key: cord-0803653-b0fjqj4h
authors: Jefferson, T.; Dietrich, M.; Brassey, J.; Heneghan, C.
title: Understanding Definitions and Reporting of Deaths Attributed to COVID-19 in the UK - Evidence from FOI Requests
date: 2022-04-29
journal: nan
DOI: 10.1101/2022.04.28.22274344
sha: 9d3879a8ed47fec5aa7c0eda8e24c2a47e8f040c
doc_id: 803653
cord_uid: b0fjqj4h

Death is a widely used outcome to assess the severity of pandemics. Accuracy in assigning the cause of death is of vital importance to define the impact of the agent, monitor its evolution, and compare its threat with those of other agents. Throughout the COVID-19 pandemic, there has been widespread reporting of aggregate death data with little attention paid to the accuracy of the assignment of causation. We aimed to analyse public authorities' understanding of the assignment of cause of deaths during the SARS-CoV-2 pandemic in the UK by accessing Freedom of Information requests posed in three periods in 2020-21. By public authorities, we mean NHS Health Trusts, laboratories, and government agencies such as Public Health England and the Department of Health and Social Care. We searched WhatDoTheyKnow using the terms "covid and death". We excluded those requests to bodies that cannot provide an answer (e.g. Councils) and those dealing with the effects of vaccines. We grouped questions into themes addressing the definitions and causes of death relevant to the pandemic. We looked at the responses to the questions of the definition of cause of death, the accuracy of the attribution, the role of other pre-existing pathologies and how these were reported and quantified. We found 800 requests from over 90 individuals. There was no consistency in the definition of cause of death or contributory cause of death across national bodies and in different bodies within the same nation. Nursing home providers, as well as medical practitioners, can assign a cause of death according to the Care Quality Commission. Post-mortem examinations were uncommon, the ONS did not incorporate their results in the summary of deaths by cause during the pandemic period. The meaning of the words "test" or "swab" was never clarified by any of the respondents. In care homes in England 1,304 out of 17,264 COVID-19 (7.6%, range 0% to 63%) mentioned COVID-19 in the absence of contributory or other factors in the death certificate, making it impossible to ascertain a chain of causality. The inconsistencies already noted hinder the ascertainment of the role of each factor leading to death and the quantification of the importance of infection. Some responses indicate that SARS-CoV-2 negative individuals or those whose death was not caused by COVID-19 were classified as "COVID-19 deaths". We found 14 different ways of attributing the causes of death mentioned by respondents. The overall lack of consistency has confused the public and likely led to erroneous conclusions. We are unable to separate the effects on deaths of SARS-CoV-2 from those of human interventions. A coherent process based on consistent definitions across the devolved nations is required. Furthermore, to enhance the accuracy of causation in pandemics a subset of deaths should be verified using autopsies with full medical documentation.

The two years of the COVID-19 pandemic restrictions were primarily underpinned by concerns over rising cases, hospital admissions and deaths. Deaths remain the most troubling as a marker of the severity of the disease. However, throughout the pandemic, it has not been possible to determine who has died 'from' or 'with' COVID.

To answer this vital question, data availability has not been the problem -several platforms have reported global daily updates on the latest number of deaths. However, there has been a scarcity of critical thinking about what individuals labelled as 'COVID deaths' died from. For example, we have documented that some studies attributed mortality in nursing homes in the absence of testing or a medical diagnosis. 1 The authors inferred the cause of death from the period in which the deaths occurred.

It is often problematic for clinicians to determine the direct cause of death or the underlying cause: none of the publicised daily death tolls differentiates between the two. The true impact of the SARS-CoV-2 virus is difficult to determine if we cannot interpret the actual cause of death. Also, misunderstanding can easily be misinterpreted and lead to underestimations or misleading statements about the effect on deaths. Attributable deaths are an important yardstick to monitor the trajectory of a pandemic.

Accounting for those that died from COVID-19 related illness is complex. 2 We are aware of several methods that label a death as causative, leading to significant differences in the reported number of COVID-19 deaths. For example, The Office for National Statistics' weekly death counts in England and Wales are based on mentions on the death certificate. Daily figures published by the Department of Health and Social Care differ because they report deaths occurring within 28 days of a positive test. 3 At the beginning of the pandemic, Public Health England linked data on positive cases to the NHS central register of patients who died. This definition meant that a patient who tested positive would be counted as a COVID death even if they were run over by a bus several months later. This method over-estimated the COVID-associated deaths. To overcome this limitation, the UK daily COVID death counts on the COVID-19 dashboard were changed to report deaths within 28 days of a first positive laboratory-confirmed test.

The ONS method establishes the cause of death through death certificates. However, anyone who has completed a death certificate will acknowledge their constraints and the potential inaccuracies for assigning causation.

Some commentators consider inaccuracies are not a problem in the certification process. 4 However, a review of their accuracy in an intensive care unit showed up to 10% were completed to a poor standard, and just over a half to a minimum standard. 5 A further study showed 82% of death certificates contained one or more errors, 6 and a substantial discrepancy has been shown between the diagnoses on death certificates compared with autopsy. 7 For example, certificates from 433 autopsied hospital patients in Iceland matched against post-mortem examinations showed significant discrepancies in 50% of patients and incorrectly stated the immediate cause of death in 25%. 8 This problem in the assignment of the cause of death has not escaped the notice of the public who, through the Freedom of Information Act ("FOIA"), have made requests to substantiate causation of those patients labelled as COVID deaths. A clear understanding of the different methods for defining causation is vital before analysing the substantial amount of data collected during the COVID-19 pandemic.

The FOIA 2000 allows for a public "right of access" to information held by public authorities in the UK. The website WhatDoTheyKnow facilitates FOI requests by forwarding requests to the appropriate authority and publishing the subsequent responses. We used a similar approach in a previous report in the role of PCR Testing in the UK During the SARS-CoV-2 pandemic. 9 We similarly collated FOI responses to preliminary understand the current knowledge of the causes of COVID-19 deaths in the UK -how such causation is assigned amongst public bodies and to what extent definitions differ depending on the organisation. We also provide an interpretation of the various FOI responses and recommendations to improve the understanding of the cause of COVID-19 deaths.

We set out to analyse public authorities' understanding of how deaths from SARS-CoV-2 are measured in the UK by accessing Freedom of Information requests posed in 2020-21. By public authorities, we mean NHS Health Trusts, laboratories, and government agencies such as Public Health England and the Department of Health and Social Care. We searched WhatDoTheyKnow using the terms "covid and death" to understand the ascertainment and cause of deaths in the UK. We found 800 requests, from over 90 individuals.

We searched through FOI requests using a cascade approach, keeping only those that gave informative results. We excluded those requests to bodies that cannot provide an answer (e.g. Councils) and those dealing with the effects of vaccines. We recorded the example and the respondent type and excluded any requests on causes of death other than COVID-19 and explanatory requests based on baseline characteristics such as age, place of death, etc. We also excluded questions posed to bodies such as councils with no statutory duty to collect death data. We concentrated on 3 periods of the pandemic assessing FOI answers provided in the spring and late 2020 and 2021, and we grouped questions into themes addressing the definitions and causes of death relevant to the pandemic.

Any emphasis of quoted sections is ours.

The ONS provides a weekly return 10 on the number of deaths based on Medical Certificates of Cause of Death (MCCD). When someone dies, a medical practitioner writes the medical certificate based on the cause of death, which is then recorded at a local authority registration office. This information is then sent electronically to the ONS to produce statistics about causes of death. ONS here clearly indicates that "due to", is either 1a (the only condition) or 1b or c (the events trigger) in the death certificate which is used "to produce statistics about causes of death".

In a response dated 23 Dec 2020 ONS state:

'We use the term 'Due to COVID-19' when referring to deaths where COVID-19 was recorded as the underlying cause of death. We use the term 'involving COVID-19' when referring to deaths that had the illness mentioned anywhere on the death certificate, whether as an underlying cause or not.' This is not the same wording as that of the previous answer as the explanation of underlying and only are missing, as the cause that started the train of events leading to death. In this version of the definition of "Directly leading to death" (1a in the death certificate) is missing. This is strange given that ONS has a dedicated FOI answering apparatus and one would expect identical answers to identical questions.

On another response on 28 Feb 2022 ONS state:

'We use the term "due to" a cause of death (e.g. when referring only to deaths with that underlying cause of death. We use the term "involving" when referring to all deaths that had the cause mentioned anywhere on the death certificate, whether as an underlying cause or not.

This version is missing the immediate cause of death that could occur if you used the 21 Sept 2021 definition.

NHS England and NHS Improvement publish the number of patients who died in hospital and tested positive for COVID-19, whether they had a pre-existing condition or whether COVID-19 was mentioned on the death certificate. This can be found in the weekly file on their website. 11 These figures don't state whether or not COVID-19 was the single cause of death. All deaths are recorded against the date of death rather than the date they were announced. Figures CC-BY-ND 4.0 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 April 29, 2022. ; https://doi.org/10.1101/2022.04.28.22274344 doi: medRxiv preprint

Facsimile of a death certificate. Image credit: https://geekymedics.com/certification-death-uk-osce-guide/.

'Part 1. The immediate, direct cause of death is reported on the certificate. Then the medical practitioner should go back through the sequence of events or conditions that led to death on subsequent lines until reaching the one that started the fatal sequence. The condition on the lowest completed line of part I will have caused all of the conditions on the lines above it. This initiating condition, on the lowest line of part I will usually be selected as the underlying cause of death, following the ICD coding rules. The WHO defines the underlying cause of death as "a) the disease or injury which initiated the train of morbid events leading directly to death, or b) the circumstances of the accident or violence which produced the fatal injury". From a public health point of view, preventing this first disease or injury will result in the greatest health gain.

Other diseases, conditions, or events that contributed to the death, but were not part of the direct sequence, are recorded in part two of the certificate. The conditions mentioned in part two must be known or suspected to have contributed to the death, not merely be other conditions that were present at the time.'

NHS England records the number of patient deaths, therefore, includes any mention on the death certificate, whether as an underlying cause or not, and can consist of those who did not test positive for SARS-CoV-2.

The UKHSA collates reports from multiple sources to provide a daily number of deaths in those with a positive PCR SARS-CoV-2 test or rapid lateral flow test in England, regardless of where they died.

. CC-BY-ND 4.0 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 April 29, 2022. ; https://doi.org/10.1101/2022.04.28.22274344 doi: medRxiv preprint 'There are two definitions of death in a person with COVID-19 in England, one broader measure and one measure reflecting current trends:

2) A death in a person with a positive SARS-CoV-2 test and died within (equal to or less than) 28 days of the first positive specimen date of the most recent episode of infection.'

The Care Quality Commission (CQC) publishes statistics on deaths involving COVID-19 in care homes in England.

The Northern Ireland Statistics and Research Agency. Cause of Death Information in Northern Ireland: A user guide Definition: The Cause of Death section of the MCCD is set out in two parts as shown below.

'Part 1 contains the disease or condition directly leading to death and any antecedent causes (i.e. conditions that gave rise to the condition leading to death) with the underlying cause stated last.

. CC-BY-ND 4.0 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 April 29, 2022. ; Part 2 contains significant conditions that contributed but were not related to the diseases causing the death. 

There is no consistency in the definition of cause of death or contributory cause of death across national bodies and in different bodies within the same nation.

Medical practitioners and nursing home providers can both assign a cause of death. However, when COVID-19 ONLY is mentioned on the death certificate it is impossible to ascertain a chain of causality. It is also very unlikely that infections with SARS-CoV-2 per se could cause death, in the absence of contributory factors, comorbidities, and a pathology directly deriving from the infection leading to death (such as respiratory failure due to bilateral bronchopneumonia).

Some statements are not clear, and the lack of consistency has confused the public and led to erroneous conclusions. For example, the underlying cause of death is essential for establishing the chain of events. However, there is no clear unified method to establish this vital chain of causation. Further confusion is added because guidance can contradict itself when updates occur. It also seems to be possible to report a death as COVID-19 despite COVID-19 not being the cause of death.

We found 14 different ways to express the cause of death (highlighted above in bold text). The lack of clarity affects the ascertainment and establishment of the causes of COVID-19 deaths and affects the responses reported in sections 2, 3 and 4 of this report.

Background Several FOIs on the ascertainment of COVID-19 deaths asked about the use of post-mortems (an examination of a body after death), the use of specific tests, or the reporting of different sections on the death certificate.

. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 'I am writing to confirm that the Office for National Statistics has now completed its search for the information which you requested and the response can be found here' 'Some of the analysis requested is available through different releases. On 16 April 2020, we released an Analysis of deaths involving COVID19, which stated that, including registrations up to 6 April, there were 3,912 deaths involving the coronavirus (COVID-19) that occurred in March 2020 in England and Wales; of these, 3,372 (86%) had COVID-19 assigned as the underlying cause of death. This analysis will be updated on 15 May to include deaths occurring in April.

The number of deaths involving COVID-19 by where it is mentioned on the death certificate is available to commission as a bespoke analysis. Such services would fall outside of the Freedom of Information regime and would be subject to legal frameworks, disclosure control, resources and agreement of costs, where appropriate.' 2d. Please provide the criteria by which death by COVID-19 is distinguished from death by influenza, pneumonia, asthma, and other respiratory illnesses.

'Most clinicians would ascribe a death to COVID if a) there was a positive swab in the lead up to the death, or b) the clinical pattern of disease is such that, despite a negative swab, COVID was the most likely cause'

The dearth of post-mortem examinations, added to the lack of consistency in the definition of what is meant by a COVID-19 death and inconsistent testing, enhances the difficulty to ascertain the direct cause of death.

Deaths are required by law to be registered within 5 days of their occurrence unless there is to be a coroner's post mortem or an inquest. If no doctor attended the deceased within 28 days of death or the deceased was not seen after death by a doctor, the MCCD can still be completed. However, the doctor will be obliged to refer the death to the coroner before it can be registered. The coroner may instruct the registrar to accept the certifying doctor's MCCD for registration.

Where a cause of death cannot be ascertained, the death cannot be certified, and the doctor should refer the death directly to the coroner with any supporting information. This information will be used for mortality statistics, but the death will be legally "uncertified" if the coroner does not investigate and make a determination as to the cause of death.

Strictly speaking, the law requires that an MCCD should be completed even when a death has been referred to the coroner. In practice, if the coroner has decided to order a post-mortem examination and/or to hold an inquest, he may tell a doctor not to complete the MCCD. However, the coroner can only legally certify the cause of death if he has investigated it through autopsy, inquest or both. This means that, if the coroner decides not to investigate, the registrar will need to obtain an MCCD from a doctor who attended the deceased before the death can be registered.

The 'sole cause of death' FOIs relates to the underlying health and attempts to understand those that died with no other related health conditions. Underlying health conditions also includes FOIs on the number of do not resuscitate orders that were in place at the time of death.

related deaths. Whilst the number of infections appear to be rising, the number of COVID-19 deaths from July onwards are very small and pose a disclosure risk, making it possible to identify individual records.' 3b. I would like to know how many COVID 19 ONLY deaths have been registered in 2020, in Scotland. NOT COVID 19 related deaths.

Response No113. National Records of Scotland, Jan 21:

'In our weekly Covid-19 statistics publication, we provide figures on both the number of deaths where Covid-19 was mentioned on the death certificate (either as the underlying cause or as a contributory factor) and on deaths where Covid-19 was the underlying cause of death. For this reason, the information you require for deaths where Covid-19 was the underlying cause of death is available from 'Furthermore, patients may have 'solely' died of Covid-19 in terms of what has been determined as their cause of death, however, the Trust does not undertake post mortems and therefore we are unable to say whether they had underlying, undiagnosed co-morbidities which contributed to their death. The Trust is following national government guidance which states that anyone who has died within 28 days of a positive Covid-19 test is classified as dying with Covid-19 and included in our figures which are returned to the national team. The Government/NHS England may have further information that is not kept or recorded at a Trust Level.'

'for the specified time period, there have been 338 deaths where COVID-19 is mentioned on the death certificate, with 292 where COVID-19 is cited as the primary cause of death. Of these deaths, there have been 11 where no other conditions are mentioned apart from COVID-19.'

The weekly data published on 10 December 2020 showed 43,537 deaths of patients who have died in hospitals in England and have tested positive for COVID-19. The daily total deaths publication on the same day showed (separately) a further 2,092 deaths where a positive test result for COVID-19 was not received but COVID-19 is mentioned on their death certificate.

The weekly data set shows on Table 3 that of the 43,537 patients that had tested positive for COVID-19, 41,683 patients had one or more pre-existing conditions and 1,854 patients had no pre-existing condition. We do not publish data on pre-existing conditions for the patients with no positive test result for COVID-19.' 'A review has been undertaken for all deaths where COVID was recorded between July to September 2020. A total of 43 COVID related deaths occurred at the Trust in this time period. However, 33 records were available for review at the time of this request. Of the 33 records reviewed, 30 had a DNACPR form in place at the time of death.'

'How many of these 515 patients had DNACPR orders (or similar decisions) on their records? 440'

'The Health Board can confirm that we do record do not attempt resuscitation (DNAR) status. However, in order to gather this information, we would have to carry out a specific exercise of reviewing all the Medical Certificate Cause of Death (MCCD) books for each patient in each hospital to establish whether COVID-19 was in present and then check DNAR against our records. We have established that to comply with your request would exceed the appropriate costs limit under Section 12 of the Freedom of Information Act 2000.'

Identification of the main cause of death is vital to assessing the severity of the pandemic and permitting meaningful comparisons of death data. The already noted vagueness of the meaning of testing, as well as inconsistencies in definition use, undermine assessments of the role of infection with SARS-CoV-2 and cause of death in relation to other underlying health conditions. COVID-19 as the sole cause of death is uncommon in frail home healthcare individuals in Sweden; however, it isn't uncommon on death certificates in the UK (see NISRA).

ONS produce a quarterly report on the pre-existing conditions of people who died due to COVID-19, in England and Wales (see here). 12 In December 2021, ONS published an FOI response asking about deaths from COVID-19 with no other underlying causes, and a blog post in January 2022 that reported claims that only 17,000 people died from COVID-19 was 'highly misleading'. 13 Whether or not the deceased had a do-not-resuscitate order (DNR) is not recorded on the death certificate. In the trusts reporting this number, it was high. For example, in the North-West University Healthcare NHS Trust 89% of 623 deceased patients positive for COVID-19 had DNACPR in place, and in the King's College Hospital NHS Foundation Trust the proportion was 85%. However, in the Barts Health NHS Trust, the proportion was lower at 32%. This variation warrants further investigation.

In an FOI response to deaths due to COVID-19 with no pre-existing conditions by county/city, the ONS produced a table of the counts of deaths registered in England and Wales by the Local Authority of residence and place of death. 14 The results for 2020 report that in care homes in England 1,304 out of 17,264 COVID-19 deaths were registered with no pre-existing conditions (unweighted average 7.6%, range 0% to 63%). See https://datawrapper.dwcdn.net/5Bo8A/4/ for the analysis by local authorities.

The lack of pre-existing conditions in care home residents (on average one in thirteen and in some homes more than half of the COVID-19 deaths had no pre-existing conditions at the time of death) further adds to the uncertainties over the assignment of causation. Care home residents have multiple comorbidities that contribute to their vulnerability, need for additional care, and usually their death.

The assignment of death with no pre-existing conditions, therefore, seems implausible and may follow on from the vagueness of the guidelines for care home certification provided by the CQC, whereby death involving COVID-19 can be based on the statement of the care home provider.

Background FOI questions in this section aim to determine who is susceptible to dying because of pre-existing conditions and therefore died 'with' COVID versus those that died as a result -"from" COVID. 'I can confirm that the department holds information that you have asked for. The information is exempt under section 21 of the FOI Act because it is reasonably accessible to you, and I am pleased to inform you that you can access it via the following link' Response No34. Liverpool University Hospitals NHS Foundation Trust, Jan 21:

'Please be advised in relation to your questions above we cannot provide data relating to the "sole diagnosis" of a patient. Neither can we provide the "direct cause" of a patient's death as we do not have access to patient's death certificates.'

'Any COVID death 575 (These deaths are people who have died in hospital and have had a positive). Deaths "from" COVID Covid test -we can not determine whether or not Covid was the cause of death, i.e. death 'from' Covid.'

'NISRA, along with the other UK jurisdictions recognised the need for a mentions based approach to Covid-19 deaths reporting given the coverage limitations of the Department of Health-related, positive test based statistics which at the start of the pandemic largely excluded deaths outside of the hospital; and also given that statistics based on Covid-19 as the underlying cause of death are not available in NI until some time after each quarter ends to allow for ICD-10 coding to take place. Figures on this basis are currently available up to the end of September 2020 at; https://www.nisra.gov.uk/publications/registrar-general-quarterly-tables-2020' 4d. Evidence that the 'COVID-19' related deaths are in fact deaths from the virus solely and not just deaths recorded with the smallest hint of being around COVID-19, a cough, temperature, etc.

Response No49. Welsh Government, Jan 21:

'The Office for National Statistics (ONS) collect mortality data for England and Wales based on information collected on the death certificate. The ONS publish the number of reported deaths by method of certification and registration (including the number of post mortems) for England and Wales in their user guide, but this is currently unavailable for the pandemic period.'

. CC-BY-ND 4.0 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) (f) Of the 171 deaths recorded at Hinchingbrooke Hospital between 20/03/2020-27/01/2021, we cannot accurately state how many patients died with COVID-19 as the underlying cause of death. Please note that the number of deaths reported includes patients whose most recent swab was positive prior to their death, as well as those patients whose most recent swab, was negative but COVID-19 was listed on their death certificate.

The inconsistencies already noted hinder the ascertainment of the role of each factor leading to death and the quantification of the importance of infection. Some responses indicate that SARS-CoV-2 negative individuals were classified as COVID-19 deaths.

The Victorian physicians John Snow and his aide Duncan laid the epidemiological groundwork in the ascertainment of death during an ugly cholera outbreak in central London. Snow and Duncan enquired door to door as to whether deaths had occurred, when and where.

They matched death certificates with the results of their enquiries as they realised that ill people could have moved away from Soho and died elsewhere. By this systematic approach, Snow was able to group like with like (all death caused by cholera), narrow down a possible source (contaminated water) and mode of transmission (waterborne). However, he was also able to define the severity of the outbreak. Although cholera has very characteristic manifestations its transmission is not 100% linear and Snow did not have the benefit of an identified agent, electron microscopy or a modern system of testing. The lack of rigorous approach to cause of death definition and attribution are possibly due to the fog of the pandemic but also point to lack of preparation and lack of control and supervision of the events from 2020.

. CC-BY-ND 4.0 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 April 29, 2022. ;

Observed differences, whether between agencies or between UK countries are affected by the definitions used. Statistical comparisons with other countries should also be treated with caution as they will also be substantially affected by the definitions chosen.

The FOI question of 'with' or 'for' COVID-19 death remains a central issue to understand the impact of the pandemic. This question cannot be answered with any certainty through the sole use of death certificates, particularly given their inherent limitations.

In hospitals, junior doctors can often be tasked with signing medical certificates of cause of death. 15 Normally, the doctor who verifies the death, or cared for the patients during their last 14 days is eligible to sign the MCCD. In the pandemic, the duration was extended to 28 days.

A death certificate is based on the probability the deceased expired of the causes on the death certificate. Prioritising the condition leading to death can prove difficult: it is affected by the experience of the clinician, their prior knowledge of the patient. In the presence of several comorbidities that may compete and co-exist with each other an individual may easily appear to die with rather than of their disease. 16 The condition listed on the bottom line of Part I of the certificate is the underlying cause of death. The cause of death is based on medical opinion, which may change as more information becomes available.

In the midst of a pandemic, the assignment of causation will be affected by availability and representativeness heuristics. Heuristics are mental shortcuts that aid problem-solving and judgments. However, they can often lead to erroneous conclusions. 17 The availability heuristic also referred to as the availability bias is 'a distortion that arises from the use of information which is most readily available, rather than that which is necessarily most representative.' 18 Determining the underlying cause of COVID-19 deaths remains an important area of future research. The high rates of Do Not Resuscitate orders in some Trusts for COVID-19 deaths and the variation in rates requires further investigation. Were these orders already in place, as may be the case for those with a terminal disease, or were they instigated during admission? These vital questions require answering -to determine the impact of the virus, particularly in those that are the most vulnerable and most likely to die.

The current FOIA prevents vital responses due to the limitations on the time required to assess some of the answers. We consider an ombudsman should override the time allowance when the answer is in the public interest and will facilitate public understanding.

The assignment of the causation of deaths can be done using several methods. The anonymised death certificate data set with the conditions and their specific placement on the death certificate would permit independent analysis of the assignment of the underlying cause of death.

the Autopsy Committee of the College of American Pathologists. Cause of death statements and certification of natural and unnatural deaths (Manual)

Judgment Under Uncertainty

Catalogue of Bias Collaboration

Effects of COVID-19 in Care Homes -A Mixed-Method Review. Collateral Global

Behind the headlines: Counting COVID-19 deaths

COVID-19) latest insights: Deaths

Mistruths and misunderstandings about COVID-19 death numbers

A Review of the Accuracy of Death Certification on the Intensive Care Unit and the Proposed Reforms to the Coroner's System

Death certification errors at an academic institution

Analysis of the sensitivity of death certificates in 440 hospital deaths: a comparison with necropsy findings

The accuracy of death certificates. Implications for health statistics

PCR Testing in the UK During the SARS-CoV-2 Pandemic -Evidence From FOI Requests

Pre-existing conditions of people who died due to COVID-19

To say only 17,000 people have died from COVID-19 is highly misleading

Deaths due to COVID-19 with no pre-existing conditions by county

the Autopsy Committee of the College of American Pathologists. Cause of death statements and certification of natural and unnatural deaths (Manual)

Judgment Under Uncertainty

Catalogue of Bias Collaboration

Confirmation of the cause of death could be established by the use of post-mortem evidence in the presence of the full documentation of the medical history of the deceased. A subset of deaths could be verified using autopsies with full medical documentation. However, the UK guidelines and the FOI answers we reviewed suggest autopsies were an uncommon practice.An FOI response by the Welsh Government stated 'the ONS publish the number of reported deaths by method of certification and registration (including the number of post-mortems) for England and Wales in their user guide, but this is currently unavailable for the pandemic period.' An FOI request to the ONS asked about the number of autopsies carried out. The response stated, the ONS 'do not hold analysis on the number of post-mortems completed.'The ONS can create an analysis on deaths involving post-mortems in 2020 and 2021 subject to legal frameworks, disclosure controls, resources, and agreement of costs. The number of post-mortems carried out in the UK is currently unclear.As we go down the causality pyramid (see figure) the uncertainty over the true cause of death increases.Analysis of a random sample of death certificates assigned as 'from', 'with', 'underlying', 'due to', or 'involving' COVID-19 and exploration of their underlying causes with independent verification of the assignment of causation using full medical documentation would reduce uncertainty. One level down is based on reporting of a positive test (usually a PCR with no further detail). At the bottom of the pyramid is Daily Death Reporting based on knowledge of the pandemic on general knowledge of viral circulation and the clinical picture.