|
Activity Number:
|
264
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #307846 |
|
Title:
|
Multiple Imputation Analysis of Association Between Cardiovascular Disease Risk Factors and Environmental Exposures
|
|
Author(s):
|
Trivellore E. Raghunathan*+ and Wei Chen
|
|
Companies:
|
University of Michigan and Karmanos Cancer Institute
|
|
Address:
|
M4071 SPH II , Ann Arbor, MI, 48109-2029 ,
|
|
Keywords:
|
Multiple Imputation ; Time Series Analysis ; Bayesian Methods ; Thin Plate Splines
|
|
Abstract:
|
There is growing public health concern about the effect of exposure to airborne particles on cardiovascular disease. The key issue is, however, how to estimate exposure levels at residential addresses which are usually not available but several measures available at various monitoring sites. This paper considers the estimation of PM10 and PM2.5 at residential addresses using the data collected by the Environmental Protection Agency (EPA) from thousands of monitoring stations nationwide and using them to relate to cardiovascular disease and risk factors. Spatial effects are modeled using a nonparametric approach based on thin-plate splines. Time effects are modeled with trend, cyclical, autoregressive effects. Models include spatial covariates population density and environmental factors: temperature, visibility and TSP. Bayesian multiple imputation framework is used for the analysis.
|
