Construction and validation of a predictive model for depression in elderly maintenance hemodialysis patients: a multicenter study

Background Depression is the most common adverse psychological state among elderly patients undergoing maintenance hemodialysis. Persistent depressive symptoms significantly impair the patients' quality of life and increase their risk of readmission and mortality. Method In this study, we selected 871 elderly hemodialysis patients from the blood purification centers of nine tertiary hospitals in Chengdu, Sichuan Province, China, during the period from November 2023 to February 2024. These patients were categorized into a depression group (538 cases) and a non-depression group (333 cases). Multivariable logistic regression analysis was employed to identify independent risk factors and to develop a risk prediction model, which included the construction of a nomogram and subsequent internal validation. Subsequently, from March to April 2024, external validation of the model was performed using 219 elderly patients undergoing maintenance hemodialysis from three additional hospitals in Chengdu. Result The depression prevalence rate among elderly patients undergoing maintenance hemodialysis was found to be 38.2%. Logistic regression analysis revealed that education level, visual impairment, frailty, cognitive impairment, malnutrition, activities of daily living, and social support were independent risk factors for depression ( p <0.05). Both internal and external validation of the model demonstrated a receiver operating characteristic curve(ROC）area under the curve (AUC) greater than 0.80, indicating good discriminative ability. Furthermore, calibration plots, the Hosmer-Lemeshow test, and clinical decision curves all demonstrated that the model has good calibration and clinical applicability. Conclusion In this study, we showed that the prevalence of depression is notably high among elderly patients undergoing maintenance hemodialysis. Risk factors for depression in this population include education level, visual impairment, frailty, cognitive impairment, malnutrition, activities of daily living, and social support. The nomogram prediction model developed based on these risk factors demonstrates good predictive efficacy, offering a valuable reference for healthcare professionals to identify at-risk individuals early in clinical practice.