OBJECTIVE: To assess the validity of self-reported diabetes and its determinants in the elderly. METHODS: Cross-sectional study including 1,492 subjects aged > 60 years, which corresponds to 86% of all residents in this age group living in the municipality of Bambuí, Southeastern Brazil, in 1997. The validity of self-reported diabetes was assessed comparing to diabetes mellitus defined by medical criteria (fasting glucose > 126 mg/dl or current treatment). Multivariate analysis was carried out to assess factors independently associated with sensitivity and specificity. RESULTS: The prevalence of diabetes mellitus and self-reported diabetes were 14.5% and 11.7%, respectively. Sensitivity and specificity of self-reported diabetes were 57.1% (95% CI: 50.3;63.8) and 96.9% (95% CI: 94.7;97.0), respectively. Medical visits in the previous two years (PR=3.78), as well as 1_3 (PR=1.90) and > 4 years (PR=1.55) of schooling were positively and independently associated with sensitivity. Medical visits in the previous two years (PR= 0.96) and female sex (PR = 0.97) were negatively and independently associated with specificity. CONCLUSIONS: The study results indicated that self-reported diabetes should not be either used to estimate the prevalence of diabetes mellitus or to identify individuals with the disease in the study population. Schooling and medical visits were determinants of the elderly awareness of being diabetics, indicating that these modifiable characteristics may play a role in secondary and tertiary prevention.
Diabetes mellitus; Health of the elderly; Sensitivity and specificity; Validation studies; Questionnaires
