We specified Bayesian hierarchical Poisson log-linear models to investigate the association of COVID-19 deaths and NO2 and PM2.5 independently.