Development and validation of a nomogram to predict protein-energy wasting in patients with peritoneal dialysis: a multicenter cohort study

Abstract

Background: Protein-energy wasting (PEW) is a common complication in patients with peritoneal dialysis (PD). Few investigations involved risk factors identification and predictive model construction related to PEW. We aimed to develop a nomogram to predict PEW risk in patients with peritoneal dialysis.

Methods: We collected data from end-stage renal disease (ESRD) patients who regularly underwent peritoneal dialysis between January 2011 and November 2022 at two hospitals retrospectively. The outcome of the nomogram was PEW. Multivariate logistic regression screened predictors and established a nomogram. We measured the predictive performance based on discrimination ability, calibration, and clinical utility. Evaluation indicators were receiver operating characteristic (ROC), calibrate curve, and decision curve analysis (DCA). The performance calculation of the internal validation cohort validated the nomogram.

Results: In this study, 369 enrolled patients were divided into development (n = 210) and validation (n = 159) cohorts according to the proportion of 6:4. The incidence of PEW was 49.86%. Predictors were age, dialysis duration, glucose, C-reactive protein (CRP), creatinine clearance rate (Ccr), serum creatinine (Scr), serum calcium, and triglyceride (TG). These variables showed a good discriminate performance in development and validation cohorts (ROC = 0.769, 95% CI [0.705-0.832], ROC = 0.669, 95% CI [0.585-0.753]). This nomogram was adequately calibrated. The predicted probability was consistent with the observed outcome.

Conclusion: This nomogram can predict the risk of PEW in patients with PD and provide valuable evidence for PEW prevention and decision-making.

Keywords: Nomogram; Peritoneal dialysis; Protein-energy wasting.

Figure 1. A nomogram for PEW prediction in patients with peritoneal dialysis.

Each predictor level indicates a specific score. A summary of each predictor score generated a total score. The total score corresponds to PEW probability. PEW, protein-energy wasting.

Figure 2. ROC curve of predictive model in the development set.

ROC, receiver operating characteristic; AUC, area under the curve.

Figure 3. ROC curve of predictive model in the validation set.

ROC, receiver operating characteristic; AUC, area under the curve.

Figure 4. Calibration plots of predictive model in the development set.

The x-axis is the predicted occurrence of PEW. The y-axis is the observed outcomes. The diagonal dotted line represents an ideal predictive outcome by an ideal model. The solid line represents the performance of nomogram. It better predicts that a solid line is close to a diagonal dotted line. The figure shows that nomogram have a good predictive ability.

Figure 5. Calibration plots of predictive model in the validation set.

The x-axis is the predicted occurrence of PEW. The y-axis is the observed outcomes. The diagonal dotted line represents an ideal predictive outcome by an ideal model. The solid line represents the performance of nomogram. It better predicts that a solid line is close to a diagonal dotted line. The figure shows that nomogram have a good predictive ability.

Figure 6. DCA of the nomogram in the development set.

Red-solid line: the patient does not apply predictive model, and the net benefit is zero; blue-solid line: all patients are treated by nomogram. The area enclosed by the three lines demonstrates the clinical utility of nomogram. DCA, decision curve analysis.

Figure 7. DCA of the nomogram in the validation set.

Red-solid line: the patient does not apply predictive model, and the net benefit is zero; blue-solid line: all patients are treated by nomogram. The area enclosed by the three lines demonstrates the clinical utility of nomogram. DCA, decision curve analysis.