key: cord-0711638-hedk5ckn
authors: Singh, R.; Rathore, S. S.; Khan, H.; Karale, S.; Bhurwal, A.; Tekin, A.; Jain, N.; Mehra, I.; Anand, S.; Reddy, S.; Sidhu, G. S.; Panagopoulos, A.; Pattan, V.; Kashyap, R.; Bansal, V.
title: Association of Obesity with COVID-19 Severity and Mortality: A Systemic Review and Meta-Regression
date: 2021-05-10
journal: nan
DOI: 10.1101/2021.05.08.21256845
sha: 9e9b58ff119dc890d9d88f47823801041e0b46a1
doc_id: 711638
cord_uid: hedk5ckn

Objective: To estimate the association of obesity with severity (defined as use of invasive mechanical ventilation or intensive care unit admission) and all-cause mortality in coronavirus disease 2019 (COVID-19) patients. Patients and Methods: A systematic search was conducted from inception of COVID-19 pandemic through January 31st, 2021 for full-length articles focusing on the association of increased BMI/ Obesity and outcome in COVID-19 patients with help of various databases including Medline (PubMed), Embase, Science Web, and Cochrane Central Controlled Trials Registry. Preprint servers such as BioRxiv, MedRxiv, ChemRxiv, and SSRN were also scanned. Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines were used for study selection and data extraction. The severity in hospitalized COVID-19 patients, such as requirement of invasive mechanical ventilation and intensive care unit admission with high BMI/ Obesity was the chief outcome. While all-cause mortality in COVID-19 hospitalized patients with high BMI/ Obesity was the secondary outcome. Results: A total of 576,784 patients from 100 studies were included in this meta-analysis. Being obese was associated with increased risk of severe disease (RR=1.46, 95% CI 1.34-1.60, p<0.001, I2 = 92 %). Similarly, high mortality was observed in obese patients with COVID-19 disease (RR=1.12, 95% CI 1.06-1.19, p<0.001, I2 = 88%). In a multivariate meta-regression on severity outcome, the covariate of female gender, pulmonary disease, diabetes, older age, cardiovascular diseases, and hypertension was found to be significant and explained R2= 50% of the between-study heterogeneity for severity. Similarly, for mortality outcome, covariate of female gender, proportion of pulmonary disease, diabetes, hypertension, and cardiovascular diseases were significant, these covariates collectively explained R2=53% of the between-study variability for mortality. Conclusions: Our findings suggest that obesity is significantly associated with increased severity and higher mortality among COVID-19 patients. Therefore, the inclusion of obesity or its surrogate body mass index in prognostic scores and streamlining the management strategy and treatment guidelines to account for the impact of obesity in patient care management is recommended.

According to WHO, the prevalence of obesity has nearly tripled in the last four decades 22 amounting to 13% of the entire world's adult population 27 . This exponential rise in the obesity 23 rates in the midst of the pandemic is a cause for concern. The interplay between obesity and 24 diabetes mellitus, cardiovascular disease, stroke, dyslipidemia, influenza has been established for 25 a long time. The presence of these comorbid determinants has been related to increased 26 predisposition and severity of COVID-19 28-31 . Many studies have reported increased rates of 27 hospitalization, mechanical ventilation, and mortality in patients with higher BMI 32-36 . 28 To mitigate the impact of heightened morbidity and mortality associated with COVID-19 29 infection in patients with obesity, it is vital to be cognizant of the implications of increased BMI 30 and its dynamic interaction with other comorbid components. Hence, we evaluated obesity as a 31 paramount risk factor for mortality and severity in COVID-19 infection, independent of potential 32 confounders via systematic review and meta-regression. information (country, sample size), patient characteristics (age, baseline comorbidities, gender), 41 treatment information and outcome data. The search strategy consisted of keywords "SARS-COVID-19 database which included articles from Medline (PubMed), Embase, Science Web, 44 and Cochrane Central Controlled Trials Registry. Studies were included from all over the world, 45 there were no language barriers. Other literature sources such as the BioRxiv (preprints), 46 MedRxiv (preprints), ChemRxiv (preprints), and SSRN (preprints) were searched as well. After 47 following a thorough search, full-length articles meeting the inclusion criteria were evaluated. In 48 an attempt to discover further eligible studies, we manually searched the reference lists of the 49 included studies, and previously published meta-analysis, systematic review, and the relevant 50 literature. We also scanned the clinicaltrials.gov registry for completed, as well as in-progress 51 randomized controlled trials (RCTs).

Eligibility Criteria: 53 The inclusion criteria for the systematic review are as follows: CC-BY-NC-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 May 10, 2021. ; https://doi.org/10.1101/2021.05.08.21256845 doi: medRxiv preprint positive hospitalized patients. Unadjusted and adjusted impact measurements were also extracted 65 where appropriate. From each study, various details including first author name, study type, 66 hospitalized total covid-19 positive patients, the definition of COVID-19 severity, definition of 67 obesity, total obese & non-obese COVID-19 positive patients, patients with high severity and 68 mortality, median age, gender (female sex proportion), hypertension proportion, pulmonary 69 disease proportion, cardiovascular disease proportion, diabetes proportion, dyslipidemia 70 proportion, liver disease proportion were mentioned in a tabulated format in excel sheet. These 71 details are exhibited in Table 1 . The included data was checked for accuracy by all authors. 72 Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines were used. The meta-analysis specifically included case-control and cohort studies comparing the effects of 81 high BMI/Obesity in COVID-19 hospitalized patients comparing them to the non-obese COVID-82 19 hospitalized patients. All outcomes were analyzed using the Mantel-Haenszel method for 83 dichotomous data to estimate pooled risk ratio (RR) utilizing the Review Manager (RevMan)- 84 Version 5.4, The Cochrane Collaboration, 2020. Meta-analysis was performed first for studies 85 reporting severity of patients in both groups followed by that for studies reporting severity of 86 disease assuming independence of results for studies that reported both. Due to anticipated 87 heterogeneity, summary statistics were calculated using a random-effects model. This model 88 accounts for variability between studies as well as within studies. Statistical heterogeneity was 89 assessed using Q value and I 2 statistics. 90 To explore differences between studies that might be expected to influence the effect size, we 91 performed random effects (maximum likelihood method) univariate and multivariate meta-92 regression analyses. The potential sources of variability hypothesized were the gender of the 93 study sample, the proportion of subjects with diabetes, pulmonary disease, cardiovascular 94 disease, and hypertension. Covariates were selected for further modeling if they significantly 95 (P < 0.05) modified the association between mortality or severity in the COVID-19 hospitalized 96 patients with high BMI/Obesity. Two models were created, one for severity and the other for 97 mortality of disease as outcomes. Subsequently, preselected covariates were included in a 98 manual backward and stepwise multiple meta-regression analysis with P = 0.05 as a cutoff point 99 for removal. P < 0.05. (P < 0.10 for heterogeneity) was considered statistically significant. All 100 meta-analysis and meta-regression tests were 2-tailed. The meta-regression was done with the 101 Comprehensive Meta-Analysis software package (Biostat, Englewood, NJ, USA)14 39 . 102 We conducted sensitivity analysis with BMI categories (BMI <18 kg/m 2 , BMI 18 kg/m 2 -25 kg/m 2 , 103 BMI 25 kg/m 2 -29.9 kg/m 2 , BMI >30 kg/m 2 , and BMI>40 kg/m 2 ) to decrease inherent selection 104 bias in observational studies 40 .

Risk of Bias assessment-The Newcastle-Ottawa (NOS) scale12 was used for measuring the risk 107 of bias in case-control studies and cohort studies. The following classes were rated per study: 108 . CC-BY-NC-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 May 10, 2021. ; https://doi.org/10.1101/2021.05.08.21256845 doi: medRxiv preprint low bias risk (9 points), moderate bias risk (5-7 points), and high bias risk (0-4 items. For a 109 cross-sectional study, we used the modified version of NOS, assigning the study in the following 110 groups: Low risk of bias (8-10), moderate risk (5-7), high risk of bias (0-4) ) 41 . Three reviewers 111 (AT, SA, and SSR) evaluated the likelihood of bias independently, and any conflict was resolved 112 by consensus (Table 2A and Meta-analysis for mortality outcome: Meta-analysis findings showed that obesity was 129 associated with increased risk of mortality from COVID 19 infections in comparison to non-130 . CC-BY-NC-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 May 10, 2021. ; https://doi.org/10.1101/2021.05.08.21256845 doi: medRxiv preprint obese patient population (RR=1.12, 95% CI 1.06-1.19, p<0.001). Heterogeneity was high with I 2 131 = 88% (Figure 3 ).

Multivariate meta-regression model for severity outcome: Multivariate meta-regression was 133 performed to explain variations in the association between COVID-19 severity and obesity. We 134 found female gender, pulmonary disease, diabetes, age, cardiovascular diseases, and 135 hypertension covariates to be significant and this explained R 2 = 50% of the between-study 136 heterogeneity in severity. The proportion of hypertension did not significantly affect the 137 between-study variations and were therefore not included in the final equation. Figure 4 shows 138 the resulting equation and individual covariate effect graphs.

Multivariate meta-regression model for mortality outcome: Multivariate meta-regression 140 performed to explain variations in the association between mortality and obesity revealed that 141 female gender, proportion of pulmonary disease, diabetes, hypertension, and cardiovascular 142 diseases to be significant together. Overall, these covariates together explained R 2 =53% of the 143 between-study heterogeneity in mortality. Figure However, in Egger's regression test the null hypothesis of no small study effects was rejected at 148 p<0.05 (estimated bias coefficient = -0.13 ± 0.42SE).

Similarly, visual inspection of the standard error plots for the mortality analysis ( Figure 7B ) 150 suggests symmetry without an underrepresentation of studies of any precision. Corroborating 151 . CC-BY-NC-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 May 10, 2021. 18 kg/m 2 -25 kg/m 2 . We also did not observe any statistically significant changes while comparing 159 BMI category 25 kg/m 2 to 29.9 kg/m 2 with respect to others in terms of mortality and severity of 

In this large meta-analysis with 100 studies, we found that obesity has a strong association with 163 increased mortality & severity of COVID-19 infection. In addition, our meta-regression analysis 164 suggests that obesity significantly increases the severity and mortality in COVID-19 patients.

Using a random effects model, we found that obese patients showed higher odds for mortality 166 and severity i.e. ICU admissions or mechanical ventilation. Our results suggest that obese 167 individuals are 1.5 times more likely to experience severe outcomes and 1.12 times more likely 168 to die when compared to non-obese individuals with COVID-19 disease. Our meta-regression 169 severity model suggested that 50% of the heterogeneity could be explained by age, gender, 170 diabetes, hypertension, pulmonary and cardiovascular diseases. The mortality meta-regression 171 model suggested that 53% of the heterogeneity could be accounted for by gender, diabetes, 172 . CC-BY-NC-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 May 10, 2021. ; https://doi.org/10.1101/2021.05.08.21256845 doi: medRxiv preprint hypertension, pulmonary and cardiovascular diseases. Through these regression models, we were 173 able to address major amount of heterogeneity seen in our meta-analysis. 174 In the existing literature, we found four meta-analysis (studies n=6, 17, 40, 76) 141-144 that 175 explored the association of obesity and worse outcomes in COVID-19 and found a similar 176 association. On the contrary, one study refuted the possibility of this association. Owing to their 177 small sample population (Studies n=2), it is likely that they were underpowered to tease out the 178 true difference or association 145 . With a much larger sample size (n=100) our study provides a 179 more robust evidence to establish this association. CC-BY-NC-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 May 10, 2021. ; https://doi.org/10.1101/2021.05.08.21256845 doi: medRxiv preprint identified the likely confounders to be age, gender, and co-morbidities such as diabetes, 196 hypertension, pulmonary and cardiovascular diseases. Through this model, we were able to 197 explain high heterogeneity with highest number of confounders, which other meta regression in 198 the recent literature were not able to reach and define 146-150 . Thus, we were able to establish a 199 strong association that obesity plays a remarkable role in worsening these outcomes in patients 200 with COVID-19 infection. In the sensitivity analyses, we were only able to find statistically 201 significant results for increased mortality in BMI<18 kg/m 2 as compared to BMI 18 kg/m 2 -25 202 kg/m 2 , however, such significance was not noted in any other BMI categories with severity and 203 mortality in COVID-19. This could be due to BMI being a very crude estimate of adiposity, may 204 not be sensitive enough to tease out the real difference. Visceral adiposity would probably be a 205 more reliable estimate to study these differences. However, in their study, Anderson et al. found 

The prime strength of this study is the large sample size. With an exhaustive search strategy, we 233 compiled 100 studies conducted globally. We also added the most recent studies to our meta-234 analysis and meta-regression model including the studies that reported contradictory information.

It enabled us to arrive at a more definitive conclusion about the risk associations. To define the 236 heterogeneity in the meta-analysis, we also conducted a meta-regression analysis. For 237 moderators, we used the most probable confounders based on the available evidence. This 238 enabled us to delineate the impact of obesity as an independent risk factor for mortality and 239 severity in COVID-19. However, our study is also subject to few limitations. We included five 240 . CC-BY-NC-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. CC-BY-NC-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 May 10, 2021. ; https://doi.org/10.1101/2021.05.08.21256845 doi: medRxiv preprint a call for the scientific community to further delve into its pathophysiology and identify potential 264 pharmacological targets, since COVID-19 is an ever-evolving disease. Finally, this information 265 must be disseminated to the general public to intensify the primary prevention of obesity. CC-BY-NC-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 May 10, 2021. . CC-BY-NC-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 May 10, 2021. (NOS) for assessing the quality of nonrandomised studies in meta-analyses. 374 . CC-BY-NC-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 May 10, 2021. . CC-BY-NC-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 May 10, 2021. . CC-BY-NC-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 May 10, 2021. . CC-BY-NC-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 May 10, 2021. . CC-BY-NC-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 May 10, 2021. CC-BY-NC-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 May 10, 2021. CC-BY-NC-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. . CC-BY-NC-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 May 10, 2021. Low CD, cannot determine; NA, not applicable; NR, not reported . CC-BY-NC-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 May 10, 2021. . CC-BY-NC-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 May 10, 2021. Was the research question or objective in this paper clearly stated and appropriate? Yes Yes Was the study population clearly specified and defined?

Yes Yes Did the authors include a sample size justification?

Yes Yes Were controls selected or recruited from the same or similar population that gave rise to the cases (including the same timeframe)?

Were the definitions, inclusion and exclusion criteria, algorithms or processes used to identify or select cases and controls valid, reliable, and implemented consistently across all study participants?

Were the cases clearly defined and differentiated from controls? Yes Yes If less than 100 percent of eligible cases and/or controls were selected for the study, were the cases and/or controls randomly selected from those eligible?

Was there use of concurrent controls? Yes No Were the investigators able to confirm that the exposure/risk occurred prior to the development of the condition or event that defined a participant as a case?

Were the measures of exposure/risk clearly defined, valid, reliable, and implemented consistently (including the same time period) across all study participants?

Were the assessors of exposure/risk blinded to the case or control status of participants? NR NR Were key potential confounding variables measured and adjusted statistically in the analyses? If matching was used, did the investigators account for matching during study analysis?

High High

Very low CD, cannot determine; NA, not applicable; NR, not reported . CC-BY-NC-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 May 10, 2021. ; https://doi.org/10.1101/2021.05.08.21256845 doi: medRxiv preprint

Obesity is associated with increased severity of disease in 657 COVID-19 pneumonia: a systematic review and meta-analysis

Obesity and its comorbid conditions

11β-hydroxysteroid dehydrogenase types 661 1 and 2 activity in subcutaneous adipose tissue in humans: implications in obesity and diabetes

The Role of Adipocytes and Adipocyte-Like Cells in the Severity of 664 COVID-19 Infections

Presence of increased stiffness of the common carotid 666 artery and endothelial dysfunction in severely obese children: a prospective study

Endothelial activation and dysfunction in COVID-669 19: from basic mechanisms to potential therapeutic approaches. Signal Transduction and 670 Targeted Therapy

Endothelial 672 Dysfunction in COVID-19: Lessons Learned from Coronaviruses

The Renin Angiotensin Aldosterone System in Obesity 675 and Hypertension: Roles in the Cardiorenal Metabolic Syndrome

Mechanisms of thrombosis in obesity

Endothelial Dysfunction in Obesity-Induced 680 Inflammation: Molecular Mechanisms and Clinical Implications

Endothelial dysfunction, obesity and insulin resistance

Endothelial Dysfunction in Obesity: Role of Inflammation. High Blood Press Cardiovasc 684 Prev

Platelet dysfunction in central obesity

An update on immune dysregulation in obesity-related 688 insulin resistance

Acute Kidney Injury, and Mortality in Critical Illness. 690 Critical care medicine

Effect of obesity on intensive care morbidity and mortality: a 692 meta-analysis

The impact of obesity on outcomes after critical 694 illness: a meta-analysis

Obesity and mortality in critically ill adults: a systematic review and meta-696 analysis

Is body mass index associated with outcomes of mechanically 698 ventilated adult patients in intensive critical units? A systematic review and meta-analysis

Obesity and Disease Severity Among 701 Patients With COVID-19

Yes Yes Yes Yes Yes Yes YesWere key potential confounding variables measured and adjusted statistically for their impact on the relationship between exposure(s) and outcome(s)?