The performance of three nutritional tools varied in colorectal cancer patients: a retrospective analysis

• Published in: Journal of clinical epidemiology 2022-09, Vol.149, p.12-22
• Main Authors: Ruan, Xiaoli, Wang, Xiaonan, Zhang, Qi, Nakyeyune, Rena, Shao, Yi, Shen, Yi, Niu, Chen, Zhu, Lingyan, Zang, Zhaoping, Wei, Tong, Zhang, Xi, Ruan, Guotian, Song, Mengmeng, Miles, Toni, Liu, Fen, Shi, Hanping
• Format: Article
• Language: English
• Subjects: Bayesian; Cancer; Cancer therapies; Chemotherapy; Clinical outcomes; Colorectal cancer; Colorectal carcinoma; Comorbidity; Epidemiology; GLIM; Inflammation; Laboratories; Leadership; Malnutrition; Medical prognosis; Metastasis; NRS-2002; Nutrition; Nutrition assessment; Nutritional status; Patients; PG-SGA; Radiation therapy; Sensitivity; Software; Survival; Tumors
• ISSN: 0895-4356; 1878-5921
• DOI: 10.1016/j.jclinepi.2022.04.026

Nutritional screening tools should be sensitive, simple, and easy to use. Differing opinions among clinicians concern the simplicity of the three tools—the Global Leadership Initiative on Malnutrition (GLIM) criteria, Nutritional Risk Screening 2002 (NRS-2002), and Patient-Generated Subjective Global Assessment (PG-SGA). For each tool, we estimated prediction of overall survival (OS) in tumor staging, sensitivity, and specificity. The NRS-2002 is favored by clinicians because it is simple to use. We compared its sensitivity and specificity with the GLIM and PG-SGA. This is an analysis of data from 1,358 adult colorectal cancer patients recruited in a multicenter from July 2013 to July 2018. In Kaplan–Meier models, each tool was found to be significantly predictive of OS: NRS-2002 (1.28), GLIM (1.49), and PG-SGA (1.42). Use of any tool improved prediction of survival at tumor staging. NRS-2002 has superior specificity (0.90) to diagnose patients without nutritional deficits (GLIM = 0.62 and PG-SGA = 0.82). This study provides evidence for the superiority of NRS-2002 to accurately identify colorectal cancer patients without nutritional limitations. Compared with the complexity of the other tools, NRS-2002 is the simplest tool to use in routine nutritional screening in busy clinical practice. •Imperfect reference standard has been proved to lead to biased estimates of diagnostic accuracy.•This study used the Bayesian latent class model (LCM) to evaluate the sensitivity and specificity of the GLIM criteria, NRS-2002, and PG-SGA, which adjusted the imperfect gold standard bias.•Survival analyses showed the association between nutritional status and overall survival of colorectal cancer patients.Harrell's concordance index showed all these three tools improved the TNM staging system for survival prediction.