Analysis of PM Components and Mortality Risk

To examine the possible role of PM components (i.e., transition metals, ions, and crustal soil tracers) on the city-to-city variation of PM₁₀ mortality risk estimates, we analyzed the association between the key FPM components from the FPM speciation network and the NMMAPS PM₁₀ daily mortality risk estimates. The speciation data were obtained from the U.S. EPA Air Quality System (AQS) for the years 2000–2003 (U.S. EPA 2003 ). The NMMAPS PM₁₀ mortality risk estimates (updated estimates using generalized linear modeling) for the 90 largest U.S. MSAs (for the time-series analysis that was conducted for 1987–1994) were obtained from the JHSPH Internet-based Health and Air Pollution Surveillance System (IHAPSS) website (JHSPH 2003 ). Although there were more than 40 FPM species, we focused on the 16 key components that were most closely associated with major source categories: aluminum, arsenic, Cr, copper, elemental carbon, Fe, manganese, Ni, nitrate, organic carbon, lead, selinium, silicon, sulfate, V, and zinc. First, for each FPM monitor, quarterly averages were computed from 24-hr average values (of at least every 6th-day schedule) when > 50% of scheduled samples were available. Second, an annual average for each FPM monitor was computed (but only when the four complete quarter averages were available). Third, the annual average values were then averaged across available monitors for each MSA. The resulting MSA-averaged FPM component values were then matched with the 60 NMMAPS MSAs that had FPM speciation data. Most of the annual speciation data were highly skewed. Therefore, we examined both raw and log-transformed data. The PM₁₀ mortality risk estimates (expressed as percent excess deaths per 10-μg/m³ increase in PM₁₀) were then regressed on each of the FPM components, with weights based on the SE of the PM₁₀ risk estimates.