Time-series analysis of health effects of particulate matter (PM) components in Detroit, MI and Seattle, WA

Background: Recent toxicological and epidemiological studies have shown associations between PM and adverse health effects, but responsible PM chemical component(s) are not known. U.S. EPA's Chemical Speciation Network (CSN) data have provided an opportunity to examine the role of PM chemical constituents, however CSN's every 3rd- or 6th-day sampling schedule limits the multi-day effects analysis and yields relatively small sample sizes. This project addresses these issues by conducting a time-series analysis of daily PM components and mortality in Seattle, WA and Detroit, MI, which have different air pollution mixtures. .

Methods: Daily FRM PM2.5 filters for Seattle and Detroit for the years 2002-2004 were obtained, and analyzed for trace elements using XRF and elemental carbon (EC) by light reflectance. A Poisson time-series model was used to estimate percent excess deaths for cardiovascular diseases (CVD) per inter-quartile-range (IQR) of PM2.5 components and gaseous pollutants at lag 0 through 3 days, adjusting for seasonal cycles, immediate and delayed temperature effects, and day-of-week.

Results: In Detroit, PM2.5, EC, and NO2 were significantly associated with warm season CVD mortality (2.7%, 2.4%, and 2.6% excess risks per IQR at lag-2 day, respectively). In Seattle, there were significant cold season associations for these three pollutants, as well as for sulfur, nickel, potassium and silica.