Evaluation of the Validity and Feasibility of the GLIM Criteria Compared with PG-SGA to Diagnose Malnutrition in Relation to One-Year Mortality in Hospitalized Patients

Abstract

BackgroundThe Global Leadership Initiative on Malnutrition (GLIM) approach to diagnose malnutrition was published in 2018. An important next step is to use the GLIM criteria in clinical investigations to assess their validity and feasibility. ObjectiveTo compare the validity and feasibility of the GLIM criteria with Patient-Generated Subjective Global Assessment (PG-SGA) in hospitalized patients and to assess the association between malnutrition and 1-year mortality. DesignPost hoc analysis of a prospective cohort study. Participants/settingHospitalized patients (n = 574) from the Departments of Gastroenterology, Gynecology, Urology, and Orthopedics at the Radboudumc academic facility in Nijmegen, The Netherlands, were enrolled from July 2015 through December 2016. Main outcome measuresThe GLIM criteria and PG-SGA were applied to identify malnourished patients. Mortality rates were collected from electronic patient records. Feasibility was assessed by evaluating the amount of and reasons for missing data. Statistical analyses performedConcurrent validity was evaluated by assessing the sensitivity, specificity, and Cohen’s kappa coefficient for the GLIM criteria compared with PG-SGA. Cox regression analysis was used for the association between the GLIM criteria and PG-SGA and mortality. ResultsOf 574 patients, 160 (28%) were classified as malnourished according to the GLIM criteria and 172 (30.0%) according to PG-SGA (κ = 0.22, low agreement). When compared with PG-SGA, the GLIM criteria had a sensitivity of 43% and a specificity of 79%. Mortality of malnourished patients was more than two times higher than for non-malnourished patients according to the GLIM criteria (hazard ratio [HR], 2.68; confidence interval [CI], 1.33-5.41). Data on muscle mass was missing in 454 of 574 (79%) patients because of practical problems with the assessment using bioimpedance analysis (BIA). ConclusionsAgreement between GLIM criteria and PG-SGA was low when diagnosing malnutrition, indicating that the two methods do not identify the same patients. This is supported by the GLIM criteria showing predictive power for 1-year mortality in hospitalized patients in contrast to PG-SGA. The assessment of muscle mass using BIA was difficult to perform in this clinical population.