Abstract
OBJECTIVE: To investigate and compare the predictive efficacy of the Nutritional Risk Screening 2002 (NRS2002) and the Global Leadership Initiative on Malnutrition (GLIM) criteria for postoperative complications in patients with gastrointestinal malignancies. METHODS: This prospective cohort study enrolled patients with gastrointestinal malignancies scheduled for surgical resection at our institution from December 2019 to December 2024. Nutritional risk and status were assessed using NRS2002 and GLIM criteria within 24 h of admission. Data on postoperative complications, hospitalization expenses, length of hospital stay, and unplanned 30-day and 60-day readmissions were meticulously collected and analyzed using logistic regression and ROC curve analysis. RESULTS: A total of 471 patients were included. Nutritional risk (NRS2002 ≥ 3) was identified in 45.01% (n = 212) of patients. According to GLIM criteria, 43.10% (n = 203) were diagnosed with malnutrition. Multivariable logistic regression analysis revealed that nutritional risk (adjusted OR 7.58, 95%CI: 4.75-12.05, p < 0.001), GLIM-defined malnutrition (adjusted OR 5.62, 95%CI: 3.59-8.76, p < 0.001), moderate malnutrition (adjusted OR 4.78, 95%CI: 2.78-8.17, p < 0.001), and severe malnutrition (adjusted OR 6.71, 95%CI: 3.82-11.78, p < 0.001) were independent risk factors for postoperative complications. The Area Under the Curve (AUC) for NRS2002 in predicting complications was 0.735, which was significantly greater than the AUC for GLIM diagnosis (0.706; p = 0.003). No significant difference in AUC was observed between NRS2002 and GLIM severity grading (0.712; p = 0.215). Neither NRS2002 nor GLIM assessments were significantly associated with unplanned readmissions in adjusted models. CONCLUSION: Both NRS2002-defined nutritional risk and GLIM-defined malnutrition are significant independent predictors of postoperative complications in patients with gastrointestinal malignancies. The NRS2002 demonstrated slightly superior predictive ability for postoperative complications compared to the dichotomous GLIM diagnosis. These findings support the routine use of both tools for preoperative risk stratification to guide targeted nutritional interventions.