Bivariate modeling of the spatial distribution of lung and tongue cancer in Limburg

Abstract

In an effort to describe the spatial distribution of diseases, a number of methods have been proposed to model Relative Risks within areas, by using Bayesian hierarchical methods, in which one typically tries to model spatially structured and unstructured extra-Poisson variance present in the data. Univariately, the Conditional Autoregressive (CAR) Convolution Model has been very popular in the last two decades, but recently, a Combined Model was proposed that 'combines' the CAR Convolution Model with the well-known Poisson-Gamma Model and that in certain settings was shown to be superior to the CAR Convolution Model. Less solutions have been provided for the bivariate case (on the county level), in which one wants to model two diseases simultaneously and also, all proposed models are based on the CAR Convolution Model. In this study, an overview is given of univariate solutions for bivariate data and bivariate models, while extensions are made to the framework of the Combined Model to