Development and validation of a risk prediction model for incident frailty in elderly patients with cardiovascular disease

Abstract

Abstract Background Cardiovascular disease (CVD) and frailty frequently coexist in older populations, resulting in a synergistic impact on health outcomes. This study aims to develop a prediction model for the risk of frailty among patients with cardiovascular disease. Methods Using data from the China Health and Retirement Longitudinal Study (CHARLS), a total of 2,457 patients with cardiovascular disease (CVD) in 2011 (n = 1,470) and 2015 (n = 987) were randomly divided into training set (n = 1,719) and validation set (n = 738) at a ratio of 7:3. LASSO regression analysis was used conducted to determine identify the predictor variables with the most significant influence on the model. Stepwise regression analysis and logistic regression model were used to analyze the risk factors of frailty in patients with cardiovascular disease. The prediction model was established by constructing a nomogram. The predictive accuracy and discriminative ability of the nomogram were determined by the concordance index (C-index) and calibration curve. The area under the receiver operating characteristic curve and decision curve analysis were conducted to assess predictive performance. Results A total of 360 patients (17.2%) had frailty symptoms. Among the 29 independent variables, it was found that gender, age, pain, grip strength, vision, activities of daily living (ADL), and depression were significantly associated with the risk of frailty in CVD patients. Using these factors to construct a nomogram model, the model has good consistency and accuracy. The AUC values of the prediction model and the internal validation set were 0.859 (95%CI 0.836–0.882) and 0.860 (95%CI 0.827–0.894), respectively. The C-index of the prediction model and the internal validation set were 0.859 (95%CI 0.836–0.882) and 0.887 (95%CI 0.855–0.919), respectively. The Hosmer-Lemeshow test showed that the model's predicted probabilities were in reasonably good agreement with the actual observations. The calibration curve showed that the Nomogram model was consistent with the observed values. The robust predictive performance of the nomogram was confirmed by Decision Curve analysis (DCA). Conclusions This study established and validated a nomogram model, combining gender, age, pain, grip strength, ADL, visual acuity, and depression for predicting physical frailty in patients with cardiovascular disease. Developing this predictive model would be valuable for screening cardiovascular disease patients with a high risk of frailty.