RESUMO
ObjectivesTo determine whether UVA exposure might be associated with COVID-19 deaths DesignEcological regression, with replication in two other countries and pooled estimation Setting2,474 counties of the contiguous USA, 6,755 municipalities in Italy, 6,274 small areas in England. Only small areas in their Vitamin D winter (monthly mean UVvitd of under 165 KJ/m2) from Jan to April 2020. Participants The at-risk population is the total small area population, with measures to incorporate spatial infection into the model. The model is adjusted for potential confounders including long-term winter temperature and humidity. Main outcome measuresWe derive UVA measures for each area from remote sensed data and estimate their relationship with COVID-19 mortality with a random effect for States, in a multilevel zero-inflated negative binomial model. In the USA and England death certificates had to record COVID-19. In Italy excess deaths in 2020 over expected from 2015-19. Data sourcesSatellite derived mean daily UVA dataset from Japan Aerospace Exploration Agency. Data on deaths compiled by Center for Disease Control (USA), Office for National Statistics (England) and Italian Institute of Statistics. ResultsDaily mean UVA (January-April 2020) varied between 450 to 1,000 KJ/m2 across the three countries. Our fully adjusted model showed an inverse correlation between UVA and COVID-19 mortality with a Mortality Risk Ratio (MRR) of 0.71 (0.60 to 0.85) per 100KJ/m2 increase UVA in the USA, 0.81 (0.71 to 0.93) in Italy and 0.49 (0.38 to 0.64) in England. Pooled MRR was 0.68 (0.52 to 0.88). ConclusionsOur analysis, replicated in 3 independent national datasets, suggests ambient UVA exposure is associated with lower COVID-19 specific mortality. This effect is independent of vitamin D, as it occurred at irradiances below that likely to induce significant cutaneous vitamin D3 synthesis. Causal interpretations must be made cautiously in observational studies. Nonetheless this study suggests strategies for reduction of COVID-19 mortality.