Relationships of mortality with the fine and coarse fractions of long-termambient PM10 concentrations in nonsmokers

In a cohort of 6338 'Califomia Seventh - day Adventists, we previously obse rved for males associations between long-term concentrations of particulate matter (PM) with an aerodynamic diameter less than 10 mum (PM10) and 15-ye ar mortality due to all natural causes (ANC) and lung cancer (LC) listed as underlying causes of death and due to nonmalignant respiratory disease lis ted as either the underlying or a contributing (CRC) cause of death. The pu rpose of this analysis was to determine whether these outcomes were more st rongly associated with the fine (PM2.5) or the coarse (PM2.5-10) fractions of PM10. For participants who lived near an airport (n = 3769), daily PM2.5 concentrations were estimated from airport visibility, and on a monthly ba sis, PM2.5-10 concentrations were calculated as the differences between PM1 0 and PM2.5. Associations between ANC, CRC, and LC mortality (1977-1992) an d mean PM10, PM2.5, and PM2.5-10 concentrations at study baseline ( 1973-19 77) were assessed using Cox proportional hazards models. Magnitudes of the PM10 associations for the males of this subgroup were similar to those for the males in the entire cohort although not statistically significant due t o the smaller numbers. In single-pollutant models, for an interquartile ran ge (IQR) increase in PM10 (29.5 mug/m(3)), the rate ratios (RRs) and 95% co nfidence intervals (CI) were 1.15 (0.94, 1.41) for ANC, 1.48 (0.93, 2.34) f or CRC, and 1.84 (0.59, 5.67) for LC. For an IQR increase in PM2.5 (24.3 mu g/m(3)), corresponding RRs (95% CI) were 1.22 (0.95, 1.58), 1.64 (0.93, 2.9 0), and 2.23 (0.56, 8.94), and for an IQR increase in PM2.5-10 (9.7 mug/m(3 )), corresponding RRs (95% CI) were 1.05 (0.92, 1.20), 1.19 (0.88, 1.62), a nd 1.25 (0.63, 2.49), respectively. When both PM2.5 and PM2.5-10 were enter ed into the same model, the PM2.5 estimates remained stable while those of PM2.5-10 decreased. We concluded that previously observed associations of l ong-term ambient PM10 concentration with mortality for males were best expl ained by a relationship of mortality with the fine fraction of PM10 rather than with the coarse fraction of PM10.