Daily mortality and PM2.5 sources in Denver estimated using only particle-phase or total semi-volatile organic compounds (SVOCs)

Abstract

Background: Source apportionment approaches have provided insights into chemical classes of fine particle matter (PM2.5) responsible for adverse health effects. The Denver Aerosol Sources and Health (DASH) study used the Positive Matrix Factorization (PMF) to identify chemical classes of PM2.5 in the Denver area. Aim: To compare estimated health effects of PM2.5 components with those of chemical classes identified by PMF based on particle-phase or total semi-volatile organic compounds (SVOCs). Methods: Daily measurements of four PM2.5 chemical components (elemental and organic carbon (EC and OC), sulfate, and nitrate), as well as a large number of organic compounds, were obtained over a 32 month period at a single monitoring station located in populated residential area in Denver. PMF factors were estimated based on particle-phase SVOCs only and total SVOCs adding estimated gas-phase SVOCs. The two approaches resulted in eight (Inorganic Ion, n alkane, summer odd n alkane, light n alkane polycyclic aromatic hydrocarbon (PAH), medium alkane alkanoic acid, PAH, winter methoxyphenol, and EC sterane) and seven (n alkane, light SVOC, PAH, summer ECOC, EC sterane, sulfate, and nitrate) factors, respectively. Daily death counts of cardiovascular and respiratory diseases were obtained from the Denver death certificate records. We used generalized additive models to estimate relative risks of mortality at lags 0 to 3 for each of PM2.5 components and PMF factors adjusting for daily humidity and temperature, long-term temporal trend, and day of week. Results: None of four PM2.5 components was associated with daily mortality. With the total SVOC source factors, the PAH factor was associated with cardiovascular mortality at lag 0; with both PMF approaches, n alkanes were associated with respiratory mortality at lag 3. Conclusions: Use of source apportionment allowed us to identify PM2.5 mortality associations that were not found using only single PM2.5 components.