Assessment of tools for the diagnosis of malnutrition and ICU outcomes in critically ill adults: A comparison study
Bruna Barbosa Stello, Elisa Loch Razzera, Danielle Silla Jobim Milanez, Flávia Moraes SilvaAbstract
Background
To evaluate the concurrent and predictive validity of the Academy of Nutrition & Dietetics/ASPEN Indicators for Malnutrition (AAIM) in critically ill patients compared with the SGA, considering disease etiology and fluid accumulation.
Methods
This was a comparison study. Malnutrition was assessed within 24 h of ICU admission using the Subjective Global Assessment (SGA) and the AAIM consensus. The AAIM approach was applied assuming all patients had acute disease (AAIM I) or differentiating acute versus chronic disease‐related malnutrition (AAIM II). Two variations were tested based on fluid accumulation: including (AAIM a) or excluding (AAIM b) this component. Concurrent validity was evaluated against SGA using diagnostic accuracy metrics, and predictive validity was assessed using hospital and ICU length of stay and mortality.
Results
Among 450 critically ill patients [median age 64 (54–71) years; 53.1% male; APACHE II score 18 (11–23)], 46.3% were malnourished by SGA, whereas AAIM prevalence ranged from 70.9% to 75.5%. AAIM showed moderate sensitivity (75.1%–78.9%) but low specificity (28.6%–32.4%) and poor agreement with SGA ( κ = 0.066–0.073). AAIM‐defined malnutrition was not associated with outcomes, while SGA‐defined malnutrition predicted prolonged hospital stay (OR 1.48; 95% CI 1.00–2.20) and ICU mortality (HR 2.26; 95% CI 1.42–3.49).
Conclusion
In critically ill patients, AAIM overestimated malnutrition prevalence and lacked concurrent and predictive validity, whereas SGA showed prognostic value.