Construction and validation of a predictive model for the risk of malnutrition in hospitalized patients over 65 years of age with malignant tumours: a single-centre retrospective cross-sectional study

• Published in: PeerJ (San Francisco, CA) CA), 2024-12, Vol.12, p.e18685
• Main Authors: Wang, Xuexing, Chu, Jie, Wei, Chunmei, Xu, Jinsong, He, Yuan, Chen, Chunmei
• Format: Article
• Language: English
• Subjects: Aged; Aged patients; Aged, 80 and over; Cross-Sectional Studies; Female; Geriatric Assessment - methods; Hospital patients; Hospitalization - statistics & numerical data; Humans; Male; Malnutrition; Malnutrition - diagnosis; Malnutrition - epidemiology; Neoplasms - complications; Neoplasms - epidemiology; Nomograms; Nutrition Assessment; Nutritional assessment; Nutritional Status; Prediction model; Retrospective Studies; Risk Assessment; Risk Factors; Tumor staging; Tumors; Type 2 diabetes
• ISSN: 2167-8359; 2167-8359
• DOI: 10.7717/peerj.18685

Nutritional status is a critical indicator of overall health in individuals suffering from malignant tumours, reflecting the complex interplay of various contributing factors. This research focused on identifying and analysing the factors influencing malnutrition among older patients aged ≥65 with malignant tumours and aimed to develop a comprehensive risk model for predicting malnutrition. This study conducted a retrospective analysis of clinical data from 3,387 older inpatients aged ≥65 years with malignant tumours collected at our hospital from July 1, 2021, to December 31, 2023. The dataset was subsequently divided into training and validation sets at an 8:2 ratio. The nutritional status of these patients was evaluated using the Nutritional Risk Screening Tool 2002 (NRS-2002) and the 2018 Global Leadership Initiative on Malnutrition (GLIM) Standards for Clinical Nutrition and Metabolism. Based on these assessments, patients were categorized into either malnutrition or non-malnutrition groups. Subsequently, a risk prediction model was developed and presented through a nomogram for practical application. The analysis encompassed 2,715 individuals in the development cohort and 672 in the validation cohort, with a malnutrition prevalence of 40.42%. A significant positive correlation between the incidence of malnutrition and age was observed. Independent risk factors identified included systemic factors, tumour staging (TNM stage), age, Karnofsky Performance Status (KPS) score, history of alcohol consumption, co-infections, presence of ascites or pleural effusion, haemoglobin (HGB) levels, creatinine (Cr), and the neutrophil-to-lymphocyte ratio (NLR). The predictive model exhibited areas under the curve (AUC) of 0.793 (95% confidence interval (CI) [0.776-0.810]) for the development cohort and 0.832 (95% CI [0.801-0.863]) for the validation cohort. Calibration curves indicated Brier scores of 0.186 and 0.190, while the Hosmer-Lemeshow test yielded chi-square values of 5.633 and 2.875, respectively ( > 0.05). Decision curve analysis (DCA) demonstrated the model's clinical applicability and superiority over the NRS-2002, highlighting its potential for valuable clinical application. This study successfully devised a straightforward and efficient prediction model for malnutrition among older patients aged 65 and above with malignant tumours. The model represents a significant advancement as a clinical tool for identifying individuals at high risk of malnutrition