Estimating causal effects of long-term PM_2.5 exposure on mortality in New Jersey

Background: Many studies have reported the associations between long-term exposure to PM_2.5 and increased risk of death. However, to our knowledge, none has used a causal modeling approach or controlled for long-term temperature exposure, and few have used a general population sample. oBjective: We estimated the causal effects of long-term PM_2.5 exposure on mortality and tested the effect modifications by seasonal temperatures, census tract-level socioeconomic variables, and county-level health conditions. Methods: We applied a variant of the difference-in-differences approach, which serves to approxi-mate random assignment of exposure across the population and hence estimate a causal effect. Specifically, we estimated the association between long-term exposure to PM_2.5 and mortality while controlling for geographical differences using dummy variables for each census tract in New Jersey, a state-wide time trend using dummy variables for each year from 2004 to 2009, and mean summer and winter temperatures for each tract in each year. This approach assumed that no variable changing differentially over time across space other than seasonal temperatures confounded the association. results: For each interquartile range (2 μg/m³) increase in annual PM_2.5, there was a 3.0% [95% confidence interval (CI): 0.2, 5.9%] increase in all natural-cause mortality for the whole population, with similar results for people > 65 years old [3.5% (95% CI: 0.1, 6.9%)] and people ≤ 65 years old [3.1% (95% CI:-1.8, 8.2%)]. The mean summer temperature and the mean winter temperature in a census tract significantly modified the effects of long-term exposure to PM_2.5 on mortality. We observed a higher percentage increase in mortality associated with PM_2.5 in census tracts with more blacks, lower home value, or lower median income. conclusions: Under the assumption of the difference-in-differences approach, we identified a causal effect of long-term PM_2.5 exposure on mortality that was modified by seasonal temperatures and ecological socioeconomic status.