Exploration of PM mass, source, and component-related factors that might explain heterogeneity in daily PM-mortality associations across the United States.

Multi-city epidemiologic studies examining short-term (daily) differences in fine particulate matter (PM) provide evidence of substantial spatial heterogeneity in city-specific mortality risk estimates across the United States. Because PM is a mixture of particles, both directly emitted from sources or formed through atmospheric reactions, some of this heterogeneity may be due to regional variations in PM toxicity. Using inverse variance weighted linear regression, we examined change in percent change in mortality in association with 24 "exposure" determinants representing three basic groupings based on potential explanations for differences in PM toxicity - size, source, and composition. Percent changes in mortality for the PM-mortality association for 313 core-based statistical areas and their metropolitan divisions over 1999-2005 were used as the outcome. Several determinants were identified as potential contributors to heterogeneity: all mass fraction determinants, vehicle miles traveled (VMT) for diesel total, VMT gas per capita, PM ammonium, PM nitrate, and PM sulfate. In multivariable models, only daily correlation of PM with PM and long-term average PM mass concentration were retained, explaining approximately 10% of total variability. The results of this analysis contribute to the growing body of literature specifically focusing on assessing the underlying basis of the observed spatial heterogeneity in PM-mortality effect estimates, continuing to demonstrate that this heterogeneity is multifactorial and not attributable to a single aspect of PM.