Real-world implementation, documentation, and prognostic implications of routinely recorded nutritional assessment based on the Global Leadership Initiative on Malnutrition criteria in critically ill patients: A nationwide retrospective cohort study.

BACKGROUND & AIMS: We examined real-world documentation under the Global Leadership Initiative on Malnutrition (GLIM) framework, the proportion of critically ill patients with a recorded, classifiable GLIM nutritional status, and associations of recorded GLIM nutritional status with mortality. METHODS: This retrospective cohort study used a nationwide Japanese Diagnosis Procedure Combination database of adults admitted to an intensive care unit (June 2024-September 2025). The exposure was database-recorded, classifiable GLIM nutritional status (malnutrition or non-malnutrition). Multivariable Cox regression compared malnutrition with non-malnutrition for in-hospital mortality within 30 days of admission. Secondary outcomes were overall in-hospital mortality and length of hospital stay (LOS) among survivors. Additional analyses included the not-assessed and no-documentation categories. RESULTS: Among 51,580 patients (mean age 70.4 years [SD 14.6]; 60.4% male), 27,613 (53.5%) had a recorded, classifiable GLIM nutritional status; 10.2% were malnourished. Malnutrition was associated with higher in-hospital mortality within 30 days of admission than non-malnutrition (adjusted hazard ratio [aHR] 1.23, 95% confidence interval [CI] 1.08-1.41), higher overall in-hospital mortality (adjusted odds ratio [aOR] 1.47, 95% CI 1.28-1.68), and longer LOS among survivors (mean difference 2.27 days, 95% CI 1.44-3.10). The not-assessed and no-documentation categories were also associated with higher in-hospital mortality within 30 days of admission than non-malnutrition. CONCLUSIONS: A classifiable GLIM nutritional status was recorded in only approximately half of critically ill patients. Among these patients, recorded malnutrition was associated with higher mortality and longer hospital stay. Associations of non-classifiable categories with mortality require further study.