Benefits of better air: pinpointing effects on subpopulations. (Science Selections).

Reducing air pollution confers health benefits to the population as a whole, but researchers at the Harvard School of Public Health have found a way to predict benefits that may accrue to particular subpopulations, such as lower-income individuals and minorities, who suffer from higher rates of illnesses affected by air pollution [EHP 110:1253-1260]. This information could help federal and state policy makers further refine debates over the costs and benefits of proposed air pollution controls.

Cost-benefit analyses typically use a conventional unstratified approach, applying the same relative risks to all individuals in an at-risk age group, and assuming baseline disease or health care utilization rates to be uniform across large geographic areas. To better quantify health benefits associated with implementing Best Available Control Technologies (BACT) for reducing emissions, Jonathan Levy and colleagues have developed a model to estimate health benefits for different demographic groups across small-scale geographic areas. Theirs is one of the first studies to attempt to capture benefits for specific subpopulations.

The team modeled health benefits from the hypothetical installation of BACT at five older power plants located within a 50-mile radius of Washington, D.C. They focused on the effects of reducing emissions of primary fine particulate matter and secondary sulfate and nitrate particles formed through emissions of sulfur dioxide and nitrogen oxide, respectively.

The team performed atmospheric modeling using CALPUFF, a regional-scale model recommended by the U.S. Environmental Protection Agency for long-range transport modeling. They modeled concentration reductions at small geographic scales--census tracts for areas within 100 kilometers of Washington and counties outside of that--to better show the impacts that such reductions would have on smaller subpopulations.

The team then established a grid encompassing a 400-kilometer radius around Washington. For the population contained within this grid, they evaluated three health end points: premature mortality, cardiovascular-related hospital admissions for the elderly, and asthma-related emergency room visits for children. Premature mortality was stratified by subjects' educational level, cardiovascular-related hospital admissions were stratified by diabetic status and age, and asthma-related emergency room visits were stratified by race and age.

The team analyzed outcomes both by using a conventional unstratified approach and by considering susceptible populations. Using the conventional approach, they estimated that implementing BACT would result in 210 fewer deaths per year in the target area, with approximately 25% of the benefits accruing to individuals with less than a high school education. However, the susceptibility model predicted that 51% of the estimated mortality benefits would accrue to that subpopulation--more than double the prediction of the conventional model. Similarly, the conventional model showed only 13% of health benefits from pollution controls accruing to diabetics, whereas the susceptibility model showed 54% of the benefits accruing to diabetics. Finally, the conventional model estimated 140 fewer pediatric asthma emergency room visits per year, with 27% of those benefits accruing to African-American children. The stratified model estimated 160 fewer visits per year, with 64% of the benefits going to African-American children.

Levy and colleagues write, "Although our ability to characterize subpopulations is constrained by the available information, our analysis demonstrates that incorporation of susceptibility information significantly affects demographic and geographic patterns of health benefits and enhances our understanding of individuals likely to benefit from emission controls." According to the researchers, the influence of the susceptibility assumptions on the distribution of benefits highlights the need for more epidemiological studies targeting high-risk subpopulations.