Validation of the Finnish Diabetes Risk Score and development of a country-specific diabetes prediction model for Turkey.

Abstract

Aims: Diabetes is a global health concern, and early identification of high-risk individuals is crucial for preventive interventions. Finnish Diabetes Risk Score (FINDRISC) is a widely accepted non-invasive tool that estimates the 10-year diabetes risk. This study aims to validate the FINDRISC in the Turkish population and develop a specific model using data from a nationwide cohort.

Method: The study used data of 12249 participants from the Türkiye Chronic Diseases and Risk Factors Survey. Data included sociodemographic variables, lifestyle factors, and anthropometric measurements. Multivariable logistic regression was employed using FINDRISC variables to predict incident type 2 diabetes mellitus (T2DM). Two country-specific models, one incorporating the waist-to-hip ratio (WHR model) and the other waist circumference (WC model), were developed. The least absolute shrinkage and selection operator (LASSO) algorithm was used for variable selection in the final models, and model discrimination indexes were compared.

Results: The optimal FINDRISC cut-off was 8.5, with an area under the curve (AUC) of 0.76, demonstrating good predictive performance in identifying T2DM cases in the Turkish population. Both WHR and WC models showed similar predictive accuracy (AUC: 0.77). Marital status and education were associated with increased diabetes risk in both country-specific models.

Conclusion: The study found that the FINDRISC tool is effective in predicting the risk of type 2 diabetes in the Turkish population. Models using WHR and WC showed similar predictive performance to FINDRISC. Sociodemographic factors may play a role in diabetes risk. These findings highlight the need to consider population-specific characteristics when evaluating diabetes risk.