On the lagged non-linear association between air pollution and COVID-19 cases in Belgium

Exposure to air pollution has been proposed as a determinant of COVID-19 dynamics. While the connection between air pollution and COVID-19 has been established for several countries worldwide, few such analyses exist in Belgium. Therefore, we examine this potential association in Belgium, using COVID-19 cases of all 581 municipalities between September 2020 and January 2022. We employ a Bayesian spatio-temporal negative binomial model, allowing for potential non-linear and lagged effects of pollution. Comparing different single-pollutant models, we find that the model providing the best fit to the data contains black carbon. At the median pollution level, a cumulative risk of 1.66(1.57,1.74) over 8 weeks is found for this pollutant, compared to the 5% pollution quantile. In addition, the study reveals a remarkable similarity in COVID-19 incidence between adjacent municipalities in Belgium. Our findings suggest paying careful attention to highly air polluted areas when preparing for future pandemics of respiratory diseases. •We investigate the association between air pollution and COVID-19 in Belgium.•We apply the distributed lag non-linear model (DLNM) to flexibly capture the lagged nature of pollution.•We use a Bayesian spatio-temporal negative binomial model.•We find significant positive association between black carbon and COVID-19 incidence.•We find a prominent separation between COVID-19 cases in the municipalities in the North (Flanders Region) and South of Belgium (Walloon Region).