Influencing factors of malnutrition in patients with colorectal cancer and value of Nomogram prediction model

Objective To analyze the risk factors of malnutrition in patients with colorectal cancer (CRC) and construct a Nomogram prediction model.

Methods A total of 402 hospitalized CRC patients in the Anning First People's Hospital Affiliated to Kunming University of Science and Technology from July 2021 to December 2023 were selected as research objects. The Global Leadership Initiative on Malnutrition (GLIM) criteria was used as the diagnostic criteria for malnutrition, and patients were divided into malnutrition group and well-nourished group. A multivariate Logistic regression model was used to analyze the influencing factors of malnutrition in CRC inpatients. A Nomogram prediction model was constructed based on predictive factors, and the discrimination and accuracy of the model were validated by the receiver operating characteristic (ROC) curve and Hosmer-Lemeshow goodness-of-fit test. Finally, the clinical application value of the model was verified by calibration curves and clinical decision curves.

Results Among the 402 CRC patients, 111 cases had malnutrition, with a malnutrition rate of 27.61%. There were significant differences in age, tumor stage, long-term bedridden status, the Karnofsky Performance Scale (KPS) score, body mass index (BMI), the Nutritional Risk Screening 2002 (NRS-2002), red blood cell (RBC), white blood cell (WBC), hemoglobin (HGB), albumin (ALB), prealbumin (PAB), alanine aminotransferase (ALT), and urea levels between the malnutrition and well-nourished groups (P < 0.05). Multivariate Logistic regression analysis showed that age, tumor stage, long-term bedridden status, HGB, KPS score, and PAB were independent risk factors for malnutrition in CRC patients, and the sensitivity, specificity and area under the curve (AUC) of the Nomogram prediction model constructed based on these factors were 57.4%, 88.0% and 0.821 (95%CI, 0.773 to 0.870, P < 0.001) respectively. Based on internal validation, 1 000 samples were drawn by the Bootstrap self-sampling method, with a consistency index of 0.821. The calibration curve and clinical decision curve indicated that the Nomogram prediction model had good clinical application value.

Conclusion The Nomogram prediction model constructed on 6 factors such as advanced age, TNM classification of stage Ⅳ, poor KPS score, long-term bedridden status, decreased HGB and decreased PAB has a high predictive value for the risk of malnutrition in CRC patients.