![[logo] US EPA](https://webarchive.library.unt.edu/eot2008/20081105174936im_/http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
Final Report: Quantifying Exposure Error and its Effect on Epidemiological Studies
EPA Grant Number: R827353C002Subproject: this is subproject number 002 , established and managed by the Center Director under grant R827353
(EPA does not fund or establish subprojects; EPA awards and manages the overall grant for this center).
Center: EPA Harvard Center for Ambient Particle Health Effects
Center Director: Koutrakis, Petros
Title: Quantifying Exposure Error and its Effect on Epidemiological Studies
Investigators: Suh, Helen H. , Sarnat, Jeremy , Schwartz, Joel , Zanobetti, Antonella
Institution: Harvard University
EPA Project Officer: Stacey Katz/Gail Robarge,
Project Period: June 1, 1999 through May 31, 2005 (Extended to May 31, 2006)
Project Amount: Refer to main center abstract for funding details.
RFA: Airborne Particulate Matter (PM) Centers (1999)
Research Category: Particulate Matter
Description:
Objective:Theme I: Assessing Particle Exposures for Health Effects Studies: A large data set on personal exposures and indoor and outdoor concentrations was collected for panels of susceptible individuals across the US (Sarnat, et al., 2000; Sarnat, et al., 2001; Sarnat, et al., 2002; Koutrakis, et al., 2005). These investigations suggest that personal exposures to PM2.5 of ambient origin are highly correlated with outdoor concentrations. However, the regression slopes of personal exposures on outdoor concentrations, which are usually less than one, vary substantially depending on house characteristics, season, and city climatic conditions. The strong correlations between personal and ambient concentrations were unique to PM2.5, as personal exposures to O3, SO2 and NO2 were substantially lower than, and weakly correlated with, corresponding outdoor concentrations (Sarnat, et al., 2005).
The primary focus of Theme I was to assess human exposures to particles and gaseous co-pollutants in order to better understand their heath effects. As such, research conducted as part of Theme I contained five main objectives:
- to characterize the inter- and intra- variability in personal particulate and gaseous exposures;
- to identify factors affecting the relationship between personal exposures and outdoor levels;
- to determine the contribution of outdoor and indoor particles to personal particulate exposures;
- to quantify the effect of measurement error for fine particles and their co-pollutants (coarse mass and the criteria gases) on risk estimates from epidemiological studies; and
- to differentiate the health effects of particles from outdoor and indoor sources.
These objectives were addressed by three inter-related research projects, which made use of our database of personal, indoor, and outdoor particulate and gaseous exposures.
The main objective of this project was to quantify exposure error and to investigate its effect on the observed associations between exposure and health outcome.
Summary/Accomplishments (Outputs/Outcomes):As mentioned above, the preliminary findings of Project Ia (R827353C001) suggest that home characteristics, particularly home ventilation, are the primary determinant of the fraction of outdoor particles that penetrate indoor environments and thus are an important determinant of personal exposures to particles of outdoor origin as well. Through its impact on exposures to particles of outdoor origin, it is possible that home ventilation may also affect the association between outdoor particle concentrations and health risk. To test this hypothesis, we used data from 14 cities located across the U.S. to examine the relationship between air conditioning prevalence and the coefficient for the relationship between ambient PM10 concentrations and cause-specific hospital admissions (Janssen, et al., 2002). In addition, we examined whether observed variability in the risk coefficients was specifically related to PM10 emissions from mobile, combustion, and other sources.
Our research examined the impact of exposure-related factors on risk estimates from time-series studies of PM10 and hospital admissions. In a paper published in 2002, we used data from 14 cities located across the US to examine the relationship between air conditioning prevalence and the coefficient for the relationship between ambient PM10 concentrations and cause-specific hospital admissions (Janssen, et al., 2002). In addition, we examined whether observed variability in the risk coefficients was specifically related to PM10 emissions from mobile, combustion, and other sources. Results from this study indicate that air conditioning use explains a substantial amount of the variability in the risk coefficients from the different cities. Furthermore, PM10 emissions from mobile and diesel sources were also found to be important determinants of the variability in the risk coefficients, particularly for cardiovascular disease (CVD)-related hospital admissions. To validate these findings, we used the same data to examine whether ventilation and source emission profiles explain season-specific risks of PM10 on hospital admissions in each of these 14 cities. This analysis is nearly complete, but a paper has not been submitted for review.
As part of our work to assess exposure error, we developed new methods to correct for measurement error in hierarchical models (Schwartz and Coull, 2003). We showed that existing standard two-stage estimators will be biased in the presence of exposure measurement error and that this bias can be away from the null hypothesis of no effect. We proposed two alternative methods for estimating the independent effects of two predictors in a hierarchical model. We applied the new methodology to show that the estimated effect of fine particles on daily deaths, independent of coarse particles, was downwardly biased by measurement error in an original analysis that did not correct for measurement error. We also used the methods to estimate the effect of gaseous air pollutants on daily deaths. The resulting effect size estimates were very small and the confidence intervals included zero. We applied this approach to a reanalysis of the National Morbidity, Mortality, and Air Pollution Study (NMMAPS) mortality study conducted by Johns Hopkins University researchers, which was published as a report to the Health Effects Institute (HEI, 2003). Also, using data from multi-pollutant exposure studies in Boston and Baltimore, simulations were conducted to assess the feasibility of health risks attributed to gases and particles. Results provided evidence that the gaseous pollutants are unlikely confounders of PM health risk estimates for these locations. These results were presented in a meeting abstract (Schwartz and Sarnat, 2002), and a manuscript has been submitted to the Journal of Exposure Science and Environmental Epidemiology.
We have also been working on the development and application of near-source and long-range atmospheric dispersion models to better quantify the relationship between emissions and concentrations of primary and secondary PM. This analysis will allow for improved spatially resolved exposure estimates and reduced exposure misclassification. A paper is under preparation, but it has not been finalized for submission.
Spatial-Temporal Modeling of Exposure
We developed spatial-temporal models of spatially varying exposures, such as traffic pollution, in the Boston area. Given a good model for exposure, this approach yields more accurate measures of spatially heterogeneous exposures than central site monitoring, and allows for examination of longer averaging times than the limited personal exposures. This approach can decrease the amount of measurement error associated with the central-site measurements and in turn yield more powerful tests of health effects. This manuscript was published. A revised manuscript describing the methodology and results of this analysis has been accepted by the Journal of the Royal Statistical Society (Gryparis, et al., 2007).
Quantifying Model Uncertainty in Epidemiological Analyses
A criticism of existing PM epidemiologic analyses is the multiple sources of uncertainty involved in obtaining health effect estimates. One key uncertainty is the shape of the concentration-response relation. Another is estimating how long one would have to wait after lowering pollution before the health improvements arrive. That is, are the associations with twenty-year average exposures, which will change slowly, or are they with recent exposures? We examined the use of Bayesian model averaging as a way of addressing these two forms of model uncertainty in a reanalysis of the Harvard Six Cities study. This approach avoids relying on an effect estimates from a single “final” model, which ignores uncertainty associated with model choice and thus can underestimate the variability associated with these effect estimates. Rather this method takes a weighted average of estimates from a range of plausible models. We implemented this approach to average over plausible models for the dose-response relationship of PM as well as the lag structure in the model. Preliminary results suggest that the dose-response curve is approximately linear and the strongest lagged effects occur during the current year (i.e. lag 0) and the immediately preceding year (i.e. lag 1). A paper describing the analysis has been submitted and is currently under review.
Conclusions:Results from our work indicate that air conditioning use explains a substantial amount of the variability in the relationship between ambient PM10 concentrations and cause-specific hospital admissions from 14 different cities studied. Additional results provided evidence that gaseous pollutants are unlikely confounders of PM health risk estimates. Finally, our results suggest that the dose-response curve for PM health effects is approximately linear, and the strongest lagged effects occur during the current year (i.e. lag 0) and the immediately preceding year (i.e. lag 1).
References:
Gryparis A, Coull B, Schwartz J, HH S. Semiparametric regression models for spatio-temporal modeling of mobile source particles in the greater Boston area. Journal of the Royal Statistical Society, Series C (in press, 2007).
Janssen NAH, Schwartz J, Zanobetti A, Suh H. Air conditioning and source-specific particles as modifiers of the effect of PM10 on hospital admissions for heart and lung disease. Environmental Health Perspectives 2002;110:43-49.
Koutrakis P, Suh H, Sarnat J, Brown K, Coull B, Schwartz J. Characterization of particulate and gas exposures of sensitive subpopulations living in Baltimore and Boston. Health Effects Institute, Research Report. Health Effects Institute, Boston, MA, 2005.
Sarnat JA, Brown KW, Schwartz J, Coull BA, Koutrakis P. Ambient gas concentrations and personal particulate matter exposures: implications for studying the health effects of particles. Epidemiology 2005;16(3):385-395.
Sarnat JA, Koutrakis P, Suh H. Assessing the relationship between personal particulate and gaseous exposures of senior citizens living in Baltimore. Journal of the Air & Waste Management Association 2000;50(7):1184-1198.
Sarnat JA, Long CM, Koutrakis P, Coull BA, Schwartz J, Suh HH. Using sulfur as a tracer of outdoor fine particulate matter. Environmental Science & Technology 2002;36(24):5305-5314.
Sarnat JA, Schwartz J, Catalano P, Suh H. Gaseous pollutants in particulate matter epidemiology: confounders or surrogates? Environmental Health Perspectives 2001;109(10):1053-1061.
Schwartz J, Coull BA. Control for confounding in the presence of measurement error in hierarchical models. Biostatistics 2003;4(4):539-553.
Schwartz J, Sarnat J. Effects of measurement error on associations between ambient air pollution and daily mortality: A simulation study using the covariance of personal and ambient measurements. Epidemiology 2002;13(4):081.
Journal Articles on this Report: 3 Displayed | Download in RIS Format
| Other subproject views: | All 3 publications | 3 publications in selected types | All 3 journal articles |
| Other center views: | All 149 publications | 149 publications in selected types | All 148 journal articles |
| Type | Citation | ||
|---|---|---|---|
|
|
Janssen NAH, Schwartz J, Zanobetti A, Suh HH. Air conditioning and source-specific particles as modifiers of the effect of PM10 on hospital admissions for heart and lung disease. Environmental Health Perspectives 2002;110(1):43-49. |
R827353 (Final) R827353C002 (2000) R827353C002 (2001) R827353C002 (2002) R827353C002 (2003) R827353C002 (Final) |
|
|
|
Schwartz J, Coull BA. Control for confounding in the presence of measurement error in hierarchical models. Biostatistics 2003;4(4):539-553. |
R827353 (Final) R827353C002 (Final) R827353C005 (Final) |
|
|
|
Schwartz J. Is the association of airborne particles with daily deaths confounded by gaseous air pollutants? An approach to control by matching. Environmental Health Perspectives 2004;112(5):557-561. |
R827353 (Final) R827353C002 (Final) |
|
, Air, Scientific Discipline, Health, RFA, Susceptibility/Sensitive Population/Genetic Susceptibility, Molecular Biology/Genetics, Toxicology, Biology, indoor air, Risk Assessments, genetic susceptability, Epidemiology, air toxics, Environmental Microbiology, particulate matter, Environmental Chemistry, tropospheric ozone, ambient measurement methods, cardiopulmonary, risk assessment, exposure and effects, ambient air quality, cardiovascular disease, health effects, indoor air quality, inhalation, developmental effects, epidemelogy, animal inhalation study, respiratory disease, inhalation toxicology, air quality, ambient air, cardiopulmonary response, indoor exposure, molecular epidemiology, measurement methods, cardiopulmonary responses, human health risk, interindividual variability, monitoring, genetic susceptibility, particle exposure, quantifying exposure error, mortality studies, air pollutants, human health effects, particulates, respiratory, sensitive populations, measurement methods , ambient particle health effects, air pollution, ambient monitoring, inhaled, metals, stratospheric ozone, ambient air monitoring, chemical exposure, dosimetry, exposure, inhaled particles, pulmonary, human health, atmospheric monitoring, human susceptibility, biological mechanism , health risks, human exposure, Human Health Risk Assessment, PM, pulmonary disease
Relevant Websites:
http://www.hsph.harvard.edu/epacenter/epa_center_99-05/index.html
Progress and Final Reports:
1999 Progress Report
2000 Progress Report
2001 Progress Report
2002 Progress Report
2003 Progress Report
Original Abstract
Main Center Abstract and Reports:
R827353 EPA Harvard Center for Ambient Particle Health Effects
Subprojects under this Center:
(EPA does not fund or establish subprojects; EPA awards and manages the overall grant for this center).
R827353C001 Assessing Human Exposures to Particulate and Gaseous Air Pollutants
R827353C002 Quantifying Exposure Error and its Effect on Epidemiological
Studies
R827353C003 St. Louis Bus, Steubenville and Atlanta Studies
R827353C004 Examining Conditions That Predispose Towards
Acute Adverse Effects of Particulate Exposures
R827353C005 Assessing Life-Shortening Associated with Exposure to
Particulate Matter
R827353C006 Investigating Chronic Effects of Exposure to Particulate
Matter
R827353C007 Determining the Effects of Particle Characteristics on Respiratory Health of Children
R827353C008 Differentiating the Roles of Particle Size, Particle Composition,
and Gaseous Co-Pollutants on Cardiac Ischemia
R827353C009 Assessing Deposition of Ambient Particles in the Lung
R827353C010 Relating Changes in Blood Viscosity, Other Clotting Parameters,
Heart Rate, and Heart Rate Variability to Particulate and Criteria Gas Exposures
R827353C011 Studies of Oxidant Mechanisms
R827353C012 Modeling Relationships Between Mobile Source Particle Emissions and Population Exposures
R827353C013 Toxicological Evaluation of Realistic Emissions of Source Aerosols (TERESA) Study
R827353C014 Identifying the Physical and Chemical Properties of Particulate Matter Responsible for the Observed Adverse Health Effects
R827353C015 Research Coordination Core
R827353C016 Analytical and Facilities Core
R827353C017 Technology Development and Transfer Core