Background: In 2015 the European Society for Clinical Nutrition and Metabolism (ESPEN) presented new consensus criteria for the diagnosis of malnutrition. Whereas most previous definitions were based on involuntary weight loss and/or a low BMI, the ESPEN definition added Fat Free Mass Index (FFMI) to the set of criteria. Aim: To study the predictive value of the new ESPEN diagnostic criteria for malnutrition on survival, with specific focus on the additional value of FFMI. Methods: Included were 335 hospitalized adult patients of the VU University Medical Center Amsterdam (60% female, age 58 ± 18 y). Three sets of criteria for malnutrition were used to study the predictive value for survival: Dutch definition for malnutrition, ESPEN diagnostic criteria for malnutrition and ESPEN diagnostic criteria for malnutrition without FFMI criterion. The association between malnutrition and three-months and one-year overall survival was analyzed by log rank tests and Cox regression. In multivariate analyses, adjustments were made for gender, age, care complexity and length of stay. Results: Ninety patients (27%) were classified as malnourished by any of the sets of criteria; malnourished patients had significant lower survival rates than non-malnourished patients at three months (84% vs 94%; p = 0.01) and one year (76% vs 87%; p = 0.02). After adjustments, malnutrition remained significantly associated with three-months survival for the Dutch definition for malnutrition (HR:2.25, p = 0.04) and the ESPEN diagnostic criteria for malnutrition (HR:2.76, p = 0.02). Malnutrition remained significantly associated with one-year survival for the ESPEN diagnostic criteria for malnutrition (HR:2.17, p < 0.02) and the ESPEN diagnostic criteria for malnutrition without FFMI (HR:2.66, p < 0.01). Conclusion: The new ESPEN definition for malnutrition is predictive for both three-months and one-year survival in a general hospital population, whereas definitions without FFMI are predictive for either three-months or one year survival.
- ESPEN diagnostic criteria
- Predictive value