Background Short-term exposure to air pollution is linked to adverse health outcomes, yet most studies assume constant associations over time and focus on densely populated regions. Evidence on temporal changes and impacts in less populated areas remains limited. Objective We aimed to (1) estimate non-linear associations between mortality and fine particulate matter (PM 2.5 ), as well as between mortality and a mixture of PM 2.5 , ground-level ozone (O 3 ) and nitrogen dioxide (NO 2 ), (2) characterize non-linear temporal trends, (3) expand analysis beyond the most populated regions, and (4) assess whether trends differ by population size. Methods Daily mortality counts (non-accidental, including cause-specific mortality) and air pollutant concentrations were analyzed for 101 Canadian census divisions from 2001–2015. We applied generalized additive mixed models to estimate time-varying, non-linear associations, adjusting for temperature, age, sex, day of the week, holiday, seasonal variation and long-term trends. Stratified analyses explored differences between large- and medium-populated areas. Results PM 2.5 showed slightly non-linear associations with non-accidental mortality in the single-pollutant model, with a rate ratio of 1.014 (95% confidence interval 1.010, 1.019) comparing the 3rd to 1st quartile at the beginning of the study period. Associations declined over time for non-accidental, circulatory, and non-cardiopulmonary mortality. The mixture associations followed similar patterns, but with larger magnitudes. Large-populated areas mirrored national trends, while medium-populated areas showed stable risks for non-cardiopulmonary mortality. Discussion Temporal trends in air pollutant-mortality associations varied by population size and cause of death, underscoring the need for further research into regional and cause-specific drivers.
Pan et al. (Mon,) studied this question.