Applicability of Global Leadership Initiative on Malnutrition (GLIM) criteria in nutrition risk screening for patients over 60 years old with digestive system tumors – A retrospective study

Abstract

Objective

This study aimed to identify the most effective nutritional risk screening tool for elderly patients with digestive system tumors.

Methods

Nutritional risk screening was performed using Nutritional Risk Screening 2002 (NRS2002), prognostic nutritional index (PNI) and Geriatric Nutritional Risk Index (GNRI) for elderly patients who underwent surgical resection of digestive tumor. Comparative analysis of each nutritional screening tool was conducted through Kappa test and ROC curve.

Results

Malnourished individuals diagnosed by the Global Leadership Initiative on Malnutrition (GLIM) criteria had lower levels of hematological parameters (serum albumin, pre-albumin, total protein and hemoglobin) and poorer body composition parameters (body mass index, body fat percentage, edema index, upper arm muscle circumference, skeletal muscle index, and phase angle; P < 0.05). NRS2002 demonstrated a sensitivity of 75.56% and specificity of 38.16% under the GLIM criteria, showing moderate agreement (Kappa = 0.346, P < 0.001). PNI had an average sensitivity of 71.11% and the highest specificity at 87.23%, but exhibited no consistency with GLIM (Kappa = 0.152, P = 0.062). GNRI showed the highest sensitivity at 91.11% and general specificity at 37.14%, aligning closely with GLIM criteria (Kappa = 0.711, P < 0.001). GNRI had the highest predictive value (AUC = 0.870, 95% CI: 0.801–0.939), followed by NRS2002 (AUC = 0.687, 95% CI: 0.589–0.785). Patients diagnosed with malnutrition via GNRI-GLIM exhibited higher rates of surgical site infection (χ² = 15.534, P < 0.001) and 3-month readmission (χ² = 4.499, P = 0.034).

Conclusion

GNRI and NRS2002 demonstrate good performance as GLIM criteria for nutritional screening in elderly patients with digestive tumors, with GNRI being potentially more suitable. Moreover, GNRI-GLIM can predict short-term prognosis in these patients.