DAILY PM(10) MORTALITY ASSOCIATIONS - AN INVESTIGATION OF AT-RISK SUBPOPULATIONS/

Recent time-series epidemiological studies have reported significant a ssociations between daily air pollution and mortality. These studies t ypically report a short-term excess increase in deaths as the fraction al increase of total (nonaccidental) deaths per unit of air pollutant. The relative risk (RR) calculated for the total population in these s tudies, however, may underestimate the risk for the most sensitive sub population(s) at risk. In this study, race, gender, and cause-specific counts of daily mortality in Cook County, Illinois (which encompasses the city of Chicago) during 1985-1990 were analyzed to determine if t here was any heterogeneity in air pollution/weather/mortality associat ions across these various population subcategories. Seasonal cross-cor relations between mortality and environmental variables first were exa mined to identify appropriate lag structures. Of the pollution variabl es considered - particulate matter less than 10 mu m (PM(10)), ozone, carbon monoxide, sulfur dioxide, and visual range-derived extinction c oefficient both PM(10) and ozone showed significant associations with same-day and next-day mortality. The Poisson regression models employe d included seasonal cycles (sine/cosine series), square and linear ter ms of lagged temperature, trend line, day-of-week dummy variables, and the average of the same day's and previous day's PM(10) or ozone. The RR for total nonaccidental mortality per 100 mu g/m(3) increase in PM (10) was 1.05 (95% CI: 1.03-1.08). The respiratory (RR 1.14; 95% CI: 1 .04-1.25) and cancer (RR = 1.12; 95% CI: 1.06-1.18) categories showed higher estimates than the circulatory category (RR = 1.03; 95% CI: 0.9 8-1.07), while the residual of the total from these three categories s howed no association with PM(10) (RR = 1.01; 95% CI: 0.95-1.08). Among the race- and gender-specific categories, black (African-American) fe males showed the highest RRs for the total (RR = 1.11; 95% CI: 1.03-1. 21), respiratory (RR = 1.31; 95% CI: 0.98-1.75), and cancer (RR = 1.25 ; 95% CI: 1.07-1.46) mortality categories. Neither ozone nor hot tempe rature showed such cause-specificity in mortality associations. Cold t emperature lagged by two days was a significant predictor of circulato ry and respiratory mortality. This study suggests the importance of ra ce- and gender-specific analysis. The greater mortality risk To Chicag o's black women from exposure to urban air pollution indicated by this finding should be rested in other metropolitan areas.