Validation of Globorisk in Turkish people and development of a new model for cardiovascular diseases

Abstract Background Cardiovascular diseases (CVD) is a major cause of death globally, and accurate risk assessment is important for identifying high-risk individuals. This study aimed to validate the laboratory-based Globorisk score for predicting CVD in the Turkish population and to develop a Turkish population-specific model. Methods We analyzed data from Turkey's Chronic Diseases and Risk Factors study, which examined CVD incidence from 2011 until 2017. After excluding those with prior CVD history, a total of 7239 individuals aged 40 to 80 years were included in the analysis. The performance of Globorisk in predicting CVD was assessed using the C-index. With demographic, dietary, anthropometric and Globorisk variables; we used backward stepwise logistic regression to select the final model (Turkish CVD-TCVD). Lastly, the TCVD model was internally validated and calibrated with 200 bootstrap replicates. Results Out of 7239 participants (mean age: 53.9±10.3), women: 52.3%); 766 developed CVD within six years (cumulative incidence rate: 10.6%). The C-index of the Globorisk was 0.72 with sensitivity and specificity being 68.2% and 67.3%. In the final TCVD model, backward stepwise selection identified age (odds ratio-OR: 1.06, 95% Cl: 1.05-1.07), diabetes (OR:1.83, 1.47-2.28), body mass index (OR:1.02, 1.01-1.04), high waist-hip ratio (OR:1.37, 1.13-1.66) and systolic blood pressure (OR:1.01, 1.00-1.01) as significant predictors for CVD. C-index was 0.73 with sensitivity and specificity being 72.9%, and 62.9%. Examination of the calibration plot showed signs of overprediction when the actual CVD probability was >20%. Conclusions Laboratory-based Globorisk score had a good fit in the Turkish population. TCVD model had better sensitivity than Globorisk. Adding of waist-hip ratio to the Globorisk score could improve predictive CVD models in the Turkish population. Key messages • External validation of laboratory-based Globorisk showed good predictive accuracy in the Turkish population. • Although more challenging to measure, adding of waist-hip ratio to the laboratory-based Globorisk score could improve predictive cardiovascular disease models in the Turkish population.