Background. Recent studies of long-term exposure to ambient air pollution suggest that particulate matter in urban air is a risk factor for mortality. However, these studies are potentially biased because they have not taken into account spatial processes. This thesis examines the effect of exposure to sulfate air pollution when place-specific covariates, scale and spatial autocorrelation are controlled for. Methods. This thesis builds upon the American Cancer Society (ACS) Study by introducing place-specific covariates into the original Cox models. In addition, a new two-stage analysis using both metropolitan and county scale covariates is employed in the analysis of the ACS data. Finally a method of filtering out spatial autocorrelation is applied to the data. Results. For all-cause mortality, the original ACS Study found a RR of 1.15 (95% confidence interval of 1.09, 1.22) from sulfates. This was reduced to 1.06 (0.99, 1.13) by the inclusion of the place-specific covariate population change. Using the two-stage method to allow for clustering within metropolitan areas, the RR from sulfates increased to 1.25 (1.13, 1.37). When regional trends were controlled for, the RR decreased to 1.19 (1.06, 1.34). When this metropolitan data was filtered to remove spatial autocorrelation, the RR decreased again to 1.09 (1.01, 1.19). However, when counties were used instead the RR increased to 1.50 (1.30, 1.73). Conclusions. Taking into account spatial processes affects the observed association between ambient sulfate air pollution and mortality. That the sulfate effect is larger and more robust to the inclusion of place-specific covariates at the county scale provides further evidence that ambient sulfate air pollution is a population health hazard.
| Identifer | oai:union.ndltd.org:uottawa.ca/oai:ruor.uottawa.ca:10393/6412 |
| Date | January 2001 |
| Creators | Willis, Alette. |
| Contributors | Krewski, Daniel, |
| Publisher | University of Ottawa (Canada) |
| Source Sets | Université d’Ottawa |
| Detected Language | English |
| Type | Thesis |
| Format | 218 p. |
Page generated in 0.002 seconds