Abstract
Introduction: Non-invasive tools are recommended to identify individuals at high risk of type 2 diabetes. Agreement between validated models is unclear. This study assessed agreement between four risk-assessment models among adults attending health posts in Kerman, Iran.
Methods: In this cross-sectional study, 430 adults aged≥30 years attending health posts in Kerman, Iran, were selected by multistage random sampling. Data collected by interview and examination were used to calculate FINDRISC, ModAsian FINDRISC, ADA, and AUSDRISK risk scores. Agreement was assessed with Cohen’s kappa; logistic regression identified correlates of discordant classification.
Results: Agreement between AUSDRISK and ADA was poor (kappa=-0.06, 95% CI -0.07 to -0.05, P<0.001); 45.1% were high-risk by AUSDRISK but low-risk by ADA. Discordant high-risk individuals were younger, with higher BMI and waist circumference, lower physical activity, lower fruit/vegetable intake, and more hyperglycaemia. Waist circumference, male sex, and employment independently predicted discordance. Agreement between FINDRISC and ModAsian FINDRISC was moderate (kappa=0.536, 95% CI 0.48–0.60, P<0.001); 65.3% were classified into the same category by both tools, with most discordance reflecting ModAsian FINDRISC assigning a higher category than FINDRISC. Male sex was the strongest independent predictor of discordance between these two tools; BMI and low socioeconomic status were not statistically significant.
Conclusion: Agreement between these tools was poor to moderate and varied by demographic context, indicating a need for population-based calibration. As no glycaemic outcome was measured, this study cannot establish predictive accuracy; prospective validation and calibration studies are needed to determine the most appropriate model.