Effect Modification by Air Conditioning on Particulate Matter Health Effect Estimates

Several studies have reported an association between short-term exposure to particulate matter (PM) and increased risk of health effects, including mortality and hospital admission. The effect estimates, however, differ by region and season, and less is known about what factors cause this variation. One hypothesis is that the use of air conditioning (AC) could alter PM exposure patterns, which might explain the heterogeneous effect estimates. We investigated whether community-specific PM health effect estimates are associated with prevalence of AC using national U.S. datasets. We applied Bayesian hierarchical modeling to explore whether AC prevalence modified the associations between: (1) PM₁₀ and mortality (n=84 communities); (2) PM_2.5 and cardiovascular hospitalizations for those >65 years (n=168 counties); and (3) PM_2.5 and respiratory hospitalizations for those >65 years (n=168 counties). We found that, in general, communities with higher AC prevalence had lower PM effects. Associations were statistically significant for cardiovascular hospitalizations and central AC. An additional 20% of households with central AC was associated with a 42.5% decrease in PM_2.5 cardiovascular hospitalization effect estimates. Central AC prevalence explained 17% of the between-community variability in PM_2.5 cardiovascular hospitalization effect estimates. Other associations did not reach to significant level. A link between AC and PM health effect estimates is plausible due to the lower penetration of outdoor pollutants into indoors with use of central AC compared to homes using open windows for cooling. Our findings provide evidence for the hypothesis that AC prevalence explains part of the variation in PM health effect estimates.