Adaptation and Validation of GLICODATA: A Brazilian Diabetes Risk Score for Primary Care Using Nationally Representative Data

Background The growing burden of type 2 diabetes disproportionately affects low- and middle-income countries (LMICs) such as Brazil. Although non-invasive risk scores are essential for early detection in resource-constrained settings, most have been developed in high-income countries and lack validation in diverse epidemiological contexts. Methods We adapted and externally validated GLICODATA—a diabetes risk score derived from the Australian AUSDRISK model—for use in Brazilian primary care. Validation used cross-sectional data from 7,438 adults in the nationally representative 2013 Brazilian National Health Survey (PNS). Diabetes was defined by fasting plasma glucose ≥ 126 mg/dL, HbA1c ≥ 6.5%, or self-reported diagnosis. Model performance was evaluated using discrimination (AUC), calibration, and decision curve analysis (DCA). GLICODATA was compared head-to-head with the Brazilian version of the Finnish Diabetes Risk Score (FINDRISC-Br). Results Diabetes prevalence was 7.68%. At the optimal cut-off (≥ 12.5 points), GLICODATA showed balanced screening performance (sensitivity 68.0%, specificity 70.8%; accuracy 70.6%). The AUC was 0.563 (95% CI 0.555–0.571). DCA demonstrated positive net benefit for thresholds between 5–20%, the range most relevant to primary care. The strongest predictors were age ≥ 65 years (aOR 4.84), elevated waist circumference (aOR 3.36), and antihypertensive medication use (aOR 2.11). GLICODATA outperformed FINDRISC-Br in both classification accuracy and net clinical benefit. Conclusions GLICODATA is a culturally adapted, non-invasive diabetes risk score validated with nationally representative biochemical data. Despite modest discrimination—consistent with similar non-laboratory tools—its balanced performance and clinical utility support its use for diabetes risk stratification within Brazil’s Family Health Strategy and potentially in other LMIC primary care systems.