A Bayesian hierarchical spatial approach for constructing disease risk maps for administrative regions using fine scale exposure

A Bayesian hierarchical spatial approach is developed to construct disease risk maps at a finer scale when the outcome variable is available at an administrative area level and the exposure variables are collected at a finer scale. Give some more detail about the model. The method is demonstrated using data on the number of lung cancer cases at each town in Connecticut in 2000, and the covariates are the traffic related pollution exposure estimates and median household income at each census block group. This Bayesian hierarchical model provides estimates of the standardized morbidity ratio (SMR) for lung cancer at the census block group level, using incidence cases reported at town level. Moreover, measurement error and variation within census block group associated with the traffic measurements are considered in the model. For model selection, we use DIC to compare different models. The results show that the traffic measurement coefficient is positive; the higher the traffic measurement, the higher the risk of lung cancer. On the other hand, the median household income coefficient is negative; the higher the median household income, the lower the risk of lung cancer.