Construction and validation of a risk predictive model for protein-energy wasting in hemodialysis patients

Abstract

Objective To explore the incidence and risk factors of protein-energy wasting （PEW） in patients on maintenance hemodialysis （MHD） and construct a visual risk predictive model and verify its predictive performance. Methods From February 2020 to February 2023, 313 MHD patients were recruited and assigned into two groups of PEW （n = 125） and non-PEW （n = 188） based upon the diagnostic criteria of PEW by International Society of Renal Nutrition and Metabolism （ISRNM）. The relevant clinical data and blood biochemical tests of two groups were compared by single factor and PEW risk factors screened by Logistic regression model. R software was utilized for establishing a nomogram model. Bootstrap method was employed for internal verification. Receiver operating characteristic （ROC） curve was plotted for calculating the area under the curve （AUC） and C-index utilized for evaluating the accuracy of the model. Calibration curve was employed for evaluating the consistency and Hosmer-Lemeshow test for determining goodness of fit. Results Single factor comparison revealed that females and diabetes mellitus predominated in PEW group. NT-proBNP and urea clearance index rose while education time, body mass index （BMI）, mid-arm muscle circumference （MAMC）, dietary protein intake （DPI）, albumin, serum creatinine, urea nitrogen, total cholesterol （TC）, vitamin D and hemoglobin declined （P＜0.05）. Univariate and multivariate regression analysis revealed that females （OR = 1.571, 95%CI：1.060-1.464, P = 0.023） and NT-proBNP （OR = 1.493, 95%CI：1.045-2.133, P = 0.032） were risk factors for PEW while TC （OR = 0.820, 95%CI：0.704-0.956, P = 0.003） and vitamin D （OR = 0.882, 95%CI：0.818-0.953, P＜0.001） were protective factors. ROC indicated that AUC of the model for predicting PEW was 0.867 （95%CI：0.812-0.952, P＜0.001） with a C-index of 0.875. Moreover, the model had excellent consistency and goodness of fit. Conclusion MHD patients tend to have a higher risk of PEW. And nomogram model based upon gender, NT-proBNP, TC and vitamin D has great potential for guiding clinical risk assessment of PEW.

Keywords

• hemodialysis
• protein - energy expenditure
• nomograms
• n-terminal - brain natriuretic peptide precursor
• total cholesterol
• vitamin d