Integrated Models for Health Outcomes, Air Pollutants, and Biomarkers

• Activity Number: 230
• Type: Topic Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Sandrah Eckel*+ and Kiros Berhane and Duncan C Thomas
• Companies: University of Southern California and University of Southern California and University of Southern California
• Keywords: Latent variable modeling ; Bayesian modeling ; Environmental epidemiology

Abstract

Air pollution has important public health impacts, particularly in susceptible subgroups such as children. Data from the Southern California Children's Health Study (CHS) motivates an integrated model, where the goal is to quantify the association of short- and long-term exposure to air pollution with a respiratory health outcome, while taking into account information on a biomarker (B) of airway inflammation, namely exhaled nitric oxide. We use a latent variable approach to tie together the outcome and the biomarker and employ Bayesian and Frequentist methods to compare the "net effect" of the exposure (E) on the outcome (Y) in the integrated model to the analogous value in a standard regression of Y on E without using B. We show that the latent variable model is identifiable and find that the two estimators are equivalent, but the integrated model provides additional insights over the standard approach by quantifying hypothesized pathways. Finally, we apply an integrated model to data in the CHS.