Abstract for presentation at the American Statistical Association Albany Chapter Conference at Rensselaer Polytechnic Institute on March 24, 2001

Glen D. Johnson

New York State Department of Health

Bureau of Environmental and Occupational Epidemiology

Abstract: Mapping human disease rates can aid in identifying potential geographic clusters of disease. Such maps can in turn help to generate hypotheses of potential causes and to help prioritize areas for more rigorous epidemiological investigations. However, geographic areas with relatively small populations will yield highly unstable rate estimates. Therefore methods have been developed for "smoothing" maps in ways that borrow strength from neighboring areas to obtain more stable estimates for areas with small populations.

A quick summary of various smoothing approaches will be presented, followed by a more in-depth discussion of a spatially autoregressive fully Bayesian hierarchical linear model for smoothing human disease incidence rates. The method and ways to view the results will be illustrated for ZIP code-level mapping of disease rates in New York State.