RESUMO
BACKGROUND & AIMS: The Global Leadership Initiative on Malnutrition (GLIM) criteria has been recognised as major diagnostic criteria for malnutrition in adults worldwide; however, its validity in rehabilitation settings remains unclear. This study investigated the concurrent and predictive validity of the GLIM criteria for adult patients in convalescent rehabilitation wards. METHODS: This retrospective cohort study was conducted using pre-established datasets from convalescent rehabilitation wards in a hospital. The inclusion criteria were adults aged ≥18 years admitted to the wards between November 2018 and October 2020 who were available for body composition assessment. Malnutrition diagnoses were determined by registered dietitians (RDs) using the GLIM criteria. The Subjective Global Assessment (SGA) was performed by another RD and used for the malnutrition reference standard. The GLIM criteria sensitivity and specificity were examined for SGA. The odds ratios and hazard ratios of GLIM-defined malnutrition for the total score of the Functional Independence Measure (tFIM) effectiveness and non-home discharge were calculated using univariable and multivariable logistic regression analyses and Cox proportional hazard models. RESULTS: Data from 723 patients were extracted from the dataset. GLIM-defined malnutrition was confirmed in 207 (28.6%) patients, 87 (12.0%) with moderate malnutrition and 120 (16.6%) with severe malnutrition. The SGA graded 146 (20.2%) patients with moderate malnutrition (grade B) and 86 (11.9%) with severe malnutrition (grade C). The GLIM criteria (malnutrition/no malnutrition) had fair sensitivity (76.7%, 95% confidence interval [CI]: 70.7-82.0%) and good specificity (94.1%, 95% CI: 91.6-96.0%), indicating acceptable concurrent validity. GLIM-defined moderate malnutrition had poorer sensitivity than severe malnutrition (42.5% vs 81.4%). Logistic regression analyses revealed no evidence for the association between GLIM-defined malnutrition and poor tFIM effectiveness (adjusted odds ratio [AOR]: 1.09, 95% CI: 0.71-1.69) and non-home discharge (AOR: 1.19, 95% CI: 0.76-1.84). The Cox proportional hazard analyses also showed no effect of malnutrition on outcomes. CONCLUSION: The GLIM criteria had fair sensitivity and good specificity, indicating acceptable criteria for diagnosing malnutrition in rehabilitation settings. However, its predictive validity for functional recovery and discharge outcomes was insufficient.