An interactive dynamic online nomogram for predicting the frailty trajectories among older people with chronic heart failure

To delineate the heterogeneous frailty trajectories in older people with chronic heart failure during one year after discharge, and further explore their predictors to construct a nomogram for prediction. Longitudinal data on 757 older chronic heart failure patients (CHF) over 60 years was used to delineate the heterogeneous frailty trajectories using a growth mixture model (GMM). The least absolute shrinkage and selection operator and logistic regression model was employed to determine their predictors and further construct a nomogram based on the predictors. A 1000-fold bootstrap resampling was used for internal validation of the nomogram, and its discrimination, calibration, and clinical values were evaluated by the area under the receiver operating characteristic curve (AUROC), calibration curves, and decision curve analysis (DCA), respectively. GMM identified three heterogeneous frailty trajectories: "frailty deteriorating trajectory", "frailty moderately improving trajectory", and "frailty slightly improving trajectory". Logistic regression analysis showed 6 independent predictors of "frailty deteriorating trajectory", and a dynamic online nomogram was constructed. The AUROCs of the nomogram on the training and validation sets were 0.752 (95 %CI, 0.713-0.792) and 0.753 (95 %CI, 0.682-0.833), respectively. The calibration curves demonstrated that probabilities predicted by the nomogram had high consistency with the actual probability, and the DCA showed that the nomogram had excellent clinical utility on both the training and validation sets. Older CHF patients have heterogeneous frailty trajectories, and the nomogram we constructed can serve as clinical tool for early identifying the high-risk groups, and promoting the personalized and precise frailty management.