ], [ace+ ] ). this means that about % of those who come in contact with an infected individual will become infected themselves. however, we incorporate a transmission rate that decreases as infections increase due to increased caution between people, as discussed in [coc] . in addition, we consider an inter-group interaction factor, ρ, as in [ace+ ] . all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . this reflects a lower rate of interactions between groups. for example, working people aged - will interact more with their peers than with those in the high-risk group. combining these two ideas, we represent the transmission rate as the scale factor α i is chosen such that the rate of transmission can be decreased by a factor of e − . α i when % of the population is infected with the virus. the questions of how long and how severely to lock down the population during a pandemic can be thought of as a planning problem. given the above dynamics, we model the optimization problem that must be solved by a social planner using the following objective function, which represents the overall societal costs of a given lockdown policy: . . attainable lockdown levels. there are certain jobs that must be done even during a pandemic, preventing the population from attaining full lockdown. these essential professions include healthcare workers, grocery store employees, delivery workers, and the postal service, among others. because of this, we set an upper limit on the possible lockdown level, denotedl j , and the set of possible lockdown policies is written as Λ = [ ,l ] × [ ,l ]. we setl at . to account for essential workers in the low-risk group. on the other hand,l is set at since we assume that the high-risk group does not work. . . herd immunity. previous models considered either an infinite time horizon with stochastic vaccine arrival or a fixed horizon with deterministic vaccine arrival when formulating their objective function. we contribute a new approach that sets reaching herd immunity as the end of the problem. this can be reached either naturally via infection spread and recovery (which confers immunity) or via the arrival of a vaccine. we assume that those who have been infected with covid- once will remain immune for the rest of the outbreak. additionally, we assume that if a vaccine is approved for distribution, vaccination levels will be high enough to produce herd immunity. we define σ as the time at which herd immunity is reached. we assume a herd immunity threshold of % recovered , so we set . . output loss. the most noticeable result of lockdown measures is economic slowdown. many workers who are not deemed essential and cannot work remotely have found themselves jobless as companies lose revenue. as in [ace+ ] , we take the average wage of a full-time worker and normalize it to and assume that on average, those in the high risk group do not earn any wages. we do not assume the existence of an "immunity passport" given to those who have recovered and are immune, so we set a flag parameter p = (this can be set to to be consistent with the cases in [cdc] table (scenario : current best estimates), the basic reproduction number of covid- is r = . . herd immunity is calculated as − /r = . . all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint of [aal ] and [ace+ ] ). on the other hand, we do consider some proportion h of the workforce who are able to work from home. therefore, we denote the purely salary-based cost of lockdown as when presenting our numerical results, we refer to the output loss due to lockdown. this is not the value of the objective function presented in ( . ), but rather the losses in output caused by requiring people to stay home and not work. the output loss is represented by and is compared to annual "normal" output. this baseline output is calculated as the amount of output produced until the expected vaccine arrival time, /nu, if there is no lockdown, annualized using the expected vaccine arrival time. this is given by . . cost of death. we calculate the cost of a covid- death in group j in the same manner as in [ace+ ] . here, χ is the non-pecuniary cost of life, which we consider as a measure of the public impact of deaths due to covid- . this can be thought of as a measure of the planner's priorities. lower values of χ lead to prioritizing output loss minimization, while higher values are chosen to encourage longer lockdowns and decrease mortality at the expense of output. to ensure that this cost is on the same order of magnitude as wages, we scale by the interest rate when we choosing χ, similarly to [aal ] . note that χ = . /r is consistent with [ace+ ] where χ = and r = . . ∆ j is the number of years left in an average individual's career. we set ∆ = and ∆ = . therefore, the cost per death due to covid- is given by deaths indirectly caused by the lockdown are not explicitly categorized and counted, and so can be considered "invisible deaths". for this reason, we do not include χ in the cost of these deaths and only count lost productivity. we also account for similar future deaths due to lack of preventative healthcare using a constant f (the number of indirect deaths in the future relative to those that occur during lockdown). these deaths do not appear in the dynamics, as they have not yet occurred, but they are considered when calculating the costs of lockdown. for this reason, f appears in ( . ) but not in ( . ), and the total cost of indirect deaths is given by . . future loss of employment. another addition to the model acknowledges the long-lasting economic impacts of a period of economic slowdown. already some large corporations like jcpenney and hertz have filed for bankruptcy ( [mon] ), and there are certainly more companies, both large corporations and small business, who are under financial strain. federal stimulus measures may alleviate some of this burden, but they cannot completely compensate for current drops in consumption. effects may manifest in a variety of ways, but we choose to express them as a "future loss in employment", in which day in lockdown results in some α e days of lost employment (on average) after lockdown ends. we set this to be . (reflects current . % unemployment [bls] and average days of unemployment for one day of lockdown based on median unemployment duration of . weeks ( months) in [cun] ). the cost of future unemployment is modeled by all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint . . icu overcapacity. a major incentive for lockdown measures is "flattening the curve"slowing the spread of the virus so that hospitals and icus will not get overwhelmed by a flood of patients in need of ventilators and other specialized medical equipment. this is already reflected in the death rate which increases as infections increase, but we add an additional penalty on top of that. we assume that a fixed proportion of infected patients, ι j , require icu care. we set this level to be . % for people without underlying conditions and . % for high-risk groups [cdc] . then, we incorporate a penalty η (representing a daily penalty scaled by the level of overcapacity) for hospitalizations exceeding the estimated average icu capacity, which is beds per , people [pw ] . this is done via the function . . numerical method. we used value iterations, first introduced in [bel ] , to solve the optimization problem presented by our model. the model is discretized using first-order taylor approximations and the value function is calculated over a regular grid. because the change in population due to deaths is very small, we follow the precedent set in [aal ] and iterate over a four-dimensional (s , s , i , i ) grid to determine optimal lockdown policy instead of the larger and more computationally expensive (but more accurate) six-dimensional grid (s , s , i , i , r , r ). by this, we mean that instead of separately keeping track of the recovered and dead populations, they are considered together as one unit when determining the optimal lockdown policy. since the vast majority of this "non-susceptible" group is recovered, this simplification, which removes two state variables, has a small effect on accuracy but a large effect on computational complexity. and, when determining the pandemic trajectory for given initial conditions and a given lockdown policy, total deaths can still be calculated via the population dynamics shown in ( . ). we choose ∆s = . and ∆i = . in our discretization, and set days as the unit of time. for all models, we take the initial conditions to be uniform across groups (if applicable) and set them at the level of % susceptible, % infected, % recovered unless otherwise specified. to test the validity of our numerical models, we use the parameter values of [aal ] and [ace+ ] and compare our model's recommendations to their results. a full list of the parameter values used in this section is presented in table . in figure , we compare a one-group version of our model with the one presented in [aal ] , whose recommended optimal lockdown reaches % lockdown after about one month, and then slows reduces in intensity until lockdown is lifted approximately days ( . months) after the outbreak begins. our version of this model maintains the maximum lockdown of % for slightly longer and ends slightly later. interestingly, the ending of lockdown nearly coincides with the population reaching herd immunity, though we did not add any such considerations when running this example. in figure , we set up our model to mimic the semi-targeted policy from [ace+ ] and find similar levels of output loss, total deaths, and general lockdown recommendations. namely, the optimal strategy keeps the high-risk group in lockdown until the arrival of a vaccine, while the low-risk group is able to emerge and return to work after approximately days of lockdown have elapsed. in this figure, note that the population reaches herd immunity well before the arrival of a vaccine, implying that the lockdown on the high-risk group could have been ended earlier. now, we investigate the results of our new model using comparable parameter levels. we incorporate herd immunity, deaths indirectly due to lockdown, ability to work remotely, and behaviordependent transmission rates. additionally, we consider the possibility of lost employment after the end of the pandemic, as well as the costs of missed health screenings and a monetary penalty for exceeding icu capacity. to allow comparisons with previous works, we use death rates of a all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint similar magnitude to those in [aal ] and [ace+ ] , but we change some parameters to better fit the current situation. interest rates have dropped significantly, so we use a . % interest rate, instead of the % used by [aal ] or the % used by [ace+ ] . note that since the interest rate is extremely low, there is little to no discounting applied to wages. we also lengthen the projected average career length in the low-risk group. finally, we adjust the population distribution slightly from % high-risk to % high-risk, based on data from the united states census [hm] . the results of our model using the parameters listed in table is shown in figure . most noticeably, lockdown rates for both groups fall to after the entire population reaches herd immunity, which is explicitly imposed by our model. additionally, the lockdown for the low-risk group is slightly shorter but more intense. unsurprisingly, incorporating deaths of despair increases the total number of deaths due to the epidemic, but this effect is kept small by the shorter lockdowns. long-term costs of lockdown (an additional penalty for icu overcrowding, future deaths due to current health negligence, and future unemployment beyond the lockdown) increase output loss while not directly contributing to deaths during lockdown. however, these output losses are offset by the proportion of the population that is able to work remotely from home and the shorter lockdown periods. rates. recent cdc reports [cdc] indicate that the death rates are much lower than those used in subsection . . to increase the realism of our model, we update the model death rates according to this newer data and use them for all subsequent results. these death rates are listed in table and the result, shown in figure , predicts total mortality of . % and total output loss of . %. the negligible output loss is due to the negligible lockdown for the low-risk group. however, lockdowns have already been imposed for both groups (and indeed we might desire a death rate lower than . %), so we increase the non-pecuniary value of life, χ, and observe how the model changes. by increasing χ from . /r to /r, figure shows that both groups experience levels of lockdown similar to that of figure , but with an output loss of . % and a lower total death toll of only . %. we designate this the benchmark situation, which uses death rates from table and χ = /r but keeps all other parameter values consistent with those in table . we also compare the results of the optimal lockdown to those generated by an uncontrolled scenario with the same parameters, shown in figure and table . without lockdown there is no output loss, but final mortality numbers are approximately twice as high. . . varying initial conditions. since we are currently in the middle of the pandemic, we investigate at how different initial conditions change the recommended lockdown levels. we model a situation where the pandemic is ongoing and lockdown measures have been lifted, but a sudden spike in infections occurs which prompts new lockdown measures. we consider a case where % of the population has recovered and . % has died, similar to estimates of the current situation in new york city [sta] . in figure a , a small infection spike affects % of the population before lockdown measure are put in place. in this case, we see additional deaths of . %. in figure b , a large infection spike affects % of the population, causing . % additional deaths. note that the lockdown is actually shorter for larger infection spikes, since the larger infection level (which occurs before lockdowns are imposed) moves the population closer to herd immunity. the price of this shorter lockdown, though, is higher mortality rates. . . parameter robustness. it is natural to ask how changes in other parameters affect the optimal controls. in general, changes in parameters create the expected changes in lockdown length and intensity, output loss, and mortality. the more interesting question asks about the level we use the data in [cdc] [hm] and multiply by . , since the cdc estimates that % of cases are asymptomatic. for the same reason, we multiply the symptomatic case fatality ratio for the + group by . to determine δ . we set δ j such that a % infection level causes a five-fold increase in deaths, as in [ace+ ] . all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint of impact of various parameters. the effects of non-pecuniary value of life (χ) have already been discussed and are displayed in figures and and table , but now we discuss other parameters. in table , we list the lockdown levels, mortality, and output loss for various other configurations of parameter choices. there are some elements of the model which do not have large impacts on the results. the icu overcapacity constraint barely affects results, likely due to a combination of low infection rates and a sufficiently high number of icu beds (on average) in the united states. this allows icu admittance rates to remain at or below the threshold. the expected vaccine arrival date also does not have much effect on the optimal lockdown levels, since the population is expected to reach herd immunity well before its introduction. based on the current situation, it seems extremely unlikely that a vaccine will be developed less than months after the start of the outbreak, so we considered expected vaccine arrival times of year (ν = ) and months (ν = . ) and found that neither adjustment had much effect. it can also be seen from the table that removing f , the cost representing future deaths due to current lack of health maintenance, lengthens the lockdown only slightly, and has little effect on output loss and mortality. in contrast, loss of future employment, ability to work remotely, indirect death rate in lockdown, inter-group interaction, and behavior-dependent infection transmission have significant effects on lockdown length and severity. adding output savings from remote work, future employment loss due to lockdown, and indirect deaths of lockdown affects the wider population and so produce similar changes in outcomes. the model produces uniformly better outcomes when the level of remote work, h, is increased, though lockdowns do last longer as seen in figure . intuitively, this follows from the idea that more people working remotely helps to maintain economic activity without increasing risk of infection. changing α i and α l in the opposite direction of remote work creates similar effects on the optimal lockdown policy, though outcomes move differently. increasing the length of projected future unemployment, α e , leads to a shorter lockdown and less output loss, while deaths increase. this suggests that α e influences the trade-off between output loss and mortality. and, when we look at varying values of α l , we can see what happens when the optimization tries to minimize deaths due to covid- while also trying to avoid deaths due to lockdown. when α l = and the model doesn't take indirect deaths into account, the lockdown extends for longer and has a larger effect on output, but we see much lower mortality levels. however, when α l is increased to deaths per , individuals at full lockdown, the model dramatically shortens the lockdown, which decreases output loss and indirect deaths but which leads to higher deaths due to covid- . we now discuss the effects of behavior-related parameters. interestingly, these are the only parameters we discuss which have also an effect on the uncontrolled outcomes. if the level of interaction between groups, ρ, is lowered, there is less interaction between the low-risk and highrisk groups and so lockdowns are more effective. in figure , when ρ goes from . to . , the low-risk group is able to begin easing the lockdown earlier since there is less worry about transmission to high-risk individuals. however, the lockdown lasts longer overall, since it is harder to reach herd immunity. this increase in output is offset by a drop in mortality. with the optimal lockdown, mortality decreases to . % compared to the benchmark of . % when ρ = . . with no lockdown, mortality decreases to . % from . %. the opposite occurs when ρ is increased to , meaning that the groups mix freely. herd immunity arrives sooner due to increased inter-group transmission, however mortality increases to . % with lockdown and to . % without it. this suggests that it is beneficial for high-risk individuals to exercise extra caution in their interactions with members of the low-risk group. the other parameter that reflects individual behavior, α i , also has a notable impact on optimal lockdown policies. this parameter determines the efficacy of personal actions taken to slow transmission of covid- . in the benchmark case, α i = . if transmission rates are constant (α i = ), then the lockdown lasts longer due to the increased likelihood of transmission and mortality increases to . %. in the uncontrolled case, all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint the mortality increases too, to . %. but, if α i is set very high (say , which implies that personal caution can reduce the transmission rate by % when infections reach %), then a lockdown is barely necessary, as shown in figure . in this case, uncontrolled mortality is a mere . %. this second scenario is perhaps too optimistic, but it demonstrates the potential power of social distancing. if we examine α i = and α i = , we see that the lockdowns increase as α i decreases. if we consider the more modest change from α i = to α i = and decrease the interaction level between groups from ρ = . to ρ = . , then imposing the optimal lockdown decreases overall mortality from . % to . %. if the population is less obedient and disregards lockdown measures, for example if θ decreases from . to . , we see that the lockdown is shorter because herd immunity is reached sooner. this decreases output loss, but leads to more deaths. if the population is more obedient, however, for example if θ = . , the effect of a given level of lockdown is larger with respect to the same level of output loss so lockdowns are less intense but last longer. this slows the arrival of herd immunity, leading to higher output loss, but has the benefit of lower mortality rates. finally, in figure we investigate the effect of increasing the herd immunity threshold σ(x) = min{t ≥ : r t ≥ x}. in all of our examples, the low-risk group is able to leave lockdown before the arrival of herd immunity, so moving this threshold does not have much impact on output loss. however, this parameter has important implications for the high-risk group. from figure , we see that there is a clear change in dynamics for thresholds of % and above, where the lockdown policy becomes comparable to that of [ace+ ] . the length of lockdown for the low-risk group increases by about days, but the high-risk group remains in lockdown until the vaccine arrives -an increase of almost days. this abrupt change in strategy arises because eventually the population reaches a steady state with very low infections. the number of susceptible individuals decreases very slowly and is driven largely by deaths due to lockdown, so will not achieve thresholds of % and over before the vaccine is expected to arrive. in this case, the pandemic has been neutralized even though the population has not reached the threshold prescribed by the parameters. these experiments demonstrate that there is not much benefit to be found from a planner overestimating the herd immunity threshold. if the herd immunity threshold is %, this is equivalent to removing the herd immunity exit time. the impact of σ can therefore be observed as an decrease in output loss and an increase in mortality, while decreasing lockdown length for the high-risk group by days and for the low-risk group by days. all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . table output loss: . %, total deaths: . % figure . recreation of [ace+ ] model (two groups and no herd immunity), parameters from table output loss: . %, total deaths: . % all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint figure . our model (two groups, herd immunity), parameters from table herd immunity: days, output loss: . %, total deaths: . % (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . table . death rates based on [cdc] , γ = / is recovery rate as listed in table figure table . a comparison of mortality rates for different non-pecuniary value of life, death rates from table all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . table . parameter robustness results (note: lockdown for + ends at herd immunity) benchmark parameters: χ = /r, r = . %, α e = . , h = . , α l = − , α i = , ρ = . , f = , θ = . , ν = . , η = , death rates from table , herd immunity threshold = % all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint figure . comparison of optimal lockdown policy to no lockdown, using benchmark parameters, optimal lockdown deaths: . % vs uncontrolled deaths: . % all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . figure . results for varied initial conditions using benchmark parameters all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . figure . robustness results for ρ (inter-group interaction level) χ = /r, r = . %, ν = . , α l = − , α i = , α e = . , η = , f = , h = . all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint (a) herd immunity = %, output loss: . %, total deaths: . % (b) herd immunity = %, output loss = . %, total deaths = . % (c) herd immunity = %, output loss: . %, total deaths: . % figure . robustness results for σ (arrival of herd immunity) ρ = . , χ = /r, r = . %, ν = . , α l = − , α i = , α e = . , η = , f = , h = . (cont.) all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint a simple planning problem for covid- lockdown. working paper . national bureau of economic research optimal targeted lockdowns in a multi-group sir model. working paper a markovian decision process the employment situation cdc. covid- pandemic planning scenarios an sir model with behavior great recession, great recovery? trends from the current population survey students under lockdown: assessing change in students social networks and mental health during the covid- crisis age and sex composition a contribution to the mathematical theory of epidemics beware of bankrupt stocks like jcpenney and hertz international comparisons of intensive care: informing outcomes and improving standards immediate and long-term impact of the covid- pandemic on delivery of surgical services. advance online publication amid ongoing covid- pandemic, governor cuomo announces results of completed antibody testing study of , people showing . percent of population has covid- antibodies unprecedented disruption of lives and work: health, distress and life satisfaction of working adults in china one month into the covid- outbreak key: cord- - uiaruhg authors: balmford, ben; annan, james d.; hargreaves, julia c.; altoè, marina; bateman, ian j. title: cross-country comparisons of covid- : policy, politics and the price of life date: - - journal: environ resour econ (dordr) doi: . /s - - - sha: doc_id: cord_uid: uiaruhg coronavirus has claimed the lives of over half a million people world-wide and this death toll continues to rise rapidly each day. in the absence of a vaccine, non-clinical preventative measures have been implemented as the principal means of limiting deaths. however, these measures have caused unprecedented disruption to daily lives and economic activity. given this developing crisis, the potential for a second wave of infections and the near certainty of future pandemics, lessons need to be rapidly gleaned from the available data. we address the challenges of cross-country comparisons by allowing for differences in reporting and variation in underlying socio-economic conditions between countries. our analyses show that, to date, differences in policy interventions have out-weighed socio-economic variation in explaining the range of death rates observed in the data. our epidemiological models show that across countries a further week long delay in imposing lockdown would likely have cost more than half a million lives. furthermore, those countries which acted more promptly saved substantially more lives than those that delayed. linking decisions over the timing of lockdown and consequent deaths to economic data, we reveal the costs that national governments were implicitly prepared to pay to protect their citizens as reflected in the economic activity foregone to save lives. these ‘price of life’ estimates vary enormously between countries, ranging from as low as around $ , (e.g. the uk, us and italy) to in excess of $ million (e.g. denmark, germany, new zealand and korea). the lowest estimates are further reduced once we correct for under-reporting of covid- deaths. electronic supplementary material: the online version of this article ( . /s - - - ) contains supplementary material, which is available to authorized users. sars-cov- , the virus which causes the covid- disease, is a zoonotic pathogen which emerged in wuhan in late (huang et al. ) . at the time of writing, in early july , it had already claimed the lives of over half a million people globally (beltekian et al. ). in the usa covid- deaths now exceed the number of us military deaths arising from all conflict since the second world war (statista ) while in the uk the four weeks to th april saw more londoners lose their lives to covid- than during the deadliest four week period of the blitz (morris and barnes ) . this death toll is only the extremely saddening tip of the much larger iceberg of disruption that covid- has caused and continues to cause. confirmed cases across the world now exceed eleven million (beltekian et al. ) and the true infection rate is likely far higher. each case imposes a real cost on every infected individual. while symptoms may sound innocuous, including a dry cough, fever, and tiredness (who a; verity et al. ) , longer term this morbidity is likely to impose significant costs on sufferers' health, including potentially permanent lung damage or fibrosis associated with impacts upon the heart, kidneys and brain (citroner ) , all of which are likely to have negative consequences for future well-being and productivity. moreover, alongside the vast disruption that the virus itself has caused directly, preventative measures have caused further disarray in the economy. at present, there are no known specific treatments or available vaccines to either cure or prevent covid- infections (who b). therefore governments world-wide have relied upon preventative measures which aim to reduce the number of people exposed to the virus, and lower the effective reproductive number (the average number of new cases per infection, known as r), ideally suppressing it below a value of at which point the number of active cases decreases over time (ferguson et al. ) . while some of these measures impose relatively little personal or economic cost (such as simple hand hygiene and the use of face masks), the failure of such measures to stem the rapid world-wide spread of the virus has necessitated international "stay at home" lockdown requirements, entailing significant impacts across the global economy. the international monetary fund (imf) predicts a contraction in global gdp of three percent in -a decline of . % relative to its october forecast-and a decrease which it describes as being "much worse than during the - financial crisis" (imf a). short term effects are even more extreme. for example, in the uk, gdp fell by . % in april (ons a), while those claiming unemployment benefits rose nearly % to over million (ons b), although even this is dwarfed by the % increase in us unemployment over the same period (aratani ). globally sovereign debt is also soaring: predicted to grow nearly % to $ trillion in (standard and poor ) as administrations around the world race to protect cash-strapped companies from going out of business in order to prevent further unemployment. at the human level, lives and livelihoods have been turned upside-down. hence the true economic costs are more diverse and quite possibly more severe than that captured by financial metrics alone. they include negative ramifications for people's mental health (pancani et al. ; chaix et al. ; branley-bell and talbot ) ; increased kurmann et al. ( ) note that small business employment contracted by % (over . million) between mid-february and mid-april since when over million had been rehired to the end of june . prevalence of domestic violence (mclay ); and likely reduce the educational achievement of today's children (pinto and jones ; van lancker and parolin ) . as with previous financial crises (hoynes et al. ) and pandemics (nikolopoulos et al. ) , the virus and the economic fall-out are disproportionately affecting people from disadvantaged groups and lower-income households. black, asian and minority ethnic people are more likely to be infected and die (bhala et al. ; garg ; khunti et al. ; yancy ; public health england ) ; and lower-income households are less likely to be able to work from home, so face greater negative income shocks (hanspal et al. ; hensvik et al. ) , just as poorer countries are likely to suffer more than richer nations (hevia and neumeyer ) . as is well known, different countries have had very different death tolls. the usa currently has the highest death toll in the world, already exceeding , deaths (as of th july ). in contrast, vietnam-which recorded its first case just days after the usais yet to experience a single death. understanding what drives these differences is clearly crucial, potentially enabling improved responses to the continuing covid- outbreak and future pandemics. this paper begins to answer the critical question of why different countries have suffered different death rates, and what we can learn for future policy. the remainder of the paper is set out as follows. in sect. we first compare the numbers of deaths attributed to covid- across all oecd countries. the paper briefly focusses upon the uk as an example of a broader pattern; that public reporting of numbers related to the pandemic can be somewhat misleading. next, we control for any within-country under-reporting by analysing the overall increase in all deaths above what would be seasonally expected. assessing these 'excess deaths' data suggests that in most nations for which information is available official reporting of covid- tends to explain most of this unexpected mortality. however, analysis also reveals some clear exceptions, such as in the netherlands, spain and the uk where more than % of all covid- deaths seem likely to have not been counted as such. addressing such reporting problems is an essential element of providing the informational base required for an evidence-based policy response to this and any future pandemics. in sect. we assess the impact of government decisions regarding lockdown, their effectiveness and the policy trade-off between economic activity and health risk that they reveal. accepting that they are a conservative estimate of the total impact of the pandemic, officially attributed covid- deaths are used to investigate the price of life implied by lockdown policies. first we use a simple regression analysis to show that differences in mortality rates between countries are not driven by factors which are beyond the short term control of policy makers-such as differences in income and equality which, at least within the time available to fight coronavirus are effectively fixed. this in turn allows us to examine the degree of control which policymakers do have at their disposal, such as the rapidity of lockdown imposition and the duration of such controls. we use country-specific susceptible-exposed-infected-recovered (seir) models, similar to the approach of ferguson et al. ( ) , to ask how changes in the timing of lockdown measures affect the current death toll. our analyses provide good evidence that these policy tools actually determine the majority of variation in covid- impacts between countries. finally, we link these estimates to financial data to reveal a huge variation in the implied price of life across countries. section concludes. table presents the number of tests, cases and deaths that are officially recorded as (at least in part) caused by covid- across all oecd countries as of th june (data from our world in data; beltekian et al. ) . as mentioned, and considered in greater detail subsequently, these official estimates are likely to under-estimate deaths from covid- . however, the degree of under-reporting is far from constant across countries. for example, while almost all countries only counted deaths which had been confirmed to be linked to covid- , belgium adopts a much broader approach also including deaths where covid- is merely suspected as a contributory factor (chini ) . this results in much higher death rates than in other countries. arguably adopting the belgian approach internationally might provide a more accurate picture of covid- mortality. it is worth drawing attention to the very substantial variation in tests, recorded covid- case numbers and official death tolls across countries. adjusting for population, iceland has undertaken far more testing per capita than any other oecd country, at over k/ million compared to just k/million in mexico. much media attention has been expended upon reporting cumulative covid- numbers in each country. in terms of cases the roughly million cases reported in the usa is indeed a prominent result. however, unsurprisingly it is the total numbers of deaths by country which has attracted more attention and again the us total of well over , deaths is eye-catching. however, this media and policy-maker focus upon totals disguises the true comparison of these figures in failing to make even the most basic of adjustments for variation in population size between countries. once this is done then the death rate per million shown in the final column of table reveals a substantially different story. here we need to rule belgium out of comparison because its addition of suspected covid- deaths to the confirmed deaths reported by other countries, upwardly inflates its death rate. given this, the death rate reported in the uk is the highest amongst all of the oecd, exceeding even those of spain and italy which experienced their first major outbreaks much earlier on in the pandemic. it is worth highlighting how reporting elsewhere can be somewhat misleading. we do so by focussing on the uk as this is the country we are most familiar with, but the story is highly likely to be similar elsewhere. figure graphs the development of total recorded deaths (vertical axis) for a selection of countries over roughly the first days since each country recorded its th death (horizontal axis). this graph and its selection of countries is dictated by that which the uk government chose to highlight for comparison at its daily coronavirus press briefings. setting aside for the moment the us trend, clear separation can be observed between those countries such as germany and korea, which rapidly entered into lockdown and quickly controlled the growth of the virus, and those countries such as the uk and spain, the figure is a redrawing of one which was displayed daily at the uk press briefing from th march until being left out of daily briefings from th may onwards. speeches by the prime minister on covid- had been conducted before then (for example on the th and th march) but they only became a daily occurrence with a relatively standardised format from th march onwards. slides from these briefings are available here: https ://www.gov.uk/gover nment /colle ction s/slide s-and-datas ets-to-accom pany-coron aviru s-press -confe rence s. where lockdown was delayed resulting in a higher plateau. this is the first indication of the positive effects of early lockdown action, which we consider further subsequently. the uk government's decision to only display the total number of deaths in each of the countries shown took no account of even basic differences between countries such as population size; and as table has already shown, this makes fair comparison of death rates difficult. it might seem unusual to fail to make such basic adjustments, however the choice of such a display by the government is one which shows the uk cumulative total initially below that of european neighbours such as italy and spain and consistently dwarfed by that for the us, rising to more than twice the uk level. the fact that the us population is more than five times that of the uk, and that therefore per capita rates were much higher in the uk, is not obvious in this display. during the early days of the coronavirus outbreak, this omission of per capita data and focus upon cumulative totals allowed the uk government to make cross country comparisons which indicated that the country appeared to be faring better than many international counterparts (such sentiments are clear in transcripts of the verbal explanation which accompanied the graph, presented in online appendix ). for example, on the st april, the graph was described by the uk government as showing "it has not been as severe here as in france, and we are just tucked in under the usa and obviously italy on a different trajectory". however, as the pandemic developed so the performance of the uk relative to these other countries worsened. this situation was exacerbated by an outcry against the uk government's use of statistics based only upon deaths within hospitals rather than also including those in the community, ignoring obvious discrepancies such as a clear rise in deaths within care homes into which elderly hospital patients had been moved without testing for coronavirus (discombe ; grey and macaskill ) . shifting to reporting deaths from all settings revealed that the uk was faring far worse than nearly all other cumulative deaths (vertical axis) plotted for various countries (as selected for comparison in uk government briefings) over approximately the first days since each country recorded its fiftieth death (horizontal axis). note that spain's apparent decrease in cumulative deaths around day is an artefact of their reporting problems countries and indeed in per capita terms was experiencing one of the highest death rates globally (beltekian et al. ) . the impact upon the official narrative presented at uk press briefings was swift and noticeable. while initially much emphasis had been placed upon the uk's apparently favourable performance compared to other nations, now government officials started to mention the difficulty of making cross country comparisons, as highlighted by the pink dots at the top of fig. (and data presented in online appendix ). these caveats increased in both regularity and stridency until, on th may , cross country comparisons were removed from government press conferences. we have no reason to suspect that the uk government was unique in attempting to provide a positive representation of trends. however, a failure to provide clear and objective information is a well acknowledged cause of mistrust in authority (kavanagh and rich ) and is corrosive to public life at any time, but particularly in a pandemic where trust in institutions is vital. in undertaking cross-country comparisons of the impacts of covid- a first issue to be tackled is the difference in national approaches to reporting. this can be seen even in the reporting of testing statistics, differences which some authorities have argued may be politically motivated (norgrove ) . likewise, some countries (e.g. belgium) are far more likely than others to ascribe a death as caused by covid- (chini ) . given these concerns, we complement our comparisons of official covid- statistics with analysis of patterns in excess mortality data. here we define excess mortality for a country as the deviation in mortality rate during the period january to april compared to a baseline of expected deaths from previous years. excess mortality data is therefore not biased by differential rates of covid- testing or legislation on ascribing cause of death. there are however important caveats to the excess mortality figures. such numbers do not exclusively capture the increase in mortality that is directly caused by the presence of the novel virus. in addition, people may be less likely to visit hospital and therefore less likely to get treated for what are, in normal times curable diseases, thus tragically dying at a higher rate (thornton ) . similarly, first response services may get overwhelmed and therefore be less able to respond to life threatening emergencies such as heart attacks and strokes, again causing higher than expected death rates (oke and heneghan ) . acting in the opposite direction, government responses to coronavirus such as lockdown, may reduce the number of deaths from other causes; transmission rates for other communicable diseases are likely to be suppressed while a reduction in travel reduces the mortality associated with traffic accidents (alé-chilet et al. ). it is therefore not a priori obvious whether excess mortality is positive or negative. nonetheless, comparison of excess mortality with official covid- deaths will provide a more informed picture of the overall impacts of the pandemic within and across countries. table presents excess mortality data for the subset of oecd countries for which it is available. in general, the data are from the economist ( ) but are supplemented for some countries by data from other sources. baseline mortality is typically calculated as the mean number of deaths occurring in january-april - . excess deaths are calculated as the difference between the number of deaths observed in january-april and baseline mortality. the final column is the ratio of excess death to cumulative deaths at the end of april for each country, as reported by our world in data (beltekian et al. ), calculated as: the heterogeneity that was present in the statistics of officially recorded covid- deaths is also present in the excess mortality data. some countries, such as austria, iceland and portugal see only very marginal increases in death rates as compared with background death. there are countries which appear to do even better; denmark, finland, germany, israel and norway all observing fewer deaths than expected. as discussed above, these negative excess death numbers could be the result of measures to combat covid- reducing other-cause mortality, or from previous years used to calculate the baseline number of deaths being particularly bad. indeed does seem to have been a year with relatively few deaths from influenza (center for disease control ). at the other extreme, countries which appear worst hit based upon the officially recorded per capita death data are also those experiencing the highest percentage increase in mortality: belgium, spain and the uk all record deaths that are more than % higher than expected. note that italy too may well have been in this list, but the data for italy is only available to th march, about the time the country experiences its peak daily mortality. turning to the ratio of excess deaths to officially reported deaths, again there appears considerable heterogeneity across countries, suggesting countries are indeed measuring the death toll from the pandemic by very different yard sticks. generally, countries officially reporting high deaths tolls are also those which have the highest ratio of excess deaths to officially reported deaths. indeed, austria, iceland and portugal report more covid- deaths than the excess deaths they experience. it is worth noting this is not to say that these ( ) ratio = excess deaths∕officially reported deaths the other data sources used for particular countries are: austria -http://www.stati stik.at/web_de/stati stike n/mensc hen_und_gesel lscha ft/bevoe lkeru ng/gesto rbene /index .html); belgium-https ://epist at.wivisp.be/momo/; finland -https ://pxnet .stat.fi/pxweb /pxweb /en/kokee llise t_tilas tot/kokee llise t_tilas tot__vamuu _koke/statfi n_vamuu _pxt_ ng.px/; iceland-https ://hagst ofa.is/utgaf ur/tilra unato lfrae di/danir -tt/; ireland (note these are death registrations rather than government figures)-https ://rip.ie/death notic es/all; israel-https ://www.healt h.gov.il/units offic e/hd/ph/epide miolo gy/pages /epide miolo gy_repor t.aspx?wpid=wpq &pn= ; new zealand-https ://www.newsr oom.co.nz/ / / / , , /arethere -hidde n-covid - -death s-in-nzs-stati stics ; spain (importantly accessed on th june, after there was a major addition to the figures)-https ://www.scb.se/conte ntass ets/edc b f ad d e ed / - - %e % % pre limin ar-stati stik-over-doda-inkl-eng.xlsx; usa-https ://data.cdc.gov/ nchs/exces s-death s-assoc iated -with-covid - /xkkf-xrst. even among the countries for which data is available, mortality data are only available for a few months of the year, generally at least to the end of april, hence the focus january-april deaths. data tend to be aggregated to the week level, hence the exact endpoint is rarely th april . rather, the last day used in is determined by the data availability, and chosen to be as close as possible to th april. in all cases, we compare like-for-like, such that the baseline deaths are recorded over the same time period. likewise, the cumulative death toll we use to calculate the ratio of excess to reported death is that which was officially reported on the last day of the mortality data we use for each country. for some countries data availability means this is not possible. for austria, belgium and germany it is - ; iceland and usa use - ; for spain baseline deaths are modelled by momo. countries are recording deaths as covid- when they are not; rather it is entirely plausible the interventions to prevent covid- in these countries have suppressed other deaths too. at the other extreme, some countries, notably the netherlands, spain and the uk, have ratios which imply upwards of % of covid- deaths that are occurring are not being officially recorded. there are of course outliers to the overall pattern. belgium, france and sweden, have ratios below despite having high per capita death tolls. likewise, chile and new zealand have very high ratios, but these are almost certainly an artefact of them having so few covid- deaths by the end of april, rather than because of under-reporting in each nation. to recap, there are vast differences in the number of cases and deaths caused by coronavirus in different countries. this heterogeneity does not merely disappear when we account for potentially different reporting guidelines in each country; rather it may even be exacerbated. so what could be driving these patterns? while most countries chose to implement a relatively similar policy response, they did so at different times in their respective pandemics and some have been criticised for only belatedly imposing lockdown. there is some early correlative evidence that differences in current death tolls could be explained by lockdown date (burn-murdoch and giles ) and we now move to consider this issue in greater detail. our investigations of the potential impact of different approaches to reporting show the usefulness of an internationally agreed standard for assessing the impact of the pandemic. however, in the absence of such a standard we use national official estimates of covid- mortality to understand the impact of lockdown policies. data is supplied by the our world in data programme (beltekian et al. ). an initial task was to estimate the overall impact which policy responses could plausibly have had on covid- mortality. to achieve this we undertook regression analysis examining the extent to which variation in covid- deaths across all oecd countries might be explained by socio-economic and demographic differences which no government could reasonably be expected to address during the timescale of a pandemic. a number of such exogenous determinants have already been highlighted in the literature. of these one of the most clearly established mortality risk factors is a positive association with age; all other things considered, older sufferers are more likely to die from contracting covid- than are younger people (dowd et al. ) . therefore, across countries, populations which include a greater proportion of elderly people are likely to report higher death tolls. similarly, those living in closer proximity to others may be more likely to pass on and contract the respiratory disease, hence variation in population density across nations may be a determinant of covid- deaths (rocklöv and sjödin ) . beyond simple average population density, the degree to which populations are clustered in large urban centres may influence covid- -related mortality (stier et al. ) . health outcomes might also differ because of within-country variation in wealth (marmot ) which we capture in our regression by controlling for the gini coefficient of income inequality for each country. richer nations are likely better placed to limit the spread of pandemics (e.g. hosseini et al. ), hence we use per capita gdp as a regressor to net-out cross-country differences owing to wealth. finally, previous studies (e.g. fraser et al. ) have highlighted that early detection may play a crucial role in halting virus spread, hence it seems plausible that countries which were exposed to covid- earlier in the pandemic, and that therefore had less time to prepare, faced worse consequences. to account for this, we use the regressor "warning days"-the length of time (in days) between the who declaring that the covid- outbreak was a "public health emergency of international concern" on th january and the country recording its th confirmed case (who c). the linear regression we use, details of which are presented alongside full results in online appendix , is deliberately simple and we are not claiming that the model necessarily captures causal relationships. however, even after including the list of exogenous factors which have been hypothesised to be major socio-economic and demographic drivers of cross-country variation in mortality rates, over % of the cross-country variation in covid- mortality differences remains unexplained. covid- deaths vary greatly across countries due to factors beyond socio-economics and demographics; the major remaining determinant is the policy responses implemented by national governments of which the most obvious difference is when different countries implemented lockdown. to investigate the impact of lockdown upon cross-country variation in covid- mortality we calibrate country-specific seir models. seir models have a long history of development (li and muldowney ) with applications across a variety of infectious diseases including measles (bolker ), hiv (shaikhet and korobeinikov ) and ebola (lekone and finkenstadt ) . more recently seir models have also been applied to covid- (e.g. annan ; flaxman et al. ; pei et al. ) . however, as far as we are aware, ours is the first study to use the seir modelling framework to examine the effects of lockdown timing across multiple countries in the same study, and the first to combine these results with financial forecasts to obtain cross-country implied price of life estimates. price of life estimates derived in this paper are of critical importance given that government intervention has the ability to save life, yet trades-off against other goods. for example, closing schools is expected to reduce the transmission of infectious disease, hence decreasing the number of lives lost in a pandemic by imposing a human capital cost on today's children (viner et al. ) . likewise, there is evidence that the more stringent the government intervention to reduce the spread of coronavirus, the fewer lives that have been lost (stojkoski et al. ). this too is not free: we all pay with restrictions on our basic freedoms. beyond coronavirus, governments spend money and introduce legislation which imposes significant costs on society in a variety of sectors: healthcare (nice ), road safety (dft ), and safety at work legislation (hse ). governments also often have to consider multiple policy options for issues of environmental concern, be that considering pollution (ackerman and heinzerling ) , climate change (stern ) or biodiversity loss (ellis et al. ) . here too, lives can be saved and lost as a consequences of policy decisions. hence understanding how governments should value life is of critical concern. indeed, a significant section of relevant policy documents is occupied by discussion of the value which a government should place on statistical life when evaluating policy (e.g. the green book; h.m. treasury ). in the case of coronavirus, there are already studies which aim to assess the economic value of particular policy interventions by reducing the number of lives lost. hale et al. ( ) ask: how much of one year's consumption would an individual be willing to forgo in order to reduce the mortality associated with covid- , suggesting the answer lies in the range one-quarter to one-half depending on exact mortality rates. underpinned by assumptions about the rate of transmission and how policies may affect this, greenstone and nigam ( ) show the economic benefit of social distancing measures in the usa to be very substantial-about $ trillion. similarly, thunström et al. ( ) use initial global estimates for the basic reproductive rate, and assume decreases to transmission from policy intervention from studies on spanish flu, to go further. they conduct a cost-benefit analysis for similar measures, again in the usa, showing that the net benefits exceed $ . trillion. gandjour ( ) and holden and preston ( ) conduct similar cost-benefit style analyses for germany and australia, respectively, both highlighting that lockdown comes out net positive. here we ask a different but related question. not whether lockdown makes economic sense, but rather what the timing of interventions reveal about the relative prices different governments place on their citizens' lives. we focus on countries with very different mortality rates and intervention timing-if there are discrepancies between countries for the price of life, they are most likely to be shown in this set of countries. in china, lockdowns were implemented on a province-by-province basis on very different dates. therefore, at the country-level our gdp calculations would be incomparable with other nations. to overcome this challenge, we additionally parameterise an epidemiological model for hubei, the province worst hit by the pandemic. we use the results from hubei in our price of life calculations to maintain comparability across countries. to be clear, the implied price of life should not be regarded as comparable to the value of a statistical life (vsl). specifically, vsl is a concept from normative economicshow much consumption should governments be willing to trade-off for an increase in the number of lives saved. this is a question which can be answered through stated-preference methods as has been done elsewhere (e.g. alberini ; carthy et al. ; jones-lee ) . rather, the implied price of life we calculate can be seen as an answer to the positive economics question of how governments actually do price lives saved in terms of consumption lost when making policy decisions. the key insight is that as the pandemic progressed governments continually had to decide when the moment was right to introduce a lockdown. earlier lockdowns would save more lives, but likely impose greater immediate costs upon the economy. likewise, delaying lockdown also delays the point at which a government becomes either morally or legally responsible for addressing the costs which such restrictions impose upon business. therefore, ex-ante the expectation was that earlier lockdown meant greater financial cost. expost, it seems governments may have been somewhat wrong to make that assumption as longer-term earlier lockdowns actually appear to be associated with shorter overall lockdown length, as is clear in online appendix , which in turn result in lower long-term economic costs (balmford et al. ). nonetheless, early imposition of lockdown imposed the certainty of cost, while a delay held out the possibility that the epidemic may turn out to be less severe than expected. gambler governments chose to delay rather than act. the chosen date of lockdown reveals a government's preferences regarding the trade-off between avoided deaths and gdp losses. relative to the chosen lockdown date, a later lockdown would have cost more lives, but reduced the financial impact. in its choice of lockdown date a government implicitly accepted the associated gdp loss rather than bear a greater death toll. earlier lockdowns would have had the reverse effect; saving more lives but at a greater cost to the economy. in choosing not to enter lockdown earlier, the government rejected the higher financial cost of earlier lockdown in favour of more deaths. hence, we are able to calculate both accepted and rejected prices for human lives: upper and lower bounds for the implied price of life in each country. a criticism of this method may be that decision makers at the time were unaware of the benefits of lockdown for public health. the evidence, however, points to the contrary. for example, it was reported in the print media at least as early as th march that the lockdown in wuhan was showing signs of slowing the spread of coronavirus (qin ) . within the uk there is evidence that scientific advisors notified the uk government of the benefits of lockdown two weeks prior to its imposition (barlow ) . calculations of the implied price of life for each country require two data points. first, the differential effect on human lives lost from a marginal change in lockdown date. second, the marginal effect on gdp from the same change in lockdown date. we use a compartmental epidemiological model to simulate the epidemic in each country and in particular to predict the outcomes of the counterfactual scenarios in which lockdown dates are changed. in this type of model, at any moment in time the population of a region or country is distributed between compartments according to disease status, and the function of the model is to describe (and predict) how the population flows between these that such a trade-off is inevitable and in principal morally defensible is not questioned, indeed it follows logically from the vsl. increasing economic costs impact upon human welfare. an approach which says that every life is of infinite value would impose infinite costs upon the economy, resulting in far greater losses of human wellbeing (and almost certainly life) than acting in a way which imposes an implicit and non-infinite price on life. it is the cross-country comparison of that implicit price which is examined here. our focus on gdp reflects both the ubiquity of this measure and a lack of available, robust, economic estimates of the wider welfare impacts of lockdown. to better understand some of those wider costs, we direct the interested reader to: branley-bell and talbot, ; burki, ; cash and patel, ; chaix et al., ; mclay, ; pancani et al., ; pinto and jones, ; sud et al., ; van lancker and parolin, . while driven out of necessity, we think that a focus solely on gdp is also justified. our interest is in the relative price of life across country. even accounting for the external costs, the relative pattern for price of life would remain; it could only be eroded if these external costs are disproportionately larger for countries with lower gdp-based price-of-life estimates. indeed grant shapps, a uk government minister, was questioned on th march , a full week before the uk entered lockdown, regarding why the uk was following the example of other countries in implementing a lockdown given evidence that such a response seemed to work. a summary of the interview is available on the sky website here: https ://news.sky.com/video /coron aviru s-uk-appro ach-entir elyscien ce-led-grant -shapp s- . there is also a video of the interview on the sky facebook channel here: https ://www.faceb ook.com/watch /?v= . compartments as the epidemic progresses. in the seir model which we are using, there are four compartments corresponding to susceptible (i.e., not infected, but vulnerable to the disease), exposed (a latent stage usually lasting a few days, where the victim has been infected but is not yet infectious), infectious (at which point they can pass the disease on to others), and removed (meaning they are no longer infectious and may be either recovered from the disease and immune, or else dead). in more complex models, the population may also be subdivided according to age and other factors, with each subdivision being compartmentalised according to disease status as previously described. this would allow for a more detailed representation of the structure of society and the progress of the epidemic as it spreads through the population, but such detail would greatly increase computational demands (especially for large ensembles of simulations as we are using here) and is not necessary for this work. for a full description of the model we are using, see annan and hargreaves ( ) and also house ( ) where the underlying model equations were originally presented. the flow of the population between the compartments depends on parameters which we estimate by fitting the model to observational data for each country. this model fitting process follows the standard bayesian paradigm of defining prior distributions for uncertain parameters, running the model numerous times with parameters sampled from these priors, and calculating the likelihood on the basis of how well the model outputs match the specific observational data that we are using. this process (using a markov chain monte carlo approach) is described in detail in annan and hargreaves ( ) . this approach requires around , model simulations for each experiment (i.e. country) and the results are represented by an ensemble of model simulations that samples our posterior probability distribution. one critical parameter of the model, which has been widely discussed in the literature and media, is the reproductive number or r, which is the number of new cases that each infectious case generates in a fully susceptible population. if r is greater than , the epidemic initially exhibits exponential growth until it infects a sufficiently high proportion of the population that the remaining susceptible fraction substantially shrinks. if r is less than , the epidemic decays, again exponentially. in our estimation procedure, we assume that all uncertain model parameters are fixed in time apart from r, which is treated as piecewise constant. we consider three discrete periods within which r is constant. first, there is an initial period prior to "lockdown" controls being imposed by governments. a new, lower value for r is then assumed to apply during the period of strict controls, with a third value applying after the controls are significantly relaxed. country specific lockdown dates that we use are detailed in online appendix . in reality, r and other model parameters are likely to vary somewhat during these periods but this piecewise constant approach has been widely used and captures the dominant features of the system (e.g. flaxman et al. ) . due to serious limitations in the testing and reporting of case numbers, we rely exclusively on daily reported death numbers for the calibration of our model. again, this is a common approach which is justified on the basis that the reporting of deaths is usually far more consistent and reliable than case numbers which depend strongly on testing capacity and policy. an alternative approach would be to use the number of excess death. while this may better reflect the number of deaths caused by covid than reported death statistics, daily excess death data are not available. moreover, the key results in the model are driven by changes in the rate of infection, hence even if death numbers in a particular country are underestimated due to systematic biases, this will not usually bias the estimates of model parameters. therefore to calibrate the models we use daily reported deaths from our world in data up to th june (beltekian et al. ) , and later suggest how accounting for excess mortality would alter our estimates. the prior estimate for r after the release of lockdown is taken to be n( , . ) which represents our assumption that the policies are intended to be as open as possible while keeping the epidemic controlled. in many cases, there are insufficient data to constrain this prior estimate strongly, and therefore it plays a greater role in our results than the priors used in earlier phases of the epidemic. estimates of all the r values, as well as our priors, are detailed in online appendix . lockdown clearly reduces the infection rate across the board. easing lockdown allows the infection rates to increase again. figure compares observed and modelled deaths in the uk, showing deaths on the (exponential) vertical axis over time. modelled mortality (the solid line) closely matches the actually observed deaths (circles), illustrating that the modelling framework is flexible enough and the methodology sufficiently rigorous that the epidemiological model well replicates the observed patterns in the uk. indeed, only on days do observed deaths fall outside the % confidence interval (shaded area), and all such occurrences are in the postlockdown period when the number of daily deaths is comparatively low. similarly, close relationships are displayed for the other countries in the equivalent plots (online appendix ), highlighting that the model well captures the country specific pandemic pathways. in order to calculate the effects of changing the dates of lockdown, we use the fitted parameter values, and perform simulations in which the date of imposing lockdown is changed-either delayed or advanced by days. we also explore advancing or delaying lockdown by or days, results of which are presented in online appendix . this approach is similar to that of others (e.g. flaxman et al. ) in which the effects of policies have been analysed. since we are using a single date to represent the net effect of multiple policies which were introduced across a period of several days, it would be more precise to interpret these scenarios as representing a change in the timing of all such policies by the given number of days. likewise, we identify the impact of lockdown using within-country variation in the rate of infection. therefore, to the extent that the stringency of policy interventions vary between countries, our simulations reflect the same countryspecific set of policy interventions of the same stringency being implemented either earlier or later. that said, the lockdown is widely believed to be the most important of these measures (flaxman et al. ) and so we consider our interpretation to be a reasonable approximation of the impacts of lockdown and variation therein. differences in total mortality for each country dependent on date of lockdown are calculated to th june . we also calculate the number of deaths that likely would have occurred were no lockdown implemented, again to the th june . for illustrative purposes, the graph of predicted daily deaths for the uk under such a scenario is in online appendix . in all cases, no correction is made for the possibility that hospitals got overwhelmed, causing an increase in infection-fatality ratios. to the extent that such an outcome would have occurred, yet more lives would have been lost under the delayed-and no-lockdown scenarios. the graphs are similar for all other countries, and hence not displayed here. table highlights the likely impacts of lockdown policy. it is clear that the imposition of lockdown likely saved in excess of million lives across the countries we examine. this overall analysis of lockdown is similar to that of flaxman et al. ( ) and comparison of overlapping results shows that they are in most cases strikingly similar. however, we caution against over-interpreting the result: it is likely that even without a formal lockdown, people would have socially distanced and engaged in other behaviours to limit covid- deaths. nevertheless, earlier governmental action would have saved a large numbers of lives, particularly in countries such as the uk and us who acted relatively late. prelockdown reproduction rates are substantially greater than one, hence across all countries, longer delays result in exponentially greater losses of life. the previous sub-section presented clear evidence that the choice of when to impose lockdown drastically affects the likely number of deaths. moreover, there is significant heterogeneity across countries in the number of lives that would have been saved had lockdown been implemented just days earlier or later. how does this heterogeneity translate into the implied price of life across countries? to assess the price of life we require estimates of the financial cost of lockdown on gdp. we first assume that the full cost of any extension to the length of lockdown is felt in the year . therefore, we estimates the cost to gdp by comparing the last imf forecasts of national gdp in prior to the pandemic (from october ; imf ) with their most recent forecast for (april , imf b). further assumptions are needed to understand the cost of a marginal extension to lockdown. the first is the relationship between lockdown length and cost to gdp. in line with the best available evidence, from studies in the us (walmsley et al. ) and thirty panglobal countries (with a focus on european nations, fernandes ), length of lockdown appears to be directly proportional to the percentage gdp loss. of course, not all of the gdp loss associated with an extended lockdown is the result of the policy decision alone: progression of the pandemic sufficient to warrant a lockdown (extension) would reduce gdp outlook anyway and there is good evidence that people were changing their behaviours to enact social distancing in advance of direct regulations (gupta et al. ) . moreover, it is not just the domestic pandemic which causes gdp losses-some is also driven by the state of the virus in other nations owing to trade (mandel and veetil ) . hence we must also make an assumption about how much of the loss in gdp in any given country is the result of the lockdown policy, rather than other factors associated with the ongoing pandemic. andersen et al. ( ) , chronopoulos et al. ( ) and goldsztejn et al. ( ) have all teased apart the effects of lockdown policy from the wider pandemic. all three suggest that the gdp loss caused by lockdown policy is approximately % of the total gdp loss experienced by each country. we note of course that there are reasons to believe this figure could be an over-or under-estimate of the proportion of cost attributable to the lockdown policy, and that this could also vary somewhat by country given that lockdown policy may have different impacts on different industries. nonetheless, we see the . estimate as offering a reasonable ball-park figure, and so adjust predicted gdp losses as per eq. : equation states that the gdp loss caused by changing the length of lockdown by some amount (either , or days; denoted i ), in country j , is calculated as the relative change in lockdown length, multiplied by the predicted change in gdp as forecast by the imf, and ( ) Δgdp ij = Δlockdown length i actual lockdown length j × imf forecast gdp loss j × . andersen et al use data from individual-level transaction data either side of the border between denmark, which imposed a lockdown, and sweden, which did not. denmark saw transactions reduce % in the immediate aftermath of lockdown imposition compared to sweden's % reduction. this suggests that . % of the gdp loss denmark experienced is caused by the lockdown rather than mere pandemic progression. chronopolous et al present evidence from either side of the uk lockdown, again using individualconsumer-level transaction data. this suggests a similar proportion of the overall cost is attributable to the lockdown policy: spending drops by . % in the week following lockdown (week beginning rd march ) relative to the previous period. goldsztejn et al conduct a modelling exercise linking economic data to an seir model for the uk again. this suggests that lockdown accounts for % of the overall economic downturn. as more accurate estimates of this key parameter become available, we would encourage the interested reader to replicate our calculations but with an updated estimate of the proportion of gdp loss attributable to lockdown policy to provide more accurate estimates of the price of life. the proportion of the loss attributable to the policy decision ( . ). we adopt the imf metric for measuring gdp in terms of purchasing power parity international dollars (ppp$) which is held constant such that it is equal to the us dollar. for hubei, we use the same formula as above, however the imf only publishes estimates gdp forecasts at the national level. therefore we partition the effect for hubei alone by multiplying by the proportion of china's gdp which hubei makes up ( . , ) . the necessary data, and calculated gdp outcomes, are presented in online appendix . it is worth highlighting two further implicit assumptions. first, we assume all of the gdp loss a country experiences occurs during the lockdown period. clearly, countries' economies were already contracting pre-lockdown, and likely will take a long time to return to normal functioning post-easement. however, our assumption ensures that the implied price of life we calculate is an upper bound. second, we assume that the date on which lockdown is eased is independent of the date on which lockdown was imposed. this is an open empirical question as it may be that earlier lockdowns halt the spread of the virus quicker, allowing an earlier end to lockdown. if earlier lockdowns result in earlier release this would lower the overall financial burden of lockdown. hence, again our assumption tends towards an upper bound estimate on the price of life. the additional assumption made for hubei may underestimate the price of life there: the contraction in china's gdp is likely most keenly felt in hubei, the worst hit province. our estimates of price of life would increase if we adjusted for this. aside from the caveat with respect to china, while our assumptions influence absolute estimates of the price of life, the only variables affecting the relative prices across countries are: ( ) the number of lives a change in the length of lockdown would save; ( ) the original length of lockdown in a country; and ( ) a country's gdp. these key variables are not assumed. to underscore the point, our assumptions cannot substantially influence the implied relative price of life across countries. to calculate the implied price of life from a change in the length of lockdown of a set number of days, i , for country, j , we link the predicted change in gdp to the change in number of lives lost as in eq. : our primary focus is for the most marginal change in length of lockdown we calculate: imposing lockdown either days earlier or later than its actual date. results for different changes in lockdown date, of and days, are presented in appendices and . these show that relative patterns remain unchanged. table showed that the exponential growth in infections means more lives are lost from a delay, than would be saved by shifting lockdown earlier by the same number of days. in contrast the modelled impact on gdp from moving the lockdown date by a fixed number of days is exactly the same; the only difference is in the sign (earlier lockdowns are a cost to gdp, later lockdowns a benefit). hence, the implied price of life is higher for moving lockdown earlier as opposed to later. ( ) implied price of life ij = Δgdp ij ∕Δlives lost ij moreover, as explained previously, by choosing not to impose lockdowns days earlier governments rejected saving more lives when the price was relatively high. similar logic reveals them to have accepted the implied price of life from a delay; they would rather bear the cost in terms of gdp than as further human lives lost. results from these analysis are presented in table . obviously, estimates for prices countries were willing to pay (accepted) are lower than estimates for the prices countries rejected. in almost all cases the estimates of the price of life are below thresholds typically used to estimate the vsl in cost-benefit analyses. hence, ex-post, it is highly likely lockdown enhance social welfare. as with progression of the pandemic, there is huge heterogeneity in the price of life across countries. comparing across countries those who pursued an early lockdown strategy reveal they are willing to pay a high price to save their citizen's lives, only rejecting prices above $ , , . the highest implied prices are in korea (> $ , , ) and new zealand (> $ , , ), both countries who acted swiftly to suppress the pandemic. however, those countries which imposed lockdown relatively late-on in their respective pandemics were clearly only willing to pay far less to protect lives. belgium, italy and the uk reject prices of life around $ , . clearly, delayed action in the face of exponential growth cost lives, and implied low price of life in those countries imposed lockdowns relatively late in the pandemic. two comparisons make this cross-country variation in the implied price of life particularly clear. first, the accepted price of life in china ($ , ) is about % higher than that for an american ($ , ). this is despite our methods meaning the calculated price of life for china is likely an underestimate. second, compare the acceptable price of life in germany ($ , ) with that in the uk ($ , ). the price of life for a german is nearly an order of magnitude greater than that for a british citizen. that vast difference is despite the two countries being very similar in terms of gdp per capita. these relative implied price of life comparisons are particularly pertinent. our methodology uses ex-post estimates of the number of lives saved to infer what government policy implies for the price of life. yet, these governments were clearly making the decisions ex-ante. nonetheless, these governments were making lockdown decisions at around the same time (except hubei which was far earlier), with nearly identical information sets. thus any differences in relative estimates would hold true even if the pandemic had proved to be far less deadly than it actually is. moreover, this heterogeneity in the price of life is not explained by different values for life. indeed, the implied prices are often far lower than official vsl estimates-seemingly, cash flowing through the market is worth much more than value passing through wellbeing, at least to some countries. the low rejected prices also imply that very few quality ideally we would assess all of the consequences of interventions (e.g. the mental health costs of lockdown) before making such an assertion. however, the difference between vsl values and our price of life estimates suggest that our statement is defensible (certainly for those countries where the latter measures are particularly low). moreover, vietnam would have been included in the modelling exercise, but we were unable to robustly parameterise our epidemiological models as so few cases (let alone deaths, of which there have been none) have occurred. this is true to the extent to which officially reported covid- deaths in china are accurate. if officially reported deaths are far lower than the number of deaths which have actually occurred, this figure may well be an overestimate of the price of life in china. we have not found data from china on excess mortality and so cannot speculate on the degree to which mortality data are accurate. adjusted life years (qalys) are assumed to be saved by governments in reducing covid- -related mortality; otherwise delays to lockdown seem nonsensical. for reference, in the uk the national institute for health and clinical excellence views a qaly costing between £ , and £ , as good value (nice ). as we mentioned when discussing table , those countries with high reported covid deaths, tend to be countries with high ratios of excess mortality to reported death, i.e. there is substantial under-reporting. to examine the extent to which our estimates change when we account for this under-reporting, we focus on the set of countries for which we have reliable estimates of that ratio, and where under-reporting appears prevalent. these countries are: italy, the uk and the usa. the estimates reported in table are calculated by dividing the estimates of the price of life by the ratio of excess mortality to reported deaths (from table ). the intuition behind this is that our estimates of lives saved by lockdowns (used in table ) are based upon reported death data, and hence should be scaled upwards by the degree of under-reporting of deaths. implicit in this correction is the assumption that the ratio of excess death to reported death is constant within a country throughout the pandemic. it is possible that the ratio declines during the tail of the pandemic when covid cases and deaths are less common, and tests more available. nonetheless, our correction offers what is currently the most comparable cross-country figure. table shows that for those countries which under-report covid- deaths, implied price of life is substantially reduced, highlighting once again that earlier lockdowns would have increased social welfare tremendously. for example, in the uk, the country for which we estimate a relatively high rate of under-reporting of covid- deaths, the adjusted rejected price of life is just $ , (equivalent to just over £ , ). the accepted price of life is lower still, at $ , (£ , ). this study has begun to disentangle the extent to which cross-country comparisons of responses to covid- are valid despite difficulties caused by both exogenous factors and differences in testing rates and the recording of cases and deaths. the results presented in this paper suggest that policy interventions may well explain the majority of cross-country variation in officially reported covid- deaths. for some countries, deficiencies in official approaches to the recording of covid- mortality mean that estimates based upon deviation of overall deaths away from the seasonally expected norm may provide a more accurate depiction of fatalities caused by the pandemic. such 'excess death' estimates suggest that in some, highly impacted, countries the actual number of covid- deaths may considerably higher than indicated in official statistics. for example, within the uk it seems that more than a third of covid- deaths may have gone unrecorded. where under-recording is prevalent, then the number of lives lost by delayed intervention (as well as those saved relative to even further delay) is likely to be substantially higher than estimated in this paper. any such under (over) estimation of true deaths would result in an over (under) estimation of the price of life implicit in lockdown decisions. careful consideration of cross-country differences is required if we are to glean the important natural experiment evidence afforded by countries implementing different policy approaches to the pandemic. the results presented in this paper highlight that welldesigned policy can save life. while the economic burden of lockdown is large, comparison with prior decision criteria suggest that such policies generate net benefits for society. pricing the priceless: cost-benefit analysis of environmental protection what is a life worth? robustness of vsl values from contingent valuation surveys activity and the incidence of emergencies: evidence from daily data at the onset of a pandemic pandemic, shutdown and consumer spending: lessons from scandinavian policy responses to covid- tweet posted th model calibration, nowcasting, and operational prediction of the covid- pandemic antconc (version . . ) [computer software covid- : analogues and lessons for tackling the extinction and climate crises tweet posted th coronavirus pandemic (covid- ) (online) sharpening the global focus on ethnicity and race in the time of covid- . the lancet bolker b ( ) chaos and complexity in measles models: a comparative numerical study exploring the impact of the covid- pandemic and uk lockdown on individuals with experience of eating disorders burki tk ( ) cancer care in the time of covid- uk suffers second-highest death rate from coronavirus (online) the contingent valuation of safety and the safety of contingent valuation, part : the cv/sg 'chained' approach has covid- subverted global health? psychological distress during the covid- pandemic in france: a national assessment of at-risk populations why does belgium have so many coronavirus deaths? (online) consumer spending responses to the covid- pandemic: an assessment of great britain what we know about the long-term effects of covid- (online) transport analysis guidance data book, department for transport (online) government has misled public over uk deaths being lower than france (online) demographic science aids in understanding the spread and fatality rates of covid- do pollinators contribute to nutritional health? impact of non-pharmaceutical interventions (npis) to reduce covid mortality and healthcare demand fernandes n ( ) economic effects of coronavirus outbreak (covid- ) on the world economy estimating the effects of non-pharmaceutical interventions on covid- in europe factors that make an infectious disease outbreak controllable the clinical and economic value of a successful shutdown during the sars-cov- pandemic in germany hospitalization rates and characteristics of patients hospitalized with laboratory-confirmed coronavirus disease -covid-net, states public policy and economic dynamics of covid- spread: a mathematical modeling study does social distancing matter? britain left many of the weakest exposed (online) tracking public and private response to the covid- epidemic: evidence from state and local government actions (no. w ) oxford covid- government response tracker, blavatnik school of government. data use policy: creative commons attribution cc by standard exposure to the covid- stock market crash and its effect on household expectations which jobs are done from home? evidence from the american time use survey a perfect storm: covid- in emerging economies (online) the costs of the shutdown are overestimated-they're outweighed by its $ trillion benefit (online) predictive power of air travel and socio-economic data for early pandemic spread modelling herd immunity (online) who suffers during recessions? appraisal values or 'unit costs', health and safety executive (online) clinical features of patients infected with novel coronavirus in wuhan global manufacturing downturn, rising trade barriers (online) imf ( a) world economic outlook reports (online) the great lockdown (online) the value of changes in the probability of death or injury is ethnicity linked to incidence or outcomes of covid- ? bmj kurmann a, lalé e, ta l ( ) the impact of covid- on small business employment and hours: real-time estimates with homebase data statistical inference in a stochastic epidemic seir model with control intervention: ebola as a case study global stability for the seir model in epidemiology the economic cost of covid lockdowns: an out-of-equilibrium analysis social determinants of health inequalities shelter-in-place" isn't shelter that's safe: a rapid analysis of domestic violence case differences during the covid- pandemic and stay-at-home orders coronavirus: which regions have been worst hit? (online) assessing cost effectiveness, national institute for health and care excellence (online) an ecological study of the determinants of differences in pandemic influenza mortality rates between countries in europe sir david norgrove response to matt hancock regarding the government's covid- testing data (online) covid- -collateral damage in scotland (online) ons ( a) gdp monthly estimate office for national statistics (online) forced social isolation and mental health: a study on italians under covid- quarantine differential effects of intervention timing on covid- spread in the united states the long-term effects of educational disruptions (online) covid- : review of disparities in risks and outcomes (online) china may be beating the coronavirus the economic consequences of covid- lock-down in the uk. an input-output analysis using consensus scenarios high population densities catalyse the spread of covid- the uk's public health response to covid- stability of a stochastic model for hiv- dynamics within a host sovereign debt : global borrowing to increase to $ . trillion amid favorable financing conditions (online) number of killed soldiers in u.s. wars since world war i as of covid- attack rate increases with city size. mansueto institute for urban innovation research paper forthcoming the socio-economic determinants of the coronavirus disease (covid- ) pandemic the economist's tracker for covid- excess deaths (online) covid- : a&e visits in england fall by % in week after lockdown the benefits and costs of using social distancing to flatten the curve for covid- the green book: central government guidance on appraisal and evaluation covid- , school closures, and child poverty: a social crisis in the making estimates of the severity of coronavirus disease : a model-based analysis. the lancet infectious diseases school closure and management practices during coronavirus outbreaks including covid- : a rapid systematic review. the lancet child & adolescent health walmsley t, rose a, wei d ( ) impacts on the us macroeconomy of mandatory business closures in response to the covid- who ( a) coronavirus: symptoms (online who ( b) coronavirus: overview (online who ( c) coronavirus: overview (online) covid- and african americans (online) the authors are responsible for all views expressed in this paper, and any errors which may remain. that said, we are extremely grateful to a number of people and institutions for making this research possible. joseph billingsley contributed enormously to us being able to successfully gather excess death data for a number of countries. discussions with andrew balmford significantly improved the analyses we conducted. as sources of crucial data, and for their commitment to facilitating open science, we are extremely grateful to: our world in data, the oxford government response tracker, johns hopkins university's coronavirus resource center, the economist (and particularly james tozer and martín gonzález), and the financial times. key: cord- -iq njzoi authors: martin-olalla, j. m. title: age disaggregation of crude excess deaths during the spring covid- outbreak in spain and netherlands date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: iq njzoi spanish and dutch official records of mortality and population during the st century are analyzed to determine the age specific crude death rate in the spring covid- outbreak. excess death rate increases exponentially with age showing a doubling time [ . , . ]a (spain) and [ . , . ]a (netherlands), roughing doubling every five years of increase in age.the effective infection fatality rate in spain also shows this doubling time. statistically significant mortality increase is noted above a (spain) and a (netherlands). a statistically significant increase of mortality is also noted in spain for the youngest age group. the illness designated covid- caused by the severe acute respiratory syndrome coronavirus (sars-cov- ) caught worldwide attention since its identification in late [ ] . spain is one of the european countries most impacted by the disease during the spring of . confirmed covid- cases climbed up to several hundred thousands -some few thousands per one million population-and confirmed covid- deaths to some or six hundred deaths per one million population. [ ] lock-down measures came into effect on march to help diminishing the outbreak. total excess deaths or crude excess deaths is a key quantity to understand the acute impact of a pandemic [ ] . its determination sensibly requires a reference mortality or baseline and a time interval over which the excess is determined. baseline is determined from previous records of mortality. european national statistical institutes and eurostat are doing a great effort in disseminating weekly deaths in europe in the past years. this manuscript takes official records of weekly crude deaths in spain and netherlands during the st century and population records to ascertain the impact of the covid- in age specific death rates. in april eurostat set up "an exceptional temporary data collection on total week deaths in order to support the policy and research efforts related to covid- " [ ] . the data collection is disaggregated by sex, five year age group and nuts regions in several countries of the european union and elsewhere. data are provided by national statistical institutes on a voluntary basis. the weekly deaths still have the flag of "estimate". this manuscript will analyze the age disaggregation in spanish and dutch weekly death rate. the data set is not fully coherent. adding up deaths for every age group does not usually result in a number equal to the total number of weekly deaths included in the catalog. in total number of weekly deaths is larger than the sum of age grouped weekly deaths by % (spain) and % (netherlands). in a previous pre-print [ ] the total (all-age) weekly crude death rate in spain was analyzed and a total excess equal to death per one million population from w to w in the year was reported. the % confidence interval (ci) was [ , ] . age group population values until january , for spain and netherlands were also collected from eurostat demo pjan table. this table disaggregates age year by year which allows to build up five year age groups matching to the weekly death data. linear interpolation was used to compute population a weekly basis. total population values for january , can be collected from table ts in eurostat. unluckily there is no age disaggregation. however these figure were obtained in spain from the table at the instituto nacional de estadística. for the dutch set the last two available shares of population for every age group were used to extrapolate the shares of population in . eurostat requested the national statistical institutes the transmission of "a back series weekly deaths for as many years as possible, recommending as starting point the year ". in this manuscript spanish and dutch weekly . cc-by-nc-nd . international license it is made available under a perpetuity. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted august , . . https://doi.org/ . / . . . doi: medrxiv preprint deaths will be analyzed from on-wards. as eurostat points out "a long enough time series is necessary for temporal comparisons and statistical modelling". in age specific analysis the population n is assumed to be a blending, a mixture, or a multicomponent system composed of several age groups of size n i whose behaviour in relation to death is homogeneous under some circumstances. age specific weekly death rate [ ] d i is nothing but the ratio between weekly deaths d i within a group i and its population n i . all age specific deaths sum up the total weekly deaths d = i d i . also considering the definition of age specific death rate this can be written as d = i n i × d i from which the total weekly death rate d = d/n can be obtained as d = x i d i where x i = n i /n is the shares of population by age group. total weekly death rate is then the weighted average of age specific death rates. figure shows the age specific weekly crude death rate d i in spain for age groups (color from a gradient palette) and the total weekly crude death rate d (orange). some interesting points must be remarked. first, in the figure it is clearly perceptible that higher age groups are equally spaced along the y-axis, which is logarithmic. therefore the age specific weekly death rates are distributed exponentially with age and follows the gompertz exponential law [ ] d(a) ∝ a/τ , where a is the age and τ is a characteristic age time for which d(a) doubles. this empirical law suggests the predominance of age specific contributions to the cause of death instead of external causes like wars, murders, plagues or the like. one could collapse the lines in figure into a universal age-independent weekly death rate just by scaling weekly death rates with some exponential function whose characteristic time τ is obtained by fitting age specific weekly death rates. however no such scaling would fit to every age group. as an alternative it is useful to look for age specific baselines, which will model the behaviour of weekly death rate back in time. on the second hand and globally speaking, every age specific weekly death rate is decreasing with time in spain. just as an example a thin horizontal line at the level . % w − is plotted to highlight the evolution of the oldest age group. on the contrary the total weekly death rate stands still for nineteen years (see also ref. [ ] ). this result can be understood considering two competing phenomena. first the improvements in the public health system which helps decreasing age specific death rates. second, the ageing of population helped by the decrease in fertility rates and by elder people being less prone to die. as a consequence death rates tend to increases simply because population tends to be elder and more prone to die even though at age specific smaller rates. for the specific set up in spain in the . cc-by-nc-nd . international license it is made available under a perpetuity. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted august , . . past years the resulting total weekly death rate stands still. the ageing of population is perceptible in figure by looking at total weekly death rate in the background of age specific weekly death rates. at the beginning of the st century d (orange line) matched to the seven to last age group, which is - . in it matches to the six to last age group, - . following the ideas in ref. [ ] this work will analyze the cumulative weekly death rate from w to w which, in means from monday march , to sunday may , . in this period of time weekly deaths exhibits an extremely large anomaly in spain (z− score equal to ) and a moderate anomaly in netherlands (z ∼ ). the data collection allows to compute deaths recorded in this elapsed time in the past years for every age group. scaling by the group population at the given year gives us the cumulative death rate in this period of time and year. figure shows the results of this statistics grouped by age group and where age specific death rates are plotted against calendar year. every plot contains data points per country. spanish data are displayed by solid blue circles. dutch results, by open black circles. notice that the death rates are shown in their normal scores -the score after sample average is removed and sample is scaled by sample standard deviation-so that every age specific y-axis extends equally. figure also displays in a solid line the linear fitting from to . the broken lines shows the % confidence (prediction) interval for the residuals of the fit. this lines were extended until yielding the baseline or reference value of death rate for every age group and its confidence (prediction) interval. as discussed previously every age group displays decreasing values of observed death rates until . it is easily perceived in the figure that every age group older than years in spain showed an observed death rate outside the confidence interval of the normal behaviour in the previous years. the death rate in is an outlier from a statistical point of view. the same happens in netherlands above age a. it is unnecessary to collect every pearson's r correlation coefficient or every p-value in these analyses: no age group sustains the null hypothesis of no relationship between age specific death rates and years. the largest recorded p-value is p = . but usually p falls below − . contrastingly the total death rate in spain and netherlands does sustain the null hypothesis (p = . and p = . ) at the standard level of confidence. it is also worthy to note that the youngest age group also exhibits an increase in , which is equal in both countries. however it is only statistically significant in spain. it could be related to failures in the public health system amidst the strong stress lived in the spring of or to parents being reluctant to show up at hospital facilities. table and table summarizes the results displayed in figure . the tables first lists the shares of population in x for every age group. then the observed cumulative death rates o in and for the observation period. next group of columns displays the results in figure : first the predicted value or reference r for followed by the death rate excess e = o − r and three statistics related to it: the % confidence interval, the p −score e/r and the z−score computed as the excess e divided by the standard deviation of the residuals. [ ] if the ratio excess death to ci falls below then the ci includes e = suggesting that the excess is not statistically significant: in those cases p -scores and z−scores are not listed. for the sake of clarity the table then lists the excess age specific deaths e × x × n . finally in table last two columns display the results from the nationwide seroprevalence study [ ] s and the effective infection fatality rate ifr computed as e/s.[ ] to the best of our knowledge there is not a seroprevalence study in netherlands yet. mimicking figure one may think in plotting age specific cumulative death rates as a function of age group for every year in the collection. as an example first two columns of table and table shows a steadily progressive increase with age group. same happens for excess death rates and for ifr in table which perceptibly doubles in every step above group - . indeed such plots would have shown up the gompertz law in these magnitudes as observed in figure (panels a, b and c) . age specific excess death rates (panel c) gives τ = . a (ci [ . , . ]a) in spain and τ = . a (ci [ . , . ]a) while ifr in spain shows τ = . a (ci [ . , . ]a), as suggested by table : the ifr doubles every step in the age group staircase climbing until % for the most aged group. for age specific death rates (panels a and b) the analysis can be extended back to . figure shows the distribution of τ (y) in the past twenty years. errorbars display the confidence interval for τ (y). solid lines display the fitting τ versus y in the range to with broken lines showing the confidence interval. spanish characteristic times sustain the null hypothesis (p = . ) of non-relationship with calendar year. the observed value in is τ = . a, % down from the reference τ = . a. dutch times do not sustain (p = . ). the observed value in is τ = . a, % down the predicted value τ = . a. one of the most striking results is the anomaly observed for first age group -less than five years age-in table and table where it exhibits large p -score and z−score. somehow these scores may be biased by the model and is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted august , . figure ): the predicted value r and the excess e = o − r followed by % confidence interval, p −score e/r and the z−score computed as the ratio e to the standard deviation of residuals; and the age specific total deaths (e − r) × x × n . last two columns shows the seroprevalence in the country [ ] and the resulting effective age specific infection fatality rate ifr. figure ): the predicted value r and the excess e = o − r followed by % confidence interval, p −score e/r and the z−score computed as the ratio e to the standard deviation of residuals; and the age specific total deaths (e − r) × x × n . . cc-by-nc-nd . international license it is made available under a perpetuity. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted august , . . https://doi.org/ . / . . . doi: medrxiv preprint might be unrealistic. nonetheless, panel d in figure shows the ratio of observed age specific death rate in to observed age specific death rates in a magnitude which does not depend on modelling. in the panel the increase of mortality in elder groups is clear in spain (above a) and netherlands (above a) peaking at a. the first age group usually gets high mortality rates compared to the following groups due to its specific fragility. this is noted in figure by the lightest coloured age specific death rate in the middle of darker shades starting at × − w − . in figure the relative increase for the first age group in spain (× . ) matches to that observed at a and the modelling shows o well outside the predicted range of observations (see figure ). it should also be taken into account that age specific death rate for the first age group matches to that of the - age group. if the latter is high enough to make its ratio (× . ) relevant (making a % of increase) then the former is high enough to make its ratio (× . ) relevant (making a % of increasing, and doubling the impact of the latter). it is out of the scope of this manuscript to address the cause of this striking anomaly. whether this is a direct impact of covid- in the youngest age group or an indirect impact due to poor public health responses or to parents reluctant to show up children at hospital for health. either way it is showing the well-known fragility of this age group. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted august , . age specific excess death rates in spain and netherlands during the spring (w to w ) covid- outbreak follows gompertz exponential law with doubling times in the range [ . , . ]a (spain) and [ . , . ]a (netherlands) and roughing double every five years. same happens for the effective ifr in spain. this result is not far from the characteristic time observed for age specific death rates in the past twenty years which are . a (spain) and . a (netherlands). excess death for the eldest age group climbs to . % of the population of this group (spain) and . % (netherlands) which must be added to the usual . % (spain) and . % (netherlands) mortality rate for this group in this period of time. the youngest group age also shows the impact of the outbreak with a % increase of mortality rate in spain relative to after a long standing decrease in the past twenty years. is the author/funder, who has granted medrxiv a license to display the preprint in (which was not certified by peer review) preprint the copyright holder for this this version posted august , . the author declares no conflict of interest. every piece of data in this work comes from public collection found in eurostat and the instituto nacional de estadística. a pandemic primer on excess mortality statistics and their comparability across countries this manuscript will print death rates as a number times − where the number stands for the mortality per one million population. sometimes rates will be computed per capita and per some unit of time, like a number times − a − which means per one million population and per year and not "per one million years notice that, however, figure is scaled by the standard deviation of the sample the palette used in figure is named "dense", from cmocean palettes by kristen thyng https://matplotlib. org/cmocean/.jmmo thanks national statistics institutes and eurostat for releasing public records of weekly deaths. jmmo knew of eurostat effort from a tweet posted by kiko llaneras https://twitter.com/kikollan/status/ on july , .this work was not founded. this work was performed using free software running on xubuntu . . lts. data bases have been imported into gnu octave- . . (https://www.gnu.org/software/octave/). pictures were developed thanks to gnuplot- . . (http://www.gnuplot.info/). the manuscript was typeset in gnu emacs- . . (https: //www.gnu.org/software/emacs/) assisted by auct e x(https://www.gnu.org/software/auctex/). data in pictures and in tables have been exported directly from octave.this project started on july , . key: cord- - a dfzs authors: bermudi, patricia marques moralejo; lorenz, camila; aguiar, breno souza de; failla, marcelo antunes; barrozo, ligia vizeu; chiaravalloti-neto, francisco title: spatiotemporal dynamic of covid- mortality in the city of sao paulo, brazil: shifting the high risk from the best to the worst socio-economic conditions date: - - journal: nan doi: nan sha: doc_id: cord_uid: a dfzs currently, brazil has one of the fastest increasing covid- epidemics in the world, that has caused at least thousand confirmed deaths until now. the city of sao paulo is particularly vulnerable because it is the most populous in the country. analyzing the spatiotemporal dynamics of covid- is important to help the urgent need to integrate better actions to face the pandemic. thus, this study aimed to analyze the covid- mortality, from march to july , considering the spatio-time architectures, the socio-economic context of the population, and using a fine granular level, in the most populous city in brazil. for this, we conducted an ecological study, using secondary public data from the mortality information system. we describe mortality rates for each epidemiological week and the entire period by sex and age. we modelled the deaths using spatiotemporal and spatial architectures and poisson probability distributions in a latent gaussian bayesian model approach. we obtained the relative risks for temporal and spatiotemporal trends and socio-economic conditions. to reduce possible sub notification, we considered the confirmed and suspected deaths. our findings showed an apparent stabilization of the temporal trend, at the end of the period, but that may change in the future. mortality rate increased with increasing age and was higher in men. the risk of death was greater in areas with the worst social conditions throughout the study period. however, this was not a uniform pattern over time, since we identified a shift from the high risk in the areas with best socio-economic conditions to the worst ones. our study contributed by emphasizing the importance of geographic screening in areas with a higher risk of death, and, currently, worse socio-economic contexts, as a crucial aspect to reducing disease mortality and health inequities, through integrated public health actions. deaths from covid- , caused by the novel coronavirus of the severe acute respiratory syndrome (sars-cov- ), are considered avoidable because some collective and individual measures can prevent infection and because appropriate health assistance greatly reduces death risks. nevertheless, the risk of dying extremely varies according to some individual and geographic risk factors resulting in health inequity observed in several parts of the world (maroko et al., ; wang and tang, ) since the beginning of the pandemic. in brazil, covid- was first reported in the city of são paulo on february , (souza et al., ) . until august , four months after the first reported death, the disease had already resulted in , , confirmed cases and , deaths in the country (brasil. ministério da saúde, a) . at the moment, the united states of america and brazil are the epicentre of the disease. brazil is the fifth country in the world in territory and population, classified as an uppermiddle-income economy (the world bank, ) . almost % of the brazilian population is concentrated in % of the big cities among which são paulo is the largest. the brazilian unified health system (sus) guarantees healthcare for all citizens and thousands of foreigners residing or passing through the country (santos, ) . despite this universal and whole care, geographic differences in mortality rates across areas have been observed in the national and intra-urban scales. spatial heterogeneity in population features, such as age, underlying health, household densities, partial lack of sanitation, socio-economic status, contact networks and, mobility patterns (yang et al., ) have emerged as potential propellants of the spatiotemporal spread of the disease. since the beginning of the pandemic, mapping the disease occurrence and spread has become a powerful tool to track and support measures to slow down the infection transmission (kamel boulos and geraghty, ) around the world and at the local scale. web-based geographical information systems (gis) have allowed near real-time monitoring using map-centric dashboards (early alert, ; johns hopkins university, ; world health organization, ) . despite advances in the use of technology to reduce the impact of the pandemic, little is known about the spatiotemporal patterns of mortality from covid- , especially in an intra-urban setting. studying the spatiotemporal dynamic of deaths instead of cases may help better evaluate inequity. all the health disadvantages accumulated over decades of life due to any kind of deprivation increases the risk of dying from covid- . the lack of a robust spatiotemporal analysis undermines the comprehension of the mitigation strategies to potentialize the disease-control efforts. thus, this study aims to unveil the spatiotemporal dynamic of covid- mortality at a fine granular level in the city of são paulo considering the socio-economic context of the population. this approach may shed light on the urgent need for solid evidence on health inequities during the covid- outbreak. this ecological study, based on covid- secondary mortality data, was delineated in the city of são paulo, state of são paulo, brazil (figure ) , where live an estimated population of , , inhabitants whose distribution over the area produces a mean demographic density of , inhabitants/km², in (fundação seade, ) . when using covid- data, underreporting is always an issue that deserves attention, even for mortality. to minimize the effect of possible sub notification in deaths, we analyzed confirmed and suspected deaths from covid- . thus, death data comprised confirmed and suspected deaths occurred between march and june , , extracted on june from the mortality information system (sim), mortality information improvement program (pro-aim) from the epidemiology and information coordination (ceinfo) of the são paulo health secretariat (sms-sp). confirmed deaths due to coronavirus infection corresponded to the code b . (coronavirus infection disease), according to the international classification of diseases tenth revision (icd- ). suspected deaths are coded as u . (severe acute respiratory syndrome). the places of residence of the covid- deaths were geocoded by ceinfo/sms-sp according to the residence place using its databases and google maps api geocoding script that uses public places as the base map. resulting geocoded addresses were validated comparing the road or zip code where the record was allocated with the original zip code. geocoded data were assigned to the sample areas of the brazilian institute of geography and statistics (ibge), for which demographic and socio-economic census data are available (ibge, ) . we considered these areas as the spatial units in our models (figure ) . the records, including the basic cause, age, sex, the date of the death according to the epidemiological week (ew) (brasil. ministério da saúde, b) and the sample areas of residence were obtained by formal request to the são paulo electronic information system (e-sic, protocol ), whose data information is hosted in an open session, on the municipality's transparency portal for public access (são paulo. prefeitura, ) . here, we named this information as e-sic database. it was not necessary to submit this study to an ethics committee because we did not have access to the names and addresses. thus, the use of secondary data, without personal identification and in the public domain, dispenses with the need for prior approval by the ethics committee on research with human beings as per resolution no. / , of the national health council (guerriero, ) . we also used the data available in tabnet-datasus and named this information as tabnet database. tabnet is an app, available in <https://www.prefeitura.sp.gov.br/cidade/secretarias/saude/tabnet/>, provided by the municipal health department of são paulo and developed by datasus. this app provides free access (to any user) to population databases and to the database information systems of sus, such as the mortality information system (sim), which is supplied by the secretariat's program (pro-aim). through the tabnet app, it is possible to perform tabulation and crossing several variables of interest such as epidemiological week, sex, age group, and specific cause. the databases are updated periodically. as a measure of the socio-economic context of the population, since individual-level data are not available in the mortality database, we used a socio-economic index especially elaborated for health research. the geoses index (barrozo et al., ) was developed using principal component analysis, starting with variables. the index conceives the socio-economic context preserving seven dimensions based on the theoretical background (duncan et al., ; krieger et al., ) : education, mobility, poverty, wealth, income, segregation, and deprivation of resources and services. the index was defined in three scales: national, federative unit and, intra-municipal. figure presents geoses for the sample areas of the city of são paulo, showing that the areas with the best socio-economic conditions (geoses equal to or close to ) are located in the central part of the city and that these conditions deteriorate towards the periphery, where they reach the worst levels (geoses equal to or close to - ). it has been shown to be useful in studies of mortality from avoidable causes of deaths (from to years old) due to interventions at the brazilian health system in the national scale and, mortality from circulatory system diseases in the city of são paulo (barrozo et al., ) . we used the information of confirmed (b . ) and suspected (u . ) death from covid- available for the entire city of são paulo from ew th to th of tabnet database to calculate the weekly mortality rates of confirmed (b . ), suspected (u . ) and total (b . + u . ) covid- deaths. we did the same using the e-sic database from ew th to th. we exclude from the information e-sic database the covid- deaths occurred on th because the data from this week was incomplete (it was extracted on th june and included only part of ew th). these rates were obtained dividing the respective numbers of deaths in each week by the total population of the city and presented as death per , inhabitants-week. in the sequence, we obtained the mortality rates for confirmed, suspected, and total covid- deaths by sex and age for the entire period from ew th to th using tabnet and e-sic databases. these comparisons between the data of these two sources were useful to evaluate how complete was the data we used for the spatial and spatiotemporal analysis. for the calculation of the mortality rates by sex and age, we excluded the data without this information. as only one suspected covid- death occurred in ew th, we restricted our spatial and spatiotemporal analysis from ew th to th and spatial or spatiotemporal architecture was considered in all the models we performed. we first modelled the confirmed and total covid- deaths using spatiotemporal models only with the intercept and random effects accounting for spatial and temporal autocorrelation and the interaction between them. the spatial dependence was modelled considering a besag-york-mollié (bym) model with two components representing the spatially structured and non-structured random effects (blangiardo et al., ; blangiardo and cameletti, ) . these two components were considered independent one for another and followed the parametrization proposed by simpson et al. ( ) . the temporal dependence was modelled by a non-structured random effect and a structured random effect given by a random walk autoregressive model of firstorder (rw ). the interaction between space and time was modelled considering spatial and temporal non-structured random effects (blangiardo and cameletti, ) . the number of confirmed and total covid- deaths by ew and sample areas were modelled using poisson and zero-inflated poisson probability distributions in a latent gaussian bayesian model approach. we considered the expected confirmed and total covid- death for each ew and spatial unit as an off-set in these models. the expected deaths were estimated by indirect standardization taking into account the age and sex structure of each sample area and the mortality rates for the entire study period and city. this, therefore, enables us to interpret the outcomes of our analysis as relative risks (rr) concerning the mortality rates for the entire period and city. we obtained, from these models, the temporal and spatiotemporal rr. subsequently, we introduced the socio-economic covariate (geoses) in these models and obtained the corresponding rr. finally, we used a spatial approach to model the confirmed and total covid- death by ew to evaluate the role of the socio-economic covariate in each one of the ew. for doing this, we considered spatial models with intercept, bym spatial random effects, and the geoses as a covariate. the expected covid- deaths were obtained in a similar way for the spatiotemporal models, but considering the entire city mortality rates for each ew, allowing us to interpret the rr concerning the entire city mortality rates for each ew. we did our models in a bayesian context using the integrated nested laplace approximation (inla) approach (rue et al., ) . we selected our best models using the deviance information criterion (dic) so that the best-adjusted models were those with lower dic values (blangiardo and cameletti, ) . we used non-informative priors for the fixed effects and priors with penalized complexity for the precision parameters of the random effects (simpson et al., ) . we ran our models in the r environment (r core team, ). we found , confirmed and suspected covid- deaths in the tabnet database, from ew th to th , and , in the e-sic database, from ew th to th , in the city of são paulo. we removed deaths from the e-sic database because they were referent to the ew th which was not completed when the data were extracted. figure shows the mortality rates for covid- for both sources of data, considering the confirmed, suspected and total deaths. the curves from the e-sic database are similar to the curves with data from the tabnet database and the differences among these curves in ew th and th are related to a delay in the notification of the covid- deaths. table shows the numbers and mortality rates of confirmed, suspected, and total covid- deaths obtained from e-sic and tabnet databases from ew th to th by sex and age. to build table , we excluded six deaths with ignored age and three deaths with ignored sex from the , ( , - ) death in the e-sic database and we excluded six ignored age and four deaths with ignored sex from the , deaths in the tabnet database. we can observe that mortality was higher for males and that it increases as age increases, corresponding to . deaths per , inhabitants (in fourteen weeks) for people aged years or older. this pattern of increased mortality as age increases is also observed when making a greater stratification among people aged or older, as shown in supplementary material . table shows that the data we used to build our spatial and spatial models (e-sic database) is close to the municipality's official data on the pandemic (tabnet-datasus). for doing these models we excluded, from the e-sic database, a suspected covid- death occurred on ew th (the first one in the city) and nine with ignored sex or age and whose addresses were not geocodified and did not have the sample area codes. we achieved a high geocoding success rate of . %, once , records were geocoded using address data out of a total of , (excluding the first death in ew th ). of these, , refer to icd b . ( . % of the initial total of , records) and , to icd u . ( . % of the total of , records). during this period, . and . % of the sample areas had zero confirmed and total covid- deaths and, therefore. the dic values of the spatiotemporal models with poisson probability distribution were lower than the value of zero-inflated poisson distribution (supplementary material ) . considering each week separately, the amount of zeros deaths in the sample areas varied from . to . %. the dic values for the spatial models with poisson probability distribution were, in most cases, lower than the values with zero-inflated poisson distribution, and, in the cases that this did not occur, they are very close to each other (supplementary material ). from these results, we considered the best-adjusted spatial and spatiotemporal models that ones with the poisson distribution. first, we present the results of the models with spatiotemporal architecture only with intercept. we have, in figure , the temporal rr from ew th to th , where we can see that maximum rr occurred on ew th for the total covid- deaths and in ew for the confirmed ones in our study period. these results, even adjusted for the temporal autocorrelation, are similar to those presented in figure . considering the data presented in figures and , the apparent pattern of the temporal curves shows a tendency to stabilize, but a new rise in mortality cannot be ruled out soon. figures and show the posterior means of the spatiotemporal rr for the sample areas and ew, respectively, for confirmed and total covid- deaths. apart from the fact that the rr is greater for the total deaths than the confirmed ones, the distribution of the rr is similar between them, following the behaviour of the temporal rr. in the first two ew, the sample areas presented lower values of rr that increased over time. however, this increase occurred with greater intensity in the peripheral areas. next, we performed the spatiotemporal models, now taking into account the socioeconomic variable (geoses). table shows the spatiotemporal rr and the % credibility intervals for geoses obtained for the models with confirmed and total covid- deaths. in both models, it is noted that the high socio-economic level was shown to protect against the risk of dying from covid- throughout the study period. thus, the increase of one unit in the socioeconomic indicator represented a % reduction in the risk of dying from covid- , for the model using confirmed deaths, and a % reduction in the risk of dying, for the model using covid- total deaths. moreover, the risk of dying from covid- in the sample areas with the best socio-economic conditions (geoses close to ), about the areas with the worst ones (geoses close to - ) was % lower for the model with confirmed deaths and % lower for the model with the total deaths. finally, we performed the spatial modelling of confirmed and total covid- deaths in the ew separately, taking into account the socio-economic covariate. figure shows rr and % credibility intervals for geoses, according to the confirmed and total covid- deaths for each one of the ew. we identified a shift in the pattern of the relationship between covid- mortality and socio-economic status over time. the best socio-economic level presented itself as a risk factor for covid- deaths in the first two ew in the city of são paulo. from ew th , for the total death, and from ew th , for confirmed and total death, the worst socioeconomic level became a risk factor. even if some values were not significant, there was a continuous decrease in rr from ew th to th , followed by stabilization. this is the first population-based study on the evolution of the spatiotemporal pattern of mortality from covid- in the intra-urban setting of the largest city of brazil. using two different datasets, analyzing confirmed and confirmed + suspected deaths separately allows evaluating how uncertainty would impact the association between rr and the socioeconomic context. the robust models by ew clearly show when the high risk of death shifted from the best to the worst socio-economic conditions in the city. our findings showed that the most critical period regarding mortality by covid- in the city of sao paulo occurred between ew th and th, followed by an apparent stabilization of the temporal trend. however, it is not possible to predict a future scenario, as social distancing measures have been relaxed in the city since the th week (ms, ) this could increase the number of infected people and, consequently, the number of deaths. albeit social distancing alone seems not to be enough to contain covid- , many studies frequently concluded that it is a critical component of outbreak control (nussbaumer-streit et al., ) . it is important to point out that both total deaths and only confirmed deaths showed similar patterns in our study, despite its differences. the suspected deaths need, for the one hand, to be treated with caution, because they may be not covid- and, for the other hand, to consider them is one of the strengths of the study. first of all, there is a delay in the confirmation of the suspected cases, consequently part of the suspected deaths will be confirmed as covid- deaths in the future. furthermore, part of them would be confirmed, if there were no difficulties related to the strict case definition, which requires testing often not available or not performed in the appropriate time window. from this point of view, the amplitudes of variation in rates and relative risks obtained from confirmed and total deaths could be considered as lower and upper limits (or vice versa) for the magnitudes of these measures. the elderly population represents one of the groups that are more prone to the infection and symptomatology by covid- in the city of sao paulo, with a higher risk of death for men after the seventh decade of life, similar to statistics found for china and the united states (cdc. the novel coronavirus pneumonia emergency response epidemiology team, ). recently souza et al ( ) analyzed the brazilian population and also found that most covid- deaths were male, and the most frequent comorbidities were cardiovascular disease and diabetes. behavioural elements, especially posture that may prejudice adherence to lockdown measures, have been demonstrated as potentially crucial in determining susceptibility to sars-cov- (pawlowski et al., ; raisi-estabragh et al., ) . this unequal death ratio in men may be interpreted considering a lot of factors: their comparatively higher presence of comorbidities (i.e., hypertension, diabetes, cardiovascular disease, and chronic lung disease) (sharma et al., ) , higher risk behaviours (i.e., smoking and alcohol use), occupational exposure (global health , ) and, sex differences in immune responses (klein and flanagan, ) . in contrast, there may be other social and behavioural characteristics that favour women, with previous studies proposing women are more likely than men to adopt hand hygiene practices (johnson et al., ) and seek preventive care (bertakis et al., ) . the spatial distribution of suspected and confirmed deaths by covid- in the city of sao paulo shows inequalities, with spatial dependence and positive correlation associated with socio-economic factors of the areas, remarkably similar to the results of maciel et al. ( ) . our findings reveal that socio-economic status acts as a protective factor against the risk of dying from covid- . in the models with only confirmed deaths and with all deaths, the increase of one unit in the socio-economic indicator represented, respectively, a % and % decrease in the risk of dying. the first observation is that, when considering all deaths, the protective effect of the socio-economic level is more evident, showing that there must be a higher incidence of suspected deaths in the less-favoured areas concerning the most favoured areas (in areas with better socio-economic level, access to confirmation for covid- probably is more available). a study conducted by souza et al. ( ) reinforces this finding. they compared the spatial pattern of confirmed cases of covid- and severe acute respiratory infection with unknown aetiology with the socio-economic level in the metropolitan region of são paulo and found that the firsts were more associated with better levels and the seconds. they pointed out the degree of underreporting of covid- cases should increase with a decrease in socio-economic status. therewith, our results have confirmed the association between covid- and human development, pointing to the importance of geographic screening in locations with a potential for local infectious transmission as a fundamental aspect to coordinate better actions against the pandemic (maciel et al., ) . the low levels of socio-economic position reveal that not only the vulnerability of the population but also the difficulties in health services concerning diagnosis and treatment of the condition, similar to the overview of fragility expected from health services in brazil (ribas et al., ) and in the countries of latin america to face the pandemic (rodriguez-morales et al., ). besides, living conditions also may be strongly influenced by the low income in different ways, such as residence in more poor neighbourhoods and housing conditions, particularly confined or overcrowded housing (khalatbari-soltani et al., ) , which has been related with a greater risk of contagion from several other pathogens, such as helicobacter pylori (webb et al., ) , tuberculosis (gupta et al., ) or epstein-barr virus (gares et al., ) . besides, a person's employment may expose them to different risks related to the type of job (khalatbari-soltani et al., ) . work requiring continuous human contact, such as caring for people or interaction with others means that risk of infection dissemination through droplets of aerosol is higher (rule et al., ) . regarding covid- , studies showed that occupation is an explicit determinant of contagion and a secondary determinant of covid- severity and deaths by the association between occupational social class and comorbidities (khalatbari-soltani et al., ) . for example, workers such as cleaners, retail staff, teachers, or healthcare workers suffer the direct impact of the covid- incidence (koh, ) . people with underprivileged socio-economic conditions are more prone to be exposed to job stress including burnout syndrome and unemployment, which may contribute to disrupted immune and inflammatory system responses (berger et al., ; nakata, ) as well as a higher risk for comorbidities for covid- (kivimäki and kawachi, ) . until now, both debilitated immunity and the existence of comorbidities are recognized risk factors of covid- severity (khalatbari-soltani et al., ) . we showed that the first cases of deaths occurred in the neighbourhoods with the best socio-economic position in the city of sao paulo. this may be related to the fact that all of the infected subjects had been abroad (ms, ). in the first two weeks, the best socioeconomic level was presented as a risk factor. then there was a change in the spatial pattern: from the fourth week onwards, the worst socio-economic level becomes a risk factor. similarly, souza et al. ( ) showed a higher risk of diagnosed covid- cases in census tracts with higher per-capita income in the sao paulo metropolitan region during the early phase of the covid- epidemic. after these first cases in richer areas, the virus started to circulate in the suburbs of the city, with high population density and worsened sanitary conditions (silva and muniz, ) , and probably explain its fast transmission. the city of sao paulo is particularly vulnerable because it is the most populous in the country, with approximately million inhabitants (ibge, ), and is highly connected within brazil and around the world. its main airport, the são paulo-guarulhos international airport, is the largest in brazil, with non-stop passenger flights to destinations in countries (rodriguez-morales et al., ) . in this study, we used the sim database instead of sivep-gripe, unlike other studies (souza et al., ) . the recommendation of the ministry of health to register the notification of death and the monitoring of mortality from this system, in practice, leads to a longer time between the event and the use of information by the teams since the registration of death occurs initially in the sim, based on the death certificate and requires health service teams and/or covisa, who access the systems, to insert the evolution for each notified case or the notification of death for cases not previously notified. these characteristics, combined with the coping strategy adopted by sms-sp, which did not include mass testing, motivated the option of using sim data considering the confirmed and suspected diagnoses for analyzing mortality caused by coronavirus infection in the analyzes performed. our study findings must be considered in the context of several assumptions and data limitations. we associate patient's addresses or postcodes to area-based socio-economic positions using the geolocalisation, which may provide some insight into the likelihood of exposure to health factors and covid- risks. this approach is frequently used as representative for the individual socio-economic condition; nevertheless, they are not a perfect picture of individual circumstances, could underestimate the magnitude of social disproportion related to individual social measures (lamy et al., ) and are best employed in complement with individual-level variables to reflect geographical or aggregate-level risks (khalatbari-soltani et al., ) . we highlight that our spatial analysis is subject to methodological limitations caused by ecological fallacy and the modifiable areal unit problem. these constraints are intrinsic to any spatial analysis using aggregated data (subramanian et al., ) . even so, our study contributes to healthcare planning measures and for future precision studies focusing on the effects of social health factors on covid- deaths. also, one of the strengths of our study was dealing with deaths by covid- instead of using the cases, due to the better accuracy and reliability of the data. when we consider only the cases, there may be many underreporting asymptomatic patients that can hamper the conclusions. we used models with spatial and spatio-temporal architectures to investigate the spatial and spatio-temporal patterns of confirmed and total (confirmed and suspected) covid- deaths in the city of são paulo. the obtained results from considering both categories showed differences regarding the magnitude of rates and rr, however, there were no differences concerning the conclusions we achieved. the maximum risk of dying from covid- occurred between ew th and th, followed by an apparent stabilization of the temporal trend, but we did not rule out a new rise in mortality soon. we identified that the high socio-economic level was shown to protect against the risk of dying from covid- throughout the study period. however, this was not a uniform pattern, since we identified a shift in the risk of dying from covid- in the city of são paulo over time: from high risk in the best socio-economic contexts in the first two ew to high risk in the worst contexts, from ew th ahead. concerning sex and age, men and elderly people presented the highest risk from dying of covid- . our study has corroborated the relationship between covid- mortality and socio-economic condition, revealing the importance of geographic screening in areas with a higher risk for deaths as a crucial aspect to integrate better actions to face the pandemic. tables table . number and mortality rates (per , inhabitants in fourteen weeks) of suspected (u . ), confirmed (b . ), and total (u . + b . ) covid- deaths, according to e-sic and tabnet databases, sex, and age. city of são paulo, th to th epidemiological weeks, . data source: deaths: sistema de informações sobre mortalidade -sim/pro-aim -ceinfo -sms-sp. population: fundação seade. tabnet database was updated on / / and e-sic database was provided on / / . . distribution of mortality rates (per , inhabitants-week) of suspected (u . ), confirmed (b . ), and total (u . + b . ) covid- deaths, according to e-sic and tabnet databases and epidemiological week. city of são paulo, . data source: deaths: sistema de informações sobre mortalidade -sim/pro-aim -ceinfo -sms-sp. population: fundação seade. tabnet database was updated on / / and e-sic database was provided on / / . . posterior means of the relative risks and % credible interval for the socioeconomic covariate obtained with spatial models for confirmed and total covid- deaths, according to each one of the epidemiologic weeks. city of são paulo, th to th epidemiology week, . geoses: a socioeconomic index for health and social research in brazil multi-cohort study identifies social determinants of systemic inflammation over the life course gender differences in the utilization of health care services -pubmed spatial and spatio-temporal bayesian models with r-inla spatial and spatio-temporal models with r-inla sinan -calendário epidemiológico [www document vital surveillances: the epidemiological characteristics of an outbreak of novel coronavirus diseases (covid- ) optimal indicators of socioeconomic status for health research coronavirus (dashboard) [www document the role of the early social environment on epstein barr virus infection: a prospective observational design using the millennium cohort study covid- sex-disaggregated data tracker resolução n o de de abril de que trata das especificidades éticas das pesquisas nas ciências humanas e sociais e de outras que utilizam metodologias próprias dessas áreas role of socioeconomic factors in tuberculosis prevalence coronavirus covid- global cases (dashboard) [www document sex differences in public restroom handwashing behavior associated with visual behavior prompts geographical tracking and mapping of coronavirus disease covid- /severe acute respiratory syndrome coronavirus (sars-cov- ) epidemic and associated events around the world: how st century gis technologies are supporting the global fight against outbr importance of collecting data on socioeconomic determinants from the early stage of the covid- outbreak onwards work stress as a risk factor for cardiovascular disease sex differences in immune responses occupational risks for covid- infection measuring social class in us public health research: concepts, methodologies, and guidelines using ecological socioeconomic position (sep) measures to deal with sample bias introduced by incomplete individual-level measures: inequalities in breast cancer stage at diagnosis as an example análise inicial da correlação espacial entre a incidência de covid- e o desenvolvimento humano nos municípios do estado do ceará no brasil covid- and inequity: a comparative spatial analysis of new york city and chicago hot spots psychosocial job stress and immunity: a systematic review quarantine alone or in combination with other public health measures to control covid- : a rapid review sex differences in everyday risk-taking behavior in humans r: a language and environment for statistical computing non-white ethnicity, male sex, and higher body mass index, but not medications acting on the renin-angiotensin system are associated with coronavirus disease (covid- ) hospitalisation: review of the first cases from the uk biobank coronavirus disease (covid- ) and healthcare-associated infections: emerging and future challenges for public health in brazil covid- in latin america: the implications of the first confirmed case in brazil approximate bayesian inference for latent gaussian models by using integrated nested laplace approximations healthcare personnel exposure in an emergency department during influenza season sus anos: o início pedidos de informação protocolados à prefeitura via sistema e-sic sex differences in mortality from covid- pandemic pandemia do coronavírus no brasil: impactos no território cearense penalising model component complexity: a principled, practical approach to constructing priors epidemiological and clinical characteristics of the covid- epidemic in brazil revisiting robinson: the perils of individualistic and ecologic fallacy world development indicators combating covid- : health equity matters relation between infection with helicobacter pylori and living conditions in childhood: evidence for person to person transmission in early life novel coronavirus (covid- ) situation (dashboard) [www document mathematical model describing covid- in são paulo, brazil -evaluating isolation as control mechanism and forecasting epidemiological scenarios of release supplementary material . number and mortality rates (per , inhabitants in fourteenweeks) of suspected (u . ), confirmed (b . ) and total (u . + b . ) covid- deaths, according to e-sic and tabnet databases and the age groups among people over years of age. city of são paulo, th to th epidemiological weeks, . key: cord- -mz cvc p authors: bone, a. e.; finucane, a. m.; leniz, j.; higginson, i. j.; sleeman, k. e. title: changing patterns of mortality during the covid- pandemic: population-based modelling to understand palliative care implications date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: mz cvc p background covid- has directly and indirectly caused high mortality worldwide. aim to explore patterns of mortality during the covid- pandemic and implications for palliative care provision, planning, and research. design descriptive analysis and population-based modelling of routine data. participants and setting all deaths registered in england and wales between th march and th may . we described the following mortality categories by age, gender and place of death: ) baseline deaths (deaths that would typically occur in a given period) ) covid- deaths ) additional deaths not directly attributed to covid- . we estimated the proportion of covid- deaths among people who would be in their last year of life in the absence of the pandemic, using simple modelling with explicit assumptions. results during the first weeks of the pandemic there were , baseline deaths, , covid- deaths and , additional deaths. deaths in care homes increased by % compared to home and hospital deaths which increased by % and %. hospice deaths fell by %. additional deaths were among older people ( % aged [≥] years), and most occurred in care homes ( %) and at home ( %). we estimate that % ( % to %) of covid- deaths occurred among people who would have been in their last year of life in the absence of the pandemic. conclusions healthcare systems must ensure availability of palliative care to support people with severe covid- in community and hospital settings. integrated models of palliative care in care homes are urgently needed. approaches to death and dying reveal much of the attitude of society as a whole to the individuals who compose it (cicely saunders, ) . the covid- pandemic has led to an unprecedented surge in mortality around the world, with some countries such as united kingdom (uk), the united states, and italy particularly affected. in the uk and elsewhere the government response has been focused on prevention and curative treatments. there has been very little emphasis internationally on care needs, including palliative care needs, of people dying during the covid- pandemic. clinical research and observation has found that people dying with covid- are symptomatic, frequently experiencing breathlessness and agitation near the end of life. clinical uncertainty around the illness trajectories of those infected increases the complexity of clinical decision-making and communication with patients and family. deaths occurring during the covid- pandemic are likely to be associated with poorer bereavement outcomes for family and friends and higher distress and support needs among frontline health and social care staff. there is an important role for palliative care but these needs have not been described at a population level. large scale population-based studies exploring patterns of mortality are important for quantifying numbers in need of care, and for informing governments about the need to plan and reorganise services. in england and wales, official mortality data from the office for national statistics (ons) reveal that during the first weeks of the pandemic, ( th march to th may), there were over , covid- deaths, most occurring in hospital ( %), with % in care homes and few elsewhere ( %). the ons estimated that there were also over , 'additional' deaths which are deaths over and above the number of deaths expected based on the -year average, that were not certified as caused by . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org / . / . / this study aims to explore patterns of mortality during the first weeks of the covid- pandemic in england and wales ( th march to th may ) to understand implications for palliative care. the objectives are: ) to explore trends in place of death; ) to explore the age and gender distribution of baseline deaths, covid- deaths and additional deaths; ) to estimate the proportion of people who died from covid- who would have been in their last year of life, and differences by age; ) to use this information to discuss implications for palliative care provision, service planning, and research. descriptive analysis and population-based modelling using routinely collected mortality data for england and wales. we used publicly available data from the office for national statistics including: • weekly registered deaths in england and wales by age group for to . • weekly registered deaths in england and wales in the weeks corresponding to the covid- pandemic, between th march (week ) and th may (week ) . this includes all deaths and covid- deaths by age, gender and place of occurrence. • principal population projections of the total population of england and wales by age for ( -based) we first defined the following mortality categories: baseline deaths, covid- deaths and additional deaths, as detailed in box . . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . deaths in excess of baseline deaths that are not accounted for by deaths certified as being as a direct result of covid- . this group might include deaths due to covid- but not recorded as such, or deaths indirectly related to the pandemic for example through avoidance of hospital care. the numbers of people in each mortality category were calculated for england and wales between th march and th may as follows: we estimated the baseline deaths according to the average deaths in the -week period of interest (weeks to ) over the years to , by age and gender. data on covid- deaths was obtained from the ons weekly registered death, which includes deaths by age, gender and place of occurrence. the ons produce figures for deaths involving covid- , based on any mention of covid- on the death certificate, including where the physician suspects covid- based on symptoms but where no test was available. these figures include deaths outside of hospital. . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint first, we derived an estimate of the proportion of the population infected in each age group, by re-arranging the following equation: the information we used in this equation included the number of registered covid- deaths reported by ons ; the total population from the ons principal projections ; and infection fatality ratios by -year age groupings obtained from ferguson et al. second, we hypothesised that someone who is in their last year of life will have an increased risk of dying from covid- compared to someone in the same age group who is not in their last year of life. to account for this, we applied a risk of dying multiplier to this group based on evidence of increased risk of death among people with serious conditions (supplementary table ). we applied a range of multipliers, with our primary estimate based on the median risk of death across different conditions. therefore, the number of people who die with covid- who would be expected to die within a year was calculated using the following equation: we calculated additional deaths by subtracting the estimated baseline deaths and covid- deaths from all registered deaths on a per week basis over the -week period of interest (weeks to ). this was calculated for each age and gender group separately to identify additional deaths within different demographic categories. where this resulted in a negative value, the estimated baseline figure was reduced by this amount and the value for additional deaths was zero. age, gender and place of death . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint to understand the age and gender distribution of the covid- and additional deaths we grouped data into age groups (< , - , - , and > years) and compared them with baseline deaths over the period of interest (weeks to ). we described trends in place of occurrence of deaths, using the ons defined categories of 'hospital' (acute and community, not psychiatric), 'care home' (nursing and residential), 'home', 'hospice' and combined 'other communal' and 'elsewhere'. to estimate additional deaths by place of death over the period, we used place of death data for week to estimate baseline place of death, as weekly place of death information is unavailable for previous years. we sought the views and experiences of members of our patient and public network regarding their concerns around covid- pandemic (johnson et al, submitted). these consultations informed the focus and design of this study. a representative from the network provided feedback on the manuscript. this study uses routinely collected anonymised and publicly available data. no ethical approvals were necessary. between th march and th may (weeks to ) in england and wales there were , deaths of which , were from covid- . we estimate that this included , baseline deaths (as estimated by average weekly deaths over the same period during - ), and , additional deaths (figure ). . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint where people die in england and wales has changed ( figure ). taking all deaths together, weekly hospital deaths increased by % from week (n= , ) to a peak in week (n= , ). we observe a % increase in home deaths between weeks (n= , ) and (n= ), while care home deaths increased by % during the same period (week n= , , week n= , ). in week , care homes temporarily became the most common place to die. hospice deaths fell by % between weeks (n= ) and week (n= ), representing % of deaths during the week period. . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint most covid- deaths occurred in hospital ( %) and care homes ( %) with few occurring at home and hospice ( % and %). the location of the additional deaths has not been described. comparing place of death for additional deaths occurring in weeks - with week , we found the majority of additional deaths occurred in care homes ( %) and at home ( %) (figure ). during weeks - there were fewer additional deaths in hospital and hospice compared to week , indicating that there were fewer non-covid- deaths in these settings than usual. . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint we found the age distribution of covid- deaths to be overall older compared to baseline deaths, with % of covid- deaths aged > years compared to % of baseline deaths ( figure ). additional deaths are older still, with % aged > years and % aged > years. a greater proportion of covid- deaths are male ( %), compared to % of baseline deaths and % of additional deaths. . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . using routine data and modelling scenarios to understand mortality patterns during the covid- pandemic, we highlight that care homes temporarily became the most common place to die in england and wales, and that hospital and home deaths increased by over % while deaths in hospices fell by %. we show that people in the additional deaths category are proportionately more likely to be older, and to die in care homes and at home, than those who are certified as dying with covid- . finally, we estimate for the first time that almost half of covid- deaths occur among people who would be in their last year of life in the absence of the pandemic, and that this increases to two thirds of those aged over years. deaths in hospitals, care homes and at home increased during the first weeks of the covid- pandemic. rapid adaptation of models of palliative care is needed to support the . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint increased number of people dying in these settings. guidance on remote consultations in primary care and on managing distressing symptoms in the community are available to help support care for people with suspected covid- , but challenges remain in terms of workforce and equipment. specialist palliative care provision in care homes is currently variable, [ ] [ ] [ ] and guidance to support palliative care for people with covid- in nursing homes has been found to be lacking and limited in focus. in contrast, deaths in hospices fell during the time period. a similar phenomenon was found during the sars pandemic and highlighted the need for seamless continuity of care between settings. systems that enable hospice resources to be shifted from inpatient to community settings are likely to be important in future pandemic peaks. little has been described about the characteristics of the additional deaths that have occurred during the covid- pandemic. the aetiology of these deaths is not known, in particular whether they are undiagnosed covid- deaths, or deaths associated with the pandemic without being as a result of the disease (for example due to avoidance of urgent healthcare). there is emerging evidence that older people with covid- may present with non-specific and atypical symptoms such as delirium or low-grade fever. we found that the additional deaths group comprises predominantly older people in care homes, suggesting that these may be undiagnosed covid- deaths. irrespective of aetiology, this group are likely to have significant palliative care needs. the characteristics of the additional deaths population is likely to change as the pandemic progresses and deferred harm from missed routine check-ups or screening becomes more likely. at a population level we estimate that % of covid- deaths occur among people who would have been in their last year of life in the absence of the pandemic, and that among those aged > years % are expected to be in their last year of life. for those with advanced conditions living at home and in care homes, guidelines recommend that general practitioners (gps), community healthcare staff and community geriatricians urgently review . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint advance care plans with patients, and discuss their care preferences in the case of contracting covid- . , our analysis indicates that this group is potentially very large and that enacting this recommendation will take significant resource. while the focus of this paper is covid- , it should be noted that baseline deaths outnumber covid- and additional deaths. previous modelling has shown that around - % of the baseline group are likely to have palliative care needs. covid- will impact on this group for example through fewer face to face contacts, and potentially less availability of specialist palliative care support as services are suddenly stretched beyond their usual bounds. across all mortality categories there is the need for bereavement support as a result of higher prevalence of complicated grief. there will be the typical grief associated with death and dying, but on a much larger scale. in addition there will also be novel grief processes, linked to social isolation and distancing, self-blame around infection, and not being able to attend funerals. palliative care specialists with particular expertise in bereavement support have a role in supporting families in coping with bereavement, as well as and supporting colleagues to deliver this care alongside other relevant organisations. this is the first paper to estimate and characterise population-level patterns of mortality during the covid- pandemic in relation to palliative care. we have produced simple models with explicit assumptions to raise questions and stimulate discussion. we calculated additional deaths for each age and gender group separately to improve precision. however, this study has limitations. we have been limited by the available official data, including the lack of historical data on place of death by week in england and wales. we have used data for england and wales, and patterns are likely to vary by country and region, particularly with respect to care home deaths. the infection fatality ratio and risk of dying multipliers for . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . this study has focused on mortality only. other aspects of palliative care that are important to consider, for example symptom control, support with difficult conversations, advance care planning, and complex clinical decision-making. there are also longer-term needs among those surviving covid- , such as rehabilitation and palliative care, which are likely to be significant and important for service planning but were not modelled here. we have a population level understanding of the palliative care needs of people with severe covid- is essential to guide efficient health service planning, but has been missing from political rhetoric around the world, much of which has been focused on intensive care capacity. covid- is now the focus of much research but very few studies are focused on palliative and end of life care. , data and research are urgently required to answer the following questions: -what are the best integrated models of palliative care in care homes, that can be adapted for use during the covid- pandemic? -what is the impact of covid- on the hospice sector and how might hospice services best use their available resources? -what is the impact on the baseline group of the additional demand for palliative care? -what is the symptom burden among people dying with covid- in different settings? . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . -what will be the effect of covid- on medium and long-term mortality projections and palliative care need? our study has highlighted the high level of need for palliative care during the covid- pandemic. healthcare systems responding to covid- must consider the need for palliative care surge capacity, including the workforce and equipment to enable people to be cared for appropriately in all settings. the need for integrated models of palliative care in care home settings is imperative. the rapid changes to practice during the pandemic provide important opportunities for research, evaluation and learning and it is essential that new models are underpinned by evidence. we show that a high proportion of people who die with covid- would be in their last year of life in the absence of the pandemic. while these data are estimates, and underlying assumptions may change, they highlight the need for careful consideration of the palliative care needs, in addition to intensive care needs, of people with severe covid- . . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . data sharing: all data is publicly available ethical approval: not required competing interests: all authors have completed the icmje uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work. . cc-by-nc-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint world health organisation. coronavirus disease (covid- ) pandemic characteristics, symptom management and outcomes of patients with covid- referred for hospital palliative care the importance of addressing advance care planning and decisions about do-not-resuscitate orders during novel coronavirus (covid- ) bereavement support on the frontline of covid- : recommendations for hospital clinicians the epidemiological characteristics of an outbreak of novel coronavirus diseases (covid- ) in china priorities for palliative care research during the covid- pandemic: a rapid online stakeholder consultation with people affected by serious illness national institute for health and care excellence. covid- rapid guideline: managing symptoms (including at the end of life) in the community van den block l, et al. palliative care provision in long-term care facilities differs across europe: results of a cross-sectional study in six european countries (pace) hospice utilization during the sars outbreak in taiwan the role and response of palliative care and hospice services in epidemics and pandemics: a rapid review to inform practice during the covid- pandemic what is happening to non-covid deaths? atypical presentation of covid- in a frail older person national institute for health research. urgent public health covid- studies uk research and innovation. covid- research and innovation supported by ukri key: cord- -uvnbgsds authors: salazar, james w.; sharpe, bradley; raffel, katie title: sunset rounds: a framework for post-death care in the hospital date: - - journal: j gen intern med doi: . /s - - - sha: doc_id: cord_uid: uvnbgsds nan notifying the donor network and medical examiner were relatively straightforward. others felt more nuanced and skillbased such as debriefing with loved ones and discussing autopsy. a loud sob alerted me that his friend had entered the ward. i shared with her a few words of condolence and reflection. she was in disbelief. i hoped we could quickly manage the logistics so she could grieve in peace. gingerly, i introduced autopsy. however, before i could get to my untrained summary of possible benefits, she declined. she had more pressing concerns including what would happen to the body. my answer felt unhelpful. in a hopeful and guilty fabrication of staff members i would only later learn definitively existed, i explained that others would come to help and that, for now, she should grieve as necessary and be with the patient. i went to follow up with her a couple hours later, only to find an empty room. although we achieved so much in my patient's final days, the emptiness of that room and my final memory of his friend in distress would stay with me. i knew there had to be a better approach to post-death care for survivors and providers alike. as a resident, heroic accomplishments in end-of-life care have felt tarnished by an ambiguous set of post-death care responsibilities often performed in isolation and without formal training. the discomfort and awkwardness surrounding postdeath processes illustrated in the patient vignette are not unique to the plight of a resident though, but rather emblematic of an aspect of patient care that is broadly neglected by the healthcare system. to move post-death care beyond an afterthought, several changes should be implemented. first, clear institutional guidance on roles, responsibilities, and resources is needed. limited literature exists to guide best practices in post-death care. of the most thorough, the american academy of pediatrics published a review for pediatric death in the emergency department. they provide guidance on several essential aspects of post-death care including organ donation, autopsy, family bereavement, and care for the care provider. to support this need, we propose "sunset rounds" as a concise framework to address post-death issues (table ) . second, structured communication should be employed by a multidisciplinary team. sunset rounds can function as an interprofessional timeout, wherein a group consisting of primary physicians and nurses, relevant consultants (e.g., palliative care), spiritual personnel, and decedent affairs team members could gather to address post-death care. many hospitals have a decedent affairs team to assist family members with navigating post-death logistics. however, in our experience, the primary medical team typically has limited interface and awareness of this important, yet often understaffed, resource, a missed opportunity for a more effective, coordinated approach. third, trainees should have formal training and feedback on post-death care. autopsy is an example of the many educational opportunities in post-death care. my hospital requires us to inquire about autopsy. however, without formal training on the details and value of autopsy, it is unsurprising that many discussions unfold like mine did and that autopsy rates are "vanishing." it is only through my own research on out-of-hospital sudden cardiac death that i became familiar with autopsy. i learned that presumed cause of death is often wrong in cases of diagnostic uncertainty; almost half of sudden cardiac deaths by the world health organization (who) clinical criteria were found to have non-arrhythmic cause on autopsy (e.g., occult overdose, pulmonary embolism, intracranial hemorrhage). i also learned that incisions are made to facilitate open casket viewing and that autopsies typically do not delay funeral proceedings. autopsy is a surgical procedure; as with procedures on the living, trainees should receive formal training on how to appropriately inform consent. amidst the coronavirus disease (covid- ) pandemic, the fog of death looms particularly large. unique challenges-limited workforce, racial disparities, lack of patient and family contact, and specialized guidance on safe post-death arrangements in patients with covid- from the who -have compounded the difficulties of post-death transitions. overcoming these difficulties in post-death care will require a significant, sustained investment in education and resources coordinated across multiple disciplines. fortunately, as with other neglected areas of the healthcare system brought to light by covid- , we are beginning to see long needed recognition of and innovation in post-death care; these range from novel approaches to death disclosure training and condolence communication to renewed attention to the proper completion of the death certificate. now, more than ever, it as soon as possible after death, gather as a multidisciplinary care team for an interprofessional timeout to develop a plan and assign a responsible party for each of the following aspects of post-death care: notification of survivors -determine the most appropriate patient contact and the team-member best suited to disclose -use "spikes" principles and the words "died" or "death" -offer assistance in sharing the news with other friends or family -consider saying a few closing words honoring the patient care team should include primary physicians and nurses, relevant consultants (e.g., palliative care), spiritual personnel, and decedent affairs team members death in the context of covid- : for specific guidance on the safe management of a dead body in the context of covid- and how it may inform the above framework, please refer to the world health organization interim guidance *state reporting guidelines can be found at: https://www.cdc.gov/phlp/publications/topic/coroner.html †local networks can be found at: https://www.organdonor.gov/awareness/organizations/local-opo.html is important that we work collectively to care for each other, support our survivors, and honor the sunsets of our patients. death of a child in the emergency department spikes-a six-step protocol for delivering bad news: application to the patient with cancer infection prevention and control for the safe management of a dead body in the context of covid- : interim guidance the vanishing nonforensic autopsy refining the world health a call to include death disclosure training alongside cardiopulmonary resuscitation training: after the code bereavement care in the wake of covid- : offering condolences and referrals the importance of proper death certification during the covid- pandemic key: cord- -j u t j authors: berenguer, juan; ryan, pablo; rodríguez-baño, jesús; jarrín, inmaculada; carratalà, jordi; pachón, jerónimo; yllescas, maría; arribas, josé ramón title: characteristics and predictors of death among , consecutively hospitalized patients with covid- in spain date: - - journal: clin microbiol infect doi: . /j.cmi. . . sha: doc_id: cord_uid: j u t j objectives: we aimed to analyse the characteristics and predictors of death in hospitalized patients with covid- in spain. methods: retrospective observational study of the first consecutive patients hospitalized with covid- confirmed by real-time polymerase chain reaction (rt-pcr) assay in spanish centres until march , . the follow-up censoring date was april , . we collected demographic, clinical, laboratory, treatment, and complications data. the primary endpoint was all-cause mortality. univariable and multivariable cox regression analyses were performed to identify factors associated with death. results: of the , patients, males accounted for , / , ( . %), the median age was years, and , / , ( . %) had > comorbidity. the most common symptoms were a history of fever, cough, malaise, and dyspnoea. during hospitalization , / , ( . %) patients developed acute respiratory distress syndrome, / , ( . %) were admitted to intensive care units, and / , ( . %) underwent mechanical ventilation. viral or host-targeted medications included lopinavir/ritonavir , / , ( . %), hydroxychloroquine , / , ( . %), interferon-beta , / , ( . %), corticosteroids , / , ( . %), and tocilizumab / , ( . %). overall , / , ( %) patients died. mortality increased with age ( . % occurring in older than years). seventeen factors were independently associated with an increased hazard of death, the strongest among them included advanced age, liver cirrhosis, low age-adjusted oxygen saturation, higher concentrations of c-reactive protein, and lower estimated glomerular filtration rate. conclusions: our findings provide comprehensive information about characteristics and complications of severe covid- and may help to identify patients at a higher risk of death. carried out a block-wise forward procedure allocating the predictor variables into five clusters: sociodemographic characteristics, comorbidities, admission signs and symptoms, vital signs, and laboratory parameters. a multivariable regression analysis was fitted within each block using two criteria to achieve the best set of predictors: relevance to the clinical situation and statistical significance (p< . ). we used variance inflation factors to detect collinearity among predictors included in the multivariable models. we carried out a sensitivity analysis in which the order of entry of the blocks was inverted. we checked the proportional hazards assumption. variables with more than % of missing values have not been considered, and missing values were treated as a separate category for analysis. heterogeneity introduced by different hospitals was accounted for by using robust methods to estimate standard errors and, thus, to calculate % confidence intervals (ci) and p-values. statistical analyses were done using stata software (version . ; stata corporation, college station, texas, usa. this study is registered with clinicaltrials.gov, the final cohort included , hospitalized patients (see web-only supplementary figure s ) in which sars-cov- was detected by , pharyngeal swabs patients' characteristics, categorized by survival, are shown in table . in brief, males accounted for . %, the median age was years, and . % were > years old. most patients were spanish born whites. the age distribution of patients stratified by sex is shown in figure a . at least one comorbidity was present in . % and . % had at least three comorbid conditions. the most common comorbidities were arterial hypertension ( . %), chronic heart disease ( . %), diabetes mellitus ( . %), chronic pulmonary disease (not asthma) ( . %), and obesity ( . %). only . % of patients had hiv. before admission, . % patients were on angiotensin- converting enzyme (ace) inhibitors, and . % were receiving angiotensin ii receptor blockers (arbs) ( table ) . the median duration of symptoms before hospitalization was (iqr - ) days, and the most commonly reported were history of fever ( ( . %), presumed bacterial pneumonia ( . %), heart failure ( . %), and blood-stream infection ( . %). during the study period, . % of patients died, . % were discharged, and . % remained hospitalized. the median (iqr) time to death since the beginning of symptoms and since hospital admission was ( - ) days and ( - ) days, respectively. death was particularly high among patients ≥ years ( . %) ( figure b) and those with ≥ comorbid conditions ( . %). death was also very high among those with ards ( . %), those who were admitted to icu ( . %), and those who underwent mechanical ventilation ( . %). the median (iqr) length of stay was ( - ) days for patients who were discharged; and ( - ) days for those who remained hospitalized at the censoring date. independent predictors of death in the different clusters of variables are shown in table . in the final adjusted analysis, we found factors independently associated with an increased hazard of death: male sex, older age, arterial hypertension, obesity, liver cirrhosis, chronic neurological disorder, active cancer, dementia, dyspnoea, confusion, low age-adjusted sao on room air, higher white cell blood count (wbc), higher neutrophil-to-lymphocyte ratio, lower platelet count, prolonged inr, lower egfr, and higher concentrations of crp (figure ) . no collinearity was detected, the proportional hazards assumption was fulfilled, and the results were not changed when the order of entry of the blocks was inverted. kaplan-meier plots for death according to age and sex are shown in figure . the adjusted hazard ratio (ahr) of death for being admitted early in the epidemic (before march) vs later was . ( % confidence interval [ci]: . ; . ), p= . . the variable unilateral or bilateral lung opacities had missing values in % individuals and was not included in the final model. however, when this variable was included in the model, the ahr ( %ci) of death for bilateral opacities in comparison with unilateral opacities was . ( . ; . ) p= . . we also carried out two post-hoc analyses (data not shown). in the first one, the predictors of mortality among patients ≤ years were not substantially different from those found in the whole data set. in the second analysis, the mortality hazard did not change depending china ( , ) and one from uk. the majority of patients in all four cohorts were male. however, in comparison with chinese patients, those from spain and the uk were on average, two decades older and had a prevalence three times higher of comorbid conditions. it is not surprising thus that mortality was substantially higher in spain ( %) and the uk ( %) than china ( . and . %). presenting features were similar in all cohorts. however, dyspnoea was less frequent in chinese patients suggesting a more severe course in the older spanish and british patients. in our cohort, age was the main determinant of death, as has been in other series of hospitalized patients with , , , ) . independently of the higher prevalence of comorbidities, it cannot be ruled out that older patients could not have been prioritized to receive icu treatment. death was also significantly higher among men than in women, as has also been described in other cohorts ( , , , , ) . there are sex differences in innate and adaptive immune responses that might have an impact on the inflammatory response and outcomes of covid- and deserve further investigation ( ). hypertension was not only the most common comorbidity in our cohort, as in other studies, but also an independent predictor of mortality. the association between hypertension and poor outcomes in covid- does not seem to be simply a matter of high prevalence; alternative explanations include pre-existing hypertensive end-organ or endothelial damage and interactions between covid- and antihypertensive medications ( ). many patients with hypertension were receiving ace inhibitors or arbs, but they did not increase j o u r n a l p r e -p r o o f mortality. obesity was the fifth most common comorbidity in our cohort but one with the highest hazard of mortality. obesity has been found to increase the risk of hospitalization and severe outcomes during influenza seasons ( ). recent studies with covid- patients indicate that younger hospitalized individuals were more likely to be obese ( ) and that obesity is associated with severe pictures ( - ) and increased mortality ( ). other underlying conditions associated with an increased hazard of death were active cancer and cirrhosis and as has been reported elsewhere ( , ); meaning that clinicians should consider patients with these underlying conditions as a high-risk category for . we identified several routine laboratory markers as predictors of mortality, including the neutrophil-to-lymphocyte ratio, an indicator of systemic inflammation that has been found of prognostic utility in sepsis ( ), and covid- ( , our study is limited by the retrospective design and the high number of sites, which might have jeopardized the quality of the data. we tried to solve this by selecting simple and well-defined variables and by careful monitoring of the data. admission criteria might have differed between the sites; nevertheless, we controlled the site effect in the analysis. we could not include in the multivariable model some potentially interesting laboratory parameters, nor changes in laboratory findings over time. the study's strengths include the large sample size, which allowed the identification of a high number of predictors of death at admission, the analysis of clinical and laboratory variables, and the inclusion of sites from areas with different incidence rates. in summary, here we report the clinical characteristics of a large cohort of patients with covid- consecutively admitted to hospitals in spain during the first month of the epidemic. our findings ministerio de ciencia, innovación y universidades -co-financed by european development regional fund "a way to achieve europe", operative program intelligent growth - . distribution -no./with data (%) abbreviations: iqr, interquartile range; egfr, estimated glomerular filtration rate median (iqr) -cells/ x < , cells/ μl -no./with data (%) mg/dl median (iqr) /with data (%) pg/ml blood cell count; egfr, estimated glomerular filtration rate; ckd-epi, chronic kidney disease epidemiology collaboration; aptt, activated partial thromboplastin time; inr, international normalized ratio key: cord- - qlk y authors: rahmandad, h.; lim, t. y.; sterman, j. title: estimating the global spread of covid- date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: qlk y limited and inconsistent testing and differences in age distribution, health care resources, social distancing, and policies have caused large variations in the extent and dynamics of the covid- pandemic across nations, complicating the estimation of prevalence, the infection fatality rate (ifr), and other factors important to care providers and policymakers. using data for all countries with reliable testing data (spanning . billion people) we develop a dynamic epidemiological model integrating data on cases, deaths, excess mortality and other factors to estimate how asymptomatic transmission, disease acuity, hospitalization, and behavioral and policy responses to risk condition prevalence and ifr across nations and over time. for these nations we estimate ifr averages . % ( . %- . %). cases and deaths through june , are estimated to be . and . times official reports, respectively, at . ( - . ) million and ( - ) thousand. prevalence and ifr vary substantially, e.g., ecuador ( %; . %), chile ( . %; . %), mexico ( . %; . %), iran ( . %; . %), usa ( . %; . %), uk ( . %; . %), iceland ( . %, . %), new zealand ( . %, . %), but all nations remain well below the level needed for herd immunity. by alerting the public earlier and reducing contacts, extensive testing when the pandemic was declared could have averted . ( . - . ) million cases and ( - ) thousand deaths. however, future outcomes are less dependent on testing and more contingent on the willingness of communities and governments to reduce transmission. absent breakthroughs in treatment or vaccination and with mildly improved responses we project ( - ) million cases and . ( . - . ) million deaths in the countries by spring . explanatory mechanisms: a) differences in population density, lifestyle, and interaction patterns create variations in reproduction number. b) risk perception, behavioral change and policy responses alter transmission rates endogenously. c) testing is prioritized based on symptoms and risk factors, so detection rates depend on both testing rates and current prevalence. d) limited hospital capacity is allocated based on symptoms, influencing fatality rates. e) weather may have a role in transmission ( ) while age, socio-economic status, and pre-existing conditions impact severity ( , ) , with the poor and minorities suffering disproportionately ( ) . prior research has addressed different parts of this puzzle, including estimating the infection fatality rate (ifr) in well-controlled settings ( ) or using statistical extrapolation ( ) , assessing the asymptomatic fraction in smaller populations or through random testing ( , ) , and estimating prevalence using random antibody testing in highly affected cities and regions ( ) . yet we lack a global view of the pandemic that is both consistent with these more focused findings and simultaneously explains the large variance in official cases and fatality across different countries. in this paper we build and estimate a multi-country model of the covid- pandemic at global scale. our model captures transmission dynamics for the disease, as well as how, at the country level, transmission rates vary in response to risk perception and weather, testing rates condition infection and death data, and fatality rates depend on demographics and hospitalization. estimating this model using data from every country with more than cases for which testing data is available, we infer some of the key characteristics of covid- pandemic globally and inform how alternative control policies may have played out across the globe. we use a multi-country modified seir model to simultaneously estimate the transmission of covid- in countries. the model tracks community transmission, excluding the global travel network and instead separately estimating the date of introduction of patient zero for each country. within each country, the core of the model tracks the population through susceptible, pre-symptomatic, infected pre-testing, infected post-testing, and recovered states. basic transmission dynamics reflect a classical seir process, with a baseline transmission rate, incubation period, and duration of infection. building on this basic structure the model explicitly represents the dynamics of covid testing, including demand for tests, sensitivity of pcr based testing, and allocation of test capacity between positive and negative cases, as well as the dynamics of hospital system capacity, both with concomitant effects on transmission and mortality. it also endogenously generates population-level behavioral and policy responses due to the perceived risks and government intervention. infected people and those with other risk factors (e.g. hospital workers) or similar symptoms due to other illnesses. we assume, based on prior research ( ) , that infected individuals are potentially tested on average days after the onset of symptoms. on any given day, individuals with more severe symptoms get priority for testing. following ( ) , this prioritization exerts a selection effect that results in different average severity of covid in the tested vs. untested infected populations. we also assume a % false negative rate for pcr-based covid tests ( , ) . a positive test results in reduced contact rates for tested individuals, changing the transmission rates endogenously. thus the model tracks various mechanisms generating the observable measures: a daily reported infection rate that combines the total positive results from each day's tests with daily post-mortem infection confirmations, and a daily reported death rate of confirmed cases. unobserved parameters, such as ifr and asymptomatic fraction, can then be estimated by fitting the model against data for observables. we estimate the effective hospital capacity for each country as a function of total hospital beds available, modified for population density. this modification approximates the effect of geographic heterogeneity in demand for hospital capacity -in larger and less densely populated countries, geographic heterogeneity in the progression of the epidemic means that some areas (e.g. new york city) may be overwhelmed with demand even when other parts of the country have unused hospital capacity. as with testing, hospital capacity is allocated between more severe covid cases (both those tested and those not tested) and demand from non-covid patients. hospitalization can reduce the mortality rate for covid patients, while other determinants of fatality rates are the severity of symptoms and the age distribution of the population ( , ) . each country also exhibits a response to the perceived risk of covid. the response is endogenously driven by a weighted combination of reported and actual hazard of infection. perceived risk drives reduction in contact rate and hence transmission rates through non-pharmaceutical interventions such as social distancing measures. specific government policies are not explicitly represented, but treated as part of this endogenous risk response. different countries could vary in their risk perception, response time, as well as the magnitude of their response based on the perceived risk. overall each country is represented by a system of ordinary differential equations tracking seven population stocks (susceptible, pre-symptomatic infected, infected pre-testing, and four groups of infected post-testing -positively tested or not × hospitalized or not) and a state variable for perceived risk. the model also includes multiple additional state variables for measurement and accounting of e.g. deaths and recoveries. model parameters are specified based on prior literature and formal estimation. main parameters adopted from the literature include incubation period ( days ( , ) ), onset-to-detection delay ( days ( ) ), sensitivity of pcr-based testing ( % ( , ) ), and post-detection illness duration ( days ( ) ). we estimate the remaining parameters using a maximum likelihood approach. using testing rate time series and various country-level data points (e.g. population, hospital capacity, comorbidities, age distribution), the model endogenously simulates confirmed new daily cases and deaths over time and matches them against observed data by maximizing the likelihood of observing those data given the model parameters. we use a negative binomial likelihood function to accommodate excess dispersion in infection and death flows compared to more common gaussian specifications. where data on mean) are . % ( . % for median across countries, med) and . % (med: . %) for cumulative infections and deaths, respectively. r-square values exceed . for % of country-specific cumulative and incidence time series. our projections are also consistent with excess mortality data, where available (see appendix s ). main findings include estimated parameters of covid- epidemics across countries, as well as trajectories of key measures. for estimated parameters we report a) the mean of most likely values across countries (mean); b) the standard deviation of those values (std), which quantifies variability of the underlying concept across countries; and c) the mean of country-level interquartile ranges (miqr), which captures the inherent uncertainty in parameter estimates for each country. the model brings together infection, testing, risk perception, behavioral and policy responses, hospitalization, and fatality using data from across the globe. here we focus on three main results. first, the magnitude of epidemic is widely under-reported with much variation globally. we estimate the asymptomatic fraction to be % (mean; std: . %; miqr: . %). combined with a sensitivity of % ( , ) for pcr tests, the fraction of infections that could potentially be identified without mass testing is limited to %. most countries do not reach this upper bound, however, and testing is the binding constraint. figure -a reports the ratio of estimated to reported infections and deaths, ranging between . and for infections and . to for deaths. in our sample of countries (total population of . billion), we estimate total infections at . ( - . , % credible interval (ci)) million and (ci: - ) thousands by june , . and . times larger than reported numbers respectively. under-counting has been larger earlier in the epidemic. for example, usa faced peak infection rate (of thousands per day) when detection rate was less than % of daily new cases. the most affected countries (based on percent infected) to date include ecuador ( %), peru ( . %), chile ( . %), mexico ( . %), iran ( . %), qatar ( . %), spain ( . %), usa ( . %), uk ( . %), and netherlands ( . %) (see figure -b) . despite the order-of-magnitude under-reporting of cases, however, herd immunity due to infection remains far from reach (figure -b) . the closest a a b . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint country has come to herd immunity is chile, which at its peak infection rate was (ci: days away from infection of % of its total population. second, we quantify both significant heterogeneity, and notable consistency, across countries in terms of basic parameters of the epidemic. the estimated initial reproduction number re ranges between . (indonesia) and . (iran), with the median of . (iceland) (re estimates include significant uncertainty; see figure s for details). even absent endogenous reductions in contact due to realized risk, these initial rates may be higher than longer-term levels for most countries because of early exposures happening in denser social networks. cross-country variations also reflect differences in population density, cultural practices, hygiene, as well as the timing of infection and thus early preparedness. we find support for the impact of weather on reproduction number, with the multiplier (rw) developed by xu and colleagues ( ) explaining transmission rates closely (rw . ). we estimate the global infection fatality rate (ifr) to have been . % (ci: . %- . %), with a wide range (figure -c) between . % (ci: . - . ; qatar) and . % (ci: . - . ; italy). the variations in ifr are primarily due to demographics with elderly having much higher risks ( ) . hospital availability and treatment effectiveness provide a second mechanism affecting fatality rates. we find that hospitalization can bring down risk of death to . % (std: . %; miqr: . %) of baseline. the model explains much variability in ifr with minimum country-level variability in basic fatality parameters. nevertheless, the quality of fit deteriorates in a handful of countries. for example, a ratio of simulated to reported deaths below one in a handful of countries signals that the model expected fewer deaths than observed. notably, we find it hard to reconcile, using only the model's mechanisms, the low fatality rates in qatar, singapore, and thailand relative to infection statistics, and high fatality rates in belgium and france. third, testing regulates transmission dynamics though a few mechanisms. by providing early warning, testing sets in motion the behavioral and policy responses essential to reducing transmission rate and controlling the epidemic. the exponential nature of contagion amplifies small early differences, such that a few days' difference in response time can lead to large differences in peak infection and final epidemic size. behavioral and policy responses to changes in infection risk are adopted rapidly, on average reducing effective contact (and thus transmission) rates in . (std: . ; miqr: . ) days. those responses are relaxed when risks attenuate, though down-regulating risk perception is slower (mean: . ; std: . ; miqr: . days). given limited data on rebounds in infections to date, how quickly perceived risk is downgraded is a major source of uncertainty in projecting the future of the pandemic. we also find that those with a positive test reduce their infectious contacts to a fraction of the original level (mean: . % of original contact rate; std: . %; miqr: . %). these reductions are especially important because asymptomatic individuals are estimated to be less infectious than symptomatic ones (mean: %; std: . %; miqr: . %). finally, by regulating infection rates through the previous two channels, testing also 'flattens the curve', allowing a larger fraction of those in need of medical care to be hospitalized. together, these mechanisms create an early race between testing and the spread of infection ( figure ). when infection is ahead, detection is compromised (moving up into the darker shades in figure) , responses lag further, and exponential infection growth can overwhelm hospital capacity. high early test capacity and its rapid expansion (moving to the right in the figure), brings the epidemic to light, activates policy responses, and allow for control. (panels a and b) shows one alternative testing scenario and counter-factual trajectories of pandemic to date. in this scenario, all countries shift their testing rates from baseline to . % of population per day (a rate currently achieved by multiple countries, e.g. usa, australia, in figure ). the shift happens on march , (the date covid- was officially designated as a pandemic). such enhanced testing would have reduced total cases from . millions to . (ci: . - . ) millions. corresponding reduction in deaths would have been from k to k(ci: k- k). by making additional assumptions on future testing and responses, the model can inform future trajectories. we explore a few projections out to spring that exclude vaccine and treatment availability. figure (panels c and d) shows projections under three scenarios: i) using the current country-specific testing rates and response functions moving forward; ii) if enhanced testing (of . % per day) is adopted on july st p; and iii) if sensitivity of contact rate to perceived risk is set to (approx. th percentile of estimated value across countries), leaving testing at current levels. contrary to the impacts of early differences in testing ( figure -a) , the reductions in future cases from additional testing (ii) are rather modest (from . in i to . billions in ii), because recognition of epidemic is no longer the bottleneck to response. both these scenarios project a very large burden of new cases in the fall , with hundreds of millions of cases concentrated in a few countries estimated a b c d . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint to have insufficient responses given perceived risks (primarily india, but also bangladesh, pakistan, and usa). in contrast, changes in response policies would make a major difference. scenario iii brings down future cases sharply, to as low as (ci: - ) millions (cumulatively) by the end of winter. across scenarios, we find that the infection rate in each country usually peaks and then stabilizes at a lower and slowly declining linear rate, in some cases after damped oscillations (i.e. second waves). these (approximately) steady-state infection rates are at levels that motivate just enough policy and behavioral response to bring effective reproduction number re to . faster rates lead to exponential growth, raise alarms and bring down contacts; slower ones lead to relaxation of policies bringing the reproduction number back up to . those post-peak infection rates vary widely across countries and depend on risk perception parameters, time constants, and contact sensitivity to risk, as well as weather and susceptible population size. in this scenario we shift both testing and contact sensitivity to perceived risks to arguably more realistic values between current ones and those in scenarios ii and iii in figure -c. specifically, on july st we shift test rates to a value % between the current rates and . % per day (which increases testing in most countries and reduces it in a few). similarly, sensitivity of contacts to perceived risk is shifted to % of the way between estimated country-level values and a high value of . resulting infection and death rates at the end of this period reflect the post-peak burden of the disease discussed above and are estimated at . % (ci: . - . ) and . e- % (ci: . e . appendix s provides projections for all countries. note that projections are highly sensitive to assumed testing, behavioral, and policy responses in the scenario. as such they should be interpreted as indicators of potential risk and not precise predictions of future cases; more rigorous testing and reductions in contacts in response to risk perception will significantly reduce future cases while laxer response and normalization of risks can lead to overwhelming breakouts. to enable empirical estimation and considering limits to data availability at a global scale, the model includes important simplifications and uncertainties that should temper the interpretation of the results. first, by analyzing an aggregate, country-level model, we abstract away much heterogeneity, from social networks to super-spreaders and local events. moreover, we offer no explanation for the underlying heterogeneity in reproduction numbers across countries. second, lacking hospitalization data across the world, the model's hospital sector includes additional uncertainty. similarly, our estimated behavioral and policy response functions do not tease apart the impact of various determinants such as national and local policies, business and event closures, use of personal protective equipment, reductions in physical interactions, and a host of other contributors. thus, results are not informative about the effectiveness of specific policies, and extrapolation of response function in future scenarios includes unknown uncertainties. for example, if normalization of risk reduced response magnitude in the second wave, our projections would miss that and prove optimistic. finally, absent explicit travel network our results would under-estimate the risk of the reintroduction of the disease in locations that have contained the epidemic. this paper provides a systemic view of the covid- pandemic globally. by incorporating testing capacity and prioritization, hospitalization, and risk perception and responses into an epidemiological model, we are able to closely match widely varying country-specific trajectories of the epidemic. our model explains the observed heterogeneities primarily through three pathways. first, estimates point to much variance in initial reproduction numbers across countries. this heterogeneity is not surprising given that reproduction number is very much a function of human behaviors and interaction is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . environments. nevertheless, our study was not designed to tease out the sources of that heterogeneity, an important task that requires other types of data and analysis. second, the course of the epidemic is also significantly impacted by risk perception and behavioral and policy responses that vary notably across different countries. finally, we show that differences in testing rates play an important role in shaping those trajectories. despite those variations, we also find much that is consistent across the globe: a) we estimate that just over half of the infections are asymptomatic, consistent with detailed estimates from smaller samples ( , , ) . b) we estimate that asymptomatic individuals are about a third as infective as symptomatic patients. c) in our model, hospitalization brings down fatality substantially, highlighting the crucial importance of keeping cases below hospital capacity. d) we find an average global infection fatality rate close to . %. this is consistent with growing evidence on ifr ( , ) . our estimates for ifr change predictably with age, and require few other predictors and no country-specific factors to stay consistent with the data. e) an order of magnitude under-reporting of cases is the norm across most countries, with a few percent of population already infected across many nations; nevertheless, herd immunity remains distant. f) finally, absent notable improvements in country level responses or breakthroughs in vaccination or treatment, the outlook for the epidemic remains grim, with most nations settling into a steady state of cases and deaths that, while below their peaks, are troublingly large. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . the model simulates the evolution of covid- epidemic, risk perception and response, testing, hospitalization, and fatality at the level of a country. here we explain key equations and structures in each sector, followed by complete listing of model equations and parameters in s . the model is a derivative of the well-known seir (susceptible, exposed, infectious, recovered) framework for simulating infection dynamics. figure s provides an overview of key population groups and the population movements among them . . we assume demand for testing and hospitalization are driven by symptoms, so all asymptomatic patients will be in the latter category. in the equations below we use short-hands to simplify mathematical notations. the full model documentation uses full variable names. table s provides the mapping between the short-hands and the full names, as well as the sources and equations for those variables and parameters discussed below. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint from these infectious categories, resolution flows (r…) take individuals to either recovered (r…) or dead (d…) states, with corresponding subscripts u, c, ch, and uh for stocks and uu, uhch etc. for flows. given the differences in severity and potential survival extension due to hospitalization, we distinguish between resolution delay for those in hospital (hospitalized resolution time; τh) and those not hospitalized (post-detection phase resolution time; τr). we use first order exponential delays for all lags, though sensitivity analyses showed very little impact of using higher order delays. the infection rate (rsp) controls the flow from s to p and depends on infectious contacts (ci), fraction of total population (n) that is susceptible, and weather effect on transmission (w). the latter is a function of rw, the country-level projections for impact of weather on covid- transmission risk year-round developed by xu and colleagues ( ) and a parameter, sensitivity to weather (sw), to be estimated: infectious contacts depend on the reference force of infection (β), various infectious sub-populations (and their relative transmission rates; ma for asymptomatic and mt for confirmed), and contacts relative to normal (fc), which captures behavioral and policy responses as a fractional multiplier to baseline infectious contacts: in this equation we separate various stocks (of i and p) into asymptomatic (a superscript) and symptomatic (s superscript). that distinction is treated analytically using a zero-inflated poisson distribution that is discussed in the next section. in light of evidence on the short serial interval for covid- , likely below the incubation period ( , ), we do not distinguish the infectivity of pre-symptomatic individuals from those post onset. contagion dynamics start from patient zero arrival time, t , another estimated parameter. the key mechanisms regulating the population flows among these stocks are discussed below, and a schematic of important relationships is provided in figure s . five parameters are estimated in the equations discussed above. one of them (sw) is global (i.e. assumed identical across countries; see the estimation section below for details on the distinction between global and country-specific parameters) and the remaining four are country-specific: β, mt, ma, and t . . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . covid- infection varies in acuity, from asymptomatic to life-threatening. acuity of disease affects notonly baseline fatality risk, but also testing and hospitalization decisions which impact official records of infection and fatality rates. since movement between population groups via testing or hospitalization is itself a function of acuity, to allow for consistent inference of mean acuity across different population groups, we use an analytical framework to track acuity levels. this framework, which we adapted from prior research ( ) , obviates the need to disaggregate the population by different acuity levels (which would prohibitively raise the computational costs for estimation). specifically, we represent acuity using a zero-inflated poisson distribution. this distribution combines two subpopulations -one with poisson-distributed acuity levels with mean covid acuity (αc), and another additional asymptomatic fraction with zero acuity, which is the zero-inflated component. the sum of those with zero acuity from the poisson part of the population and the second group is the total asymptomatic fraction (a). we assume this asymptomatic group is not given priority in testing or hospitalization, and is not at risk of death. thus they will always follow the → → → → pathway. the pathways for the remaining population depend on acuity and its impacts on testing, hospitalization, and death. note that the concept of acuity defined here only needs to have a monotonic relationship with tangible symptoms and risk factors and it does not have a one-to-one relationship with any real world measure of acuity, and as such is better seen as a mathematical construct that informs modeling rather than a real-world variable with clinical definition. from this framework two parameters, a and αc, are estimated as country specific parameters with limited variability across countries. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint the testing sector reads the active test rate (tt) for each country as exogenous input data (see appendix s for pre-processing details for this data). a fraction of this total test rate, typically small, is allocated to postmortem testing of covid- victims who have not been previously confirmed (post mortem tests total, tpm). specifically, of the deaths of unconfirmed infectious individuals (whether hospitalized or not), a certain fraction of fatalities screened post mortem (npm) will be identified true post-mortem tests. we anchor the npm to fraction covid death in hospitals previously tested (ndch). the rationale for this anchoring is that on the margin if there are many unidentified covid patients in hospitals, the chances are that the system lacks enough testing capacity and thus post-mortem testing should also be less thorough: we experimented other functional forms with a free parameter connecting the two constructs, but following our conservative estimation principle decided against including that free parameter in the final model. we feared that absent clear observables to identify this additional parameter (e.g. on country-specific policies regulating post-mortem testing) the degree of freedom would improve the fit but potentially for the wrong reason. the where the multiplier recent infections to test (mit), captures the sensitivity of negative test demand to recent infection reports. to allocate the available tests (tnet) between these two sources of demand, we use an analytical logic that allocates testing based on acuity of symptoms. the basic idea is that through self selection and screening by testing centers, people who have more symptoms or other signals that correlate with covid infection (e.g. high exposure risk) are more likely to be tested. we assume each unit of acuity increases the likelihood that an individual gets tested, based on a variable prob missing symptom, pms. this variable represents the probability that each acuity unit fails to convince the testing decision process to test a given individual, i.e. how selectively and sparingly tests are conducted. specifically, in this model an individual with k acuity units is tested with probability: we assume the negative test demand is coming from a population with a poisson-distributed, unit average acuity level (αn= ) for symptoms of non-covid influenza-like illnesses. the test demand from covid . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint patients also comes from a poisson distribution of acuity, but with mean αc. with the poisson distribution and given a level of α and pms, one can calculate the fraction of each demand source that would be tested: we therefore need to find the pms that allows test supply to match demand that is satisfied, specifically, by solving the following equation for pms*: having solved for p*ms (numerically), we analytically calculate the average acuity level for those positively tested (αcp: average acuity of positively tested ) and those either not tested or having received a false negative result (αcn). specifically, if test sensitivity was %, the average acuity for those not tested would be: the acuity level for those tested could then be found based on the conservation of total acuity across those positively tested and those not. starting with this basic specification we further account for the sensitivity of covid test (st) to calculate the values of αcp and αcn. we parametrize sensitivity at %, which is the estimated sensitivity for the pcr-based tests used as the primary diagnosis method of current infections of covid- ( , ) . overall, the testing rates that are determined by solving for * , combined with sensitivity of tests, inform the fraction of covid positive individuals transitioning from pre-detection (ip) to confirmed vs. unconfirmed states (ic or ich vs. iu or iuh), while the calculated α values inform the likelihood of hospitalization and fatality rates, as discussed next. the testing sector includes the following two country level parameters that are estimated: nst, mit. the hospitalization sector of the model starts with each country's nominal hospital capacity (hn) in total hospital beds. in practice, geographic variation in hospital density and demand creates imperfect matching of available beds with cases of covid- at any point in time, e.g. because some potential capacity is physically distant from current covid hotspots. this imperfect matching means some of the nominal hospital capacity is effectively unavailable at any time, especially in larger, less densely populated countries. we therefore calculate effective hospital capacity (he) by considering geographic density of hospital beds (bed per square kilometer; dh): where the * represents a large reference hospital density of . beds per km (which is the value of for south korea). the parameter sdh (impact of population density on hospital availability) is estimated. effective capacity is allocated between potential hospital demand (hcd) from covid- cases and the regular demand for hospital beds from all other conditions (which we assume equals pre-pandemic effective hospital capacity). we assume that covid- patients will have higher priority for hospitalization compared to regular demand. specifically, we assume that fraction of regular demand allocated (mhr) would be the . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint square of that for covid demand (mhc), = , and solve the resulting hospital capacity allocation problem analytically: we determine the covid demand for hospitalization based on a screening process similar to that for testing. two types of covid patients may seek hospitalization: those with confirmed test results and those without. the former are more likely to seek hospital treatment. we first calculate a parameter analogous to pms in the testing sector that informs the demand from confirmed covid patients for hospitalization. this parameter, the pmas confirmed for hospital demand (pmhc) is determined based on acuity level of confirmed (αct) and reference covid hospitalization fraction confirmed (rh), an estimated parameter capturing the overall need for hospitalization among covid patients: for unconfirmed covid patients we scale the analogue of this parameter (pmhu) based on how much priority non-covid patients generally receive: this formulation ensures that: ) confirmed covid patients are more likely to be hospitalized, but also that ) if there is ample hospital capacity (mhr~ ), then confirmed and unconfirmed covid patients will receive similar priority for the same level of acuity. in short, the pm. values determine hospital demand by confirmed and unconfirmed covid patients, which add up to hcd. the latter determines the fraction of hospital demand that is met. analogous to the testing sector, this fraction along with demand determines the flow of individuals from the pre-detection (ip) state to hospitalized vs. non-hospitalized states (ich or iuh vs. ic or iu). matching demand to allocated capacity also allows us to calculate the realized probability of missing acuity signal at hospitals (p*m) for confirmed and unconfirmed patients. as in the testing sector, those probabilities let us approximate for the expected acuity levels for covid patients in and out of hospital, as well as tested vs. not-tested, i.e. αct, αch, αu, and αuh. these average acuity levels in turn inform fatality rates for each group. the hospital sector includes two country level estimated parameter with limited variation across countries: sdh and rh. for patients in each of the u, c, ch, and uh groups we specify the infection fatality rate (f), as: the parameter base fatality rate for unit acuity (fb) sets the baseline for fatality rate. sensitivity of fatality rate to acuity (sf) determines how fatality changes with estimated acuity levels; more severe cases are expected to have higher fatality rates. hospitalization reduces fatality rates, expressed as the relative impact of treatment on fatality rate (shf). the function incorporates the impact of age distribution on fatality rates. for age effect, we calculate a risk factor for each country based on its age distribution and the relative fatality risks for covid patients in different ages documented in prior work ( ) . we normalize this factor . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint against its value for china where the original study was conducted. given the well-established impact of age on fatality, this factor is directly multiplied into the infection fatality equations. overall, the fatality sector includes three parameters that are estimated at the country level, with limited variance across countries, those are: , , and . note on comorbidities and fatality: we also explored including three comorbidities but found the estimates unreliable and therefore they are not included in the main specification of the model. those comorbidities include obesity, chronic disease, and liver disease. the effects we explored for each were: (.) = (.) (.) , where we used the following country-level indicators from the world health organization ( ), normalized by the average across all countries (d(.)): for obesity: prevalence of obesity among adults, bmi ≥ (age-standardized estimate) (%) for chronic health issues: probability (%) of dying between age and exact age from any of cardiovascular disease, cancer, diabetes, or chronic respiratory disease for liver disease: liver cirrhosis, age-standardized death rates ( +), per , population in equation we noted that contacts relative to normal (fc) regulates infection rates. this factor ranges between a minimum (min contact fraction; cmin) and as a function of the relative utility from normal activities (un) compared to utility from limited activities (ul) in light of additional risks associated with normal activity patterns: the utilities from normal and limited activities are specified as . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint overall, the risk perception and response sector includes the following five country-specific parameters that are estimated: cmin, τru, τrd, λ, and wr. table s summarizes the main equations discussed in appendix s , providing the mapping between full variable names and the short forms. it also includes all estimated model parameters, as well as those specified based on prior research. table s -mapping between full variable names and their short form for the subset of variables and parameters discussed in s . also included are equations explained above and sources for other variables. is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint the model we estimate is nonlinear and complex, and any estimation framework is unlikely to have clean analytical solutions or provable bounds on errors and biases. therefore, in designing our estimation procedure, we set out to be conservative. specifically: we use a likelihood function that accommodates overdispersion and autocorrelation (negative binomial); we utilize a hierarchical bayesian framework to couple parameter estimates across different countries which reduces the risk of over-fitting the data; and we use the conceptual definitions of parameters and their expected similarity across countries to inform the magnitude of that coupling across countries. compared to more common choices in similar estimation settings, these choices tend to widen the credible regions for our estimates and reduce the quality of the fit between model and data. in return, we think the results may be more reliable for projection, more informative about the underlying processes, and better reflective of uncertainties in such complex estimation settings. the model is a deterministic system of ordinary differential equations with a set of known and unknown parameters. the known parameters are those specified based on the existing literature and do not play an active role in estimation. the unknown parameters can be categorized into those that vary across different countries and those that are the same across all countries (i.e. general parameters). the estimation method is designed to identify both the most likely value and the credible regions for the unknown parameters, given the data on reported cases and deaths (and for a subset of countries, the excess deaths). this is done through a combination of estimating the most likely parameter values in a likelihood based framework, and using markov chain monte carlo simulations to quantify the uncertainties in parameters and projections. we first introduce the different components of the likelihood function we use: the fit to time series data, the random effects component coupling country-level parameters, and the penalty for excess mortality. then we explain the implementation details. define model calculations for expected reported cases and deaths for country i as μij(t) (with index j specifying cases and deaths) and the observed data for those variables as yij(t); the country-level vector of unknown parameters as and the general unknown parameters as ϕ. note that vector includes several parameters, each specifying an unknown model parameter, such as impact of treatment on fatality, or total asymptomatic fraction, for country i. the model can be summarized as a function f that produces predictions for expected cases and deaths for each country given the general and country-specific parameters: we use a negative binomial distribution to specify the likelihood of observing the y values given θ and ϕ. specifically, the logarithm of likelihood for observing the data series y given model predictions μ(θ ,ϕ) is: where (dropping time index for clarity): . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . summing the lt function over time provides the full (log) likelihood for the observed data given a parameterization of the model. the negative binomial likelihood function includes two parameters, μ and ε which determine the mean and the scaling/shape of the observed outcomes. the second parameter, ε, provides the flexibility needed fit outcomes with fat tails and auto-correlation. this parameter could itself be subject to search in the optimization process. specifically, we assume that: thus we create a (set of) country specific parameter(s) ( ) and two general parameters ( ) which should be estimated along with the conceptual model parameters. the country level scale ( ) implicitly assesses the reliability and inherent variability in country level reports, and the general ones inform the variability in case data vs. deaths. we augment the vectors ϕ and θ to include these scaling parameters as well. up to this point we have not included any relationship among country specific parameters, . this independence assumption would allow parameters representing the same underlying concept to vary widely across different countries. such treatment, by providing more flexibility, enhances the model's fit to historical data. however, it ignores the conceptual link that exists for a given parameter across countries, potentially allowing the model to fit the data for the wrong reasons (i.e. using parameter values that do not correspond to meaningful real world concepts). the result would likely be less reliable and also not robust for future projections. we therefore define a hierarchical bayesian framework to account for the potential dependencies among model parameters. specifically, we assume the same conceptual parameters (e.g. impact of treatment on fatality), across different countries, are coming from an underlying normal distribution with an unknown mean (to be estimated) and a pre-specified standard deviation. this assumption is similar to the use of "random effect" models common in regression frameworks, though we deviate from canonical random effect models by pre-specifying the standard deviation. in fact it is possible to estimate the standard deviation across countries as well (and to obtain better fits to data), but adding those degrees of freedom ignores qualitatively relevant insights about the level of coupling across different countries for each parameter, and thus results may fit the data better but for the wrong reasons. for example, some parameters, such as patient zero arrival time, could be very different across countries, whereas parameters reflecting innate properties of the sars-cov- virus itself (e.g. total asymptomatic fraction (a)) or those determining fatality (e.g. base fatality rate for unit acuity (fb)) should be very similar across different countries. allowing the model to determine the variance for the latter will lead to better fits: the model can find baseline fatality rates that easily match fatality variations across countries, and would expand the corresponding variance parameter accordingly. however, as a result the estimation algorithm will have too easy a job: it will not require a precise balancing between hospitalization, impact of acuity on fatality, and post-mortem testing decisions to fit fatality data. thus, the estimates may well be less informative, or further from true underlying processes and the general characteristics of the disease which we care about. the implementation of this random effect introduces another element to the overall likelihood function: here θik represents the k th parameter for country i, and ̅ is the (estimated) average across countries for the k th parameter. σk is the pre-specified allowable variability for the k th parameter across different countries. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint specifying these standard deviations adds a subjective element to the estimation process. however, we note that subjective elements are ultimately indispensable in any modeling activity: from specifying the model boundary to the level of aggregation, use of various functional forms, and choice of likelihood functions, these choices are built on subjective assessments that experts bring to a modeling project. absent our conceptually informed variability factors, we would need to make the assumption that country-level parameters are independent, or that our complex estimation process would correctly identify the true dependencies among those parameters. we think both those alternatives are inferior in the chosen method. so here we focus on transparently documenting and explaining those assumptions. table s summarizes the estimated model parameters, their estimated values (mean across countries and mean of inter-quartile range) and the assumed variability factor (σk) for each. sw sensitivity to weather . (global parameter so the other metrics do not apply) *given the wide range and potential long tail for these parameters the log transformation is used in specifying the dispersion penalty (equation ) and variability factors are reported as σ , where σ is used in equation . ** these parameters are expected to be less variable across countries and thus are assigned small variability allowances compared to their mean. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint finally, we include a likelihood-based penalty term to allow model predictions be informed by excess mortality data collected by various news agencies and researchers for a subset of countries in our sample. these data provide snapshots of excess mortality (compared to a historical baseline) for a window of time in each country. subtracting from total excess mortality the covid- deaths officially recorded in that window offers a data point for excess mortality not accounted for in official data (ei). we can calculate in the model the counter-part for this construct: the simulated mortality that is not included in the simulated reported covid- deaths ( ̅ ). there is uncertainty in these excess mortality data: the historical baselines used by various sources do not adjust for demographic change, excess mortality may be due to factors other than covid- , and some of it may be due to changes in healthcare availability and utilization motivated by covid- but not directly attributable to the disease (for example when surgeries are delayed, hospitalization is avoided, or heart conditions are ignored). excess mortality may also be reduced due to reduced traffic accidents (in light of physical distancing policies) and pollution related deaths. given these uncertainties, we use the following penalty function to keep the simulated unaccounted excess mortality close to data: this penalty could be seen as a likelihood coming from the probability distribution ( ) = should be attributed to covid- deaths, but that there is significant uncertainty around this, so some % variation across this figure is quite plausible ( %- % of data). however, numbers outside of this range start to impose increasingly large penalties, so that very large deviation becomes unlikely. combining these three components, we obtain the full likelihood function used in the analysis: for each country we include the lt component from the first day they have reached . % of their cumulative cases to-date, or a minimum of cumulative cases. this excludes very early rates that are both unreliable and which, given very small estimated model predictions for infection, could lead to unreasonably large likelihood contributions. the model includes a large number of parameters to be estimated: a general parameter for the impact of weather, general parameters for , and parameters for each country that are coupled together based on the random effects framework described above. out of those parameter is for and the rest are informing various features of disease transmission, testing, hospitalization, and risk perception and response. with a sample of countries, this would lead to about parameters to be estimated. a direct optimization approach to this problem suffers from potential risk of getting stuck in local optima, and direct use of mcmc methods to find the promising region of parameter space suffers from the curse of dimensionality. we therefore designed the following -step procedure to find more reliable solutions to both problems. ) we estimate the model with the full parameter vector for a smaller number of countries with larger outbreaks ( - countries). we use the powell direction search method implemented in vensim™ simulation software for this step. the method is a local search approach though it has features that allows it to escape local optima in some cases. we restart the optimization from various random points in the feasible parameter space and track the convergence of those restarts to unique local peaks. we stop this process when we are repeatedly landing on the same local peaks in the parameter . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint space. this procedure showed that local peaks do exist, but they are not many; for example within restarts we may find - distinct peaks. this provides a coherent set of starting points for ϕ and ̅ for next steps. ) we go through iterations of the following two steps: a) conduct country-specific optimizations with restarts to find the vector of θi given the ϕ and ̅ from first optimization or from the step b. b) conduct a global optimization, including all countries but fixing θi and optimizing on ϕ (and ̅ ; though that is simply the mean across country level parameters from previous round). we stop when iterations offer little improvement from one round to the next (less than . % improvement in loglikelihood). ) we conduct a full optimization allowing all parameters (θi, ϕ and ̅ ) to change, starting from the point found in the last iteration of step . this step finds the exact peak on the likelihood landscape which is the best-fitting parameter set for the model. ) for the mcmc, theoretically one should conduct the sampling from all model parameters in the full model. however, our experiments showed that the large dimensionality of the parameter space requires an infeasible number of samples to achieve adequate mixing and ensure reliable credible regions for parameters and projections. to overcome this challenge we note that the parameters of different countries are connected to each other only through ϕ and ̅ , and these general parameters are rather insensitive to dynamics in each country. the insensitivity is due to the fact that a single country only contributes about % to the general parameters' values, and within a typical mcmc the country-level parameters often can't change more than % before the resulting samples become highly unlikely. therefore, one can conduct an approximate country-level mcmc by fixing the general parameters at those from step , and only sampling from the θi for each country. the mcmc algorithm used is one designed for exploring high dimensional parameter spaces using differential evolution and self-adaptive randomized subspace sampling ( ) . using this method we obtain good mixing and stable outcomes (robin-brooks-gelman psfr convergence statistic remaining under . ) after about , samples (the burn-in period). we continue the mcmc for each country for million samples and then randomly take a subsample of those points after the burn-in period for the next step. ) the resulting subsamples for different countries from step are assembled together to create a final sample of parameters for the full model to conduct projections and sensitivity analysis at the global scale. uncertainties in the handful of global parameters is not identified in this procedure, but can be quantified by assessing the sensitivity of the global likelihood surface to changes in those parameters. the process above is automated using a python script that controls the simulation software (vensim). we conduct the analysis using a parallel computing feature of vensim on a windows server with cores. after compiling the simulation model into c++ code (which speeds up calculations significantly), and using a simulation time step of . days, it takes about hours to complete the estimation for countries. full analysis code is available online at https://github.com/tseyanglim/covidglobal. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint s -data pre-processing getting contemporaneous, comprehensive, national-level data on covid- is a challenge. the most widely-cited data aggregators, such as the johns hopkins center for systems science and engineering's covid- database ( ) , ourworldindata portal ( ), and the us-focused covid tracking project ( ), get their data from the same few official sources, such as the us centers for disease control and prevention (cdc), the european cdc (ecdc), and the world health organization (who). these official agencies in turn get their data from national and subnational public health authorities, which ultimately rely on reports from hospitals, clinics, and private and public health labs. as a result, idiosyncrasies in the ground-level data collection processes permeate virtually all sources of aggregate data. most notably, data collection involves time lags, which can differ from source to source. daily death counts could reflect the date of actual death or the date a death is registered or reported; different uk government sources, for instance, use each of these metrics. daily infection or case counts could include the total new cases reported on a given date, or the total cases confirmed from that date; the latter would result in some 'backfill' whereby case counts for previous days can continue to increase for some time as delayed confirmations come in. daily counts of tests conducted could report samples collected, samples processed, results reported, or a mix of these; the us cdc, for instance, reports a mix of testing by date of sample collection and date of sample delivery to the cdc. aside from differences in unit of measure (people vs. tests vs. samples), there may be different time lags involved as well. in addition to these idiosyncrasies, testing data in particular is also patchy for many countries, even as testing has become more widespread. the who does not report country-by-country testing, nor does the jhu covid map outside the us. furthermore, there are sometimes irregular delays in the reporting of test results, which can create occasional unexpected spikes in reported numbers of tests, infections, or both. depending on the specifics of how daily infection and test counts are reported, there can in some cases be a disjunction between the two. because confirmed case counts largely depend on positive test results, test and infection counts should be correlated -ceteris paribus, a day with a lot of samples collected for testing should see more confirmed cases attributed to it, while a day with no sample collection should see no cases. but since cases may not be reported by the date of the test, and tests may not be reported by the date of sample collection, officially reported numbers can get out of sync in either direction. this problem is most salient when there are clear weekly cycles in daily rates. in most of the world, particularly western countries, daily test rates are far lower on weekends than during the week. as a result, infection numbers show a clear weekly cyclical component as well. but the weekly cycles in testing and infection numbers for a given country do not always line up. our model explicitly accounts for the effect of testing on reported infections, but we do not explicitly model the country-level idiosyncrasies of reporting and how they vary between test data and infections. instead we account for any such lags in pre-processing of the data to align testing and case data. the weekly cycle occurs in many countries' death rate data as well, where it presents a different problem. a weekly cycle in testing is a behaviourally realistic part of the data-generation process, as many labs, clinics, or other testing sites for instance may be closed on weekends. as testing provides the window on the state of confirmed infections, a comparable cycle in confirmed cases is to be expected as well. by linking case confirmations to testing, our model explicitly accounts for this limited visibility on the true state of the epidemic. however, a weekly cycle in death rates almost certainly reflects different limitations of the data- https://blog.ons.gov.uk/ / / /counting-deaths-involving-the-coronavirus-covid- / https://www.cdc.gov/coronavirus/ -ncov/cases-updates/previous-testing-in-us.html see e.g. https://www.wcvb.com/article/massachusetts-coronavirus-reporting-delay-due-to-quest-lab-itglitch/ # . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint generation process, typically to do with hospital staffing, which we do not explicitly model. as such we need to address any weekly cycle in death rates through data pre-processing as well. to deal with these challenges, we developed a multi-step algorithm to pre-process our data before feeding it into the model for calibration. the algorithm is described below. it was implemented in python, largely using the pandas and numpy packages, and the code is available in full at: https://github.com/tseyanglim/covidglobal. the algorithm proceeds country-by-country, following these steps on each country. ) examine daily cumulative test data; if data are insufficient ( or fewer data points), drop country from the dataset. ) interpolate any missing daily cumulative test data points using a piecewise cubic hermite interpolating polynomial (pchip) spline. if the first reported infection is before the first reported cumulative test, also extrapolate cumulative tests back to the date of first reported infection. a. extrapolation to the date of first reported infection is necessary since both in the model and, to a large extent, in reality, reported infections require testing for confirmation. b. pchip spline interpolation yields a continuous monotonic function with a continuous first derivative, thus avoiding generating any anomalous rapid change in daily test rate. c. we used the implementation of pchip interpolation from the widely used scipy package for python. ) calculate daily test rate as daily cumulative tests less the preceding day's cumulative test total: ) examine the original daily cumulative test data to estimate how much of the calculated daily test rate is based on interpolated vs. original data. a. daily test rates calculated based on mostly original data should be expected to include any weekly cycles or occasional irregularities that would also be reflected in daily infection counts. conversely, daily test rates calculated from cumulative test counts that are largely interpolated would not be expected to fully reproduce any such cycles or irregularities, since the interpolation produces a relatively smooth function. b. as a rule of thumb, we examine the cumulative test data for the second half of the time from the first test to the latest test. if fewer than half the days in that window have original cumulative test data, we consider the test data to be 'sparse', requiring further processing. it may be argued that there are weekly cycles in large-scale human behaviour that may drive some true weekly cyclicality in the true rates of infection and death, and as such it may be wrong to consider such cycles to be artefacts of the datageneration process. however, we find this unlikely for a few reasons. first, weekly cycles in human interactions, largely driven by the work and school week and weekend, will have been significantly attenuated by widespread adoption of social distancing measures around the world. second and more importantly, variation in incubation period and time before development of symptoms means that any true cyclicality in the timing of initial infection will be further attenuated in the timing of symptom development. by the same logic, wide variability in the delay from symptom development to death means there should be minimal cyclicality, if any, in the timing of deaths, meaning any such cycles visible in the data are due to measurement and reporting lags. https://docs.scipy.org/doc/scipy/reference/generated/scipy.interpolate.pchipinterpolator.html . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint ) if the test data are not sparse, account for any potential lag or other reporting delay differences between daily test rate and daily infection rate using a time-shift algorithm to estimate any such lags or delays from the data and shift the test rate time series accordingly. the time-shift algorithm ensures that any weekly cycles present in the daily infection rate data are reflected in the daily test rate data and aligned as best as possible on date, thereby accounting for the fact that model-generated reported infections depends on testing but with no time lag between test and result. a. first, identify the weekly component of the time series of daily infection rate and daily test rate using a seasonal-trend decomposition based on loess (stl) procedure. i. stl deconstructs time series data into several components, including a trend and a seasonal component over a specified period (weekly, in this case) as well as a residual. stl is an additive decomposition, and has the advantage of allowing the seasonal component to change over time (rather than being a fixed pattern repeated exactly across the whole time series). ii. we used the stl implementation from the statsmodels package for python. b. shift the time series over a one-week range (from - to + days of lag between test and infection reporting), calculating the cross-correlation between the weekly seasonal component of the daily infection rate data and the daily test rate data for each time shift. c. identify the time shift within this range that maximizes the cross-correlation between the infection rate and test rate data, and shift the test rate data accordingly. ) if the test data are sparse, the spline interpolation will generally cut out some of any weekly cyclicality that may be present. visual inspection of daily test rates for countries with sparse test data also shows large, irregular spikes in reported tests are not uncommon, without necessarily having concomitant irregular spikes in reported daily infection rates. as such, rather than attempting to eliminate differences in reporting lags through the time-shift algorithm described above, we instead apply a data-smoothing algorithm to both daily test rate and daily infection rate, in order to reduce any cyclicality and irregular spikes. this smoothing allows the calibration of the main model to focus on matching the underlying trends in the data. in all cases, whether daily cumulative test data are sparse or not and whether infection and test rate data are smoothed or not, since weekly cycles in death data are reflective of reporting lags not captured in the model, daily death rate data is smoothed using the same algorithm. ) the smoothing algorithm used is designed first to conserve the total number of reported cases (tests, infections, or deaths), and second to preserve some degree of variation in the time series, as some noise may be informative and retaining some is important to the calibration of the model. a. starting from when the time series of daily rate (test or infection) exceeds a specified minimum value ( /day), calculate the rolling mean of the daily rate, using a centred moving window of days. b. calculate the residual between each day's data point and the rolling mean for that day, and divide by the square root of the rolling mean, to get an adjusted deviation value: i. dividing by the square root of the rolling mean reflects a heuristic assumption that each daily rate (of infections, deaths, or tests) behaves as a poisson process (stdev of pois() =  . ). is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint ii. the functional result of this adjustment is that both absolute and relative magnitudes of deviations from the rolling mean are given some weight -large relative deviations when absolute values are small (and data are noisier) are not ignored, but neither do they outweigh larger absolute (but smaller relative) deviations that occur when the mean is large, which is important since most of the time series data are growing significantly over the time horizon of the model. c. calculate thresholds for identifying dips and peaks in the data based on the median of the adjusted deviations, ± one median absolute deviation (mad) of the adjusted deviations: i. using the median absolute deviation to determine thresholds for peaks and dips is robust to outliers in the deviations, which do arise occasionally in the data. ii. a threshold width of one mad is relatively narrow for outlier detection, but by inspection of the data, is about right for identifying most of the peaks and dips caused by weekly cycles in test, infection, and death rates, as well as larger outliers. d. once thresholds are calculated, iterate through the data points in the time series first forward in time from oldest to newest, filling in any 'dips' (data points with adjusted deviations below the lower threshold), then backward in time from newest to oldest, smoothing out any 'peaks' (data points with adjusted deviations above the upper threshold) that remain. repeat the process until all data points' adjusted deviations are within the originally calculated thresholds for the time series. i. we infer that the underlying processes generating dips and peaks are somewhat different. dips are generally the result of weekly cycles in the data, e.g. lower rates of testing or longer lags in death reporting that occur on weekends. peaks arise to some extent due to the same weekly processes, e.g. some deaths that occur on weekends only being recorded at the start of the next week. however, some peaks, especially larger ones, may result from irregular random delays in reporting, such as large batches of tests being held up due to logistical issues and then getting processed all at once. as such the smoothing procedure for dips vs. peaks is slightly different. e. the dip-filling step fills a fraction of each dip (specified as a smoothing factor) by redistributing data counts based on a multinomial draw from the subsequent few days following each dip. i. first, calculate the amount to fill based on the deviation and the smoothing factor specified, in this case . : ii. calculate the amount redistributed from each of the following few ( ) days + , + , … + , = , based on a multinomial distribution as follows: where + , + , … + are calculated as: iii. this formulation allows some redistribution from any of the subsequent few days whose adjusted deviations exceed the focal day's adjusted deviation, but with more redistribution from days with higher adjusted deviations. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint f. the peak-smoothing step similarly redistributes a fraction of each peak, specified by the smoothing factor, to the preceding several days based on another multinomial draw. i. first, calculate the amount to redistribute similarly to the dip-filling step: ii. calculate the amount redistributed to each of the preceding several ( ) iii. this formulation redistributes peaks to preceding days based on the calculated rolling mean counts of those days, on the assumption that the irregular delays that generate random spikes in counts are essentially random and equally likely to affect any given unit of data over a several-day span. as such, the probability that a unit showing up in a spike due to such delays comes from a given preceding day is simply proportional to the expected count for that day, as approximated by the rolling mean. g. by filling dips first before smoothing peaks, the combined algorithm largely addresses any peaks that are due primarily to weekly cycles during the dip-filling stage, such that remaining peaks that get smoothed tend to be the larger, irregular ones. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint table s reports two quality of fit metrics for different countries and different time series. the first four columns report mean absolute error normalized by mean (maen) and the last four report the r-squared measures. errors for cumulative infection and deaths are followed by those for the new cases and deaths (flow variables). . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint figure s shows the visualization of fit between data and simulations for all the countries in our sample. these graphs include data and model outputs for reported new cases (blue; left axis in thousands per day) and deaths (red; right axis in thousands per day) starting from the beginning of the epidemic in each country until june . figure s -comparison of data and simulation. new cases in blue (left axis, in thousands per day) and new deaths (red, right axis). . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint estimates for true cumulative cases (blue; left axis in millions) and deaths (red; right axis in thousands) across different countries up to june are reported in figure s . figure s shows the ratio of estimated excess deaths, i.e. covid- fatalities not reported as such, to reported excess deaths, i.e. deaths over historical baseline not accounted for by reported covid- deaths, for the countries for which such data are available. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint figure s -ratio of estimated excess deaths to reported excess deaths. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint figure s shows the initial reproduction number (re) occurring in each country. reproduction numbers are changing dynamically and transient dynamics may lead to larger than equilibrium numbers if maximum re values were used. we therefore use the th percentile of simulated reproduction number in this graph. also note the large credible intervals for these estimates, partly coming from changes in what is included in the th percentile (sometimes it includes the highest values and sometimes it does not), as well as the inherent uncertainty when both reproduction number and behavioral and policy responses are estimated: one can have smaller initial re and smaller response functions, or larger values for both, and stay consistent with the data. figure s -maximum reproduction number re for each country's outbreak . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint parameter estimates figure s reports most likely estimates for the vector of country-specific parameters (θi). the figure also includes % credible intervals for these parameter estimates. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint figure s -parameter estimates and % credible regions for country-specific parameters. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint figure s reports country-level projections for new cases and deaths based on the following assumptions (consistent with those reported in figure -d in the main paper). we shift both testing and contact sensitivity to perceived risks to arguably more realistic values between current ones that capture some improvements but not a completely different behavior. specifically, on july st we shift test rate to a value % between the current rates and . % per day (which increases testing in most countries and reduces it in a few). similarly, sensitivity of contacts to perceived risk is shifted to % of the way between estimated country-level values and a high value of that is. figure s -country level projections until spring . daily cases (in thousands, blue, left axis) and daily deaths (red, right axis) are graphed. . cc-by-nd . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted june , . . https://doi.org/ . / . . . doi: medrxiv preprint spread of sars-cov- in the icelandic population clinical characteristics and imaging manifestations of the novel coronavirus disease (covid- ): a multi-center study in wenzhou city detection of sars-cov- in different types of clinical specimens sensitivity of chest ct for covid- : comparison to rt-pcr stability issues of rt-pcr testing of sars-cov- for hospitalized patients clinically diagnosed with covid- weather conditions and covid- transmission: estimates and projections. medrxiv clinical characteristics of coronavirus disease in china characteristics of and important lessons from the coronavirus disease (covid- ) outbreak in china: summary of a report of cases from the chinese center for disease control and prevention covid- exacerbating inequalities in the us suppression of covid- outbreak in the municipality of vo estimating the asymptomatic proportion of coronavirus disease (covid- ) cases on board the diamond princess cruise ship universal screening for sars-cov- in women admitted for delivery incubation period and other epidemiological characteristics of novel coronavirus infections with right truncation: a statistical analysis of publicly available case data modeling the rework cycle: capturing multiple defects per task data analysis using regression and multilevel/hierarchical models accelerating markov chain monte carlo simulation by differential evolution with self-adaptive randomized subspace sampling most equations are self-explanatory. the "[…]" notation is used to subscript variables over a set of members. for example, the subscript "rgn" is used to identify different countries. therefore [rgn] indicates that a variable is defined separately for each member of the set "rgn". other subscript ranges used in the equations are: expnt: used for numerically solving the probability of missing symptoms equation. pdim: used for setting policy levels for a few variables. priors: used for implementing the random effects estimation components. each estimated parameter is mapped into an element of this subscript to simplify vector-based calculations tststs: the test status including those confirmed ('tested') and those unconfirmed complete equations and units potential test demand from susceptible population[rgn] , positive candidates interested in testing poisson subset adj active test rate[rgn] = if then else ( time < new testing time , datatestrate[rgn] , external test rate initial( inputave[priorendoave] * ( -sw endoave ) + sw endoave * calcave activities allowed by government additional asymptomatic post detection[rgn] = weighted infected post detection gate[rgn] * additional asymptomatic relative to symptomatic[rgn] / ( + additional asymptomatic relative to symptomatic additional asymptomatic relative to symptomatic[rgn] = zidz ( total asymptomatic fraction net[rgn] -exp ( -covid acuity[rgn] ) , -total asymptomatic fraction net all recovery[rgn] = recovery of confirmed[rgn] + recovery of untested[rgn] + sum ( hospital discharges expected positive poisson covid patients[rgn] + sqrt ( expected positive poisson covid patients[rgn] ^ + * effective hospital capacity[rgn] * effective hospital capacity[rgn] ) ) / ( * effective hospital capacity allocated fration noncovid hospitalized[rgn] = smoothi ( allocated fraction covid hospitalized = min ( maxalp , ialp * alpr = min ( , talp * alpr[rgn] ) units: dmnl ) alpr area of region[rgn] = get vdf constants(constant data file average acuity of positively tested[rgn] * xidz ( ( -probability of missing acuity signal at hospitals[rgn,tested] * fraction poisson not hospitalized[rgn,tested] ^ ) , -fraction poisson not hospitalized average acuity of untested poisson subset[rgn] * ( -probability of missing acuity signal at hospitals[rgn,notest] * fraction poisson not hospitalized[rgn,notest] ^ ) , -fraction poisson not hospitalized average acuity not hospitalized[rgn,notest] = zidz ( average acuity not hospitalized poisson[rgn,notest] * infectious not tested or in hospitals poisson average acuity not hospitalized[rgn,tested] = average acuity not hospitalized poisson average acuity not hospitalized poisson[rgn,tested] = max ( , probability of missing acuity signal at hospitals[rgn,tested] * average acuity of positively tested[rgn] * fraction poisson not hospitalized probability of missing acuity signal at hospitals[rgn,notest] * average acuity of untested poisson subset[rgn] * fraction poisson not hospitalized -prob missing symptom[rgn] * fraction interested not tested[rgn] ^ ) , -fraction interested not tested[rgn] , * prob missing symptom average acuity of untested poisson subset[rgn] = zidz ( poisson subset reaching test gate[rgn] * covid acuity[rgn] -positive tests of infected[rgn] * average acuity of positively tested[rgn] , poisson subset not tested passing gate positive candidates interested in testing poisson subset adj[rgn] -potential test demand from susceptible population[rgn] , positive candidates interested in testing poisson subset adj base fatality rate for unit acuity base fatality rate for unit acuity net[rgn] = initial( base fatality rate for unit acuity baseerror = units: person baseline daily fraction susceptible seeking tests demographic impact on fatality relative to china[rgn] * base fatality rate for unit acuity net baseline risk of transmission by asymptomatic[rgn] = initial( baseline transmission multiplier for untested symptomatic * multiplier transmission risk for asymptomatic net baseline transmission multiplier for untested symptomatic = units: dmnl bed per square kilometer[rgn] = initial( nominal hospital capacity[rgn] / area of region beds per thousand population[rgn] = get vdf constants(constant data file zidz ( lnymix deaths of symptomatic untested[rgn] -post mortem test rate[rgn] * frac post mortem from untreated chronic death rate[rgn] = get vdf constants(constant data file chronic impact on fatality[rgn] = initial cml death frac in hosp[rgn] = xidz ( cumulative deaths at hospital cml death fraction in hospitals large enough = sum ( if then else ( cml death frac in hosp cml known death frac hosp[rgn] = xidz ( cumulative deaths at hospital cmlterrpw = units: dmnl cmltpenaltyscl = units: dmnl constant data file :is: 'covidmodelinputs -constantdata contacts relative to normal[rgn] = min ( voluntary reduction in contacts[rgn] , activities allowed by government reaching testing gate[rgn] -symptomatic infected to testing[rgn] -untested symptomatic infected to hospital excess death start count[rgn] = :na:, , if then else ( time >= excess death start count flu acuity * covid acuity relative to flu net covid acuity relative to flu net[rgn] = initial( covid acuity relative to flu total covid hospitalized[rgn] , infectious not tested or in hospitals poisson[rgn] + infectious confirmed not hospitalized[rgn] + total covid hospitalized cumulative cases[rgn] = integ( new cases[rgn] , ) units: person cumulative confirmed cases[rgn] = infectious confirmed not hospitalized[rgn] + hospitalized infectious[rgn,tested] + cumulative deaths of confirmed confirmed recovered[rgn] + cumulative recovered at hospitals cumulative death fraction cumulative deaths at hospital[rgn,tststs cumulative deaths of confirmed[rgn] = cumulative deaths at hospital[rgn,tested] + cumulative deaths of confirmed untreated cumulative deaths of confirmed untreated[rgn] = integ( deaths of confirmed cumulative deaths untested untreated cumulative fraction total cases hospitalized[rgn] = zidz ( sum ( cumulative deaths at hospital[rgn,tststs!] + cumulative recovered at hospitals[rgn,tststs!] + hospitalized infectious[rgn,tststs!] ) , cumulative cases cumulative missed death[rgn] = integ( count missed death[rgn] , ) units: person cumulative negative tests cumulative recovered at hospitals[rgn,tststs] = integ( hospital discharges cumulative recoveries[rgn] = integ( all recovery cumulative tests conducted cumulative tests data current test rate per capita dalp = . units: dmnl data excess deaths[rgn] = get vdf constants(constant data file , 'dataconstants[rgn]', ) units: person data pseudo case fatality raw: := datacmltinfection raw: := datacmltdeath raw: units: person/day raw: := dataflowinfection raw: := dataflowdeath raw: := datatestrate raw: units: person/day cumulative confirmed cases new testing time ) ) ) units: person/day datalimitfromtime = if then else ( time > max time data used , , ) units: dmnl = units: day/year death rate[rgn] = deaths of confirmed[rgn] + deaths of symptomatic untested[rgn] + sum ( hospitalized infectious deaths tested untreated resolution[rgn] * fatality rate untreated post-detection phase resolution time" * fatality rate untreated delay order = units: dmnl demographic impact on fatality relative to china[rgn] = get vdf constants(constant data file discount rate annual = . units: /year discount rate per day = initial( discount rate annual / days per year ) units: /day dread factor in risk perception dread factor in risk perception net[rgn] = if then else ( response policy time on < time , dread factor policy * response policy weight[dfcp] + ( -response policy weight[dfcp] ) * dread factor in risk perception[rgn] , dread factor in risk perception dread factor policy = units: dmnl effective hospital capacity[rgn] = nominal hospital capacity[rgn] * normalized hospital density[rgn] ^ impact of population density on hospital availability excess death end count[rgn] = get vdf constants(constant data file excess death mean frac = . units: dmnl excess death range frac = . units: dmnl zidz ( cumulative missed death[rgn] -excess death mean frac * data excess deaths[rgn] , excess death range frac * data excess deaths excess death start count[rgn] = get vdf constants(constant data file expected positive poisson covid patients[rgn] = sum ( potential hospital demand[rgn,tststs!] ) * "post-detection phase resolution time" units: person ) expnt external test rate[rgn] = population[rgn] * policy test rate utility from normal activities[rgn] / ( utility from limited activities[rgn] + utility from normal activities units: dmnl baseline fatality multiplier[rgn] * impact of treatment on fatality rate[rgn] * average acuity hospitalized baseline fatality multiplier[rgn] * average acuity not hospitalized poisson[rgn,tststs] ^ sensitivity of fatality rate to acuity net final test rate per capita[rgn] = initial( current test rate per capita[rgn] + weight max in test goal * ( max test rate per capita -current test rate per capita final time = units: day flu acuity relative to covid deaths of symptomatic untested fraction covid death in hospitals previously tested[rgn] = zidz ( hospitalized infectious deaths fraction covid hospitalized positively tested[rgn] = zidz ( hospitalized infectious fraction interested not tested[rgn] = -zidz ( total test on covid patients[rgn] , positive candidates interested in testing poisson subset fraction interseted not correctly tested[rgn] = -( -fraction interested not additional asymptomatic relative to symptomatic[rgn] / ( + additional asymptomatic relative to symptomatic fraction of fatalities screened post mortem[rgn] = indicated fraction post mortem testing[rgn] * switch for government response fraction of population hospitalized for covid[rgn] = total covid hospitalized fraction poisson not hospitalized[rgn,tested] = exp ( -average acuity of positively tested[rgn] * ( -probability of missing acuity signal at hospitals fraction poisson not hospitalized[rgn,notest] = exp ( -average acuity of untested poisson subset[rgn] * ( -probability of missing acuity signal at hospitals fraction seeking test fraction tests positive[rgn] = zidz ( positive tests of fraction tests positive data[rgn] = min ( , zidz ( dataflowinfection[rgn] , active test rate[rgn] ) ) units: dmnl ) fractional value of limited activities = . units: dmnl global cases = sum ( cumulative cases[rgn!] ) units: person global deaths = sum ( cumulative deaths[rgn!] ) units: person global deaths , global cases ) units: dmnl government response start time hazard of symptomatic infection[rgn] = infection rate[rgn] / susceptible[rgn] * ( -total asymptomatic fraction net herd immunity fraction = . units: dmnl hospital admission infectious[rgn,tststs] = hospital admits all hospital admit ratio[rgn,tststs] = xidz ( hospital admits all hospital admits all[rgn,tested] = hospital demand from tested[rgn] * allocated fraction covid hospitalized hospital admits all[rgn,notest] = hospital demand from not tested[rgn] * allocated fraction covid hospitalized poisson subset not tested passing gate[rgn] * ( -exp ( -average acuity of untested poisson subset[rgn] * ( -pmas unconfirmed for hospital demand positive tests of infected[rgn] * ( -exp ( -average acuity of positively tested[rgn] * ( -pmas confirmed for hospital demand = ( -fatality rate treated[rgn,tststs] ) * hospital outflow covid positive hospital outflow covid positive[rgn,tststs] = hospitalized infectious[rgn,tststs] / hospitalized resolution time units cumulative deaths at hospital hospitalized infectious deaths hospitalized infectious deaths[rgn,tststs] = fatality rate treated[rgn,tststs] * hospital outflow covid positive hospitalized resolution time = units: day zidz ( sum ( hospitalized infectious deaths[rgn,tststs!] ) , sum ( hospital outflow covid positive zidz ( sum ( cumulative deaths at hospital[rgn,tststs!] ) , sum ( cumulative deaths at hospital[rgn,tststs!] + cumulative recovered at hospitals units: day flu acuity * ( prob missing symptom indicated fraction post mortem testing[rgn] = fraction covid death in hospitals previously tested[rgn] ^ sensitivity post mortem testing to capacity indicated risk of life loss[rgn] = min ( , switch for government response[rgn] * perceived hazard of infection[rgn] * dread factor in risk perception net[rgn] * pseudocfr / discount rate per day ) units: dmnl symptomatic infected to testing[rgn] -untested symptomatic infected to hospital[rgn] , ) units: person ) transmission multiplier presymptomatic[rgn] + infected pre detection[rgn] * transmission multiplier pre detection[rgn] + ( additional asymptomatic post detection[rgn] + "poisson not-tested asymptomatic transmission multiplier for confirmed[rgn] + sum ( hospitalized infectious[rgn,tststs!] * transmission multiplier for hospitalized[rgn,tststs!] ) ) * reference force of infection infectious not tested or in hospitals poisson constant data file , 'dataconstants[rgn]', ) units: person ) initial time = units known death fraction in hospitals large enough = sum ( if then else ( cml known death frac hosp[rgn!] < minhspdtreshadv , , ) * advcntrs liver disease impact on fatality[rgn] = initial( ( liver disease rate liver disease rate[rgn] = get vdf constants(constant data file , 'dataconstants[rgn]', ) units: dmnl max test rate per capita = . units: /day max time data used = units: day ) maxalp = units: dmnl units: person maxrtresh = units: dmnl constant data file , 'meanchronic', ) units: dmnl constant data file , 'meanliver', ) units: dmnl constant data file , 'meanobesity', ) units: dmnl min contact fraction minadjt = units: day multiplier recent infections to test multiplier transmission risk for asymptomatic multiplier transmission risk for asymptomatic net[rgn] = initial( multiplier transmission risk for asymptomatic + alp negative test results new cases[rgn] = infection rate new testing time = units: day nominal hospital capacity[rgn] = initial( initial population[rgn] * beds per thousand population[rgn] / ) units: person normalized hospital density[rgn] = initial( bed per square kilometer[rgn] / reference hospital density ) units: dmnl not too few susceptibles = sum ( if then else ( suscfrac[rgn!] < minsusctresh , , ) ) units: dmnl sens obesity impact net infection rate[rgn] , incubation period , , delay order ) units: person/day ) onset to detection delay = units overall death fraction[rgn] = zidz ( death rate[rgn] , all recovery[rgn] ) units: dmnl patient zero arrival[rgn] = if then else ( time < patient zero arrival time patient zero arrival time units: person ) payoff = units: dmnl ) pdim : tstp,dfcp,dgtp,scup perceived hazard of infection[rgn] = ( weight on reported probability of infection[rgn] * reported hazard of infection[rgn] + ( -weight on reported probability of infection[rgn] ) * hazard of symptomatic infection indicated risk of life loss[rgn] -perceived risk of life loss[rgn] ) / if then else ( indicated risk of life loss[rgn] > perceived risk of life loss pmas confirmed for hospital demand[rgn] = ( -reference covid hospitalization fraction confirmed[rgn] ) ^ ( / average acuity of positively tested pmas confirmed for hospital demand[rgn] + ( -pmas confirmed for hospital demand[rgn] ) * untested pmas gap with tested infectious not tested or in hospitals poisson[rgn] * exp ( -average acuity not hospitalized poisson poisson subset not tested passing gate[rgn] = poisson subset reaching test gate[rgn] -positive tests of infected poisson subset reaching test gate[rgn] = reaching testing gate[rgn] / ( + additional asymptomatic relative to symptomatic policy test rate[rgn] = if then else ( time < new testing time , current test rate per capita[rgn] , final test rate per capita infected unconfirmed post-detection"[rgn] + susceptible[rgn] + recovered unconfirmed[rgn] + confirmed recovered[rgn] + infectious confirmed not hospitalized[rgn] + "pre-symptomatic infected positive candidates interested in testing poisson subset[rgn] = poisson subset reaching test gate[rgn] * fraction seeking test positive candidates interested in testing poisson subset adj[rgn] = max ( . * potential test demand from susceptible population[rgn] , positive candidates interested in testing poisson subset positive testing of infected untreated[rgn] = positive tests of infected[rgn] * fraction poisson not hospitalized positive tests of infected[rgn] = positive candidates interested in testing poisson subset[rgn] * ( -fraction interseted not correctly tested post mortem test rate[rgn] = post mortem tests total[rgn] * sensitivity of covid test units post mortem test untreated[rgn] = post mortem test rate[rgn] * frac post mortem from untreated deaths of symptomatic untested[rgn] + hospitalized infectious deaths[rgn,notest] ) * fraction of fatalities screened post mortem post mortem tests total[rgn] = min ( post mortem testing need[rgn] , active test rate[rgn] ) units: person/day hospitalized infectious[rgn,notest] / minadjt , post mortem test rate[rgn] * ( -frac post mortem from potential hospital demand[rgn,notest] = hospital demand from not tested potential hospital demand[rgn,tested] = hospital demand from recovered unconfirmed[rgn] + cumulative recovered at hospitals[rgn,notest] ) * ( baseline daily fraction susceptible seeking tests[rgn] * fraction seeking test[rgn] + multiplier recent infections to test infection rate[rgn] + patient zero arrival[rgn] -onset of symptoms[rgn] , ) units: person ) priorendoave : upadj mtrasym ) priors : upadj prob missing symptom probability of missing acuity signal at hospitals[rgn,tested] = zidz ( ln average acuity of untested poisson subset[rgn] ) + units: dmnl ) pseudocfr units: dmnl effective reproduction rate[rgn] = zidz ( infection rate[rgn] , total weighted infected population reaching testing gate realistic r = sum ( if then else ( r effective reproduction rate recovery of untested[rgn] , ) units: person ) recovery of confirmed[rgn] = tested untreated resolution[rgn] * ( -fatality rate untreated post-detection phase resolution time" ) -deaths of symptomatic untested reference covid hospitalization fraction confirmed units: person/(km*km) regionalinputs[rfi,rgn] = reference force of infection sensitivity post mortem testing to capacity baseline daily fraction susceptible seeking tests weight on reported probability of infection multiplier recent infections to test min contact fraction confirmation impact on contact impact of population density on hospital availability impact of treatment on fatality rate log ( dread factor in risk perception reference covid hospitalization fraction confirmed base fatality rate for unit acuity net covid acuity relative to flu net sensitivity of fatality rate to acuity net total asymptomatic fraction net sens obesity impact net sens chronic impact net sens liver impact net multiplier transmission risk for asymptomatic net relative risk of transmission by hospitalized = units: dmnl relative risk of transmission by presymptomatic = units: dmnl reported hazard of infection[rgn] = positive tests of infected response policy time on = units: day response policy weight[pdim] = units: dmnl ) rgn : albania saveper = units: day sens chronic impact = e- units: dmnl sens chronic impact net[rgn] = initial( sens chronic impact * ( -sw gen sens liver impact = e- units: dmnl sens liver impact net[rgn] = initial( sens liver impact * ( -sw gen sens obesity impact = e- units: dmnl sens obesity impact net[rgn] = initial( sens obesity impact * ( -sw gen senscoviduntestedadmission = units: dmnl sensitivity of contact reduction to utility sensitivity of contact reduction to utility policy * response policy weight[scup] + ( -response policy weight[scup] ) * sensitivity of contact reduction to utility sensitivity of contact reduction to utility policy = units: dmnl sensitivity of covid test = . units: dmnl sensitivity of fatality rate to acuity sensitivity of fatality rate to acuity net[rgn] = initial( sensitivity of fatality rate to acuity sensitivity post mortem testing to capacity ^ cmlterrpw ) / ( baseerror + datacmltovertime sim pseudo case fatality[rgn] = zidz ( cumulative deaths of confirmed[rgn] , cumulative confirmed cases simcmltovertime[rgn,infection] = cumulative confirmed cases post mortem test rate[rgn] + positive tests of infected post mortem test rate total simulated tests = if then else ( flowresiduals[rgn,series] = :na:, :na:, flowresiduals units: dmnl units: dmnl ) switch for government response[rgn] = if then else ( time > government response start time symptomatic fraction in poisson[rgn] = initial( -exp ( -covid acuity[rgn] ) ) units: dmnl symptomatic fraction negative symptomatic infected to testing[rgn] = positive testing of infected untreated[rgn] + hospital admission infectious testing capacity net of post mortem tests[rgn] = active test rate positive candidates interested in testing poisson subset[rgn] * symptomatic fraction in poisson[rgn] + potential test demand from susceptible population[rgn] * symptomatic fraction negative testing on living[rgn] = min ( testing capacity net of post mortem tests indicated fraction negative demand tested[rgn] * potential test demand from susceptible population[rgn] , indicated fraction negative demand tested[rgn] * potential test demand from susceptible population[rgn] + indicated fraction positive demand tested[rgn] * positive candidates interested in testing poisson subset tests per million[rgn] = cumulative tests data[rgn] / initial population time step = . units: day time to adjust testing = units: day time to downgrade risk time to downgrade risk policy * response policy weight[dgtp] + ( -response policy weight[dgtp] ) * time to downgrade risk time to downgrade risk policy = units: day time to herd immunity[rgn] = xidz ( herd immunity fraction * susceptible time to upgrade risk total asymptomatic fraction total asymptomatic fraction net[rgn] = initial( total asymptomatic fraction total covid hospitalized[rgn] = sum ( hospitalized infectious total disease duration = onset to detection delay + "post-detection phase resolution time" + incubation period units: day total simulated tests[rgn] = post mortem tests total[rgn] + testing on living total test on covid patients[rgn] = max ( , min ( positive candidates interested in testing poisson subset total to official cases simulated[rgn] = zidz ( cumulative cases[rgn] , simcmltovertime[rgn,infection] ) units: dmnl pre-symptomatic infected transmission multiplier for confirmed[rgn] = initial( baseline transmission multiplier for untested symptomatic * confirmation impact on contact transmission multiplier for hospitalized[rgn,tststs] = initial( baseline transmission multiplier for untested symptomatic * relative risk of transmission by hospitalized * if then else ( tststs = , confirmation impact on contact transmission multiplier pre detection[rgn] = initial( baseline transmission multiplier for untested symptomatic * ( -total asymptomatic fraction net[rgn] ) + total asymptomatic fraction net[rgn] * baseline risk of transmission by asymptomatic baseline transmission multiplier for untested symptomatic * relative risk of transmission by presymptomatic ) * ( -total asymptomatic fraction net[rgn] ) + total asymptomatic fraction net[rgn] * baseline risk of transmission by asymptomatic[rgn] * relative risk of transmission by presymptomatic ) units: dmnl active test rate[rgn] units: person/day ) tststs : tested untested pmas gap with tested[rgn] = ( -allocated fration noncovid hospitalized[rgn] ) ^ senscoviduntestedadmission units: dmnl untested symptomatic infected to hospital[rgn] = hospital admission infectious utility from limited activities[rgn] = exp ( sensitivity of contact reduction to utility net[rgn] * fractional value of limited activities ) units: dmnl utility from normal activities[rgn] = exp ( sensitivity of contact reduction to utility net[rgn] * ( -perceived risk of life loss voluntary reduction in contacts[rgn] = f[rgn] / f [rgn] * ( -min contact fraction[rgn] ) + min contact fraction zidz ( sum ( average acuity hospitalized[rgn,tststs!] * hospitalized infectious[rgn,tststs!] ) , sum ( hospitalized infectious weather effect on transmission weight max in test goal = units: dmnl infected unconfirmed post-detection"[rgn] + infectious confirmed not hospitalized[rgn] + sum ( hospitalized infectious[rgn,tststs!] ) * "post-detection phase resolution time" / hospitalized resolution time units spread of sars-cov- in the icelandic population clinical characteristics and imaging manifestations of the novel coronavirus disease (covid- ): a multi-center study in wenzhou city detection of sars-cov- in different types of clinical specimens sensitivity of chest ct for covid- : comparison to rt-pcr stability issues of rt-pcr testing of sars-cov- for hospitalized patients clinically diagnosed with covid- weather conditions and covid- transmission: estimates and projections. medrxiv clinical characteristics of coronavirus disease in china characteristics of and important lessons from the coronavirus disease (covid- ) outbreak in china: summary of a report of cases from the chinese center for disease control and prevention covid- exacerbating inequalities in the us estimating the infection and case fatality ratio for coronavirus disease (covid- ) using age-adjusted data from the outbreak on the diamond princess cruise ship estimating the infection fatality rate among symptomatic covid- cases in the united states. health aff (millwood) suppression of covid- outbreak in the municipality of vo estimating the asymptomatic proportion of coronavirus disease (covid- ) cases on board the diamond princess cruise ship universal screening for sars-cov- in women admitted for delivery incubation period and other epidemiological characteristics of novel coronavirus infections with right truncation: a statistical analysis of publicly available case data modeling the rework cycle: capturing multiple defects per task data analysis using regression and multilevel/hierarchical models accelerating markov chain monte carlo simulation by differential evolution with self-adaptive randomized subspace sampling an interactive web-based dashboard to track covid- in real time simulation-based estimation of the early spread of covid- in iran: actual versus confirmed cases estimates of the severity of coronavirus disease : a model-based analysis. the lancet infectious diseases serial interval of covid- among publicly reported confirmed cases serial interval of novel coronavirus (covid- ) infections an interactive web-based dashboard to track covid- in real time key: cord- -y zxipiz authors: dagpunar, j. s. title: sensitivity of uk covid- deaths to the timing of suppression measures and their relaxation date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: y zxipiz in this paper i examine the sensitivity of total uk covid- deaths and the demand for intensive care and ward beds, to the timing and duration of suppression periods during a day period. this is achieved via a seir model. using an expected latent period of . days and infectious period of . days, r_ was first estimated as . using observed death rates under unmitigated spread and then under the effects of the total lock down (r_ = . ) of march. the case fatality rate given infection is taken as %. parameter values for mean length of stay and conditional probability of death for icu and non-icu hospital admissions are guided by ferguson et al.( ). under unmitigated spread the model predicts around , deaths in the uk. starting with one exposed person at time zero and a suppression consistent with an r_ of . on day , the model predicts around , deaths for a first wave, but this reduces to around , if the intervention takes place one week earlier. if the initial suppression were in place until day and then relaxed to an r_ of . between days and , to be followed by a return to an r_ of . , the model predicts around , deaths. this would increase to around , if the release from the first suppression takes place one week earlier. the results indicate the extreme sensitivity to timing and the consequences of even small delays to suppression and premature relaxation of such measures. in order to model the spread of total covid- deaths in the uk over the next days say, one can build a detailed stochastic micro simulation model that includes spatial and age heterogeneity as in davies et al. ( ) and ferguson et al. ( ) . in order to gain some understanding of the sensitivity with respect to the timing of suppression measures and their relaxation, a simple deterministic seir model, see for example hethcote ( ) and blackwood and childs ( ) , can be useful. that is the approach used in this paper. seir is a compartment based model with transitions as shown in diagram . s stands for the number of susceptibles, those who have not yet been infected. initially, almost the entire population is in that category. e is the number who are currently exposed, that is they are infected but not yet infectious. after a latent period, an exposed person becomes infectious and is able to infect susceptible persons. i is the number who are currently infectious. initially, both i and e are small numbers in comparison to n, the population size. each infectious person remains so for an infectious period. an infective person is one who is either exposed or infectious. finally, r stands for those who are removed in the sense they have either recovered or died. in the simple form of the model considered here it is assumed that + + + = . the seir model is more appropriate than a sir model for covid- as sir does not separate the exposed and infectious states. a key controllable parameter is the reproduction number ! which is the expected number of susceptibles that an infector will infect, when the susceptible proportion is close to . it is thought that a significant proportion of those infected may be asymptomatic or presymptomatic in which case they might spend all or part of the infectious period unknowingly infecting susceptibles, although there is a question as to how much virus they shed. on the other hand, those who are symptomatic may spend some of that time in self-isolation which reduces the ! for them. one of the simplifying assumptions of the model developed here is that people can transmit the virus throughout their infectious period. a person will transition between these compartments. the seir model will keep track of how many are in each compartment by noting the rate at which people join and leave a compartment. the key dynamic element of this process is the transmission rate at which an infectious person infects susceptibles. diagram shows the rates of movement between the compartments. the mean latent and infectious periods are respectively ! " and ! # , while " is the probability that an infectious person is admitted to hospital. r is the cumulative number who do not require hospital treatment. it is assumed that all such persons survive, but it may be more accurate to say with hindsight that it may include an appreciable number of people in the community who never went to hospital and did not survive. in that sense the model underestimates the cumulative number of deaths. the model is an idealisation of what happens in practice as some people will enter hospital while still infectious and some will self-isolate during the infectious state. this can be compensated for by a suitable choice of . apart from predicting the number of deaths it is also important to predict the demand for ward and intensive care beds. the uk national health service (nhs) has approximately , ward beds and around , intensive care beds. clearly, one could not expect to access all of these as the occupancy rate is high in normal times. there is the option of building surge capacity. ideally, that would need to be matched by similar increases in medical, nursing, and ancillary staff; equipment such as ventilators; personal protective equipment; and testing for presence of the virus, all of which have become major issues. turning now to what happens in hospitals, diagram gives a schematic view with the transition rates. s e i r . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . an infectious person will move with probability " to hospital. of these a proportion # will never require an intensive care bed and will occupy a ward bed. of those a proportion # will die and the rest will recover. either way their average length of stay will be ! $ days. the remaining proportion − # will spend an average of ! % days in a ward bed followed by an average of ! & days in intensive care. a proportion $ of those entering intensive care will die. for simplicity, those who recover are not re-directed to a ward bed. the symbol in each compartment gives the number of persons currently occupying it, # and $ being the cumulative number of deaths from non-icu and icu patients respectively. cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . diagrams and lead directly to differential equations ( - ) in this section we obtain the important results for the case when / ≈ . equations ( - ) can be written as the negative root will give a solution that decays rapidly and can be ignored, see for example ma ( ) . so, while the susceptible proportion is close to we have the solution and where = ( ) ( ) ' ) $ ( ) that is . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint ( + )( + ) = . ( ) the expected number of secondary infections from one infectious person is for the special case of = substituting ( ) into ( ) while > the doubling time is and when < the halving time is ( ) from ( ) and ( ) , will all change at rate so one could use any of these to estimate . in the uk there were a large proportion of infectious persons (including asymptomatic and pre-symptomatic) who were never tested and the availability of tests changed markedly over time. therefore, it is felt that the number of positive test results is not a reliable proxy for the number of infectious persons. if is the case fatality rate then the expected number of deaths up to time is [ − ( − )] where is the mean time between exposure and death. differentiating, the death rate at − is − the government told people who showed symptoms to stay at home, without testing for seven days before contemplating hospital admission. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . for this reason, was estimated by fitting the logarithm of daily deaths nonparametrically in the period march to april. the dates were selected as to be largely uninfluenced by the gentle suppression introduced on march and the full lockdown beginning march, given that there is a lag between exposure and death of the order of days. the fit gave = . , % confidence interval ( . , . ) with the amount explained by the regression as $ = . . the estimated doubling time in the early stages of unmitigated spread is in order to estimate ! using ( ) we need not only , but estimates of mean latent and infectious periods. ferguson et al. ( ) assumed an incubation period of . days with a hour pre-symptomatic infectious period for some, together with a . day mean generation interval. anderson et al. ( ) thought the incubation period to be around - days with a serial interval of similar duration. davies et al. ( ) assumed a gamma distributed latent period with mean and standard deviation of and days respectively, a mean incubation period of . days, and a serial interval of . days. in this paper we take the mean latent and infectious period to be . and . days respectively, leading to a crude estimate of the serial interval to be . + . x . = . days. using these in ( ) gives ! = . . the death data was that pertaining to english hospital deaths by date of death . this data set was chosen as many other accessible sets are by date of recording the death which can be several days after actual death. we assumed that the growth/decline rate is similar in all four countries of the uk. the model assumes all deaths occur in hospital and applies to the entire uk of population million. for illustrative purposes this is considered reasonable although in the event many deaths occurred in the community including care homes. the model calculates bed occupancy assuming an ideal scenario where all who need hospital care receive it. it is assumed that % of those infected die and . % of those infected require hospital treatment. of these % will never require intensive care and spend an average of days in a ward bed. of these % will die. of the remaining % all will spend days in a ward followed by days in intensive care and % of these will die, these parameter values give a case fatality rate given infection of %. these figures are adapted from those of ferguson et al. ( ) . table summarises the model results over a period of days with an initial condition of one exposed person at time zero and ! = . . figure shows the death rate, required ward and intensive care beds, numbers of susceptible, exposed, and infectious, and cumulative deaths over time. of note is the great speed at which the disease spreads, with almost all deaths occurring within days. secondly, is the fact that at the peak of the wave, the capacity of the heath service, even assuming it was exclusively available to covid- patients, is exceeded by a factor of around and for intensive care and ward beds respectively. https://www.england.nhs.uk/statistics/statistical-work-areas/covid- -daily-deaths/ . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . ( ) , ( ) next we consider the effect of a lockdown such as the one implemented on march . we used death data for the period to april to estimate = − . , with % confidence interval (- . , - . ) and an explanatory $ of . . thus, the estimated halving time in declining the peak, once the effect of the previous ! of . has been largely eliminated, is exposed and infectious . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint from a behavioural survey of people's distancing following lockdown. table shows summary statistics and figure the associated graphs. ( ) of note is the speed at which peak death rate is reached, some days after starting suppression. figure suppression to ! = . on day we note that icu capacity is exceeded by a factor of , while ward bed demand is approximately % of capacity. exposed and infectious . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . now suppose that the suppression in table takes place one week earlier on day . table and figure show the model results ( ) we see a very large reduction in absolute death numbers from around , to , , this extreme sensitivity is a result of the exponential rise in infectious numbers and in hindsight clearly illustrates that earlier action was needed and would have saved many lives. the peak death rate occurs days after the beginning of suppression. exposed and infectious . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . . releasing the suppression to a relaxed of . a lockdown to an ! of . from day to day is unrealistic as it causes great economic and societal damage as well as increasing other mortality and morbidity. now suppose that the suppression of figure is released to a more relaxed ! of . between days and . that will mean there is a second wave so we limit its effect by suppressing again to ! of . between days and . the results are shown in table and figure . . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . the total deaths have been limited to around , by driving down infectious numbers to very low levels in the first wave. the first suppression was released on day . what happens if the release from the first suppression is brought forward from day to day ? table and figure show the results. exposed and infectious ward beds intensive care beds ward capacity icu capacity . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . we note that the number of deaths has increased by around , , an increase of approximately %. the model constructed here is a simple one. it cannot compete with detailed micro simulation models. many simplifying assumptions are made. there is much uncertainty about the parameter values for the examples shown. in reality, they will be influenced by the degree of stress a health system is exposed to. rather than precise prediction, the intention has been to show the modelling principles and to demonstrate the extreme sensitivity of the total number of deaths to the timing of suppression measures and their release/relaxation. literally, each day's delay in starting suppression can result in thousands of extra deaths. after the peak has been passed it is equally important to time the relaxation so that the further death toll is not excessive. these conclusions are the incontrovertible consequence of the exponential growth and decline of a managed epidemic. a more sophisticated model will enable more precise predictions on cumulative deaths under a range of control strategies, such as shielding and segmentation of the population. these can then be put alongside the economic and other effects of prolonged suppression. the political decisions can then be well informed. at the time of writing on may some days after the lockdown of march, we note that total deaths were , and rising, which is perhaps quite close to the scenario considered in figure . comparing with figure it does pose the question as to why lockdown did not occur earlier? it is quite possible that the cumulative deaths obtained from the model are an underestimate, since in the event a substantial proportion of deaths occurring in care homes are lagging those in hospitals and were not used in the estimate of ! . further, no account has been taken in the model of a gradual transition from the unmitigated ! of . to the estimated value of . following lockdown. going forward, ! will be monitored to update predictions. a disadvantage of using death data is that one is essentially estimating ! some days previously. but the alternative of using positive test numbers might lead to biased estimates due to the volatility of the proportion of infectious persons who are represented in such figures. perhaps the compromise is to use hospital admissions as a proxy for infectious numbers. how will country based mitigation measures influence the course of the covid- epidemic ? an introduction to compartmental modeling for the budding infectious disease modeler the effect of non-pharmaceutical interventions on covid- cases, deaths and demand for hospital services in the uk: a modelling study impact of non-pharmaceutical interventions (npis) to reduce covid- mortality and healthcare demand the mathematics of infectious diseases quantifying the impact of physical distance measures on the transmission of covid- in the uk. centre for mathematical modelling of infectious diseases, department of infectious diseases, london school of hygiene and tropical medicine, london wc e ht . first online estimating epidemic exponential growth rate and basic reproduction number key: cord- - j ekiy authors: lyu, m.; hall, r. title: dynamic modeling of reported covid- cases and deaths with continuously varying case fatality and transmission rate functions date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: j ekiy in this paper, we propose an enhanced seird (susceptible-exposed-infectious-recovered-death) model with time varying case fatality and transmission rates for confirmed cases and deaths from covid- . we show that when case fatalities and transmission rates are represented by simple sigmoid functions, historical cases and fatalities can be fit with a relative-root-mean-squared-error accuracy on the order of % for most american states over the period from initial cases to july ( ). we find that the model is most accurate for states that so far had not shown signs of multiple waves of the disease (such as new york), and least accurate for states where transmission rates increased after initially declining (such as hawaii). for such states, we propose an alternate multi-phase model. both the base model and multi-phase model provide a way to explain historical reported cases and deaths with a small set of parameters, which in the future can enable analyses of uncertainty and variations in disease progression across regions. covid- has challenged the world to react to a new contagious virus in the absence of effective medical treatment and vaccines. over the course of nine months from the outbreak in december , when the first cases were confirmed in wuhan, china, until april th , countries and territories reported nearly million confirmed cases and a death toll exceeding , persons [ ] . in the meantime, waiting for effective clinical care and vaccination, countries have reacted to the pandemic by controlling travel, implementing large-scale quarantine and restricting gatherings and contact among people, as well as requiring hygiene measures and screening for possible cases. in this paper, we seek to improve understanding of the dynamics of how covid- is spread, utilizing a variation of the susceptible-exposed-infectious-recovered-death (seird) model. our key innovation is representation of the transmission rate and case fatality rate as continuously varying functions, which are optimally fit to historical data on confirmed cases and deaths. we surmise that neither parameter is static, as they are influenced by the enactment and adherence to public health measures and medical care, neither of which is constant over time. we have applied our model to all american states to derive insights into how the disease has spread in different localities, which are influenced by population health, disease exposure, localized public health interventions and messaging, in addition to other place specific factors. prior research prior research on covid- has estimated disease-specific parameters, such as the basic reproduction number and latent period [ ] [ ] [ ] , demonstrating why the disease is interventions, such as closures of schools and restaurants [ ] . another use of disease transmission models has been to predict and plan for future demands on the health care system, such as demands for hospital beds (icu in particular) and needs for health care resources, such as ventilators. toward that goal, [ ] provides a statistical model of death data to predict future fatalities, assuming that social distancing measures are maintained. from the projected fatality data, they estimated hospital utilization with an individual-level microsimulation model based on due to the limits of testing methods, the long incubation period, and cases with mild or no symptoms and delayed reporting, there is potentially a huge (and unknown) number of unreported cases, the extent to which could affect the future evolution of the epidemic. some researchers, therefore, have used the sir (symptomatic-infectious -recovered) model and seird to estimate the number of undetected cases [ ] and [ ] . some approaches also incorporate transportation information (such as human migration data and community mobility data) to analyze the impact of travel on disease transmission and thus the effect of travel restriction [ ] [ ] [ ] . however, most studies using seird or sir assume the transmission rate and death rate to be constant over time. with improvement in clinical treatment and changing intervention policies, the transmission rate and fatality rate will be time variable. therefore, seird models with constant parameters cannot accurately depict the spread of disease. [ ] and [ ] in our research we explore use of a concise formulation through which continuously time varying transmission and case fatality rates are modeled with a small number of parameters, which are fit to historical data. like [ ] , we utilize a type of logistic function (i.e., a sigmoid function), but not simply to model reported deaths over time, but to instead model transmission rate and case fatality rate within the seird model. our objective is to improve the classical seird model through an approach that adapts to the dynamic pattern of transmission under different epidemic scenarios. thus, we provide insights into transmissibility of the disease while modeling historical data on confirmed cases and confirmed deaths. the proposed time varying model we draw from the seird compartmental model, which divides the population into five groups: susceptible(s), exposed(e), infected(i), recovered(r) and dead(d). seird infectious people and turn into the state of exposed. exposed people are in a latent state and then progress to the infectious state with a rate inversely related to the incubation period (thus, exposed is defined as a state in which people are not yet infectious). infected people eventually progress into either the dead state, if they succumb to the disease, or the recovered state, with different rates. those who have recovered are assumed to be no longer susceptible to contracting the disease. we introduce death rate α (t) as a time varying function, representing the proportion of infectious individuals who eventually die from the disease, by date. those who eventually die transfer from the infected to the died state at a rate of ρ, representing the inverse of the time from becoming infectious until time of death. in our model, ρ is assumed to be constant over time. those who eventually recover do so at the γ, representing the inverse of the time from becoming infectious until recovery. we will also later derive the effective reproduction number rep (t), representing the average number of persons who are exposed to the disease by each infectious person, as a function of time. taking these factors into account, the system of equations of the proposed seird model is given by:: where: preprint (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this this version posted september , . . https://doi.org/ . / . . . doi: medrxiv preprint r (t) = number of people who have recovered at time t n = total number of people β (t) = transmission rate at time t σ = transformation rate from exposed to infectious, which is the reciprocal of the incubation period while, in theory, these functions may change erratically as a consequence of discrete events, such as new public health measures, we hypothesize that such discrete events do not suddenly alter either function. therefore, we seek to understand whether a simple continuous model, with a minimal set of parameters, might accurately represent historical data. for illustration, at the enactment of a new intervention policy, the public may not react immediately, and neither do the transmission parameters. the public will get used to the policy after a period of adaptation, and eventually the effective reproduction number will stabilize. in addition, the public responds to both government policies and communication about the disease. communication comes from many, sometimes conflicting, sources. how the public at large absorbs and responds to such often confusing messages may be gradual. a natural function to describe this pattern of change is the sigmoid function. thus, we define the function for transmission rate and death rate as: september , / all rights reserved. no reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this this version posted september , . . as mentioned in the cdc reports [ ] [ ] [ ] , the median incubation period is days, preprint (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this this version posted september , . . (as of march st few states had reported cases). prior research suggests that the initial effective reproduction number is around [ ] , equivalent to a transmission rate of . , which we use for initialization. because transmission rates vary significantly among the number of reported deaths is smaller than the number of reported cases in all locations. thus, treating errors in death estimation and case estimation the same will lead to underfitting of the death data, in preference to minimizing the errors in case data. therefore, considering the accuracy of the reported death data and the fitting accuracy, we optimized a weighted sum of squared death and case data, multiplying w by deaths during the fitting process. the adjusted objective function is: the parameters are estimated by solving the nonlinear constrained least-squares reporting has also shown a consistent day-of-week variation across many locations, with weekend data differing from weekday data. this variation is more likely the consequence of different patterns of healthcare staffing, and differences in how patients present for testing by day of week, rather than differences in disease transmission by day of week. to smooth out these effects, we model the moving -day average data instead of the daily reported data where y n is the case/death number on the n t h day. wyoming. figure shows that the proposed seird model with time-dependent preprint (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this this version posted september , . . and hawaii, the cdc data follows a pattern of two phases, which is not as well captured by our model. especially for hawaii, the curve flattened for a period and then rose. as discussed later, our model characterizes the transmission dynamic for a period with one phase, i.e. the curve should become flat at most once. all rights reserved. no reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. effective reproduction number at any time t, which we define as rep (t), is the average number of people in a population who are infected per infectious case, where everyone is susceptible to the disease. rep (t) measures the transmission potential of infectious diseases [ ] . when rep (t) > , the rate of new cases will increase over time, until the population loses susceptibility to the disease. when rep (t) < , the rate of new cases will decline over time. rep (t) can be estimated with the next generation matrix method [ ] . we define x as the vector of infected class (i.e. e, i) and y as the vector of uninfected class (i.e. s, r, d). then the effective reproduction number is the spectral radius of m, expressed as: at the beginning of the epidemic, rep (t) reflects the natural transmissibility of preprint (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this this version posted september , . we divide the hawaii timeline into two periods, the first from march th until may th , and the second from may th to july th . we fit the first stage with the initialization of only one exposed people at the start. to initialize the second phase, we use the predicted number of exposed people, infectious people and recovered people from the first phase, combined with the reported deaths as of may th . with this modification, the rrmse for cases declines below . % and the rrmse for deaths declines below . %. the fitting results in figure show that our two-phase model captures the transmission pattern more precisely than the single-phase model. the history of effective reproduction number and death rate are shown in figure . the first phase showed a decline in the reproduction number after the initial announcement of the stay-at-home order. however, with the reopening, the reproduction number increased, explaining increases in case rates. death rates, by contrast, exhibit a peculiar behavior, increasing over time in each phase, with a discontinuity when transitioning from the first phase to the second. beyond exhibiting two phases, hawaii has a small number of deaths, with no deaths occurring in the transition period between phases. we surmise that the function, while representing the data well, is peculiar because of the unusual pattern in deaths within hawaii. preprint (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. we have developed an extension of the seird model that represents changing transmission and death rates over time as a continuous sigmoid function, under the hypothesis that these rates change gradually, rather than immediately upon implementation of public health policies or treatments. the model fits historical data for the united states well for most states. those with poorer fits exhibit patterns of multiple waves of the disease. using hawaii as an example, a multi-phase extension of the model provides a more accurate fit, where the transition from one phase to the next is defined by a change in public health policy. an advantage of the model is that it is defined by a small set of parameters. thus, it provides an efficient method for quantifying differences among regions in the spread and outcomes of the disease. in the future, we intend to examine the predictive value for the model, taking into consideration ranges of uncertainty in model estimates. by examining historical trends, we can understand how variations in simple parameters can lead to fewer or more cases and deaths. in the future, we will develop multi-region extensions of the model, which permit representation of spread of disease from one region to another, or perhaps within sub-regional groups. our research is premised on the method of modeling case and death data as they are reported. we recognize that the true number of cases may differ from reported values, as might the number of deaths, both of which are unknown. the variations from state to state reflect in part the actual spread and outcomes of disease, as well as the extent to which cases are detected and reported, as well as how deaths have been classified. transmission dynamics of the covid- outbreak and effectiveness of government interventions: a data-driven analysis coronaviruses and acute respiratory syndromes (covid- , mers, and sars) -infectious diseases. merck manuals incubation period of -ncov) infections among travellers from wuhan preliminary estimation of the basic reproduction number of novel coronavirus ( -ncov) in china, from to : a data-driven analysis in the early phase of the outbreak the reproductive number of covid- is higher compared to sars coronavirus new york state task force on life & the law nysdoh. ventilator allocation guidelines need , ventilators very soon far short of that. the new york times forecasting covid- impact on hospital bed-days, icu-days, ventilator-days and deaths by us state in the next months. medrxiv projecting hospital utilization during the covid- outbreaks in the united estimating the a data-driven modelling analysis of the early outbreak estimation of the asymptomatic ratio of novel coronavirus infections (covid- ). international journal of infectious diseases management strategies in a seir model of covid community spread estimating the prevalence and risk of covid- among international travelers and evacuees of wuhan through modeling and case reports the association between domestic train transportation and novel coronavirus ( -ncov china from to : a data-driven correlational report. travel medicine and infectious disease seir modeling of the italian epidemic of using computational swarm intelligence monitoring italian covid- spread by a forced seird model clinical characteristics of coronavirus disease in china characteristics and clinical outcomes of adult patients hospitalized with covid- -georgia characteristics of persons who died with covid- -united mmwr morbidity and mortality weekly report the levenberg-marquardt algorithm: implementation and theory first case of novel coronavirus in the united states the covid tracking project. the atlantic the effective reproduction number as a prelude to mathematical and statistical estimation approaches in epidemiology key: cord- -a t u authors: nan title: alphabetic listing of diseases and conditions date: - - journal: handbook of autopsy practice doi: . / - - - - _ sha: doc_id: cord_uid: a t u part ii begins with a list of special histologic stains, their for use and their corresponding references. at the end of this list is a procedure for removal of formalin precipitate from tissue sections. diseases. there may also be a list of possible associated conditions. these entities are generally linked pathogenetically to the main disease entry. any asterisk after a related disease indicates that that disorder is also listed as a disease entry. many disease entries will be followed by a three-column table that provides the reader with a listing of the pathologic findings to be expected with the disease as well as the prosection and dissection procedures necessary to demonstrate those findings. it is expected that routine hematoxylin-eosin stains will be done on all sections submitted for histologic examination. special stains will be recommended in the procedures column of the tables, when indicated. any table immediately following the two columns of disease entries always refers to the disease in the right column. prepare smears of undiluted blood. obtain blood for molecular studies for preservation of small intestinal mucosa and for preparation for study under dissecting microscope, see part i, chapter . submit sample for histologic study. submit stool for chemical analysis. record weight and submit sample for histologic study. freeze liver for molecular studies record appearance of spine (see also chest roentgenogram). for removal and specimen preparation, see chapter . request luxol fast blue stain. for removal and specimen preparation, see chapter . below-normal weight in infants. kyphoscoliosis. very low concentrations of cholesterol and decreased triglycerides; serum~-lipoprotein or absent; a.-lipoproteins present. acanthocytosis (spiny red cells). gene mutations ( ) . abnormal shape of villi; vacuolation of epithelial cells. fatty stools fatty changes. gene mutations ( ) systemic manifestations of malabsorption syndrome* and of vitamin a deficiency. * kyphoscoliosis. axonal degeneration of the spinocerebellar tracts; demyelination of the fasciculus cuneatus and gracilis ( ) . possible involvement of posterior columns, pyramidal tracts, and peripheral nerves. atypical retinitis pigmentosa ( ) with involvement of macula. angioid streaks ( ) . synonym: cerebral abscess. note: for microbiologic study of tissues and abscesses, see part i, chapter . include samples for anaerobic culture. it is best to study the brain after fixation but if specimen is examined fresh, aspirate and prepare smears of abscess content. photograph surface and coronal slices of brain. request giemsa stain, gram stain, pas stain, and grocott's methenamine silver stain for fungi. external examination if there is evidence of trauma, see also under "injury, head." prepare roentgenograms of chest and skull. submit for microbiologic study. for removal and specimen preparation, see chapter . for microbiologic study, photography, and special stains, see under "note." for exposure of venous sinuses, see chapter . sample walls of sinuses for histologic study. for exposure of paranasal sinuses, mastoid cells, and middle ears, see chapter . for removal and specimen preparation, see chapter . procedures depend on suspected lesions as listed in right-hand column. skin infections in upper half of face. edema of forehead, eyelids, and base of nose, proptosis, and chemosis indicate cerebral venous sinus thrombosis. * trauma; craniotomy wounds. skull fracture and other traumatic lesions. for possible intrathoracic lesions, see below under "other organs." the national transportation safety board (ntsb)* has authority over aircraft wreckage and the legal authority to investigate and to determine the cause of air crashes. ( ) the dead are the responsibility of the medical examiner or coroner. local police will seal off the area of the crash. other than for the purpose of determining that death has occurred, no one should be allowed to approach the bodies or any objects until the identification teams and the medical examiner or coroner have taken charge. the sudden influx of bodies after a commercial air carrier accident and the request for speedy identification of the victims would overburden almost any institution. managing such a disaster is eased by writing a contingency plan beforehand. temporary morgue facilities may have to be established near the scene of the crash. refrigerated trucks may serve as storage space. a practical approach is to deal first with those bodies that seem to be the easiest to identify, in order to narrow the field for the more difficult cases. if bodies are scattered, their locations can be referenced to stakes in the ground or spray paint on pavement; only then should these bodies (or parts) and personal effects be collected. for large-scale crashes a locations can be referenced to a string-line grid benchmarked to gps coordinates. records and diagrams of the relative positions of victims are prepared during this phase. if bodies are still within the airplane, their positions are recorded, and photographed. the personnel of the medical examiner or coroner can augmented by d-mort team staffed by forensic pathologists, anthropologists, dentists, morgue technicians, and investigators supplied by the national disaster medical system. ** the airline will provide a list of the passengers and the federal bureau of investigation (fbi) disaster team will make itself available to take and identify fingerprints and aid in the acquisition of other identifying data such as age, race, weight, height, and hair color and style. if dental records can be obtained, this provides one of the most certain methods of identification. a medical history indicating amputations, internal prostheses, or other characteristic surgical interventions or the presence of nephrolithiasis, gallstones, and the like will be helpful. fingerprints (and footprints of babies) should be taken in all instances. wallets with identification cards,jewelry, name tags in clothing, or other personal belongings may provide the fastest tentative identification. the medical examiner may elect to autopsy only the flight crew but not the passengers of an aircraft crash. however, the grossly identifiable fatal injuries should be described, photographed, and x-rayed. this may reveal identifying body changes. if comparison of somatic radiographs, dental records, fingerprints, or photographs do not identify the victim, dna comparison must be considered. burned or fragmented bodies of passengers and the bodies ofcrew members, and particularly the pilots, must have a complete autopsy, including roentgenographic and toxicologic examinations, which must always include alcohol and carbon monoxide determinations. internal examination might reveal a coronary occlusion, or roentgenograms may disclose a bullet as evidence that violence preceded the crash. in some airplane crashes, particularly in light airplane accidents, suicide must be considered. in such cases police investigation is required to determine if the pilot exhibited suicidal ideation in the recent past.. when resources permit, autopsies should be performed on all deceased occupants of aircraft crashes, including passengers, in order to distinguish among blunt impact trauma, smoke inhalation, and flash fires as causes ofdeath, and to answer future questions concerning pain and suffering, intoxication, and sequence of survivorship. after a crash victim has been identified, the coroner or medical examiner will issue a death certificate. if remains of a decedent cannot be found, a judge can, upon petition, declare a passenger dead and sign a death certificate prepared by a medical examiner. *phone # ofntsb command center: - - **phone # of dmort: - - . entry should be followed. usually, the circumstances that led to drowning are not apparent from the autopsy findings but can be reconstructed from reports of witnesses and the police. because the reflex drive to seek air is triggered by hypercarbia, not hypoxia, loss of consciousness and drowning can ensue after hyperventilation and breath-holding by experienced swimmers who then drown without a struggle. there are no specific autopsy findings. a search for trauma, including a posterior neck dissection, should be made in all instances. head and cervical injuries may be responsible for loss of consciousness and drowning, usually in individuals diving into shallow water. toxicologic examination as described below for scuba diving accidents is always indicated. with scuba diving fatalities, investigation of the equipment and circumstances is usually more important than the autopsy. scuba fatalities should be studied by or with the aid of diving experts-for instance, members of a diving club or shop (not the one providing the gear used by the decedent) or the u.s. navy. ( ) careful investigation of the scene and study of reports of witnesses and the police are essential. the investigation should ascertain the site of diving (currents and other underwater hazards), the estimated depth, the water temperature (exposure to cold), and a description of water clarity. electrocution should be considered if the site has electric underwater cables (see "injury, electric"). cerebral concussion should be considered if explosives were used in the vicinity. knowledge of the method of recovery of the body and the type of resuscitation efforts can aid in the interpretation of apparent wounds. the medical history of the diving victim should be sought, as it may lead to a diagnosis for which the autopsy is typically silent, such as seizure disorder, or may reveal asthma, emphysema, or chronic bronchitis, all of which increase the risk of air trapping and arterial air embolism. although drowning may be the terminal event in some scuba deaths, the investigation should be focused on the adverse environmental and equipment factors that place a capable swimmer at risk of drowning (see "embolism, air" and "sickness, decompression"). because scuba divers risk arterial air embolism if they ascend with a closed glottis, on can attempt to document gas bubbles at autopsy, but their interpretation is problematic: bodies recovered immediately are subjected to resuscitation efforts, which can by themselves produce extra-alveolar air artifacts, and bodies not recovered immediately tend to be found in a putrefied condition, full of postmortem gas. in the remaining cases, the pathologist must consider the potential of introducing artifactual gas bubbles by the forcible retraction of the chest plate and by sawing the calvarium. the following procedures apply primarily to scuba diving accidents. interrogation of witnesses is important; the behavior and complaints of the decedent, if any, might help distinguish between a natural death by heart disease and an unnatural death by air embolism. external examination eyes and ears head (skull and brain) chest blood (from heart and peripheral vessels) heart tracheobronchial tree and lungs a procedures photograph victim as recovered and after removal of wet suit and other diving gear. record condition of clothing and gear. impound all diving equipment for study by experts, particularly scuba tank, breathing hoses, and regulators. residual air in tank should be analyzed. record color of skin (including face, back, soles, palms, and scalp). palpate skin and record presence or absence of crepitation. record extent and character of wounds. prepare histologic specimens. record appearance of face (including oral and nasal cavities) and of ears. prepare roentgenograms. if air embolism must be expected, as in the presence of pneumomediastinum, follow procedures described under "embolism, air." for evaluation of findings, see also above under "note." if decompression sickness (caisson disease) is suspected, also prepare roentgenograms of the elbows, hips, and knees. otoscopic examination. funduscopic examination. save vitreous for possible toxicologic and other studies. for removal of brain, see chapter . record contents of arteries of the circle of willis and its major branches and basilar artery. strip dura from base of skull and from calvarium. for removal and specimen preparation, see chapter . for demonstration of pneumothorax, see under "pneumothorax". if gas is visible in coronary arteries, photograph. photograph and aspirate gas in heart chambers. submit samples of heart blood and peripheral blood for toxicologic study and drug screen. examine lungs in situ. save bronchial washings for analysis of debris. fresh dissection is recommended. if decompression sickness is suspected, prepare sudan stains from fresh-frozen lung sections. complete toxicologic sampling should be carried out (see chapter ). record nature of gastric contents. remove neck organs toward end of autopsy. for posterior neck dissection, see chapter . incise tongue. for removal, see chapter . for removal, see chapter . for removal, prosthetic repair, and specimen preparation, see chapter . consult roentgenograms. in decompression sickness, fatty change of liver, and ischemic infarctions of many organs. interstitial emphysema. aspiration (see above). trauma to cervical spine. mottled pallor of tongue after air embolism. contusion of tongue after convulsive chewing. nitrogen bubbles in spinal cord arteries may occur after rapid ascent. air embolism;' cerebral edema in decompression sickness. aseptic necroses (infarcts, "dysbaric osteonecrosis"), most often in head of femur, distal femur, and proximal tibia. infarcts indicate repeated hyperbaric exposures. nitrogen bubbles in and about joints and in periosteal vessels ("bends") occur during rapid ascent. related terms: automobile accident; motorcycle accident. note: a visit to the scene can make the interpretation of the autopsy findings easier. the vehicle can also be inspected in a more leisurely fashion at the impound lot. this is particularly useful for correlating patterned injuries with objects in the vehicle. most vehicular crashes occur as intersection crashes or because a vehicle with excessive speed left a curved road. the medical examiner or coroner should gain a basic understanding of the crash mechanism so that informed descriptions can be rendered, e.g., "impact to the b pillar of the decedent's automobile by the front of a pickup truck which failed to stop for a stop sign at an intersection, resulting in a -feet intrusion into the cabin; restraint belts not employed; air bag deployed; extrication required which took minutes." police are responsible for determining mechanical and environmental risk factors for the crash and for determining some human risk factors such as suicidal or homicidal intent. the pathologist determines other risk factors for crashes such as heart disease, a history of epilepsy, and intoxication by carbon monoxide, drugs, and alcohol. suicide as a manner of death should be considered when a single-occupant vehicle strikes a bridge abutment or a large tree head-on, with no evidence of evasive action or braking. in such a situation, the standard police traffic investigation should be supplemented of interviews of the victim's family and friends. the ambulance run sheet is an invaluable source of observations that often are not available from the police. this document should be acquired in all instances, even if the paramedics determined that death occurred and did not transport. the basic autopsy procedures are listed below. most traffic victims who die at the scene or who are dead on arrival at the hospital died from neurogenic shock caused by wounds of the head or vertebral column, or from exsanguination from a tom vessel or heart. as such, they have little lividity, and little blood is found in the vehicles. presence ofintense lividity may indicate suffocation or heart disease as a cause of death. if postural asphyxia is suspected, the first responders to the scene should be interviewed to determine the position of the decedent in the vehicle, and the vital signs, ifany, ofthe decedent from the time of the crash to the time of extrication. posterior neck dissection is indicated in these instances. if manifestations of heart disease, intense lividity, and absence oflethal wounds suggest that a crash occurred because the driver was dead, other drivers on the road may have observed that the victim was slumped at the wheel before the crash. the determination of heart attack at the wheel is usually simple, because most such victims realize that something is wrong, and bring the vehicle to a stop at the side of the road, or coast gently into a fixed object. in such instances, damage to the vehicle is minor, and wounds to the decedent are usually trivial. while pattemed wounds can often be matched to objects (see below), patternless wounds usually cannot be visually matched to specific objects, although an opinion can sometimes be given as to what object was struck, based on the direction of motion and position ofthe body with respect to the vehicle. impacts with the a-pillar produce narrow vertical zones of facial laceration and fractures extending from forehead to jaw. tempered glass shatters into small cubes on impact, and leaves so-called "dicing" wounds, which are abraded cuts arranged in a somewhat rectilinear pattern. windshield glass leaves shallow, abraded, vertically oriented cuts on the face or scalp. with pedestrians, the lower extremities are of particular forensic interest, to determine the height and direction of impact from vehicles that left the scene. scalp hair and blood should be collected from such "hit and run" victims and from occupants of a suspect car if police have a question as to which occupant was the driver; these exemplars can be compared to fibers and tissue recovered from the vehicle in question. likewise, foreign material in wounds can sometimes be matched to suspect vehicles, and should be sought and retained as evidence. for pedestrians, the distance between the impact point on the lower extremities and the soles of the feet should be recorded. the legs should be opened to inspect tibial fractures; cortical fractures initiate propagation opposite to the side of impact, where they usually have a pulled-apart appearance, and then splinter the cortex at the side of impact. abrasions are better impact markers than contusions, because subcutaneous blood extravasation can be caused not only by impact to the skin, but also from blood extravasating from underlying fractures. if no cutaneous abrasions or fractures of the leg bones are found, the skin of the legs should be incised to expose contusions. fracture descriptions should include location in the bone (e.g., proximal metaphysis or shaft), whether the fracture is complete or incomplete, and whether the fracture is displaced or distracted. lacerations of intervertebral disks, facet joint capsules, and ligamenta flava should not be loosely termed "fractures." the presence or absence of blood extravasation in soft tissue adjacent to the fractures should be recorded, and its volume estimated if it appears severe enough. venous air embolism from tom dural sinuses cannot be diagnosed without a pre-autopsy chest radiograph or an in situ bubble test. if an x-ray machine is readily available, an anterior-posterior chest radiograph should be obtained in every traffic victim who dies at the scene or after a failed resuscitation attempt. if a hemothorax is suspected, the rib cuts should be placed further lateral and the chest plate reflected so that the internal mammary vessels can be inspected before the chest plate is removed. after measuring and removing the bloody effusion, the underlying serosal surfaces should be inspected for defects. lacerations of the heart and aorta will be obvious. tamponaded lacerations of the aorta, around which the adventitia still holds, must be noted as such. if no lacerations are found at the usual sites, lacerations of the azygous veins must be considered, especially in association with fracture dislocations of the thoracic vertebral column; other sites are the internal mammary arteries, especially with fractures of ribs i and or of the sternum, and intercostal arteries with displaced rib fractures. only after the serosal defect is identified should the organs be removed, because that procedure creates many more holes in the serosa. for that reason, as much information as possible should be gained by in situ observation. the only evidence of concussion of the heart may be a cardiac contusion or a sternal fracture. the usual clinical history suggests cardiovascular instability that is not associated with craniocerebral trauma and which does not respond to the infusion of intravenous volume agents. the autopsy assistant may saw but should not retract the skull cap and remove the brain. the pathologist should observe in situ whether shallow lacerations of the pontomedullary junction with stretching of the midbrain are present. these lesions cannot be distinguished from artifact by examining the brain later. thus, only after appropriate in situ inspection should the pathologist remove the brain. a posterior neck dissection is required if no lethal craniocerebral or cardiovascular trauma is found, or if suffocation is suspected; neck trauma must be ruled out to diagnose suffocation in a traffic fatality. sudden death in a patient with seemingly trivial wounds may be caused by undiagnosed trauma of the craniocervical articulation. a posterior neck dissection is required in these instances. the diagnosis of diffuse axonal injury of the brain in victims with no appreciable survival interval requires that suffocation be ruled out and that no resuscitation from a cardiac arrest has been attempted. clinicians are quick to apply the label "closed head injury" when a victim of a traffic crash has cerebral edema on a computerized axial tomogram of the head, even if no cerebral contusions, scalp contusions, or skull fractures are evident. this may be a misinterpretation, because cerebral edema can be caused by hypoxic encephalopathy made evident after resuscitation from a cardiac arrest, or from hypoxia caused by suffocation. procedures possible or expected findings record presence of lividity. photograph all external wounds; measure all lacerations and any abrasions or contusions with a pattern. collect scalp hair and blood (see below) from victims of hit and run accidents. collect foreign material in wounds. intense lividity and absence of lethal wounds may indicate that the crash occurred because the driver was dead from heart disease or suffocation. wound documentation. patterned injuries often sometimes be matched to objects in or about the vehicle (the most common patterned wound is that from tempered glass; see above under "note"). impact patterns in pedestrians may help to reconstruct the accident. hair and blood of the victim may be matched to transfer evidence on a vehicle suspected of having left the scene. part ii / diseases and conditions internal examination of body cavities heart and great vessels abdomen skull and brain; neck soft tissue compartments at any location prepare roentgenograms of chest is cases with head impact and skull fractures. collect samples for toxicologic study from all victims, including passengers. create pleural window to detect pneumothorax. if blood is seen, examine internal mammary vessels (see under "note"). measure volume of blood in cavity bleeds, and note whether chambers of heart and great vessels are collapsed or filled. record evidence of cardiac contusion, sprain of intracardiac inferior vena cava, laceration of pericardial sac, and fracture of sternum. laceration of heart or great vessels (measure volume of blood). follow routine procedures for dissection of heart and great vessels (see chapter ) . in situ bubble test to confirm venous air embolism. record evidence of trauma and volume of blood in peritoneal cavity; estimated volume of blood in retroperitoneal soft tissues. autopsy assistant may saw the skull but pathologist should inspect brain in situ and remove it personally. for removal and specimen preparation of brain, see chapter . record brain weight. posterior neck dissection is indicated if there is no craniocerebral or cardio-vascular trauma, or if suffocation is suspected. record evidence of trauma and estimate volume of blood. venous air embolism.' evidence of alcohol or drug intoxication. pneumothorax, hemothorax, e.g., after laceration of internal mammary vessels. evidence of significant hemorrhage. indirect evidence of cardiac concussion. evidence of exsanguinating wounds. evidence of cardiovascular disease that may have felled the driver before the crash. in european countries, the concentration is expressed in promille (grams per liter). in the united states, it has become customary to refer to concentration by percentage (grams per deciliter), and values in these units have been written into legislation and included in the uniform vehicle codes. unless qualified, the use of promille or percentage does not indicate whether the result of the analysis is weight/weight, weight/ volume, orvolume/volume. another common way ofexpressing concentration, milligrams per deciliter, has also been used to indicate alcohol concentrations. the method ofexpressing concentration must be clearly specified whenever the alcohol level is mentioned. the desired expression canbe derived from the toxicologic report by using the following equation: i, ~g/ml = mg/dl = . g/dl = . mmolll = . promille = . % what is the legal interpretation of alcohol (ethanol) intoxication? objective impairment of driving ability is observed at threshold blood alcohol concentrations of . -. g/dl. as of august all states and the district of columbia have adopted laws that make it criminal offense for a driver to operate a motor vehicle with a blood alcohol concentration of . g/ dl or greater. many states have an enhanced penalty for high concentrations such as . g/dl or above. several states have zero tolerance laws, under which drivers who are minors are legally operating only if their blood alcohol concentration is . g/dl or less, and in some states, not detectable at all. blood alcohol concentrations obtained at autopsy are valid until putrefaction begins. specimen tubes with sodium fluoride should be used, and the specimen should be stored in the refrigerator. if the air space above the blood samples in the container is large, alcohol can evaporate and a falsely low blood alcohol level can result. putrefactive changes before autopsy or during storage may cause a falsely high blood alcohol concentration. ethanol can be produced in the specimen container; this is more likely in the absence of a preservative. because fluoride inhibits bacteria far more than fungi, higher fluoride concentrations are required for the inhibition of fungal growth ( ) . although there is no major difference in the alcohol concentrations ofblood samples from the intact heart chambers and the femoral vessels ( ), autopsy samples from pooled blood in the pericardial sac or pleural cavity are unsatisfactory. we therefore recommend that blood be withdrawn from peripheral vessels. is there normal "endogenous" blood alcohol (ethanol) in a living person? blood alcohol concentrations are generally believed to be negligible in the absence of ingested alcohol. "endogenous" ethanol in human blood exists at a concentration of about . g/dl, which is below the limit of detection for most methods ( ) . first in such a list would be postural asphyxia, for example, in drunks who fall asleep face down. also, depressant drugs in the tricyclic, analgesic, barbiturate, and benzodiazepine classes all potentiate the effect of alcohol ( ) . also included in such a list would be infancy and childhood; ischemic heart disease;' chronic bronchitis and emphysema;' other chronic debilitating diseases; poisoning with carbon tetrachloride' or carbon monoxide;' and other causes of hypoxia.' how can one estimate blood alcohol (ethanol) concentrations from vitreous, urine, or tissue alcohol levels and from alcohol in stomach contents? the ratio of serum, plasma, urine, vitreous, and various tissues has been compiled by garriot ( ) . the values may vary considerably. for vitreous, the ratios varied from . - . . these variations may depend on whether blood alcohol concentrations were increasing or decreasing at the time of death. most other body fluids and tissues showed ranges closer to . most urine values were above the blood alcohol concentrations. in another study ( ) , the blood/vitreous (bn) ratio in the early absorption phase was . (range, . - . ; sd . ) and in the late absorption and elimination phase, the bn ratio was . (range, . - . ; sd . ). blood ethanol concentrations probably can be estimated using b = . v for early absorption and b = . v for later phases. a urinelblood ethanol ratio of . or less indicates that the deceased was in the early absorption phase. how can one use alcohol (ethanol) concentrations in postmortem specimens to estimate the blood alcohol concentration at various times before death? with certain limitations, one can base calculations of this kind on the assumption that the blood alcohol level decreases from its peak at a fairly constant rate of . -q. g/dl/h until death ( ) . if blood is not available, conversion factors (see above) must be used. alcoholics have been reported to metabolize at a rate of up to . g/dl/h ( ) . example: the driver of an automobile drinks at a party until midnight. he leaves his host at about : a.m. and is involved in a head-on collision at : a.m. he dies in the emergency room at : a.m. there are multiple injuries and the patient exsanguinates. the autopsy is done at : p.m. although this appears quite unlikely, let us assume that no satisfactory blood sample was obtained before death and that no blood or plasma expanders were given. if under such circumstances the alcohol concentration in the vitreous was found to be . g/dl, what was the alcohol concentration in the blood at the time of the accident? vitreous and blood alcohol concentrations may be assumed to have remained unchanged after death. therefore, the blood alcohol level at the time of death must have been approx . (vitreous humor alcohol) x . (conversion factor, see above) = . g/dl. the time interval between the accident ( : a.m.) and death ( : a.m.) is hand min or / h. if we assume that the decedent was not an alcoholic and that the blood alcohol concentration was decreasing from its peak at a constant rate of . g/dl/h, then the concentration at the time ofthe accident is estimated to have been . (concentration at time of death) + ( / x . ) = . + . = . g/dl or . %. the blood alcohol concentration at the time of the accident could have been lower if the victim stopped drinking later than h or / h before the accident. in the latter case, the peak alcohol level would have occurred after the accident, reflecting the time to absorb the latest drink. the blood alcohol concentration at the time of the accident could have been lower or higher if the time when the patient stopped drinking, the time of the accident, or the time of the death is uncertain. the blood alcohol concentration at the time of the accident could have been higher if the victim was a chronic alcoholic. the elimination rate in such persons may be as high as . mg/dl, which would change the figures in our example above to . + ( / x . ) = . + . = . g/d or . %. only rough estimates are possible. first, the peak blood alcohol level must be determined or calculated, as described in the previous paragraphs. tables (see below) are available that relate blood alcohol level to the minimal amounts of whiskey, wine, or beer that must have been consumed ( ) . however, tables of this type are often based on the minimum amount of alcohol circulating in the body after specific numbers of drinks; such tables do not yield reliable results if used conversely. furthermore, inasmuch as drinking and elimination of alcohol may take place concomitantly, over a longer period the total amount of alcohol consumed may have been much greater than the tables would indicate. it cannot be lower. according to these tables, pints of ordinary beer or fl oz of whiskey would be the minimal amounts needed to produce a blood alcohol level of about mg/dl in a person weighing - pounds. the total body alcohol can be calculated from the blood alcohol level by using widmark's formula: average concentration of alcohol in entire body = . concentration of alcohol in the blood in a person weighing kg, the blood alcohol concentration would be increased mg/dl ( . %) by the absorption of oz of ethanol ( z of -proof whiskey). strength of alcohol is measured in "proof'; absolute alcohol is proof. therefore, in the united states, alcohol content as volume percent is half the proof (for example, -proof whiskey contains % alcohol by volume). the alcohol content of various beverages is shown in the following table. approximate alcohol content in various beverages t toata from glaister, rentoul e. medical jurisprudence and toxicology, th ed. e & s livingstone, edinburgh, with permission. twithin h after consumption of diluted alcohol (approx %) on an empty stomach, assuming body weight of - pounds ( . - . kg) reproduced from ( ) with permission. *one ounce (about ml) of whiskey or z (about ml) of beer. what is the toxicity of alcohol other than ethanol? in general, the toxicity increases as the number of carbon atoms in the alcohol increases. thus, butyl alcohol is two times as toxic as ethyl alcohol: but isopropyl alcohol is only twothirds as toxic as isobutyl alcohol and one-half as toxic as amyl alcohol. primary alcohols are more toxic than the corresponding secondary isomers ( ) . anemia, hemolytic synonyms and related terms: acquired hemolytic anemia; extracorpuscular hemolytic anemia; hereditary hemolytic anemia (hereditary elliptocytosis, pyropoikilocytosis, stomatocytosis. spherocytosis); immunohemolytic anemia; intracor-puscular hemolytic anemia; microangiopathic hemolytic anemia; spur cell anemia. possible associated conditions: disseminated intravascular coagulation;* eclampsia;* glucose- -phosphatase deficiency (g pd); hemolytic uremic syndrome;* malignant hypertension; lymphoma* and other malignancies; paroxysmal nocturnal hemo-globinuria; sickle cell disease;*thalassemia;* thrombotic thrombocytopenic purpura.* (see also below under "note.") note: hemolysis also may be caused by conditions such as poisoning with chemicals or drugs, heat injury, snake bite,* or infections or may develop as a transfusion reaction* or be secondary to adenocarcinoma, heart valve prostheses (see below), liver disease (see below), renal disease, or congenital erythropoietic porphyria. * procedures prepare skeletal roentgenograms. jaundice; skin ulcers over malleoli. in young patients: thickening of frontal and parietal bones with loss of outer table ("hairon-end" appearance); paravertebral masses caused by extramedullary hematopoiesis; deformities of metacarpals, metatarsals, and phalanges. osteonecrosis* of femoral heads. remove and place in fixative as early as possible in order to minimize autolysis (alternatively, formalin can be injected in situ; see below). samples should include oxyntic corpus and fundus mucosa. record weights. submit tissue samples for histologic study. record weight of thyroid gland. for removal and specimen preparation, see chapter . request luxol fast blue stain. for removal and specimen preparation, see chapter . if there is a clinical diagnosis of anemia-related amblyopia, follow procedures described under "amblyopia, nutritional." jaundice. manifestations of malnutrition. * stomatitis with cheilosis and perianal ulcerations due to folic acid deficiency. chronic exfoliative skin disorders. vitiligo. macrocytosis; poikilocytosis; macroovalocytes; hypersegmentation of leukocytes; abnormal platelets. atrophic glossitis with ulcers. pharyngoesophagitis (folic acid deficiency). previous total or subtotal gastrectomy. carcinoma of stomach. autoimmune gastritis (diffuse corporal atrophic gastritis) with intestinal metaplasia. crohn's disease;* sprue;* other chronic inflammatory disorders; jejunal diverticula; intestinal malignancies; fish tapeworm infestation; previous intestinal resection or blind intestinal loop; enteric fistulas. hepatosplenomegaly. alcoholic liver disease. * giant epithelial cells. hyperthyroid goiter; thyroiditis. demyelination of cerebral white matter (in advanced cases). demyelination in posterior and lateral columns of spinal cord, most frequently in thoracic and cervical segments. demyelination of peripheral nerves. retinal hemorrhages; demyelination of optic nerves. hypercellular; megaloblastic. myeloproliferative disorder. brain other organs if mycotic aneurysms are expected and microbiologic studies are intended, follow procedures described below under "aneurysm, mycotic aortic." request verhoeff-van gieson, gram, and grocott's methenamine silver stains. for cerebral arteriography, see chapter . if arteriography cannot be carried out, rinse fresh blood gently from base of brain until aneurysm can be identified. record site of rupture and estimated amount of extravascular blood. for paraffin embedding of aneurysms, careful positioning is required. expected findings depend on type of aneurysm. mycotic aneurysms are often multiple and deep in brain substance. berry aneurysms are the most frequent types and often are multiple. most frequent sites are the bifurcations and trifurcations of the circle of willis. saccular atherosclerotic aneurysms are more common than dissecting aneurysms, which are very rare. with congenital cerebral artery aneurysm: coarctation of aorta;* manifestations of hypertension;* and polycystic renal disease. with mycotic aneurysm: infective endocarditis;* pulmonary suppurative processes; and pyemia. aneurysm, dissecting aortic (see "dissection, aortic.") aneurysm, membranous septum of heart note: for general dissection techniques, see chapter . most aneurysms ofthe membranous septum probably repre-sent spontaneous closure of a membranous ventricular septal defect by the septalleafiet of the tricuspid valve. aneurysm, mycotic aortic note: (i) collect all tissues that appear to be infected. ( ) request aerobic, anaerobic, and fungal cultures. ( ) request gram and grocott methenamine silver stains. ( ) no special precautions are indicated. ( ) no serologic studies are available. ( ) this is not a reportable disease. chest and abdominal organs aorta other organs submit blood samples for bacterial culture. en masse removal of adjacent organs is recommended. photograph all grossly identifiable lesions. aspirate material from aneurysm or para-aortic abscess and submit for culture. prepare sections and smears of wall of aneurysm and of aorta distant from aneurysm. request verhoeffvan gieson and gram stains. septicemia and infective endocarditis. * streptococcus, staphylococcus, spirochetes, and salmonella can be found in mycotic aneurysm. para-aortic abscess. septic emboli with infarction or abscess formation. aneurysm, syphilitic aortic part ii / diseases and conditions heart and aorta other organs en masse removal of organs is recommended. for coronary arteriography, see chapter . request verhoeff-van gieson stain from sections at different levels of aorta, adjacent great vessels, and coronary arteries. see also under "syphilis." aneurysm usually in ascending aorta. may erode adjacent bone (sternum). syphilitic aortitis may cause intimal wrinkling, narrowing of coronary ostia, and shortening of aortic cusps. disruption of medial elastic fibrils. aortic valvulitis and insufficiency;* syphilitic coronary arteritis; syphilitic myocarditis. external examination aorta prepare chest and abdominal roentgenograms. open aorta along line of blood flow, or bisect into anterior and posterior halves. photograph tear(s). measure bloody effusions in body cavities. measure or estimate amount of blood in mediastinum. request verhoeff-van gieson stain. cutaneous impact trauma. mediastinum widened by hemorrhage in case of tarnponaded dissection. a bleed into a body cavity of less-thanexsanguinating volume should point to an alternate mechanism of death such as neurogenic shock or lethal concussion; a posterior neck dissection may be required in such instances. microscopy may show transmural rupture, false aneurysm, or localized dissection. angiitis (see "arteritis, all types or type unspecified.") angina pectoris note: see under "disease, ischemic heart" and chapter . angiokeratoma corporis dittusum (see "disease, fabry's.") angiomatosis, encephalotrigeminal (see "disease, sturge-weber-dimitri.") angiopathy, congophilic cerebral synonyms and related terms: beta amyloid angiopathy due to~-amyloid peptide deposition (~a ) (associated with alzheimer's disease; hereditary cerebral hemorrhage with amyloid angiopathy of dutch type; or sporadic beta amyloid angiopathy); hereditary cerebral amyloid angiopathy, due to deposition of other amyloidogenic proteins such as cystatin c (icelandic type) and others (e.g., transthyretin, gelsolin) ( ). procedures possible or expected findings request stains for amyloid, particularly congo red, and thioflavine s (examine with polarized and ultraviolet light, respectively). request immunostain for~a . some tissue should be kept frozen for biochemical studies. multiple recent cerebral cortical infarctions or small cortical hemorrhages, or both, or massive hemispheric hemorrhages, both recent and old. amyloid deposition in leptomeninges and cortical blood vessels. senile plaques are usually present. in some cases, angiopathy is part of alzheimer's disease. * other organs a prepare material for electron microscopy. electron microscopic study permits definite confirmation of diagnosis. organs and tissues may be minimally affected by amyloidosis. anomaly, coronary artery possible associated conditions: with double outlet right ventricle; persistent truncal artery; tetralogy of fallot;* and transposition of the great arteries.* note: coronary artery between aorta and pulmonary artery, often with flap-valve angulated coronary ostium. coronary artery may communicate with cardiac chamber, coronary sinus, or other cardiac veins, or with mediastinal vessel through pericardial vessel. saccular aneurysm of coronary artery with abnor-mal flow, infective endarteritis of arteriovenous fistula, and myocardial infarction may be present. ifone or both coronary arteries originate from pulmonary trunk, myocardial infarction may be present. heart perform coronary angiography. if infective endarteritis is suspected, submit blood sample for microbiologic study. ectopic origin of coronary arteries or single coronary artery. sudden death. for a detailed description of possible additional findings, see above under "note." anomaly, ebstein's (see "malformation, ebstein's") anorexia nervosa note: sudden death from tachyarrhythmias may occur in advanced cases and thus, autopsy findings may not reveal the immediate cause of death. external examination all organs record height and weight, and prepare photographs to show cachectic features. record abnormalities as listed in righthand column. follow procedures described under "starvation." record weight of endocrine organs and submit samples for histologic study. cachexia, often with preserved breast tissue; hirsutism; dry, scaly, and yellow skin (carotenemia). mild edema may be present. parotid glands may be enlarged. manifestations of starvation.* ovaries tend to be atrophic; other endocrine organs should not show abnormalities. synonyms: cutaneous anthrax; gastrointestinal anthrax; pulmonary (inhalational) anthrax. note: ( ) collect all tissues that appear to be infected. this is a reportable disease. bioterrorism must be considered in current cases. external examination and skin blood photograph cutaneous papules, vesicles, and pustules. prepare smears and histologic sections. submit samples for bacteriologic study. submit sample for serologic study. disseminated anthrax infection may occur without skin lesions. edema of neck and anterior chest in nasopharyngeal anthrax. anthrax septicemia. see above under "note." part ii i diseases and conditions lungs gastrointestinal tracts and mesentery neck organs record character and volume of effusions. after sampling for bacteriologic study (see above under "note") perfuse one or both lungs with formalin. extensive sampling for histologic study is indicated. extensive sampling for histologic study is indicated. photograph meningeal hemorrhage in situ. pleural effusions;* hemorrhagic mediastinitis; anthrax pneumonia (inhalational anthrax; woolsorter's disease). histologic sections reveal hemorrhagic necrosis, often with minimal inflammation and gram-positive, spore-forming, encapsulated bacilli. gastrointestinal anthrax with mucosal edema and ulcerations. hemorrhagic mesenteric lymphadenitis. tongue, nasopharynx, and tonsils may be involved. hemorrhagic meningitis (hemorrhage tends to predominate). external examination distal colon and rectum photograph perineum. measure depth of anal pit, if any. dissect distal colon, rectum, and perirectal pelvic organs in situ (as much as possible). search for opening of fistulous tracts from lumen. use roentgenologic study or dissection, or both, to determine course of tract. absence of normally located anus; anal dimple. abnormal termination of the bowel into the trigone of the urinary bladder, the urethra distal to the verumontanum, the posterior wall of the vagina, the vulva, or the perineum. aortitis note: see also under "arteritis" and "aneurysm, ascending aortic." heart and aorta other organs and tissues remove heart with whole length of aorta and adjacent major arteries. record width and circumference of aorta at different levels. describe and photograph appearance of intima and of orifices of coronary arteries and other aortic branches. submit multiple samples for histologic study and request verhoeff-van gieson stain. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. secondary aortic atherosclerosis or intimal fibroplasia. widening of aorta; syphilitic aneurysm. * giant cell aortitis; rheumatoid aortitis; syphilitic aortitis; takayasu's arteritis.* manifestations of rheumatoid arthritis, * syphilis,* systemic sclerosis,* hodgkin's lymphoma, and many other diseases associated with vasculitis. external examination brain spine and spinal cord other organs prepare roentgenogram of spine. for removal and specimen preparation, see chapter . for removal of spinal cord and specimen preparation, see chapter . expose nerve roots. record appearance and photograph spinal cord in situ. submit samples of spinal cord and inflamed tissue for histologic study. request gram, gomori's iron, and grocott's methenamine silver stains. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. signs of previous spinal surgery or lumbar puncture (myelography). evidence of previous trauma or previous myelography. cerebral arachnoiditis. fibrous arachnoidal adhesions and loculated cysts. tuberculosis;* syphilis;* fungal or parasitic infection. systemic infection (see above). ascending urinary infection or other manifestations of paraplegia. arch, aortic, interrupted synonym: severe coarctation. note: the basic anomaly is a discrete imperforate region in the aortic arch, with a patent ductal artery joining the descending thoracic aorta. type a interruption is between the left subclavian and ductal arteries; type b between the left subclavian and left common carotid arteries; and type c (rare) between the left common carotid and brachiocephalic (innominate) arteries. for general dissection techniques, see part i, chapter . possible associated conditions: bicuspid aortic valve (with type a); di george syndrome* with thymic and parathyroid aplasia (with type b); hypoplasia of ascending aorta (with all types); persistent truncal artery (truncus arteriosus); ventricular septal defect. arrhythmia, cardiac note: see also under "death, sudden cardiac." toxicologic studies may be indicated, for instance, if digitalis toxicity (see "poisoning, digitalis") is suspected. if a cardiac pacemaker had been implanted, the instrument should be tested for malfunction. arteriosclerosis (see "atherosclerosis.") arteritis, all types or type unspecified synonyms and related terms: allergic angiitis and granulomatosis (churg-strauss);* allergic vasculitis; anaphylactoid purpura* and its synonyms; angiitis; buerger's disease;* cranial arteritis; giant cell arteritis;* granulomatous arteritis (angiitis); hypersensitivity angiitis; infectious angiitis; necrotizing arteritis; polyarteritis nodosa;* rheumatic arteritis; rheumatoid arteritis, syphilitic arteritis; takayasu's arteritis;* temporal arteritis; thromboangiitis obliterans; and others (see also below under "note"). note: autopsy procedures depend on ( ) the expected type of arteritis, such as giant cell arteritis,* polyarteritis nodosa,* or thromboangiitis obliterans (buerger's disease*); and ( ) the nature of suspected associated or underlying disease, such as aortic arch syndrome,* beh~et's syndrome,* cogan's syndrome, degos' disease,* dermatomyositis,* erythema nodosum and multiforme,* goodpasture's syndrome,* polymyositis, rheumatic fever, * rheumatoid arthritis,* syphilis,* and other nonspecific infectious diseases, systemic lupus erythematosus,* systemic sclerosis (scleroderma),* or takayasu's disease. for histologic study of blood vessels, verhoeff-van gieson stain or a similar stain is recommended. temporal and ophthalmic arteritis. arteritis of ciliary and retinal vessels. clinically, polymyalgia. anemia. arteritis, takayasu's synonyms: aortic arch syndrome; pulseless disease. external examination heart, aorta, and adjacent great vessels kidney eyes and optic nerve brain for in situ aortography, clamp distal descending thoracic aorta and neck vessels as distal as possible from takeoff at aortic arch. remove heart together with aorta and long sleeves of neck vessels. for coronary arteriography, see chapter (method designed to show coronary ostia). test competence of aortic valve. open aortic arch anteriorly and measure (with calipers) lumen at origin of great neck vessels. photograph aorta and neck vessels and submit samples for histologic study. request verhoeffvan gieson stain. submit tissue for histologic examination. for removal and specimen preparation, see chapter . for removal and specimen preparation, see chapter . facial muscular atrophy and pigmentation. narrowing at origin of brachiocephalic arteries. dilated ascending aorta. narrowing of coronary arteries at origins. myocardial infarction. aortic insufficiency. * aortic atherosclerosis. thromboses of brachiocephalic arteries. giant cell arteritis. * diffuse mesangial proliferative glomeulonephritis ( ) . atrophy of optic nerve, retina, and iris; cataracts; retinal pigmentation. ischemic lesions. artery, patent ductal synonym: patent ductus arteriosus. note: the basic anomaly is persistent postnatal patency of the ductal artery, usually as an isolated finding (in % of cases in infants, and in % in adults). it is more common in premature than full-term infants and at high altitudes than at sea level. possible complications in unoperated cases include congestive heart failure, * plexogenic pulmonary hypertension,* ductal artery aneurysm or rupture, fatal pulmonary embolism,* or sudden death. in some conditions, such as aortic atresia* or transposition with an intact ventricular septum,* ductal patency may be necessary for survival. possible associated conditions: atrial or ventricular septal defect;* coarctation ofthe aorta;* conotruncal anomalies; necrotizing enterocolitis in premature infants; postrubella syndrome; and valvular or vascular obstructions. artery, persistent truncal synonym and related terms: type i, pulmonary arteries arise from single pulmonary trunk (in %); type , pulmonary arteries arise separately but close-by (in %); type , pulmonary arteries arise separately but distal from one another (in %). note: the basic anomaly is a common truncal artery, with truncal valve, giving rise to aorta, pulmonary arteries, and coronary arteries, usually with a ventricular septal defect. interventions include complete rastelli-type repair, with closure of ventricular septal defect, and insertion of valved extracardiac conduit between right ventricle and detached pulmonary arteries. possible associated conditions: absent pulmonary artery (in %); atrial septal defect (in %); absent ductal artery (in %); coronary ostial anomalies (in %); di george syndrome;* double aortic arch; extracardiac anomalies (in %); interrupted aortic arch* (in %); right aortic arch (in %); truncal valve insufficiency (uncommon) or stenosis (rare); trun-cal valve with three (in %), four (in %), or two (in %) cusps. heart and great vessels if infective endocarditis is suspected, follow culture procedures for endocardial vegetation described in chapter . request verhoeff-van gieson stain. infective endocarditis,* usually of truncal valve. late postoperative conduit obstruction. postoperative late progressive truncal artery dilation with truncal valve insufficiency. hypertensive pulmonary vascular disease. cerebral abscess,* if right-to-ieft-shunt was present. arthritis, all types or type unspecified note: for extra-articular changes, see under the name of the suspected underlying conditions. infectious diseases that may be associated with arthritis include bacillary dysentery, * brucellosis, * gonorrhea, rubella,* syphilis, * tuberculosis, * typhoid fever, * and varicella. * noninfectious diseases in this category include acromegaly,* beh<;et's syndrome,* felty's syndrome,* gout,* rheumatoid arthritis,* and many others, too numerous to mention. remove synovial fluid and prepare smears. submit synovial fluid for microbiologic and chemical study. for removal of joints, prosthetic repair, and specimen preparation, see chapter . for removal and specimen preparation, see chapter . in the polyarticular variant, facial asymmetry may be noted. rheumatoid factor positive in some cases. pericarditis.* interstitial pneumonitis; pleuritis. (see also under "arthritis, rheumatoid.") lymphadenopathy. splenomegaly. monarthritis or severe, erosive polyarthritis; see also under "arthritis, rheumatoid" and above under "externalexamination and skin." ankylosing spondylitis* may be present. chronic iridocyclitis. see "arthritis, rheumatoid." arthritis, rheumatoid synonyms and related terms: ankylosing spondylitis;* felty's syndrome;* juvenile rheumatoid arthritis* (still's disease); rheumatoid disease; and others. possible associated conditions: amyloidosis;* polymyositis (dermatomyositis*); psoriasis;* sjogren's syndrome;* systemic lupus erythematosus;* systemic vasculitis, and others. subcutaneous rheumatoid nodules on elbows, back, areas overlying ischial and femoral tuberosities, heads of phalangeal and metacarpal bones, and occiput. deformities and subluxation of peripheral joints (see also below under "joints"). subaxial dislocation of cervical spine may be cause of sudden death. pneumothorax;* pleural empyema.* t-cell abnormalities ( ) . bacteremia. positive rheumatoid factor. rheumatoid granulomas in myocardium (septum), pericardium, and at base of aortic and mitral valves; constrictive pericarditis;* aortic stenosis;* coronary arteritis. systemic vasculitis (arteritis*). rheumatoid granulomas in pleura and lung (with pneumoconiosis*); bronchopleural fistula; rheumatoid pneumonia with interstitial pulmonary fibrosis and honeycombing; bronchiectasis;* bronchiolitis with cystic changes; pulmonary arteritis. pneumoconiosis* in caplan arthrogryposis ( ) may be a primary muscle disease, or it may involve abnormalities of the brain, spinal cord, and/or peripheral nerves. etiologies are numerous, as are the modes of inheritance. critical to making the appropriate diagnosis is the collection of muscles from various sites for routine histology, muscle histochemistry, and electron microscopy. portions of peripheral motor nerves must also be prepared for histology and electron microscopy. abdominal cavity intra-abdominal lymphatic system puncture abdominal cavity and submit fluid for microbiologic study. record volume of exudate or transudate and submit sample for determination of fat and cholesterol content. prior to routine dissection, lymphangiography (see below) may be indicated. possible associated conditions: with pulmonary aspergillosis-bronchiectasis; * bronchocentric granulomatosis;* sarcoidosis;* tuberculosis. * with systemic aspergillosisleukemia;* lymphoma;* and other conditions complicated by immunosuppression (l, ) . other organs a carefully make multiple parasagittal sections through the unperfused lungs. culture areas of consolidation. if diagnosis was confirmed, perfuse lungs with formalin. prepare histologic sections from walls of cavities, cavity contents, and pneumonic infiltrates. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. assault note: all procedures described under "homicide" must be followed. asthma note: spray death* may occur in asthma sufferers from pressurized aerosol bronchodilators. record thickness and position. perfuse one lung with formalin. because mucous plugs may block bronchial tree, attach perfusion apparatus to pulmonary artery or to bronchus and pulmonary artery. monitor perfusion to ensure proper inflation. prepare photograph of fixed cut section. submit samples of pulmonary parenchyma and bronchi for histologic study. request azure-eosin and verhoeff-van gieson stains. record weight and thickness of walls. leave attached to stomach. photograph and submit samples for histologic study. eczema. conjunctival hemorrhages and subcutaneous emphysema may be present after fatal attack. pneumothorax;* mediastinal emphysema. low diaphragm (see below). increased igeconcentrations in fatal asthma; postmortem tryptase determination is of doubtful value in this regard ( ) . hypertrophy. low position of diaphragm. hyperinflated lungs. thick-walled bronchi with prominent viscid mucous plugs. typical microscopic inflammatory changes ( ) . asthmatic bronchitis with eosinophilic infiltrates. bronchocentric granulomatosis.* pulmonary atherosclerosis with breakup of elastic fibers. paucity of ecosinophils in mucous ( ) . cor pulmonale. refl ux esophagitis ( ) . peptic ulcer. * pneumatosis of small intestine; emphysema of colon. centrilobular congestion and necrosis. petechial hemorrhages in hypothalamus; necrosis of cerebellar folia; anoxic changes in cortex, globus pallidus, thalamus, sommer's sector of hippocampus, and purkinje cells of cerebellum. suspected changes in anterior hom cells of spinal cord in patients with asthma-associated poliomyelitis-like illness (hopkins syndrome) ( ). allergic polyps and other allergic inflammatory changes ( ) . increased erythropoiesis. atresia, aortic valvular synonym: aortic atresia; aortic atresia with intact ventricular septum; hypoplastic left heart syndrome. note: the basic anomaly is an imperforate aortic valve, with secondary hypoplasia ofleft-sided chambers and ascending aorta. for possible surgical interventions, see two-stage norwood and modified fontan procedures in chapter . possible associated conditions: atrial septal defect* (or patent foramen ovale, usually restrictive); dilatation of myocardial sinusoids thatcommunicate with coronary vessels; dilatation of right atrium, right ventricle, and pulmonary trunk; fibroelastosis ofleft atrial and left ventricular endocardium; hypertrophy of ventricular and atrial walls; hypoplastic left atrium, mitral valve, left ventricle, and ascending aorta; mitral atresia* with minute left ventricle; patent ductal artery (ductus arteriosus); small left ventricle with hypertrophic wall; tubular hypoplasia of aortic arch, with or without discrete coarctation. synonyms and related terms: congenital biliary atresia; extrahepatic biliary atresia; infantile obstructive cholangio-pathy; syndromic (alagille's syndrome) or nonsyndromic paucity of intrahepatic bile ducts ("intrahepatic" biliary atresia). possible associated conditions: alpha]-antitrypsin deficiency;* choledochal cyst;* congenital rubella syndrome;* polysplenia syndrome* ( ); small bowel atresia; trisomy - ; trisomy ; turner's syndrome;* viral infections (cytomegalovirus infection;* rubella*). dissect extrahepatic bile ducts in situ or leave hepatoduodenalligament intact for later fixation and sectioning (see below). record appearance and contents of gallbladder and course of cystic duct. in postoperative cases, submit sample of anastomosed hepatic hilar tissue for demonstration of microscopic bile ducts. remove liver with hepatoduodenalligament. prepare horizontal sections through ligament and submit for histologic identification of ducts or duct remnants. prepare frontal slices of liver and sample for histologic study. request pas stain with diastase digestion. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. jaundice. congenital rubella and other viral infections. alpha]-antitrypsin deficiency;* defects in bile acid synthesis. chromosomal abnormalities. in atresia of the hepatic duct, the gallbladder will be empty. in isolated atresia of the common bile duct, the gallbladder contains bile but it cannot be squeezed into the duodenum. atresia or hypoplasia of bile duct(s); choledochal cyst(s). biliary drainage created by kasai operation. obliterative cholangiopathy ( ) . intrahepatic cholelithiasis; postoperative ascending cholangitis; secondary biliary cirrhosis; giant cell transformation; paucity of intrahepatic bile ducts. pas-positive inclusions in alphal-antitrypsin deficiency.* polysplenia syndrome* ( ) with malrotation, situs inversus, preduodenal portal vein, absent inferior vena cava, anomalous hepatic artery supply, and cardiac defects. for other abnormalities outside the biliary tree, see under "possible associated conditions"). nephromegaly ( ) . atresia, cardiac valves (see "atresia, aortic valvular," "atresia, mitral valvular," "atresia pulmonary valvular, with intact ventricular septum," "atresia, pulmonary valvular, with ventricular septal defect," and "atresia, tricuspid valvular.") atresia, duodenal possible associated conditions: with membranous obstruction of the duodenum-annular pancreas; atresia of esophagus* with tracheoesophageal fistula; congenital heart disease; cystic fibrosis;* down's syndrome;* hirschsprung's disease; imperforate anus* or other congenital obstructions of the intestinal tract ( ); intestinal malrotation; lumbosacral, rib-, and digitllimb anomalies; single umbilical artery; spinal defects; undescended testis ( ). see also under "atresia, small intestinal." the basic anomaly is an imperforate pulmonary valve, with a hypoplastic right ventricle. in unoperated cases, ductal closure is the most common cause of death. for possible surgical interventions, see modified blalock-taussig shunt, mod-ified fontan procedure, and pulmonary valvulotomy in chapter . for general dissection techniques, see chapter . possible associated conditions: dilated myocardial sinusoids that may communicate with epicardial coronary arteries or veins; patent ductal artery (ductus arteriosus); patent oval foramen (foramen orale); tricuspid atresia with minute right ven-tricle; tricuspid stenosis with hypoplastic right ventricle (in %); tricuspid insufficiency with dilated right ventricle (in %). synonym: tetralogy of fallot with pulmonary atresia. note: the basic anomaly is atresia of the pulmonary valve and ofvariable length ofpulmonary artery, and ventricular septal defect (membranous or outlet type), with overriding aorta, and with pulmonary blood supply from ductal or systemic collateral arteries. for possible surgical interventions, see rastelli-type repair and unifocalization of multiple collateral arteries in chapter . possible associated conditions: right ventricular outflow tract a short blind-ended pouch ( %) or absent ( %); atresia of pulmonary artery bifurcation, with nonconfluent pulmonary arteries; right aortic arch ( %); atrial septal defect ( %); persistent left superior vena cava; anomalous pulmonary venous connection; tricuspid stenosis or atresia; complete atrioventricular septal defect; transposed great arteries; double inlet left ventricle; asplenia, polysplenia, or velocardiofacial syndromes; dilated ascending aorta, with aortic insufficiency. related term: jejuno-ileal atresia. possible associated findings: esophageal atresia* with tracheoesophageal fistula; lumbosacral, rib-, or digit/limb anom -alies; undescended testes (l) . note: see also under "atresia, duodena ." fascia lata, blood, or liver these specimens should be collected using aseptic technique for tissue culture for chromosome analysis (see chapter ) . intestinal tract for mesenteric angiography, see chapter . leave mesentery attached to small bowel, particularly to the atretic portion. trisomy . multiple atresias; proximal dilatation; volvulus; malrotation; meconium impaction; other evidence of cystic fibrosis. anorectal malformation (l) . annular pancreas ( ). atresia, tricuspid valvular note: the basic anomaly is an absent right atrioventricular connection ( %) or imperforate tricuspid valve ( %), with a hypoplastic right ventricle ( %), muscular ventricular septal defect ( %) that is restrictive ( %), and a patent oval atresia, urethral foramen ( %) or secundum atrial septal defect ( %). for possible surgical interventions, see modified fontan or glenn procedures in chapter . for general dissection techniques, see chapter . possible associated conditions: juxtaposed atrial appendages; large left ventricular valvular orifice; large left ventricular chamber; persistent left superior vena cava; pulmonary atresia; transposition of the great arteries ( %), with aortic co-arctation ( % of those); anomalies of musculoskeletal or digestive systems ( %); down's,* asplenia, or other syndromes. heart aorta and cervical arteries brain if infective endocarditis* is suspected, culture using the method described in chapter . for dissection of carotid and vertebral arteries, see chapter . for removal and specimen preparation, and cerebral anteriography, see chapter . if a foreign body is discovered during a medicolegal autopsy or if the discovery of a foreign body may have medicolegal impli-cations (e.g., presence of a surgical instrument in the abdominal cavity), the rules of the chain of custody apply. for the handling of bullets or bullet fragments, see "injury, firearm." if analysis offoreign material is required, commercial laboratories may be helpful. bolus (see "obstruction, acute airway!') burns note: fatal bums should be reported to the medical examiner's or coroner's office. the questions to be answered by the pathologist depend on whether the incident was accidental, sui-cidal, or homicidal, and whether the victim survivied to be treated in the hospital. a pending death certificate should be issued if the fire and police investigators are not sure of the circumstances at the time of the autopsy. for electrical bums, see under "injury, electrical." for victims who were treated at the hospital, autopsy procedures should be directed toward the discovery or confirmation of the mechanism of death, such as sepsis or pulmonary embolism.* death can be caused primarily by heart disease, with other-wise minor bums and smoke inhalation serving as the trigger that leads to lethal ventricular arrhythmia. because carbon monoxide concentrations are halved approx every min with % oxygen therapy, the pathologist must obtain the first clinical laboratory test results for co-hemoglobin. soot can be detected with the naked eye or d after inhalation of smoke. ambulance records should be examined to determine whether a persistent coma might have been caused by hypoxic encephalopathy following resuscitation from cardiac arrest at the scene. admission blood samples should be acquired to test for cohemoglobin and alcohol. this may not have been done in the emergency room. persons suffering from chronic alcoholism succumb to fire deaths more often than persons who do not drink. a very high initial serum alcohol concentration suggests a risk factor for the fire and presence of chronic alcoholism. patients with chronic alcoholism typically are deprived of alcohol when they are in the bum unit and this can cause sudden, presumably cardiac, death,just as it occurs under similarcircum-stances, not complicated by bums. under these circumstances, the heart fails to show major abnormalities. this mode of dying seems to have no relationship to the presence or absence of liver disease. if the body is found dead and charred at the scene, prepare whole body roentgenograms, before and after removal of remanants of clothing. see also under "identification of the body" and "external examination" in chapter ). one or two fingerpads may yield sufficient ridge detail for identification. if this is not possible, ante-and postmortem somatic and dental radiographs must be compared for identification, or dna comparison must be used. external examination, heart and lungs abdominal cavity and liver see below under "cardiomyopathy, dilated." record volume of ascites. record actual and expected weight of liver. request iron stain. see below under "cardiomyopathy, dilated." alcoholic cirrhosis and alcoholic cardiomyopathy rarely coexist. however, in genetic hemochromatosis,* cirrhosis and heart failure are common findings. cardiomyopathy, dilated (idiopathic, familial, and secondary types) note: for general dissection techniques, see chapter . external examination heart other organs and tissues record actual and expected weights. record ventricular thicknesses and valvular circumferences. evaluate relative atrial and ventricular chamber sizes. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. note: huntington's disease maps to the short arm of chromosome . the gene is widely expressed but of unknown function; it contains a cag repeat sequence, which is expanded (range, to ) in patients with huntington's disease. a sensitive diag-nostic test is based on the determination of this cag sequence, which can be done on fresh-frozen tissue or blood ( ) . in the absence of genetic confirmation, sampling of organs and tissues cannot be excessive because a complex differential diagnosis must be resolved. note: disseminated intravascular coagulation (dic) often is a complication of obstetrical mishaps such as abruptio placentae or amniotic fluid embolism,* or it complicates malignancies (such as adenocarcinomas or leukemia*) or bacterial, viral, and other infections. other conditions such as aortic aneurysm* or hemolytic uremic syndrome* are known causes also. ifthe nature of the underlying disease is known, follow the procedures under the appropriate heading also. note: this is a cause of diarrhea. microscopic colitis is associated with older age; collagenous colitis is associated with female sex ( ). the colon is grossly normal but microscopically, increased lymphocytes in the lamina propria and a subepithelial band of collagen is found. if only the lymphocytic infiltrate is found, the term "lymphocytic colitis" or "microscopic colitis" should be applied. a trichrome stain should be ordered in all instances, because the collagen band may be difficult to see without the special stain. i death, anaphylactic synonym: generalized anaphylaxis. note: autopsy should be done as soon as possible after death. neck organs should be removed before embalming. if death is believed to be caused by drug anaphylaxis, inquire about type of drug(s), drug dose, and route of administration (intravenous, intramuscular, and oral or other). this will determine proper sampling procedures-for instance, after penicillin anaphylaxis. allergy to bee stings, wasp stings, fire ants, and certain plants may also be responsible for anaphylaxis. however, envenomation also can be fatal in the absence of anaphylaxis. external examination search for injection sites or sting marks. if such lesions are present, photograph and excise with -cm margin. freeze excised tissue at - °c for possible analysis. prepare chest roentgenogram. foam in front of mouth and nostrils. swelling of involved tissue. antigen-antibody reaction in involved tissues. antibodies against suspected antigen. laryngeal edema may recede soon after death. foamy edema in trachea and bronchi; diffuse or focal pulmonary distention ("acute emphysema") alternating with collapse; pulmonary edema and congestion; accumulation of eosinophilic leukocytes. eosinophilic leukocytes in red pulp. death, anesthesia-associated. note: there are many possible causes of anesthesiaassociated death that are not drug-related, such as acute airway obstruction* by external compression, aspiration, arrhythmia of a heart not previously known to be diseased, tumor, or an inflammatory process. some ofthe complications are characteristically linked to a specific phase of the anesthesia, and many are not revealed by customary morphologic techniques. the task for the pathologist charged with investigating an anesthesia-associated death is to reconstruct the chain of physiologic events culminating in cessation of vital signs. autopsy morphology plays a supporting role; the main investigations center around the record left by the anesthesiologist, testing of anesthesia equipment, and toxicological testing. a consulting anesthesiologist can divine much more information from the anesthesia and recovery room records than can the pathologist, and can suggest avenues of further investigation. therefore, the most important step in these autopsies is to obtain the anesthesiaassociated records and to secure the consulting services of an independent anesthesiologist. the changes in the vital signs during and after anesthesia will help to focus the investigation toward a cardiac mechanism ofdeath or depression ofbrainstem function as a terminal mechanism. when information is gathered about drugs and chemical agents that have been administered or to which the victim may have had access, the pathologist must keep in mind that some non-medical chemicals and many drugs are known to affect anesthesia. drugs and their metabolic products, additives, stabilizers, impurities, and deterioration products (one of which can be carbon monoxide) may be present and can be identified in postmortem tissues. therefore, all appropriate body fluids and solid tissue should be submitted for toxicological examination. if the anesthetic agent was injected into or near the spinal canal, spinal fluid should be withdrawn from above the injected site into a standard toxicologist's collection tube with fluoride preservative. if the anesthetic agent was injected locally, tissue should be excised around the needle puncture marks at a radius of - em. serial postmortem analysis of specimens may permit extrapolation to tissue concentration at the time of death. the time interval between drug administration and death sometimes can be calculated from the distribution and ratio ofadministered drugs and their metabolic products. for a review of anesthetic death investigation, see ref. ( ) . halothane anesthesia and some other anesthetic agents may cause fulminant hepatitis and hepatic failure. the autopsy procedures suggested under "hepatitis, viral" should be followed. note: for special autopsy procedures in postoperative deaths, see chapter . in some instances, procedures described under "death, anesthesia-associated" may be indicated. for a review of investigational procedures and autopsy techniques in operating-room-associated deaths, see ref. ( ) . if the autopsy will involve anatomy or dissection techniques that are unfamiliar, the pathologist should not hesitate to invite the surgeon to the autopsy. in patients who develop a cerebral infarction after open heart surgery, arterial air embolism should be considered as a possible cause. the diagnosis often must be based on excluding other causes because the air has been absorbed prior to death. if a patient dies rapidly, the hospital records may be incomplete or scanty. for example, if a patient bleeds to death despite attempted repair of hepatic lacerations, hospital records may not suffice to reach the correct cause-of-death opinion; personal accounts from the surgeon and anesthesiologist may be needed. autopsy data on patients dying following thoracic surgery may be found in ref ( ) . d death, restaurant (see "obstruction, acute airway.") death, sniffing and spray related terms: glue sniffing; sudden sniffing death syndrome. note: no anatomic abnormalities will be noted at autopsy. sudden death may occur after cardiac dysrhythmia or respiratory arrest. procedures possible or expected findings lungs brain if poison had been inhaled at the time when death occurred, tie main bronchi. submit lungs in glass container for gas analysis. submit samples of small bronchi for histologic study. for removal and specimen preparation, see chapter . submit samples of fresh or frozen brain for toxicologic study. submit samples in glass containers (not plastic) for toxicologic study. trichloroethane, fluorinated refrigerants, and other volatile hydrocarbons are most often involved in the "sudden sniffing death syndrome." spray death may occur in asthma sufferers using pressurized aerosol bronchodilators. freons and related propellants may also be responsible for sudden death. toxic components of glue-such as toluene-accumulate in the brain of glue sniffers. also present in various glues are acetone, aliphatic acetates, cyclohexane, hexane, isopropanol, methylethyl ketone, and methylisobutyl ketone. aerosols may occlude the airway by freezing the larynx. carbon tetrachloride sniffing may cause hepatorenal syndrome (see also under "poisoning, carbon tetrachloride"). death, sudden unexpected, of adult note: medicolegal autopsies are usually indicated, and appropriate procedures should be followed. ifanaphylactic death is suspected, see also under that heading. for all unexpected deaths, the pathologist should learn the circumstances of the death, in order to determine whether the mechanism of death was rapid or slow, and to guide the selection of ancillary tests. whenever paramedics attended a person, the run sheet should be obtained to look for a history of recent drinking or ofchronic alcoholism may be an important clue. the combination of a history ofalcoholism, a negative test for ethanol, and absence ofcardiovascular disease, should suggest alcohol withdrawal as the cause ofa sudden death. the list of"possible or expected findings" below is not complete. for general toxicologic sampling, see chapter . possible associated conditions: atrial septal defect;*bicuspid aortic valve;* coarctation,* hypoplasia, or interruption (type a) of aortic arch; coronary artery from main pulmonary artery; right atrial arch; patent ductal artery;* right pulmonary artery from ascending aorta; subaortic stenosis;* tetralogy of fallot;* ventricular septal defect. * (in approx % of the cases, one or more of these associated conditions are found.) defect, atrial septal note: the basic anomaly is a defect of the atrial septum, usually at the oval fossa (in %). possible complications in unoperated cases include atrial arrhythmias, congestive heart failure; paradoxic embolism; plexogenic pulmonary hypertension « %), and pulmonary artery aneurysm. possible surgical interventions include surgical and transcatheter closure of defect. for deficiency, vitamin c synonyms: hypovitaminosis c; scurvy. external examination and skin other organs bones, joints, and soft tissues record extent and character of skin lesions; prepare sections of skin. describe appearance of gums, and prepare sections. record evidence of bleeding. for removal, prosthetic repair, and specimen preparation of bones and joints, see chapter . hyperkeratotic hair follicles with perifollicular hemorrhages (posterior thighs, anterior forearms, abdomen); petechiae and ecchymoses (inner and posterior thighs); subcutaneous hemorrhages. gingivitis. in rare instances, gastrointestinal or genitourinary hemorrhages. hemorrhages into muscles and joints. subperiosteal hemorrhages occur primarily in distal femora, proximal humeri, tibiae, and costochondral junctions (scorbutic rosary). deficiency, vitamin d synonyms: hypovitaminosis d; rickets. note: features or rickets may be found in familial hypophosphatemia (vitamin d-resistent rickets; fanconi syndrome). vitreous or blood (serum) other organs prepare skeletal roentgenograms. in infants with suspected rickets, record size of anterior fontanelle and shape of head; state of dentition; and shape of costochondral junctions, wrists, long bones, and spine. submit samples for calcium, magnesium, and phosphate determination. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. weigh parathyroid glands and submit samples for histologic study. submit samples of intestine for histologic study. for removal, prosthetic repair, and specimen preparation, see chapter . in infantile rickets, diagnostic sites for histologic sampling are costochondral junctions, distal ends of radius and ulna, and proximal ends of tibia and humerus. for adults, see under "osteomalacia." in infants, rachitic changes at costochondral junctions; in adults, osteoporosis* and osteomalacia*-with or without pseudofractures (milkman's syndrome ( ) . note: the term spinocerebellar degeneration encompasses a variety of lesions whose classification is controversial. a new approach has come from linkage analysis and molecular biology. for instance, friedreich's ataxia, the classic form of hereditary ataxia, is due to an intronic expansion of a gaa tri-nucleotide repeat. other forms are also identified by their specific gene loci. neuropathologic examination still is important and ample sampling is suggested, which should include cerebral cortex, basal ganglia (caudate nucleus, putamen, and globus pallidus), thalamus, subthalamic nucleus, midbrain (red nucleus and substantia nigra), pons (pontine nuclei), spinal cord (at cer-vical, thoracic, and lumbar levels), optic tract, optic nerves with lateral geniculate nucleus, and sensory and motor peripheral nerves. for removal and specimen preparation, see chapter . enlargement of head. poor demarcation between cortex and gelatinous white matter. extensive demyelination and vacuolation of white matter, particularly subcortically. optic atrophy. degeneration, striatonigral (see "atrophy, multiple system.") related term: thirst. note: possible underlying conditions not related to inaccessibility of water include bums, exposure to heat, gastrointestinal diseases, recent paracentesis, renal diseases, and use of diuretic drugs. see also under "disorder, electrolyte(s)." external examination vitreous urine prepare histologic sections of blisters, ulcers, or skin abrasions. submit sample for sodium, chloride, and urea nitrogen determination. skin turgor may be decreased and eyes may be sunken. microscopic changes help to decide whether skin lesions are antemortem or postmortem. sodium concentrations more than meqll, chloride concentrations more than meq/ and urea nitrogen concentrations between and meq/dl indicate dehydration. absence or minimal amount of urine. dementia (see "disease, alzheimer's.") drug abuse, amphetamine(s) note: methamphetamine abuse may be suggested by poor condition of the dentition. methylenedioxymethamphetamine ("ecstasy") abuse is often suggested by friends with whom the decedent was abusing drugs. follow procedures described under "dependence, drug(s)." drug abuse, cocaine note: cocaine is spontaneously hydrolyzed by blood esterases, even after death. however, one of its major metabolite, benzoylecgonine, is routinely identifiable by immunoassay screening tests. when cocaine is abused concurrently with heroin or other depressant drugs, it may be difficult to ascribe deth to a single agent, unless circumstances clearly point to a rapid cardiac mechanism or a slow brainstem depression mechanism. note: if narcotic paraphernalia and samples of the drug itself are found at the scene of the death, they should be submitted for analysis. helpful information about the nature of a drug may be obtained from witnesses. state crime laboratories may provide much assistance. if name of drug is known, see also under "poisoning,..." the slang name of a drug may be insufficient for identification because these names often are used for different compounds at different times of places. opoid narcotics can be injected intravenously, or subcutaneously, or snorted. death may occur with such speed that the bodies may be found with needles and syringes in the veins or clenched in the hands. drug abuse may be associated with a multitude of local (see below) or systemic complications, including malaria* and tetanus. * as stated in chapter , for a growing number of analytes, most notably tricyclic antidepressants, peripheral blood is preferred over central blood. peripheral blood is aspirated by percutaneous puncture before autopsy, from the femoral vein or the subclavian vein. the authors prefer the femoral approach in order to avoid any question of artifact in the diagnosis of venous air embolism. it may be pru-dent to add naf to some of the samples. related term: childhood dermatomyositis (or polymyositis) associated with vasculitis; dermatomyositis (or polymyositis) associated with neoplasia or collagen vascular disease; primary idiopathic dermatomyositis; primary idiopathic polymyositis. possible associated conditions: carcinoma (lung, stomach, intestine, and prostate in males; breast, ovary, and uterus in females; miscellaneous sites in both sexes); lymphoma* (rare) and other malignancies ( ); lupus erythematosus;* mixed connective tissue disease; progressive systemic sclerosis;* rheumatoid arthritis;* sjogren's syndrome;* and others. vasculitis of childhood polymyositis (dermatomyositis). external examination and skin heart lungs esophagus and gastrointestinal tract photograph grossly involved skin. prepare sections of involved (anterior chest, knuckles, knees) and grossly uninvolved skin and subcutaneous tissue. prepare roentgenograms. submit samples from myocardium for histologic study. perfuse one lung with formalin. submit samples from all segments for histologic study. arteritis* and phlebitis* with thrombosis, fibrosis, and infarctions. steatohepatitis and manifestations of diabetes mellitus* may be found ( ) . myositis with muscular atrophy and fibrosis; vasculitis in childhood cases. polyneuropathy (rare) ( ). arthritis. diabetes mellitus synonyms: type i (insulin-dependent or juvenile-onset) diabetes mellitus; type ii (insulin-independent or adult onset) diabetes mellitus; secondary diabetes mellitus (e.g., due to drugs or pancreatic disease). note: in infants of diabetic mothers, macrosomia and congenital malformations must be expected. record size and weight of placenta and total weight and length, crown to rump length, and crown to heel length of infant. compare with expected measurements (see part iii). expected histologic finding in-clude hyperpla-sia with relative increase ofb cells of the islands of langerhans with interstitial and peri-insular eosinophilic infiltrates, decid-ual changes of the endometrium, enhanced follicle growth in the ovaries, and leydig cell hyperplasia. possible associated conditions: acanthosis nigricans; acro-megaly;* amyotrophic lateral sclerosis; * ataxia telangiectasia;* fanconi's anemia;* friedreich's ataxia;* gout;* hemochro-matosis; *hyperlipoproteinemia; * hyperthroidism;* obesity;* turner's syndrome;* and many others, too numerous to mention. note: the term "caroli's syndrome" often is used for cases that also show histologic features of congenital he-patic fibrosis or other manifestations of fibropolycystic liver disease,* whereas the name "caroli's disease" refers to idiopathic dilatation of intrahepatic bile ducts, without associated abnormalities. possible associated conditions: choledochal cyst* and related extrahepatic biliary abnormalities ( ); congenital hepatic fibrosis; * cysts of kidneys (renal tubular ectasia or medullary sponge kidney; autosomal-recessive polycystic kidney disease, and rarely, autosomal-dominant polycystic kidney disease [ ] )* and of pancreas. record volume of effusions. prepare smears of fresh blood or of buffy coat, or make thick-drop preparation. submit sample for xenodiagnosis or animal inoculation and for serologic study. record weight. in chronic chagas' disease, perfuse intact heart with formalin (chapter ) and slice fixed heart in a frontal plane so as to create anterior and posterior halves. prepare photographs. histologic samples should include conduction system. include several sections of atrial (auricular) walls for histologic study of autonomous ganglia. perfuse at least one lung with formalin. leave affected hollow viscera intact and fill with formalin. cut fixed organs in half, photograph, and cut histologic sections on edge. record liver weight and submit samples for histologic study. record weight. prepare photographs of abnormalities. weigh and examine. prepare histologic sections. for removal and specimen preparation, see chapter . autopsy is desirable in suspected cases because the diagnosis can only be firmly established after neuropathologic examination. serologic studies are not available. unfortunately, all tissues (not just the brain and spinal cord) may remain infectious even after prolonged fixation and histologic processing. thus, the autopsy recommendations for most other infectious diseases do not apply here. this is a reportable disease in some states. special precautions are indicated and therefore, the procedures described here should be followed strictly ( ) ( ) ( ) ( ) : all persons in the autopsy room must wear disposable long-sleeved gowns, gloves, and masks. contamination of the autopsy table should be prevented by covering it with a disposable, non-permeable plastic sheet. autopsy generally should be restricted to the brain. if organs in the chest or abdomen need to be examined, this is best done in situ. to prevent aerosolization of potentially infectious bone dust, a hood or other protective device should be used while opening the skull with a stryker saw. after completing the autopsy, instruments and other potentially contaminated objects should be autoclaved in a steam autoclave ( h at °c). porous load is considered more effective than gravity displacement autoclaves. immerse autopsy instruments in distilled water before and during autoclaving, in order to protect them from corrosion. ifno autoclave is available, chemical disinfection (see below) is a satisfactory alternative. disposable items should be put in a container for infectious hospital waste and ultimately incinerated. contaminated objects not suitable for autoclaving (such as the stryker saw) should be soaked with a nnaoh solution for h (alternatively, nnaoh may be used for h). contaminated surfaces should be thoroughly washed with the same solution. aluminum should be treated for h with a fresh % naoci (sodium hypochlorite) solution with at least , ppm free chloride. wash waters should be collected; if no autoclave is available, n naoh or > volumes of % sodium hypochlorite bleach should be added to the water and left for a minimum of h before being discarded. before removing the body from the autopsy room, it should be sponged with % sodium hypochlorite. to deactivate cjd infectivity, tissue blocks, mm or less in thickness, should be fixed in formalin in a formalin-totissue ratio of at least : for at least h and then soaked in concentrated formic acid ( - %) for i h, followed by another h of formalin fixation. the fixation fluid should be collected and decontaminated, as described earlier for wash water. glassware and tissue carriers should also be decontaminated as previously described. after this deactivation, the tissue blocks can be processed in a routine fashion. at any stage of these procedures, special care must be taken to avoid cuts with potentially contaminated glassware, blades, or other objects. parenteral exposure to potentially contaminated material also should be avoided. remains of patients who have died of the disease should not be accepted for anatomy teaching for students. if specimens are prepared for pathology collections, they should be handled with great caution. morticians and mortuary workers should be warned of possible hazards posed by tissues of patients with transmissible spongiforme encephalopathies; they should be advised about proper use of disinfectants. clinical laboratories that receive autopsy tissues or fluids must be warned about the infectious nature of the material. if possible, decontamination should be done at the site where the autopsy was done. for the shipping of potentially infected material, see chapter . increased concentrations of nse ( ). spongiforme changes, astrocytosis, neuronal loss, amyloid plaque formation, prp deposition, and proliferation of activated microglia ( ). cerebrospinal fluid brain submit sample for neuron-specific enolase (nse). for removal and specimen preparation, see chapter and above under "note." submit fresh-frozen material for confirmation of diagnosis by histoblot technique on protease k-digested frozen tissue or western blot preparations on brain homogenates. immunohistochemical localization ofprp and hla-dr protein on paraffin-embedded tissue is possible. disease, demyelinating (see "degeneration, spongy, of white matter," "encephalomyelitis, all types or type unspecified," "leukodystrophy, globoid cell," "leukodystrophy, sudanophilic," "sclerosis, multiple;' and "sclerosis, schilder's cerebral.") disease, diffuse alveolar synonym: diffuse pulmonary disease. note: autopsy procedures are listed under the more specific diagnoses, such as "hemosiderosis, idiopathic pulmonary," "lipoproteinosis, pulmonary alveolar," "microlithiasis, pulmonary alveolar," "pneumonia, lipoid," and "syndrome, goodpasture's." glycosphingolipid storage in cornea; lens opacities; dilated vessels in conjunctiva and lens; thrombi in blood vessels ( ). disease, fibropolycystic, of the liver and biliary tract note: "fibropolycystic disease of the liver and biliary tract" comprises a group of well defined conditions, which may occur together and hence need a collective designation. the conditions include autosomal-recessive (infantile) and auto-somal dominant (adult) polycystic disease of the liver; caroli's disease or syndrome;* choledochal cyst,* congenital hepatic fibro-sis,* multiple biliary microhamartomas, and related disorders. for autopsy procedures, see also under more specific designations. disease, glycogen storage synonyms: andersen's disease or brancher deficiency (glycogenosis, type iv); cori's or forbes' disease (glycogenosis, type ill); cyclic amp dependent kinase (type x); glycogen synthetase deficiency (type ); hers' disease (glycogenosis, type vi); mcardle's disease (glycogenosis type v); phosphorylase b kinase deficiency (types ixa, b, and c); pompe's disease (glycogenosis, type it); tarui disease (glycogenosis type vii); von gierke's disease (glycogenosis, type ia); x-linked glycogenosis (type vill). note: if the diagnosis had not been confirmed prior to death, samples of liver, skeletal muscle, blood, and fascia (for fibroblast culture, see below) should be snap-frozen for enzyme assay, which will determine the specific deficiency. types ia and b, iii, vi, and hepatic phosphorylase b kinase deficiency (types ixa, b and c) are hepatic-hypoglycemic disorders, whereas types v and vii affect muscle energy processes. type ii also affects the musculature, whereas type iv may cause cirrhosis and death in infancy from extreme hypotonia. determination of type of glycogenosis usually can be based on (i) pattern of glycogen storage in liver, ( ) presence or absence of nuclear hyperglycogenation in liver, ( ) cytoplasmic lipid in liver, ( ) presence or absence of liver cirrhosis, and ( ) presence or absence of glycogen and basophilic deposits in skeletal muscles. possible associated conditions: fanconi syndrome* or gout* with type ia glycogenosis; neutropenia, recurrent infections, and crohn's disease with types ib or ie. glycogen primarily in retinal ganglion cells and ciliary muscle. glycogen in sympathetic nerve ganglia and neurons of cranial nerves in type vii. gouty arthritis. disease, graft-versus-host note: this disease occurs most commonly after bone marrow transplantation. the disease has also occurred after transfusion of viable lymphocytes, for example, to patients with cancer or leukemia. * in patients with graft-versus-host disease (gvhd), autopsy also may reveal recurrence of the underlying disease such as leukemia. possible associated conditions: alphal-antitrypsin deficiency;* amyloidosis;* ankylosing spondylitis;* primary sclerosing cholangitis;* sjogren's syndrome. * see also below under "possible or expected findings." note: in many instances, either chronic ulcerative colitis or crohn's disease* had been diagnosed clinically, but sometimes, the distinction is difficult to make, even at autopsy. many features described below occur in chronic ulcerative colitis but some manifestations of crohn's disease or conditions that may occur in all types of inflammatory bowel disease also are listed so that both positive and negative findings can be recorded properly. osteoporosis;* ankylosing spondylitis;* arthritis of peripheral joints; periarthritis; hypertrophic osteoarthropathy;* tendinitis (particularly of ankle and achilles tendons). disease, iron storage (see "hemochromatosis.") related terms: atherosclerotic heart disease. note: the most common anatomic finding at autopsy in subjects older than yr is coronary atherosclerosis. unusual under-lying or associated conditions include chronic aortic stenosis or regurgitation; coronary artery anomalies; coronary artery dissection; coronary embolism; coronary ostial stenosis (due to calcification of aortic sinotubular junction or, rarely, to syphilitic aortitis); coronary vasculitis (for instance, in polyarteritis nodosa* or acute hypersensitivity arteritis); hyperthyroidism,* gastrointestinal hemorrhage; * hypothyroidism, * idiopathic arterial calcification of infancy; intramural coronary amyloidosis; pheochromocytoma, polycythemia vera; * pseudoxanthoma elasticum,* radiationinduced coronary stenosis; severe pulmonary hypertension (with right ventricular ischemia); sickle cell disease;* and others. if bypass surgery had been performed, see "surgery, coronary bypass." macular rash ( ). multifocal fibrinopurulent pneumonia with sparing of the bronchi and bronchioles. exudate is rich in phagocytes, fibrin, and karyorrhectic debris. synonym: lyme arthritis note: this infection is caused by the spirochete, borrelia burgdoiferi, which is transmitted from rodents to human by the hard deer ticks, ixodes dammini, . ricinus, and others. brain and spinal cord for removal and specimen preparation, see chapter . request luxol fast blue stain for myelin. symmetric and zonal demyelination in corpus callosum, anterior commissure, optic chiasm, optic tracts, and white matter of frontal lobes. external examination and skin; oral cavity lungs aorta record distribution of skin lesions and submit tissue samples for histologic study. for preparation of angiograms of the pulmonary arterial and venous vasculature, see chapter . if aneurysm or dissection is present, follow procedures described under those headings. telangiectatic (often papular) lesions most commonly found in cheeks, scalp, nasal orifices, oral cavity, ears, neck, shoulders, fingers, toes, and nail beds. cyanosis and clubbing may be prominent. arteriovenous malformations/fistulas. aneurysm; * aortic dissection. * if cirrhosis is present, prepare angiograms of hepatic arteries and veins (chapter ). photograph and prepare sections of angiomatous lesions. note: parkinson's syndrome is caused by conditions that may simulate parkinson's disease; these include carbon monoxide* and manganese poisoning, corticobasal degeneration, druginduced parkinsonism, huntington's disease, multiple system atrophy,* progressive supranuclear palsy* (steele-richardson-olszewski syndrome), space-occupying lesions (rare), trauma (dementia pugilistica), and causes related to tumors and vascular diseases. brain for removal and specimen preparation, see chapter . histologic sections should include midbrain (substantia nigra), upper pons (locus ceruleus), medulla, nucleus basalis (substantia innominata), and basal ganglia. if parkinsonian syndrome was diagnosed, follow procedures described under the name of the suspected underlying condition (see above under "note"). depigmentation of substantia nigra and locus coeruleus; neuronal loss and reactive gliosis; eosinophilic intracytoplasmic inclusion bodies (lewy bodies) in some of the surviving neurons; no significant changes in basal ganglia. disease, pelizaeus-merzbacher synonyms: sudanophilic (orthochromatic) leukodystrophy. brain and spinal cord for removal and specimen preparation, see chapter . request luxol fast blueipas stain for myelin and bielschowsky's stain for axons. prepare frozen sections for sudan stain. brain generally atrophic. myelin loss in centrum ovale, cerebellum, and part of brain stem, with a tigroid pattern of residual myelin near vessels. axons are preserved. diffuse gliosis with relatively few lipoid-containing macrophages, compared to the myelin loss. lipoid material stains with sudan. brain and spinal cord for removal and specimen preparation, see chapter . request silver stains (bielchowsky or bodian stain). histochemical stains in pick's cells and bodies reveal phosphorylated neurofilaments, ubiquitin, and tubulin. some tissue should be kept frozen for biochemical studies. severe cerebral atrophy, involving primarily frontal and anterior temporal lobes (knifeblade atrophy; walnut brain). microscopically, severe neuronal loss accompanied by astrocytosis. characteristic argyrophilic, intracytoplasmic inclusions (pick's bodies), particularly in hippocampus and swollen, distended "ballooned" neurons (pick's cells). these changes are not always present. external examination, skin, and adipose tissue blood cerebrospinal fluid heart liver and kidneys brain, spinal cord, and peripheral nerves eyes submit sample for determinaion of phytanic acid concentration and for molecular studies. for obtaining a sample, see chapter . sample for histologic study. for removal and specimen preparation, see chapter . for removal and specimen preparation, see chapter . ichthyosis. phytanic acid accumulation in adipose tissues. phytanic acidemia, mutation of phyh or pex ( ). increased protein concentrations. cardiomyopathy.* phytanic acid accumulation. axonal neuropathy. retinitis pigmentosa. hypoalphalipoproteinemia. lymphadenopathy with diffuse deposition of cholesterol esters. premature atherosclerotic cardiovascular disease ( ). hepatosplenomegaly with foam cells. enlarged tonsils with characteristic orange discoloration. polyneuropathy ( ) . in adults, corneal infiltrates. foam cells. request pas stain. in granulomas, bacilli are not always pas positive ( ) . section all grossly involved tissues for histologic examination. submit section for electron microscopy. emaciation. hyperpigmentation, particularly of exposed skin and in scars. hyperkeratosis. arthritis involving ankles, knees, shoulders, and wrists. ascites; fibrinous peritonitis. * nodules in peritoneum containing sickle-form particlecontaining cells (spc cells submit sample for determination of sodium, potassium, chloride, glucose, urea nitrogen, and creatinine concentrations. calcium and phosphate concentrations can also be tested. if sample is small, indicate priority for testing. if indicated, submit sample for chemical study. submit tissue samples for histologic study. considerably increased or decreased values for sodium (more than meqll or less than meqll) and chloride (more than meqll or less than meqll) indicate that changes were present before death. for further interpretation, see chapter . postmortem electrolyte concentrations are quite unreliable. may be useful for calcium determination. vacuolar nephropathy (vacuolar changes in proximal convoluted tubules) in potassium deficiency (may also occur after infusion of hypertonic solutions). disorder, hemorrhagic (see "coagulation, disseminated intravascular," ''disease, christmas:' ''disease, von willebrand's," "hemophilia," and "purpura,.••") disorder, inherited, of phagocyte function note: several conditions represent phagocyte function disorders. autopsy procedures for one of these disorders can be found under "disease, chronic granulomatous." consult this entry for other phagocyte function disorders. synonyms and related terms: fabry's disease* (angiokeratoma corporis diffusum); gangliosidosis;* gaucher's disease;* glycogenosis,* type ii; leukodystrophies (krabbe's or globoidcell,* metachromatic leukoencephalopathy*); mucopolysaccharidoses* (hunter, hurler, morquio, and sanfilippo disease); mucolipidosis; niemann pick disease* (type a, b, c, or sphingomyelinase deficiency); neuraminidase deficiency; neuronal ceroid lipofuscinosis (batten's disease or kufs' disease). hypopharyngeal pulsion diverticulum (zenker's diverticulum) at lower margin of inferior constrictor muscle of pharynx. traction diverticulum at midesophagus after an inflammatory process-for instance, tuberculous lymphadenitis. epiphrenic diverticulum may also occur. luxtacardiac or juxtapyloric diverticulum. heterotopic tissue in meckel's diverticulum, with or without peptic ulceration. colonic muscular hypertrophy and stenosis, usually in sigmoid colon. diverticulitis with perforation, fistulas, or peritonitis. * diving (see "accident, diving (skin or scuba).") related terms: dry drowning; fresh-water drowning; near-drowning; salt (sea)-water drowning (see the following table). primary drowning ("immediate drowning") deaths occurring within minutes after immersion, before or without resuscitative measures deaths from hypoxia and acidosis caused by glottal spasm on breath holding. there may be no evidence of water entering stomach or lungs and no appreciable morphologic changes at autopsy. note: the diagnosis is one of exclusion. the pathologist should help the police to determine: i) how did the person (or dead body) get in the water, and ) why could that person not get out of the water? it is not enough to ask if a person could swim but investigators should find out how well (what strokes did the victim know?) and how far he or she could swim. the inquiry must include the depth of the water and must address hazards such as undertow or underwater debris, and the behavior deaths occurring from within min to several weeks after resuscitation, because of metabolic acidosis, pulmonary edema, or infective or chemical pneumonitis deaths from hypoxia and acidosis caused by obstruction of airway by water related to: hypervolemia hemolysis hyponatremia hypochloremia hyperkalemia of the victim immediately before submerging. deaths of adults in bathtubs and swimming pools are usually from natural, cardiac causes, or they are suicides, unless the victim was drunk. diatom tests ( ) have not proven useful in the united states but there is enthusiasm for such tests among european pathologists. the distinction between hyponatremic deaths in fresh water and hypernatremic deaths in salt water derives from experimental studies; in practice, one cannot reliably predict the salinity of the immersion medium from autopsy studies. because many bodies of drowning victims are recovered only after the body floats to the surface, decomposition will often obscure even the nondiagnostic findings such as pleural effusions, which are often associated with drowning. external examination and skin (wounds) organ samples for diatom search serosal surfaces and cavities if identity of drowning victim is not known, record identifying features as described in chapter . prepare dental and whole-body roentgenograms. submit tissue samples for histologic study of wounds. inspect inside of hands. collect fingernail scrapings. record appearance and contents of body orifices. record features indicative of drowning. photograph face from front and in profile. take pictures of all injuries, with and without scale and autopsy number. remove vitreous for analysis. if diatom search is intended, clean body thoroughly before dissection to avoid contamination of organs and body fluids with algae and diatoms (see below). submit sample for toxicologic study. sample early during autopsy, before carrying out other dissections. use fresh instruments for removal of specimens to avoid contamination. submit subpleural portion of lung: subcapsular portions of liver, spleen, and kidneys; bone marrow; and brain. store samples in clean glass jars. for technique of diatom detection, see below. record volume of fluid in pleural spaces. photograph petechial hemorrhages. photograph layerwise neck dissection if strangulation* is suspected. open airways posteriorly, and photograph, remove and save mud, algae, and any other material in tracheobronchial tree. record size and weight of lungs. there may be wounds that were inflicted before drowning occurred-for instance, in shipwrecks or vehicular and diving accidents. other wounds may be inflicted after deathfor instance, from ship propellers or marine animals. sometimes, premortem and postmortem wounds can be distinguished histologically. object (hair?) held by hands in cadaveric spasm. cutis anserina and "washerwoman" changes of hands and feet are of no diagnostic help. foreign bodies; semen (see also under "rape"). foam cap over mouth and nose. in the autopsy room, water running from nose and mouth is usually pulmonary edema or water from the stomach. high concentrations of alcohol indicate intoxication (see under "alcoholism and alcohol intoxication"). evidence of alcohol intoxication may be found. diatoms may occur in the liver and in other organs of persons who have died from causes other than drowning. comparison with diatoms in water sample from area of drowning may be helpful. penny-sized or smaller hemorrhages may indicate violent respiratory efforts or merely intense lividity. presence of pleural fluid suggests drowning. for diatom detection (l) , boil - g oftissue for -- min in rnl of concentrated nitric acid and . rnl of concentrated sulfuric acid. then, add sodium nitrate in small quantities until the black color of the charred organic matter has been dispelled. it may be necessary to warm the acid-digested material with weak sodium hydroxide, but the material must soon be washed free from alkali to avoid dissolving the diatoms. the diatoms should be washed, concentrated, and stored in distilled water. for examination, allow a drop of the concentrate to evaporate on a slide, and then mount it in a resin of high refractive index. all equipment must be well-cleaned, and distilled water must be used for all solutions. there are several variations and adaptations of this method. drug abuse, amphetamine(s) note: methamphetamine abuse may be suggested by poor condition of the dentition. methylenedioxymethamphetamine ("ecstasy") abuse is often suggested by friends with whom the decedent was abusing drugs. follow procedures described under "dependence, drug(s)." ductus arteriosus, patent (see "artery, patent ductal.") synonyms and related terms. achondroplastic dwarf; asexual dwarf; ateliotic dwarf; micromelic dwarf; normal dwarf; pituitary dwarf; true dwarf; and many other terms, too numerous to mention. external examination bones and joints record height and weight. prepare skeletal roentgenograms. for removal, prosthetic repair, and specimen preparation, see chapter . growth retardation. abnormal growth of epiphyseal cartilage with enlargement of metaphysis. long bones and pelvis most commonly affected. cavernous hemangiomas (maffucci's syndrome). see above under "external examination." chondrosarcoma. dyscrasia, plasma cell note: these conditions are characterized by abnormally proliferated b-immunocytes that produce a monoclonal immunoglobulin. multiple myeloma, * plasma cell leukemia, plasma-cytoma, and waldenstrom's macroglobulinemia* as well as heavy-chain diseases and monoclonal gammopathies of unknown type belong to this disease family. amyloidosis* is closely related to these conditions. for autopsy procedures, see under "amyloidosis," "macroglobulinemia," or "multiple myeloma" and under name of condition that may have caused the plasma cell dyscrasia. such conditions include carcinoma (colon, breast, or biliary tract), gaucher's disease,* hyperlipoproteinemia, * infectious or noninfectious chronic inflammatory diseases, and previous cardiac surgery. synonym: shigella dysentery. note: (i) collect all tissues that appear to be infected. blood bowel eyes joints submit sample for culture and for serologic study. submit sample of feces or preferably bloodtinged mucus for culture. if bacteriologic diagnosis has already been confirmed, pin colon on corkboard, photograph, and fix in formalin for histologic study. submit sample of vitreous for study of sodium, potassium, chloride, and urea nitrogen concentrations. for removal and specimen preparation of eyes, see chapter . for removal, prosthetic repair, and specimen preparation, see chapter . escherichia coli septicemia. colitis with microabscesses; transverse shallow ulcers and hemorrhages, most often in terminal ileum and colon. dehydration* pattern of electrolytes and urea nitrogen. serous arthritis* of knee joints is a late complication. external examination record extent of pigmentation, facial features, and primary and secondary sex characteristics. prepare skeletal roentgenograms. for removal, prosthetic repair, and specimen preparation, see chapter . record size of apertures of cranial nerves in base of skull. unilateral skin pigmentation and precocious puberty in females (albright's syndrome), less commonly in males. synonyms and related terms: becker's muscular dystrophy; congenital muscular dystrophy; duchenne's progressive muscular dystrophy; dystrophinopathy; em-ery-dreifuss mucular dystrophy; facioscapulohumeral dystrophy; limb girdle dystrophy; myotonic muscular dystrophy. external examination record pattern of scalp hair. record status of skeletal musculature. obtain sections for histologic examination. dystrophin staining of the sarcolemma is absent in duchenne's muscular dystrophy and patchy in becker's dystrophy. frontal baldness (in myotonic muscular dystrophy). atrophy and wasting of muscles (generalized or local: predominantly distal in myotonic muscular dystrophy). pseudohypertrophy of calf muscles in duchenne's muscular dystrophy. dystrophic changes include variations in fiber size, fiber degeneration and regeneration, peri-and endomysial fibrosis, and fatty replacement of muscle. the liver, especially the right lobe, is the most common site of involvement. secondary infection or calcification may be present. the lung is the second most common site of involvement. fluid and air may be visible on the roentgenogram. cysts may be present in the abdominal cavity, muscles, kidneys, spleen, bones, heart, and brain. eosinophilia. edema, angioneurotic synonym: angioedema. note: possible causes and suggested autopsy procedures are described under "death, anaphylactic." related term: silo-filler's disease. n<yfe: this condition is causedby inhalationoftoxic gases, such as oxides of nitrogen (silo-filler's disease) and phosgene (coci ). see also "bronchitis, acute chemical" and "poisoning, gas." ( ) gunshot and shotgun wounds of the head involving dural sinuses; ( ) injury to large veins, particularly cranial sinuses (during neurosurgical procedures) and veins ofthe neck (knife wounds, surgery) or uterus (criminal abortion); ( ) insufflation offallopian tubes (particularly in pregnancy or during menstrual period); ( ) infusion of blood components or crystalloid; ( ) malfunctioning of dialysis machines; ( ) subclavian vein catheterization in the semi-fowler position; and ( ) fracture in the hub of a central venous catheter used for parenteral nutrition. possible causes of arterial air embolism include: ( ) open heart surgery involving the aorta, left atrium, or left ventricle; ( ) positive-pressure ventilation in newborn infants; and ( ) underwater ascent with closed glottis (see accident, diving) ifair embolism is suspected, the autopsy should be performed as soon after death as possible. decomposition gases may be produced within a few hours. roentgenography of the whole body may detect large quantities of air, and the roentgenograms may serve as a guide to the most advantageous way of dissection. the postmortem diagnosis of arterial air embolism is made largely on the basis of medical history and circumstances. the autopsy serves to rule out competing conditions. the diagnosis of venous air embolism is made by chest roentgenography before the autopsy ( ). the air in the right chambers of the heart can be confirmed at autopsy by aspiration into a syringe. the formerly recommended procedure of clamping the internal mammary vessels below the sternoclavicular joints, and then cutting across the sternum distal to these clamped vessels so that the sternoclavicular joint area remains intact, is designed to prevent the production of a vacuum that pulls air into the veins. the procedure is not necessary if the air is welldocumented by roentgenography before autopsy. at autopsy, a large fatal pulmonary air embolism is readily apparent. the right atrium and ventricle are distended with fine, frothy, brightred blood, which also may distend the pulmonary arteries and superior vena cava. microbiologic examination ofblood and pericardial sac contents may help to rule out the presence of gas-forming bacteria that may simulate air embolism (fig. ii-i). however, the presence of putrefactive emphysema in any part of the body points toward a bacterial origin of the gas. differentiation of air and decomposition gases at the autopsy table with the pyrogallol test is described below. a % pyrogallol solution is prepared (it should be waterclear). two lo-ml syringes (syringe a and syringe b) are loaded with ml of the pyrogallol solution in each, without permitting any air to enter the system. immediately before the solution is used, drops of . n naoh is aspirated through the needle of syringe a to adjust the ph to about ( drop per ml of solution); the mixture will turn faint yellow. six ml of gas is then aspirated from the heart or blood vessels. the needle is immediately sealed with a cork or replaced by a cap, and the syringe is vigorously shaken for about min. in the presence of air, the pyrogallol solution will turn brown. if the solution remains clear, decomposition gases were present. in the latter instance, drops of . n naoh and ml of room air should be aspirated into syringe b, which is then also sealed and shaken for min. the mixture should turn brown, thus serving as a fig. · . gas-forming bacteria simulating air embolism. the pericardial sac is opened and filled with water. the heart is kept submerged with a pair of scissors. the coronary arteries have been incised with a scalpel. note gas bubbles and foam on the water surface. no discoloration of % pyrogallol was noted. blood cultures were positive for enterococcus organisms. microscopically, gram-negative rods were found in most tissues. control that the pyrogallol solution had been properly prepared. syringe b may also serve as a reserve. if only one syringe is used, the decomposition gas can be expel ed and room air can be aspirated for the control test. if only small amounts of gas can be aspirated, the volume of the pyrogallol solution should be decreased so that the gas-fluid volume ratio is at least : . if no bacterial gas formation is present, the edges of the pericardial incision are elevated and the pericardial sac is fil ed with water. clamping of the ascending aorta and venae cavae prevents the escape of gas into these vessels. the heart is held under water while the coronary arteries are incised, and the escape of bubbles is recorded. when the right coronary artery is opened, care must be taken that the right atrium is not incised. air in the coronary arteries indicates systemic embolism. the heart chambers are then incised. when there is gas in any of the arteries or heart chambers, gas bubbles rise to the surface of the water in the pericardial sac ("bubble test"). sometimes the vessels have to be somewhat compressed in order to cause the gas to escape. because large amounts of air or other gases cause the heart to float, it must be kept submerged before the vessels and chambers are incised. basically the same procedure is used for demonstrating the presence of gas in the superior or inferior vena cava and the pelvic veins (for example, in cases of criminal abortion). in this situation, the abdominal cavity is filled with water and the inferior vena cava and its tributaries are incised. in support of the diagnosis of systemic arterial air embolism, the skull vault may be removed without puncturing the meninges, so that the cerebral arteries can be inspected for gas bubbles. the demonstration of gas bubbles in the meningeal vessels and in the circle of willis is meaningful only when the neck vessels are still intact and the internal carotid artery and basilar artery have been clamped before the brain is removed. in acute cases, gas bubbles will be visible within the cerebral vessels. they are released under water when the clamps are removed and the vessels are slightly compressed. for the collection of gas from blood vessels or cavities, a system oflittle quantitative reliability is an air-tight, water-filled glass syringe with a needle. the needle is inserted into the vessel or cavity in question and gas is carefully aspirated. a combined qualitative and quantitative method has been described by kulka ( , ) (see fig. - ). he devised an apparatus for gas collection and described it as shown in fig. - and caption. the entire system is filled with mineral oil so that, when the funnel is level with the upright bottle, the oil fills only about half of the funnel. in operation, the funnel is first raised to a position - cm above the level of the upright bottle. all the cocks are opened and the position is held until every trace of gas has been driven from the system through the needle, which is thereby coated on the inside by a film of oil. after all air has been expelled, the cocks are closed and the funnel is lowered to its original position. as a precautionary measure and control, the air-tightness of the whole system should be tested before operation. this is done by inserting the needle into musculature or skin and attempting aspiration in the manner described in the next paragraph. to make the test, the bottle is inverted and the needle is inserted into the cavity in question. when the needle is in position, all cocks are opened. the funnel is lowered about - cm, or until adequate suction is created. this aspirates the contents of the cavity, which may consist of air or other gases, either pure or mixed with blood or other liquid. any gas or liquid entering this system may be observed through the wall of the short bent glass tubing. in a positive test, gas bubbles will collect in the bottle above the level of the oil. if desired, this gas can be saved for further examination by closing all the cocks and returning the bottle to its upright position ( ). the volume ofintravenous air needed to cause death in adults depends of the cross sectional area of the cardiac cavity or great vessel that is the site of the gas lock, which in tum depends on the position of the decedent at the time of the embolism ( ). ml of gas can pass through an intravenous hub in just a few seconds. small amounts of air entering the systemic circulation may cause death within minutes. delayed air embolism with fatal outcome may also occur. other organs if purpura is present, prepare photographs and record extent. submit sample (from right atrium) for microbiologic study. collect blood from right atrium and right ventricle. after the heart has been removed, allow blood in pulmonary vessels to pool in the pericardial sac. centrifuge this blood and submit sample of flocculent layer above buffy coat for microscopic study ( ) . submit a section of lung for bacteriologic study. dissect pulmonary arteries; prepare histologic sections of all lobes; request mucicarmine stain and the aldan blue and phloxinetartrazine stain of lendrum. also request sudan stain on frozen sections. skin purpura. vernix, lanugo hairs, and meconium can be found in pericardial blood pool. meconium-type material in blood vessels. in histologic sections, squamous epithelium, meconium, and fat from vernix caseosa. complete or incomplete lower uterine tear; chorioamnionitis. manifestations ofdisseminated intravascular coagulation* and fibrinolysis. intrauterine pneumonia. large amounts of debris in the blood vessels of all sections of the lungs may be considered to be lethal if there is no other cause of death. small amounts in one or more blocks of pulmonary tissue are more likely incidental ( ). a small uterine tear is more likely followed by a fatal amniotic fluid embolization than is a large tear, which may result in fatal hemorrhage or fibrinogen depletion. chorioamnionitis, intrauterine pneumonia, and positive lung cultures of the mother indicate infection of the amniotic fluid. record weight and sample for histologic study. for removal and specimen preparation, see chapter . photograph horizontal sections through brain, brain stem, and spinal cord. prepare frozen sections and request sudan stain. prepare frozen sections of one-half of gland and paraffin sections of the other half. petechial hemorrhages of skin (chest, neck, and face). wounds; other traumatic lesions. bone fractures. petechiae of conjunctivas and retinas. fat may accumulate on surface ofblood pool. no useful technique is available for estimating the amount of fat globules in the blood. fat emboli in lumen of small vessels and in pulmonary air spaces. severe fatty changes may be the cause offat embolism. fractures are the most common cause of fat embolism. petechial hemorrhages, fat emboli ( ). abdominal cavity submit sample of subphrenic exudate for gram stain and cultures. record location and volume of subphrenic exudate. possible causes of subphrenic empyema include appendicitis, cholecystitis, * diverticulitis, intrahepatic abscess, pancreatitis,* ruptured viscus; penetrating abdominal wound(s), perforated ulcer of stomach or duodenum,* and other conditions. pleural cavities and lungs e procedures record volume of effusion or exudate in pleural space. basal pleuritis and pneumonia, adjacent to empyema. encephalitis, au types or type unspecified synonyms and related terms: acute disseminated encephalomyelitis;* acute hemorrhagic encephalitis; acute infective encephalitis or encephalomyelitis; acute poliovirus encephalitis or encephalomyelitis; amoebic encephalitis; arbovirus encephalitis (japanese encephalitis; eastern encephalitis, western encephalitis, venezuelan equine encephalitis, st. louis encep-halitis); bulbar encephalitis;* brain stem encephalitis;* herpes encephalitis (cytomegalovirus encephalits, epstein-barr virus encephalitis, varicella-zoster encephalitis); herpes simplex ence-phalitis; hiv encephalitis; measles encephalitis; measles inclusionbody encephalitis; postinfectiousencephalitis; postvac-cinal encephalitis; progressive multifocal leukoencephalitis or leukoencephalopathy; rabies* encephalitis; subacute encephalitis; subacute sclerosing panencephalitis; viral encephalitis, west nile encephalitis and other terms ( ), too numerous to mention. see also under "note" and under "possible or expected findings." note: if the condition that caused the encephalitis is known, see also under that heading. if the cause of the encephalitis is unknown, submit samples of tissue for microbiologic and toxicologic study, particularly if there is a suspicion of lead poisoning.* see also under "encephalitis, brain stem," "encephalomyelitis,...," "encephalopathy" and "myelopathy, myelitis." encephalitis, brain stem synonyms and related terms: brain stem abscess; infectious brain stem encephalitis; listeria monocytogenes brain stem encephalitis; viral brain stem encephalitis. note: see also under "encephalitis, limbic." brain for removal and specimen preparation, see chapter . necrotizing encephalitis, with or without abscess formation ( enterocolitis, other types or type undetermined note: a multitude of infectious and noninfectious agents may cause inflammation of the small bowel, large bowel, or both. if the condition is not listed under "colitis," "enteritis," or "enterocolitis" or under another specific heading such as "dysentery, bacillary," obtain sufficient material for microbio-logic and histologic study to identify organisms such as clos-tridium, chlamydia, shigella, salmonella, yersinia, helicobacter, verotoxic e. coli, and others. if lymphogranuloma venereum,* or tuberculosis* are suspected, see also under these headings. see also under "disease, inflammatory bowel" and "disease, crohn's." synonyms and related terms: c. difficile colitis; darmbrand; hemorrhagic necrosis (gangrene; infarction) of intestine; ischemic enteritis or enterocolitis;* neutropenic enterocolitis;* pseudomembranous colitis. note: the name "pseudomembranous enterocolitis" is descriptive; the condition may be infectious, ischemic, or both. if the intestinal changes are clearly ischemic, see above under "enterocolitis, ischemic." if the cause is in doubt and if pseudomembranes can be identified, follow the procedures described here. (l) collect all tissues that appear to be infected. for other infectious intestinal diseases, see under specific names, such as "enterocolitis, neutropenic" or "enterocolitis, staphylococcal." intestinal tract and mesentery other organs collect material from pseudomembranes for culture and for c. difficile toxin assay. sample intestinal wall with pseudomembranes for histologic study. if the condition is suspected to be caused by ischemia, follow procedures described above under "enterocolitis, ischemic." note: myoclonic seizures also have been described in a number of progressive encephalopathies with complex neurological symptoms, such as gmi and gm gangliosidosis,* and niemann-pick* and krabbe's disease but also acquired disorders, including alzheimer's disease,* creutzfeldt-jakob epilepsy, myoclonus (continued) disease,* posthypoxic encephalopathy, and subacute sclerosing panencephalitis. mitochondrial encephalomyopathy also can present with myoclonus epilepsy and a mitochondrial myopathy with ragged red fibers (merrf syndrome) in skeletal muscles. lafora-body-type material in the heart, liver, retinas, peripheral nerves, skeletal muscles, and sweat gland ducts (especially axillary). ragged red fibers in mitochondrial myopathies. epilepsy, symptomatic note: possible causes or underlying conditions include cerebrovascular diseases, congenital malformations ofthe brain, degenerative and demyelinating diseases of the brain, head injury,* intracranial and cerebral infections, toxic or metabolic disorders (alcoholism,* barbiturate,* carbon monoxide,* and lead poisoning,* hemodilution, hypocalcemia, or hypoglycemia),* and tumors of the brain. * f failure, congestive heart note: coronary atherosclerosis and manifestations of ischemic heart disease, * valvular heart disease, congenital cardio-vascular diseases, and manifestations of systemic or pulmonary hypertension are the most frequent findings in patients dying of or with congestive cardiac failure. other causes include cardiomyopathies* and secondary myocardial disease (such as amyloid or pericardial constriction). if the cause of the congestive cardiac failure is unknown or not immediately evident after dissection of the heart and of the great vessels, myocardium and other appropriate tissues may be submitted for microbiologic study-including viral cultures-and for electron microscopy. specimens can also be snap-frozen for possible immunofluorescent, biochemical, or histochemical studies, particularly of the myocardium. possible causes of congestive cardiac failure are too numerous to mention. dilatation of heart, with or without mural thrombosis. myocardium may be soft, normal, or firm. pulmonary congestion, with or without hemosiderosis; congestion of viscera with organomegaly. other organ manifestations include bowel edema or hemorrhagic enteropathy (without mechanical vascular occlusion) and zonal hepatic steatosis, fibrosis, or necrosis, with or withoutevidence of liver failure. acute renal tubular necrosis may be present also. synonyms and related terms: acute kidney failure; chronic kidney failure; renal failure; uremia. note: if acute kidney failure had been diagnosed, the autopsy procedures will depend on the expected causes, such as poisoning with ethylene glycol,* lead,* mercury,* or methyl alcohol;* disseminated intravascular coagulation* and its various underlying conditions; glomerulonephritis* and its various underlying conditions; diabetes mellitus;* or multiple myeloma. * the procedures described below deal primarily with chronic renal failure. if the patient had had dialysis, see also under "dialysis (for chronic renal failure )." if transplantation had been carried out, see also under "transplantation, kidney." lassa fever is a highly communicable disease and autopsy studies are not recommended in the usually available surround-ings. if lassa fever is suspected, contact the state health department and the centers for disease control and prevention, atlanta, ga, for disposition and further studies ( ).ifan autopsy is done, disinfection can be accomplished by washing instruments with . % phenol in detergent (i.e., lysol), . % hypochlorite solution, formalin, or paracetic acid. for shipping procedures, see chapter . this is not a reportable disease. synonyms: borreliosis; louseborne (epidemic) relapsing fever; tickborne (endemic) relapsing fever. note: (i) collect all tissues that appear to be infected. ( ) rat/mouse inoculation with infected blood is the most sensitive method for the detection of the organism. consult the state health department. ( ) before the autopsy, consultation with the microbiology laboratory is advised. ( ) request direct dark-field examination and giemsa or wright stain. ( ) no special precautions are indicated. ( ) serologic studies are of questionable value due to the antigenic variability of the organism. ( ) fever, scarlet note: usually, this disease is a nonfatal group a streptococcal tonsillitis and pharyngitis with skin rash. potentially fatal complications include suppurative otitis media,* mastoiditis, and pharyngeal abscess ( ), with or without septicemia. ( ) collect all tissues that appear to be infected. ( ) atresia, but may also be seen with viral myocarditis, type ii glycogen storage disease (pompe disease) and carnitine deficiency. thus, a metabolic disorder must be considered. it is ideal to perform the autopsy as soon after death as possible ( ) . urine plasma for removal, prosthetic repair, and specimen preparation, see chapter . malnutrition ( ) muscle necrosis (clostridial myonecrosis) and accumulation of gas; little leukocytic infiltration. other organs g procedures depend on expected findings or grossly identified abnonnalities as listed in right-hand column. see also above under "note" and under "skeletal muscles." synonyms and related terms: acute postinfectious glomerulonephritis (nonstreptococcal postinfectious glomerulonephritis;*minimalchangedisease;mesangialproliferativeglomerulonephritis;* focal and segmental glomerulosclerosis with hyalinosis (focal sclerosis); poststreptococcal glomerulonephritis; idiopathic nephrotic syndrome; iga nephropathy (berger's disease); membranous glomerulonephritis; membranoproliferative glomerulonephritis; mesangial proliferative glomerulonephritis; rapidly progressive glomerulonephritis (associated with systemic infectious or immunologic multisystem diseases; drug idiosyncrasy; or as primary crescentic glomerulonephritis or superimposed on another primary glomerular disease). possible associated conditions: acquired immunodeficiency syndrome (aids);* alport's syndrome;* amyloidosis; anaphylactoid purpura; bee stings; chronic allograft rejection; dennatomyositis;* dennatitis herpetiformis; diabetes mellitus;* drug dependence;*fabry's disease;* goodpasture's syndrome;* guillain-barre syndrome;* henoch-schonlein purpura;* hemolytic uremic syndrome;* infective endocarditis;* leprosy;* malig-nancies;mixedconnectivetissuedisease;myxedema;polyarteritis nodosa;* rheumatoid arthritis;* preeclamptic toxemia; renovascular hypertension; sarcoidosis;* serum sickness;* sjogren's syndrome;*syphilis;*systemiclupuserythematosus;*systemic sclerosis;* thrombotic thrombocytopenic purpura;* thyroiditis;* vasculitis; viral hepatitis;* wegener's granulomatosis;* and many other conditions. urate deposits in scleras and corneas. granulocytopenia note: midline granulomas may belong to the angiocentric immunoproliferative lesions, which are related to lymphomatoid granulomatosis* and malignant lymphoma. the name "idiopathic midline granuloma" should be reserved for the few cases without evidence of malignant lymphoma or wegener's granulomatosis* ( ). the diagnosis of idiopathic midline granuloma also can be ruled out if studies reveal fungal organisms or features of leishmaniasis;* leprosy,* rhinoscleroma, pseudotumor of the orbit or tuberculosis. * complications of nasal cocaine abuse also may mimic midline granuloma ( ). eosinophilic pneumonitis (degenerating eosinophils with charcot-leyden crystals) and granulomas. angiitis (mostly arteritis), typically with giant cells in tunica media ( ) . findings resemble those in polyarteritis nodosa. * heart ( ), grastrointestinal tract ( ), skin, muscles, and joints are commonly involved. however, renal disease is often (but not always) mild or absent. necrotizing vasculitis of small arteries and veins is present, with extravascular granulomas and eosinophilic infiltration of vessels and perivascular tissues. optic neuritis may be found ( ). may be affected by the vasculitis ( ). pcr studies on paraffin sections via rna in situ hybridization may confirm presence of epstein-barr virus-positive b-cell proliferations combined with dense t-cell accumulations ( ) ( ) ( ) . the condition closely resembles angiocentric t-/nk cell lymphoma ( ). infiltration of lymphocytoid cells, plasma cells, and macrophages with necroses and granulomatous features, which are found primarily in the vicinity of blood vessels. special stains may reveal evidence of infection. lymphoreticular and granulomatous infiltrates (see "lung"). lymphoreticular and granulomatous infiltrates (see "lung"). characteristic infiltrates may be present in all organs and tissues. involvement of spleen, lymph nodes, and bone marrow is uncommon. in rare instances, the disease is confined to the abdomen. in most instances, characteristic infiltrates are present ( ). septicemia; circulating immunoglobulin complexes, and antiendothelial antibodies ( ). angiocentric granulomatosis ( ) hemorrhage, intracranial (see under "hematoma, subdural" and under name of suspected underlying condition, such as "aneurysm, cerebral artery," "infarction, cerebral," "injury, head," and "tumor of the brain." hemorrhage, subarachnoid (see under name of suspected underlying condition, such as "aneurysm, cere-bral artery," "infarction, cerebral," "injury, head," and "tumor of the brain." hemosiderosis, idiopathic pulmonary note: this diagnosis is made by exclusion; involvement of organs other than the lungs (see below under "kidneys") suggests another disease. hepatitis, chronic synonyms and related terms: autoimmune hepatitis; chronic viral hepatitis b (with or without d) and c. note: the term "chronic hepatitis" is not a complete etiologic diagnosis and related names such as chronic active (aggressive) hepatitis, chronic active liver disease, and chronic per-sistent hepatitis are obsolete. most cases in these categories represent autoimmune hepatitis or chronic viral hepatitis b or c. many other liver diseases, including drug-induced hepatitis, inborn errors ofmetabolism (e.g., alphal-antitrypsin deficiency* or wilson's disease*), developmental disorders, and chronic biliary diseases such as primary biliary cirrhosis or primary sclerosing cholangitis also may present as chronic hepatitis. ifchronic hepatitis was the cause ofdeath, submassive hepatic necrosis or cirrhosis* is usually present, often with manifestations ofportal hypertension,*hepatic encephalopathy, hepatorenal syndrome,* and with ascites or spontaneous bacterial peritonitis. possible associated conditions: see also below under "possible or expected findings." conditions that may be associated with hepatitis c include ( ): autoimmune hepatitis; behcet's disease; diabetes mellitus (type );* glomerulonephritis;* guillain-barre syndrome;* idiopathic pulmonary fibrosis; idiopathic thrombocytopenic purpura; iga deficiency; lichen planus; mixed essential cryoglobulinemia; mooren's corneal ulcers; polyarthritis; porphyria cutanea tarda;* thyroiditis. * note: coinfection with other hepatitis viruses (e.g., c and g) and/or systemic viruses such as the immunodeficiency virus are common, particularly in drug addicts ( , ) . in many cases of fulminant hepatitis, tests for known hepatitis viruses are negative ( , ) . homicide note: not all of the following procedures can be carried out or will be required in all cases, nor will this checklist be sufficient in all instances. consult also the entries indicating the cause of death, such as "injury, firearm" or "injury, stabbing." if the victim is an infant, see all under "infanticide." before the body arrives or before autopsy is begun: . investigate the scene where the body was found and where the crime may have been committed (these may be two separate locations). if this is not possible, study the report and photographs submitted by the investigator or the police. study all available information. request previous medical records and roentgenograms of the victim. . emphasize to first responders and autopsy personnel that the body of the victim should not be undressed, washed, or otherwise disturbed until it has been inspected by the pathologist. embalming is not permitted before completion of the autopsy. advise first responders or transporters to put paper bags over the hands of the victim. this will protect possible evidence, such as hair from the assailant. . prepare record sheets to document the time of arrival and release of the body; chain of custody for the body and specimens; the name, age, sex, and other information about the victim; the names ofthe technician, pathologist and guests; and the specimens taken. . prepare roentgenographic and photographic equipment. after body arrives but before autopsy is begun: . if the body is left unguarded-for instance, overnight-a lock should be placed on the refrigerator or cool room where the deceased is kept, and the key should be retained by the technician. . if the pathologist is not accustomed to a busy autopsy room, he or she should feel free to ask all persons other than the pathologist(s), technicians, and law enforcement personnel who are not immediately involved in the actual performance of the autopsy to leave the morgue. . the body temperature of the victim can be recorded at the autopsy facility, but this is rarely useful, particularly if the victim seems to have been deadfor longer than a day ortwo or ifthe body had previously been stored in arefrigeratoror cool room. insert thermometer deep into the anus (about - cm). record temperature and manner and time of procedure. . aspirate vitreous from one or both eyes and store the specimen in a refrigerator. . prepare roentgenograms. in each case, a decision must be made whether roentgenograms should be made of the entire body or only of parts of it-or not at all-and whether they should be made before the victim is undressed, after the victim is undressed, or at both times. . take overall survey photographs that depict the anterior surfaces of the body as it is on arrival to the facility, before any undressing, manipulation for roentgenographs, or cleaning. . take close-up photographs of any trace evidence such as fibers or flakes of gunpowder before undressing or cleaning. . photograph the face of the victim for identification purposes after it has been cleaned of any blood and foreign matter. to avoid perspective distortion from wide angle lenses, the lens should be about ft from the face. . photograph all injuries and other forensically significant findings after blood and foreign matter have been cleaned off the body. a moist towel is useful; brushes are too abrasive. two photographs should be taken of each finding. one should include the autopsy number and a scale. a second photograph should be taken without any extraneous objects. collection and documentation of evidence: . fingerprinting can be done before or after the autopsy but should be done in all homicide cases. in cases involving close contact between assailant and victim, fingernail scrapings or clippings (use a new clipper) and hair with roots should be collected, and its source should be identified (hair pulled from scalp, axillae, and pubis is identified as that of the victim; hair in hands or under fingernails or on clothing of victim may be from the assailant). hair exemplars can be collected in cases of homicide by firearm but are rarely of use. if the body is decomposed or mutilated or for any other reason has not been identified, prepare roentgenograms ofthe head, neck, and torso before the autopsy (the radiographic appearance is altered by the autopsy). these can be used for comparison purposes if antemortem films are located. if films of the extremities are needed they can be prepared after the autopsy. dental roentgeno-.grams are taken, usually after the autopsy, when investigators have located antemortem dental records. detailed descriptions and sizes of clothing, with photographs of clothing, can be useful then there is no putative identity. . in cases in which sexual assault may have been involved, collect pieces of clothing that may contain seminal fluid stains. follow procedures described under "rape." . preserve clothing or part of clothing as evidence. wet clothing should be dried before it is stored in labeled and sealed bags. the autopsy: . the measured length and weight, extent of rigor, and color and distribution of livor should be recorded, along with the state of preservation, nutrition, and hydration. do not omit examination of the hands, particularly of the volar surfaces. . if air embolism is suspected, see procedure described under part ii "embolism, air". if pneumothorax is suspected, see procedure described under part ii "pneumothorax". if there is evidence of strangulation or other neck injury, perform a layerwise dissection of the neck that includes the hyoid bone and tongue. . prepare diagrams of wounds and identify their location by anatomic region. for wounds of the torso, state the distance from the soles of a foot and the distance laterally from to the right or left of the midline. . submit samples of wounds for histologic study when there is any question of the age of the wounds. . the records should show when body fluids and tissues were collected for laboratory analysis, the volume and appearance of such specimens, and what was done with them. if most of the toxicology specimens are collected immediately after the internal examination has commenced, the time of the internal examination serves as the collection time. in all instances, the following items should be collected: blood ofvictim for dna comparison and toxicologic study (determination of alcohol, carbon monoxide, and drug concentrations will be requested most frequently); vitreous; urine; gastric contents; at least one solid organ specimen for toxicologic study (liver and brain have the most comparison data if an extracerebral congenital malformation is suspected, follow procedures described under "malformation, arnold-chiari." if cerebrospinal fluid is aspirated with a syringe, record volume. record weight of brain; record size of brain in relation to inner dimensions of skull. describe size of ventricles. other procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. if a surgical shunt is present, its location should be recorded and both ends of the implanted specimen that was used for shunting may be submitted for microbiologic study. if the shunt is not patent, record site and nature of obstruction. related terms: antibody-mediated hydrops fetalis; erythroblastosis fetalis;* nonimmune hydrops fetalis. note: the classic example of antibody mediated (rh incompatibility) hydrops fetalis is hemolytic disease of the newborn (erythroblastosis fetalis*). however, nonimmune hydrops fetalis also may be caused by hematologic disorders, e.g., alpha-thalassemia* ( ) or it may have no known cause. infec-tions, e.g., with human parvovirus bl ( ), cytomegalovirus, or syphilis;* heart and vascular diseases ( ) (cardiac tumors, cardi-omyopathy,* myocarditis,* arterial calcification, and others); storage disease ( ); tumors (including neonatal leukemia*); and many other fetal (e.g., congenital chylothorax* or lymphatic dysplasia; pulmonary sequestration and cystic adenomatoid malformation) or maternal conditions, e.g., maternal thyrotoxicosis, also may cause nonimmune hydrops fetalis. if coronary insufficiency or myocardial infarction is suspected, follow procedures described under "disease, ischemic heart." for coronary arteriography, see chapter to. record distribution of atherosclerotic lesions in aorta. samples for histologic study should include aorta, coronary arteries, and peripheral arteries. see also above under "heart." submit samples of head, corpus, and tail for histologic study. if applicable, see also under "diabetes mellitus." for special procedures and stains, see above under "heart." other procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. for removal and specimen preparation, see chapter . obesity* is a common finding. eruptive xanthomas (palms, elbows, knees) in some but not all types of hyperlipoproteinemia. most prominent in apoprotein cii deficiency. xanthomas of tendons (knees, elbows, dorsum of hands), xanthelasmas, and arcus comeae in familial hypercholesterolemia. gangrene of lower extremities (see below under "arteries"). severe coronary atherosclerosis and myocardial infarcts in type hyperlipoproteinemia and multiple lipoprotein-type hyperlipidemia. atherosclerosis of abdominal aorta and its branches and of carotid arteries. coronary atherosclerosis (see above). pancreatitis* in familial apoprotein cii deficiency. foam cells with triglycerides in liver, spleen, and bone marrow in familial lipoprotein lipase deficiency. manifestations of diabetes mellitus* and hypothyroidism* in some cases of type hyperlipoproteinemia. cerebral infarct (stroke*) in type hyperlipoproteinemia. hypernatremia (see "disorder, electrolyte(s).") hyperoxaluria synonyms and related terms: oxalosis; primary hyperoxaluria type i (a anineglyoxylate aminotransferase deficiency); primary hyperoxaluria type ii (d-glyceric acid dehydrogenase deficiency); secondary hyperoxaluria (see under "note"). note: in ethylene glycol poisoning,*oxalatecrystals in media ofsmall arteries, with associated ischemic lesions. similardeposits may occur after long-term hemodialysis ( ). these conditions must be distinguished from the genetic disease. also, oxa-iate nephropathy may be a complication of short bowel syndrome. if patient with congenital hyperoxaluria underwent liver or combined liverlkidney transplantation ( ) , see also under these headings. remove vitreous for biochemical evalvation. submit tissue (i.e., fascia lata) for karyotype analysis. polycystic ovaries, testicular adrenal rests ( note: there are no diagnostic autopsy findings for hypoxia. possible causes of acute hypoxia include anesthesia-associated death,' compression of the torso by a vehicle, diving accident,' exposure to toxic gases-forinstance, carbon monoxide poisoning;' mechanical failure ofan oxygen supply system, as in an airplane; and sudden mechanical airway obstruction,' as in aspiration ofa foreign body orstrangulation. causes ofchronic asphyxia include prolonged exposure to high altitude and chronic pulmonary disease. profound medial hypertrophy of small pulmonary arteries and of pulmonary veins from chronic exposure to hypoxia. acute pulmonary edema may also occur in chronic hypoxia. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. malnutrition. empty and collapsed colon; small amounts of gray and dry meconium in terminal ileum; meconium masses in dilated and hypertrophied mid-ileum; liquid contents in proximal intestine ( ); appendicitis; bowel perforation. glandular inspissation and atrophy of intestinal mucosa. volvulus, infarction, necrosis, perforation, peritonitis, and acquired intestinal atresia may be present. absence of neural plexuses in congential megacolon* (hirschsprung's disease). manifestations of cystic fibrosis, with or without cirrhosis;* biliary atresia* ( ). immunodeficiency (see "syndrome, acquired immunodeficiency" and "syndrome, primary immunodeficiency.") impression, basilar note: basilar impression constitutes an upward bulging of the margins of the foramen magnum. when occipital condyles are displaced above the plane of the foramen magnum, basilar invagination is present. possible associated conditions: klippel-feil syndrome;* osteogenesis imperfecta;* osteomalacia;* paget's disease of bone;* rheumatoid arthritis; rickets; syringobulbia; syringomylia. * compression and secondary ischemic injury to lower brain stem and spinal cord. see also above under "skull and spine" and under "possible associated conditions." incompetence,••• (see "insufficiency,..•") related terms: battered-child syndrome; child-abuse or child-neglect death. note: vitreous should not be aspirated until the skull has been opened and trauma ruled out, because there is a risk of artifactual damage to the retina. in the presence of craniocerebral trauma the pathologist can opt to examine the eyes by removal, fixation, and histologic study; or retinal photography. follow general procedures described under "homicide" and, if necessary,. the procedures described under "rape." external examination see above under "note." prepare roentgenograms of entire body. see above under "note." in other situations, or after study of the fundus, submit sample of vitreous for chemical and possible toxicologic study. umbilical cord attachment prepare histologic sections of umbilicus. and end of umbilical cord blood submit sample for toxicologic study. consider retaining dried blood on filter paper for possible dna testing for paternity investigation. if there is a possibility that the cadaver represents a stillbirth, see part ii, "stillbirth". if infant is thought to have died shortly after birth, determine the location of air in the intestinal tract; roentgenograms may be helpful. save stomach contents and record amount. other organs see also under "homicide." in the abused child, bruises, hematomas, burn marks, or other patterned injuries. in any child, diaper rash, mongolian spots, self-inflicted fingernail scratches, skin infections, and emaciation. in the previously battered child, fractures in various states of healing. in hypertonic dehydration,' sodium concentrations more than meq/l. this may be caused by organic disease, improper medical treatment, or physical neglect. if infant was born alive, inflammatory changes may be found, depending on survival interval. for the hydrostatic lung test, see "stillbirth." the test is unreliable. extraneous material in air passages indicates that child was alive. air reaches the stomach after min, the small intestine after - h, the colon after - h, and the rectum after l h. there is no difference in the speed of gas propulsion between full-term and premature infants. bacterial gas production and previous resuscitation attempts are potential sources of errors. in questionable stillbirth, presence of milk proves that infant was alive and had been nursed. traumatic lesions and signs of neglect may be present. unilateral ulcers in visceral zoster. in rare instances, diffuse meningoencephalitis may occur. limited necrotic and inflammatory lesions in the cord or brain stem at the level of the affected ganglion are common. posterior hom myelitis ( ). ganglion cell necrosis; lymphocytic infiltration, hemorrhage, and, later, fibrosis. as a rule, only one ganglion is severely involved, but less severe lesions may occur in the ganglia that are immediately adjacent. diffuse lymphocytic infiltration; demyelination; axonal destruction; fibrosis. conjunctivitis; keratitis; iridocyclitis; retrobulbar neuritis; neuroretinitis; occlusion of retinal vessels. evidence of hematologic malignancies ( ) or other conditions, as listed above under "note." infection, middle ear (see "otitis media.") infection, pneumocystis carinii synonym: pneumocystosis. note: ( ) collect all tissues that appear to be infected. ( ) this organism cannot be cultured at present but is frequently associated with viral, bacterial, or fungal infections that can be diagnosed by culture. ( ) for rapid staining of aspirates, use gram-weigert stain, which will stain cysts of pneumocystis carinii and also fungi and bacteria. pneumocystis carinii is best demonstrated with grocott's methenamine silver stain. ( ) no special precautions are indicated. ( ) usually, no serologic studies are available. ( ) this is not a reportable disease. injury, fire (see "burns.") injury, firearm note: protect all drains of autopsy table, which will prevent accidental loss of bullets or bullet fragments. recovery of bullet fragments and pellets can also be improved by passing tissue fragments, blood, and other appropriate materials through a fine nonmetallic sieve. caution must be exercised because sharp edges andjagged projections ofbullets and bullet fragments may cause injury ( ). this applies particularly to the black talon bullet. bullets and bullet fragments should not be touched with forceps or other metal instruments that may produce artifactual markings; they must be placed in properly labeled evidence containers, and the chain of custody must be preserved. from a birdshot wound, at least pellets should be recovered. the firearms examiner will divide the total pellet weight by and derive median pellet weight. from a buckshot wound, all pellets should be recovered. in many of these cases, procedures described under "homicide" must also be followed. do not excise wounds before completion of autopsy (see below). as an academic exercise, excised firearm wounds may be used later for analysis of metal traces by neutron activation or for tests for carboxyhemoglobin. in practice, however, a high quality photograph is adequate to establish the presence or absence of gunpowder stippling or soot deposition toward the end of establishing the range of fire. the dimensions of the stippled areas and soot deposits should be recorded. after completion of external examination, dry the wet clothing and keep as evidence. clothing may also be used for possible test firing or for stain, powder, or particle analysis. external examination if identity of victim is unknown, follow procedures described in chapter . prepare initial roentgenograms of the body regions that have been shot, before disrobing of body, and then lateral films of body regions with bullets. follow up with films of other body regions as needed in the course of autopsy. if the body is decomposed to the extent that the external examination cannot be relied upon to determine cutaneous gunshot perforations, prepare roentgenograms of the entire body before autopsy. record location and number of bullet holes in all layers of garments, indicating whether the involved area is bloodstained or shows gunpowder or soot. if there are several cutaneous perforations, number or letter each consecutively and refer to these numbers in all records. location of bullet holes should be described by recording distance from soles of feet or top of head, and from midline of body. prepare diagrams and photograph holes, with and without labels. photographs should show surrounding garments and extent of blood stains. in hairy areas, shave around bullet wounds for better photographic documentation. for evaluation of distance from where a shot had been fired, see fig. additional roentgenograms during the autopsy may be needed to find bullets embolized to the femoral veins or down the spinal canal. systemic roentgenograms may reveal bullets from obscure entrance wounds, particularly in decomposed bodies, old bullets, bullets in the skin flaps reflected at autopsy, or bullets external to the body in clothing. bullet(s) may have been arrested in hollow viscus or blood vessel and may have been transported to distant sites by peristalsis or bloodstream ("bullet embolus"). bullets may be deflected in unexpected directions from bony surfaces. soot ("smudging") and gunpowder stippling may occur around entrance wound, depending on the muzzle-to-skin distance. there may also be an impression from a recoiling handgun or from a power piston. wounds may be stellate, round, jagged, or slit-like, with or without wide margins of abrasion. primer residues on hand of suicide victim after use of a handgun. fatal secondary injuries, particularly hemorrhages. alcohol intoxication is a frequent finding. fig. · . bullet wounds. differences in the appearance of cutaneous wounds of entrance and of exit and differences in wounds of entrance according to the distance between muzzle and skin at the moment of fire. entrance wounds often differ from exit wounds in that the former are usually surrounded by a narrow zone of abrasion. if the muzzle of a gun is in close contact with the skin at the moment of fire, the combustion products will be blown into the wound and will not be visible on the surface. for close-range wounds, the stippling of the skin around the wound, produced by particles of burned and unburned powder, becomes progressively more dispersed as the range of fire increases and is ordinarily not perceptible if the range has been greater than inches ( synonyms and related terms: acute radiation syndrome; chronic (delayed) radiation injury; radiation enterocolitis; radiation nephritis; radiation pneumonitis. note: procedures and expected findings depend on type of radiation damage, whether acute or chronic, localized or whole-body irradiation. if suspected radiation injury was associated with the administration of radionuclides such as p, , or au, follow procedures suggested in chapter . in fatal whole-body irradiation, findings are most likely related to bone marrow injury, and the suggested procedures and expected findings are those described under "pancytopenia." record extent of oropharyngeal and intestinal ulcerations and hemorrhages. late complications include malignant tumors (carcinoma, leukemia,* lymphoma*), manifestations of hypothyroidism,* and cataracts. organ changes in localized radiation injury depend on site of irradiation (lung, brain, kidney, intestine). the skin is involved in both acute (erythema) and chronic (atrophy, epilation) radiation injury. related terms: cutting injury; knife injury. note: in many of these cases, procedures described under "homicide" must also be followed. insuffiency, adrenal synonyms and related terms: adrenocortical insufficiency; polyglandular autoimmune syndrome (type i, usually in childhood, with parathyroid insufficiency and chronic mucocutaneous moniliasis; type ii, usually in adults, with two or more autoimmune endocrine disorders such as thyroiditis and diabetes mellitus*); primary adrenal insufficiency (addison's disease); secondary adrenal insufficiency (in hypothalamic-pituitary disease or steroid induced); x-linked adrenal hypoplasia. is caused by anatomic changes (see below under "adrenal glands"), drugs (enzyme inhibitors or cytotoxic drugs), or acth-blocking antibodies. possible associated conditions: aids with systemic infections ( ); antiphospholipidantibody syndrome ( ); autoimmune hepatitis; chronic lymphocytic thyroiditis; type i diabetes mellitus;* hypoparathyroidism;* hypothyroidism;* megaloblastic anemia;* surgical procedures, such as heart surgery or orthopedic procedures. large cell lymphoma ( ). should be removed as soon as possible (before embalming), either from cervical midline incision or from chest (neck of cadaver must be well-extended). photograph obstruction before larynx is opened, and inside of larynx and epiglottis after larynx is is opened in posterior midline. make gram-stained touch preparations. make histologic sections of larynx and epiglottis. hemophilus injluenzae (cannot always be isolated from larynx synonyms and related terms: acute or chronic lymphocytic leukemia, acute or chronic myelogenous leukemia, and hairy cell leukemia. multiple subtypes have been identified; they are characterized by cell surface markers, chromosomal abnormalities, staining reactions, and morphologic features. note: at the time of autopsy, most leukemias have been properly classified and often treated. in these cases, the goal of the autopsy is to document the extent of the disease and the presence of complications. if the leukemia had not been classified or if the features might have changed from the time of the last work-up (e.g., in suspected cases of superimposed diffuse large cell lymphoma [richter's syndrome]), material should be snap-frozen and studied in more detail. if the patient was treated by bone marrow transplantation,* see also under that heading. possible associated conditions: agnogenic myeloid metaplasia with myelofibrosis; bloom's syndrome;* cutaneous mastocytosis; down's syndrome;* immunodeficiency syndromes* (bruton's disease; louis-bar syndrome; wiskott-aldrich syndrome); infantile genetic agranulocytosis; klinefelter's syndrome;* lymphoma;* multiple myeloma;* polycythemia;* and many others. for sampling and specimen preparation, see chapter . increased protein concentration. material in sediment stains red with toluidine blue. metachromatic material. arylsulfatase-a deficiency. excessive loss of myelin with large amounts of metachromatic material (see above under "urine") in white matter and also in some neurons. demyelination with metachromatic material (see above under "urine"). leukoencephalopathy, progressive multifocal possible associated conditions: aids* and other immunosuppressed states; carcinoma; malignant myeloproliferative or lymphoproliferative disorder; sarcoidosis;* tuberculosis. * procedures submit portion of fresh brain for viral culture. fix remainder of brain and spinal cord in formalin and submit samples for histologic study. in situ hybridization for ic virus is available on paraffin-embedded tissue. small patches of demyelination with tendency to form confluent areas in the cerebral white matter. white matter necrosis may be present. eosinophilic intranuclear inclusions occur in affected oligodendroglia cells. subsequently, large, bizarre astrocytes develop. no inflammatory (lymphocyte) cell reaction is present. lightning (see "injury, lightning.") lipoproteinemia (see "hyperlipoproteinemia.") lipoproteinosis, pulmonary alveolar synonym: idiopathic alveolar lipoproteinosis. note: the condition has been described in allografts ( ); in such cases, see also under "transplantation, lung." elephantiasis of penis, scrotum, or vulva (in chronic cases); skin rash (l) and conjunctivitis (in acute cases); genital ulcers. suppurative or fibrosing inguinal, iliac, or pelvic lymphadenitis, with or without sinus tracts. rectal strictures and fistulas in chronic cases ( ) . systemic involvement in the acute stage with pericarditis,* and arthritis,* and meningitis,* proctitis ( ) lymphoma synonyms and related terms: aids-related lymphoma; adult t-cellieukemiallymphoma; angioimmunoblastic lymphadenopathy; b-celllymphoma; burkitt's lymphoma; cutaneous t-cell lymphoma (includes mycosis fungoides and sezary syndrome); hodgkin's disease; non-hodgkin's lymphoma (low grade, intermediate grade, high grade, all with multiple subtypes too numerous to list, which vary in natural history); natural killer cell neoplasms; post-transplant lymphoproliferative disorders (ptld); t-cell lymphoma. note: at the time of autopsy, most lymphomas have been properly classified and often treated. in these cases, the goal of the autopsy is to document the extent ofthe disease and the presence of complications. if the lymphoma had not been classified or if the features might have changed from the time of the last work-up, material should be snap-frozen and studied in more detail. if the patient was treated by bone marrow transplantation, * see also under that heading. for current terminologies of hodgkin's disease and non-hodgkin lymphomas as well as for staging criteria, appropriate hematologic textbooks should be consulted. possible associated conditions: acquired immunodeficiency diseases (e.g., after organ transplantation; human immunodeficiency virus-l infection); autoimmune disease (e.g., celiac sprue,*rheumatoid arthritis,* systemic lupus erythematosus,* or sjogren's syndrome*); chemotherapy with or without radiation treatment; epstein-barr virus infection; human t-cell leukemia virus infection; inherited diseases with immunodeficiency (e.g., klinefelter syndrome; * common variable immunodeficiency disease, wiskott-aldrich syndrome); radiation. the lesions are found most commonly in the urinary bladder and other parts of the urinary tract, but may also occur at many other sites, including skin ( ) and upper respiratory tract ( ). note: (i) collect all tissues that appear to be infected. ( ) usually, cultures are not indicated. ( ) request giernsa or wright stain. tissue blocks should be rinsed of blood and be as thin as possible before fixation in refrigerated buffered and neutral % formalin solution. this is to avoid precipitationofformalin pigment that may be confused with malaria pigment. the formalin solution should be used in a ratio of parts formalin to one part tissue and should be agitated every few hours. if one is dealing with tissues that contain formalin pigment, this pigment may be removed with a bleaching solution. ( ) usually, no special precautions are indicated. ( ) possible associated conditions: histoplasmosis* and other chronic fungal infections; immune connective tissue diseases such as rheumatoid arthritis* (l) ; malignancies (l); sarcoidosis;* silicosis;* tuberculosis* (l) . note record location and extent of skin changes or skin defects on back. prepare skeletal roentgenograms. if meningitis is suspected, submit sample of cerebrospinal fluid for culture. dse the posterior approach to remove the spinal cord. atrophic skin over meningocele, lacking rete pegs and skin appendages; skin defect in complete rachischisis. bony defects of spine. in meningocele and spina bifida occulta, arachnoid and dura herniate through the vertebral defect. spinal cord and roots are generally not involved. lumbosacral mass in meningomyelocele, with a highly vascular mass (area medullovasculosa) in spina bifida aperta. neural defects in complete rachischisis. diastematomyelia, hydrocephalus ( ) . pyelonephritis;* enlarged urinary bladder ("neurogenic bladder"). external examination heart lungs procedures prepare chest roentgenogram. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. record weights. prepare photographs and roentgenograms of fresh and fixed lung specimens. perfuse one lung with formalin. for preparation of paper-mounted sections, see chapter . submit one lobe for microbiologic and chemical study. decalcify tissue for histologic study. request van gieson's, hale's colloidal iron, pas, and sudan stains. miliary mottling in lower lung fields. mitral stenosis* may be a cause of microlithiasis (mostly intra-alveolar ossification). * cor pulmonale and heart failure ( ) (l) , see also under that heading. ( ) collect all tissues that appear to be infected. ( ) viral isolation, especially with ebv, is not diagnostically useful, due to long incubation periods. ( ) note: these diseases are characterized by a deficiency of a variety of hydrolases, resulting in accumulation of gly-cosaminoglycans (mucopolysaccharides) and glycolipids within lysosomes of fibroblasts, macrophages, white cells, and parenchymal cells of many organs ( , ) . mucopolysaccharides are also excreted in the urine. the general approach is similar in all types. formalin may dissolve all of the stored material and leave empty vacuoles in the involved cells. therefore, frozen sections should be utilized or tissues should be fixed in absolute alcohol. the accumulated material will show intense metachromasia if stained with toluidine blue. it will also stain with pas, alcian blue, and colloidal iron. oil red a will also stain the material from frozen sections. for specimen preparation for elec-tron microscopy, see chapter is. characteristic external features include dwarfism; thickened long bones; coarse facial features; coarse hair; macrocephaly; prognathism; hypertelorism; malformed teeth, and scaphocephalic skull with hyperostosis of sagittal suture; short neck; chest deformity; umbilical and inguinal hernias; lower thoracic and lumbar gibbus; genu valgum and coxa valga; pes planus and other joint deformities; wide hands (clawhands) and feet. coarse thickened skin, covered with lanugo-like hair. hyperlordosis; ovoid deformities of vertebrae; odontoid hypoplasia; large, shoe-shaped sella turcica; kyphosis; hypoplasia of femoral heads; osteoporosis.* if patient underwent bone marrow transplantation ( ), see also under that heading. chronic upper respiratory infections. large cytoplasmic granules in neutrophils. storage of mucopolysaccharides in osteocytes, chondrocytes, and fibroblasts of periosteum, tendons, fasciae, and other connective tissues; dysostosis multiplex. other tissues histologic sampling cannot be excessive. ( ) collect all tissues that appear to be infected. ( ), see also under that heading. blood; other organs and tissues record extent of skin lesions and prepare photographs; prepare sections of affected and unaffected skin. request gram stain of sections and smears. submit samples of blood and grossly affected organs or tissues for microbiologic study. other procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. extensive epidermal necrosis. shedding of granular and horny layers of epidermis. septicemia; staphylococcal infection of nose, throat, ears, eyes, heart valves, urogenital tract, and other sites. bronchial epithelial detachment and bacterial pneumonia ( ). synonyms and related terms: acute kidney failure; *acute tubular necrosis; lower nephron nephrosis. note: the morphologic diagnosis of this condition may be difficult to discern from autolysis. the features that suggest tubular necrosis are: tubular dilation with epithelial flattening, intertubular edema and necrotic epithelial cells in the collecting ducts. the autopsy should be performed as soon as possible. needle specimens of the kidneys obtained in the immediate postmortem period may yield acceptable material. ifnephrotoxic drugs or chemicals are thought to be responsible for tubular necrosis, submit samples for toxicologic study (see also under synonyms and related terms: multiple endocrine neoplasia (men), type (parathyroid hyperplasia or adenoma; pancreatic islet cell hyperplasia, adenoma, or carcinoma; pituitary hyperplasia or adenoma); or type a (medullary thyroid carcinoma, parathyroid hyperplasia or adenoma; and pheochromocytoma); or type b (medullary thyroid carcinoma, pheochromocytoma, mucosal and gastrointestinal neuromas, and marfanoid features). note: in men, type , foregut carcinoids and subcutaneous and visceral lipomas also may be found. in type a, cutaneous lichen amyloidosis may be observed. mixed syndromes include (i) familial pheochromocytoma and islet cell tumor, ( ) von hippel-lindau syndrome, pheochromocytoma and islet cell tumor, ( ) neurofibromatosis* with features of men or , and myxomas, spotty skin pigmentation, and generalized endocrine overactivity (carney complex). in all instances, the autopsy should be done as soon as possible so that tissues for biochemical study can be frozen without delay. nephritis note: see under specific designation, such as "glomerulonephritis" and "pyelonephritis," or under name of suspected underlying condition, such as "gout" or "lupus erythematosus, systemic." nephroblastoma (see "tumor of the kidney(s).") synonyms and related terms: renal stones; urolithiasis. possible associated conditions: carcinomatosis; cushing's syndrome;* cystinuria;* fanconi syndrome;* hyperoxaluria;* hypervitaminosis d;* gout;* multiple myeloma;* osteoporosis;* polycystic renal disease;* primary hyperparathyroidism;* rheumatoid arthritis* ( ); sarcoidosis* ( ). kidneys ureters, urinary bladder, and urethra other organs remove kidneys, ureters, and urinary bladder en bloc. excise and bivalve kidneys to reveal renal pelves. open ureters. record size, number, and appearance of stones, and save for chemical analysis ( ). record heart weight. dissect and record weights of all parathyroid glands. other procedures depend on suspected underlying conditions, as listed above under "possible associated conditions." nephropathy note: see under name of suspected underlying condition, such as "diabetes mellitus," "disorder, electrolyte(s)" (hypercalcemia, potassium depletion), "gout," "hypertension (arterial), all types or type unspecified," or "poisoning,.. ." (heavy metal). if kidney failure was present, procedures under that heading should also be followed. renal tissue may need decalcification or fixation in water-free solution. ( ) note: in this condition, lesions in the brain and cranial nerves, spinal cord, and spinal roots may be schwannomas (including bilateral vestibular schwannomas); multiple meningiomas; gliomas (generally of spinal cord), mostly ependymomas ( %) and pilocytic astrocytomas; or schwannosis of spinal dorsal root entry zones. intracortical meningioangiomatosis; glial hamartia (intracortical, basal ganglia, thalamus, cerebellum, and dorsal horns of spinal cord) and cerebral calcifications also may be found. cutaneous abscess (j). acute necrotizing nocardial pneumonia with abscesses or sinus formation into surrounding tissues; empyema. pulmonary and mediastinallymph nodes other organs submit samples for culture and histologic study. submit samples from all organs and tissues with suspicious gross lesions for culture and histologic study. poorly encapsulated abscesses of any organ; endocarditis ( ) . related term: total parenteral nutrition. note: air embolism* or blood loss may have occurred if line became detached from the catheter hub. metabolic complications such as fluid overload or disturbances of acid-base and electrolyte balance often cannot be diagnosed reliably at autopsy. the disease(s) that may have necessitated parenteral nutrition therapy are not considered here; they include the acquired immunodeficiency syndrome (aids);* cancer cachexia; inflammatory bowel disease; liver or kidney failure;* severe pancreatitis;* short bowel syndrome, and others. obstruction, acute airway synonyms and related terms: aspiration; bolus death; "cafe coronary"; croup; restaurant death. note: ifpermission has been obtained, remove neck organs through straight incision from chest to chin to avoid dislodging a foreign body. ifdissection has to be accomplished from chest, neck of unembalmed cadaver should remain well extended during procedure. inspect larynx and trachea from above and below, respectively, before opening them carefully along the posterior midline. for removal, prosthetic repair, and specimen preparation, see chapter . submit samples of osteocartilaginous and synovial tissues for histologic study; sagittal or frontal saw section through spine provides for best routine evaluation. heberden's nodes at interphalangeal joints of fingers. degenerative changes, primarily of spine, hip joints, and knee joints. histologic and macroscopic degeneration of cartilage; exposure of subchrondral bone; formation of marginal osteophytes; bone cysts; synovial fibrosis; hypertrophic synovitis. o procedures if artificial joints (e.g., hip or knee) had been implanted, record state of implant site. loose joint prostheses. synonyms: hypertrophic pulmonary osteoarthropathy; pac-hydermoperiostosis (idiopathic hypertrophic osteoarthropathy [ j). note: in all instances, an underlying disease must be identified; they include cardiac disease (cyanotic congenital heart dis-ease with right-to-left shunt; infective endocarditis*); gastroeso-phageal reflux ( ); neoplasms of lungs, esophagus, intestine, and liver; chronic liver disease; inflammatory bowel disease;* pulmonary disease (e.g., abscess;* bronchiectasis;* cystic fibrosis;* emp-yema;* emphysema;* lipoid pneumonia* ( ); pneumocystis pneumonia; sarcoidosis;* tuberculosis*); hyperthyroidism. * external examination elastic arteries other organs prepare roentgenograms of extremities. for removal, prosthetic repair, and specimen preparation, see chapter . record presence of aneurysms (thoracic aorta, subclavian) or arteriovenous fistula of brachial vessels. if abdominal aortic aneurysm had been surgically repaired, search for evidence of graft infection ( ) . procedures depend on expected findings or grossly identified abnormalities as listed above under "note." clubbing of fingers or toes (see below under "elastic arteries"); swelling of extremities. periosteal new bone formation in distal shafts of bones of forearms and legs. all bones of extremities may be involved. see above under "external examination." unilateral clubbing. clubbing of toes but not of fingers. swelling; fistulas. bone defect(s) and focal osteosclerosis. bacterial or fungal infections, most common in metaphyseal region of bones; paravertebral or psoas abscess in osteomyelitis of spine. bacterial or fungal osteomyelitis, with or without cavitation and fistulas. trauma, surgery, insertion of prosthesis, generalized ( ) or contiguous infection, inflammatory bowel disease ( ), diabetes mellitus,* and peripheral vascular diseases, deep vein thrombosis ( ). synonyms and related terms: aseptic necrosis of bone; avascular necrosis of bone; idiopathic osteonecrosis; perthes' dis-ease; postfracture osteonecrosis; renal transplant associated osteonecrosis. note: possible underlying conditions include chronic alcoholism,* decompression sickness,* diseases that have been treated with high doses of corticosteroids ( ), gaucher's disease,* human immunodeficiency virus infection* ( ), tuberculosis* ( ), sickle cell disease, * systemic lupus erythematosus,* tuberculosis* ( ), and bisphosphonate therapy ( ). other organs submit sample for biochemical study. for removal, prosthetic repair, and specimen preparation, see chapter . if osteonecrosis is suspected because one of the possible underlying conditions (see above) is present, prepare saw sections through femoral heads, medial femoral condyles, and heads of humeri. procedures depend on suspected underlying conditions, as listed above under "note." osteopetrosis synonymsand related terms: albers-schonberg disease; autosomal-dominantosteopetrosis; carbonic anhydrase-it deficiency; malignant infantile osteopetrosis ( ); marble bone disease. note: manifestations of renal tubular acidosis may have been a clinical complication. possible associated conditions: bone marrow transplantation* (for infantile osteopetrosis). anemia, recurrent infections, bleeding, and bruises may have resulted from myelophthisis associated with osteopetrosis. upper airway obstruction in malignant infantile osteopetrosis may have necessitated tracheostomy ( ). in heritable osteoporotic disorders of connective tissue (osteogenesis imperfecta,* marfan' s syndrome,* and others) are pre-sented separately. for"osteomalacia," see above. for "osteodystrophy," see under "failure, kidney." possible associated conditions: acromegaly;* chronic alco-holism;* chronic obstructive pulmonary disease; chronic kidney failure* ( ); cushing's syndrome;* debilitating disease (various kinds, often with immobilization); diabetes mellitus* ( ); epilepsy;* hyperthyroidism ( );* hypogonadism; malabsorption syndrome;* malnutrition;* primary biliary cirrhosis;*rheu-matoid arthritis;* scurvy;* steroid therapy or anticonvulsant medication. normal parathyroid glands in uncomplicated cases. hyperparathyroidism* ( ) (without osteitis fibrosa) may be present, however. features of osteitis fibrosa (osteoclastic osteoporosis; see "hyperparathyroidism") or osteomalacia* exclude the diagnosis of uncomplicated osteoporosis. osteoporosis, which may be localized, occurs in various neoplastic (e.g., mastocytosis) and inflammatory diseases. external examination brain and base of skull with middle ears examine external ear canal. for removal and specimen preparation, see chapter . culture any lesion appearing to be infectious. draining, foul-smelling greasy material associated with acquired cholesteatoma. bacterial or viral infection, with or without mastoid osteitis. neck abscess, sinus thrombosis ( ) and brain abscess* and meningitis* may complicate otitis media. otosclerosis ( , ) note: otosclerosis may be a measles-virus-associated disease ( ) and therefore, studies to rule out a past infection may be indicated. for removal and specimen preparation, see chapter . for current methods of temporal bone studies, see ref. ( ) . spongy bone in the capsule of the labyrinth. trabeculae of woven bone show pagetoid changes. if stapedectomy had been done, a prosthesis may be in place. brain is externally normal or mildly atrophic. characteristic is midbrain atrophy with aqueduct dilatation and depigmentation of the substantia nigra. neuronal loss with reactive gliosis, associated with neurofibrillary tangles (globose tangles). primarily affected are globus pallidus, subthalamic nucleus, red nucleus, substantia nigra, tectum, periaqueductal gray matter, and dentate nucleus (grumose degeneration). palsy, pseudobulbar (see "disease, motor neuron.") related terms: acute pancreatitis; alcoholic pancreatitis; chronic (or chronic fibrosing) pancreatitis; interstitial pancreatitis. note: some causes of pancreatitis such as adverse drug reactions (e.g., due to azathioprine, furosemide, or estrogens) cannot be diagnosed at autopsy. possible associated conditions: acute fatty liver of pregnancy; apolipoprotein cli deficiency; cystic fibrosis; *hyperparathyroidism;* kidney transplantation. * status post endoscopic retrograde cholangiopancreatography. synonyms and related terms: calcifying panniculitis; histiocytic cytophagic panniculitis; mesenteric panniculitis (mes-enteric lipodystrophy); "sclerema neonatorum." note: the term weber-christian disease described systemic panniculitis with fever, bleeding, pulmonary and pancreatic lesions, and other abnormalities also involving lungs and pancreas, among others. however, this condition is not a specific entity and thus, the name has become obsolete. the term histiocytic cytophagic panniculitis is more descriptive and preferred. leukemia* and subcutaneous t-celllymphoma* may closely simulate panniculitis. possible associated conditions: alphaj-antitrypsin deficiency;* dermatomyositis;* rheumatoid arthritis;* scleroderma and morphea; sjogren's syndrome;* systemic lupus erythematosus.* external examination, skin, subcutaneous tissue, and breasts chest heart lungs liver and spleen record extent and character of skin lesions. record size, location, and gross appearance of subcutaneous nodules. submit tissue samples for histologic study of grossly unaffected skin, skin lesions, and subcutaneous nodules. request verhoeff-van gieson, gram, and grocott's methenamine silver stains. request s-l protein stain. ascertain that cell infiltrates are not leukemic or lymphomatous ( ) . dimpling of skin; necroses; fistulas. panniculitis with subcutaneous nodules in trunk, breasts, and thighs ( ) . calcification may occur in breast tissue but also at other sites (e.g., in kidney failure*). pancreatic enzymeinduced fat necroses associated with severe pancreatitis. widespread hardening of fat tissue with rupture of fat cells and crystal formation in "sclerema neonatorum." focal traumatic panniculitis also may occur in newborns. s-ioo stain negative in panniculitis but positive in rosai-dorfman disease (sinus histiocytosis with massive lymphadenopathy). mediastinal panniculitis. pericarditis;* interstitial myocarditis. * pneumonia;* pleuritis. fat embolism. * features of alphal-antitrypsin deficiency* ( ) . fatty changes of liver; splenitis. intestinal mucosal erosions and ulcers, with or without perforation. massive gastrointestinal hemorrhage in histiocytic cytophagic panniculitis. blind intestinal loop ( ). for sampling and specimen preparation, see chapter . prepare samples for electron microscopy. vacuolar myopathy. ( ) . external examination, skin, and mouth record extent of skin lesions; photograph; submit multiple samples for histologic study, preferably of areas that are free of secondary infection. prepare sections for immunofluorescent staining. bullous and other lesions of scalp, eyelids, nose, axillae, umbilicus, inframammary areas, back, hands, groins, genitalia, anus, knees, and feet. similar lesions may occur in the mouth. acantholysis is suprabasal or near granular layer. igg deposits on the surfaces of keratinocytes. note: (i) collect all tissues that appear to be infected. collect inguinal, popliteal, axillary, and supraclavicular lymph nodes for aerobic culture. photograph enlarged lymph nodes, and prepare smears of fresh cut surfaces. submit consolidated areas for microbiologic study (see above under "note"). perfuse both lungs with formalin. submit samples of pulmonary tissue and bronchial lymph nodes for histologic study. submit samples of grossly abnormal organs, exudates, or drainage fluids for culture and gram stain. hemorrhagic necrosis; variable suppuration. severe hemorrhagic edema; pneumonia with lobular to lobar features. large areas of necrosis teeming with organisms and minimal to suppurative inflammation. plasmacytoma (see "myeloma, multiple.") platybasia note: possible causes or associated conditions include arnold-chiari malformation;* basilar impression* orinvagination; fusion ofatlas to the foramen magnum; klippel-feil syndrome;* malpositioning of odontoid process; osteitis deformans (paget's disease of bone*); osteogenesis imperfecta;* osteomalacia;* rickets; syringobulbia, syringomyelia. * skull and spine prepare roentgenograms of skull and cervical spine. flattening of the base of the skull (angle formed bythe planeoftheclivus andtheplane of the anterior fossa exceeds °). pleuritis (see "effusion(s) and exudate(s), pleural.") pleurodynia, epidemic synonyms: bornholm disease; devil's grip; epidemic myalgia; epidemic myositis. blood heart and pericardium submit samples for viral culture. sample for histologic study and for viral cultures (coxsackievirus a and b; echoviruses). if pericardia! fluid can be obtained, submit for viral culture also. cultures may be negative and search for viral rna by in situ hybridization may be indicated. chest organs other organs dissect thoracic duct and its tributaries (see chapter ). perfuse one or both lungs with formalin. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. submit samples of cystic lesions for histologic study. submit samples of cystic lesions for histologic study. gas in thoracic duct. emphysema* of lungs. necrotizing enterocolitis in premature infants. pyloric stenosis, colitis ( ), redundant sigmoid colon, ischemic bowel disease. mucosal pseudolipomatosis ( ) . usually, cysts are subserosal in adults and located between muscularis mucosae and muscularis propria in infants and children. mucosa may be inflamed. extraintestinal cysts, as listed above under "abdomen." cysts may also occur in vaginal mucosa. include bright-field and polarized light microscopy, transmission electron microscopy, scanning or transmission electron microscopy with energy dispersive x-ray analysis ( ), ion beam instrumentation, and secondary ion mass spectrometry. the last three methods are time-consuming, require much skill and expensive equipment, and do not lend themselves well to quantitation. ideally, bulk analysis and in situ microanalysis should be used in combination. analyses can be conducted in specialized laboratories at the following centers (for charges and other information, consult the appropriate laboratories): prepare chest roentgenogram. submit sample for microbiologic study. refrigerate sample for possible immunologic study. submit one lobe or samples of all lobes for bulk analysis and for in situ microanalysis (see above under "note"). microincineration may permit preliminary dust analysis. for demonstration of asbestos fibers, see chapter and ref. ( ) . submit one lobe or segment for microbiologic study. for pulmonary arteriography and bronchography, see chapter . prepare roentgenograms of entire lungs and slices. perfuse one lung (or both, if only routine studies are intended) with formalin. submit samples for histologic study of nodular dust lesions, diffuse fibrotic lesions, grossly uninvolved areas, bronchi, and bronchopulmonary lymph nodes. for preparation of paper-mounted sections, see chapter . prepare samples for transmission and scanning electron microscopic study. for removal, prosthetic repair, and specimen preparation, see chapter . alveolar rupture with dissection of air into the mediastinum in neonates with respiratory distress (see "syndrome, respiratory distress, of infant") and in adult patients ventilated on volume respirators. external examination prepare chest roentgenogram. photograph mediastinum. explore major veins and record compression in cases of tension pneumomediatinum. roentgenograms provide the best permanent record of a pneumomediastinum. air bubbles in mediastinal soft tissues. part ii / diseases and conditions abdomen and neck organs perfuse one lung with formalin. other procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. search for evidence of trauma of other lesions that may have allowed air to enter soft tissues. intubation injury (j); perforation or rupture of major bronchus, trachea, or esophagus. alveolar rupture, e.g., in asthma* ( ), may not be discernible. bronchiolitis obliterans ( ), interstitial pneumonia* ( ) and other lung conditions also may lead to pneumomediastinum. dissection of air from lesions in abdomen (e.g., retroperitoneal colonic perforation [ ] ) or in neck area. pneumonia, all types or type unspecified note: if the type of underlying infection is known, follow procedures suggested under the name of the infectious disease. if the etiologic agent of the pneumonia is unknown, proceed as follows: (l) collect all tissues that appear to be infected. and grocott's methenamine silver stains. ( ) special precautions may be required (see chapter ). ( ) serologic studies may be helpful once a specific etiologic agent is suspected. thus, collect serum at the time of autopsy or procure serum that was collected prior to death. ( ) this may be a reportable disease. related terms: acute interstitial pneumonia (hamman-rich syndrome); desquamative interstitial pneumonia (dip); idiopathic organizing pneumonia (or "bronchiolitis obliterans with patchy organizing pneumonia [boop]" or "cryptogenic organizing pneumonitis"); idiopathic pulmonary fibrosis; lymphoid interstitial pneumonitis (lip); nonspecific interstitial pneumonia (nsip) (or "cellular interstitial pneumonia" [eip]); usual interstitial pneumonia (uip); fibrosing alveolitis; pulmonary alveolitis; and many others. note: the conditions listed under "related terms" are histologic variants of idiopathic interstitial pneumonia. vip is synonymous with idiopathic pulmonary fibrosis (ipf) ( ). diffuse alveolar damage is the histologic finding in patients with the adult respiratory distress syndrome* (ards) causing interstitial and intra-alveolar fibrosis. this is an acute condition, referred to as acute interstitial pneumonia when it occurs as an idiopathic form of rapidly progressive interstitial pneumonia. possible associated conditions: acquired immunodeficiency syndrome* (aids) in patients with with lip or nonspecific interstitial pneumonia ( , ); trauma and shock in ards. secondary pulmonary fibrosis from inhalants (extrinsic allergic alveolitis or pneumonia; pneumoconiosis*), in drug-induced pneumonia, hypersensitivity pneumonitis (microgranulomatous hypersensitivity reaction of lung), rheumatoid arthritis,* sjogren's syndrome,* and other collagen-vascular diseases; primary biliary cirrhosis in patients with nonspecific interstitial pneumonia ( ). external examination postmortem chest roentgenograms provide the only reliable and permanent record of a pneumothorax and its main complication, mediastinal shift due to a tension pneumothorax (fig. - ) . if roentgenograms cannot be prepared, a reasonably reliable diagnosis still can be made if the prosector inserts a needle through the lateral chest wall. the needle should be connected to a water-filled flask. ifa pneumothorax is present, gas bubbles appear in the flask as shown in fig. - . one can also expose the intact parietal pleura and observe if the lung tissue is separated from the parietal pleura by gas. incising the thorax at the base of a water-filled skin pocket is the least reliable method. infants can be totally submerged under water before the chest cavity is incised. however, care must be taken not to make an accidental incision into the underlying lung tissue. poisoning, all types or type unspecified note: if a specific substance is suspected-for instance, arsenic or ethylene glycol-follow procedures described under the appropriate entry. similar entries can be found for poisons whose general character or source is known-for instance, gas or mushroom poisoning. for some substances, the appropriate entry can be found under "abuse,...," "death,...," or "dependence, ..." in all instances, routine sampling of toxicologic material should be done as described in chapter . if no specific substances can be incriminated, the toxicologist must be provided with all available clinical information, as shown in chapter . in most cases of fatal poisoning, the coroner or medical examiner must be notified. poisoning, alkaloid note: see under specific name of alkaloid-for instance, "dependence, cocaine," "poisoning, atropine," "poisoning, digitalis," or "poisoning, strychnine." only a fraction of this large group of plant poisons has been listed. whether the specific name of the alkaloid is known or unknown, complete toxicologic sampling is recommended. poisoning, antimony ote: toxicologic material should be submitted for anal ysis, as suggested under "poisoning, arsenic." iatrogenic antimony toxicity may occur after treatment with antimony compounds for conditions such as filariasis, fungal infections, and schistosomiasis.* external examination and eyes pharynx and gastrointestinal tract heart, liver, and kidneys record skin changes and eye abnormalities. for toxicologic sampling of contents, see chapter . submit tissue samples for histologic study. request frozen sections for sudan stain. if exposure was from du t in smelting work, dermatitis and conjunctivitis may be present. severe gastroenteritis in acute poisoning. if victim drank antimony trichloride, ulcerative pharyngitis and gastritis may be present. fatty changes of myocardium and hepatic and renal parenchyma. poisoning, arsenic note: toxicologic material will be contaminated by bringing it in contact with fluids. keratinized tissues take up arsenic from solutions. put plastic bags over hands of victim. if exhumed bodies are investigated for arsenic poisoning, include material from surrounding soil and coffin along with tissues submitted for chemical analysis. interpretation of toxicologic findings: after fatal poisoning, arsenic concentrations in the liver tend to exceed . - . mg/ioo g wet tissue. in acute poisoning, arsenic concentrations in hair may reach ! g/g ( ) and in nails ! g/g. organs collect all contents, particularly in accidental poisoning in children. body dry and warm after death. mydriasis. iatrogenic atrioventricular block ( ) . fruits of atropa belladonna or seeds of datura stramonium may be found. no characteristic findings. poisoning, barbiturate(s) note: this type of poisoning has become uncommon. barbiturates may cause sudden death. in all instances, concomitant alcohol intoxication* must be ruled out. standard toxicologic sampling is sufficient. blood bile urine esophagus and stomach liver and brain procedures submit samples of blood from portal vein and peripheral veins or heart for toxicologic study. refrigerate for possible toxicologic study. record total volume and ph value. request tests for protein, glucose, and ketones; request drug screen. submit all contents and record their character. analyze for barbiturates and alcohol. submit samples for toxicologic study. evidence of alcohol intoxication.* poisoning by other addictive drugs. gritty residues of unabsorbed tablets, powder, or capsules. mucosal corrosion, ulceration, and discoloration from capsules may occur. concentration of barbiturate in parenchyma important for interpretation. poisoning, bismuth note: accidental poisoning is common (industrial exposure or drugs with soluble bismuth compounds). search also for other heavy metals. acute kidney failure* may be the cause of death. external examination and oral cavity stomatitis with bluish black discoloration of gums; loose teeth; sticky white membranous patches in mouth and throat. jaundice. protein casts and tubular epithelial cells in sediment. gray or black mucosal membranes; swelling of mucosa; intestinal ulcers that may be perforated. hemosiderosis. fatty changes; hemosiderosis. fatty changes; renal tubular degeneration with amorphic basophilic deposits in epithelium of convoluted tubules. hemosiderosis. see above under "external examination." for removal and specimen preparation, see chapter . peripheral neuritis. synonyms: bromine poisoning; bromism. note: the lethal dose is about . g in children and i g in adults (ingested). after fatal methyl bromide poisoning, headspace gas chromatography revealed a subclavian blood concentration of . microgram/ml whereas inorganic bromide concentrations were micrograms/ml in the blood ( ). tissue concentrations were lower than those in the blood. toxicologic samples should include liver and kidneys. chemical bums on face and conjunctivitis indicate direct exposure; skin pustules over body and nodose bromoderma of the legs indicate bromism. blood is best suited for bromide determination. after ingestion of bromide, necrosis with brown discoloration of mucosa of upper gastrointestinal tract may be present. after inhalation of bromide, swelling and inflammation ofmucous membranes in upper and lower respiratory tracts may be present. there may be pulmonary edema. pneumonia occurs in bromism. organs to be analyzed for cd should have no contact with water or be contaminated with blood; they should be sealed in polyethylene bags. cd leaks into fixation fluid. postmortem blood concentrations are very high and no indicator of the antemortem values ( ). external examination and oral cavity blood urine gastrointestinal tract other organs procedures submit sample for toxicologic study. submit sample for determination of cadmium concentration. submit sample for toxicologic study and a portion of one lobe for microbiologic study. perfuse one lung with formalin. fix bowel as soon as possible. collect renal tissue for light microscopic and electron microscopic study. sample for toxicologic study. submit samples for histologic study also. poisoning, carbon monoxide note: if the victim had been in a fire, see also under "bums." carbon monoxide poisoning may rarely be responsible for automobile accidents. relatively low carboxyhemoglobin concentrations may contribute to death if there is concomitant poisoning-forinstance, with alcohol or drugs, particularly sedatives. anemia, atherosclerotic heart disease, and chronic pulmonary disease also increase sensitivity to carbon monoxide. ifblood had been withdrawn at time of hospital admissionfor instance, for crossmatching-submit this for carboxyhemoglobin determination. if no blood can be obtained, see under "heart, kidneys, and other organs." for quick-orienting qualitative tests, for quantitative methods of carbon monoxide determi-nation, and for interpretation of toxicologic findings, see below. request also determination of hemoglobin concentrations and of blood alcohol. request drug screen. for shipping of blood and tissues for carbon monoxide determination, see chapter . it should be noted that losses of up to % of the original saturation occurred when blood was kept in uncapped container at room temperature for yz wk or at °c for wk. external examination blood heart, kidneys, and other organs brain record color of fingernails, particularly in heavily pigmented persons in whom lividity is difficult to discern. record appearance of blood and submit sample of postmortem blood for toxicologic study. see also above under "note." if no blood can be obtained, prepare water extract of spleen, kidneys, or other organs. request determination of carbon monoxide content and of carbon monoxide-binding capacity of this mixture. submit tissue samples for histologic study. for removal and specimen preparation, see chapter . pink skin and fingernails; bullous edema of skin; decubital ulcers. blood tends to be cherry red. for interpretation of toxicologic findings, see below. necrosis of papillary muscles in the heart or myocardial infarction may occur. renal tubular degeneration may also be found. acute kidney failure* has been observed after rhabdomyolysis complicated by compartment syndrome ( ) . hemorrhagic necrosis of basal ganglia (lenticular nucleus in globus pallidus); diffuse petechial hemorrhages in white matter; cerebral edema. acute hydrocephalus in infants ( ) . many methods of carbon monoxide determination have been described. currently, carboxyhemoglobin is detected in most medical examiner toxicology laboratories by visible spectrophotometry or gas chromatography. in hospitals, carboxyhemoglobin is frequently detected and reported in the course of routine arterial blood gas analysis. pink discoloration of skin and organs usually indicates the presence of more than % carboxyhemoglobin (but rule out cyanide poisoning* and exposure to cold*). in a healthy, middle-aged person, a carboxyhemoglobin concentration greater than - % is usually fatal. if the victim was anemic or suffered from chronic lung disease, particularly emphysema* or atherosclerotic heart disease, the concentration may be lower. in association with alcohol, sedatives, and other drugs, carboxyhemoglobin levels may also be much lower and yet fatal. a heavy cigarette smoker may have a carboxyhemoglobin concentration of - %, and higher levels may occur in police officers and other persons exposed to automobile exhaust in dense traffic. if the victim survived the carbon monoxide poisoning for several hours, postmortem blood samples usually will fail to show the presence of carboxyhemoglobin. in these instances, blood taken at the time of admission to the hospital may still be available and of particular value. if the victim had spent h in fresh air before death, - % of the carbon monoxide will have been removed, and - % will have been removed during each subsequent hour. even though clearance may be complete, death may still occur-primarily from brain damage and infectious complications in prolonged coma. or delayed by only a few hours, particularly after inhalation of carbon tetrachloride. sudden death probably is caused by cardiac dysrrhythmia. alcohol concentrations should be determined in all cases or, if death was delayed, evidence of drinking at the time of exposure should be sought. alcohol considerably increases the hazards of carbon tetrachloride. note: see also under "bronchitis, acute chemical" and under "poisoning, gas." hydrochloric acid is sold by plumbing supply houses and pool supply companies as muriatic acid. it is a liquid. chlorine is a water-soluble gas. as supplied for pool sanitation, liquid chlorine is usually acidic. for convenience, chlorine and hydrochloric acid are discussed here together. prepare photographs of face. conjunctivitis and cyanosis in chlorine gas poisoning; burns of lips from hydrochloric acid. see also above under "larynx and trachea." photograph opened esophagus and stomach. sample for histologic study, particularly if there is doubt whether a perforation was antemortem or postmortem. for removal and specimen preparation, see chapter . severe pulmonary edema, bronchopneumonia, and swelling of mucous membranes in chlorine poisoning. arterial thrombosis may occur. pulmonary fibrosis may develop after prolonged survival. swelling and ulceration of mucous membranes in chlorine poisoning; acute laryngotracheitis. tracheoesophageal perforation. corrosion of mucosa with thickening, hemorrhage, and blackish discoloration after ingestion of hydrochloric acid. antemortem and postmortem perforation may occur. glomerular capillary thromboses. hemorrhages in white matter ( ). poisoning, cyanide synonym: hydrocyanic acid (hydrogen cyanide) poisoning. note: hydrocyanic acid (hydrogen cyanide, hcn) is a water-soluble gas. its salts, sodium cyanide and potassium cyanide are sold as "eggs" to the jewelry industry. hydrocyanic acid is formed when cyanide salts are dissolved in acidic solutions. containers from which the poison might have been ingested or inhaled should also be submitted for toxicologic examination. for cyanide screening tests in the autopsy room and for interpretation offindings, see below. caution: stomach may still contain cyanide gas, formed by acidic reaction of cyanide salt. it may be best to open the stomach under a hood ( ). the odor is quite characteristic for cyanide poisoning but most persons are unable to smell this odor. it is helpful to know in advance if any person in an office or laboratory can smell cyanide. forensic pathologists who can smell the compound state that it has its own specific odor, which differs from the often quoted smell of bitter almonds. autopsies also can be done in a negatively pressured isolation room ( ). for odor, see above under "note." bright red skin color is not always present. corrosion around mouth and in oral cavity may be found after ingestion of potassium or sodium cyanide. blood is fluid and sometimes bright red. if potassium or sodium cyanide was ingested, brown-red mucosal corrosion may be present in stomach or in upper digestive tract. pulmonary edema. for odor, see above under "note." if death was not instantaneous, there may be hyaline thrombi in small blood vessels, minute hemorrhages, and necroses of lenticular nuclei. this test can be used for blood and gastric contents ( ) . dip squares of filter paper in a small amount ofsaturated picric acid. let these squares dry until barely moist. place a drop ofthe material to be tested-e.g., blood or gastric contents--on a piece of paper. let material dry for a moment, and then place one drop of % sodium carbonate in the center of the material to be tested. ifcyanide is present, a reddish purple color will chromatograph out from the material. the higher the concentration of cyanide the more blue the color will be. it is possible to recognize whole blood because the blood turns a rather dark brown and the reddish to purple color is clearly visible. high concentrations of sulfide interfere by giving a false-positive test. another screening test is done as follows ( ) . dip filter paper into normal blood. then treat the paper with potassium chlorate, whereupon brown methemoglobin forms. place this preparation into the fluid suspected of containing cyanide (e.g., blood, gastric contents, pulmonary edema fluid). if bright red cyanmethemoglobin forms, the reaction is positive. if the concentration of cyanide in the stomach is high and the concentration in the lungs is low, cyanide was ingested. alternatively, if the pulmonary cyanide concentration is high and the concentrtaion in the gastric contents is low, hydrogen cyanide most likely was inhaled. occasionally, minimal cyanide levels will be present in decomposed bodies. related term: digoxin toxicity. note: certain drugs may interfere with correct digitalis determination. blood heart vitreous procedures submit sample of peripheral blood for digoxin radioimmunoassay. freeze fresh myocardium for digitalis extraction. submit sample for digoxin radioimmunoassay. poisoning, drug(s) (see "dependence, drug(s), all types or type unspecified" or under "poisoning,•••" followed by specific name of drug.) poisoning, ethanol (ethyl alcohol) (see "alcoholism and alcohol intoxication," "cardiomyopathy, alcoholic," "disease, alcoholic liver," "syndrome, fetal alcoholic," and syndrome, wernicke-korsakorr.") poisoning, ethylene glycol related term: antifreeze poisoning. note: pulmonary and cerebral manifestations are the main findings in acute poisoning, and renal tubular necrosis is the primary finding in chronic poisoning. for general toxicologic sampling, see chapter . calcium oxalate crystals can be demonstrated in routine histologic sections but also in scanning electron micrographs of thick deparaffinized sections. eyes blood urine for removal and specimen preparation, see chapter . in acute cases, submit sample for ethylene glycol determination; in chronic cases, request determination of calcium concentrations. prepare sediment. papilledema; optic nerve atrophy. ethylene glycol in serum (i) . protein casts; calcium oxalate crystals ( ) . crystals are light yellow and birefringent, arranged as sheaves, rhomboids, or prisms. poisoning, food related terms: bacillary dysentery* (shigella food poisoning); botulism;* clostridium perfringens food poisoning; favism; mushroom poisoning;* salmonella food poisoning; staphylococcal food poisoning. note: if cause of food poisoning is unknown, submit suspected food for aerobic and anaerobic cultures, gram stain of smears, and routine toxicologic study. this should include tests for heavy metals (antimony, cadmium, and lead) that may have leaked from old cooking utensils. test for the presence of staphylococcal enterotoxin are done only in specialized laboratories. if botulism is suspected, follow procedures described under that heading. mushroom poisoning also is listed as a separate entity. if salmonella food poisoning is suspected, see under "fever, typhoid." see also under "enteritis" or "enterocolitis" or under another specific heading such as "dysentery, bacillary." obtain sufficient material for microbiologic and histologic study to identify organisms such as chlamydia, clostridium (type f strains), salmonella, shigella, verotoxic e. coli, yersinia, and others. external examination gastrointestinal tract submit contents for aerobic and anaerobic cultures (see above under "note"); prepare smears of contents for gram stain. submit samples for histologic study. debilitated states; patients in extremes of life. enteritis or enterocolitis. poisoning, gas note: anesthesia-associated death, * carbon monoxide poisoning,* and sniffing and spray death* are presented under the appropriate headings. procedures discussed here deal with other volatile substances, including chemical irritants such as ammonia (nh ), chlorine or hydrochloric acid poisoning (see also under that heading); methylene chloride, phosgene (coci ), sulfurous acid (h s ), or sulfur dioxide (s ) ' gases from body cavities, heart chambers, or blood vessels can be removed as described under "embolism, air." gases can also be trapped with a rubber dam after cutting organs under water. samples from various organs should be shipped in hermetically sealed nonplastic containers or in analyzing solutions. external examination, and oral cavity blood larynx and trachea record extent of chemical bums. submit sample for gas analysis. in many instances, inhaled gases can be demonstrated chromatographically in gas from head space above sealed blood specimen. chemical bums in and around mouth or conjunctivas. chemical bums. other organs if gas was inhaled and is to be analyzed, submit intact lungs with bronchi ligated in airtight, nonplastic container to laboratory that can conduct gas analysis. if survival was short, formalin perfusion of lungs is not recommended; it may cause artifactual ballooning and internal ruptures of organ. submit samples of tissue for routine histologic study. see above under "note." chemical pneumonia; pulmonary edema. after longer survival, obliterating fibrous bronchiolitis, chronic bronchitis, and saccular bronchiectasis* may occur. poisoning, glycol (see "poisoning, ethylene glycol.") poisoning, halogen (see "fluorosis," "poisoning, bromide," "poisoning, chlorine or hydrochloric acid," "poisoning, gas?, and "poisoning, iodine!') poisoning, heavy metal (see "poisoning, antimony," "poisoning, arsenic," "poisoning, cadmium," "poisoning, lead," poisoning, mercury," "poisoning, thallium!') poisoning, insecticide (see "poisoning, organophosphate(s)*) poisoning, iodine related terms: lugol's solution; tincture of iodine. for toxicologic sampling, see chapter . external examination and eyes urine, blood, and parenchymal organs lungs and upper respiratory tract stomach record color of skin and extent of corrosive lesions. submit samples for toxicologic study. submit samples for histologic study. submit contents for toxicologic study; photograph mucosa; prepare histologic sections. see above under "stomach." perioral corrosive lesions; yellow discoloration of skin; conjunctivitis after exposure to vapors. acute inflammation of respiratory tract after inhalation of vapors. corrosive gastritis; if starch was used as antidote, gastric lining will be bluish. histologically, well-preserved mucosa is present because of in vivo fixation. mucosa may show same changes as stomach. swelling of tubular epithelium. synonyms: propanol; rubbing alcohol. procedures see under "alcoholism and alcohol intoxication." nonspecific autopsy findings: visceral congestion; pulmonary and cerebral edema. poisoning, lead note: lead-free syringes and lead-free polyethylene containers should be used. blood lead concentrations can be determined by inductively coupled plasma mass spectrometry ( ). for screening methods, see ref. ( ) . this is a reportable disease in some states. external examination, oral cavity, and hair bluish lead line at gingival margin in victims with poor oral hygiene. external examination, oral cavity, and hair for diagnosis of chronic plumbism, analysis of scalp hair may be useful. analysis is done by neutron activation (see also under "poisoning, arsenic"). remove samples with lead-free syringe (see above under "note") or -ml vacutainer tubes (becton, dickinson and company). do not add anti-coagulant or preservative. for coliection procedures, see above under "blood" and under "note." request also determination of coproporphyrin concentrations. submit sample for histologic study; submit remaining tissue for toxicologic analysis. submit with contents for toxicologic study, particularly in acute poisoning. submit sample from each kidney for histologic study; submit remaining tissue of both kidneys separately for toxicologic study. submit at least g of fresh bone for toxicologic study. for removal and specimen preparation, see chapter . poisoning, lsd (d-lysergic acid diethylamide) (see "abuse, hallucinogen(s).") poisoning, lye related terms: ammonium hydroxide poisoning; calcium oxide or quicklime poisoning; poisoning by alkaline corrosives; potassium hydroxide poisoning; sodium hydroxide poisoning. external examination and oral cavity blood neck organs, esophagus, trachea, and lungs record extent of oral, perioral, and other facial corrosive injuries. photograph lesions. prepare histologic sections of tissue from inside of lips or mouth. submit sample for toxicologic study. after removal of heart, remove neck organs with hypopharynx, esophagus, larynx, and trachea. leave stomach attached to esophagus. open pharynx and esophagus along posterior midline. in acute cases, formalin perfusion of lungs is not recommended; it may cause artifactual baliooning and internal ruptures of organ. lye bums on face and chest in acute cases; scars and manifestations of malnutrition* in chronic cases. sweliing, edema, and necrosis of mucous membranes in acute poisoning. fibrosis and strictures in chronic cases. bronchitis* and bronchopneumonia. submit specimen from pharynx for histologic study. for removal and specimen preparation, see chapter . submit sample of brain for toxicologic study. exfoliative dermatitis. blue line at gingival margin; hypertrophy of gum; acute and chronic gingivitis; exfoliation and loss of teeth ( ) . degeneration of myocardium. increased concentrations of mercury ( ) . congestion. erosive gastritis and colitis. increased concentrations of mercury ( ) . degeneration of proximal tubules; calcifications. chronic kidney failure* may be the cause of death. induration of mucosa. increased concentrations of mercury ( ) . cortical hemorrhages. cholinesterase activity will be low. pulmonary edema if poison was inhaled. airways may contain aspirated material. cholinesterase activity can be determined reliably even after decomposition and embalming. clinically, guillain-barre syndrome has been observed after poisoning with organophosphate. in acute poisoning, the cholinesterase levels may be % of the normal values (see below). the cholinesterase levels in the blood are not affected by the duration of the postmortem interval; measurement may be attempted even on decomposed or exhumed bodies. severe fatty changes; periportal necroses. fatty changes in myocardium, skeletal muscles, and other organs. for toxicologic sampling (gastric contents, urine, blood, brain, and other organs), see chapter . rigor mortis after fatal strychnine poisoning may occur very soon after death and may be very severe (opisthotonos); it may persist until decomposition sets in. congestion of viscera; no characteristic morphologic autopsy findings. acute pancreatitis has been observed ( ). high strychnine concentrations (also demonstrable in exhumed bodies [ ] ). poisoning, thallium note: for toxicologic sampling, see below and chapter . thallium is used as a rodenticide and pesticide, and has some industrial applications. retrobulbar neuritis. chapter and . submit brain for toxicologic study. submit sections of brain and optic nerves for histologic study. submit specimens for toxicologic study (take from lower extremity). for removal, prosthetic repair, and specimen osteomalacia;* osteomyelofibrosis. preparation of bones, see chapter . for preparation of sections and smears of bone marrow, see chapter . submit samples for toxicologic study. synonym: acute anterior poliomyelitis. note: the disease has been nearly eliminated in the usa but not in many other countries. ( ) collect all tissues that appear to be infected. ( for removal and specimen preparation, see chapter . in acute cases, submit portions of brain and spinal cord for viral culture. neurogenic atrophy of skeletal muscles in areas of paralysis. electrolyte disorder. * hypertensive heart disease; myocarditis.* aspiration or bronchopneumonia (or both); edema; atelectasis; embolism;* alveolar wall necrosis (acute or organizing diffuse alveolar damage) after oxygen toxicity. acute ulcers. acute gastric dilatation; acute gastroduodenal ulcers; gastrointestinal erosions and hemorrhages. dilatation of colon; perforation of cecum. urolithiasis and nephrolithiasis,* pyonephrosis and pyelonephritis. * phlebothrombosis of legs, most commonly on left side. necrosis of anterior hom cells of spinal cord, with neuronophagia and perivascular inflammatory reaction. old lesions show neuronal loss and gliosis. medulla ("bulbar polio") and other areas of brain stem, cerebellum, and cerebrum, particularly the motor cortex, may be affected in various degrees. part ii / diseases and conditions pregnancy note: in some instances, procedures described under "death, abortion-associated," "embolism, amniotic fluid," or "toxemia of pregnancy" may be indicated. synonym: werner syndrome. external examination and skin abdomen procedures record volume of intraperitoneal fluid; prepare smears; submit samples of peritoneum for histologic study. colloid carcinoma of stomach or colon. well-differentated adenocarcinoma of ovary or, rarely, of appendix; ruptured appendiceal mucinous adenoma is the most common cause of peritoneal adenomucinosis. pseudotumor cerebri synonym: benign intracranial hypertension; meningeal hydrops. note: this condition, which generally affects young, obese females, is characterized by symptoms and signs of increased intracranial pressure without a demonstrable cause. hence, intra-cranial mass lesions, infections, and related conditions should be excluded. such conditions include adrenal insufficiency;* guillain-barre syndrome* (increased colloid-osmotic pressure); hyperadrenalism; hypervitaminosis a* (e.g., after treatment of acne); hypoparathyroidism;* hypothroidism;* infectious mono-nucleosis;* lyme disease; pregnancy; sydenham's chorea; throm-bus ofthe lateral or superior sagittal sinus (otitic hydrocephalus); and wiskott-aldrich syndrome. submit samples of all accessible organs and tissues, with or without gross lesions. submit material for electron microscopy. for removal and specimen preparation, see chapter . for removal, prosthetic repair, and specimen preparation, see chapter . heart small bowel liver procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. fix the bowel as soon as possible. record weight. submit samples for histologic study. for removal, prosthetic repair, and specimen preparation, see chapter . aortitis involving ascending aorta and aortic valve with regurgitation; mitral valve and adjacent myocardium and conduction system also may be involved. sprue-like changes with loss of villi. fibrosis or cirrhosis and other abnormalities, particularly in patients who had been taking methotrexate. psoriatic arthritis and spondylitis. synonyms and related terms: allergic purpura; anaphylactoid purpura; hypersensitivity vasculitis. external examination and skin rabies note: (i) in most instances, an autopsy limited to the brain is sufficient to confirm the diagnosis. ( ) rabies virus is especially infectious and thus, universal precautions should be strictly followed (see chapter ) . avoid the use of scalpels whenever possible. the generation ofaerosols should be assidu- see above under "note." if a complete autopsy is done, sample heart, lungs, kidneys, pancreas, submaxillary salivary glands, and adrenal glands; freeze or refrigerate sampled tissues and submit for viral study (see below under "brain and spinal cord"). obtain serum for serologic studies. freeze or refrigerate cerebral tissue immediately after removal and submit frozen to special laboratory for fluorescent antibody staining. take these sections from hippocampus or brain stem. place the refrigerated tissues in % neutral glycerol saline solution for preservation. fix remaining brain and spinal cord tissue in % formalin and submit for histologic study. as an alternative to freezing of tissue, immunoperoxidase methods for detection of rabies viral antigens can now be applied to formalin-fixed tissue ( in almost all instances, the procedures described under "assault" and under "homicide" must also be followed. see also refs. ( ) and ( ) . for the evaluation of bite marks, photographs with and without scales and black and white film should be used. a forensic dentist is required to compare the victim's bite marks to the teeth of suspects. swabs of possible sperm or other fluids must be sent to the crime lab for proper evaluation. external examination other organs and tissues genital organs comb pubic hair over a towel and then pull sample; also, pull samples of hair from head. collect fingernail clippings and place in containers marked "right" and "left." collect blood, hair, fibers, and residues of urine or saliva and/or semen that may have stained the victim's clothing or that may be found on the skin of the victim. for study of stains and scrapings, see below. introduce sterile dry cotton swab into the posterior vault of the vagina or-preferablyinto the cervical canal. prepare smear on glass slide and send to crime laboratory for identification of sperm (see also above under "note" using a sterile cotton swab moistened with sterile saline, lift suspected dried spermatic fluid from hair or skin of external genitalia, perineum, buttocks, or other sites. stains on clothing will be extracted by forensic laboratory personnel. let specimens air dry in absence of sunlight and then place in paper bags or envelopes and seal. smears are made for the demonstration of spermatozoa, cytologic changes, bacteria, and other possible findings (some crime labs prefer to make their own smears from the swabs). other samples are used for the determination of acid phosphatase, as described below. usually, the survival time of morphologically recognizable spermatozoa in the vagina is less than h, but on occasion this may be much longer. thus, within the first h after coitus, % of cervical smears have been found to be positive for spermatozoa. at day , spermatozoa have been found in % of the smears. obviously, if the assailant had had a vasectomy, spermatozoa will not be found at any time. spermatozoa have been found in the vagina of some women several days or weeks after death. dried stains (see above) may give positive results after longer intervals. acid phosphatase activities greater than bodansky units (for method, see above) indicate that probably less than h have elapsed since the time of intercourse (i) . in another study, vaginal swab specimens showed acid phosphatase activities of more than , king if there is evidence of parotid involvement, other salivary glands should also be studied. parotid gland can be biopsied from scalp incision. submaxillary gland can be removed with floor of mouth. for sampling and specimen preparation, see chapter . for removal, prosthetic repair, and specimen preparation, see chapter . chronic meningitis; space-occupying lesions (submeningeal nodular granulomas). granulomatosis. iridocyclitis; chorioretinitis; papilledema; posterior uveitis; keratoconjunctivitis sicca; conjunctival follicles; cataracts. involvement of lacrimal glands and other orbital tissues ( ) . sarcoidosis of parotid gland is common. sarcoid neuropathy and sarcoid myopathy. cystic changes, primarily of small bones of hands and feet. sarcoidosis of joints ( ). synonyms and related terms: bilharziasis; schistosoma haematobium infection; schistosoma japonicum infection; schis-tosoma mansoni infection. note: unless specifically stated, the changes listed below refer to chronic schistosomiasis mansoni and japonica. ( ) collect all tissues that appear to be infected. ( ) request direct examination for schistosoma. the following procedures have been described and recommended ( ). for demonstration of schistosoma eggs, compress -mm tissue fragments-for instance, mucosa of the urinary bladder-between glass slides. if this gives negative results, digest a -g portion of tissue in potassium hydroxide. for the recovery of adult worms of schistosoma mansoni and schistosoma haematobium, remove the viscera en bloc and rinse with water. separate the intestines from the mesentery. subsequently, perfuse the portal vein system, the liver, and one lung with saline ( ) . then pass the perfusion fluid through a monofilament nylon cloth with an aperture size of m. submerge the cloth in water and examine with a dissecting microscope. fix worms in formalin solution. examine the intestinal mucosa directly. from the urinary bladder, ureters, and surrounding connective tissue, worms can be recovered as follows. inject water into the tissue until the tissue increases or times in thickness. then compress the tissue gently with a glass plate. cut slices . - . cm in thickness. compress these slices between the glass plate and the stage of a dissecting microscope and examine for the presence ofadult worms. many worms also will be present in the fluid expressed from the tissue as it is cut. for the counting of eggs in tissues, urine, and feces, see ref. ( ) . immunodiagnostic methods (elisa and imrnunoblot) also have been developed prepare skeletal roentgenograms. there are no diagnostic laboratory tests but it still is advisable to save a blood sample. record heart weight. test competence of valves (chapter ). open heart in line of blood flow. photograph and measure valvular lesions. prepare sections of valves, myocardium, conduction system (if there was a history of heart block), and ascending aorta. perfuse at least one lung with formalin. submit any consolidated area for microbiologic study. fix intestines as soon as possible. abnormalities as listed in right-hand column. sample for histologic study and request amyloid stains or renal parenchyma. prepare roentgenograms of spine, sacroiliac joints, symphysis ossium pubis, and manubriosternal, sternoclavicular, and humeroscapular joints. if spine cannot be removed in its entirety, it can be split in midline and one half can be removed, with costovertebral and costotransversal joints. hip joints should be exposed. histologic sections should include synovia and periarticular tissue. secondary and peripheral osteoarthritis* may be present. leukemia* developed in some patients who had had radiation treatment ( or earlier). acute anterior uveitis. ( ) request fungal culture. ( ) request grocott's methenamine silver stain. ( ) record weight of all organs (for expected weights, see part iii). submit samples of all major organs, including endocrine glands, lymphatic and fat tissue, bone, and bone marrow for histologic study. hunger edema; skin changes secondary to vitamin deficiencies. hypoproteinemia. atrophy; hemosiderosis of spleen. degree of atrophy varies from organ to organ; atrophy of fat tissue, lymphoid tissue, and gonads usually is most pronounced. severe infections, such as tuberculosis,* may be present without having been apparent clinically. steatohepatitis, nonalcoholic (nash) note: the morphologic findings in the liver are indistinguishable from those in alcoholic liver disease* ( ). follow autopsy procedures described under "disease, alcoholic liver." if the patient had received a liver transplant, procedures described under "transplantation, liver" should be followed also. possible associated conditions: celiac sprue (rare); diabetes mellitus* (type ); hyperlipidemia; malnutrition* from bulemia (rare); morbid obesity with liver failure particularly after episodes of rapid weight loss (e.g., after recent gastroplasty); lipodystrophy; mild obesity;* short bowel syndrome (rare); total parenteral nutrition. * stenosis, congenital valvular pulmonary related terms: isolated (pure, simple, or dome-shaped) pulmonary stenosis. note: the pulmonary valve is usually acommissural in isolated congenital pulmonary stenosis. with other coexistant congenital heart disease (such as tetralogy), the pulmonary valve is usually bicuspid and hypoplastic, but may be unicommissural or dys-plastic and tricuspid. heart and great vessels; peripheral pulmonary arteries brain procedures culture any grossly infected valve. for general dissection techniques, see chapter . record weight of heart, thickness of ventricles, and annular circumferences. for removal and specimen preparation, see chapter . infective endocarditis* of pulmonary valve and tricuspid valve; infective endarteritis at bifurcation of pulmonary trunk. it is preferable to have autopsies of fetuses and stillborns performed by pathologists experienced in perinatal pathology. if such personnel are not immediately available, the attending pathologist may nevertheless collect important information. if attempts to induce abortion appear to have caused the death of the mother, see "death, abortion-associated." the placenta should be immediately procured from the delivery room should it not arrive in the laboratory with the fetus. this will avoid the possibility of the placenta being discarded by the delivery room staff. no fetal autopsy is complete without a careful examination of the placenta (see part i, chapter ). pathologic changes of placenta that are causative of stillbirth are summarized in ref. . fascia lata or other aseptically obtained tissue should be collected for tissue culture for karyotype analysis. if the fetus is at all autolyzed, then the fascia lata cells may not grow in tissue culture. therefore, if the fetus shows any evidence of autolysis, tissue should be taken aseptically, from placenta immediately beneath the chorionic plate. a portion ofplacenta and liver should also be snap-frozen for possible molecular analysis. the initial stage of the autopsy should include photography and radiography, taking anterior-posterior and lateral views. the photographs will record the degree of maceration, which can be roughly correlated with the duration of fetal demise before delivery [ ] external measurements should include body weight, circumference ofhead, chest and abdomen, crown-rump length, crown-heel length, and foot length. these measurements are compared with tables of standards for normal fetuses (see part iii of this book). assessment of growth retardation may be based on these data. a careful external examination should search for abnormalities such as jaundice, bulging fontanel, cranial bone softening, hyper-or hypotelorism, choanal atresia, external ear anomalies, cleft lip, cleft palate, macroglossia, micrognathia, colobomata, cystic hygroma, shortened neck, contractures, omphalocele, gastroschisis, abnormal external genitalia, anal atresia, absent vagina, sacral pits, open neural tube defects, hemihypertrophy, syndactyly, clinodactyly, transverse palmar creases, or incomplete descent of testes. the organ bloc may be removed in the manner similar to the adult, using the technique of letulle (see chapter ) . if the thyroid gland is noted to be in its usual location and if it appears normal, then the tongue may be left in the body. in macerated fetuses, it is suggested that the organ bloc be fixed overnight in formalin solution prior to dissection. to aid adequate fixation, the following simple steps may be done: i) wash the organ bloc thoroughly prior to fixation; ) place multiple transverse cuts through the liver and lungs; ) dissect the posterior leaves of the diaphragm away from the adrenal glands and kidneys; ) bivalve the adrenal glands and kidneys in the coronal plane; ) instill formalin in the lumen of the intestine, using a syringe; and ) gently instill formalin into both ventricles of the heart, being careful to avoid the ventricular septum. the procedure for dissection of the organs is similar to that of the adult except: ) the venous and arterial connections of the heart, including the patency of the ductus arteriosus must be determined before the heart is removed; ) the esophagus must be opened posteriorly prior to its complete removal so that esophageal atresia or tracheo-esophageal fistula may be recognized and photographed prior to further dissection; ) the location of the appendix and of the testes should be recorded. until the intestine has been examined for stenosis or atresia, the mesentery should be left attached. the degree of autolysis as seen grossly and with histologic examination of both the fetus and the placenta can be used to estimate the duration of time between fetal death and delivery ( , , ) . trichrome stain is useful for better visualizing histologic features in severely autolyzed tissue. to remove the brain, benecke's technique of one of its modifications may be used. stillbirth vs livebirth. a decision has to be made whether the infant was born alive or was stillborn. the hydrostatic lung test, described in the previous edition, appears unreliable. the presence of gas in the lungs does not rule out stillbirth. after death, air can be introduced into the lungs, or putrefaction gases might be present. however, air artificially introduced after death will not distend the alveoli and can be squeezed out, whereas this does not seem to be the case after active ventilation. the distribution of fat in the fetal zone of the adrenal cortex may indicate whether intrauterine death was acute, more prolonged, or chronic ( ). this is particularly helpful if the stillborn baby is macerated. if the mother died also, see under "death, abortion-associated" strangulation note: in many instances, procedures described under "hom-icide" must also be followed. if a rope or some other material had been used (see also under "hanging"), leave ligature in place until autopsy can be done. see also under "hypoxia." toxicologic sampling, particularly for alcohol, should be done in all instances (chapter ). results than body fluids. ( ) universal precautions should be strictly followed. the generation of aerosols should be minimized. to sterilize tissue surfaces in preparation for culture, swab the surface with povidone iodine. avoid searing the tissue surface since this will produce an aerosol. keep no more than one scalpel in the dissecting area at anyone time. have an assistant available with clean, gloved hands to receive specimens in containers, so as to minimize the degree of contamination on the outside of the containers. for cleaning procedures and related information, see chapter . ( ) hiv-l in tissue may be demonstrated using polymerase chain reaction (pcr), immunohistochemistry, in situ hybridization, or immunofluorescence. ( ) for immunocytochemical and molecular studies, fix tissue in ethanol or carnoy's solution. the virus can be identified using pcr in most tissues, whether fresh, frozen, or fixed in ethanol. ( ) serologic tests as well as direct fluorescent antibody tests are avail-able for many of the expected infections. ( ) this is not a reportable disease. external examination and skin record body weight and evidence of lipodystrophy following aids medications. record and photograph any skin lesions. examine oral cavity. if the diagnosis is in doubt, samples can be submitted for enzyme immunoassay. cachexia. severe loss of subcutaneous fat in face and extremities, associated with large fat deposits on upper back and upper abdomen ("protease paunch"). cutaneous kaposi's sarcoma (l) . test may be positive for many weeks after death ( ) . syndrome, myelodysplastic synonyms and related terms: chronic myelomonocytic leukemia;* refractory anemia; refractory anemia with excess of blasts; refractory anemia with excess of blasts in transformation; refractory anemia with ringed sideroblasts; refractory dysmyelopoietic anemias. note: the myelodysplastic syndromes are represented by a heterogeneous group of normocytic anemias, often with neutropenia, thrombocytopenia, and monocytosis. for expected bone marrow changes, see above under "synonyms and related terms." autopsy procedures are similar to those recommended for most cases of leukemia, with particular attention paid to intercurrent infections and thrombocytopenic hemorrhages. in all instances, material should be collected using aseptic technique for tissue culture for chromosome analysis. common findings in these conditions are deletion of the long arm ofchromosome , deletion of chromosome or , or trisomy . syndrome, nephrotic note: see under name of suspected underlying condition, such as amyloidosis,* anaphylactoid purpura,*diabetes mellitus,* glomerulonephritis,* goodpasture's syndrome,* heavy metal poisoning, hemolytic uremic syndrome,* infective endo-carditis,* polyarteritis nodosa,* syphilis, * or systemic lupus erythematosus. * if accelerated hypertension or constrictive pericarditis is the suspected underlying condition, see under "hypertension (arterial), all types or type unspecified" or "pericarditis," respectively. in all instances, the renal veins and the inferior vena cava should be opened in situ. "en masse" removal of organs is recommended for this purpose. if thrombosis is found, record exact location and size of clot and submit sample of clot with wall of veins for histologic study. see also under "thrombosis, venous." coronary atherosclerosis and its complications seem to be increased in patients with the nephrotic syndrome. submit sample for bacteriologic, viral, and serologic study. record weight and submit samples for histologic study. if heart block was present, submit samples of conduction system for histologic study (chapter ). submit consolidated areas for microbiologic study. perfuse at least one lung with formalin. for removal of urethra, see chapter . procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. for removal and specimen preparation, see chapter . consult roentgenograms (see above under "external examination, skin, and oral cavity"). keratoderma (keratosis blennorrhagica) with vasculitis ( ) of sole of feet, of palms, and of circumcised glans penis. arthritis* of knees, ankles, and metatarsal and midtarsal joints, with or without ankylosis. osteoporosis.* usually, sterile culture. pericarditis; * myocarditis.* aortic valve lesions. pleuritis and pneumonia.* pulmonary fibrosis involving upper lobes. erosions, papules, and plaques in urethral or bladder mucosa. enteritis. thrombophlebitis. conjunctivitis; multifocal choroiditis; acute anterior uveitis; iritis; keratitis. arthritis* (see above under "external examination, skin, and oral cavity") resembling rheumatoid arthritis. * nonspecific synovitis. for removal and specimen preparation, see chapter . syndrome, reye's note: hypoglycemia* may have been present, but that condition is difficult or impossible to confirm after death. an immediate autopsy is indicated (see below under "liver" and "pancreas"). vitreous blood urine heart lungs submit sample for sodium, chloride, and urea nitrogen determination. submit sample for microbiologic (viral) and serologic study, for ammonia and bilirubin determination, and for determination of salicylate levels if there is a history of treatment with this drug. obtain sample for biochemical and toxicologic analysis. record weight. request frozen sections for sudan stain. submit fresh tissue for bacterial and viral culture. request paraffin sections and frozen sections for fat stain (see above under "heart"). influenza;* varicella.* manifestations of liver failure. evidence of salicylate administration. assay for organic acids should rule out acylcoenzyme a dehydrogenase deficiency, and toxicity, e.g., of acetaminophen and valproic acid (l) . cardiomegaly; fatty changes of myocardium and patchy myocytolysis. viral pneumonia rarely present. acute interstitial pneumonia;* bronchitis;* hemorrhages. lipid-laden histiocytes in alveoli. record body weight and length, stature, and distribution of head, axillary, and pubic hair. record and prepare photographs of features of face and neck. prepare skeletal roentgenograms. submit samples of breast tissues for histologic study. these specimens should be collected using aseptic technique for tissue culture for chromosome analysis (see chapter ) . procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. submit for histologic study. submit samples of all portions for histologic study. if biliary atresia is suspected, follow procedures described under that heading. dissect kidneys and ureters in situ and record findings. submit samples of gonads or-if gonads cannot be identified-equivalent ridges on mesosalpinx for histologic study. also submit samples of endometrium, cervix, and vagina. record weight and submit sample for histologic study. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. for removal and specimen preparation, see chapter . procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. for sampling and specimen preparation, see chapter . peroxidase-labeled gastrin antibodies seem to react with cells in all gastrinomas. aberrant gastrinoma may occur at hilus of spleen. metastases are found in regional lymph nodes and liver. there may be manifestations of multiple endocrine neoplasia. * secondary syphilis. levaditi's stain or warthin-starry stain is recommended for paraffin sections, and labeled fluorescent antibody techniques are recommended for frozen sections. india ink preparations or the fontana-masson silver stain has been used for the study of fresh lesions, and electron microscopy has also been employed. ( ) in adult autopsies, no special precautions are indicated (see below under "liver"). ( ) serologic studies are available from local and state health department laboratories. ( ) this is a reportable disease. external examination record and prepare photographs of all abnormalities listed in right-hand column. prepare smears and sections of acute lesions; prepare sections of older skin lesions or anogenital mucosal lesions. hunterian chancre in primary syphilis; condylomata lata. noduloulcerative gummas and scarring in later stages. cerebrospinal fluid lymph nodes heart and aorta other organs and tissues prior to wk gestation, the destructive effects of syphilis may not be seen. gummata are rare in neonates. tabes dorsalis is also uncommon. serologic diagnosis is difficult in the neonate because of transplacental transfer of maternal igg antibodies. acquired syphilis (syphilis in adulthood) is presented above under a separate heading. ( ) collect all tissues that appear to be infected. ( ) culture methods are not available, but animal inoculation can be performed. consultation with a microbiology laboratory is recommended. ( ) special stains for treponema pallidum rarely are positive except with material from fresh lesions of primary or secondary syphilis and of syphilitic hepatitis of the newborn. levaditi's stain or warthin-starry stain is recommended for paraffin sections, and labeled fluorescent antibody tech-niques are recommended for frozen sections. india ink preparations or the fontana-masson silver stain has been used for the study of fresh lesions, and electron microscopy has also been employed. ( ) in neonates, special precautions are indicated (see below under "liver") ( ) serologic studies are available from local and state health department laboratories. ( ) this is a report· able disease. (i) collect all tissues that appear infected. ( ) request aerobic and anaerobic cultures. however, the presence of tetanus bacilli established in culture is not diagnostic, since spores of c. tetani frequently contaminate wounds. record body weight and length and appearance of wound(s); photograph and excise wound(s) for histologic study. record evidence of parenteral drug abuse, especially subcutaneous injection (i.e., "skin popping"). prepare chest roentgenogram. evidence of weight loss; subcutaneous abscesses. tetanus may occur in drug addicts. tension pneumothorax* after mechanical ventilation. submit sample for microbiologic study. submit any consolidated area for microbiologic study. bacterial meningitis must be ruled out. aspiration and bronchopneumonia; embolism;* atelectasis. tetany note: see under name ofsuspected underlying conditions, such as hyperparathyroidism,* malabsorption syndrome,* or vitamin deficiency. * respiratory or metabolic alkalosis* cannot be confirmed after death; determination of blood ph is not helpful because acidity increases rapidly after death. the concentration of serum phosphates also increases after death. synonym: large ventricular septal defect with pulmonary stenosis* or atresia.* note: the basic anomaly consists of subpulmonary stenosis, ventricular septal defect, overriding aorta, and secondary right ventricular hypertrophy. for general dissection techniques, see chapter . surgical interventions include modified blalock-taus-sig subclavian-to-pulmonary arterial shunt; complete repair with patch closure of ventricular septal defect, and reconstruction of right ventricular outflow tract (with a patch or with an extracardial conduit). possible associated conditions: origin of left pulmonary artery from aorta; minor abnormalities of the tricuspid valve; absent ductal artery ( %); atrial septal defect* (in %; pentalogy of fallot); bicuspid pulmonary valve;* dextroposition of aorta; double aortic arch; hypoplastic pulmonary arteries; patent ductal artery;* patent oval foramen; complete atrioven-tricular septal defect* (usually with down's syndrome*); persis-tent left superior vena cava; pulmonary valve atresia (see "atresia, pulmonary valve, with ventricular septal defect"); right aortic arch ( %); second ventricular septal defect; origin ofleft anterior descending (lad) or right coronary artery (rca) from contralateral aortic sinus or coronary artery ( %); syndrome with absent pulmonary valve and massively dilated pulmonary arteries (rare). chest cavity heart lungs other organs record course of superior vena cava and of its tributaries. record course of thoracic aorta and of its main branches. record origin of pulmonary arteries. if infective endocarditis is suspected, follow procedures described under that heading (chapter ). for coronary arteriography, see chapter . perfuse both lungs with formalin. request verhoeff-van gieson stain. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. persistent left superior vena cava. right aortic arch with or without right-sided ductal artery; double aortic arch; absent ductal artery. origin of the left pulmonary artery from descending aorta. for possible additional findings, see above under "note" and under "possible associated conditions." infective endocarditis* (usually of pulmonary or aortic valve). see also above under "possible associated conditions." marked right ventricular hypertrophy. right ventricular dilatation and fibrosis with late postoperative right-sided heart failure or sudden death due to arrhythmia. old in situ thrombosis of small pulmonary artery branches. paradoxic embolism; cerebral abscess. * thalassemia synonyms and related terms: congenital hemolytic anemia; alpha-thalassemia; beta-thalassemia major (cooley's anemia); beta thalassemia minor (beta-thalassemia trait). note: the changes described below are observed primarily in beta-thalassemia major. unless the cause of the renal vein thrombosis and the nature of the complicating or underlying renal disease are known, follow procedures described under "glomerulonephritis." procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. in infants, manifestations of dehydration. * pulmonary embolism. * tumor of the kidney* (renal cell carcinoma); aortic aneurysm; lymphadenopathy; other mass lesions. inferior vena cava thrombosis involving orifice or whole length of renal veins. tumor thrombus (e.g., from renal cell carcinoma). femoral vein thrombosis. rare venous malformations also may cause renal vein thrombosis (l) . hemorrhagic infartion. parenchymal renal disease with nephrotic syndrome,* including amyloidosis* and vasculitis* (these last conditions may be associated with renal vein thrombosis). acute gastroenteritis in infancy. hyperparathyroidism,* pregnancy,* trauma, and other conditions. thrombosis, venous note: for peripheral venous thrombosis, the autopsy procedures are essentially similar to those described under "phlebitis." see also under "hypertension, portal," "syndrome, budd-chiari," "thrombosis, cerebral venous sinus," "thrombosis, lateral sinus," and "thrombosis, renal vein." thymoma (see "'fumor of the thymus.") thyroiditis synonymsand related terms: chronic fibrosing thyroiditis (riedel's struma); chronic thyroiditis with transient thyrotoxicosis; hashimoto's thyroiditis; pyogenic thyroiditis; subacute (granulomatous, giant cell, de quervain's) thyroiditis. possibleassociatedconditions: withhashimoto's thyroiditisautoimmune (chronic) hepatitis,*megaloblastic anemia,*rheumatoid arthritis,*sjogren's syndrome,*and systemic lupus erythematosus.* with riedel's struma-retroperitoneal fibrosis,* sclerosing cholangitis,* sclerosing mediastinitis,*and other conditions with idiopathic fibrosis. with subacute thyroiditis-viral infection. submit sample for determination of protein concentration; record appearance of sediment. for removal and specimen preparation, see chapter . thromboses of small vessels. glomerulonephritis.* hemorrhagic necroses. proteinuria; abnormal sediment. abruptio placentae, small placenta with accelerated maturation of villi; infarcts; fibrinoid necrosis of decidual arterioles. thrombotic occlusion of small vessels with hemorrhagic necroses; anterior lobe of pituitary gland may contain necroses (see also "syndrome, sheehan's"). submit samples for histologic study. if infection is expected, submit material for microbiologic study. other procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. transposition, complete, of the great arteries note: the basic anomaly is the origin of the aorta from the right ventricle, and of the pulmonary artery from the left ventricle, with a shunt, and usually with a right anterior aorta. there are four major types: ( ) with an intact ventricular septum ( %), ( ) with a ventricular septal defect* ( %), ( ) with a ventricular septal defect and subvalvular pulmonary stenosis* ( %), and ( ) with an intact ventricular septum and subvalvular pulmonary stenosis* ( %). for general dissection techniques, see chapter . interventions include atrial septostomy or septectomy; mustard or senning atrial switch procedure; rastelli-type repair with a valved extracardiac conduit; and jatene arterial switch procedure with lecompte maneuver. possible associated conditions: abnormal origin ofcoronary arteries ( %); atrial septal defect* ( %); coarctation of the aorta;* interrution of the aortic arch;* juxtaposition of atrial appendages ( %); overriding aorta ( %); overriding pulmonary artery ( %); patent ductal artery;* patent oval foramen; subvalvular pulmonary stenosis* ( %); tubular hypoplasia ofthe aortic arch;* ventricular septal defect* ( %), often mal-alignment type ( %). synonyms: atrioventricular and ventriculoarterial discordance; l-transposition. note: the basic anomaly is a mirror-image ventricular inversion, with a left anterior aorta, and with blood flow from right atrium to left ventricle to pulmonary artery, and from left atrium to right ventricle to aorta. for general dissection techniques, see chapter . interventions include patch closure of the ventricular septal defect; relief of pulmonary stenosis; relief of tricuspid insufficiency; and insertion of pacemakers. possible associated conditions: anterior and posterior av nodes ( %), prone to develop complete heart block; dysplasia or epstein's malformation* ofleft-sided tricuspid valve ( %); mirror-image epicardial coronary artery distribution ( %); right-sided mitral valve anomalies; subvalvular pulmonary stenosis* ( %); ventricular septal defect* ( %). kinyoun's, or other acid-fast stains. polymerase chain reaction for mycobacterial dna may be helpful in the differential diagnosis ofgranulomas ( ). ( ) universal precautions should be strictly followed and aerolization should be avoided. ( ) reliable serologic studies are not available. a definite diagnosis requires isolation of the organism. ( ) this is a reportable disease. cavitary, fibrocalcific, miliary, bronchial, and other types of pulmonary tuberculosis; granulomas in hilar lymph nodes. most organs and tissues may be involved, including liver, pancreas, spleen, kidneys, adrenal glands and other endocrine glands, gonads, and lymph nodes. tuberculous meningitis; tuberculoma. iridocyclitis or panophthalmitis. tuberculous arthritis and synovitis (hips, spine, knee); tuberculous osteomyelitis (anterior aspect of vertebrae; metaphysis of long bones). part ii / diseases and conditions procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. chronic ulcerative colitis may be associated with carcinoma of bile ducts, typically arising in psc.* metastases common in regional lymph nodes and lungs. i for removal, prosthetic repair, and specimen preparation and decalcification procedures, see chapter . if osteoblastic metastases appear to be present, search for primary tumor in prostate and breast, carcinoid tumors and hodgkin's lymphoma. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. cachexia. evidence of hypercalcemia (particularly in presence of osteolytic metastases). metastases to bone (e.g., from carcinoma of prostate, breast, bronchi, thyroid gland, kidney, or urinary bladder) usually involve red bone marrow. therefore, distal extremities are rarely involved by metastatic tumors in adults. metastases, commonly in lungs. metastatic calcification in patients with hypercalcemia. eosinophilia. myopathy.* metastases in liver and regional lymph nodes. see also above under "possible associated conditions." malakoplakia (rare) ( ) . part ii / diseases and conditions other organs and peripheral arteries tumor(s). if tumor is within a heart chamber (usually a myxoma), record extent of mobility and capability of tumor to obstruct a valvular orifice. submit samples of tumor(s) for histologic study and, particularly if the tumor is of unknown type, for immunohistochemical study and electron microscopy (chapter ). procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. metastatic carcinoma or melanoma. secondary cardiac effects by tumors include carcinoid heart disease, amyloid heart disease in multiple myeloma, and lymphocytic myocarditis in pheochromocytoma. treatment effects such as adriamycin cardiotoxicity may be noted also. tumor (myxoma) emboli in pulmonary or peripheral vessels ( ) . multiple aneurysms* of cerebral and other arteries may rarely be associated with atrial myxomas. procedures depend on expected findings or grossly identified abnormalities as listed above and in right-hand column. part ii / diseases and conditions if there was evidence of endocrine activity (see above), snap-freeze portion of fresh tumor for hormone assay. perfuse lungs with formalin. sample neoplastic and non-neoplastic tissue for histologic study. procedures depend on expected findings or grossly identified abnormalities as listed above and in right-hand column. for removal and specimen preparation, see chapter . for removal and specimen preparation, see chapter . for removal and specimen preparation, see chapter . for removal and specimen preparation, see chapter . for removal and specimen preparation, see chapter . see above under "possible associated conditions." note that hypoglycemia* generally cannot be confirmed at autopsy. nonbacterial thrombotic endocarditis.* carcinoma may be associated with asbestosis or other types of pneumoconiosis,* chronic bronchitis,* emphysema,* interstitial pneumonia,* and many other bronchopulmonary diseases. see above under "possible associated conditions." metastases (regional lymph nodes, liver, bones, brain, and many other sites) and metastatic calcification. migratory thrombophlebitis.* encephalomyelitis;* cerebellar corticoid degeneration;* subacute spinocerebellar degeneration. * crooke cell hyperplasia. myasthenic syndrome (eaton-lambert syndrome); myopathy;* dermatomyositis.* peripheral neuropathy. see above under "external examination and skin." bones (with red marrow) are common sites of metastases. record presence or absence of testes. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. thmor of the soft tissues the possible sites and characteristics of soft tissue tumors vary so much that no universally applicable autopsy techniques can be presented. in all instances, the size, weight, and location of the tumor(s) must be recorded and tissue must be sampled for histologic study. if the tumor had not been classified prior to death, samples should be snap-frozen for immunohistochemical study. other samples should be prepared for electron microscopic study. evidence of paraneoplastic syndromes (see below) may require additional procedures. possible associated conditions: only a few paraneoplastic syndromes or systemic complications can be presented here. for the type of soft tissue tumor that was associated with each condition, see title of reference. kasabach-merritt syndrome ( ) (thrombocytopenia, microangiopathic hemolytic anemia, and acute or chronic coagulopathy associated with a rapidly enlarging hemangioma); liver function abnormalities ( ); neurofibromatosis ( ); osteomalacia ( ). heart esophagus, stomach, and duodenum other organs and tissues t in most instances, detennination of vitreous sugar concentration and of blood group is not indicated. inspect valves and prepare photographs and sections of vegetations. leave esophagus and part of duodenum attached to stomach. submit samples of tumor and of grossly uninvolved stomach for histologic study. request pas stain and warthin-starry stain for identification of h. pylori. portions of endocrine gastric tumor should be snap-frozen for immunohistochemical study and possible hormone assay. record location of tumor metastases. other organs procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. for dissection of the thoracic duct (for search of tumor cell clusters), see chapter . tumor of the uterus (with cervix) possible associated conditions: acquired immunodeficiency syndrome* (aids) ( ) if peritonitis is present, submit exudate for bacteriologic study. record volume of exudate and location of perforation. see "disease, ischemic heart." other procedures depend on grossly identified abnormalities as listed in right-hand column. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. perfuse at least one lung with formalin. for celiac arteriography, see chapter . open stomach and duodenum in situ and record site of perforation or penetration. record measured or estimated volume of blood in gastrointestinal tract. rinse ulcer with saline to locate eroded vessel(s). pin stomach and duodenum on corckboard (serosa toward board) and fix specimen in formalin before sectioning. prepare histologic sections of ulcer(s) and of remainder of stomach. request warthin-starry stain for h. pylori. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. free air in abdomen suggests perforation of ulcer. peritonitis. * perforation of ulcer. coronary atherosclerosis and manifestations of coronary insufficiency are commonly associated with peptic ulcer disease. in rare instances, pericardial fistula may be present from ulcer in hiatus hernia. peptic ulcers may be associated with emphysema,* tuberculosis,* and other chronic pulmonary diseases. vasculitis (see "aortitis," "arteritis, .•.," "phlebitis," and "purpura,•••") ventricle, double inlet left ( ) synonyms: single functional ventricle; univentricular heart; univentricular atrioventricular connection; holmes heart. note: the basic anomaly is the connection of both atrioventricular valves to the left ventricle, often with transposed great arteries and a restrictive ventricular septal defect. there are four major types, based on the ventriculoarterial connection: ( ) with congenitally corrected transposition ( %); ( ) with complete transposition ( %); ( ) with normally related great arteries (holmes heart; %); and ( ) double outlet, persistent truncal artery, or pulmonary atresia ( %). for general dissection techniques, see chapter . possible associated conditions: bicuspid pulmonary valve; bilateral mirror-image mitral valves (without tricuspid morphology); subvalvular aortic stenosis,* often with hypoplasia, coarctation, or interruption of the aortic arch; subvalvular pulmonary stenosis;* dual av nodes and progressive heart block in patients with congenitally corrected transposition. ventricle, double outlet right synonym: origin of both great arteries from right ventricle; taussig-bing anomaly. note: the basic anomaly is the origin of the aorta and pulmonary artery primarily from the right ventricle, usually with a ventricular septal defect, and often with subpulmonary stenosis. for general dissection techniques, see chapter . possible associated conditions: complete atrioventricular septal defect (often with asplenia syndrome); muscular discontinuity between aortic and mitral valves; right ventricular infundibular stenosis; ventricular septal defect* that may be subaortic, subpulmonary, doubly committed, or remote. examine histologically with h&e and oil red a stains. hepatomegaly and splenomegaly, jaundice. deficiency of acid lipase ( ). foam cells in the lamina propria. hepatomegaly, severe steatosis with cholesterol clefts in hepatocytes and kupffer cells. fibrosis. splenomegaly with lipid-laden foam cells. symmetric enlargement with dystrophic calcifications, giving the adrenals a gritty texture. vacuolated cells of the inner zona fasciculata and entire zona reticularis with cholesterol clefts. foam cells may be found in the lungs, bone marrow, lymph nodes, vessel walls, as well as schwann and ganglion cells. new insights into lipoprotein assembly and vitamin e metabolism from a rare genetic disease angioid streaks associated with abetalipoproteinemia familial hypobetalipoproteinemia: genetics and metabolism aviation pathology scuba diving accidents: decompression sickness, air embolism aetiology and occurrence of diving injuries. a review of diving safety scuba decompression illness and diving fatalities in an overseas military community diving-related emergencies bodies found in water achalasia and squamous cell carcinoma of the esophagus: analysis of patients esophageal achalasia and adenocarcinoma in barrett's esophagus: a report of two cases and a review of the literature a technique for necropsy evaluation of stenosis of the foramen magnum and rostral spinal canal in osteochondrodysplasia mutation analysis of the men i gene in multiple endocrine neoplasia type i, familial acromegaly and familial isolated hyperparathyroidism. i increased incidence of neoplasia in females with acromegaly benign and malignant tumors in patients with acromegaly pathology of acromegaly ethanol in sequestered hematomas stages of acute alcoholic influence/intoxication electrolyte imbalance in alcoholic liver disease collection and storage of specimens for alcohol analysis analysis for alcohol in postmortem specimens disposition of alcohol in man medicolegal aspects of alcohol. garriott jc blood, urine and other tissue specimens for alcohol analysis relationship between postmortem blood and vitreous humor ethanol levels larson cpo alcohol: fact and fallacy gradwohl's legal medicine left ventricular hypertrophy is more prominent in patients with primary aldosteronism than in patients with other types of secondary hypertension intracranial aneurysm and hemorrhagic stroke in glucocorticoid-remediable aldosteronism association of adrenal medullae and cortical nodular hyperplasia: a report of two cases with clinical and morpho-functional consideration amyloidosis: recognition, confirmation, prognosis, and therapy electron microscopy of primary and secondary cutaneous amyloidosis and systemic amyloidosis ocular amyloidosis current status review: cerebral amyloid spinal cord compression by amyloid deposits the burden of "sticky" amyloid: typing challenges quantification of cerebral amyloid angiopathy and parenchymal amyloid plaques with congo red histochemical stain kidney involvement in takayasu arteritis pathogenesis of rheumatoid arthritis splenic involvement in rheumatic diseases aspergillus sinusitis in patients with aids: report of three cases and review invasive aspergillosis in systemic lupus erythematosus aspiration (see "obstruction, acute airway aspergillus endocarditis, myocarditis and pericarditis complicating necrotizing fasciitis. case report and subject review aspergillus-related lung disease biliary atresia and the polysplenia syndrome: its impact on final outcome biliary atresia nephromegaly and elevated hepatocyte growth factor in children with biliary atresia coccidiomycosis pneumonia in a nonendemic area associated with inftiximab codeine (see "dependence, drug[s], all types or type unspecified all types or type unspecified (see "enterocolitis, other types or type undetermined chronic ulcerative (see "disease, inflammatory bowei cytologic diagnosis of cryp -tococcus neofonnans in hiv-positive patients cryptococcosis as an opportunistic infection in immunodeficiency secondary to paracoccidioidomycosis hypereosinophilia in disseminated cryptococcal disease cryptococcosis-a review of autopsy cases from a -year period in a large hospital cryptosporidiosis in persons with hiv infection broncho-pulmonary cryptosporidiosis in four hiv-infected patients cryptosporidiosis and inflammatory bowel disease sclerosing cholangitis associated with chronic cryptosporidiosis in a child with a congenital immunodeficiency disorder cyanide (see "poisoning, cyanide cyst(s), choledochal choledochal cyst-dinical features and classification cyst(s), liver (see "disease, fibropolycystic, of the liver and biliary tract pulmonary related terms: congenital cystic adenomatoid malformation; congenital pulmonary lymphangiectasis; intralobular bronchopulmonary sequestration. references infection or calcification of cysts; pyelonephritis;* perinephric abscess. obstructive uropathy in recessive polycycstic renal disease, congenital hepatic fibrosis is present and cystic bile ducts may be present in dominant cases ( ). see above under "possible associated conditions the pathology ofhuman renal cystic disease cystic kidney: genetics, pathologic anatomy, clinical picture, and prenatal diagnosis hepatic fibrosis associated with hereditary cystinosis: a novel form of noncirrhotic portal hypertension ophthalmic manifestations and histopathology of infantile nephropathic cyctinosis: report of a case and review of the literature pathological investigations of deaths following surgery, anaesthesia and medical procedures clinical outcome versus post-mortem finding in thoracic surgery: a lo-year experience order from american registry of pathology sales office intra-alveolar pulmonary siderophages in sudden infant death: a marker for previous imposed suffocation a contribution to a possible differentiation between sids and asphyxiation sudden infantdeath syndrome (sids)-standardized investigations and classification: recommendations investigation of the sudden deth of infants: a multicenter analysis alpha l-antitrypsin deficiency-associated panniculitis: resolution with intravenous alpha i-antitrypsin administration and liver transplantation clinical manifestations of alpha l-antitrypsin deficiency risk of hepatobiliary disease in adults with severe alpha i-antitrypsin deficiency (pizz): is chronic viral hepatitis b or c an additional risk factor for cirrhosis and hepatocellular carcinoma? alpha i-antitrypsin deficiency and inflammatory bowel disease severe alphal-antitrypsin deficiency (piz homozygosity) with membranoproliferative glomerulonephritis and nephrotic syndrome, reversi-bleafterorthotopic livertransplantation klippel-feil syndrome associated with malfonned larynx. case report klippel-feil syndrome in association with a craniocervical dennoid cyst presenting as aseptic meningitis in an adult: case report klippel-feil syndrome accompanied by an aneurysm of the non-coronary sinus of valsalva the hereditary ataxias clinical and genetic characterizations of q-linked autosomal dominant spinocerebellar ataxia (ad-sca) and frequency analysis of ad-sca in the japanese population cocain-related medical problems. consecutive series of cases direct cocaine cardiotoxicity demonstrated by endomyocardial biopsy the pathology and etiology of cocaineinduced heart disease ischemic colitis related to cocaine abuse hepatic dysfunction accompanying acute cocaine intoxication high incidence of malignancies in patients with dennatomyositis and polymyositis: an li-year analysis partial lipodystrophy associated with juvenile dennatomyositis: report of two cases chronic pneumomediastinum and subcutaneous emphysema: association with dennatomyositis spontaneous esophageal rupture in adult dennatomyositis polyneuropathy in juvenile dermatomyositis combined glucose and lactate values in vitreous humor for postmortem diagnosis of diabetes mellitus diabetes is related to cerebral infarction but not to ad pathology in older persons hepatic lesions resembling alcoholic liver disease relationship between cardiomyopathy and liver disease in chronic alcoholism cyanamide-associated alcoholic liver disease: a sequential histologic evaluation caroli's disease: - experiences caroli 's disease and autosomal dominant polycystic kidney disease: a rare asso-ciation? spontaneous rupture of a bile duct and its endoscopic management in a patient with caroli's syndrome bartonella henselae endocarditis in an immunocompetent adult cat-scratch disease and bacillary angiomatosis attack, transient cerebral ischemic" and "infarction, cerebral!') cholesteryl ester storage disease: hepatopathology and effects of therapy with lovestatin clinical, biochemical and histological analysis of seven patients with cholesterol ester storage disease cutaneous manifestations of chronic granulomatous disease. a report of four cases and review of the literature aspergillus endocarditis in chronic granulomatous disease colitis complicating chronic granulomatous disease. a clinicopathological case report hypertrophic cranial pachymeningitis with spinal epidural granulomatous lesion ocular pathologic findings of chronic granulomatous disease of childhood review: creutzfeldt-jakob disease survival of creutzfeldt-jakob disease virus in formol-fixed brain tissue guidelines for high risk autopsy cases: special precautions for creutzfeldt-jakob disease tissue handling in suspected creutzfeldt-jakob disease and other human spongiforme encephalopathies (prion diseases) cerebrospinal fluid concentration of neuron-specific enolase in diagnosis of creutzfeldt-jakob disease autopsy case of sporadic creutzfeldt-jakob disease presenting with signs suggestive of brainstem and spinal cord involvement metastatic crohn's disease: a rare cutaneous manifestation crohn's disease with pulmonary involvement in a year old boy mesenteric fibromatosis asso-ciated with crohn's disease cholangiocarcinoma and crohn's disease the pancreas as a site of granulomatous inflammation in crohn's disease dermatomyositis associated with crohn's disease cardiocyte storage and hypertrophy as a sole manifestation of fabry's disease. report on a case simulating hypertrophic non-obstructive cardiomyopathy an atypical variant of fabry's disease with manifestations confined to the myocardium pulmonary involvement in fabry disease cerebrovascular complications offabry's disease high incidence of thrombosis in fabry's disease oral involvement in chronic graft versus host disease after allogenic bone marrow transplantation massive pericardial effusion complicating the course of chronic graft-versus-host disease (cgvhd) in a child with acute lymphoblastic leukemia following allogeneic bone marrow transplantation the histological spectrum of pulmonary graft-versushost disease in bone marrow transplant recipients bronchiectasis in bone marow transplantation oncocytic metaplasia of bile duct epithelium in hepatic gvhd immune-mediated myelopathy after allogeneic marrow transplantation gunther's (see ''porphyria, congenital erythropoietic heavy-chain synonyms and related terms: gamma heavy-chain inflammatory bowel disease (first of two parts) takayasu's arteritis associated with idiopathic ulcerative colitis cerebrovascular complications ofinflammatory bowel disease pathology of inflammatory bowel disease legionnaires' disease source of infection for sporadic legionnaires' disease: a review legionnaire's disease associated with macular rash; two cases severe skin loss after meningococcal septicemia: complications in treatment osteonecrosis following meningococcemia and disseminated intravascular coagulation in an adult: case report and review rhabdomyolysis during the subacute stage of meningococcal sepsis primary meningococcal arthritis in a child: case report and literature review chondrosarcoma as a complicating factor in paget's disease of bone lymphoma arising in paget's disease osteoarthritis and paget's disease sickle cell disease sickle cell disease and sudden death: a retrospective/prospective study of autopsy cases and literature review vascular lesions of the liver in sickle cell disease. a clinicopathologic study in living patients whipple's disease and "tropheryma whippelii periodic acid-schiff-negative granulomatous lymphadenopathy in patient with whipple's disease. localization of the whipple bacillus to noncaseating granulomas by electron microscopy serial imaging studies of cerebral whipple's disease: from onset to end the wilson's disease gene is a putative copper transporting p-type atpase similar to menke's gene the liver biopsy diagnosis of wilson's disease. methods in pathology hepatolenticular degeneration (wilson's disease) venous air embolism in homicidal blunt impact head trauma: case reports a practical device for demonstrating air embolism amniotic fluid . kulka w. laboratory methods and technical notes: a practical device for demonstrating air embolism proof of fatal air embolism venous air embolism from head and neck wounds hantavirus pulmonary syndrome: a clinical description of patients with a newly recognized disease liver involvement in hemorrhagic fever with renal syndrome hantavirus infection induces a typical myocarditis that may be responsible for myocardial depression and shock in hanta virus pulmonary syndrome lassa fever in the united states. investigation ofa case and new guidelines for management lassa fever: review of virology, immunopathogenesis, and algorithms for control and therapy early diagnosis of lassa fever by reverse transcription-pcr rheumatic pneumonia: reappearance of a previously recognized complication of rheumatic fever acute rheumatic fever and poststreptococcal acute glomerulonephritis caused by t serotype streptococcus scleritis, uveitis, and glaucoma in a patient with rheumatic fever comparison of clinical features and pathologic findings in fatal cases of typhoid fever during the initial and later stages of the disease typhoid fever complicated by acute renal failure and hepatitis: case reports and review hemophagocytosis and granulomas in the bone marrow of a child with down syndrome diseases involving intrahepatic bile ducts idiopathic eosinophilic esophagitis with stricture formation in a patient with longstanding eosinophilic gastroenteritis eosinophilic gastroenteritis presenting with colitis and cholangitis eosinophilic gastroenteritis presenting with acute pancreatitis idiopathic midline destructive disease: does it . mendenhallwmetai.lethalmidlinegranoloma-nasalnaturalkillerff-celi exist? nasal cocain abuse mimicking midline granuloma pulmonary lymphomatoid granulomatosis in acquired immunodeficiency syndrome: lesions with epstein-barr virus infection recurrent bilateral spontaneous pneumothoraces associated with pulmonary angiocentric immunoproliferative lesion lymphomatoid granulomatosis: pathogenesis, pathology and clinical implications wegener's granulomatosis: histological documentation of common and uncommon manifestations in patients necrotizing granulomatosis of the breast wegener's granulomatosis with gangrene of toes pathology of pulmonary granulomatous vasculitis. sarcoidosis gunshot (see "injury, firearm splenic involvement in rheumatic diseases wegener's granulomatosis, acute laryngotracheal airway obstruction and death in a -year-old female: case report and review of the literature antiendothelial antibodies in patients with wegener's granulomatosis: prevalence and correlation with disease activity and manifestations. i rheumatoi transfusion-transmitted disease references flattening, broadening, and possible fusion of villi. phlegmonous inflammation and edema, mainly in ileocecal region encephalitis.* leukopenia; thrombocytopenia; aplastic anemia ( ) (pancytopenia*) synovitis (arthritis*) hepatitis g virus infection in hepatitis c virus-positive patients co-infected or not with hepatitis b virus and/or human immunodeficiency virus hepatitis g and c coinfection in children tumor of the liver diaphragmatic related terms: congenital diaphragmatic hernia fulminant hepatitis in patients undergoing liver transplantation: evidence for anon-a, non-b, non-c marrow transplantation for hepatitis-associated aplastic anemia: a follow up oflong-term survivors langerhans cell histiocytosis research. past, present, and future an update on c onality, cytokines, and viral etiology in langerhans cell histiocytosis langerhans cell histiocytosis in adults pulmonary langerhans cell granulomatosis central nervous system disease in langerhans cell histiocytosis immunohistochemical analysis oflangerin in langerhans cell histocytosis and pulmonary inflammatory and infectious diseases pathological approach to the diagnosis of hydrocephalus thoracic lipomeningocele associated with diastematomyelia, tethered spinal cord, and hydrocephalus. case report infectious causes ofhydrops fetalis human parvovirus b infection associated with hydrops fetalis cardiac abnormalities associated with hydrops fetal is gmi-gangliosidosis presenting as nonimmune hydrops fetalis: a case report a case of recurrent infantile polycystic kidney associated with hydrops fetalis the pathologist and the hydropic placenta, fetus, or infant chronic diarrhea associated with thymoma and hypogammaglobulinemia (good's syndrome) systemic amyloidosis in a patient with hypogammaglobulinemia. i hcv infection in patients with primary defects of immunoglobulin production encepahomyelitis in primary hypogammaglobulinemia retinitis pigmentosaand hypogammaglobulinemia intestinal obstruction in the newborn with congenital syphilis meconium ileus and its equivalent as a risk factor for the development of cirrhosis: an autopsy study in cystic fibrosis biliary atresia and meconium ileus associated with nieman-pick disease postmortem cranial mri and autopsy correlation in suspected child abuse cytomegalovirus-associated pseudotumor simulating gastric malignancy in acquired immuno-deficiency syndrome: a case report with review of literature bone marrow fibrin ring granulomas and cytomegalovirus infection cutaneous reactions following herpes zoster infections: report of three cases and a review of the literature herpes zoster and internal malignancy posteriorhorn varicella-zoster virus myelitis death or injury caused by electrocution myocardial hemorrhagic necrosis in delayed death from electrocution safety in bullet recovery procedures: a study of the black talon bullet lightning injuries: prognostic signs of death lightning injuries adrenal insufficiency, recurrent bacteremia, and disseminated abscesses caused by nocardia asteroides in a patient with acquired immunodficiency syndrome ards and adrenal insufficiencyassociated with the antiphospholipid antibody syndrome adrenal insufficiency from bilateral adrenal hemorrhage after total knee replacement surgery non-hodgkin's lymphoma presenting with adrenal insufficiency and hypothyroidism: an autopsy case report a case ofprimary unilateral adrenal burkitt-like large cell lymphoma presenting as adrenal insufficiency effect of growth hormone on fatty liver in panhypopituitarism increased fracture frequency in adult patients with hypopituitarism and gh deficiency pituitary abscess the liver: an atlas and text of ultrastructural pathology intubation (see "injury, intubation iodine (see "poisoning, iodine attack, transient cerebral ischemic" and "infarction, cerebral heart (see "disease, ischemic heart isomorism (see "syndrome, polysplenia and asplenia a study on performance of two serological assays for diagnosis of leprosy detection of mycobacterium leprae infection by pcr pathology of eye in leprosy references intervillous abscesses containing grampositive, non-acid-fast bacilli bacteria are predominantly intracellular. enterocolitis. listeria monocytogenes can be cultured from meconium. generalized lymphadenitis. disseminated abscesses and/or granulomas, particularly after transplacental infection listeria monocytogenes empyema in an hiv infected patient fatal listeria meningitis, endocarditis and pericarditis in a patient with haemochromatosis liver abscesses due to listeria monocytogenes yust . brainstem abscess and meningitis due to listeria monocytogenes in an adult with juvenile chronic arthritis pulmonary hypertension associated with systemic lupus erythematosus: predominantly thrombotic arteriopathy accompanied by plexiform lesions the liver in systemic lupus erythematosus: pathologic analysis of cases and review of japanese autopsy registry data chronic pancreatitis: a complication of systemic lupus erythematosus disseminated perivenous necrotizing encephalomyelitis in systemic lupus erythematosus: report of an autopsy case skeletal muscle lymphocytic vasculitis in systemic lupus erythematosus: relation to disease activity clinically occult avascular necrosis of the hip in systemic lupus erythematosus storage histiocytes and hemophagocytosis: a common finding in the bone marrow of patients with active systemic lupus erythematosus extensive medium vessel vasculitis with sle: an unsual association cutaneous manifestations of sexually transmitted ing cause of proctitis in men who have sex with men lymphogranuloma venereum as a cause cutaneous malakoplakia in a patient with acquired immunodeficiency syndrome (aids) extensive malakoplakia ofthe nasopharynx: management of a rare disease malakoplakia and colorectal adenocarcinoma coinfection is common in measles associated pneumonia giant cell pneumonia: light microscopy, immunohistochemical, and ultrastructural study of an autopsy case subacute sclerosing panencephalitis manifesting as viral retinitis: clinical and histopathologic findings intestinal neuronal dysplasia thoracic lipomeningocele associated with diastematomyelia, tethered spinal cord, and hydrocephalus. case report disease, meningococcal encephalitis, all types or type unspecified" and "meningitis mercury (see "poisoning, mercury with myelofibrosis synonym: idiopathic myelofibrosis acute cardiac tamponade associated with pericardial extramedullary hematopoiesis in agnogenic myeloid metaplasia methyl alcohol) (see "poisoning, methanol (methyl alcohol) thrombotic thrombocytopenic (see "purpura, thrombotic thrombocytopenic epstein-barr virus in inflammatory diseases of the liver and liver allografts: an in situ hybridization study fulminant hepatitis due to epstein-barr virus infection the mucopolysaccharidoses: a clinical review and guide to management biochemical diagnosis of mucopolysaccharidoses: experience of diagnoses in a -year period ( - ) mucormycosis synonym: phycomycosis; zygomycosis. note: diseases that may be complicated by mucormycosis include bums,* diabetes mellitus,* leukemia,* lymphoma,* and tuberculosis cardiac involvement in mucopolysaccharidosis: effects of allogeneic bone marrow transplantation mitral and aortic regurgitation in patients with mucopolysaccharidosis toxic epidermal necrolysis possibly linked to hyperacute graft-versus-host disease after allogeneic bone marrow transplantation pulmonary complications in toxic epidermal necro-iysis: a prospective clinical study references possible or expected findings pyelonephritis;* nephrocalcinosis; granulomas; tumor infiltrates; manifestations of the conditions listed below under "other organs renal stones in patients with rheumatoid arthritis nephrolithiasis as a presenting feature ofchronic sarcoidosis: a prospective study nephrolithiasis and risk of hypertension posterior lens opacities; retinal hamartomas. peripheral neuropathy with focal schwannomatous changes or onion-bulb-like schwann cell or perineural cell proliferation neurofibromatosis type i. in: pathology and genetics of tumours of the nervous system neurofibromatosis type . in: pathology and genetics oftumours ofthe nervous system neurological complications involving the central nervous system in neurofibromatosis type i primary cutaneous nocardia asteroides infection after heart transplantation native valve endocarditis due to nocardia-like organisms hepatobiliary complications of total parenteral nutrition total parenteral nutrition: a histopathologic analysis of the liver changes in children obesity and breast cancer risk nonalcoholic steatohepatitis obesity cardiomyopathy: pathogenesis and pathophysiology bennett cpo pachydennoperiostitis in childhood unilateral hypertrophic osteoarthropathy associated with aortofemoral graft infection gastroesophageal reflux and esophagitis-associated hypertrophic osteoarthropathy hypertrophicosteoarthropathy caused by lipoid pneumonia achondroplasia" and dyschondroplasia, oilier's osteogenesis imperfecta synonyms and related terms: lobstein's syndrome osteogenesis imperfecta congenita; type i, ii, and iii; osteogenesis imperfecta cystica hypercalcemia in osteogenesis imperfecta treated with pamidronate gastrointestinal problems in patients who have type-iii osteogenesis imperfecta osteomalacia related terms: osteoporosis;* renal osteodystrophy; rickets. note: many possible causes of osteomalacia may not be apparent at autopsy, e.g., sodium fluoride or diphosphonate toxicity or use of anticonvulsant drugs osteomalacia associated with adult fanconi syndrome: clinical and diagnostic features epstein-barr virus in b-celllymphomas associated with chronic suppurative inflammation osteomyelitis and osteonecrosis in inflammatory bowel disease references hyperlipidemia; hyperuricemia. fracture or traumatic dislocation of hip see above under "note osteomyelitis and osteonecrosis in inflammatory bowel disease osteonecrosis and human immunodeficiency virus infection report of fifteen cases. therapeutic recommendations malignant infantile osteopetrosis: otolaryngological complications and management renal tubular acidosis and osteopetrosis with carbonic anhydrase ii deficiency: pathogenesis of impaired acidification spondylolysis in children who have osteopetrosis osteopetrosis with arnold chiari malformation type and brain stem compression osteoporosis and atherosclerosis in chronic renal failure endocrine disorders and osteoporosis otosclerosis: a measles virus associated inflammatory disease correlations between pathologic changes in the stapes and conductive hearing loss in otosclerosis the aetiology of otosclerosis: a review of the literature etiologic links between chronic pancreatitis and pancreatic cancer hyperparathyroidism and pancreatitis during pregnancy pancytopenia related terms: agranulocytosis;* aplastic anemia; fanconi's anemia.* note: some causes of pancytopenia such as adverse drug reactions (e.g., due to antimetabolites, sulfa drugs thrombotic thrombocytopenic purpura/hemolytic uremic syndrome secondary to pancreatitis retinopathy as asys-temic complication of acute pancreatitis references mesenteric panniculitis necrotizing pancreatitis. * vasculitis with thrombi; adrenocortical infarctions. vasculitis with thrombi and infarctions. relapsing polychondritis ( ). see also above under "possible associated conditions subcutaneous panniculitic t-cell lymphoma is a tumor of cytotoxic t lymphocytes abecassism. alpha i-antitrypsindefi-ciencyassociated panniculitis: resolution with intravenous alpha l-anti-trypsin administration and liver transplantation blind loop syndrome: an unusual cause of panniculitis. j am acad paracoccidioidomycosis synonyms: paracoccidioides brasiliensis infection cutaneous panniculitis and relapsing polychondritis: two cases normal subcutaneous fat, necrosis of adipocytes and classification of the panniculitides paraneoplastic pemphigus a case of pemphigus vulgaris with esophageal involvement rheumatoid constrictive pericarditis (clinical conference) benign paroxysmal (see "fever, familial mediterranean pneumatosis intestinalis in pediatric acquired immunodeficiency syndrome pneumatosis cystoides intestinalis: an unusual complication of systemic amyloidosis pneumatosis intestinalis in children with primary combined immunodeficiency pneumatosis cystoides intestinalis with abdominal free air in a -year-old girl after allogeneic bone marrow transplantation pneumatosis intestinalis: a review pneumatosis coli: a proposed pathogenesis based on study of cases and review of the literature collagenous inorganic dust pneumoconiosis, diffuse type: aluminosis; asbestosis; chronic pulmonary berylliosis; talcosis; collagenous inorganic dust pneumoconiosis, nodular type: silicosis; noncollagenous inorganic dust pneumoconiosis (includes mixed-dust fibrosis): baritosis; china clay pneumoconiosis; chromite pneumoconiosis; coal worker's pneumoconiosis; fuller's earth pneumoconiosis; hematite or magnetite miner's lung; siderosis or welder's lung; stannosis; organic dust pneumoconiosis: byssinosis. note: in addition to the aforementioned classic forms of pneumoconiosis, many other airborne substances have been impli-cated in recent years, for example, cerium uncommon causes ofoccupational interstitial lung diseases rare earth (cerium oxide) pneumoconiosis: analytical scanning electron microscopy and literature review diagnostic criteria for chronic beryl ium disease (cbd) based on the uk registry - pneumocystis carinii (see "infection, pneumocystis carinii this condition is diagnosed by inspection, palpation, and roentgenography. tension pneumomediastinum may autopsy evaluation of asbestos exposure: retrospective study of cases with quantitation of ferruginous bodies in digested lung tissue atypical mycobacteriosis as a complication of talc pneumoconiosis silicosis, mixed dust pneumoconiosis, and lung cancer complications associated with the use ofthe esophageal-tracheal combitube pneumomediastinum: an unusual complication of asthma in a young man spontaneous pneumomediastinum in a patient with bronchiolitis obliterans after bone marrow references i. katzenstein a-l, myers j. idiopathic pulmonary fibrosis: clinical relevance of pathologic classification update on lymphoid interstitial pneumonitis lymphocytic interstitial pneumonitis and nonspecific interstitial pneumonitis histologic diagnosis of extrinsic allergic alveolitis role of electron microscopy in interstitial lung disease detennination of arsenic in hair using neutron activation arsenic intoxication associated with tubulointerstitial nephritis atrioventricular block after administration of atropine in patients follow-ing cardiac transplantation death following intentional methyl bromide poisoning: toxicological data and literature review pulmonary edema, alveolar wall damage, and interstitial pneumonia after acute inhalation. severe pulmonary fibrosis may develop in chronic cases. gastroenteritis after nonlethal food poisoning. degeneration of proximal tubules and proteinuria in acute poisoning degenerative changes of liver and myocardium problems in the analysis of cadmium in autopsied tissues elevated urinary cadmium concentrations in a patient with acute cadmium pneumonitis cadmium-associated renal disease the effects of storage conditions on the stability of carbon monoxide in postmortem blood acute renal failure, poisoning, carbon tetrachloride synonym: tetrachloromethane poisoning. note: toxicologic sampling of body fluids and organs should be done routinely in all cases. in many instances, however, death occurs i wk to d after exposure, and by this time no carbon tetrachloride is demonstrable. death may be sudden compartment syndrome, and systemic capillary leak syndrome complicating carbon monoxide poisoning southern je acute hydrocephalus following carbon monoxide poisoning prevention of occupational cyanide exposure in autopsy prosectors gradwohl's legal medicine references possible or expected findings digoxin values in digitalis toxicity are greater than ng/ml ( ) digoxin concentration may be higher or lower than concentration in serum, depending on how long before death drug was taken (i) digoxin concentrations in postmortem specimens after overdose and therapeutic use a fluorimetric determination ofmyocardial digoxin at autopsy, with identification of digitalis leaf, digitoxin and gitonin ethylene glycol poisoning: toxicokinetic and analytical factors affecting laboratory diagnosis ethylene glycol poisoning: case report of a record-high level and a review lead concentrations in human plasma, urine and whole blood trace metals (lead and cadmium screening) an emg case report oflead neuropathy years after a gunshot injury aminoaciduria and glycosuria following severe childhood lead poisoning risk of nervous system cancer among workers exposed to lead demonstration of mercury in the human brain and other organs years after metallic mercury exposure spontaneous exfoliation of teeth following severe elemental mercury poisoning: case report and histological investigation for mechanism. oral surg oral med oral pathoi acute chemical pancreatitis associated with nonfatal strychnine poisoning homicide by strychnine poisoning coronary arteritis, with or without aneurysms and infarctions, primarily in childhood. myocardial hypertrophy secondary to hypertension. * minimal or no involvement by polyarteritis nodosa; considerable involvement in other types ofnecrotizing vasculitis (see "arteritis, all types or type unspecified") with or without formation of aneurysms and infarctions, may occur in all organs. the liver may show bile duct injury and rarely, nodular regenerative hyperplasia ( ) more frequently involved in giant cell elastic arteritis.* polyarteritis of small muscular arteries, including vasa nervorum. arthritis* may rarely be present with swollen joints. infarctions;* subarachnoid hemorrhage; arteritis of cerebral arteries. papillitis; retinal hemorrhages polyarteritis nodosa-like vasculitis in human immunodeficiency virus infection the coexistence of familial mediterranian fever and polyarteritis nodosa: report of a case microscopic polyarteritis during polymyalgia rheumatica remission development of polyarteritis nodosa in the course of inactive systemic lupus erythematosus bone marrow pathology in relapsing polychonditis: high frequency of myelodysplastic syndrome relapsing polychondritis associated with subclinical sjo-gren's syndrome and phlegmon of the neck giant cell arteritis and polymyalgia rheumatica posterior scleritis and polymyalgia rheumatica polymyalgia rheumatica in patients with ankylosing spondylitis: a report of cases inflammatory changes of hip synovial structures in polymyalgia rheumatica polymyositis (see "dermatomyositis polyneuritis (see "syndrome, guillain-barre familial, and related syndromes related terms: cowden's disease (multiple hamartoma syndrome); familial colonic polyposis gardner's syndrome; non-polyposis syndrome (hereditary nonpolyposis colorectal cancer syndrome) sampling for histologic study depends on expected findings or grossly identified abnormalities as listed in right-hand column. (see also below under "soft tissues syndrome; mucocutaneous pigmentations (buccal, perioral, priorbital, distal extremities) in peutz-jeghers syndrome;* papules in face and oral mucosa in cowden's disease; tumors of skin and subcutis (see below under "soft tissues gardner's syndrome. osteomas or exostoses (mandible, calvara desmoid in familial adenomatous polyposis malignant potential in intestinal juvenile polyposis syndromes ampullary carcinoma in familial adenomatous polyposis adenoma of the common bile duct in gardner's syndrome may cause relapsing acute pancreatitis orbital osteoma in gardner's syndrome myelopathylmyelitis," and "syndrome, guillain-barre.") references possible or expected findings hypertrichosis; erythrodontia; scarring and mutilation of hands and face. hemolytic anemia. erythrocytes contain large amounts of uroporphyrin i; nonnoblasts and reticulocytes exhibit intense red fluorescence uroporphyrin i and coproporphyrin i in high concentrations. splenomegaly (may have been treated by splenectomy) correction of congenital erythropoietic porphyria by bone marrow transplantation successful cord blood stem cell transplantation for congenital erythropoietic porphyria (gunther's disease) congenital erythropoietic porphyria associated with nephrotic syndrome atypical red cell inclusions in congenital erythropoietic porphyria possible associated conditions: adverse drug reaction; chronic alcoholism;* chronic hepatitis c ( ); human immunodeficiency virus infection porphyria cutanea tarda and human immunodeficiency virus: two cases associated with hepatitis c porphyria cutanea tarda and antibodies to hepatitis c virus an unhappy triad: hemochromatosis, porphyria cutanea tarda and hepatocel ularcarcinoma-a case report variegate porphyria possible associated conditions: ebstein's malformation of tricuspid valve. references possible or expected findings chronic eczematous skin lesions or superficial scarring; nail changes. perivascular pas-positive hyaline ( ) brown protoporphyrin deposits with red to yellow birefringence with a maltese cross configuration in hepatocytes, kupffer cells, and bile canaliculi erythropoietic protoporphyria cytofluorometry as a diagnosis of protoporphyria recessive inheritance of erythropoietic protoporphyria with liver failure pseudogout synonym: calcium pyrophosphate dihydrate (cppd) deposition disease puncture grossly affected joints and submit sample of synovial fluid for crystal analysis under compensated polarized light ( ) references possible or expected findings punctate calcium deposits in kneejoints and, less commonly, in joints of hips, ankles, shoulders, or wrists or in symphysis ossium pubis and intervertebral disks. arthritis* with synovitis. crystals of calcium pyrophosphate dihydrate in periarticular tissue ( ) and synovial fluid. cartilage contains calcium salts of pyrophosphate, hydroxyapatite, and orthophosphate. alkaptonuria;* gout calcium pyrophosphatedihydratedeposition disease presenting as tumoral calcinosis (periarticularpseudogout) calcifications in dermis; calcifications of blood vessels. diaphragmatic hernia.* hypertensive cardiomegaly. characteristic plaques in pericardium and endocardium, with or without mitral valve involvement. coronary atherosclerosis may have been a causeofangina( ).coronarythrombosisand myocardial infarction may be present also. accelerated atherosclerosis; calcification of peripheral vessels. gastrointestinal hemorrhage;* peptic ulcer degeneration of bruch's membrane with retinal hemorrhages; sclerosis of choroid vessels calcification of elastic fibers in pseudoxanthoma elasticum new report of severe coronary artery disease in an eighteen-year-old girl with pseudoxanthoma e asticum. case report and review of the literature hivrelated thrombotic thrombocytopenic purpura: report of cases and a review of the literature beham-pyelonephritis synonyms and related terms: acute ascending pyelone angiotropic large cell lymphoma presenting as thrombotic micro-angiopathy (thrombotic thrombocytopenic purpura) splenic pathology in thrombotic thrombocytopenic purpura emphysematous pyelonephritis in diabetic patients nephrobronchialfistulaandlungabscessresulting from nephrolithiasis and pyelonephritis a case ofan enterorenal pyothorax (see "empyema, pleurai.") fistula and pyelonephritis with airin renal pelvis systemic amyloidosis secondary to pyonephrosis. resolution after nephrectomy medicolegal investigation of death the laboratory's role in detecting sexual assault toluidine blue in the detection at autopsy of perineal and anal lacerations in victimes of sexual abuse sarcoidosis presenting as heart disease sarcoidosis and its ocular manifestations rheumatic features of sarcoidosis cutaneous sarcoidosis: differential diagnosis schistosoma mansoni and s. haematobium infections in egypt. i. evaluation of techniques for recovery of worms and eggs at necropsy. am i a quantitative post-mortem study of schistosomiasis mansoni in man immunodiagnosis of schistosomiasis. screen with fast-elisa and confinn with imrnunoblot hepatic connective tissue changes in hepatosplenic schistosomiasis schistosomiasis in women: manifestations in the upper reproductive tract sclerosis, systemic amyotrophic lateral (see "disease, motor neuron diffuse (see "sclerosis, schilder's cerebrai multiple synonyms and related terms: acute and subacute variants: acute multiple sclerosis (marburg type), acute necrotizing myelopathy; concentric sclerosis (ba type); concentric lacunar leukoencephalopathy; encephalitis periaxialis diffusa (schilder's type; see also next entry silicone ganuloma in acral skin in a patient with silicone-gel implants and systemic sclerosis organizing diffuse alveolar damage associated with progressive systemic sclerosis bayle , fiessinger in. brain involvement in scleroderma: two autopsy cases skeletal muscle pathology in systemic sclerosis mechanisms of scleroderma-induced lung disease thberous sclerosis complex and subependymal giant cell astrocytoma bilateral temporal arachnoid cysts associated with tuberous sclerosis poisoning, alkaloid accident, diving [skin or scuba scurvy (see "deficiency, vitamin c shigellosis (see "dysentery, bacillary shock related terms: anaphylactic shock; bacteremic shock; many others. note: see also under name of suspected underlying condition, such as "bums intestinal histological and ultrastructura inflammatorychanges in spondyloarthropathy and rheumatoid arthritis the sacroiliac joint in the spondyloarthropathies granulomatous ileitis in a patient with ankylosing spondylitis sporotrichosis synonym: sporothrix (sporotrichum) schenckii infection. note: ( ) mckiernan . partial lipodystrophy in coeliac disease coeliac disease and lymphoma: current status review: nonalcoholic steatohepatitis alcoholic and non-alcoholic wernicke's encephalopathy. be alert to the preventable and treatable disease non-alcoholic fatty liver disease in the metabolic syndrome the placenta in stillbirth estimating the time of death in stillborn fetuses: i. histologic examination of fetal organs: an autopsy study of stillborns estimating the time of death in stillborn fetuses: ii. histologic evaluation ofthe placenta; a study of stillborns estimating the time ofdeath in stillborn fetuses: iii. external fetal examination; a study of stillborns fat distribution in the adrenal cortex as an indication of the mode of intrauterine death stimulant(s) (see "dependence, drug(s), all types or type undetermined cutaneous manifestations of the acquired immunodeficiency syndrome aids: cardiac findings from autopsies the liver in hiv infection neuropathology of the spinal cord in the acquired immunodeficiency syndrome postmortem enzyme immunoassay for human immunodeficiency virus bk virus-associated renal failure among hiv patients - or write to american registry ofpathology sales office pathology of acquired immunodeficiency syndrome (aids) in children thymus involution in the acquired immunodeficiency syndrome micrencephaly in children congenitally infected with human immunodeficiency virus-a gross-anatomical morphometric study a rapid method for detecting the predominant ashkenazi jewish mutation in the bloom's syndrome gene bloom syndrome: multiple retinopathies in a chromosome breakage disorder severe eczema in a patient with di-george's syndrome digeorge's syndrome orthe iii -iv pharyngeal pouch syndrome: pathology and a theory of pathogenesis basal ganglia calcification and psychosis in q . deletion syndrome down's synonyms and related terms: mongolism possible associated conditions: acute lymphocytic, my references epicanthal folds; cleft lip/palate; high arched palate; furrowed tongue; rhagades. depressed nasal bridge small, broad or fiat nose; slanted palpebral fissures; fiat occiput; brachycephaly; palmar "simian crease"; short fifth middle finger ventricular septal defect;* complete atrioventricular septal defect* ( ). duodenal stenosis and atresia; imperforate anus. microcephaly; poorly developed secondary gyri hypoplastic brain stem, medulla and cerebellum; polymicrogyria; neurofibrillary tangles; meningomyelocoele. acute lymphocytic leukemia; acute myelocytic leukemia; acute megakaryocytic leukemia cardiovascular disease in down syndrome possible or expected findings hyperelasticity of skin, bruises or scars, hemorrhages, and hyperpigmentation. lipomatous pseudotumors that may be calcified or ossified. normal or abnormal amounts and fragmentation of elastic tissue. abnormalities of collagen. dislocation of hip, shoulder, patella, radius, or clavicle. loose-end clavicles; spondylolisthesis mitral valve prolapse; aortic insufficiency.* aortic dissection, aneurysm, ectasia, occlusion ( ). various congenital anomalies. congenital anomalies of gastrointestinal and genitourinary tracts. bleeding from various organ sites vascular ehlers-danlos syndrome: imaging findings note: this syndrome belongs to a large family (with more photograph abnormal features as listed in right-hand column. record appearance of external genitalia. prepare skeletal roentgenograms. procure long bones and vertebral bodies short-limb dwarfism* with narrowing of the rib cage; polydactyly; dysplasia of fingernails; thin and sparse hair; premature eruption of teeth; defective dentition; eye abnormalities; upper lip bound down by multiple frenula. cryptorchidism; epispadias; hypospadias. chondrodysplasia with acromelic micromelia (shortening of the distal segment of the limb); fusion of capitate and hamate bones of wrist; defects of lateral aspect of proximal tibia see lesions listed above under "external examination empty sella synonyms and related terms: primary empty sella syndrome; secondary empty sella syndrome (e.g., after surgical removal or spontaneous infarction of pituitary adenoma) chorea, philic pneumonia: histopathologic features in nine cases. am rev resp hereditary ascariasis and hookworm chronic eosinophilic pneupophosphatemic vitamin d-resistant rickets; galactosemia;transport defect; tyrosinemia. * excludes fan- weight and external measurements; record and photograph abnormalities as listed in right-hand column. prepare skeletal roentgenograms. radiographs of long bones. submit for tissue culture for possible enzyme analysis. possible or expected findings redlblond hair, fair skin (diminished pigmentation); dehydration;* stigmata of hypothyroidism;* delay of sexual maturation. rickets; osteomalacia.* positive rheumatoid factor. various types of pneumonia after splenectomy, presence of accessory spleen may account for treatment failure. manifestations of portal hypertension.* lymphadenopathy of mediastinal and paraaortic lymph nodes. anemia;* neutropenia sinusoidal lymphocytosis of the liver in felty's syndrome with a review of the liver involvement in felty's syndrome splenic involvement in rheumatic diseases felty's and pseudo-felty's syndrome nodular regenerative hyperplasia of the liver in rheu-matic diseases: report of seven cases and review of the literature fetal alcohol syndrome: craniofacial and central nervous system manifestations goodpasture's syndrome anti-glomerular basement membrane glomerulonephritis: a morphologic study of cases gronblad-strandberg (see "pseudoxanthoma elasticum acute inflammatory polyradiculoneuropathy; guillain-barre-strohl syndrome hemolytic uremic syndrome/thrombotic thrombocytopenic purpura: pathophysiology and treatment cyclosporin-induced hemolytic uremic syndrome: factors that obscure its diagnosis enterohemorrhagic escherichia coli :h : an emerging pathogen hemolytic uremic syndrome as a presenting form of hiv infection hemolytic uremic syndrome in prostatic carcinoma diabetes secondary to genetic disorders. baillieres suprasellar tumors of maldevelopmental origin in klinefelter's syndrome. a report of two cases klippel-feu synonym: congenital fusion of cervical vertebrae primary yolk sac tumor of the prostate in a patient with klinefelter's syndrome extragonodal germ cell tumors are often associated with klinefelter syndrome organs and tissues procedures possible or expected findings external examination prepare roentgenograms of chest, neck, and head skull, spine, brain, and spinal cord myasthenia gravis disorders with dysraphia (see below), fusion of cervical vertebrae. congenital elevation of the scapula (sprengel's deformity) chiari malformation;* basilar impression;* meningomyelocele; platybasia;* spinal cord compression; weight and length; record and photograph abnormalities as listed in right-hand column. record weight and sample for histologic study. if renal transplantation ( ) had been carried out, see also under that heading. follow procedures described under "glomerulonephritis retinal dystrophy ( ) and other retinal changes the cardinal manifestations of bardet-biedl syndrome, a form of laurence-moon-biedl syndrome laurence-moon-biedl syndrome accompanied by congenital hepatic fibrosis pediatric renal transplantation in laurence-moon-biedl syn-drome possible or expected findings typical skeletal proportions. arachnodactyly; pectus excavatum genu recurvatum; dislocation of joints diaphragmatic hernia. * infective endocarditis. * atrial septal defect. * myxomatous transformation of mitral ring. mitral valve prolapse. aortic and pulmonary dilatation and valvular insufficiency.* aortic dissection* with dissection of adjacent vessels. ascending aortic aneurysm multiple cysts (see "cyst(s), pulmonary") aortopulmonary fistula: an uncommon complication in dystrophic aortic aneurysm peripartum acute myocardial infarction in marfan's syndrome aneurysm of the cervical internal carotid artery associated with marfan's syndrome--ease report noonan's syndrome in association with acute leukemia lymphatic dysplasia in a neonate with noonan's syndrome noonan syndrome: a clinical description emphasizing the cardiac findings cerebral arteriovenous malformation in noonan's syndrome clinical variability in a noonan syndrome family with a new ptpnii gene mutation peutz-jeghers syndrome bilateral large-cell calcifying sertoli cell tumor of the testes in peutz-jeghers syndrome: a case report ovarian tumors associated with the peutz-jeghers syndrome robin sequence and a deficiency of the left forearm in a girl with a deletion of chromosome g -gter congenital heart disease in the pierre robin syndrome glossoptosis-apnea syndrome chiari i malformation, caudal regression syndrome, and pierre robin syndrome: previously unreported combination risk factors for squamous cell carcinoma of the esophagus heterotaxy syndrome; ivemark's syndrome possible associated conditions: with asplenia: right isomerism (bilateral mirror-image right-sided symmetry) of heart, lungs, and abdominal viscera; common atrium; total anomalous pulmonary venous connection; absent coronary sinus; complete atrioventricular septal defect;* subpulmonary stenosis;* pulmonary valve atresia;* midline symmetric liver; malrotation of bowel; absent spleen. with polysplenia: left isomerism (bilateral mirror-image left-sided symmetry) ofheart and lungs, with variable sidedness ofabdominal viscera primary immunodeficiencies reiter's syndrome-like patternin aids-associated psoriasiformdermatitis reference pneumothorax;* pneumomediastinum;* pneumoperitoneum. septicemia. right ventricular hypertrophy and/or dilatation. intubation trauma and mural edema and hemorrhage in trachea; hyaline membrane disease intubation trauma. hemorrhages or other evidence of birth trauma; germinal matrix hemorrhages in premature infants; infarctions of subcortical white matter in term infants coalson . pathology of bronchopulmonary dysplasia pathology of arrested acinar development in postsurfactant bronchopulmonary dysplasia inborn errors of metabolism and reye's syndrome: differential diagnosis prevalence and non-specificity of microvesicular fatty changes liver histopathology in clinical reye syndrome ultrastructural study of intranuclear inclusions in the exo-crine pancreas in reye's syndrome neuropathologic findings in reye syndrome sanfilippo's (see "mucopolysaccharidosis scheie's (see "mucopolysaccharidosis segawa's (see "syndrome, dystonia severe acute respiratory (sars) note: this highly contagious illness, caused by a coronavirus (sars-cov)details, see biosafety level laboratory for autopsies of patients with severe acute respiratory syndrome: principles, practices, and prospects pathology and pathogenesis of severe acute respiratory syndrome pulmonary involvement in primary sjogren's syndrome polymyositis and sjogren's syndrome associated with bronchiolitis obliterans organizing pneumonia the gastrointestinal manifestations of sjogren's syndrome clinicopathological factors relating malignant lymphoma with sjogren's syndrome acute transverse myelopathy and cutaneous vasculopathy in primary sjogren's syndrome primary localized cutaneous amyloidosis with unusual clinical features in a patient with sjogren's syndrome disease, parkinson's erythema multiforme man related terms: armadillo disease; continuous muscle fiber activity; neuromyotonia autoimmune central nervous system paraneoplastic disorders: mechanisms, diagnosis, and therapeutic options sudden infant death (sids) (see "death, sudden unexpected, of infant.") syndrome, superior vena cava related term: superior vena cava obstruction continue dissection of veins into neck and axillas. record and photograph site of thrombosis or of compression by surrounding pathologic conditions. submit samples for histologic study. benign or malignant tumors; fibrosing mediastinitis;* postradiation fibrosis; infectious disease (tuberculosis,* histoplasmosis*); thoracic aortic aneurysm;* chronic constrictive pericarditis;* chest trauma; arteriovenous fistula between ascending aorta and superior vena cava toxic shock synonyms and related terms: staphylococcal scarlet fever. note: ( ) collect all tissues this is a reportable disease. premature aging; increased number of pigmented nevi; infantile sex organs; webbed neck; broad chest with wide spacing of nipples. short fourth metacarpal; abnormal epiphyseal fusions; osteochondrosis-like changes of spine bicuspid aortic valve;* coarctation of aorta. * rarely other anomalies ( ) decreased/absent follicles. intestinal telangiectases. biliary atresia. * horseshoe kidney; double ureters. streak gonads without germ cells or follicles hashimoto's thyroiditis. * manifestations of diabetes mellitus,* hypertension,* or thyrotoxicosis. neuroblastoma and related tumors ( ). keratoconus acute aortic dissection, aortic insufficiency, and a single coronary artery in a patient with turner's syndrome neuroblastoma and related tumors in turner's syndrome turner's syndrome associated with bilateral retinal detachments weil's (see "leptospirosis related terms: alcoholic wernicke's encephalopathy; korsakoff's psychosis; nonalcoholic wernicke's encephalopathy ( , ); and specimen preparation, see chapter . for selection of histologic samples, see right-hand column melissas . wernicke's encephalopathy after vertical banded gastroplasty for morbid obesity syndrome, respiratory distress, of infant.") syndrome, wiskott-aldrich (see "syndrome, primary immunodeficiency malformation, ebstein's" and "preexcitation, ventricular alcoholic and non-alcoholic wernicke's encephalopathy. be alert to the preventable and treatable disease cerebro-hepato-renal syndrome; infantile refsum's disease. * note: this congenital familial cholestatic syndrome results from impaired assembly of peroxisomes and has its main manifestations in the brain, liver, and kidneys postmortem findings and prenatal diagnosis ofzellwegersyndrome. case report alcoholism and alcohol intoxication" and "disease, alcoholic liver possible associated condition: multiple endocrine neoplasia primary cardiac gastrinoma causing zollinger-ellison syndrome localization, syphilis, acquired synonym: treponema pallidum infection. note: congenital syphilis is presented below under a separate heading and surgical management of gastrinoma fundic argyrophil carcinoid tumor in a patient with sporadic-type zollinger-ellison syndrome nonimmune hydrops fetalis due to congenital syphilis associated with negative intrapartum mater-nal serology screening syringomyelia synonyms and related term: hydromyelia; idiopathic syringomyelia; secondary syringomyelia; syringobulbia. possible associated conditions: with idiopathic syringomyelia-arnold chiari malformation, type i;* basilar impression;* . oppenheimer eh, dahms bb. congenital syphilis in the fetus and neonate congenital syphilis: detection of treponema pallidum in stillborns with secondary syringomyelia-intramedullary gliomas (ependymoma, pilocytic astrocytoma) and vascular tumors; spinal arachnoiditis and pachymeningitislisted in right-hand column. prepare roentgenograms of spine and joints. for removal and specimen preparation, see chapter . for histologic sampling, see right-hand column. hypertrophy of body parts. muscle atrophy of upper extremities and hands cervical spinal cord is swollen and tense, with cavitation (syrinx) containing clear fluid. wall of cavity consists of degenerated glial and neural elements, with marked gliosis. spinal cord parenchyma is markedly compressed. see also above under hydrops fetalis caused by alpha-thalassemia: an emerging health care problem limb defects in homozygous alpha-thalassemia: report of three cases poisoning, thallium thirst (see "dehydration thromboangiitis obliterans (see "disease, buerger's purpura, thrombotic thrombocytopenic" and "syndrome, hemolytic uremic the pathological manifestations of pulmonary toxoplasmosis in the acquired immunodeficiency syndrome a spectrum in the pathology of toxoplasmosis in patients with acquired immu transfusion (see "reaction to transfusion bone marrow note: if the patient had symptoms of acute or chronic graft-versus-host disease (gvhd), see "disease, graftversus-host morphology and diagnostics of human toxoplasmosis fatal myocardial coinfection by toxoplasma gondii and parvovirus b in an hiv patient if there is evidence of infection, submit material for microbiologic studies. if there is evidence of recurrent leukemia or lymphoma, sample as described under those headings. if the patient had symptoms of thrombotic thrombocytopenic purpura (tip) or hemolytic uremic syndrome (hus), see these headings. record average size. fix specimens in b-plus®. make touch preparations. request giemsa or wright stain. for preparation of sections and smears (imprints), see chapter . manifestations of graft-versus-host disease (e.g., scleroderma-like changes).* septicemia. interstitial pneumonia* and cytomegalovirus infection* or human herpesvirus infection ( ). bacterial, fungal bacterial or fungal infections in gvhd.* other manifestations of gvhd. * recurrent leukemia,* lymphoma* (including posttransplant, epstein-barr virus associated lymphoproliferative disorder). recurrent solid tumor, e.g., carcinoma of breast, lung (small cell carcinoma), ovary lymphomatous or leukemic infiltrates myeloid cells may be absent in marrow graft rejection or nonimmunologic marrow graft failure. references hematomas; hemorrhagic necroses; infarcts; bacterial or fungal infections leukoencephal-opathy; vascular siderocalcinosis; neuro-axonal spheroids ( ) human herpesvirus infection and associated pathogenesis following bone marrow transplantation pulmonary veno-occlusive disease in an adult following bone marrow transplantation. case report and review of the literature acute renal failure following bone mar-row transplantation transplantation-associated thrombotic thrombocytopenic purpura and hemolytic uremic syndrome thrombotic microangiopathies associated with drugs and bone marrow transplantation record status of all vascular anstomoses. if there are hemorrhages, record volume and site. record weight, ventricular thickness, and valve circumferences. take multiple sections from all areas of myocardium. cut coronary arteries in cross sections; request verhoeff-van gieson stain. procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. references cushingoid features (see "syndrome, cushing's") after steroid therapy. suture dehiscence. hemothorax or hemomediastinum in recent cases. left ventricular hypertrophy, from cyclosporine-related hypertension. acute transplant rejection (for grading, see ref. ). chronic transplant vasculopathy with coronary artery stenosis ( ) opportunistic infection ( ). pneumonia a working formulation for the standardization of nomenclature in the diagnosis of heart and lung rejection: heart and lung rejection study group autopsy findings in cardiac transplant patients: a lo-year experience fulminant toxoplasmosis following heart transplantation banff schema for grading liver allograft rejection: an international consensus document record status of all vascular anstomoses. if there are hemorrhages, record volume and site. record lung weights separately. if lung infection is suspected, submit material for microbiologic possible or expected findings cushingoid features (see "syndrome, cushing's") after steroid therapy. suture dehiscence. hemothorax or hemomediastinum in recent cases. opportunistic infection (in patients with a single lung transplant, infections may involve references the nontransplant lung, as well). chronic bronchitis* and bronchiectasis.* acute rejection (rare); chronic airway rejection (obliterative bronchiolitis); chronic vascular rejection. (for grading of rejection, see ref .) post-transplant lymphoprolijerative disorder a working formulation for the standardization of nomenclature in the diagnosis of heart and lung rejection: heart and lung rejection study group recurrent salmonella enteritidis and hepatic tuberculosis tularemia synonyms and related terms: francisella tularensis infection; pasteurella tularensis infection cerebral abscesses complicating tularemia meningitis endocrine (see "neoplasia, multiple endocrine any type note: ifthe tumor had been treated by surgery, irradiation, chemotherapy, or other means (the most common situation possible associated conditions: with adrenocortical adenoma-conn's syndrome with adrenocortical carci-noma-cushing's syndrome;* hypoglycemia;* virilization. with pheochromocytoma-cerebellarhemangioblastoma ery-throcytosis; hypercalcemia; hypertension,* multiple enarteries record body weight and abnormal features. photograph skin changes. record heart weight and thickness of ventricles. procure multiple sections of myocardium possible associated conditions.") focal myocarditis. * hypertensive cardiovascular disease (see "hypertension myocardial infarction without severe coronary artery disease (with pheochromocytoma) asystematicreviewoftheassociationbetween barrett's esophagus and colon neoplasm malakoplakia and colorectal adenocarcinoma an unusual case of colonic mixed adenoendocrine carcinoma: collision vs composite tumor. a case report and review of the literature submit lymph nodes for histologic study. remove neck organs with tongue together with esophagus. open pharynx and esophagus in posterior midline. if fistulas and abscesses are suspected, dissect esophagus but leave attached to mediastinum, stomach and diaphragm. record width of lumen of esophagus at various levels. photograph and record size and location of tumor; sample tumor and uninvolved esophagus for histologic study. request pas stain of uninvolved esophagus (sample from all levels). procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. references cachexia. eczematoid dermatitis with adult renal cell carcinoma. cushing's syndrome,* galactorrhea or feminization or masculinization in some cases of renal cell carcinoma. evidence of hyperalcemia. pulmonary tumor embolism after invasion of renal vein and inferior vena cava. acquired renal cystic disease prolactin, renin, and prostaglandins in some renal cell carcinomas. see above under "possible associated conditions wilms tumor associated with polycythemia: case report and review of the literature renal medullary carcinoma: a potential sickle cell nephropathy of children and adolescents acquired cystic kidney disease byler's disease, carcinoid syndrome,*cirrhosis* of any type (mostly nonbiliary), congenital hepatic fibrosis, * cushing's syndrome,* erythrocytosis, genetic hemochromatosis, * glycogen storage disease (type ),* hepatitis b or c virus infection, hereditary tyrosinemia (type i),* hypercalcemia, hypercholesterolemia, hypoglycemia,* neurofibromatosis,* osler-rendu-weber disease* (hereditary hemorrhagic telangiectasia), polycythemia,* porphyria (acute intermittent and porphyria cutanea tarda), * pseudohyperparathyroidism,* and thorium (thorotrast) deposition. with bile duct carcinoma (cholangiocarcinoma}-clonorchis sinensis or opisthorchis viverrini infection, fibropolycystic liver and biliary tract disease,* hepatolithiasis, hypercalcemia, pri-mary sclerosing cholangitis,* and thorium (thorotrast) deposition. with hepatoblastoma-alpha-fetoproteinemia, cardiac and renal malformations, cleft palate, diaphragmatic hernia,* down's syndrome,* familial colonic polyposis,* hemihypertrophy, nephroblastoma. with angiosarcoma-thorium (thorotrast) deposition. see also below under "possible or expected findings describe and photograph cut surfaces. sample tumor and nonneoplastic liver for histologic study. if thorium deposition is suspected, prepare roentgenograms of liver slices and submit samples for energy-dispersive x-ray microanalysis of paraffin sections. procedures depend on expected findings or grossly identified abnormalities as listed above and in right-hand column. cachexia. precocious puberty. feminization and gynecomastia in rare cases of hcc. spider angiomas; clubbing of fingers. see above under "possible associated conditions thorotrast storage may cause cirrhosis, hcc, bile duct carcinoma, or angiosarcoma. see above under "possible associated conditions tumor of the lung or bronchus possible associated conditions: abnormal concentrations of hormons or other metabolites in blood and tumor tissue (adrenocorticotropic, antidiuretic, growth hormone, parathyroid-like substances, or -hydroxyindolacetic acid); acanthosis nigricans s syndrome;*dermal hyperpigmentation; feminization; hyperglycemia; hypercalcemia; hypoglycemia;* hypokalemia; hyponatremia; precocious puberty. for syndromes affecting the brain, peripheral nerves or muscles, see below under "possible or expected findings organs and tissues procedures possible or expected findings external examination and skin record body weight. record abnormal features as listed in right-hand column; prepare photographs. prepare histologic sections of normal and grossly abnormal skin clubbing of fingers; spider angiomas. for other rare tumor-related changes, see above under "possible associated conditions thmor of the pancreas possible associated conditions: with carcinoma of the exocrine pancreas--cushing's syndrome;* diabetes mellitus* (rare); hypercalcemia; hyperglycemia with islet cell tumor-abnormal concentrations of hormons in blood and tumor tissue (see below under "pancreas"); features. dermal hyperpigmentation. for other rare tumor-related changes, see above under "possible associated conditions biliary obstruction caused by tumor in head of pancreas. islet cell tumor with adrenocorticotropic hormone, gastrin, glucagon, insulin, or other peptide hormones. see above under "possible associated conditions paraneoplastic pemphigus associated with pancreatic carcinoma pancreatic cystadenocarcinoma in peutz-jeghers syndrome testes may have been surgically removed to achieve androgen deprivation. urinary obstruction and hydronephrosis. * osteoblastic metastases hemolytic uremic syndrome in prostatic carcinoma thmor of the small intestine possible associated conditions: acquired immunode oncogenic osteomalacia associated with prostatic cancer gardner's syndrome) submit sample of non-neoplastic small bowel for histologic study. procedures depend on expected findings or grossly identified abnormalities as listed above. cachexia. mucocutaneous pigmentations associated with peutz-jeghers syndrome. * carcinoid tumor. lymphoma (see also above under "possible associated conditions"). familial polyposis or related syndrome. * celiac sprue. * crohn's disease. * see above under "possible associated conditions kasabach merritt syndrome in infants paraneoplastic hepatopathy associated with soft tissue sarcoma neurofibro-thmor of the spinal cord possible associated conditions: with angioma-cerebellar hemangioblastoma; segmental cutaneous vascular nevi; with matosis in children with soft tissue sarcoma tissues record abnormal features; photograph and submit nevi for histologic study. for removal and specimen preparation, see chapter . see also below under "vertebral column bone erosion and calcification of spinal canal. vertebral hemangiomas may be associated with arteriovenous malformation of spinal cord. manifestations of von hippel-lindau disease. * thmor of the stomach possible associated conditions: hypoglycemia;* megaloblastic anemia;* skin changes (as listed under "possible or expected findings)weight. record abnormal features; photograph and submit samples of normal and abnormal skin for histologic study. possible or expected findings cachexia; lower leg lymphoedema ( ) metastases common in liver, retroperitoneal and supraclavicular lymph nodes (virchow's node), ovaries (krukenberg tumor), and peritoneal cui de sac (blumer's shelf') carcinoma of the stomach with hyperkeratosis palmaris and plantaris and acanthosis of the esophagus gastric lymphoma ofmucosa-associated lymphoid tissue and helicobacter pylori gastric signet-ring cell carcinoma: unilateral lower extremity lymphoedema as the presenting feature thmor of the testis possible associated conditions: demyelinating neuropathy record location, size, and weight of both testes and of testicular tumor. submit samples of tumor and of ininvolved testis and epididymis for histologic study. snap-freeze tumor tissue for hormone assay. possible or expected findings cachexia. cryptorchism; hypospadia. gynecomastia. increased concentrations of alphafetoprotein and human chorionic gonadotropin testicular microlithiasis. secondary testicular tumors (metastases to the testes) are rare, except in association with leukemia* in children. references pulmonari embolism.* paraneoplastic diseases as listed above under "possible associated conditions chronic inflammatory demyelinating polyneuropathy (clop) associated with seminoma dermatomyositis associated with intratubular germ cell tumor and metastatic germ cell cancer testicular cancer in association with developmental renal anomalies and hypospadias submit samples of lymph nodes, spleen, peyer's plaques, and bone marrow for histologic study. other procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. references cachexia. dermal hyperpigmentation. pemphigus foliaceus (rare). hypogammaglobulinemia and other immunoglobulin abnormalities. usually, thymoma presents as an infiltrating, anterior mediastinal mass that rarely metastasizes. carcinoid tumor, malignant lymphoma* (hodgkin's disease), and metastases from carcinomaofthe breast* and other tumors also may occur in this location herpes simplex*) and fungal infections (e.g., candidiasis*) due to thymoma-related immunodeficiency ( ). manifestations of paraneoplastic diseases and conditions as listed above under "possible associated conditions thrombotic thrombocytopenic purpura accompanied by transient pure red cell aplasia and thymoma glomerulonephritis associated with myasthenia gravis thymoma and cellular immune deficiency in an adolescent possible associated conditions: with papillary carcinoma-familial adenomatous polyposis thyroid carcinoma associated with familial adenomatous polyposis an association between acromegaly and thyroid carcinoma thyroid carcinoma causing fatal laryngeal obstruction prostate cancer metastasis to thyroid gland schistosomiasis and the risk of bladder cancer in human papilloma virus infection ( ) or herpesvirus infection ( ) with invasive cervical carcinoma. erythropoietin may be found in some tumors. thrombosis. * myopathy* may be associated with carcinoma of the cervix. cerebellar cortical degeneration* may be associated with carcinoma of the uterus. manifestations of uremia (see "failure, kidney") invasive cervical cancer in human immunodeficiency virus-infected and uninfected hospital patients association of risk factors for cervical cancer and human papilloma viruses in invasive cervical cancer association of herpesvirus infection with the development of genital cancer tyrosinemia type ii: a challenge for ophthalmologists and dermatologists hereditary tyrosinemia type i (chronic form): pathologic findings in the liver richner-hanhart syndrome (tyrosinemia ii): early diagnosis of an incomplete presentation with unusual findings acute pancreatitis-associated acute gastrointestinal mucosal lesions: incidence, characteristics, and clinical significance uncinariasis (see "ancylostomiasis uremia (see "failure, kidney from: handbook of autopsy practice urticaria pigmentosa of childhood (see ''mastocytosis, systemic acute aortic dissection;* aneurysma(s) of cerebral arteries; aortic insufficiency;* calcific aortic stenosis,* coarctation of the aorta;* infective endocarditis;* shone's syndrome; turner's syndrome. * organs and tissues procedures possible or expected findings heart if infective endocarditis is suspected, see chapter . open heart in cross-sections (see chapter ). photograph aortic valve and test valvular competence note: reye's syndrome* is a possible postviral complication of varicella that must be distinguished from varicella encephalitis. varicella may also cause exacerbation of tuberculosis.* (i) collect all tissues that appear infected bicuspid aortic valves are associated with aortic dilatation out of proportion to coexistent valvular lesions varicella-zoster virus infection in children with underlying immunodeficiency virus infection group a beta-hemolytic streptococcal epiglottitis as a complication of varicella infection optic neuritis: a rare complication of primary varicella infection hsv-l-induced acute retinal necrosis syndrome presenting with severe inflammatory orbitopathy, proptosis, and optic nerve involvement varicella with acute motor axonal neuropathy descending necrotizing mediastinitis in a child with chicken pox double outlet right ventricle with intact ventricular septum respiratory syncytial (see "pneumonia, all types or type unspecified vitamin a (see ''deficiency, vitamin a" and "hypervitaminosis a vitamin bt (thiamine) (see "syndrome, wernicke-korsakott vitamin bt (see "anemia, megaloblastic deficiency deficiency, vitamin d" and "hypervitaminosis d waldenstrom's macroglobulinemia (see ''macroglobulinemia, waldenstrom's.) waterhouse-friderichsen syndrome (see "disease, meningococcal wegener's granulomatosis (see "granulomatosis, wegener's werdnig-hottman disease (see "disease, motor neuron acid esterase activity in cultured skin fibroblasts and amniotic fluid cells using -methylumbelliferyl palmitate which may be secondarily infected by bacteria. dark field microscopy will identify the organism expected findings mucosal ulcers, resembling those of typhoid fever, primarily of distal ileum and colon. villous shortening, crypt hyperplasia with mucosal microabscesses. microabscesses in regional mesenteric lymph nodes zellweger's syndrome (see "syndrome mucormycosis" and procedures under "diabetes mellitus abscesses and granulomas may occur in all organs and tissues. granulomatous lesions in central nervous system ( ) and eyes ( ). fungal osteomyelitis* that may be multifocal, including sites such as metacarpals and metatarsals. external examination other procedures depend on expected findings or grossly identified abnormalities as listed in right-hand column. osteosclerosis of vertebrae, ribs, clavicles, pelvic bones, scapulae, skull, and metaphyseal ends of femur and humerus. osteomyelofibrosis or osteoreticulosis; rarely, panhyperplasia of bone marrow; increase of megakaryocytes.changes associated with chronic granulocytic leukemia* or with polycythemia vera* may imitate idiopathic myelofibrosis. widespread extramedullary hematopoiesis ( ) with splenomegaly. infectious diseases, including tuberculosis;* gouty arthritis; ascites and other manifestations of portal hypertension;* or hepatic vein thrombosis (budd-chiari syndrome*). cardiac tamponade ( ) . synonyms and related terms: hepatic porphyria; protoporphyrinogen oxidase deficiency ( ) .note: the manifestations of the disease closely resemble those in porphyria cutanea tarda and hereditary coproporphyria. measurements of porphyrins and porphyrin precursors are the only clearly distinguishing features. for autopsy procedures, see under "porphyria cutanea tarda." fibrin and platelet thrombi, with or without microaneurysms and purpura in kidneys, adrenal glands, pancreas, heart, and brain; lesions may also be present in liver; spleen ( ), lymph nodes, muscle, bone marrow, and synovium. fibrin thrombi can be demonstrated with labeled antihuman fibrin antibodies. lesions in precapillary arterioles. hypertrophic osteoarthropathy* (with pleural mesothelioma).pleural effusions. * asbestosis may be complicated by pleural mesothelioma. other organs and tissue sample bladder tumor and uninvolved urinary bladder for histologic study.cachexia. hydronephrosis* (obstructive uropathy) and hydroureter, usually caused by distal ureteral obstruction. recurrent nephrolithiasis* or pyelitis may have been present and is a risk factor for urinary bladder carcinoma.chronic urocystitis; infestation with schistosoma haematobium ( ) (uncommon in north america). manifestations of uremia (see "failure, kidney").