You are here:
Improved population exposure factors in the meta-analysis of air pollution health effects
Citation:
Baxter, L., B. Schultz, H. A. OZKAYNAK, AND L. M. NEAS. Improved population exposure factors in the meta-analysis of air pollution health effects. Presented at International Society of Environmental Epidemiology (ISEE), Pasadena, CA, October 12 - 16, 2008.
Impact/Purpose:
research results
Description:
Numerous time-series studies have reported significant associations between ambient PM2.5 levels and increased mortality and morbidity. A recent mortality study conducted by Franklin et al. 2007 in 27 U.S. cities has reported significant heterogeneity among city-specific effect estimates. This city-to-city heterogeneity may reflect epidemiologic effect modification related to varied behaviors ofindividuals, residential characteristics, and other community-specific factors. For instance, differences in home air exchange rates can affect personal-ambient relationships, since individuals spend a majorityoftheirtime indoorsandambientPM2.5penetratesfairlyefficientlyindoors. In addition, community co-pollutant concentrations (e.g., 0 3, N02, CO) may also be a factor, depending on the correlations among the various exposure metrics for these pollutants. In this presentation, we evaluate whether variations in the personal-ambient relationships and community characteristics can explain part ofthe variability seen in the PM2.5 health effect estimates among different cities. We use the health effects estimated from the Franklin et al. 2007 multi-city study that examined the association between PM2.5 and both total and cause-specific mortality in 27 U.S. communities between 1997 and 2002. Using inverse variance weighted least-squares regression, we examine the effect ofhome ventilation and 0 3 to analyze whether differences in exposure factors may account for some ofthe variability observed across the study sites. Home ventilation will be estimated using the Lawrence Berkeley National Laboratory (LBNL) equation for normalizedleakage(NL),an indicator ofthe tightness ofthe building, determined bythe year built and the area ofhouses. Using the distribution ofthese factors, we will determine a distribution ofNLs for each community. Using the medians for each community, we calculated that the maximum NL was 30% higher than the minimum NL. Preliminary results indicate that differences in community-specific housing characteristics may explain part ofthe heterogeneity seen among city-specific PM2.5 mortality effect estimates. We outline additional studies to improve our understanding of various factors which could have a measurable influence on the magnitude ofthe PM¬mortality relationships. Although this work was reviewed by EPA and approvedfor publication. it may not necessarily reflect official Agencypolicy.