Background: Children who receive in-hospital treatment of severe acute malnutrition often have high mortality rates, and the reasons are not well understood. Objective: We assessed risk factors for death in children who were treated for malnutrition in a hospital. Design: In a prospective observational study of 120 children who were receiving in-hospital treatment of severe acute malnutrition in Uganda with therapeutic formulas F-75 and F-100, we collected data on symptoms, clinical findings, plasma markers of refeeding syndrome (electrolytes and phosphate), and acute phase reactants, and recorded the nutritional therapy given in hospital. Results: Seventeen children (14%) died. Clinical risk factors for death were the presence of oral thrush (HR: 5.0; 95% CI: 1.6, 15.2), a caretaker-reported severity of illness on a visual analog scale (HR: 1.7; 95% CI: 1.1, 2.6), impaired consciousness (HR: 16.7; 95% CI: 3.1, 90.4), and a capillary refill time .2 s (HR: 3.9; 95% CI: 1.4, 11.3). HIV infection was not associated with mortality (HR: 3.0; 95% CI: 0.7, 12.4), which was most likely due to low power. Biochemical risk factors were a plasma C-reactive protein concentration .15 mg/L on admission and low plasma phosphate that was measured on day 2 (HR: 8.7; 95% CI: 2.5, 30.1), particularly in edematous children. The replacement of F-75 with unfortified rice porridge to ameliorate diarrhea was associated with a higher risk of death, particularly if given during the first 2 d (HR: 5.0; 95% CI: 1.9, 13.3), which was an association that remained after adjustment for potential confounders (HR: 69.5; 95% CI: 7.0, 694.6). Conclusions: Refeeding syndrome may occur in children who are treated for malnutrition, even with moderately low plasma phosphate, and, in particular, in children with edematous malnutrition. The replacement of F-75 with unfortified rice porridge is associated with increased risk of death, which is possibly mediated by lowering plasma phosphate. The identified clinical risk factors may potentially improve the triage of children with malnutrition. This trial was registered at www.isrctn.com as ISRCTN55092738.
