Abstract:
|
Assessing the association between air pollutants and health outcomes requires multivariable models that assess multiple pollutants simultaneously. Here, we fit a Bayesian hierarchical model within a stochastic search variable selection framework to select pollutants associated with spina bifida risk, a birth defect of the spine. We simultaneously investigated 33 hazardous air pollutants (HAPS) as identified by the US Environmental Protection Agency (EPA). HAP levels were estimated for census tracts across Texas using the U.S. EPA's Assessment System for Population Exposure Nation Wide (ASPEN). Cases and controls were provided by the Texas Birth Defects Registry, and assigned to census tracts using the mother's home address at time of delivery. To model HAPs at the census tract level, and adjust for individual level covariates, we used a generalized linear model with a random intercept, and used stochastic search to select HAPs associated with spina bifida. We identified two HAPs associated with spina bifida risk: quinolone and trichloroethylene. This work is to appear in Environmental Health, 2015.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.