- Characteristics of the included studies
The characteristics of the 10 included studies are summarized in Table 1. These studies were published in the months of April and May 2020 from 5 countries, including China (2 studies), France (2 studies), Italy (1 study), the United Kingdom (3 studies), and the United States (2 studies). The sample size of these studies varied greatly, ranging from 41 to 17,425,445. Nine studies focused specifically on COVID-19 and obesity, and one study investigated obesity related behaviors (e.g., eating and physical activity).
The study design and populations were diverse. There were 8 cohort studies, 1 cross-sectional study and 1 case report. These studies included retrospective analysis of primary care electronic health records in the general population (1 study), retrospective analysis of cohorts of COVID-19 patients (5 studies), and prospective observational cohort studies (3 studies). The duration of data collected from these studies ranged from 1 week (2 studies), to 3 weeks (3 studies), 4 weeks (1 study), 6 weeks (1 study), 10 weeks (1 study) and 12 weeks (1 study). Some studies only included adults while others incorporated children. Subjects’ ages ranged from 0 to 104 years.
Only one study examined the influence of COVID-19 on obesity-related behaviors. This was a longitudinal observational study in Italy with 41 children. It reported changes in eating, physical activity and inactivity behaviors.
- Effect of obesity on COVID-19 infection risks
Nine studies provided results regarding the association between obesity and COVID-19 infection and severity risks. Three studies reported that obesity was more prevalent amongst COVID-19 infected patients. In a retrospective cohort study of SARS-CoV-2 confirmed patients, compared with a non-SARS-CoV-2-infected patient control group, Simonnet et al (4) observed that obesity (BMI 30-34.9kg/m2) and severe obesity (BMI ≥35 kg/m2) were significantly more frequent among COVID-19 infected patients than controls (47.6% vs 25.2% and 28.2% vs 10.8%, respectively). Williamson et al (18) used data from over 17 million UK National Health Service primary care records and observed that the incidence of COVID-19 advanced with increasing severity of obesity. They reported a dose-response relationship between degree of obesity and COVID-19 infection risk. Assuming non-obese individuals had a hazard ratio (HR) of 1.00, the fully adjusted HRs for obesity class I (BMI 30-34.9kg/m2), obesity class II (BMI 35-39.9kg/m2) and obesity class III (BMI ≥40 kg/m2) were 1.27, 1.56 and 2.27, respectively. An intensive care admissions report of confirmed COVID-19 cases from the Intensive Care National Audit and Research Centre (19) showed the majority of ICU admissions (71.7%) also suffered from overweight or obesity. Specifically, 31.6% of ICU admissions had overweight (BMI 25-29.9kg/m2), 32.8% had moderate obesity (BMI 30-39.9kg/m2) and 7.3% had severe obesity (BMI ≥40 kg/m2).
Six studies from different countries reported on the association between obesity and severity of COVID-19 infection based on varying criteria such as admission to acute or critical care such as ICU or requirement for IMV. Overall, they suggested a positive assocatoin. In a study of 291 French patients admitted to ICU for SARS-CoV-2, Caussy et al (20) observed a doubled risk of requirement for IMV in severe obese- (BMI ≥35 kg/m2) compared to lean patients (81.8% versus 41.9%). In a retrospective analysis of COVID-19 hospitalized patients in Wuhan, China, Hu et al (21) observed that while obesity (BMI ≥30 kg/m2) was not associated with diagnosis of disease severity (p=0.522), it was significantly associated with unfavorable clinical outcomes (OR=3.6, p=0.009). In a retrospective cohort study of 103 US patients admitted for COVID-19, Kalligeros et al (6) observed that obesity was associated with the severity of COVID-19 presentation. Severe obesity was associated with admission to ICU (adjusted OR=5.39) while obesity and severe obesity were associated with the need for IMV (adjusted OR=6.85 and 9.99, respectively). Similarly, Lighter et al (5) performed a retrospective analysis of COVID-19 patients in a New York hospital and found that in those under 60 years old, patients with obesity (BMI 30-34.9kg/m2) and severe obesity (BMI 35-39.9kg/m2) had a greater risk of admission to acute care (OR 2.0 and 2.2, respectively) and admission to ICU (OR 1.8 and 3.6, respectively). A similar trend was also observed in a French cohort (4) as its multivariate regression analysis showed OR for requirement for IMV was 7.36 in patients with BMI ≥35 kg/m2 compared with those with BMI <25 kg/m2 (p=0.021). Another study from China of patients with metabolic associated fatty liver disease (22) reported a significant association between overweight/obesity (BMI ≥25 kg/m2) and COVID-19 severity even after adjusting for age, sex, smoking, diabetes, hypertension, and dyslipidemia (adjusted OR=6.32, p=0.033).
One study reported the association between obesity and COVID-19 mortality. Data from about 17,000 COVID-19 patients in the UK showed that any form of obesity (as defined by hospital staff) was associated with a 37% greater risk of COVID-19 mortality (HR=1.37). (7) Note that in the study from China, Hu et al (21) included death, along with progression to more severe disease presentation, in their criteria they found unfavorable outcomes, indicating mortality data alone was insufficient from this study.
- Effect of COVID-19 on obesity-related behaviors
Only one published study investigated the effect of COVID-19 on obesity-related behaviors such as eating and physical activity. A longitudinal study of lifestyle factors involving 41 children and adolescents with obesity was based in Verona, Italy. Compared with previous data, lockdown was associated with an increase in number of meals a day (+1.15), an increase in number of servings of potato chips, red meat and sugary drinks (+0.54, 1.66 and 0.5, respectively), an increase in screen time and sleep time (+4.85 hours and +0.65 hours, respectively) and a decrease in time dedicated to sports (-2.3 hours/week). Therefore, more research on this topic is needed.
- Meta-analysis of the associations between COVID-19 and obesity
Figure 2 and Table 2 show that five studies (which provided 17 analyses and types of results) were included in the meta-analysis. The overall pooled analysis showed that obesity was associated with increased risk of COVID-19 infection (OR=1.69, 95% CI: 1.41, 2.02), and with large heterogeneity (I2= 87.3%, χ2=126.48, p< 0.001). The ORs reported by the individual studies ranged from 0.90 to 9.99.
Our subgroup analyses showed between-country differences in the associations between obesity prevalence and COVID-19 (Table 2): in the US (OR=1.80, 95% CI: 1.32, 2.46), the UK (OR=1.57, 95% CI: 1.23, 2.02), and China (where national obesity/overweight prevalence was much lower than that in the US and the UK) (OR=6.32, 95% CI: 1.16, 34.48). When stratified by study design, obese/overweight people had 1.66 times higher risk of getting COVID-19 infection than those non-overweight (OR=1.66, 95% CI: 1.39, 1.99) in cohort studies. In contrast, in a cross-sectional study, the reported OR for obese patients was 6.32 (95% CI: 1.16, 34.48). When stratified by obesity classifications, obesity (OR=1.43, 95% CI: 1.18, 1.73) and severe obesity (OR=1.96, 95% CI: 1.49, 2.59) were positively associated with COVID-19 risk, and obesity was positively associated with COVID-19 mortality (OR=1.64, 95% CI: 1.20, 2.25) and its severity: ICU (OR=2.01, 95% CI: 1.25, 3.23), and IMV use (OR=8.20, 95% CI: 2.10, 31.91).
Meta-regression analysis indicated that age affected the association between obesity and COVID-19 infection (β=-0.29; 95%: -0.47 to -0.10), but not gender (Table 3).
- Sensitivity analysis and assessment of publication bias
The sensitivity analysis consistently showed that removing individual studies from the meta-analysis did not change the association estimate (Supplemental Table 1). The funnel plot and assessment of Egger’s and Begg’s tests did not reveal any significant publication bias in the association (Egger P=0.148; Begg P=0.343) (Figure 3).