Objective To explore the risk factors for mineral and bone disorder in maintenance hemodialysis patients, and to construct and validate a nomogram prediction model.

Methods  A total of 306 patients undergoing maintenance hemodialysis at Shanghai Eighth People’s Hospital from January 2021 to May 2025 were selected as study subjects and randomly divided into a training set (n＝214) and a validation set (n＝92) in a 7∶3 ratio. In the training set, patients were divided into a normal bone mineral metabolism group and an abnormal bone mineral metabolism group, and related factors were compared between the two groups. The multivariate logistic regression analysis was used to identify the influencing factors of mineral and bone disorder in maintenance hemodialysis patients in the training set, and a nomogram prediction model was constructed. ROC curves were drawn to evaluate the ability of the nomogram model for predicting mineral and bone disorder in these patients. Calibration curves and Hosmer-Lemeshow goodness-of-fit test were used to analyze the consistency of the predictive probability of nomogram model and actual probability of mineral and bone disorder in these patients. The decision curve was used to assess the clinical benefit using nomogram prediction model.

Results  Among the 306 hemodialysis patients, 254 patients had mineral and bone disorder, accounting for 83.01%. Among the 214 patients in the training set, 177 had mineral and bone disorder, accounting for 82.71%. In the training set, age, gender, body mass index (BMI), hypertension rate, dialysis age, blood urea nitrogen (BUN), hemoglobin (Hb), albumin (ALB), alkaline phosphatase (ALP), serum creatinine (SCr), uric acid (UA), estimated glomerular filtration rate (eGFR), and rate of taking phosphate binders were statistically significant different between the two groups (P＜0.05). The multivariate logistic regression analysis showed higher age, female, hypertension, longer dialysis duration, decreased eGFR, and not taking phosphate binders were identified as risk factors for mineral and bone disorder in maintenance hemodialysis patients (P＜0.01). The nomogram prediction model was constructed. The area under the ROC curve of the model for mineral and bone disorder in the training set and validation set was 0.895 (95%CI 0.850-0.941) and 0.881 (95%CI 0.830-0.932), respectively, with maximum Youden indice of 0.650 and 0.600, sensitivity of 0.856 and 0.849, and specificity of 0.794 and 0.751. The Hosmer-Lemeshow test showed the nomogram prediction model had good consistency in predictive probabilities with actual probabilities in training set and validation set. The decision curve showed the nomogram model could bring clinical net benefits when the threshold probabilities in the training set and validation set were less than 0.96 and 0.91.

Conclusions  The nomogram prediction model constructed based on six independent risk factors including age, gender, hypertension, dialysis duration, eGFR, and using phosphate binders or not, shows good discrimination and calibration, with good clinical predictive ability, which could provide guidance for the management of maintenance hemodialysis patients.