2009 Progress Report: Estimation of the Risks to Human Health of PM and PM Components

EPA Grant Number: R832417C001
Subproject: this is subproject number 001 , established and managed by the Center Director under grant R832417
(EPA does not fund or establish subprojects; EPA awards and manages the overall grant for this center).

Center: Johns Hopkins Particulate Matter Research Center
Center Director: Samet, Jonathan M.
Title: Estimation of the Risks to Human Health of PM and PM Components
Investigators: Dominici, Francesca , Bell, Michelle L. , Peng, Roger D. , Zeger, Scott L.
Current Investigators: Peng, Roger D. , Bell, Michelle L. , Dominici, Francesca , Samet, Jonathan M.
Institution: The Johns Hopkins University , Yale University
EPA Project Officer: Chung, Serena
Project Period: October 1, 2005 through September 30, 2010
Project Period Covered by this Report: August 1, 2008 through July 31,2009
RFA: Particulate Matter Research Centers (2004) RFA Text |  Recipients Lists
Research Category: Health Effects , Air

Objective:

In this project we will develop and apply statistical methods to national data sources to:  1) carry out multi-site time series studies for estimating short-term effects of PM and PM components on mortality and morbidity  (Phase I); 2) carry out cohort-studies for estimating long-term effects of PM and PM components in susceptible populations (Phase II); and 3)  assess coherence of evidence from bioassays and epidemiological studies on PM toxicity and susceptibility; and explore linkages of sources of harmful PM components to human health risks  (Phase III). 
 
By providing individual-level health data for the entire US population of elderly, the National Medicare Cohort will allow us to take full advantage of all existing and future air quality databases on PM and its characteristics. This project will address the following objectives of the Center on a national scale: 1) mapping risks of PM and PM constituents to human health across the United States; 2) using the maps to identify a sampling frame of locations with contrasting higher and lower risks; 3) carrying out more refined epidemiological studies to estimate further the risks of the more toxic particles to susceptible individuals.

Progress Summary:

Progress is described in relation to the original specific aims for this project.

  1. Multi-site time series studies for estimating short-term effects of PM and PM components on mortality and hospitalization (Phase I)
    1. Characterize spatial and temporal variability of PM2.5, PM2.5 components, and gaseous pollutants across the US to identify locations for a more in-depth PM characterization for biological studies;

      Progress to date:

      We have characterized the spatial and temporal variability of PM2.5 components in the U.S. with the objective of identifying components for assessment in epidemiological studies (Bell et al. 2007). We constructed a database of 52 PM2.5 component concentrations for 203 U.S. counties for 2000 to 2005 from the EPA’s Speciation Trends Network (STN). First, we described the challenges inherent to analysis of a national PM2.5 chemical composition database. Second, we identified components that substantially contribute to and/or co-vary with PM2.5 total mass. Third, we characterized the seasonal and regional variability of targeted components. We identified substantial seasonal and geographical variation in PM2.5 chemical composition. Only seven of the 52 components contributed 1% or more to total mass for yearly or seasonal averages (NH4+, elemental carbon, organic carbon, NO3-, Si, Na+, and SO4=). The strongest correlations with PM2.5 total mass were with NH4+ (yearly and each season), organic carbon (winter and autumn), NO3- (winter), and SO4= (yearly, spring, and summer), with particularly high correlations for NH4+ and SO4= in summer. Components that co-varied with PM2.5 total mass, based on daily detrended data, were NH4+, organic carbon, elemental carbon, NO3-, SO4=, and bromine. As a basis for further investigation, we identified a subset of PM2.5 components that should be further investigated to determine whether: 1) their daily variation is associated with daily variation of health indicators; and 2) their seasonal and regional patterns can explain the regional and seasonal heterogeneity in PM10 and PM2.5 health risks.

    2. Develop and apply statistical methods for multi-site time series studies for estimating short-term effects of PM10, PM2.5, and coarse particles (PM10- PM2.5) on hospitalization and mortality from the National Medicare Cohort, over the entire year and by season for the largest 300 counties in the USA with PM data available. Identify locations with the largest and smallest short-term effects of PM on the health indicators (“cold and hot spots”) to identify locations for a more in-depth PM characterization for biological studies;

      Progress to date:

      Multi-site time series studies of PM2.5 and hospital admissions (1999-2005)

      As reported in Dominici et al. (2006) we have conducted a multi-site time series study to estimate risks of cardiovascular and respiratory hospital admissions associated with short-term exposure to PM2.5 for Medicare enrollees and to explore variation of risks across regions. We assembled a national database comprising daily time-series data for the period 1999-2005 on hospital admission rates for cardiovascular and respiratory diseases and injuries (as a control), ambient PM2.5 levels, and temperature and dew-point for 204 US urban counties. Daily hospital admission rates were constructed from the Medicare National Claims History Files (NCHF). Our study population included 11.5 million Medicare enrollees living on average 5.9 miles from a PM2.5 monitor. We found that short-term exposure to PM2.5 increases hospital admission risks for cardiovascular and respiratory diseases. By linking Medicare, pollutant, and weather data, we created a national database for continued research that can be updated and analyzed repeatedly to track the health risks of air pollution.

      Multi-site time series studies of PM2.5 and hospital admissions, stratified by season and geographical regions (1999-2005)

      In Bell et al 2008, we investigated whether short-term effects of fine particulate matter (PM2.5) on risk for cardiovascular and respiratory hospitalizations among the elderly vary by region and season in 202 U.S. counties for 1999-2005. We fit three types of time-series models to provide evidence for: (1) consistent PM effects across the year; (2) different PM effects by season; and (3) smoothly varying PM effects throughout the year. We found statistically significant evidence of seasonal and regional variation in PM effect estimates. Respiratory disease effect estimates were highest in winter with a 1.05% (95% posterior interval 0.29, 1.82%) increase in hospitalizations per 10 μg/m3 increase in same day PM2.5. Cardiovascular diseases estimates were also highest in winter with a 1.49% (1.09, 1.89%) increase in hospitalizations per 10 μg/m3 increase in same day PM2.5, with associations also observed in other seasons. The strongest evidence of a relationship between PM2.5 and hospitalizations was in the Northeast for both respiratory and cardiovascular diseases. Heterogeneity of PM2.5 effects on hospitalizations may reflect seasonal and regional differences in emissions and in particles’ chemical constituents. Our results can help guide the development of hypotheses and further epidemiological studies on potential heterogeneity in the toxicity of constituents of the PM mixture.

      Multi-site time series studies of Coarse PM (PM10-2.5) and hospital admissions

      As reported in Peng et al. (2008) we have estimated the risk of hospital admissions for cardiovascular (CVD) and respiratory diseases (RESP) associated with exposure to PM10-2.5, controlling for PM2.5. We ablssemed a database for 108 US counties for the period 1999 to 2005 with a study population of approximately 12 million Medicare enrollees (≥ 65) living on average 9 miles from collocated pairs of PM10 and PM2.5 monitors. We found a positive and statistically significant association between same-day concentrations of PM10-2.5 and CVD admissions. When adjusted by same-day concentrations of PM2.5, this association was positive but not statistically significant. The associations between PM10-2.5, and RESP admissions, adjusted and unadjusted by PM2.5, were positive but not statistically significant. The effect of PM10-2.5 on CVD admissions was statistically significantly higher in more urban counties compared to less urban counties. We also found continuing evidence of a positive and statistically significant association between same day concentrations of PM2.5 and CVD admissions in the most recent Medicare data.

      Rationale for Sampling Strategy for Selecting US Locations for In-depth PM2.5 Characterization and PM collection

      Underlying the general approach of the Johns Hopkins PM Center was the proposition that results of epidemiological analyses could be used to identify locations having greater and lesser risk to human health associated with PM exposure. The analyses carried out for the National Morbidity, Mortality and Air Pollution Study (NMMAPS) had shown geographic heterogeneity in the effect of PM on mortality and PM characteristics are known to vary across the country. In the PM Center application we proposed to analyze national mortality and Medicare morbidity data to create a sampling frame for selection of locations for Phase II monitoring and PM collection. We recognize that alternative sampling frames could be used, e.g., based on location or source mix; however, for identification of PM characteristics that could contribute to risk, the health risk-based approach provides the most directly relevant sampling frame. We have completed analyses to identify locations for which we have the greatest confidence that they lie at the higher or lower end of the risk distribution.

      Analytic Approach

      We use time-series data for the period 1999-2005 and for the same 203 US counties included in the study by Dominici et al. (2006) on PM2.5 and hospital admissions in Medicare enrollees. We focus our analyses on lag 0 PM2.5 concentrations because relative risk estimates are generally larger at this lag. As health outcomes, we use hospital admissions for all cardiovascular disease combined and hospital admissions for all respiratory diseases combined. Specifically, hospital admissions for cardiovascular diseases includes five cardiovascular outcomes: heart failure (ICD 9, 428), heart rhythm disturbances (426-427); cerebrovascular events (430-438); ischemic heart disease (410-414, 429) and peripheral vascular disease (440-448), Hospital admissions for respiratory diseases includes two respiratory outcomes: chronic obstructive pulmonary disease (COPD) (490-492) and respiratory infections (464-466 and 480-487).

      We then

      1. fit Bayesian two-stage normal-normal models to estimate short-term effects of PM2.5 on hospital admissions for cardiovascular and respiratory hospital admissions outcomes;
      2. map the Bayesian estimates of the county-specific relative rates and quantify evidence of spatial heterogeneity
      3. use the maps to identify geographical locations where there are larger and smaller estimates of short-term effects of PM2.5 on hospital admissions. Specifically:
        1. e group the 203 US counties in five large geographical regions North East (NE), Mid West (MW) ,South East (SE), North West (NW), South West (SW). We anticipate that these five large geographical regions have very different mixtures of PM2.5 chemical composition;
        2. North East (NE), South East (SE), North West (NW), South West (SW) and Upper MidWest (UM). Based on discussions with EPA, we have also replicated these analyses with one additional fifth region, making up the Central US. We anticipate that these four large geographical regions have very different mixtures of PM2.5 chemical composition;
        3. within each region, we fit a two-stage Bayesian model to the 203 US counties and we calculate the posterior t-statistics. A posterior t-statistic is defined as the ratio of the county-specific posterior mean divided by the county-specific posterior standard deviations of the relative risk;
        4. within each region and for each outcome, we rank the posterior t-statistics. A location having a posterior t-statistic above the 75-th percentile, consistently across the health outcomes, is identified as a location with a high risk. A location having a posterior t-statistic below the 25-th percentile, consistently across the outcomes, is identified as a location with a low risk.
        5. these locations must have at least one STN site (ESAC recommendation on July 9 2007)

          The selected locations are:

          • NW: King WA
          • NE: Bronx, NY (or Baltimore MD) and Allegheny PA
          • SW: Sacramento, CA and Maricopa AZ
          • SE: Harris, TX and Pinellas, FL
          • MW: Jefferson KY and Hennepin, MN
      4. Characterize the PM2.5 chemical composition of the selected locations

      For each location, we calculated the t statistics for the PM2.5-associated risk for the outcome, along with a categorization of the relative ranking of the estimate. We identified locations that have generally lower estimates across all outcomes while others have generally higher estimates. Within each of the regions, we identify locations in the strata of lower and higher risk. We plan to use these locations for particle collection and characterization in Project #2 so as to assure representativeness across the country.

    3. Develop and apply statistical methods for multi-site time series studies that take into account exposure measurement error for: 1) estimating short-term effects of PM2.5 components on hospitalization and mortality from the National Medicare Cohort; and investigating whether spatial and seasonal variability of the PM2.5 components explain spatio-temporal variability of short-term effects of PM10, (PM10- PM2.5) and PM2.5 on hospitalization and mortality estimated in A.2

      Progress to date:

      New methods for adjustment uncertainty

      We are developing new methods to account for “adjustment uncertainty” in time series studies of air pollution and health. Specifically in the paper by Crainicenau et al. (2007) we propose a general statistical framework for handling adjustment uncertainty in exposure effect estimation for a large number of confounders, a specific implementation, and associated visualization tools. We also show that when the goal is to estimate an exposure effect accounting for adjustment uncertainty, Bayesian Model Averaging (BMA) can fail to estimate the true exposure effect and over- or under-estimate its variance.

      Bayesian hierarchical distributed lag model for estimating the time course of hospitalization risk associated with particulate matter air pollution

      In a paper by Peng et al. 2007, we developed new methods for estimating the distributed lag function in time series studies of air pollution and health. We have proposed a Bayesian hierarchical distributed lag model that integrates information from national databases with prior knowledge of the time course of hospitalization risk after an air pollution episode. We have applied the model to a database of particulate matter air pollution monitoring and health information on 6.3 million enrollees of the US Medicare system living in 94 counties covering the years 1999–2002. We have obtained estimates of the distributed lag functions relating fine particulate matter pollution to hospitalizations for both ischemic heart disease and acute exacerbation of chronic obstructive pulmonary disease (COPDAE). We found that the effect of an increase in fine particulate matter on ischemic heart disease is immediate, with the bulk of hospitalizations occurring within 2 days of an air pollution increase, while for COPDAE the effect of fine particulate matter appears to be distributed over a week or more.

      Estimating trends in the short-term effects of PM10 on mortality

      In the paperby Dominici et al. 2007, we have evaluated change in the short-term effect of airborne particles over a period of increasingly stringent regulation that might have changed the chemical composition and toxicity of the airborne particles. We use updated data and methods of the National Mortality Morbidity Air Pollution Study (NMMAPS) to estimate national average relative rates of the effects of PM10 on all-cause, cardiovascular and respiratory mortality, and other-cause mortality for the period 1987-2000. We have estimated national average relative rates of the effects of PM2.5 (<2.5 µm) on all-cause mortality for the period 1999-2000. We found strong evidence that lag 1 exposures to PM10 and PM2.5 continue to be associated with increased mortality. We also found a weak indication that the lag 1 effects of PM10 on mortality declined over the period 1987-2000 and that this decline mostly occurred in the eastern U.S. The methodology presented here can be used to track the health effects of air pollution routinely on regional and national scales.

      Multi-site time series studies of PM2.5 and hospital admissions, stratified by season and geographical regions (1999-2005)

      In Bell et al 2008, we investigated whether short-term effects of fine particulate matter (PM2.5) on risk for cardiovascular and respiratory hospitalizations among the elderly vary by region and season in 202 U.S. counties for 1999-2005. We fit three types of time-series models to provide evidence for: (1) consistent PM effects across the year; (2) different PM effects by season; and (3) smoothly varying PM effects throughout the year. We found statistically significant evidence of seasonal and regional variation in PM effect estimates. Respiratory disease effect estimates were highest in winter with a 1.05% (95% posterior interval 0.29, 1.82%) increase in hospitalizations per 10 μg/m3 increase in same day PM2.5. Cardiovascular diseases estimates were also highest in winter with a 1.49% (1.09, 1.89%) increase in hospitalizations per 10 μg/m3 increase in same day PM2.5, with associations also observed in other seasons. The strongest evidence of a relationship between PM2.5 and hospitalizations was in the Northeast for both respiratory and cardiovascular diseases. Heterogeneity of PM2.5 effects on hospitalizations may reflect seasonal and regional differences in emissions and in particles’ chemical constituents. Our results can help guide the development of hypotheses and further epidemiological studies on potential heterogeneity in the toxicity of constituents of the PM mixture.

      Multi-site time series studies of PM2.5 chemical components and hospital admissions (1999-2005)

      In Peng et al 2008, we estimated the associations between daily levels of PM2.5 components and risk of hospital admissions in 119 US urban communities for 12 million Medicare enrollees (aged 65 years or older) using Bayesian hierarchical statistical models. We used a national database comprising daily data for 2000—2006 on hospital admissions rates for cardiovascular and respiratory outcomes, ambient levels of major PM2.5 chemical components (sulfate, nitrate, silicon, elemental carbon, organic carbon matter, sodium and ammonium ions) and weather. In multiple-pollutant models, an interquartile range (IQR) increase in elemental carbon was associated with a 0.80 (95% posterior interval [PI], 0.34, 1.27) percent increase in risk of same day cardiovascular admissions, and an IQR increase in organic carbon matter was associated with a 1.01 (95% PI: 0.04, 1.98) percent increase in risk of respiratory admissions on the same day. Ambient levels of elemental carbon and organic carbon matter, which are generated primarily from vehicle emissions, diesel, and wood burning, were associated with the largest risks of emergency hospitalization across the major chemical constituents of fine particles.

      Bayesian Model Averaging for Clustered data: Imputing Missing PM Data

      In Chang et al (2009) (submitted) we have developed an innovative statistical methodology for imputing missing PM2.5 data when PM10 data from the co-located monitor is available (and viceversa). We assume that the two daily time series of PM10 and PM2.5 from the pair of collocated monitors form a data cluster. The goal is to best predict the missing PM data at the co-located monitor pair in presence of multiple competing prediction models. In the presence of multiple competing models, Bayesian model averaging (BMA) offers a powerful tool to account for model uncertainty and improve prediction. However, the typical application of BMA to clustered data determines model weights (posterior probabilities of the competing models) by comparing how data from all clusters fit each model. In Chang et al 2009 we develop a BMA approach for clustered data that accounts for dfferences in the best fitting models among clusters. This is accomplished by allowing the weights of competing regression models to vary between clusters while borrowing information across clusters in estimating model parameters. The work is motivated by the problem of imputing missing observations in clustered data when the performance of different prediction models varies across clusters. Through simulation and cross-validation studies, we demonstrate that our approach outperforms the standard BMA. Finally, we apply the proposed method to a national dataset of daily ambient particulate matter concentrations between 2003 and 2005. We then estimate the posterior probability of coarse PM nonattainment status for 95 US counties based on the Environmental Protection Agency’s proposed 24-hour standard. .

      Estimating the Acute Health Effects of Coarse ParticulateMatter Accounting for Exposure Measurement Error

      In air pollution epidemiology there is a growing interest in estimating the health effects of coarse particulate matter (PM) with aerodynamic diameter between 2.5 and 10 µm. Coarse PM concentrations can exhibit considerable spatial heterogeneity because the particles travel shorter distances and do not remain suspended in the atmosphere for an extended period of time. In Chang et al (2009) (submitted), we develop a modelling approach for estimating the short-term effects of air pollution in time series analysis when the ambient concentrations vary spatially within the study region. Specifically, our approach quantifies the error in the exposure variable by characterizing, on any given day, the disagreement in ambient concentrations measured across monitoring stations. This is accomplished by viewing monitor-level measurements as error- prone repeated measurements of the unobserved true exposure. Inference is carried out in a Bayesian framework to fully account for uncertainty in the estimation of model parameters. Finally, by using different exposure indicators, we investigate the sensitivity of the association between coarse PM and daily hospital admissions based on a recent national multi-site time series analysis. Among Medicare enrollees from 59 U.S. counties between the period 1999 to 2005, we find a consistent positive association between coarse PM and same-day admission for cardiovascular diseases.

      Diagnosing confounding bias in studies of chronic effects of air pollution on health

      In the paper by Janes H et al. 2007 we have proposed a new method for diagnosing confounding bias under a model that estimates associations between a spatially and temporally varying exposure and health outcome. Specifically, we have decomposed the association into orthogonal components, corresponding to distinct spatial and temporal scales of variation. If the model fully controls for confounding, the exposure effect estimates should be equal at the different temporal and spatial scales. We have shown that the overall exposure effect estimate is a weighted average of the scale-specific exposure effect estimates.

      Using approach, we have estimated the association between monthly averages of fine particles (PM2.5) over the preceding 12 months and monthly mortality rates in 113 US counties from 2000 to 2002. We have decomposed the association between PM2.5 and mortality into 2 components: (1) the association between "national trends" in PM2.5 and mortality; and (2) the association between "local trends," defined as county-specific deviations from national trends. This second component provides evidence as to whether counties having steeper declines in PM2.5 also have steeper declines in mortality relative to their national trends.

      We have found that the exposure effect estimates are different at these 2 spatiotemporal scales, which raises concerns about confounding bias. We conclude that the association between trends in PM2.5 and mortality at the national scale is more likely to be confounded than is the association between trends in PM2.5 and mortality at the local scale. If the association at the national scale is set aside, there is little evidence of an association between 12-month exposure to PM2.5 and mortality.

  2. Cohort studies based on the National Medicare Cohort for estimating longer-term effects of PM and PM composition in susceptible populations and for cause-specific health outcomes (Phase II)
    1. Develop statistical methods for cohort studies for estimating associations between longer-term exposure to PM10, PM2.5, PM10, PM2.5, and PM2.5 components and hospitalization and mortality adjusted by individual-level and area-level confounders;

      Fine Particulate Matter and Mortality: A Comparison of the Six Cities and American Cancer Society Cohorts with a Medicare Cohort

      In the paper by Eftim et al. 2008, using Medicare data, we have assessed the association of PM2.5 with mortality for the same locations included in these studies. We have estimated the chronic effects of PM 2.5 on mortality for the period 2000-2002 using mortality data for cohorts of Medicare participants and average PM 2.5 levels from monitors in the same counties included in the SCS and the ACS. Using Medicare data, which lack information on some potential confounding factors, we estimated risks similar to those in the SCS and ACS, which incorporated more extensive information on individual-level confounders. We propose that the Medicare files can be used to construct cohorts for tracking the longer-term risk of air pollution over time.

      Methods development for estimation of the long-term effects of PM2.5 on mortality and morbidity outcomes is in progress. The challenge of adequately adjusting for individual-level and area-level confounders needs to be addressed.

      Mortality in the Medicare Population and Chronic Exposure to Fine Particulate Air Pollution in Urban Centers (2000-2005)

      Prospective cohort studies constitute the major source of evidence about the mortality effects of chronic exposure to particulate air pollution. Additional studies are needed to provide evidence on the health effects of chronic exposure to particulate matter < 2.5 micrometers in aerodynamic diameter (PM2.5) as most previous studies have focused on Total Suspended Particles (TSP) and on particulate matter < 10 micrometers in aerodynamic diameter (PM10). In Zeger et al (2009), we estimate the relative risk of death in a U.S. population of elderly associated with long-term exposure to PM2.5 by region and age-groups, for the period 2000-2005. By linking fine particulate matter (PM2.5) monitoring data to the Medicare billing claims by zip code of residence of the enrollees, we have developed a new retrospective cohort study, the Medicare Cohort Air Pollution Cohort Study. The study population comprises 13.2 million participants living in 4,568 zip codes having centroids within 6 miles of a PM2.5 monitor. Relative risks adjusted by socioeconomic status and smoking were estimated by fitting log-linear regression models. In the East and Central regions, a 10 µg/m3 increase in six year average of PM2.5 is associated with a 6.8% (95% CI: 4.9 to 8.7%) and 13.2% (95% CI: 9.5 to 16.9) increases in mortality, respectively. We did not find evidence of an association in the West and for persons above 85 years of age. We established a cohort of Medicare participants for investigating air pollution and mortality on longer-term time frames. Chronic exposure to PM2.5 was associated with mortality in the eastern and central regions, but not in the western U.S.

      Diagnosing confounding bias in studies of chronic effects of air pollution on health

      In the paper by Janes H et al. 2007 we have proposed a new method for diagnosing confounding bias under a model that estimates associations between a spatially and temporally varying exposure and health outcome. Specifically, we have decomposed the association into orthogonal components, corresponding to distinct spatial and temporal scales of variation. If the model fully controls for confounding, the exposure effect estimates should be equal at the different temporal and spatial scales. We have shown that the overall exposure effect estimate is a weighted average of the scale-specific exposure effect estimates.

      Using approach, we have estimated the association between monthly averages of fine particles (PM2.5) over the preceding 12 months and monthly mortality rates in 113 US counties from 2000 to 2002. We have decomposed the association between PM2.5 and mortality into 2 components: (1) the association between "national trends" in PM2.5 and mortality; and (2) the association between "local trends," defined as county-specific deviations from national trends. This second component provides evidence as to whether counties having steeper declines in PM2.5 also have steeper declines in mortality relative to their national trends.

      We have found that the exposure effect estimates are different at these 2 spatiotemporal scales, which raises concerns about confounding bias. We conclude that the association between trends in PM2.5 and mortality at the national scale is more likely to be confounded than is the association between trends in PM2.5 and mortality at the local scale. If the association at the national scale is set aside, there is little evidence of an association between 12-month exposure to PM2.5 and mortality.

      Progress to date

      Bayesian Model Averaging for Clustered data: Imputing Missing PM Data

      In Chang et al (2009) (submitted) we have developed an innovative statistical methodology for imputing missing PM2.5 data when PM10 data from the co-located monitor is available (and viceversa). We assume that the two daily time series of PM10 and PM2.5 from the pair of collocated monitors form a data cluster. The goal is to best predict the missing PM data at the co-located monitor pair in presence of multiple competing prediction models. In the presence of multiple competing models, Bayesian model averaging (BMA) offers a powerful tool to account for model uncertainty and improve prediction. However, the typical application of BMA to clustered data determines model weights (posterior probabilities of the competing models) by comparing how data from all clusters fit each model. In Chang et al 2009 we develop a BMA approach for clustered data that accounts for dfferences in the best fitting models among clusters. This is accomplished by allowing the weights of competing regression models to vary between clusters while borrowing information across clusters in estimating model parameters. The work is motivated by the problem of imputing missing observations in clustered data when the performance of different prediction models varies across clusters. Through simulation and cross-validation studies, we demonstrate that our approach outperforms the standard BMA. Finally, we apply the proposed method to a national dataset of daily ambient particulate matter concentrations between 2003 and 2005. We then estimate the posterior probability of coarse PM nonattainment status for 95 US counties based on the Environmental Protection Agency’s proposed 24-hour standard.

      Estimating the Acute Health Effects of Coarse ParticulateMatter Accounting for Exposure Measurement Error

      In air pollution epidemiology there is a growing interest in estimating the health effects of coarse particulate matter (PM) with aerodynamic diameter between 2.5 and 10 µm. Coarse PM concentrations can exhibit considerable spatial heterogeneity because the particles travel shorter distances and do not remain suspended in the atmosphere for an extended period of time. In Chang et al (2009) (submitted), we develop a modelling approach for estimating the short-term effects of air pollution in time series analysis when the ambient concentrations vary spatially within the study region. Specifically, our approach quantifies the error in the exposure variable by characterizing, on any given day, the disagreement in ambient concentrations measured across monitoring stations. This is accomplished by viewing monitor-level measurements as error- prone repeated measurements of the unobserved true exposure. Inference is carried out in a Bayesian framework to fully account for uncertainty in the estimation of model parameters. Finally, by using different exposure indicators, we investigate the sensitivity of the association between coarse PM and daily hospital admissions based on a recent national multi-site time series analysis. Among Medicare enrollees from 59 U.S. counties between the period 1999 to 2005, we find a consistent positive association between coarse PM and same-day admission for cardiovascular diseases.

      Spatial misalignment in time series studiesof air pollution and health data

      Time series studies of environmental exposures often involve comparing daily changes in a toxicant measured at a point in space with daily changes in an aggregate measure of health. Spatial misalignment of the exposure and response variables can bias the estimation of health risk and the magnitude of this bias depends on the spatial variation of the exposure of interest. In air pollution epidemiology, there is an increasing focus on estimating the health effects of the chemical components of particulate matter. One issue that is raised by this new focus is the spatial misalignment error introduced by the lack of spatial homogeneity in many of the particulate matter components. Current approaches to estimating short-term health risks via time series modeling do not take into account the spatial properties of the chemical components and therefore could result in biased estimation of those risks. In Peng et al (2009), we present a spatial-temporal statistical model for quantifying spatial misalignment error and show how adjusted heath risk estimates can be obtained using a regression calibration approach and a two-stage Bayesian model. We apply our methods to a database containing information on hospital admissions, air pollution, and weather for 20 large urban counties in the United States.

      A Spatio-Temporal Approach for Estimating ChronicEffects of Air Pollution

      Estimating the health risks associated with air pollution exposure is of great importance in public health. In air pollution epidemiology, two study designs have been used mainly. Time series studies estimate acute risk associated with short- term exposure. They compare day-to-day variation of pollution concentrations and mortality rates, and have been criticized for potential confounding by time- varying covariates. Cohort studies estimate chronic effects associated with long- term exposure. They compare long-term average pollution concentrations and time-to-death across cities, and have been criticized for potential confounding by individual risk factors or city-level characteristics.

      In Greven et al, we propose a new study design and a statistical model, which use spatio-temporal information to estimate the long-term effects of air pollution exposure on life expectancy. Our approach avoids confounding by time-varying covariates and individual or city-level risk factors. By estimating the increase in life expectancy due to decreases in long-term air pollution concentrations, it provides easily interpretable values for public policy purposes. We develop a suitable backfitting algorithm that permits effcient fitting of our model to large spatio-temporal data sets. We evaluate spatio-temporal correlation in the data and obtain appropriate standard errors. We apply our methods to the Medicare Cohort Air Pollution

      Study, including data on fine particulate matter (PM2.5 ) and mortality for 18.2 million Medicare enrollees from 814 locations in the U.S. during an average of 65 months in 2000-2006.

      QA/QC:

      The investigators, along with the Center PI and the Quality Assurance Manager of the Center have collaborated to develop the Project Quality Assurance Project Plan for Project 3. The Plan was developed and submitted to the PI and QAM in 2007. The QAPP is being implemented at this time, and the investigators have systems in place to identify and correct any QA/QC issues. The Project Director, the Director of the Data Management Core, key data analysts, the Center PI, and the QAM have met periodically and consistently through this past year to define data management policies and assessment procedures associated with secondary data analysis. The Project Director meets monthly with the Project 1 team and the Center PI for a research in progress meeting and QA issues are discussed at that time as part of the QC process. As part of the QAPP development process, the QAM has conducted an internal assessment of the project’s quality control procedures.


Journal Articles on this Report : 25 Displayed | Download in RIS Format

Other subproject views: All 41 publications 36 publications in selected types All 34 journal articles
Other center views: All 89 publications 66 publications in selected types All 64 journal articles
Type Citation Sub Project Document Sources
Journal Article Barr CD, Dominici F. Comment on article by Craigmile et al. Bayesian Analysis 2009;4(1):37-40. R832417C001 (2009)
  • Other: Bayesian Analysis
    Exit
  • Journal Article Bell ML, Peng RD, Dominici F. The exposure-response curve for ozone and risk of mortality and the adequacy of current ozone regulations. Environmental Health Perspectives 2006;114(4):532-536. R832417 (Final)
    R832417C001 (2006)
    R832417C001 (2007)
    R832417C001 (2009)
    R832417C001 (Final)
    R830548 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Journal Article Bell ML, Dominici F, Ebisu K, Zeger SL, Samet JM. Spatial and temporal variation in PM2.5 chemical composition in the United States for health effects studies. Environmental Health Perspectives 2007;115(7):989-995. R832417 (Final)
    R832417C001 (2006)
    R832417C001 (2007)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R830548 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: EHP-Full Text HTML
  • Other: EHP-Full Text PDF
  • Journal Article Bell ML, Kim JY, Dominici F. Potential confounding of particulate matter on the short-term association between ozone and mortality in multisite time-series studies. Environmental Health Perspectives 2007;115(11):1591-1595. R832417 (Final)
    R832417C001 (2007)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R830548 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Journal Article Bell ML, Dominici F. Effect modification by community characteristics on the short-term effects of ozone exposure and mortality in 98 US communities. American Journal of Epidemiology 2008;167(8):986-997. R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: OUP-Full Text HTML
    Exit
  • Abstract: OUP-Abstract
    Exit
  • Other: OUP-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. R832417 (2008)
    R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
    R833622 (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 Bell ML, Ebisu K, Peng RD, Samet JM, Dominici F. Hospital admissions and chemical composition of fine particle air pollution. American Journal of Respiratory and Critical Care Medicine 2009;179(12):1115-1120. R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
    R833863 (2009)
    R833863 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: AJRCCM-Full Text PDF
    Exit
  • Abstract: AJRCCM-Abstract & Full Text HTML
    Exit
  • Journal Article Bell ML, Peng RD, Dominici F, Samet JM. Emergency hospital admissions for cardiovascular diseases and ambient levels of carbon monoxide: results for 126 United States urban counties, 1999-2005. Circulation 2009;120(11):949-955. R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
    R833863 (2009)
    R833863 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: Circulation-Full Text PDF
    Exit
  • Abstract: Circulation-Abstract & Full Text HTML
    Exit
  • Journal Article Crainiceanu CM, Dominici F, Parmigiani G. Adjustment uncertainty in effect estimation. Biometrika 2008;95(3):635-651. R832417 (Final)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
  • Abstract: Biometrika-Abstract
    Exit
  • Journal Article Dominici F, Peng RD, Bell ML, Pham L, McDermott A, Zeger SL, Samet JM. Fine particulate air pollution and hospital admission for cardiovascular and respiratory diseases. JAMA-Journal of the American Medical Association 2006;295(10):1127-1134. R832417 (Final)
    R832417C001 (2006)
    R832417C001 (2007)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R830548 (2005)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: JAMA-Full Text HTML
    Exit
  • Abstract: JAMA-Abstract
    Exit
  • Journal Article Dominici F, Peng RD, Ebisu K, Zeger SL, Samet JM, Bell ML. Does the effect of PM10 on mortality depend on PM nickel and vanadium content? A reanalysis of the NMMAPS data. Environmental Health Perspectives 2007;115(12):1701-1703. R832417 (Final)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R830548 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Journal Article Dominici F, Peng RD, Zeger SL, White RH, Samet JM. Dominici et al. respond to "Heterogeneity of particulate matter health risks." American Journal of Epidemiology 2007;166(8):892-893. R832417 (2008)
    R832417 (Final)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
  • Full-text: OUP-Full Text PDF
    Exit
  • Abstract: OUP-Abstract and Full Text HTML
    Exit
  • Journal Article Dominici F, Peng RD, Zeger SL, White RH, Samet JM. Particulate air pollution and mortality in the United States: did the risks change from 1987 to 2000? American Journal of Epidemiology 2007;166(8):880-888. R832417 (Final)
    R832417C001 (2007)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R830548 (Final)
    R833622 (2008)
    R833622 (2009)
  • Abstract from PubMed
  • Full-text: Oxford Journals-Full Text HTML
    Exit
  • Abstract: Oxford Journals-Abstract
    Exit
  • Other: Oxford Journals-Full Text PDF
    Exit
  • Journal Article Dominici F, Wang C, Crainiceanu C, Parmigiani G. Model selection and health effect estimation in environmental epidemiology. Epidemiology 2008;19(4):558-560. R832417 (2008)
    R832417 (Final)
    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 Eftim SE, Samet JM, Janes H, McDermott A, Dominici F. Fine particulate matter and mortality: a comparison of the six cities and American Cancer Society cohorts with a Medicare cohort. Epidemiology 2008;19(2):209-216. R832417 (2008)
    R832417 (Final)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R833622 (2008)
    R833622 (2009)
  • Abstract from PubMed
  • Full-text: Epidemiology-Full Text HTML
    Exit
  • Abstract: Epidemiology-Abstract
    Exit
  • Other: PennState-Full Text PDF
    Exit
  • Journal Article Janes H, Dominici F, Zeger S. Partitioning evidence of association between air pollution and mortality. Epidemiology 2007;18(4):427-428. R832417 (2008)
    R832417 (Final)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R833622 (2008)
    R833622 (2009)
    R833622 (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. R832417 (Final)
    R832417C001 (2006)
    R832417C001 (2007)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R830548 (Final)
    R833622 (Final)
  • Abstract from PubMed
  • Full-text: Epidemiology-Full Text HTML
    Exit
  • Abstract: Epidemiology-Abstract
    Exit
  • Other: Epidemiology-Full Text PDF
    Exit
  • Journal Article Peng RD, Dominici F, Zeger SL. Reproducible epidemiological research. American Journal of Epidemiology 2006;163(9):783-789. R832417 (Final)
    R832417C001 (2006)
    R832417C001 (2007)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R830548 (Final)
  • Abstract from PubMed
  • Full-text: Oxford Journals-Full Text HTML
    Exit
  • Abstract: Oxford Journals-Abstract
    Exit
  • Other: Oxford Journals-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. R832417 (2008)
    R832417 (Final)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R833622 (2008)
    R833622 (2009)
    R833622 (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. R832417 (2008)
    R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
    R833622 (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. R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
    R833622 (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 Symons JM, Wang L, Guallar E, Howell E, Dominici F, Schwab M, Ange BA, Samet J, Ondov J, Harrison D, Geyh A. A case-crossover study of fine particulate matter air pollution and onset of congestive heart failure symptom exacerbation leading to hospitalization. American Journal of Epidemiology 2006;164(5):421-433. R832417 (Final)
    R832417C001 (2006)
    R832417C001 (2007)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R830548 (Final)
  • Abstract from PubMed
  • Full-text: AJE-Full Text HTML
    Exit
  • Abstract: AJE-Abstract
    Exit
  • Other: AJE-Full Text PDF
    Exit
  • Journal Article Thomas DC, Jerrett M, Kuenzli N, Louis TA, Dominici F, Zeger S, Schwartz J, Burnett RT, Krewski D, Bates D. Bayesian model averaging in time-series studies of air pollution and mortality. Journal of Toxicology and Environmental Health-Part A 2007;70(3-4):311-315. R832417 (Final)
    R832417C001 (2007)
    R832417C001 (2008)
    R832417C001 (2009)
    R832417C001 (Final)
    R830548 (Final)
    R831861 (2005)
  • Abstract from PubMed
  • Abstract: Taylor&Francis-Abstract
    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. R832417 (2008)
    R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
    R833622 (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. R832417 (2008)
    R832417 (2009)
    R832417 (Final)
    R832417C001 (2009)
    R832417C001 (Final)
    R833622 (2008)
    R833622 (2009)
    R833622 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: EHP-Full Text PDF
  • Abstract: EHP-Abstract & Full Text HTML
  • Supplemental Keywords:

    time series, susceptible populations, risk estimates,, RFA, Health, Scientific Discipline, PHYSICAL ASPECTS, Air, particulate matter, Health Risk Assessment, Epidemiology, Risk Assessments, Physical Processes, atmospheric particulate matter, atmospheric particles, long term exposure, acute cardiovascular effects, airway disease, exposure, human exposure, ambient particle health effects, atmospheric aerosol particles, ultrafine particulate matter, aersol particles, cardiovascular disease

    Relevant Websites:

    www.jhsph.edu/particulate_matterexit EPA

    Progress and Final Reports:

    Original Abstract
  • 2006 Progress Report
  • 2007 Progress Report
  • 2008 Progress Report
  • Final Report

  • Main Center Abstract and Reports:

    R832417    Johns Hopkins Particulate Matter Research Center

    Subprojects under this Center: (EPA does not fund or establish subprojects; EPA awards and manages the overall grant for this center).
    R832417C001 Estimation of the Risks to Human Health of PM and PM Components
    R832417C002 PM Characterization and Exposure Assessment (Project 2)
    R832417C003 Biological Assessment of the Toxicity of PM and PM Components