Seminar - Estimating the Health Effects of Air Pollution

02/12/2014 - 12:30pm

Estimating the Health Effects of Air Pollution: Multiple Exposure Estimation and Confounder Selection

Ander Wilson, Department of Statistics, North Carolina State University

Model-based estimation of the effect of air pollution exposures can be sensitive to the shape of the concentration-response function (i.e. linear verse non-linear and additive or synergistic effects between co-exposures) and the choice of confounding variables included. In this talk we present 1) a spatial model to estimate the semi-parametric effects of two co-exposures in multi-city time-series studies and 2) a decision theoretic approach to confounder selection. For the bivariate exposure model we propose a spatial, semi-parametric surface model to estimate the joint ozone-temperature risk surfaces in 95 US urban areas. Our methodology restricts the ozone-temperature risk surfaces to be monotone in ozone and allows for both non-additive and non-linear effects of ozone and temperature. We use data from the National Mortality and Morbidity Air Pollution Study (NMMAPS) to examine interaction through the estimated bivariate ozone-temperature risk surfaces both nationally and in each city. We find evidence of a non-linear ozone effect and an ozone-temperature interaction at higher temperatures and ozone concentrations. For the confounder model we propose a decision-theoretic approach to confounder selection and effect estimation. Our method can be fit easily with existing software and in many situations without the use of MCMC methods, resulting in computation on the order of the least squares solution. The proposed method has attractive asymptotic properties and performs well in a simulation study.

Date: Wednesday, February 12, 2014 - 12:30pm

Location: Mesa Laboratory - Directors Conference room