Institutional members access full text with Ovid®

Measurement Error Correction for Predicted Spatiotemporal Air Pollution Exposures

Keller, Joshua P.; Chang, Howard H.; Strickland, Matthew J.; Szpiro, Adam A.

doi: 10.1097/EDE.0000000000000623
Air Pollution

Background: Air pollution cohort studies are frequently analyzed in two stages, first modeling exposure then using predicted exposures to estimate health effects in a second regression model. The difference between predicted and unobserved true exposures introduces a form of measurement error in the second stage health model. Recent methods for spatial data correct for measurement error with a bootstrap and by requiring the study design ensure spatial compatibility, that is, monitor and subject locations are drawn from the same spatial distribution. These methods have not previously been applied to spatiotemporal exposure data.

Methods: We analyzed the association between fine particulate matter (PM2.5) and birth weight in the US state of Georgia using records with estimated date of conception during 2002–2005 (n = 403,881). We predicted trimester-specific PM2.5 exposure using a complex spatiotemporal exposure model. To improve spatial compatibility, we restricted to mothers residing in counties with a PM2.5 monitor (n = 180,440). We accounted for additional measurement error via a nonparametric bootstrap.

Results: Third trimester PM2.5 exposure was associated with lower birth weight in the uncorrected (−2.4 g per 1 μg/m3 difference in exposure; 95% confidence interval [CI]: −3.9, −0.8) and bootstrap-corrected (−2.5 g, 95% CI: −4.2, −0.8) analyses. Results for the unrestricted analysis were attenuated (–0.66 g, 95% CI: −1.7, 0.35).

Conclusions: This study presents a novel application of measurement error correction for spatiotemporal air pollution exposures. Our results demonstrate the importance of spatial compatibility between monitor and subject locations and provide evidence of the association between air pollution exposure and birth weight.

From the aDepartment of Biostatistics, University of Washington, Seattle, WA; bDepartment of Biostatistics and Bioinformatics, Emory University, Atlanta, GA; and cSchool of Community Health Sciences, University of Nevada Reno, Reno, NV.

Submitted 9 March 2016; accepted 11 January 2017.

This publication was made possible by Grants RD-834799 and RD-83479601-0 from the United States Environmental Protection Agency (US EPA). Its contents are solely the responsibility of the grantee and do not necessarily represent the official views of the US EPA. Further, US EPA does not endorse the purchase of any commercial products or services mentioned in the publication. Support was also provided by NIH/NIEHS through R01ES009411, R01ES020871, R21ES024894, R21ES022795, and the Biostatistics, Epidemiologic, and Bioinformatic Training in Environmental Health Training Grant T32ES015459.

The authors report no conflicts of interest.

Code and simulated data for implementing the methods described in this manuscript are available as an eAppendix ( on the journal website.

This study was approved by the IRB of Emory University, #45413.

Supplemental digital content is available through direct URL citations in the HTML and PDF versions of this article (

Correspondence: Joshua P. Keller, Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, 615 North Wolfe Street, Baltimore, MD 21205. E-mail:

Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.