A Spatio-temporal model for cancer incidence data with zero-inflation

MUSIO, MONICA;SAULEAU EA

2011-01-01

---

Abstract

In this work we consider a joint space-time model for cancer incidence, using data on prostate cancer collected between 1988 and 2005 in a specific area of France. Our aim is to take into account possible non linear effects of some covariates and zero-inflation due to data aggregation for Poisson regression. We assume that counts of cancer cases follow zero-inflated Poisson distribution, where the probability of zero inflation is a monotonic function of the mean. The purpose of our analysis is to check whether the French prostate screening program, which begins in 1994, results in a spatial or a spatial-temporal change of the pattern of the disease.