key: cord-0976973-5vpnmw3f
authors: Zaki, N.; Mohamed, E. A.; Ibrahim, S.; Khan, G.
title: The influence of comorbidity on the severity of COVID-19 disease: systematic review and analysis
date: 2020-06-20
journal: nan
DOI: 10.1101/2020.06.18.20134478
sha: 21d991544eaa4c57ec10fe7cd64d7e1c08b210d7
doc_id: 976973
cord_uid: 5vpnmw3f

A novel form of coronavirus disease (SARS-CoV-2) has spread rapidly across the world. This disease, originating in Wuhan, China, has become a global pandemic. What risk factors influence the severity of the disease is of considerable importance. This research is intended to offer a systematic review/meta-analysis for assessing how common clinical conditions and comorbidities correlate with COVID-19. The meta-analysis incorporated seven studies covering 4101 COVID-19 patients from Chinese hospitals who had their diagnosis confirmed through laboratory testing. The findings demonstrate that the most common comorbidities with the disease were COPD (2.53%, OR 3.24 [95% CI: 1.99-4.45], p< 0.0006), cardiovascular disease (10.76%, OR 2.89 [95% CI: 1.90-4.40], p <0.0001), coronary heart disease (5.52%, OR 2.97 [95% CI: 1.99-4.45], p <0.0001), diabetes (11.34%, OR 2.27, [95% CI: 1.46-3.53], p = 0.0003), and hypertension (22.07%, OR 2.43 [95% CI: 1.71-3.45], p <0.0001). No significant associations were found for disease severity with the comorbidities of kidney disease, liver disease, or cancer. The most frequently exhibited clinical symptoms were fever (74.52%, OR 1.37, 95% CI: 1.01-1.86, p = 0.04), cough (62.15%, OR 1.25, 95% CI: 0.97-1.60, p = 0.0823), myalgia/fatigue (38.77%, OR 1.31, 95% CI: 1.11-1.55, p = 0.0018), dyspnea (33.9%, OR 3.61, 95% CI: 2.57-5.06, p = <0.0001), respiratory failure/ARDS (20.6%, OR 11.46, 95% CI: 3.24-40.56, p = 0.0002), diarrhea (11.21%) and chest tightness/pain (16.82%, OR 2.17, 95% CI: 1.40-3.36, p = 0.0006). Meta-analysis also revealed that neither the duration of the incubation period nor current smoking status associated with disease severity.

The 2019/2020 emergence of the novel COVID-19 disease and its swift global expansion represents a health emergency for all of humanity. This novel virus, which causes severe acute respiratory disease, is believed to be a member of the same family of coronaviruses as the Severe Acute Respiratory Syndrome coronavirus (SARS-CoV) and the Middle East Respiratory Syndrome coronavirus (MERS-CoV) [1] . Unprecedented efforts are currently being made to identify the risk factors that increase the severity of the disease. The World Health Organisation (WHO) has suggested that patients with certain medical conditions and elderly people are at greatest risk of developing more severe disease [2] . As of June 13, 2020, over 7.6 million people have been infected and 425,800 have died from the disease. The majority of deaths are thought to have associations with the existence of one or more comorbidity. In general, it is believed that patients with compromised immune system are especially at high risk [3] . A number of researchers have examined the clinical/epidemiological characteristics of patients suffering COVID-19, but there has been insufficient investigation regarding mortality risk factors [4] . The identification of the principal risk factors and employing appropriate clinical interventions to mitigate them could save numerous lives. Numerous studies indicate that kidney disease, COPD, coronary heart disease, liver disease, cardiovascular disease, diabetes mellitus, and hypertension are amongst the most important risk factors for COVID-109. This paper will undertake a systemic review of the literature to illuminate the way in which the impact of the COVID-19 virus can be made more severe by the existence of such comorbidities and extant conditions.

This research looked for all relevant published articles between the dates of December 1, 2019, and May 20, 2020, i.e. those that were related to comorbidity and its influence on the severity of COVID-19. Papers were excluded if they were related to the adult patients, if they were not written in English, if they had not been peer reviewed, or if they were not original (e.g. reviews, editorials, letters, commentaries, or duplications). Papers mainly focused on severe vs nonsevere groups of patients were considered. Severe group are those who develop severe illnesses. These cases are more likely to result in ICU admission or death.

A search was undertaken on the COVID-19 Open Research Dataset (CORD- 19) , which is regularly updated [5] . This dataset contains more than 128,000 scholarly articles; 59,000 of these articles have full text that relates to coronavirus, SARS-CoV-2, and COVID-19. A search engine running over the BM25 search index was created so that all articles in this dataset could be screened [6]. The BM25 retrieval function creates a document ranking for a dataset on the basis of whether or not searched terms appear in the documents, no matter how proximate they may be to each other. Indexing of the papers occurs through simple application of preprocessing functions that clean and tokenize abstracts. Having indexed all documents, document vectors were created through loading optimized cached JSON tokens and subsequent application of a document similarity index founded on Annoy [7] . Annoy is a C++ library that has Python bindings for searching for documents within a space that make a close match to a specific query. This is an efficient and simple process as it creates substantial read-only file-based data structures used with memory mapping allowing several processes to work with identical data. More detail on this implantation can be found at HTTP://github.com/dgunning/cord19.

Articles were then filtered employing questions and keywords such as "Effects of diabetes on COVID/normal coronavirus/SARS-CoV-2/nCoV/COVID-19 disease severity?" This search was undertaken for every comorbidity under consideration in this research ( Figure 1 ). CC-BY-NC 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 June 20, 2020. . https://doi.org/10.1101/2020.06. 18.20134478 doi: medRxiv preprint Additionally, an independent search was undertaken by two researchers (NZ and EA), who systematically examined the PubMed, Europe PMC, and Google Scholar databases. These searches obtained information regarding name of author, year of publication, publication name, age, gender, severity/non-severity of the disease, and the presence of any clinical symptoms, i.e. chest tightness/pain, respiratory failure/ARDS, dyspnoea, myalgia/fatigue, diarrhea, cough, or fever. Where either vital or non-vital information was absent, requirement for admission to ICU was employed to indicate how severe the disease was.

The articles that remained underwent independent screening once more by the two researchers (NZ and EA). Any disagreement was settled either by consensus or by the casting vote of a third researcher (SI). Research was completed on May 20 th , 2020; the stages of this systemic literature search are illustrated in Figure 2 .

Meta-analysis was undertaken employing a python library "meta". Odds ratios (ORs) were employed for describing the risk factors of various comorbidities for patients with severe disease compared to those with mild disease. Due to internal and external heterogeneities, random effect modeling was employed for estimation of average effects and their precision, offering a more cautious estimate regarding the 95% confidence intervals (CIs). Statistical heterogeneity was assessed using the I 2 statistic.

Machine learning techniques based on regression, e.g. support vector machine (SVM) [8] , linear regression, multi-perceptron [9] , random forest [10] , and attribute selection techniques such as Wrapper Subset Evaluator [11] , Correlation Ranking Filter, and ReliefF Ranking Filter [12] as implemented in WEKA [13] were employed for determination of how useful and significant different elements were in terms of the prediction of levels of severe instances of COVID-19. Consideration was only given to the conditions kidney disease, coronary heart disease, COPD, liver disease, cardiovascular disease, cancer, diabetes mellitus, and hypertension. Clinical symptoms that were added to consideration as independent attributes were diarrhea, chest tightness/pain, respiratory failure, dyspnoea, fatigue, coughing, and fever, incubation period, current smoking status, and patient gender. 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 20, 2020. . https://doi.org/10.1101/2020.06.18.20134478 doi: medRxiv preprint Results 303 articles were found through the search of Europe PMC, PubMed, and the COVID-19 database; 110 of these were excised as they were duplicates. In total 74 text articles were examined to see if they were eligible once 125 articles had been removed due to the fact that they were not Englishlanguage articles, did not have a full text, were duplicates, were editorials, were letters, were reviews, were irrelevant, or not peer-reviewed. Ultimately, the meta-analysis group comprised 12 articles that met all quantitative and qualitative standards.

As Table 1 shows, the median age for severe cases was 58.8 years, and for non-severe cases 48.5 years; 53.55% of patients were male. The meta-analysis demonstrated that the highest levels of comorbidity were with hypertension (22.07%), diabetes (11.34%), cardiovascular disease (10.76%), liver disease (6.31%), coronary heart disease (5.52%), kidney disease (3.82%), and COPD (2.53%). In terms of clinical symptoms, the most common were fever (74.52%), cough (62.15%), myalgia/fatigue (38.77%), dyspnoea (33.9%) respiratory failure/ARDS (20.6%), diarrhea (11.21%) and chest tightness/pain (16.82%). Meta-analysis indicated that neither length of incubation period nor current smoking status had any associations with disease severity. 32 ], p = 0.0006; I2: 0%). No significant correlation was found between disease severity and kidney or liver disease. The I 2 test for heterogeneity ranged between nought percent and 62%, indicating a fair level of statistical heterogeneity. With COPD and coronary heart disease, I 2 was equal to 0%, meaning no heterogeneity can be found, making this research adequately homogenous. . CC-BY-NC 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 June 20, 2020. . CC-BY-NC 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 June 20, 2020. Fig. 3 . Forest-plots indicate that the comorbidities diabetes mellitus, hypertension, coronary heart disease, and cardiovascular disease had associations with composite poor outcomes, severe COVID-19, and ICU care requirements. This is not true of cancer, liver disease, and kidney disease patients.

. CC-BY-NC 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 June 20, 2020. For evaluation of publication bias in the article in the meta-analysis, we employed the R metapackage for visualization of the funnel plots. The outcomes of publication bias are shown in Figs. 5 and 6.

. CC-BY-NC 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 June 20, 2020. . https://doi.org/10.1101/2020.06.18.20134478 doi: medRxiv preprint

Coronary heart disease Cardiovascular disease This analysis employed the seven-machine learning/feature selection techniques outlined in the methodology section indicating that diabetes was listed in the top 10 crucial attributes for 6 techniques, and then heart (5/7), hypertension (4/7), and COPD (4/7). Clinical conditions like cough (6/7), fever (5/7), chest pain (4/7), and fatigue (4/7) are similarly listed in the top 10 important attributes.

. CC-BY-NC 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 June 20, 2020. . https://doi.org/10.1101/2020.06.18.20134478 doi: medRxiv preprint

This research has offered a comprehensive vision review/meta-analysis regarding the influence of comorbidities, e.g. diabetes and hypertension, on the severity of COVID-19. The meta-analysis employed data gathered from 12 studies covering the 4101 patients in Chinese hospitals with a COVID-19 diagnosis confirmed by laboratory. Patients from the severe group were all higher age (median age 58.8) than the patients in the non-severe group (median age 48.5). There were more males than females in the study (2196 males, 1702 females). Two of the studies (21 and 24) indicated a significant correlation between gender and disease severity. Every study examined agreed that being female has no significant correlation with disease severity (OR 0.74; 95% CI: 0.61-0.89); rather, as with SARS-CoV and MERS-CoV patients [26] , there is simply a greater likelihood of infection for males as they generally have higher exposure levels.

Meta-analysis of comorbidity found that diabetes mellitus (11.34%) and hypertension (22.07%) are two of the most prevalent conditions in this patient cohort; this almost exactly parallels the levels of hypertension and diabetes amongst the population of China as a whole (hypertension 23.2% [27] , diabetes mellitus 10.9% [28] [29] found that human pathogenic coronaviruses, e.g. SARS-CoV and SARS-CoV-2 achieve binding with target cells using Angiotensin-Converting Enzyme 2 (ACE2), which originates from epithelial cells in the blood vessels, kidneys, intestines, and lungs [30] . ACE2 levels are considerably higher for patients suffering type I or type II diabetes; because of this they are given treatment with ACE inhibitors and/or Angiotensin II type-1 Receptor Blockers (ARBs). These medications are also used in the treatment of hypertension, and this causes an upregulation of ACE2 [31] . ACE2 may also be increased using either ibuprofen or thiazolidinediones. Higher levels of ACE2 would therefore promote COVID-19 infection. Thus we agree with the hypothesis that treating hypertension or diabetes with ACE2-stimulators could predispose patients to more severe COVID-19. We thus concur with Fang et al [29] that patients with diabetes, hypertension, or cardiac disease being medicated with ACE2-increasing medication must be regarded as having a greater risk of developing severe COVID-19 and so they should be closely monitored for the effects of self-medication.

Zhao Q et al, in a recent review [32] found that extant COPD makes it four times more likely that severe COVID-19 will develop. These authors also examine the influence of smoking in COVID-19 severity, coming to the conclusion that there is no significant correlation. This accords with the findings of this meta-analysis. There has always been an assumption that smoking may have an association with less favorable disease outcomes, as there is copious research regarding the detrimental effects of smoking on lung health and the causal correlation between it and many different forms of respiratory disease [33] . Furthermore, smoking is injurious to the immune system and the body's ability to respond to infection, which means that smokers are more susceptible to contracting infections [34] . This accords with this paper's meta-analysis, although, as Lippi G and Henry BM [35] indicate, more research will be necessary as further evidence becomes available; from the limited data available at present, and while acknowledging that the findings above have not been adjusted for other elements that may influence the disease pathway, it seems probable that smoking has a correlation with more severe undesirable outcomes in COVID-19 cases.

. CC-BY-NC 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 June 20, 2020. . https://doi.org/10.1101/2020.06.18.20134478 doi: medRxiv preprint While patients exhibiting abnormal liver results are regarded as being significantly more likely to develop severe pneumonia, and employing ritonavir or lopinavir makes patients four times more likely to suffer liver injury [36] , this meta-analysis did not find a significant correlation between liver disease and negative outcomes. This accords with recent research from [37) , which stated that while it is common for COVID-19 patients to exhibit abnormal liver function index, this is not a significant factor in COVID-19 patients, and it may not be associated with serious negative clinical outcomes.

The study by Guan W et al [19] was the only one of the 12 articles under review that found a significant correlation between kidney disease and COVID-19 severity; examining the articles in total, no significant correlation is found (p = 0.0476). The clinical symptoms that are most frequently exhibited are fever (74.52%), cough (62.15%), myalgia/fatigue (38.7%), dyspnoea (33.9%), respiratory failure/ARDS (20.6%), diarrhea (11.21%), and chest tightness/pain (16.82%). This accords with the majority of published research. In terms of heterogeneity, in the majority of cases zero or low heterogeneity was displayed by the I 2 , so there is relative homogeneity in the majority of cases examined for this research.

This meta-analysis has certain limitations that could be mitigated in future research: firstly, only patients in Chinese hospitals were included in the studies under review, and generalisability can only be assumed if research including other populations is incorporated; secondly, the presence of multiple comorbidities in single patients has not been considered (this is an element virtually every other review as overlooked also); finally, this meta-analysis does not encompass laboratory, radiographic, clinical, or demographic data.

The estimations of the COVID-19 incubation period: a systematic review of the literature

Coronavirus disease (COVID-19) Pandemic," World Health Organization

Cancer care in the time of COVID-19

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

CORD-19: The Covid-19 Open Research Dataset

Improvements to the SMO Algorithm for SVM Regression

Multilayer feedforward networks are universal approximators

Random Forests

Wrappers for feature subset selection

Practical Approach to Feature Selection

Online Appendix for

Analysis of factors associated with disease outcomes in hospitalized patients with 2019 novel coronavirus disease

Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China

Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus-Infected Pneumonia in Wuhan, China

Clinical characteristics and risk factors associated with COVID-19 disease severity in patients with cancer in Wuhan, China: a multicentre, retrospective, cohort study

Clinical characteristics of 140 patients infected by SARS-CoV-2 in Wuhan, China

Clinical characteristics of coronavirus disease 2019 in China

Risk factors for disease severity, unimprovement, and mortality in COVID-19 patients in Wuhan, China

Risk factors for severity and mortality in adult COVID-19 inpatients in Wuhan

Clinical characteristics of 161 cases of coronavirus disease 2019 (COVID-19) in Changsha

COVID-19 with Different Severities: A Multicenter Study of Clinical Features

Host susceptibility to severe COVID-19 and establishment of a host risk score: findings of 487 cases outside Wuhan

Clinical findings of patients with coronavirus disease 2019 in Jiangsu province, China: A retrospective, multi-center study

Sex-Based Differences in Susceptibility to Severe Acute Respiratory Syndrome Coronavirus Infection

Summary of the 2018 report on cardiovascular diseases in China

Clinical and biochemical indexes from 2019-nCoV infected patients linked to viral loads and lung injury

Are patients with hypertension and diabetes mellitus at increased risk for COVID-19 infection?

Receptor recognition by novel coronavirus from Wuhan: an analysis based on decadelong structural studies of SARS

The vasoprotective axes of the renin-angiotensin system: physiological relevance and therapeutic implications in cardiovascular, hypertensive and kidney diseases

The impact of COPD and smoking history on the severity of COVID-19 A systemic review and meta-analysis

Secular trends in smoking in relation to prevalent and incident smoking-related disease: A prospective population-based study

Are healthy smokers really healthy?

Active smoking is not associated with severity of coronavirus disease 2019 (COVID-19)

Liver impairment in COVID-19 patients: A retrospective analysis of 115 cases from a single centre in Wuhan city, China

The authors would like to thank Mr. Dwight Gunning for his kind assistance with the search engine.

The authors declare no conflict of interest.

NZ and EA performed the the literature scan and review, NZ performed the meta-analysis, all authors contributed to the paper writing, GK and SI validated the findings.