[Influence of a priori parameters on bayesian relative risks estimations. Spatial distribution of bladder cancer in the urban area of Grenoble].

Bayesian estimates of disease relative risks is currently the gold standard in disease mapping when the disease is rare and/or when the geographical area is small. Its use has become quite easy with adhoc software. However, the implicit mechanisms of the choices made by the user must be clearly identified. We were interested here in the consequences of the choice of the hyper a priori parameters. We have compared results obtained using various hyper a priori parameters. The consequences of these choices are illustrated through the example of the incidence of bladder cancer among men in the urban area of Grenoble. We show that the risks can appear weak from a statistical point of view but important from an epidemiologic point of view in the presentation of the results.