A Bayesian Hierarchical Model for Estimating Personal Exposure to Traffic-Related Air Pollution Using Personal and Ambient Data

• Activity Number: 462 - Novel Spatial and Spatio-Temporal Models in Public Health
• Type: Contributed
• Date/Time: Thursday, August 6, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Jenna Krall*
• Companies: George Mason University
• Keywords: Air pollution; Environmental exposure; Hierarchical models; Latent variable models; Bayesian models

Abstract

Exposure to traffic-related air pollution has been associated with adverse health outcomes including preterm birth and low birthweight. In epidemiologic studies, estimating exposure to traffic-related air pollution is challenging because total exposure is a mixture of pollutants generated by both traffic and non-traffic sources. Source apportionment models have been extensively applied to estimate source-specific pollution, including traffic-related pollution. However, existing source apportionment models cannot reliably estimate traffic-related pollution in personal exposure studies when the sample size is small. We proposed a Bayesian hierarchical model for estimating personal exposure to traffic-related air pollution in a study of commuters, utilizing both vehicle monitoring and personal pollution monitoring data. Our model also incorporated ambient data, which provided a long time series that was leveraged to better estimate the chemical composition of traffic-related pollution. We applied our approach to a study of 49 women commuters in the DC metro area. Using our approach, we estimated personal exposures to traffic-related pollution and non-traffic pollution.