Positive correlation between long term emission of several air pollutants and COVID-19 deaths in Sweden

Abstract

Several recent studies have found troubling links between air pollution and both incidence and mortality of COVID-19, the pandemic disease caused by the virus SARS-CoV-2. Here, we investigate whether such a link can be found also in Sweden, a country with low population density and a relatively good air quality in general, with low background levels of important pollutants such as PM2.5 and NO2. The investigation is carried out by relating normalized emission levels of several air pollutants to normalized COVID-19 deaths at the municipality level, after applying a sieve function using an empirically determined threshold value to filter out noise. We find a fairly strong correlation for PM2.5, PM10 and SO2, and a moderate one for NOx. We find no correlation neither for CO, nor (as expected) for CO2. Our results are statistically significant and the calculations are simple and easily verifiable. Since the study considers only emission levels of air pollutants and not measurements of air quality, climatic and meteorological factors (such as average wind speeds) can trivially be ruled out as confounders. Finally, we also show that although there are small positive correlations between population density and COVID-19 deaths in the studied municipalities (which are for the most part rural and non densely populated) they are either weak or not statistically significant.