A web-based dynamic nomogram for predicting readmission in patients with heart failure with preserved ejection fraction.

BACKGROUND

The study aims to evaluate the efficacy of a web-based dynamic nomogram predicting the risk of heart failure (HF)-related rehospitalization within 1 year in patients with HF with preserved ejection fraction (HFpEF).

METHODS

The data of patients from two centers were categorized into training and test sets. Least absolute shrinkage and selection operator and multivariate logistic regression analysis were conducted on the training set data after selecting risk factors described in previous studies, and they were used to set up a nomogram. We then analyzed the area under the receiver operating characteristic curve (AUC-ROC) and calibration plot and conducted decision curve analysis (DCA) to confirm the efficacy of the nomogram.

RESULTS

The 1-year HF rehospitalization rates of patients with HFpEF were 23.7% and 22.8% in the two study centers, respectively. Age, body mass index, atrial fibrillation, triglyceride-glucose index, left ventricular ejection fraction, E/e, and angiotensin-converting enzyme inhibitors/angiotensin receptor blocker administration positively correlated with 1-year HF-related rehospitalization in patients with HFpEF. The dynamic nomogram was constructed based on the seven variables. The AUC-ROC of the training [0.801, 95% confidence interval (CI): 0.767-0.837] and the test datasets (0.773, 95% CI: 0.713-0.824) demonstrated that the model had good predictive ability for risk factors, the calibration plots demonstrated the excellent agreement. Additionally, the DCA curve showed that the model is highly effective with a threshold probability of 10%-80%.

CONCLUSION

The dynamic nomogram model effectively predicts HF-related rehospitalization risk within 1 year in patients with HFpEF and helps determine high-risk categories among them.