[The challenge of applying capture-recapture methodology to diabetes mellitus surveillance in the elderly: lessons from a Brazilian experience].

The purpose of this study was to evaluate the feasibility of establishing a surveillance system for diabetes mellitus in the elderly using data sources on morbidity (hospital and outpatient data) and mortality and the capture-recapture methodology. Log-linear models were adjusted for the entire sample and for gender and age strata. Seven hundred and forty patients were identified in the three combined sources. The model selected for the whole sample when the first linkage strategy was deployed estimated a total of 22,925 cases (90%CI: 11,354-57,269), an implausibly high figure that suggested a strong final negative dependency between sources. Age and gender stratification led to more plausible point values, which nevertheless still lacked precision. The only subset that displayed precise values was the 60-to-69-year-old female stratum (1,290 cases; 90%CI: 779-2,476). The factors that possibly explain these results are: (a) the low number of cases originated in the specialized and more complex facilities in the outpatient data source and (b) overall underreporting of diabetes mellitus cases.