ABSTRACT
Background: Physical activity (PA) is known to be a protective lifestyle factor against several non-communicable diseases while its impact on infectious diseases, including Coronavirus Disease 2019 (COVID-19) is not as clear. Methods: We performed univariate and multivariate logistic regression to identify associations between both objectively and subjectively measured PA collected prospectively and COVID-19 related outcomes (overall COVID-19, inpatient COVID-19, outpatient COVID-19, and COVID-19 death) in the UK Biobank cohort. Subsequently, we tested causality by using Mendelian randomisation (MR) analyses. Results: In the multivariable model, the increased acceleration vector magnitude PA (AMPA) is associated with a decreased probability of overall and outpatient COVID-19 with an odds ratio (OR) and 95% confidence interval (CI) of 0.80 (0.69, 0.93) and 0.74 (0.58, 0.95), respectively. No association is found between self-reported moderate-to-vigorous PA (MVPA) and COVID-19 related outcomes. No association is found by MR analyses. Conclusions: Our results indicate a protective effect of objectively measured PA and COVID-19 outcomes (outpatient COVID-19 and overall COVID-19) independent of age, sex, measures of obesity, and smoking status. Although the MR analyses do not support a causal association, that may be due to limited power. We conclude that policies to encourage and facilitate exercise at a population level during the pandemic should be considered.