Construction of a risk prediction model for malnutrition in maintenance hemodialysis patients

Objective To explore the influencing factors of malnutrition in maintenance hemodialysis patients, and establish the prediction model of the nomogram.

Methods Data of 255 patients undergoing maintenance hemodialysis were retrospectively analyzed, and the patients were divided into modeling group (153 cases) and verification group (102 cases) according to a vatio of 6∶4. Univariate and multivariate Logistic regression analysis was performed on data of patients in the modeling group; according to the influencing factors of malnutrition in maintenance hemodialysis patients, the prediction model of the nomogram was established. Calibration curve and receiver operating characteristic (ROC) curve were drawn to evaluate the validity and differentiation of the model.

Results Working status, whether there was diabetes mellitus or not, nutritional knowledge awareness, whether there was insufficient protein energy intake or not, dialysis frequency, urea clearance index (Kt/v), dialysis age, age, body mass index (BMI), mid-upper arm circumference (MAC), hypersensitive c-reactive protein, serum creatinine (Scr), procalcitonin (PCT), HCO₃^- and parathyroid hormone (PTH) were the factors of malnutrition in maintenance hemodialysis patients (P < 0.05). Multivariate Logistic regression analysisshowed that insufficient protein energy intake, Kt/v, age, dialysis age and hypersensitive c-reactive protein were the influencing factors for malnutrition in maintenance hemodialysis patients (P < 0.05). Calibration curve evaluation results showed that χ² of the modeling group was 7.425, and P value was 0.492; the χ² of the verification group was 6.398, and P value was 0.603. The ROC curve evaluation model differentiation results showed that the area under the curve of the modeling group was 0.877 (95%CI, 0.818 to 0.936), the sensitivity was 84.34%, and the specificity was 80.00%; the area under the curve of the verification group was 0.825 (95%CI, 0.747 to 0.903), the sensitivity was 89.09%, and the specificity was 61.70%.

Conclusion Insufficient protein energy intake, Kt/v, age, dialysis age and hypersensitive C-reactive protein are the influencing factors of malnutrition in maintenance hemodialysis patients. The established nomograph prediction model has good differentiation and effectiveness, and can be used as a tool for early screening of malnutrition in maintenance hemodialysis patients.