Seasonal impact of air particulate matter on morbidity: Interaction effect assessment in a time-stratified case-crossover design

The present study aimed to explore the interaction effect between PM10 and season as well as temperature on hospital admissions (HAs) in a region in North East of Iran. Distributed lag non-linear model in a time-stratified case-crossover design with a fixed and disjointed window was used in order to assess the impact of PM10 on HAs. Two approaches were applied to assess the interaction effect between season and the pollutant named stratification and interaction term in the regression model. In addition, a bivariate response surface model in Generalized Additive Model (GAM) was used to explore the interaction effect between PM10 and weather parameters, in order to see whether the seasonal variation of the PM10 effect is explained by weather conditions or not. The existence of interaction effect between PM10 and season was supported by both approaches; it was evident that spring significantly shared a large proportion of the adverse impact of PM10 on HAs. However, the interactive effects of PM10 and mean temperature were less obvious in this study. The results showed that all people were at risk of HAs due to high levels of PM10, with a particularly more immediate impact on elderlies. The results of the interaction effect can help policymakers to address spring when any preventive interventions are developed in the region.