Current evidence on short-term effects of air pollution on mortality often overlooks potential temporal variation and non-linear exposure-response relationships, bringing to biased effect estimates and limited accuracy of health risk assessments. Leveraging high-quality daily environmental data from local air quality monitoring stations, we examined temporal changes and non-linear associations between daily concentrations of PM10, PM2.5, NO2, and SO2 and mortality from natural, cardiovascular, and respiratory causes in a high-risk area surrounding the city of Florence (Italy), from 2008 to 2019. Missing environmental data were handled through multiple imputation. For each air pollutant, we fitted Poisson regression models assuming: (1) a linear effect which was constant over time, (2) a linear effect allowed to flexibly vary over the study period, and (3) a non-linear effect which was constant over time. Regression splines were used to model non-linearities. During the entire period, we found that higher levels of PM2.5, and SO2 were associated with increased mortality. These effects were time-varying, with a peak observed during 2012–2015, despite lower pollutant concentrations. Only PM2.5 exhibited a non-linear relationship with natural and cardiovascular mortality, with greater effects at higher concentrations. The effects of NO2 were minimal. In the study area, the harmful effects of air pollution did not decline over time despite decreasing pollutant concentrations. This pattern may reflect changes in airborne particles composition, possibly linked to variations in traffic and major infrastructures development, or complex interactions with meteorological factors. Our findings highlight the need of high-quality local studies to disentangle complex exposure-response dynamics and support effective, context-specific regulatory decisions.
Marzi et al. (Thu,) studied this question.