Bayesian approach to study the temporal trend geographical variation in the risk of type 1 diabetes. The Sardinian Conscript Type 1 Diabetes Registry

Background: A previous analysis of the Sardinian Conscript Type 1 Diabetes Registry indicated an abrupt increase in the risk for type 1 diabetes (T1D) among Sardinian male cohorts starting from the 1946 one. Objective: The aim of this study was to update the point prevalence of T I D among Sardinian male army conscripts aged 18 yr from 1992 to 1997 described in the previous survey and to update its secular trend. It also looked at the geographical distribution of prevalence. Subjects and methods: All male Italian citizens must appear before the Conscript Board for a clinical examination prior to military service. T1D patients are rejected with T1D listed as the reason. We examined, retrospectively, a total of 83 807 Sardinian males aged 18, born between 1 January 1974 and 31 December 1979. They virtually represent the whole Sardinian population of that age and sex. Results: A total of 307 subjects with T1D were identified. T1D point prevalence for the birth cohorts 1974-1979 was 3.66/1000 (95% CI 3.28-4.09). Bayesian analysis of the geographical distribution of prevalence shows the highest rate in the Southern part of the island (Cagliari province) and the lowest in the north-western part (Sassari province). A similar picture was produced by data from the Eurodiab Registry in Sardinia (1989 ongoing). No significant temporal trend of prevalence was seen in the 6 yr examined in the current study; pooling these recent birth cohorts with the previous ones, the increasing trend of prevalence was confirmed. Conclusions: Our data confirmed the South to North gradient of T1D and the progressive increase of T1D risk among males in Sardinia.