2 October 2026Abstract Background Despite growing evidence on the health effects of air pollution, few studies have examined its impact on COVID-19 hospitalizations and deaths, and none have explicitly accounted for preexisting comorbidities in a causal framework. This study addresses this gap by investigating the relationship between PM2.5 exposure and COVID-19 severity. We place particular emphasis on the role of comorbidities, which can increase vulnerability to COVID-19 and may interact with air pollution exposure. Methods We use an exhaustive dataset covering the entire French territory over 2020–1, totaling more than 3.5 million weekly observations. The data include information on COVID-19 mortality and hospitalizations, disaggregated by age, sex, and municipality, along with detailed records on preexisting conditions. To address endogeneity, we employ a two-stage least squares approach, using wind direction as an instrument for local PM2.5 concentrations. Results We find no significant effect of PM2.5 exposure on COVID-19 mortality or hospitalizations among individuals without preexisting conditions. However, when accounting for comorbidities, a 10 μg/m3 increase in PM 2.5 was associated with higher COVID-19 mortality rates (1.5 per 100 000, 95% CI: 0.8–2.2), particularly among patients with respiratory (0.9 per 100 000, 95% CI: 0.4–1.4), and cardiovascular diseases (0.6 per 100 000, 95% CI: 0.1–1.1). Conclusions These findings underscore the importance of incorporating comorbidities when assessing the health effects of air pollution during a pandemic. Our results also show that the effects of PM2.5 are driven by the oldest age groups, who are both more vulnerable to COVID-19 and more likely to have comorbidities.