Interpreting protein dose trials in critical illness: a guide for the bedside clinician

Critical illness induces a catabolic state, associated with profound muscle wasting and functional disability in survivors. Based on evidence at the time of development, international critical care nutrition guidelines recommend providing higher protein doses than in healthy populations.

Three recent international multi-centre randomised trials (total n = 5633 patients) have compared higher protein doses to usual protein doses in critically ill patients. Based on the primary outcomes, these trials concluded that higher protein doses did not improve time-to-discharge alive or number of days free of the index hospital and alive at day 90, with worse functional recovery using the EQ-5D-5L health utility score over 180 days. Supported by evidence from recent trials, it appears preferable to commence protein delivery at low doses once patients are haemodynamically stable and increase progressively over the first 5 days to deliver a maximum of 1.2 g/kg/day. Based on current trial data, this upper limit may represent a safer alternative to doses >1.2 g/kg/day, acknowledging that the optimal dose may be lower. Subgroup analyses suggest that patients with an acute kidney injury may be particularly vulnerable to higher protein. There are no data to indicate the minimum protein dose that can be safely delivered to critically ill patients over their entire ICU stay. While existing trials included patients with a prolonged ICU stay, no trial identifying this cohort pre-randomisation has been conducted. It cannot be excluded that higher protein doses may provide benefit later in recovery, when the anabolic resistance to dietary protein observed early in the ICU admission has subsided. Evidence is also lacking on optimal protein targets for patients after ICU discharge.

In critically ill adults, we suggest that protein doses be commenced at a low dose and increased progressively to a maximum of 1.2 g/kg/day based on recent randomised trials.