NRS-2002 Deemed Practical Option for Routine Nutritional Screening in Colorectal Cancer

Colorectal cancer, intestinal carcinoma, bowel neoplasia, 3D illustration showing malignant tumor in intestine
Researchers compared GLIM, NRS-2002, and PG-SGA nutritional screening tools for sensitivity and specificity in patients with colorectal cancer.

Compared with the Global Leadership Initiative on Malnutrition (GLIM) and Patient Generated Subjective Global Assessment (PG-SGA) nutritional screening tools, Nutritional Risk Screening 2002 (NRS-2002) was found to be the most practical option in a busy health care setting for patients with colorectal cancer (CRC), according to study results published in the Journal of Clinical Epidemiology.

The retrospective analysis compared the 3 nutritional screening tools using data from the multicenter, prospective Investigation on Nutrition Status and its Clinical Outcome of Common Cancers (INSCOC) project of China. Eligible participants had CRC, were aged older than 18 years, and were hospitalized for over 48 hours.

The patients were enrolled consecutively from July 2013 to July 2018 and were regularly followed up via telephone or outpatient visits until death or end of the study in December 2019. The primary endpoint was all-cause mortality.

A total of 1358 patients (40% women) with CRC were included and contributed 47626.58 total person-months of observation (49.80 months median, IQR, 28.80-60.00). The cohort had a median age of 60 (IQR, 52-67) years and included 643 (47.35%) cases of colon cancer and 715 (52.65%) cases of rectal cancer.

Among the participants, 518 (38.14%) were at risk of malnutrition, according to the NRS-2002 (score ≥3); 824 (60.68%) were malnourished per the GLIM criteria; and 841 (61.93%) patients were malnourished, according to the PG-SGA (score ≥4).

According to results of the fitting Bayesian latent class model, the GLIM criteria demonstrated a moderate sensitivity (0.78; 95% credibility interval [CrI], 0.73-0.82) and a low specificity (0.62; 95% CrI, 0.58-0.67). The NRS-2002 had a high specificity (0.90; 95% CrI, 0.87-0.92) and a low sensitivity (0.64; 95% CrI, 0.59-0.69). The PG-SGA had a high sensitivity (0.96; 95% CrI, 0.92-0.98) and specificity (0.82; 95% CrI, 0.74-0.90).

A total of 381 (28.06%) patients died during the follow-up. A worse overall survival (OS) was associated with poor nutritional status (unadjusted HR [95% CI] for GLIM: 1.59 [1.28-1.98], P <.001; NRS-2002: 1.44 [1.18-1.76], P <.001; and PG-SGA: 1.32 [1.07-1.63], P =.011).

Malnutrition was associated with poor OS after multivariate adjustment for confounders (adjusted HR [95% CI] for GLIM: 1.49 [1.20-1.86], P <.001; NRS-2002: 1.28 [1.04-1.59], P =.020; and PG-SGA: 1.42 [1.14-1.77], P =.002).

Study limitations included the retrospective analysis of an observational cohort. In addition, when a participant’s weight from 6 months ago was not available, weight loss over 2% within 1 month was used as an alternative measure for the diagnosis of GLIM-defined malnutrition. Also, the GLIM recommends use of validated body composition measures, such as bioelectrical impedance analysis or dual-energy X-ray absorptiometry to determine muscle mass, which was not available.

“This study proved the diagnostic accuracy and prognostic ability of the GLIM, NRS-2002, and PG-SGA in [CRC] patients,” the study authors wrote. “Although the sensitivity of NRS-2002 is not high enough, compared with the complexity of PG-SGA, NRS-2002 is more practical for nutritional screening in busy, resource-stretched units. As a comprehensive nutritional assessment tool, PG-SGA may then be used to provide the basis for individualized nutrition pathways.”


Ruan X, Wang X, Zhang Q, et al. The performance of three nutritional tools varied in colorectal cancer patients: a retrospective analysis. J Clin Epidemiol. Published online May 7, 2022. doi:10.1016/j.jclinepi.2022.04.026