The impact of obesity on severe disease and mortality in people with SARS‐CoV‐2: A systematic review and meta‐analysis

Abstract Background Obesity accompanied by excess ectopic fat storage has been postulated as a risk factor for severe disease in people with SARS‐CoV‐2 through the stimulation of inflammation, functional immunologic deficit and a pro‐thrombotic disseminated intravascular coagulation with associated high rates of venous thromboembolism. Methods Observational studies in COVID‐19 patients reporting data on raised body mass index at admission and associated clinical outcomes were identified from MEDLINE, Embase, Web of Science and the Cochrane Library up to 16 May 2020. Mean differences and relative risks (RR) with 95% confidence intervals (CIs) were aggregated using random effects models. Results Eight retrospective cohort studies and one cohort prospective cohort study with data on of 4,920 patients with COVID‐19 were eligible. Comparing BMI ≥ 25 vs <25 kg/m2, the RRs (95% CIs) of severe illness and mortality were 2.35 (1.43‐3.86) and 3.52 (1.32‐9.42), respectively. In a pooled analysis of three studies, the RR (95% CI) of severe illness comparing BMI > 35 vs <25 kg/m2 was 7.04 (2.72‐18.20). High levels of statistical heterogeneity were partly explained by age; BMI ≥ 25 kg/m2 was associated with an increased risk of severe illness in older age groups (≥60 years), whereas the association was weaker in younger age groups (<60 years). Conclusions Excess adiposity is a risk factor for severe disease and mortality in people with SARS‐CoV‐2 infection. This was particularly pronounced in people 60 and older. The increased risk of worse outcomes from SARS‐CoV‐2 infection in people with excess adiposity should be taken into account when considering individual and population risks and when deciding on which groups to target for public health messaging on prevention and detection measures. Systematic review registration: PROSPERO 2020: CRD42020179783.


| BACKG ROU N D
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), initially reported in November 2019 in Wuhan, China, has now claimed over 300 000 lives globally 1 and devastated the global economy. It continues to severely affect the social fabrics of most counties. The unprecedented projected mortality and economic devastation caused by the virus has led to global research efforts to identify people at greatest risk of developing critical illness and dying. Chen et al identified older age, male gender and comorbidities such as hypertension, diabetes, cardiovascular disease and chronic lung disease as risk factors for severe disease when they compared the characteristics of 113 (14.4%) patients who had died of the disease with those of 161 patients who recovered. 2 Recent studies have increasingly described obesity as an associating factor for people at an increased risk of severe disease. [3][4][5][6] The risk of many chronic diseases such as cardiovascular disease, cancer, diabetes and the mortality in individuals with these diseases increases significantly in people with obesity or overweight. 7 With the World Health Organization estimating that more than 1.9 billion adults worldwide have overweight or obesity, 8 any causal relationship or association between obesity and severe disease from SARS-CoV-2 has the potential to claim even more lives globally.
There may be pathophysiological mechanisms for worse outcomes in people with excess adipose tissues. 9 Excess adipose tissue leads to local insulin resistance and may stimulate inflammation, functional immunologic deficit and a pro-thrombotic disseminated intravascular coagulation with associated high rates of venous thromboembolism. 10,11 Conversely, the phenomenon of the 'obesity paradox' has emerged from epidemiological data in recent years to suggest counterintuitively that people with overweight and obesity may have a better prognosis than those with BMI values in the normal range.
This phenomenon has thus far been described in patients with hypertension, heart failure, coronary artery disease, peripheral artery disease and other cardiovascular and noncardiovascular conditions. 12,13 The phenomenon seems more convincing in older populations where weight and muscle mass start to decline at advanced age. Sarcopenia is a major contributor to age-related frailty. 14 It is therefore possible that people with overweight and obesity, especially in older populations, could have less severe disease and mortality compared to individuals with normal weight.
In this context, whether obesity has better or worse impact on the occurrence of severe disease or death in people with SARS-CoV-2 is uncertain. A recent systematic review on this topic identified only three studies and concluded that obesity is an independent risk and prognostic factor for SARS-CoV-2. The review was limited by the small number of included studies, the absence of some key data and the lack of quantitative synthesis. 15 In order to attempt to quantify the relationship between raised body weight and severe outcomes from COVID-19, we conducted a systematic review and meta-analysis to determine whether people with overweight or obesity and with SARS-CoV-2 have different outcomes compared to those within normal weight thresholds.

| Data sources and search strategy
This review was conducted and reported based on PRISMA and MOOSE guidelines 16,17 (Appendices S1 and S2) and in accordance with a protocol which has been registered with PROSPERO, (CRD42020179783). We searched MEDLINE, Embase, Web of Science and The Cochrane Library from inception to 16 May 2020 for studies reporting on relationships between overweight and obesity and the clinical outcomes in patients with COVID-19. The computer-based searches combined free and MeSH search terms and key words related to the exposures (such as 'Body Mass Index', 'BMI' 'Body Weight' 'Obesity'.) and population (eg 'COVID-19', 'SARS-CoV-2') in humans. The search strategy was limited to English language given the potential for duplicate reporting of same study participants. 18 The full search strategy is reported in Appendix S3.
The titles and abstracts of retrieved citations were initially screened by one reviewer (SS) for potential eligible studies based on the inclusion criteria. Full texts of potential eligible abstracts were then acquired for the detailed evaluation by 2 reviewers (SS and SKK). To potentially identify articles missed by the search strategy, reference lists of relevant articles were manually scanned.

| Study selection and eligibility criteria
The protocol was designed to include all observational studies (prospective cohort, retrospective, nested case-control and casecontrol), clinical studies, nonrandomized controlled trials (RCTs) and RCTs reporting a relationship between obesity and the clinical outcomes in patients with COVID-19. Article types such as opinion pieces, consensus reports, single case studies and narrative reviews were excluded. Outcomes evaluated included severe disease (defined as intensive care unit (ICU) care, SpO 2 < 90%, requiring supplemental O 2 ) and mortality. No limits were placed on the study follow-up duration.

| Data extraction and quality assessment
Using a predesigned data extraction form which had been piloted by one reviewer (SKK), data on patient characteristics ( 19 a validated tool for assessing the quality of nonrandomized studies, was used to assess the methodological quality of the studies. This tool measures the quality of evidence from a score of zero to nine (higher scores = better), based on three predefined domains including the following: (a) selection of participants; (b) comparability; and (c) ascertainment of outcomes of interest.

| Statistical analysis
Mean differences (95% CIs) for comparing mean levels of body mass index (BMI) across outcomes and RRs (95% CIs) for associations between BMI categories and outcomes were used as summary measures across studies. The reported HRs were assumed to approximate the same measure of RR. Multivariable-adjusted risk estimates were used for pooling if available, otherwise crude RRs were used as reported. For data reported as medians, standard errors, ranges and 95% confidence intervals (CIs), means and standard deviations were estimated using methods as described by Hozo and colleagues. 20 Due to the different cut-offs used for BMI by the included studies, we employed the risk comparison: ≥25 vs <25 kg/m 2 , to ensure Random effects models using the inverse variance weighted method (DerSimonian and Laird) were used to combine RRs to minimize the effect of heterogeneity. Heterogeneity was assessed and quantified using the Cochrane χ 2 statistic and the I 2 statistic. 21 Prespecified study-level characteristics such as geographical location, average age at baseline and study quality were explored as sources of heterogeneity, using stratified analysis and random effects meta-regression. 22 We assessed for evidence of publication bias using visual inspection of Begg's funnel plots and Egger's regression symmetry tests. STATA release MP 16 (StataCorp LP) was used for all statistical analyses.

| Study identification and selection
The flow of studies through the review process is presented in

| Study characteristics and quality
All nine studies comprised of observational cohorts (8 retrospective and 1 prospective) and comprised of 4920 patients with COVID-19, of which 841 developed severe illness and 136 died F I G U R E 1 Selection of studies included in the meta-analysis (Table 1). Four studies were based in Asia (China and Singapore) (n = 549), three from Europe (France and Italy) (n = 653) and two from North America (USA) (n = 3718). The average age at baseline ranged from approximately 43 to 64 years, with a weighted mean (SD) of 56.8 (8.3) years. All studies enrolled both male and female patients. The overall NOS methodological quality scores of studies ranged from 4 to 7.

| Subgroup analyses and observed heterogeneity
The substantial heterogeneity between the contributing studies reporting on the association between BMI ≥ 25 vs <25 kg/m 2 and risk of severe illness may be partly explained by age (P-value for meta-regression < .001) and study quality (P-value for metaregression = .02) ( Figure 5). In subgroup analyses, BMI ≥ 25 kg/m 2 was associated with a statistically significant increased risk of severe illness in older age groups (≥60 years); the point estimate was lower and the difference was no longer statistically significant in younger age groups (<60 years). In a second subgroup analysis, BMI ≥ 25 kg/ m 2 was associated with statistically significant increased risk of se- The reasons for the observed heterogeneity in our analysis comparing BMI levels in people with severe illness compared to those without is unclear; due to the small number of studies, subgroup analyses were not appropriate. However, despite high levels of statistical heterogeneity, all studies included in this analysis found statistically significant increases in BMI in people with severe disease; the direction of effect was consistent and differences in the magnitude of effect drove the statistical heterogeneity in this case. People with overweight or obesity are generally more likely to have more severe infections, decreased responses to treatments and an increased risk of death. 27 We have shown that people with higher BMI have an increased risk of severe disease in older age groups (≥60 years); the association was less clear in younger people. In a recent analysis of 265 patients with COVID-19, Kass et al 28 described a significant inverse correlation between age and BMI, in which younger individuals admitted to hospital were more likely to be obese. The authors suggest that, on this basis, public messaging to younger adults be clear that even at younger ages, they are still vulnerable to COVID-19. Though our data did not find the same pattern as Kass et al, we do not disagree with this messaging. The average age at baseline in our population ranged from approximately 43 to 64 years, which is relatively young.

| D ISCUSS I ON
Despite this relatively young population, 17% of our population had severe disease or death.

| Possible mechanisms explaining the link between raised body weight and worse outcomes from COVID-19
Several mechanisms could explain why raised body weight predisposes patients with SARS-CoV-2 to severe disease. First, raised body weight increases mechanical pressure on chest and abdomen causing diaphragmatic embarrassment, thus restricting pulmonary function, especially when lying supine. This causes a decrease expiratory reserve volume, functional capacity and respiratory system compliance. 24

TA B L E 1 Baseline characteristics of included studies of COVID-19 patients
Author, year of publication Source of participants Similarly, obesity is one of the leading risk factors for atrial fibrillation and this is a common condition present in severe forms SARS-CoV-2. 30 Given limitations in our data, we were unable to ascertain the extent to which pre-existing conditions contribute to the observed association between raised BMI and worse outcomes from COVID-19.
Thirdly, obese patients have excess ectopic visceral fat, which correlates with a cluster of metabolic abnormalities. 31 Visceral fat represents a metabolically active organ and has been strongly re- Finally, obesity enhances thrombosis, 40,41 and this increases with the severity of obesity. 42 The mechanistic pathway impli-  pro-thrombotic disseminated intravascular coagulation and higher rates of venous thromboembolism, leading to worse outcomes. 38 Therefore, increased hypercoagulability and thrombosis in COVID-19 patients may contribute to additive effects of obesity and SARS-CoV-2 infection.

| Strengths and limitations
Our analysis has several strengths. Our comprehensive search strategy yielded the highest number of published articles on this topic to date and evaluated the relationship of overweight and obesity on adverse outcomes in people with COVID-19. We were able to transform reported risk estimates from majority of contributing studies to a consistent level to allow combination of estimates across studies, which enhanced interpretation of the overall find-

| CON CLUS ION
Despite important differences between studies and relatively small sample sizes, data consistently suggest raised body weight is a risk factor for severe disease and death with COVID-19. This is particularly pronounced in older age groups and in higher BMI ranges. The increased risk of worse outcomes from SARS-CoV-2 infection in people with overweight and obesity should be taken into account when considering individual and population risks and when deciding on which groups to target with public health messaging and increased prevention and detection measures.

ACK N OWLED G EM ENTS
We acknowledge the support from the National Institute for Health

DATA AVA I L A B I L I T Y S TAT E M E N T
The corresponding author had full access to all the data in the studies used for the analysis and takes responsibility for the accuracy of the data analysis. The studies are all publicly available in the publications. The data that support the findings of this study as listed in the reference list, are openly available on PUBMED.