Final Report: Statistical Models for Estimating the Health Impact of Air Quality Regulations

EPA Grant Number: R833622
Title: Statistical Models for Estimating the Health Impact of Air Quality Regulations
Investigators: Dominici, Francesca , White, Ronald H. , Samet, Jonathan M.
Institution: Harvard T.H. Chan School of Public Health
EPA Project Officer: Hahn, Intaek
Project Period: July 1, 2007 through September 30, 2010 (Extended to September 30, 2011)
Project Amount: $500,000
RFA: Development of Environmental Health Outcome Indicators (2006) RFA Text |  Recipients Lists
Research Category: Air Quality and Air Toxics , Health Effects , Health

Objective:

In this proposal, we will develop statistical models for estimating consequences of regulation for environmental health outcome indicators, such as the number of premature deaths or morbidity events prevented by regulation. The specific aims of this proposal are to:

A.1: Assess the chronology of the implementation of the NAAQS for PM10 and the corresponding attainment and non-attainment status for all of the U.S. counties for the period 1987-2006.

A.2: Develop exposure indicators: estimates of the association between regulation and longer term trends in PM. Specifically, we propose to develop Bayesian hierarchical models for time series data that use all of the available air pollution data to:

  • A.2.1: estimate longer term national, regional and county-specific trends in PM10 for the period 1987-2006, and in fine and coarse PM  for the period 1999-2006.
  • A.2.2:  estimate the association between county-specific trends in PM10 and nonattainment status accounting for changes in population demographics, industrial activities, and energy demand.

A.3: Develop outcome indicators: estimates of the associations between longer term exposure to PM and human health on a national scale. More specifically, we plan to develop methods for estimating:

  • A.3.1: cross-sectional associations between longer term exposure to PM on mortality and morbidity accounting for individual-level risk factors and area-level confounders.
  • A.3.2: month-to-month variations in mortality rates and average PM concentrations over the previous year. These methods will provide evidence as to whether counties having steeper decline in PM levels relative to the PM national trend will also have steeper declines in mortality rates relative to the mortality national trend.

A.4: Develop environmental health outcome indicators: estimates of the number of adverse health events prevented by regulation. Specifically, we plan to develop a Bayesian statistical approach for integrating Aims 1-3. We will estimate the percentage decrease in adverse health events prevented by regulation and quantify its uncertainty. We plan to monitor this environmental health outcome indicator over time on a national scale.

 

Summary/Accomplishments (Outputs/Outcomes):

This has been a very productive grant. We have addressed Aims 1 and 2 in the paper by Rava et al. (2010) entitled Does Attainment Status for the PM10 National Air Ambient Quality Standard Change the Trend in Ambient Levels of Particulate Matter? In this paper, we focused on assessing the impact of air quality regulations on ambient levels of air pollution. We considered two groups of counties: the first group (attainment counties) includes counties that in 1991 were designated as in attainment or unclassifiable with respect to the 1987 NAAQS and maintained their status through 2006; the second group (no attainment counties) includes counties that in 1991 were designated as nonattainment and were subsequently re-designated as in attainment (see Figure 3 below from Rava et al 2010). We hypothesized that if air pollution control programs adopted to meet the NAAQS are effective in reducing air pollution levels, counties in the no-attainment group will experience a sharper decrease in PM10 levels than counties in the attainment group. To provide evidence to support this hypothesis, Bayesian hierarchical models were developed for estimating (1) the yearly percentage change in ambient PM10 levels for 100 counties and the entire United States during the period 1987-2007, and (2) the change in PM10 ambient levels in counties in the nonattainment group compared to counties in the attainment group.
 
 
We found statistically significant evidence of variability across counties in trends of PM10 concentrations. We also found strong evidence that counties transitioning from nonattainment to attainment status during the period 1987-2007 experienced a sharper decline in PM10 when compared to counties that were always in attainment.
 
Beside the implementation of air pollution controls included in the SIP, other factors could be responsible for the decline in PM10 trend. Changes in long-term particulate matter levels in ambient air can be affected by changes in multiple factors, such as population demographics, industrial activity, and energy demand. In this analysis, we only accounted for county-level population socio-economic status and did not assess the impact of any other potential confounders. Inclusion of time-varying area-level characteristics did not greatly change air pollution trend estimates. Another limitation of our study is that we did not take into account information on county-specific SIPs, for example date of implementation and type of SIP for each study area. We also excluded from our analyses counties in attainment with the NAAQS at the beginning of the study that transited to the nonattainment status. In this paper, we only considered the designation as attainment or nonattainment counties at two time points: when each county was classified as in attainment or nonattainment with respect to the 1987 NAAQS during 1991 and 2007. As a further analysis, the model used in the study could allow for random changing point corresponding to every change in the designation status. These analyses, also, could be repeated routinely for future revisions of the NAAQS. Assessment of the relationship between implementation of national and state-level air pollution control measures to changes in ambient air quality levels and ultimately to health outcomes can provide important information regarding the efficacy of air quality management policies. The statistical methods here proposed could be further applied to assess the impact of air pollution control measures on public health.
 
 
To address Aim 3, we have developed new statistical methods for estimating the chronic air pollution effects using spatio-temporal information (Janes et al 2007, Greven et al 2011). There is now substantial observational evidence that chronic exposure to particulate air pollution is associated with premature death in urban populations. Estimates of the magnitude of these effects derive largely from cross-sectional comparisons of adjusted mortality rates among cities with varying pollution levels. Such estimates are potentially confounded by other differences among the populations correlated with air pollution, for example socio-economic factors. An alternative approach is to study co-variation of particulate matter and mortality across time within a city, as has been done in investigations of acute exposures. In either event, observational studies like these are subject to confounding by unmeasured variables. Therefore, the ability to detect such confounding is a high priority. In Janes et al. 2007 and Greven et al. 2011, we have developed and applied methods for decomposing the air pollution exposure into components with variation at distinct temporal, spatial, and time by space scales. More specifically, in Greven et al. 2011, starting from a proportional hazard model, we derive a Poisson regression model and estimate two regression coefficients: the “global” coefficient, corresponding to the temporal scale and measuring the association between national trends in pollution and mortality (); and the “local” coefficient, corresponding to the space by time scale and measuring the association between location-specific trends in pollution and mortality adjusted by the national trends (). (See Figures 1 and 2 from Greven et al. web supplemental material to provide an example). Absent unmeasured confounders and given valid model assumptions, the scale-specific coefficients should be similar; substantial differences in these coefficients constitute a basis for questioning the model. We derive a backfitting algorithm to fit our model to very large spatio-temporal data sets.
 
 
 
 
We apply our methods to the Medicare Cohort Air Pollution Study (MCAPS), which includes individual-level information on time of death and age on a population of 18.2 million for the period 2000-2006. See Table 1 from Greven et al. 2011.
 
Results based on the local coefficient indicate no significant change in life expectancy for any reduction in PM2.5. Results based on the global coefficient indicate that a 10 mg/m3 reduction in the yearly national average of PM2.5 is associated with a significant increase in life expectancy of more than 3 years (See Figure 4 from Greven et al. 2011). Although the coefficient based on national trends in PM2.5 and mortality is likely to be confounded by other variables trending on the national level, confounding of the local coefficient by unmeasured factors cannot be ruled out. We use additional survey data available for a subset of the data to investigate sensitivity of results to the inclusion of additional covariates, but both coefficients remain largely unchanged.
 
To address Aim 4, we develop innovative methods for causal inference aimed at estimating the causal effects of an air quality intervention on health (see Zigler et al. 2012). Methods for causal inference regarding health effects of air quality regulations are met with unique challenges because (1) changes in air quality are intermediates on the causal pathway between regulation and health, (2) regulations typically affect multiple pollutants on the causal pathway towards health, and (3) regulating a given location can affect pollution at other locations, that is, there is interference between observations. We propose a principal stratification method designed to examine causal effects of a regulation on health that are and are not associated with causal effects of the regulation on air quality. A novel feature of our approach is the accommodation of a continuously scaled multivariate intermediate response vector representing multiple pollutants. Furthermore, we use a spatial hierarchical model for potential pollution concentrations and ultimately use estimates from this model to assess validity of assumptions regarding interference.
 
To characterize causal effects of the 1991 nonattainment designations on pollution and health, we focus on 3-year average ambient concentration of PM10 and O3 during the period 1999–2001 and on all-cause mortality in 2001 among Medicare beneficiaries living in the vicinity of an air pollution monitor. To this end, we compile data from several national sources. From the EPA Air Quality System (AQS) database, we obtain 3-year average ambient concentrations of PM10 and O3 during the preregulation period (1987–1989) and the 3 years leading up to the target date for attainment of the NAAQS (1999– 2001). From the Center for Medicare and Medicaid Services Medicare enrollee file, we obtain all-cause mortality information in 2001 for all Medicare enrollees living within 6 miles of a pollution monitor, as well as basic demographic characteristics such as age, gender, and ethnicity. From the 2000 U.S. Census, we obtain county-level demographic characteristics such as population size and income characteristics. From the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance System, we obtain county-level smoking rates in 2000. Table 1 from Zigler et al. 2012 summarizes the available data. The observational units of the analysis are the locations of air pollution monitors in AQS having data available for PM10 and/or O3 during either the pre- or postregulation years.
 
 
 
 
Next, we determined all Medicare enrollees living in U.S. zip codes having geographic centroids within a 6-mile radius of a pollution monitor and assigned these enrollees pollution exposure measured from that monitor. We restrict the analysis to monitor locations in the western United States having at least 50 Medicare enrollees because almost all initial PM10 nonattainment areas fell in this region. The resulting data set consists of ambient pollution measurements at 362 pollution monitor locations (of which 200 lie in regulated counties), county-level characteristics on 140 counties, and basic characteristics and mortality information for 6 926 338 Medicare beneficiaries. Figure 1 from Zigler et al. 2012 displays the monitor locations.
 
 
Our analysis of the CAAA estimated that 1991 nonattainment designations for PM10 did causally
reduce Medicare mortality in 2001, and that there are important causal pathways through which this effect occurred without affecting average ambient concentrations of PM10 or O3 during 1999–2001. More specifically, we found that the estimated overall average causal effect of the regulation program on mortality was 1.76 fewer deaths per 1000 Medicare beneficiaries after adjusting for the aforementioned covariates with a Poisson regression model (posterior mean [sd] deaths/1000: 63.41 [0.29] vs. 65.17 [0.34]).
 
Summary of findings
 
  • In Rava et al. 2010, we found statistically significant evidence of variability across counties in trends of PM10 concentrations. We also found strong evidence that counties transitioning from nonattainment to attainment status during the period 1987-2007 experienced a sharper decline in PM10 when compared to counties that were always in attainment.
  • In Greven et al. 2012, we developed new statistical methods to estimate the long-term effects of PM on mortality. We applied our methods to the Medicare Cohort Air Pollution Study (MCAPS), which includes individual-level information on time of death and age on a population of 18.2 million for the period 2000-2006. We found that associations between local trends in air pollution and local trends in life expectancy were not significant. Results based on the global coefficient (that is the association between national trend in PM and national trend in LE) indicate that a 10 mg/m3 reduction in the yearly national average of PM2.5 is associated with a significant increase in life expectancy of more than three years.
  • In Zigler et al. 2012, we developed innovative methods for causal inference to estimate the causal effects of the 1991 nonattainment designations on pollution and health. Here, we focus on 3-year average ambient concentration of PM10 and O3 during the period 1999–2001 and on all-cause mortality in 2001 among Medicare beneficiaries living in the vicinity of an air pollution monitor. The resulting data set consists of ambient pollution measurements at 362 pollution monitor locations (of which 200 lie in regulated counties), county-level characteristics on 140 counties, and basic characteristics and mortality information for 6,926,338 Medicare beneficiaries. We found that the 1991 nonattainment designations for PM10 did causally reduce Medicare mortality in 2001, and that there are important causal pathways through which this effect occurred without affecting average ambient concentrations of PM10 or O3 during 1999–2001. More specifically, we found that the estimated overall average causal effect of the regulation program on mortality was 1.76 fewer deaths per 1,000 Medicare beneficiaries after adjusting for the aforementioned covariates with a Poisson regression mode.

 

Conclusions:

Assessing the public health consequences of air quality interventions is of paramount importance. In this grant, we have developed new methods and conducted large epidemiological studies to estimate the public health benefits of air pollution regulatory actions. We estimated the effects of the regulatory actions on ambient air pollution levels (Rava et al. 2010), on health (Greven et al. 2011), and the effect of the regulatory action on health that is associated and dissociated with the effects of regulatory actions on air pollution (Zigler et al. 2012). Overall, the levels of air pollution are continuing to decline and this is mostly due to air pollution control polices. However, even at the smaller levels, ambient pollution continues to adversely affect public health.


Journal Articles on this Report : 21 Displayed | Download in RIS Format

Other project views: All 23 publications 23 publications in selected types All 23 journal articles
Type Citation Project Document Sources
Journal Article Barr CD, Dominici F. Cap and trade legislation for greenhouse gas emissions: public health benefits from air pollution mitigation. JAMA-Journal of the American Medical Association 2010;303(1):69-70. R833622 (Final)
R832417 (Final)
R832417C001 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Abstract: JAMA-Abstract
    Exit
  • Journal Article Barr CD, Diez DM, Wang Y, Dominici F, Samet JM. Comprehensive smoking bans and acute myocardial infarction among Medicare enrollees in 387 US counties:1999–2008. American Journal of Epidemiology 2012;176(7):642-648. R833622 (Final)
    R832417 (Final)
    R834798 (2013)
    R834798 (2014)
    R834798 (Final)
    R834894 (2012)
    R834894 (2013)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: AJE-Full Text HTML
    Exit
  • Abstract: AJE-Abstract
    Exit
  • Other: AJE-Full Text PDF
    Exit
  • Journal Article Bell ML, Ebisu K, Peng RD, Walker J, Samet JM, Zeger SL, Dominici F. Seasonal and regional short-term effects of fine particles on hospital admissions in 202 US counties, 1999-2005. American Journal of Epidemiology 2008;168(11):1301-1310. R833622 (Final)
    R832417 (2008)
    R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: OUP-Full Text PDF
    Exit
  • Abstract: OUP-Abstract & Full Text HTML
    Exit
  • Journal Article Bobb JF, Dominici F, Peng RD. A Bayesian model averaging approach for estimating the relative risk of mortality associated with heat waves in 105 U.S. cities. Biometrics 2011;67(4):1605-1616. R833622 (Final)
    R832416 (Final)
    R832417 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Abstract: Wiley Online-Abstract
    Exit
  • Other: USDA-Abstract
  • Journal Article Chang HH, Peng RD, Dominici F. Estimating the acute health effects of coarse particulate matter accounting for exposure measurement error. Biostatistics 2011;12(4):637-652. R833622 (Final)
    R832416 (Final)
    R832417 (Final)
    R832417C001 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: Oxford Journals-Full Text HTML
    Exit
  • Abstract: Oxford Journals-Abstract
    Exit
  • Other: Oxford Journals-Full Text PDF
    Exit
  • Journal Article Greven S, Dominici F, Zeger S. An approach to the estimation of chronic air pollution effects using spatio-temporal information. Journal of the American Statistical Association 2011;106(494):396-406. R833622 (2009)
    R833622 (Final)
    R832416 (Final)
    R832417 (Final)
    R834798 (2013)
    R834798 (2014)
    R834798 (Final)
    R834894 (2012)
    R834894 (2013)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: ResearchGate-Abstract & Full Text PDF
    Exit
  • Abstract: Taylor & Francis-Abstract
    Exit
  • Journal Article Janes H, Dominici F, Zeger S. Partitioning evidence of association between air pollution and mortality. Epidemiology 2007;18(4):427-428. R833622 (2008)
    R833622 (2009)
    R833622 (Final)
    R832417 (2008)
    R832417 (Final)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
  • Full-text: Epidemiology-Full Text HTML
    Exit
  • Abstract: Epidemiology-Article Preview
    Exit
  • Other: Epidemiology-Full Text PDF
    Exit
  • Journal Article Janes H, Dominici F, Zeger SL. Trends in air pollution and mortality: an approach to the assessment of unmeasured confounding. Epidemiology 2007;18(4):416-423. R833622 (Final)
    R830548 (Final)
    R832417 (Final)
    R832417C001 (2006)
    R832417C001 (2007)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
  • Abstract from PubMed
  • Full-text: Epidemiology-Full Text HTML
    Exit
  • Abstract: Epidemiology-Abstract
    Exit
  • Other: Epidemiology-Full Text PDF
    Exit
  • Journal Article Janes H, Dominici F, Zeger S. On quantifying the magnitude of confounding. Biostatistics 2010;11(3):572-582. R833622 (Final)
    R832417 (Final)
    R832417C001 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: Biostatistics-Full Text HTML
    Exit
  • Abstract: Biostatistics-Abstract
    Exit
  • Other: Biostatistics-Full Text PDF
    Exit
  • Journal Article Peng RD, Chang HH, Bell ML, McDermott A, Zeger SL, Samet JM, Dominici F. Coarse particulate matter air pollution and hospital admissions for cardiovascular and respiratory diseases among Medicare patients. JAMA-Journal of the American Medical Association 2008;299(18):2172-2179. R833622 (2008)
    R833622 (2009)
    R833622 (Final)
    R832417 (2008)
    R832417 (Final)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: JAMA-Full Text HTML
    Exit
  • Abstract: JAMA-Abstract
    Exit
  • Other: JAMA-Full Text PDF
    Exit
  • Journal Article Peng RD, Dominici F, Welty LJ. A Bayesian hierarchical distributed lag model for estimating the time course of risk of hospitalization associated with particulate matter air pollution. Journal of the Royal Statistical Society Series C--Applied Statistics 2009;58(1):3-24. R833622 (Final)
    R832417 (2008)
    R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
  • Full-text: JohnsHopkins-Prepublication Full Text PDF
    Exit
  • Abstract: Wiley-Abstract
    Exit
  • Journal Article Peng RD, Bell ML, Geyh AS, McDermott A, Zeger SL, Samet JM, Dominici F. Emergency admissions for cardiovascular and respiratory diseases and the chemical composition of fine particle air pollution. Environmental Health Perspectives 2009;117(6):957-963. R833622 (Final)
    R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
    R833863 (2009)
    R833863 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: EHP-Full Text PDF
  • Abstract: EHP-Abstract & Full Text HTML
  • Journal Article Peng RD, Bobb JF, Tebaldi C, McDaniel L, Bell ML, Dominici F. Toward a quantitative estimate of future heat wave mortality under global climate change. Environmental Health Perspectives 2011;119(5):701-706. R833622 (Final)
    R832417 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: EHP-Full Text HTML
  • Abstract: EHP-Abstract
  • Other: EHP-Full Text PDF
  • Journal Article Rava M, White RH, Dominici F. Does attainment status for the PM10 National Air Ambient Quality Standard change the trend in ambient levels of particulate matter? Air Quality, Atmosphere & Health 2011;4(2):133-143. R833622 (2009)
    R833622 (Final)
    R832417 (Final)
    R832417C001 (Final)
  • Abstract: Springer-Abstract
    Exit
  • Other: ResearchGate-Abstract
    Exit
  • Journal Article Venturini S, Dominici F, Parmigiani G. Generalized quantile treatment effect: a flexible Bayesian approach using quantile ratio smoothing. Bayesian Analysis 2015;10(3):523-552. R833622 (Final)
  • Full-text: Euclid-Full Text-PDF
    Exit
  • Abstract: Euclid-Abstract
    Exit
  • Other: Johns Hopkins University-Abstract
    Exit
  • Journal Article Wang C, Parmigiani G, Dominici F. Bayesian effect estimation accounting for adjustment uncertainty. Biometrics 2012;68(3):661-671. R833622 (Final)
    R832416 (Final)
    R832417 (Final)
    R834798 (2012)
    R834798 (2013)
    R834798 (2014)
    R834798 (Final)
    R834798C005 (2012)
    R834798C005 (2013)
    R834798C005 (Final)
    R834894 (2012)
    R834894 (2013)
  • Abstract from PubMed
  • Full-text: ResearchGate-Full Text PDF
    Exit
  • Abstract: Wiley-Abstract
    Exit
  • Other: Harvard-Full Text PDF
    Exit
  • Journal Article Welty LJ, Peng RD, Zeger SL, Dominici F. Bayesian distributed lag models: estimating effects of particulate matter air pollution on daily mortality. Biometrics 2009;65(1):282-291. R833622 (Final)
    R832417 (2008)
    R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
  • Abstract from PubMed
  • Full-text: JohnsHopkins-Full Text PDF
    Exit
  • Abstract: Wiley-Abstract
    Exit
  • Journal Article Zeger SL, Dominici F, McDermott A, Samet JM. Mortality in the Medicare population and chronic exposure to fine particulate air pollution in urban centers (2000–2005). Environmental Health Perspectives 2008;116(12):1614-1619. R833622 (2008)
    R833622 (2009)
    R833622 (Final)
    R832417 (2008)
    R832417 (2009)
    R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: EHP-Full Text PDF
  • Abstract: EHP-Abstract & Full Text HTML
  • Journal Article Zhou Y, Dominici F, Louis TA. A smoothing approach for masking spatial data. Annals of Applied Statistics 2010;4(3):1451-1475. R833622 (Final)
    R830548 (Final)
  • Full-text: Euclid-Full Text PDF
    Exit
  • Abstract: Euclid-Abstract
    Exit
  • Other: ResearchGate-Abstract and Full Text PDF
    Exit
  • Journal Article Zigler CM, Dominici F, Wang Y. Estimating causal effects of air quality regulations using principal stratification for spatially correlated multivariate intermediate outcomes. Biostatistics 2012;13(2):289-302. R833622 (Final)
    R832417 (Final)
    R834798 (2012)
    R834798 (2013)
    R834798 (2014)
    R834798C005 (2012)
    R834798C005 (2013)
    R834894 (2012)
    R834894 (2013)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: Oxford Journals-Full Text HTML
    Exit
  • Abstract: Oxford Journals-Abstract
    Exit
  • Other: Oxford Journals-Full Text PDF
    Exit
  • Journal Article Zigler CM, Belin TR. A Bayesian approach to improved estimation of causal effect predictiveness for a principal surrogate endpoint. Biometrics 2012;68(3):922-932. R833622 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Abstract: Wiley - Abstract
    Exit
  • Other: USDA-Abstract
  • Supplemental Keywords:

    Air pollution interventions, confounding, mortality, Medicare;

    Progress and Final Reports:

    Original Abstract
  • 2008 Progress Report
  • 2009 Progress Report
  • 2010
  • 2011