JOURNAL ARTICLE
RESEARCH SUPPORT, U.S. GOV'T, NON-P.H.S.
TECHNICAL REPORT
Add like
Add dislike
Add to saved papers

Causal Inference Methods for Estimating Long-Term Health Effects of Air Quality Regulations.

INTRODUCTION: The regulatory and policy environment surrounding air quality management warrants new types of epidemiological evidence. Whereas air pollution epidemiology has typically informed previous policies with estimates of exposure-response relationships between pollution and health outcomes, new types of evidence can inform current debates about the actual health impacts of air quality regulations. Directly evaluating specific regulatory strategies is distinct from and complements estimating exposure-response relationships; increased emphasis on assessing the effectiveness of well-defined regulatory interventions will enhance the evidence supporting policy decisions. The goal of this report is to provide new analytic perspectives and statistical methods for what we refer to as "direct"-accountability assessment of the effectiveness of specific air quality regulatory interventions. Toward this end, we sharpened many of the distinctions surrounding accountability assessment initially raised by the HEI Accountability Working Group (2003) through discussion, development, and deployment of statistical methods for drawing causal inferences from observational data. The methods and analyses presented here are unified in their focus on anchoring accountability assessment to the estimation of the causal consequences of well-defined actions or interventions. These analytic perspectives are discussed in the context of two direct-accountability case studies pertaining to four different links in the so-called chain of accountability, the related series of events leading from the intervention to the expected outcomes (see Preface; HEI Accountability Working Group 2003).

METHODS: The statistical methods described in this report consist of both established methods for drawing causal inferences from observational data and newly developed methods for assessing causal accountability. We have sharpened the analytic distinctions between studies that directly evaluated the effectiveness of specific policies and those that estimated exposure-response relationships between pollution and health. We emphasized how a potential-outcomes paradigm for causal inference can elevate policy debates by means of more direct evidence of the extent to which complex regulatory interventions affect pollution and health outcomes. We also outlined the potential-outcomes perspective and promoted its use as a means to frame observational studies as approximate randomized experiments. Our newly developed methods for assessing causal accountability draw on propensity scores, principal stratification, causal mediation analysis, spatial hierarchical models, and Bayesian estimation. The first case study made use of health outcomes among approximately four million Medicare beneficiaries living in the Western United States to estimate the causal health impacts of areas designated as being in nonattainment for particulate matter ≤10 μm in aerodynamic diameter (PM10*) according to the 1987 National Ambient Air Quality Standards (NAAQS). The second case study focused on developing and testing our new, advanced methodology for multipollutant accountability assessment by examining the extent to which sulfur dioxide (SO2) scrubbers on coal-fired power plants causally affect emissions of SO2, nitrogen oxides (NO(x)), and carbon dioxide (CO2) as well as the extent to which emissions reductions mediate the causal effect of a scrubber on ambient concentrations of PM2.5. Both case studies were anchored in our compilation of national, linked data on ambient air quality monitoring, weather, population demographics, Medicare hospitalization and mortality outcomes, continuous-emissions monitoring for electricity-generating units (EGUs) in power plants, and a variety of regulatory control interventions. The resulting database has unprecedented accuracy and granularity for conducting the types of accountability assessments presented in this report. A key component of our work was the creation of tools to help distribute our linked database and to facilitate reproducible research.

RESULTS: In the first case study, we focused on illustrating the most fundamental features of a causal-inference perspective on direct-accountability assessment. The results indicated that all-cause Medicare mortality and respiratory-related hospitalization rates were causally reduced in areas designated as nonattainment for PM10 during 1990 to 1995 compared with the rates that would have occurred without the designation. In the second case study, which examined power-plant emissions and illustrated our newly developed statistical methods, the results indicated that the presence of an SO2 scrubber causally reduced ambient PM2.5 and that this reduction was mediated almost entirely through causal reductions in SO2 emissions. The results were interpreted in light of the well-documented relationships between scrubbers, power-plant emissions, and PM2.5.

CONCLUSION: By grounding accountability research in a potential-outcomes framework and applying our new methods to our collection of national data sets, we were able to provide additional sound evidence of the health effects of long-term, large-scale air quality regulations. This additional, rigorous evidence of the causal effects of well-defined actions augments the existing body of research and ensures that the highest-level epidemiological evidence will continue to support regulatory policies. Ultimately, our research contributed to the evidence available to support to the U.S. Environmental Protection Agency (U.S. EPA) and other stakeholders for incorporating health outcomes research into policy development.

Full text links

We have located links that may give you full text access.
Can't access the paper?
Try logging in through your university/institutional subscription. For a smoother one-click institutional access experience, please use our mobile app.

For the best experience, use the Read mobile app

Mobile app image

Get seemless 1-tap access through your institution/university

For the best experience, use the Read mobile app

All material on this website is protected by copyright, Copyright © 1994-2024 by WebMD LLC.
This website also contains material copyrighted by 3rd parties.

By using this service, you agree to our terms of use and privacy policy.

Your Privacy Choices Toggle icon

You can now claim free CME credits for this literature searchClaim now

Get seemless 1-tap access through your institution/university

For the best experience, use the Read mobile app