Accuracy of three tools for malnutrition diagnosis in hospitalised patients: Comparison to subjective global assessment

Background Malnutrition is prevalent in hospital, and the Subjective Global Assessment (SGA) has been widely used for its identification. However, in the last decade, new tools were proposed by the Academy of Nutrition and Dietetics-American Society for Parenteral and Enteral Nutrition (AND-ASPEN), European Society for Clinical Nutrition and Metabolism (ESPEN) and Global Leadership Initiative on Malnutrition (GLIM). The diagnostic test accuracy of these tools has been scarcely investigated. Thus, we aimed to compare the accuracy of AND-ASPEN, ESPEN and GLIM for malnutrition diagnosis in hospitalised patients. Methods A cross-sectional study was conducted with hospitalised patients aged >= 18 years from a five-unit complex hospital. Malnutrition was diagnosed within 48 h of admission using SGA, AND-ASPEN, ESPEN and GLIM. The accuracy of these tools was evaluated by the area under the receiver operating characteristic (AUROC) curve, considering SGA as reference, which was compared by the DeLong test. Results Six hundred patients (55.7 +/- 14.8 years, 51.3% male) were evaluated. AND-ASPEN [AUROC 0.846; 95% confidence interval (CI) = 0.810-0.883] and GLIM presented a satisfactory accuracy (AUROC 0.842; 95% CI, 0.807-0.877), whereas ESPEN had a substantially lower accuracy (AUROC, 0.572; 95% CI, 0.522-0.622). The AUROC of AND-ASPEN and GLIM were not different from each other (p = 0.785) and both had significantly higher accuracy than ESPEN (p < 0.001). AND-ASPEN and GLIM presented sensitivity, specificity and negative predictive value > 80%, whereas ESPEN sensitivity was < 20%. Conclusions AND-ASPEN and GLIM were accurate methods for diagnosing malnutrition and could be applied in hospitalised patients. By contrast, the ESPEN criteria had unsatisfactory accuracy.