Health & Environmental Research Online (HERO)


Print Feedback Export to File
42850 
Technical Report 
Mortality and air pollution for Santa Clara County, California, 1989-1996, In: Revised analyses of time-series studies of air pollution and health. Special report. 
Fairley, D 
2003 
Health Effects Institute 
Boston, MA 
97-106 
The data are reanalyzed using generalized additive models (GAMs*) with stricter convergence criteria. The results are essentially unchanged with coefficients differing by at most +/- 0.5 standard error (SE) from the original and relative risks differing by at most +/- 0.01 SE. Various inference methods are compared. The method comparing GAM coefficients with standard errors generated from generalized linear models (GLMs) produces P values similar to simulations. The S-Plus ANOVA (analysis of variance) feature often gave conservative results. As found previously, a statistically significant relation existed between daily nonaccidental mortality and every criteria pollutant, either on the same day or lag one. Particulate matter (PM) less than 2.5 ým in diameter (PM2.5) and nitrate (NO3) predominate when included in models with carbon monoxide (CO), nitrogen dioxide (NO2), and sulfate (SO4). Coarse-fraction PM, less than 10 ým in diameter (PM10), was not statistically significant. A new ozone (O3) variable--the daily number of parts per billion (ppb)-hours greater than a 60-ppb threshold (o3ppbgt60)--was found to have a statistically significant relation with nonaccidental mortality even when included in a regression jointly with PM2.5 or NO3 and was also significantly related to cardiovascular mortality. 
10/18/2004