Clinical nomogram for predicting type 2 diabetes in elderly patients with hypertension: a novel model approach toward predictive, preventive, and personalized medicine

Background: Type 2 diabetes (T2D) is a widely prevalent disease, often asymptomatic in its initial phases. Therefore, early identification of individuals at a high risk of T2D is essential in the context of Predictive, Preventive, and Personalized Medicine (PPPM/3PM). Currently, there is a lack of a specific model for estimating the risk of T2D in elderly hypertensive patients. This study aims to develop a nomogram for predicting the 5-year risk of T2D in this specific population. Methods: This retrospective cohort study included 6041 elderly patients with hypertension initially free of T2D. The Least Absolute Shrinkage and Selection Operator (LASSO) regression model was employed to identify potential predictors. The relationship between continuous predictors and new-onset T2D was examined using the restricted cubic spline (RCS) function. Cox regression analysis was conducted to establish the optimal predictive nomogram model. The performance of the nomogram was assessed through the receiver operating characteristic (ROC) curve, calibration curves, and decision curve analysis (DCA). Time-dependent ROC analysis was conducted to evaluate the discriminative ability of the nomogram for T2D at different time points. We categorized participants into four risk categories according to their nomogram scores: low (Q1), middle (Q2), high (Q3), and extremely high (Q4). Kaplan-Meier (K-M) curve was used to assess the predictive value of the nomogram. Results: All 6,041 participants, 495 individuals (8.2%) developed diabetes during the follow-up period. The nomogram model incorporated four variables: age (HR = 1.03; 95% CI: 1.01–1.04), body mass index (BMI) (HR = 1.04; 95% CI: 1.01–1.08), fasting blood glucose (FPG) (HR = 4.32; 95% CI: 3.69–5.07), and triglyceride levels (HR = 1.30; 95% CI: 1.02–1.66). The nomogram demonstrated robust discrimination performance with an area under the ROC curve (AUC) of 0.795 for the training cohort and 0.755 for the validation cohort. Furthermore, calibration curves illustrated a close alignment between the predicted and observed probabilities of T2D risk, affirming the reliability of the nomogram's predictions. The DCA substantiated the favorable clinical utility of the nomogram. The time-dependent ROC and K-M curves underscored the nomogram's good discriminatory and predictive capabilities. Conclusions: We have formulated an easily applicable, personalized nomogram model that demonstrates precision and reliability in distinguishing the 5-year risk of incident T2D in elderly patients with hypertension.