Potential causal associations of long-term exposure to PM2.5 constituents and all-cause mortality: Evidence from the Pearl River Cohort study

Evidence of the potential causal effect of PM_2.5 and its constituents on all-cause mortality based on large population cohort is still limited. Based on a large scale cohort of 341,098 participants in southern China, we developed a marginal structure Cox model based on inverse probability weighting, an established causal inference approach, to evaluate the potential causal associations between PM_2.5 constituents and all-cause mortality, taking into account time-varying covariates. Additionally, we further explored the modifying effects of demographic and lifestyle characteristics on these associations. For each IQR increase in black carbon and organic matter, the risk of all-cause mortality increased by 51 % (95 % CI: 40-62 %) and 52 % (95 % CI: 40-58 %), followed by ammonium, nitrate and sulfate (HR = 1.32, 1.38 and 1.39, respectively). Individuals < 65 years, unmarried, urban medical insured, not consuming alcohol, or not exercising were potentially more susceptible to the adverse effects for most PM_2.5 constituents (P for interaction < 0.10). The HR estimates for these subgroups ranged from 1.51 to 2.10 for black carbon, 1.40-1.71 for nitrate, 1.42-1.53 for sulfate, and 1.52-1.77 for organic matter. Various sensitivity analyses verified the robustness of our findings. This study presented compelling evidence of potential causal links between PM_2.5 constituents and all-cause mortality.