Objectives: We investigated whether preexisting kidney function determines if ICU patients may benefit from increased (2.0 g/kg/d) protein intake.
Design: Post hoc, hypothesis-generating, subgroup analysis of a multicenter, phase 2, randomized clinical trial. All analyses were conducted by intention to treat and maintained group allocation. Ninety-day mortality was the primary outcome.
Setting: ICUs of 16 hospitals throughout Australia and New Zealand.
Patients: Adult critically ill patients expected to remain in the study ICU for longer than 2 days.
Interventions: Random allocation to receive a daily supplement of up to 100 g of IV amino acids to achieve a total protein intake of 2.0 g/kg/d or standard nutrition care.
Measurements and main results: A total of 474 patients were randomized: 235 to standard care and 239 to IV amino acid supplementation. There was a statistically significant interaction between baseline kidney function and supplementation with study amino acids (p value for interaction = 0.026). Within the subgroup of patients with normal kidney function at randomization, patients who were allocated to receive the study amino acid supplement were less likely to die before study day 90 (covariate-adjusted risk difference, -7.9%; 95% CI, -15.1 to -0.7; p = 0.034). Furthermore, amino acid supplementation significantly increased estimated glomerular filtration rate in these patients (repeated-measures treatment × time interaction p = 0.009). Within the subgroup of patients with baseline kidney dysfunction and/or risk of progression of acute kidney injury, a significant effect of the study intervention on mortality was not found (covariate-adjusted risk difference, -0.6%; 95% CI, -16.2 to 15.2; p = 0.95).
Conclusions: In this post hoc, hypothesis-generating, subgroup analysis, we observed reduced mortality and improved estimated glomerular filtration rate in ICU patients with normal kidney function who were randomly allocated to receive increased protein intake (up to 2.0 g/kg/d). We strongly recommend confirmation of these results in trials with low risk of bias before this treatment is recommended for routine care.