INTEGRATIVE STATISTICAL METHODS FOR EXPOSURE MIXTURES AND HEALTH

• Type of Material: Article
• Authors: Brian J Reich, North Carolina State University; Yawen Guan, University of Nebraska; Denis Fourches, North Carolina State University; Joshua L Warren, Yale University; Stefanie Ebelt, Emory University; Howard Chang, Emory University
• Language: English
• Date: 2020-12-01
• Publisher: INST MATHEMATICAL STATISTICS-IMS
• Publication Version: Author Accepted Manuscript (After Peer Review)
• Copyright Statement: © 2020 Institute of Mathematical Statistics
• Final Published Version (URL): http://dx.doi.org/10.1214/20-AOAS1364
• Title of Journal or Parent Work: ANNALS OF APPLIED STATISTICS
• Volume: 14
• Issue: 4
• Start Page: 1945
• End Page: 1963
• Grant/Funding Information: This work was supported by the National Institutes of Health (ES025128, ES027892, ES031651 and ES028526), the National Science Foundation (DMS-1638521), the Electric Power Research Institute (EPRI, 10002467) and the U.S. Environmental Protection Agency (USEPA, RD834799). The content of this publication is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the USEPA. Further, USEPA does not endorse the purchase of any commercial products or services mentioned in the publication. The air pollutant data were supported by the Southeastern Aerosol Research and Characterization (SEARCH) Network, the Electric Power Research Institute and the Atmospheric Research Associates.
• Keywords: Cheminformatics; REGRESSION; Science & Technology; RISK; Statistics & Probability; ATLANTA; DAILY MORTALITY; Mathematics; BAYESIAN VARIABLE SELECTION; AMBIENT AIR-POLLUTION; stochastic search; Physical Sciences; collinearity; factor analysis; principal components; POLLUTANTS; variable selection

Abstract

Humans are concurrently exposed to chemically, structurally and toxico-logically diverse chemicals. A critical challenge for environmental epidemiology is to quantify the risk of adverse health outcomes resulting from exposures to such chemical mixtures and to identify which mixture constituents may be driving etiologic associations. A variety of statistical methods have been proposed to address these critical research questions. However, they generally rely solely on measured exposure and health data available within a specific study. Advancements in understanding of the role of mixtures on human health impacts may be better achieved through the utilization of exter-nal data and knowledge from multiple disciplines with innovative statistical tools. In this paper we develop new methods for health analyses that incorpo-rate auxiliary information about the chemicals in a mixture, such as physico-chemical, structural and/or toxicological data. We expect that the constituents identified using auxiliary information will be more biologically meaningful than those identified by methods that solely utilize observed correlations between measured exposure. We develop flexible Bayesian models by specify-ing prior distributions for the exposures and their effects that include auxiliary information and examine this idea over a spectrum of analyses from regression to factor analysis. The methods are applied to study the effects of volatile organic compounds on emergency room visits in Atlanta. We find that includ-ing cheminformatic information about the exposure variables improves prediction and provides a more interpretable model for emergency room visits for respiratory diseases.