|
Abstract:
|
In developed countries where implementation of highly costly environmental regulations has led to cleaner air, a critically important questions remains: Is there strong evidence of a causal effect of adverse health outcomes at these low exposure levels? Several statistical methods exist that allow to estimate the exposure response (ER) curve flexibly. However, most of the current approaches are not formulated in the context of a potential outcome framework for causal inference. In addition, regression approaches adjust for the same set of potential confounders across all levels of exposure and often fail to account for model uncertainty regarding covariate selection. We propose a Bayesian causal inference framework for ER estimation that achieves two goals: 1) confounder selection that accounts for the possibility that different sets of variables might confound the estimation of the causal effects across the different levels of exposure; 2) flexible estimation of the whole ER that has a causal interpretation.
|
