Malnutrition Screening Tools for Geriatric Patients Presenting to Emergency Department: Agreement and Prognostic Utility for Hospital Admission and Length of Stay

Background: Malnutrition is a major global geriatric health problem, reported in approximately one-fifth of older adults worldwide and occurring even more frequently in acute care and hospital settings. Among older adults presenting to the emergency department (ED), nutritional vulnerability is often underrecognized because early ED decision-making is primarily dominated by acute physiological instability and the need for rapid disposition. Clarifying whether commonly used malnutrition screening tools provide clinically useful information beyond frailty, comorbidity burden and acute illness severity may help determine their role in early geriatric ED risk stratification, in-hospital care planning, and resource utilization. Objectives: To evaluate the prevalence, agreement, and clinical utility of three validated malnutrition screening tools the Malnutrition Universal Screening Tool (MUST), Nutritional Risk Screening 2002 (NRS-2002), and the clinician-administered Subjective Global Assessment (SGA) in older ED patients, and to examine their associations with hospital admission and length of stay (LOS). Methods: This prospective single-center study included 325 patients aged ≥65 years presenting to the ED. Nutritional status was assessed using the MUST, the NRS-2002, and the SGA. Agreement between tools was evaluated using Cohen’s kappa, positive percent agreement, and negative percent agreement. Associations with hospital admission were analyzed using multivariable logistic regression adjusted for age, sex, Clinical Frailty Scale, National Early Warning Score 2, and Charlson Comorbidity Index. Multivariable linear regression was used to identify predictors of LOS. Results: Overall, 32.6% of patients required hospital admission. Among admitted patients, the median hospital length of stay was 5 days (IQR 2–9). The prevalence of high nutritional risk varied substantially across tools, from 16.6% with the MUST to 41.5% with the NRS-2002 and 23.4% with the SGA. Agreement between tools was moderate overall (κ = 0.41–0.60), with moderate concordance in identifying low-risk and high-risk patients. After adjustment for clinically relevant covariates, none of the screening tools was independently associated with hospital admission. However, high-risk classification by the NRS-2002 was independently associated with prolonged LOS (β = 0.47, 95% CI 0.10–0.85; p = 0.01), whereas the MUST and the SGA were not. Conclusions: In older ED patients, malnutrition screening tools did not add independent value for predicting immediate hospital admission beyond frailty, comorbidity burden, and acute illness severity. However, the NRS-2002 was associated with longer hospital stay, suggesting potential value for early identification of patients who may require more complex in-hospital care and resource planning.