ABSTRACT
Introduction: Several studies have linked obesity to more severe illness and higher mortality in COVID-19 patients. However, the relationship between being underweight and COVID-19 mortality remains inconclusive. Previous dose-response meta-analyses did not enroll or specifically analyze the underweight population. Herein, we conducted a systematic review and dose-response meta-analysis to investigate the relationship between body mass index (BMI) and mortality in both obese and underweight patients with COVID-19. Methods: We searched PubMed, Embase, Cochrane Library, Scopus, and Web of Science databases from inception until August 12, 2020 using the keywords “COVID-19,” “body mass index,” “obesity,” “overweight”, and “underweight.” Three reviewers independently assessed the relevant articles, including the title, abstract, and full text, to identify eligible studies. We performed a two-stage random-effects dose-response meta-analysis, including only studies with at least three quantitative classifications for BMI. The nonlinear trend was evaluated using a restricted cubic splines model with three-knots at the 10th, 50th, and 90th percentiles. A sensitivity analysis was conducted by pooling only those studies which specifically evaluated underweight patients (BMI<18.5 kg/m2). Results: Thirteen studies comprising 25,828 patients were included in the analysis. In the linear model, the mortality of patients with COVID-19 increased by 1.5% for each 1-kg/m2 increase in BMI (pooled relative risk [RR] =1.015, 95% confidence interval [CI] =1.004−1.027). However, a significant non-linear relationship between BMI and mortality was observed (Wald test: Pnon-linearity<0.001). We demonstrated a J-shaped curve, indicating that both underweight and obese patients had a higher mortality than those with normal weight. Interestingly, overweight patients (BMI, 25−30 kg/m2) seemed to have the lowest mortality risk. Using a BMI of 15 kg/m2 as the reference, the RRs of mortality decreased with BMI, and this trend continued until BMI of approximately 28 kg/m2 (RR=0.743, 95% CI=0.576−0.959). The relationship between BMI and mortality was then reversed, and an upward trend was observed when BMI exceeded 30 kg/m2;the RRs (95% CI) at BMIs 30, 35, 40, and 45 kg/m2 were 0.745 (0.570−0.974), 0.841 (0.643−1.100), 1.082 (0.850−1.377), and 1.457 (1.129−1.879), respectively. Conclusion: This study is the first dose-response meta-analysis that showed both underweight and obese COVID-19 patients are at higher risk of increased mortality. A J-curve relationship was demonstrated between BMI and COVID-19 mortality.