Do Differences in Exposures Explain the Observed Heterogeneity in PM2.5 - Mortality Associations across U.S. cities?
Citation:
Baxter, L., K. Dionisio, P. Pradeep, AND L. Neas. Do Differences in Exposures Explain the Observed Heterogeneity in PM2.5 - Mortality Associations across U.S. cities? Annual Conference of the International Society of Exposure Science, Research Triangle Park, North Carolina, October 15 - 19, 2017.
Impact/Purpose:
Fine particulate matter has been associated with increased mortality; however, the magnitude of effects can vary in different cities. Baxter’s paper identified city-specific characteristics, such as housing stock age, size, and air conditioning use, that explain some of the differences. Identification of the characteristics will inform ways of reducing people’s exposure to air pollution, thereby reducing their health risk.
Description:
This analysis evaluates exposure factors as potential determinants of the heterogeneity in city-specific associations between PM2.5 and mortality. Exposure factor variables were created based on housing characteristics, commuting patterns, heating fuel usage, and climatic factors data from national surveys. When survey data was not available, air conditioning (AC) prevalence was predicted utilizing machine learning techniques. We examined the heterogeneity in 312 city-specific PM2.5-mortality health effect estimates using inverse variance weighted linear regression with inverse variance weights. Prevalence of central AC predicted via machine learning techniques showed a strong relationship (R2 = 0.78) with the observed prevalence in surveyed cities. The national estimate (0.96% increase in total non-accidental mortality for a 10 µg/m3 increment in PM2.5 at lag 1) decreased by 0.12% (95% confidence interval (CI) -0.27 to 0.03) for an interquartile increase in the predicted prevalence of central AC. The other determinants examined were also found to modify the PM2.5-mortality association to varying degrees: median house size (+0.09%, 95% CI 0.01 to 0.18, for a 1 room increase in the median number of rooms), percentage of heating fueled by oil (+ 0.15%, 95%CI 0.10 to 0.20, for a percentage point increase), and heating degree days (HDD) (+0.31%, 95% CI 0.12 to 0.50, for a 1 day increase). In our analysis, the health impact of PM2.5 on mortality increases in cities with larger homes, more heating fueled by oil, more HDD, and less central AC.