A Risk Prediction Model for Heart Failure with Preserved Ejection Fraction Hospitalization in Patients on Maintenance Hemodialysis

Background: Heart failure with preserved ejection fraction (HFpEF) is a common complication in patients on maintenance hemodialysis (MHD). A predictive model for hospitalization risk in this population would facilitate accurate risk assessment, patient stratification, and timely clinical decision-making. Methods: This prospective cohort study enrolled MHD patients with left ventricular ejection fraction (LVEF) ≥50% between January and May 2021. The primary endpoint was HFpEF-related hospitalization, with follow-up through December 31, 2024. Predictors were selected using Lasso regression. Patients were randomly assigned to training (70%) and validation (30%) sets. A Cox proportional hazards model was used to develop the risk prediction model. Results: A total of 340 patients were included (59.7% male; mean age 56.84 ± 12.52 years). Over a median follow-up of 27.7 months, 68 patients (20.0%) experienced HFpEF-related hospitalization. The final predictive model included aortic diameter (AOD), left atrial diameter (LAD), interventricular septal thickness (IVS), urea reduction ratio (URR), and 6-minute walk distance (6MWD). The nomogram demonstrated good discrimination with a C-index of 0.834. The area under the curve (AUC) for 1- and 1.5-year predictions was 0.867 and 0.875 in the training set, and 0.849 and 0.787 in the validation set, respectively. Calibration and decision curve analysis confirmed model robustness and clinical utility. Conclusion: We developed and validated a simple, practical model to predict HFpEF-related hospitalization in MHD patients. This tool may facilitate early identification of high-risk individuals to guide clinical management.