Effect of Exposure Measurement Error in Air Pollution Studies: Two Case Studies

EPA Grant Number: FP917345
Title: Effect of Exposure Measurement Error in Air Pollution Studies: Two Case Studies
Investigators: Kioumourtzoglou, Marianthi-Anna
Institution: Harvard University
EPA Project Officer: Lee, Sonja
Project Period: September 1, 2011 through August 31, 2014
Project Amount: $126,000
RFA: STAR Graduate Fellowships (2011) RFA Text |  Recipients Lists
Research Category: Academic Fellowships , Fellowship - Human Health: Public Health Sciences


Exposure measurement error has long been identified as an important issue in epidemiological studies of particulate matter. The proposed study will quantify measurement error in chronic PM health studies, introduced by measurements of PM2.5 and its components at centrally located stationary ambient sites. Additionally, this study will examine whether statistical methods that integrate across pollution and health measures, such as structural equation and hierarchical models, are able to reduce the impact of measurement error and de-attenuate risk estimates.


To assess the impact of measurement error on the association between PM exposures and chronic health outcomes, personal PM2.5 exposure and ambient concentration data from 10 validation datasets will be used. The estimated deattenuation factors will then be used to develop regression calibration models to adjust for bias in studies of chronic PM health effects. Further, structural equation models will be used to examine the association between specific PM2.5 sources and cardiovascular (CVD)- and respiratoryrelated emergency hospital admissions. To assess pollution sources, both source apportionment and supervised approaches will be considered. Finally, a hierarchical regression model will be employed to examine the association between pollutant properties and CVD- and respiratoryrelated emergency hospital admissions.

Expected Results:

Although many studies have linked fine particles to adverse health outcomes, the specific particulate sources or chemical components responsible for the observed effects are not known. This study aims to provide insight in the PM constituents causing the health outcomes, while using methods that reduce the impact of measurement error and de-attenuate the risk estimates. Additionally, the estimated de-attenuation factors from the developed measurement error model can be used to adjust the estimates of previous cohort studies on the chronic effects of air pollution.

Potential to Further Environmental / Human Health Protection

Results of this study will further the understanding of health effects from PM2.5, improving the ability to interpret health risks from previous studies and to design future studies. Furthermore, this study will provide critical insights on the toxicity of PM2.5 components and sources, helping to assess whether pollutants produced from different sources are more toxic than others. The findings will have significant implications for future epidemiologic studies, source-specific policy decisions and improved air quality monitoring.

Supplemental Keywords:

air pollution, exposure measurement error, fine particles, source apportionment, structural equation models

Progress and Final Reports:

  • 2012
  • 2013
  • Final