COVID-19 and perfluoroalkyl substances (PFAS) from contaminated drinking water in the Veneto Region (Italy): a spatial ecological analysis

Abstract

BACKGROUND AND AIM: Given the evidence of immunosuppression by some perfluoroalkyl substances (PFAS), there is interest in assessing if PFAS may have an impact on the risk of Coronavirus infection or COVID-19 severity. The COVID-19 pandemic in Italy is an opportunity to test this hypothesis in a large area polluted with PFAS, and data on mortality from COVID-19 by municipality. METHODS: For mortality from COVID-19 during the first wave of the pandemic (period 21 February - 15 April 2020), we fitted a Bayesian ecological regression model with spatially and not spatially structured random components on COVID-19 mortality at municipality level. We compared populations in the (red zone) of the Veneto Region, where residents were exposed for decades to drinking water contaminated by PFAS, with the rest of the Veneto. Serum PFOA in particular was raised in the Red Zone. Models included education score, background all-cause mortality (for the years 2015-2019), nursing homes provision in a sensitivity analysis, and an indicator for the Red Zone. The two random components are intended to adjust for potential hidden confounders spatially and not spatially structured. RESULTS:During the first wave of the pandemic, the COVID-19 crude mortality rate ratio for the Red Zone compared to the rest of the Veneto, was 1.55 (90% Confidence Interval 1.25; 1.92). From the Bayesian ecological regression model adjusted for education level and baseline all-cause mortality, the rate ratio for the Red Zone was 1.60 (90% Credible Interval 0.94; 2.51). CONCLUSIONS:We observed a higher mortality risk for COVID-19 in a population heavily exposed to PFAS, and plan to extend analyses to later periods of COVID mortality. If it is not simply a chance association, this may reflect a general immunosuppressive effect of PFAS, or bioaccumulation of PFAS in lung tissue, or PFAS-related pre-existing disease. KEYWORDS: PFAS, COVID-19, Spatial statistics, Hierarchical Bayesian models