The role of air pollution in the relationship between a heat stress index and human mortality in Toronto

In this study we considered confounding from air pollutants and chronological variables in the relation between humidex, a summer temperature and humidity index, and nonaccidental mortality, from 1980-1996 in Toronto, Canada. Changes in the risk of death by age group, gender, and combined cardiac-respiratory cause of death were estimated for both 1 degreesC and 50-95th percentile increases in humidex using a generalized additive linear model. With air pollution terms in the models, relative risk (RR) point estimates narrowly exceeded 1.0 for all groups. Humidex effects were most apparent for females (RR = 1.006, 95% CI = 1.004-1.008 per 1 C humidex and RR = 1.089, 95% CI = 1.058-1.121 for 50th to 95th percentile humidex). When air pollution was omitted from the model, RR in the 50-95th percentile analysis increased less than 1.71% for all groups except females, for which RR decreased 1.42%. Differences in RR per 1 degreesC humidex were all less than 0.12%. Confidence intervals narrowed slightly for all groups investigated. Heat stress has a statistically significant, yet minimal impact on Toronto populations, and air pollution does appear to have a small, but consistent confounding effect on humidex effect estimates. (C) 2003 Elsevier Science (USA). All rights reserved.