Evaluate the performance of phase angle (PA) in identifying malnutrition and in predicting clinical outcomes in critical adult patients.

A longitudinal observational study with secondary data from Nossa Senhora da Conceição Hospital (Porto Alegre) and Risoleta Tolentino Neves Hospital (Belo Horizonte) involving critically ill patients assessed for nutritional status by subjective global assessment (SGA) and by anthropometry in the first 48 h after admission to the intensive care unit (ICU). The PA was evaluated from the realization of the bioelectrical impedance. Patients were followed up until hospital discharge to verify the other outcomes of interest: death, hospitalization time and in ICU, and duration of mechanical ventilation.

A total of 169 patients (60.3 ± 16.7 years, 56.7% men, 46.7% surgical) were followed for 23.0 (14.0–40.8) days. The accuracy of standardized PA (SPA) reduced in identifying malnourished patients was 60.6% (ROC curve AUC = 0.606, 95% CI 0.519–0.694). Reduced SPA increased in about three times the chance of having malnutrition (OR = 2.79, 95% CI 1.39–5.61) and 2 times the chance of prolonged hospital stay (OR = 2.27; 95% CI 1.18–4.34) in an adjusted analysis for the origin hospital and for the severity score.

Reduced SPA showed satisfactory predictive validity for malnutrition and prolonged hospital stay in critically ill patients, reinforcing the applicability of BIA in the routine of nutritional care in ICU, since it is a simple, fast and low cost method.