Cardiometabolic Traits, Sepsis, and Severe COVID-19

Objectives: To investigate whether there is a causal effect of cardiometabolic traits on risk of sepsis and severe covid-19. Design: Mendelian randomisation analysis. Setting: UK Biobank and HUNT study population-based cohorts for risk of sepsis, and genome-wide association study summary data for risk of severe covid-19 with respiratory failure. Participants: 12,455 sepsis cases (519,885 controls) and 1,610 severe covid-19 with respiratory failure cases (2,205 controls). Exposure: Genetic variants that proxy body mass index (BMI), lipid traits, systolic blood pressure, lifetime smoking score, and type 2 diabetes liability - derived from studies considering between 188,577 to 898,130 participants. Main outcome measures: Risk of sepsis and severe covid-19 with respiratory failure. Results: Higher genetically proxied BMI and lifetime smoking score were associated with increased risk of sepsis in both UK Biobank (BMI: odds ratio 1.38 per standard deviation increase, 95% confidence interval [CI] 1.27 to 1.51; smoking: odds ratio 2.81 per standard deviation increase, 95% CI 2.09-3.79) and HUNT (BMI: 1.41, 95% CI 1.18 to 1.69; smoking: 1.93, 95% CI 1.02-3.64). Higher genetically proxied BMI and lifetime smoking score were also associated with increased risk of severe covid-19, although with wider confidence intervals (BMI: 1.75, 95% CI 1.20 to 2.57; smoking: 3.94, 95% CI 1.13 to 13.75). There was limited evidence to support associations of genetically proxied lipid traits, systolic blood pressure or type 2 diabetes liability with risk of sepsis or severe covid-19. Similar findings were generally obtained when using Mendelian randomization methods that are more robust to the inclusion of pleiotropic variants, although the precision of estimates was reduced. Conclusions: Our findings support a causal effect of elevated BMI and smoking on risk of sepsis and severe covid-19. Clinical and public health interventions targeting obesity and smoking are likely to reduce sepsis and covid-19 trait estimates, weighted for the precision of the variant-outcome risk association, and with the intercept fixed to zero. 41 Causal effect estimates are expressed as odds ratios per standard deviation increase in genetically predicted levels of the exposure for continuous traits, and per unit increase in the log odds ratio of the exposure for binary traits. All analyses were performed using the MendelianRandomization package in R. 42