A hierarchical modeling approach to estimate regional acute health effects of particulate matter sources

• Type of Material: Article
• Authors: Jenna R. Krall, Emory University; A.J. Hackstadt, Vanderbilt University; R.D. Peng, Johns Hopkins University
• Language: English
• Date: 2017-04-30
• Publisher: Wiley: 12 months
• Publication Version: Author Accepted Manuscript (After Peer Review)
• Copyright Statement: © 2017 John Wiley & Sons, Ltd.
• Final Published Version (URL): http://dx.doi.org/10.1002/sim.7210
• Title of Journal or Parent Work: Statistics in Medicine
• ISSN: 0277-6715
• Volume: 36
• Issue: 9
• Start Page: 1461
• End Page: 1475
• Grant/Funding Information: Contract/grant sponsor: This work was supported by the National Institute on Aging [T32AG000247], the National Institute Of Environmental Health Sciences [T32ES012160, T32ES012871, R01ES019560, R21ES020152], and the U.S. Environmental Protection Agency [RD83587101].; This publication was developed under Assistance Agreement No. RD83587101 awarded by the U.S. Environmental Protection Agency to Yale University.
• Keywords: FINE; source apportionment; Science & Technology; PM SOURCE APPORTIONMENT; Physical Sciences; DAILY MORTALITY; HOSPITAL ADMISSIONS; cardiovascular health; statisticalmethods in epidemiology; Mathematical & Computational Biology; AIR-POLLUTION; POSITIVE MATRIX FACTORIZATION; Public, Environmental & Occupational Health; Mathematics; Life Sciences & Biomedicine; health effects; PM2.5; particulate matter sources; RESPIRATORY-DISEASES; ASSOCIATIONS; Statistics & Probability; BLOOD-PRESSURE; Research & Experimental Medicine; Medicine, Research & Experimental; Medical Informatics
• Research Categories: Health Sciences, Epidemiology; Health Sciences, Public Health; Biology, Biostatistics

Abstract

Exposure to particulate matter (PM) air pollution has been associated with a range of adverse health outcomes, including cardiovascular disease hospitalizations and other clinical parameters. Determining which sources of PM, such as traffic or industry, are most associated with adverse health outcomes could help guide future recommendations aimed at reducing harmful pollution exposure for susceptible individuals. Information obtained from multisite studies, which is generally more precise than information from a single location, is critical to understanding how PM impacts health and to informing local strategies for reducing individual-level PM exposure. However, few methods exist to perform multisite studies of PM sources, which are not generally directly observed, and adverse health outcomes. We developed SHared Across a REgion (SHARE), a hierarchical modeling approach that facilitates reproducible, multisite epidemiologic studies of PM sources. SHARE is a two-stage approach that first summarizes information about PM sources across multiple sites. Then, this information is used to determine how community-level (i.e., county-level or city-level) health effects of PM sources should be pooled to estimate regional-level health effects. SHARE is a type of population value decomposition that aims to separate out regional-level features from site-level data. Unlike previous approaches for multisite epidemiologic studies of PM sources, the SHARE approach allows the specific PM sources identified to vary by site. Using data from 2000 to 2010 for 63 northeastern US counties, we estimated regional-level health effects associated with short-term exposure to major types of PM sources. We found that PM from secondary sulfate, traffic, and metals sources was most associated with cardiovascular disease hospitalizations.