The role of respiratory disease as a contributing cause of short-term non-respiratory mortality associations with ambient particle mass

Many time-series studies have reported associations between cardiovascular mortality (reported as the primary cause of death) and morbidity, and increased ambient particle levels. However, the mechanism(s) of these associations are not yet understood. In this thesis, pollution-associated effect estimates for individuals who died from non-respiratory primary causes who also had contributing respiratory causes listed on their death certificates were compared to those not having contributing respiratory conditions. In order to assess the short-term associations between ambient thoracic particulate matter (≤10 micrometers in aerodynamic diameter, PM₁₀) and mortality, Poisson time series analysis was used adjusting for weather effects, day-of-week patterns, and seasonal influences on day of death. Acute associations were modeled, during the years 1985–1994, in 5 U.S. cities (i.e., Chicago, IL; Detroit, MI; St. Louis, MO; Houston, TX; and Minneapolis St. Paul, MN) that had periods of daily or continuous PM₁₀ data. While New York City had only every-6th-day PM₁₀ data, it had a large daily number of deaths. The combined results for the 5 cities indicate that those with contributing respiratory disease had higher estimated relative risks [95% confidence intervals] per inter-quartile range of PM₁₀, as compared to those without contributing respiratory disease. This was the case for circulatory deaths (relative risk = 1.036 [1.012–1.061] versus 1.017 [1.006–1.027]), and cancer deaths (relative risk = 1.028 [0.995–1.062] versus 1.002 [0.990–1.015]). A similar pattern of association, with much stronger contrast and effect size estimates, was observed in New York City, particularly among those ≥75 years for circulatory (relative risk = 1.066 [1.027–1.106] versus 1.022 [1.008–1.035]), and cancer deaths (relative risk = 1.129 [1.041–1.225] versus 1.025 [1.000–1.050]). In New York City, the contrast in effect estimates was also more pronounced among those that had an infectious co-existing respiratory condition at the time of death (e.g., pneumonia), versus a chronic one (e.g., COPD), for both underlying circulatory and cancer deaths (particularly among those a 75 years). The results of this thesis support the causality of the association between respiratory disease and ambient particulate matter, and also suggest that past studies have under-estimated the contribution of respiratory disease to overall pollution-mortality associations.