Accuracy of the GLIM criteria for diagnosing malnutrition: A systematic review and meta-analysis

Abstract

Background & aims: Although malnutrition remains a global public health concern, and has proved to be a major contributor to death and illness, there has been a foundational lack of a gold standard for diagnostic testing for clinical application. The Global Leadership Initiative on Malnutrition (GLIM) criteria were established to normalize the diagnosis of malnutrition, but their use remains controversial. Therefore, we carried out a meta-analysis based on the published literature to assess the accuracy of the GLIM criteria for diagnosing malnutrition.

Methods: We utilized publication databases (including CENTRAL, MEDLINE, and EMBASE) to acquire research studies published from the initial use of the GLIM criteria in 2019 until January 22, 2022 that used the criteria to diagnose malnutrition. We conducted this meta-analysis with reference to the recommendations from the PRISMA-DTA statement. We separately calculated the amalgamated sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), and AUC with 95%CI for the GLIM criteria. Then, we aggregated and presented the data by drawing forest plots to assess the real accuracy of the criteria. A subgroup analysis was also carried out to identify the potential sources of heterogeneity.

Results: After the initial search of the CENTRAL, EMBASE, and MEDLINE databases, a total of 451 unique studies were identified. Twenty studies met our selection standards and 10,781 total patients were included in the meta-analysis. We noted that 4761 of the 10,781 patients (44.2%) were malnourished. The amalgamated sensitivity of the GLIM criteria was 0.72 (95%CI, 0.64-0.78), the specificity was 0.82 (95%CI, 0.72-0.88), the PLR was 3.9 (95%CI, 2.6-6.1), NLR was 0.35 (95%CI, 0.27-0.44), DOR was 11 (95%CI, 6-20), and AUC was 0.82 (95%CI, 0.79-0.85). Based on the results of a subgroup analysis using the SGA as a reference standard, the GLIM criteria had better diagnostic value (sensitivity, 0.81; specificity, 0.80; DOR, 17; AUC, 0.87).

Conclusions: The GLIM criteria have high diagnostic accuracy for distinguishing patients with malnutrition, and the GLIM criteria seem to have the potential to be used as a gold standard for diagnosing malnutrition in clinical practice. Moreover, the subgroup analysis showed a better diagnostic value for the GLIM criteria compared to the SGA used as a reference standard. Large-scale diagnostic trials and additional refinements to simplify the criteria are urgently needed to increase the clinical utilization of the GLIM criteria in the future.

Keywords: Diagnostic accuracy; GLIM criteria; Malnutrition; Meta-analysis.