To examine the relationship between the Mini Nutritional Assessment Short Form (MNA-SF) screening for malnutrition and the Global Leadership Initiative on Malnutrition (GLIM) diagnosis and the difference between the globally agreed GLIM two-step approach with prior screening (GLIMs+) and the alternative one-step approach without prior screening (GLIMs−) in a large group of geriatric patients.

Cross-sectional multicenter study in eight geriatric institutions in Germany.

681 geriatric patients.

Patient characteristics, MNA-SF and GLIM criteria were assessed in all patients using standardized questionnaires based on routine geriatric assessments. Malnutrition prevalence rates, Cohen’s kappa and diagnostic test parameters were calculated to compare MNA-SF results with GLIMs− and GLIMs− with GLIMs+.

88.1 % of all patients (83 ± 6.7 years; 68.0 % women) were MNA-SF positive (38.8 % malnourished, 49.3 % at risk). GLIMs− malnutrition was diagnosed in 44.6 %, GLIMs+ in 43.5 % of all patients. 8 patients (1.2 %) were normal according to MNA-SF and malnourished according to GLIMs− (false negative). Sensitivity of MNA-SF compared to GLIMs− was 0.97, specificity 0.19. Sensitivity of GLIMs− compared to GLIMs+ was 1.00, specificity 0.98. Cohen’s kappa indicated slight agreement between MNA-SF and GLIMs− (κ = 0.153) and near-perfect agreement between GLIMs− and GLIMs+ (κ = 0.976).

The very good agreement between GLIMs− and GLIMs+ may argue in favor of direct GLIM diagnosis without prior screening, but prior MNA-SF screening saved the effort of GLIM testing in 12 % of all patients with only very few false negatives and also identifies the large group of at-risk patients who may benefit from preventive interventions.