Search

Cohort study designs are often used to assess the association between community-based ambient air pollution concentrations and health outcomes, such as mortality, development and prevalence of disease, and pulmonary function. Typically, a large number of subjects are enrolled in the study in each of a small number of communities. Fixed-site monitors are used to determine long-term exposure to ambient pollution. The association between community average pollution levels and health is determined after controlling for risk factors of the health outcome measured at the individual level (i.e., smoking). We present a new spatial regression model linking spatial variation in ambient air pollution to health. Health outcomes can be measured as continuous variables (pulmonary function), binary variables (prevalence of disease), or time-to-event data (survival or development of disease). The model incorporates risk factors measured at the individual level, such as smoking, and at the community level, such as air pollution. We demonstrate that the spatial autocorrelation in community health outcomes, an indication of not fully characterizing potentially confounding risk factors to the air pollution--health association, can be accounted for through the inclusion of location in the deterministic component of the model assessing the effects of air pollution on health or through a distance-decay spatial autocorrelation function in the stochastic component of the model, or both. We present a statistical approach that can be implemented for very large cohort studies. Our methods are illustrated with an analysis of the American Cancer Society cohort to determine whether the prevalence of heart disease is associated with concentrations of sulfate particles. From a statistical point of view, it appears that a location surface in the deterministic component of the model was preferred to a distance-decay autocorrelation structure in the model's stochastic component.

The estimation of health benefits from reductions in ambient air pollution requires information both about the relationship between concentrations and health effects and about the ultimate value of health outcomes. In this commentary, I discuss significant issues and uncertainties that must be addressed for both elements. Critical issues for concentration-response estimation include the shape of the concentration-response curve at current and projected ambient levels, the relevant exposure period, the existence of high-risk subpopulations, and the ability to generalize across study settings. Health outcomes can be valued using either health-based measures like quality-adjusted life years or economic measures like value of statistical life; the choice of measure has a significant effect on the magnitude and distribution of health benefits. Analysts estimating health benefits for ultimate benefit-cost applications should provide results using multiple valuation frameworks and multiple assumptions about health evidence to facilitate interpretation and to determine whether policy decisions depend on these assumptions.

Air pollution and premature death are important public health concerns. Analyses have repeatedly demonstrated that airborne particles are associated with increased mortality and estimates have been used to forecast the impact on life expectancy. In this analysis, we draw upon data from the American Cancer Society (ACS) cohort and literature on utility-based measures of quality of life in relation to health status to more fully quantify the effects of air pollution on mortality in terms of quality-adjusted life expectancy. The analysis was conducted within a decision analytic model using Monte Carlo simulation techniques. Outcomes were estimated based on projections of the Canadian population. A one-unit reduction in sulfate air pollution would yield a mean annual increase in Quality-Adjusted Life Years (QALYs) of 20,960, with gains being greater for individuals with lower educational status and for males compared to females. This suggests that the impact of reductions in sulfate air pollution on quality-adjusted life expectancy is substantial. Interpretation of the results is unclear. However, the potential gains in QALYs from reduced air pollutants can be contrasted to the costs of policies to bring about such reductions. Based on a tentative threshold for the value of health benefits, analysis suggests that an investment in Canada of over 1 billion dollars per annum would be an efficient use of resources if it could be demonstrated that this would reduce sulfate concentrations in ambient air by 1 microg/m(3). Further analysis can assess the efficiency of targeting such initiatives to communities that are most likely to benefit.

The American Cancer Society (ACS) Study and its reanalysis are built upon in order to examine the impact of scale on the observed relationship between sulfates and mortality. The limitations of the original ACS Study (Pope et al., 1995) and the reanalysis of this study (Krewski et al., 2000) are discussed; while the latter dealt with some issues in using ecological data, it did not address scale. Next, the article outlines the county-scale study, the methods used to aggregate data, and the two-stage analysis used to derive relative risk (RR). Finally, the results of working at the county scale are compared with those obtained by the reanalysis team using larger metropolitan areas. Less than half of the cohort used in the metropolitan study were used at the county scale because of the limited availability of sulfate monitors and because five-digit ZIP codes more accurately assigned individuals to geographical areas. Therefore, the county data should be considered as new and not as a reorganization of the original data set. The reanalysis and the county studies should be considered as two separate studies that took different scales as their basic organizing principle. The RR of all-cause mortality from sulfate exposure at the county scale was 1.50 (1.30, 1.73) compared with 1.25 (1.13, 1.37) at the metropolitan scale; for cardiopulmonary mortality, the RR was 1.75 (1.48, 2.08) at the county scale compared with 1.29 (1.15, 1.46) at the metropolitan scale. Because lung cancer mortality was low in some counties, the two-stage random effects model became unstable. At the county scale, the RR from sulfates was more robust to the inclusion of ecologic covariates. Other place-specific ecologic covariates were either insignificant or barely significant (with a lower 95% confidence limit near 0.99 or 1.00) when included in the two-stage regression model for all-cause mortality with sulfates. Moreover, no ecologic covariate changed the RR of all-cause mortality from sulfates by 25% or more. Both population change and unemployment rate affected the RR for cardiopulmonary mortality from sulfate exposure by 25% or more in the county-scale analysis. However, when these two variables were entered into a multiple covariate analysis, the RR from sulfates decreased but remained strongly significant. Sulfur dioxide was not an important covariate at the county scale. Thus, at the county scale, long-term exposure to sulfates appears to be more strongly associated with increased risk of all-cause and cardiopulmonary mortality than previously indicated by the ACS study and its reanalysis.