Interactive effects between extreme temperatures and PM2.5 on cause-specific mortality in thirteen U.S. states.

Abstract

The extent and robustness of the interaction between exposures to heat and ambient PM_2.5 is unclear and little is known of the interaction between exposures to cold and ambient PM_2.5. Clarifying these interactions, if any, is crucial due to the omnipresence of PM_2.5 in the atmosphere and increasing scope and frequency of extreme temperature events. To investigate both of these interactions, we merged 6 073 575 individual-level mortality records from thirteen states spanning seventeen years with 1 km daily PM_2.5 predictions from sophisticated prediction model and 1 km meteorology from Daymet V4. A time-stratified, bidirectional case-crossover design was used to control for confounding by individual-level, long-term and cyclic weekly characteristics. We fitted conditional logistic regressions with an interaction term between PM_2.5 and extreme temperature events to investigate the potential interactive effects on mortality. Ambient PM_2.5 exposure has the greatest effect on mortality by all internal causes in the 2 d moving average exposure window. Additionally, we found consistently synergistic interactions between a 10 μg m^-3 increase in the 2 d moving average of PM_2.5 and extreme heat with interaction odds ratios of 1.013 (95% CI: 1.000, 1.026), 1.024 (95% CI: 1.002, 1.046), and 1.033 (95% CI: 0.991, 1.077) for deaths by all internal causes, circulatory causes, and respiratory causes, respectively, which represent 75%, 156%, and 214% increases in the coefficient estimates for PM_2.5 on those days. We also found evidence of interactions on the additive scale with corresponding relative excess risks due to interaction (RERIs) of 0.013 (95% CI: 0.003, 0.021), 0.020 (95% CI: 0.008, 0.031), and 0.017 (95% CI: -0.015, 0.036). Interactions with other PM_2.5 exposure windows were more pronounced. For extreme cold, our results were suggestive of an antagonistic relationship. These results suggest that ambient PM_2.5 interacts synergistically with exposure to extreme heat, yielding greater risks for mortality than only either exposure alone.