The development of the geographic information systems has improved the interest in analyzing the characteristics of the population, taking into account the place where they live. The health of the population, measured as the risk of morbidity or mortality, is conditioned by the variety of risk factors characteristic of the region which cannot be measured. Analysis of spatial models considers the dependence of the health indicators between close regions. This dependence is due to the existence of the risk factors that are not measured but are shared by the region. Thus, the spatial distribution of these indicators depends on the geographic pattern of these risk factors. In this study, some limitations of the standardized methods and the Poisson regression used to model the spatiality are discussed and the advantages of the spatial models are shown. The methodology is illustrated by the insulin-dependent diabetes type 1 data from Catalonia during 1989 and 1998.