A nomogram for risk stratification of protein-energy malnutrition in pediatric cancer patients: development and internal validation.
Source: PubMed, NCBI / U.S. National Library of Medicine
To develop and internally validate a nomogram for predicting the risk of protein-energy malnutrition (PEM) in pediatric cancer patients and to evaluate its clinical utility. A total of 375 pediatric cancer patients admitted to a tertiary hospital in Xinjiang (January-May 2025) were enrolled. Nutritional status was assessed using the WHO BMI-for-age Z-score. Candidate variables (P < 0.01 in univariate analysis) were entered into the least absolute shrinkage and selection operator (LASSO) model, and multivariate logistic regression was applied to construct the prediction model. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), calibration plots with the Hosmer-Lemeshow test, and decision curve analysis (DCA). Six readily obtainable variables were retained in the final model. Protective factors included being an only child (OR = 0.278), time since diagnosis of 0-6 months (OR = 0.264) or > 12 months (OR = 0.351), caregiver education of high school or above (OR = 0.406), and household per-capita income ≥ 1,000 yuan (OR = 0.255). Rural residence (OR = 2.643) and ≥ 6 chemotherapy cycles (OR = 2.259) were risk factors for PEM. The nomogram demonstrated robust discrimination in both the modeling (AUC = 0.823, 95% CI: 0.767-0.868) and validation (AUC = 0.
Abstract
To develop and internally validate a nomogram for predicting the risk of protein-energy malnutrition (PEM) in pediatric cancer patients and to evaluate its clinical utility. A total of 375 pediatric cancer patients admitted to a tertiary hospital in Xinjiang (January-May 2025) were enrolled. Nutritional status was assessed using the WHO BMI-for-age Z-score. Candidate variables (P < 0.01 in univariate analysis) were entered into the least absolute shrinkage and selection operator (LASSO) model, and multivariate logistic regression was applied to construct the prediction model. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), calibration plots with the Hosmer-Lemeshow test, and decision curve analysis (DCA). Six readily obtainable variables were retained in the final model. Protective factors included being an only child (OR = 0.278), time since diagnosis of 0-6 months (OR = 0.264) or > 12 months (OR = 0.351), caregiver education of high school or above (OR = 0.406), and household per-capita income ≥ 1,000 yuan (OR = 0.255). Rural residence (OR = 2.643) and ≥ 6 chemotherapy cycles (OR = 2.259) were risk factors for PEM. The nomogram demonstrated robust discrimination in both the modeling (AUC = 0.823, 95% CI: 0.767-0.868) and validation (AUC = 0.779, 95% CI: 0.685-0.859) cohorts. Calibration curves showed close agreement between predicted and observed probabilities, and Hosmer-Lemeshow tests indicated good fit (all P > 0.05). DCA confirmed meaningful net clinical benefit across a wide range of threshold probabilities. The nomogram provides an interpretable and clinically practical tool for early identification of PEM risk in pediatric cancer patients, supporting targeted nutritional screening and timely intervention.