Photo Credit: _David Gee
The following is a summary of “Higher Versus Lower Protein Delivery in Critically Ill Patients: A Systematic Review and Bayesian Meta-Analysis,” published in the December 2024 issue of Critical Care by Heuts et al.
Researchers conducted a retrospective study to update a Bayesian meta-analysis estimating the probabilities of beneficial and harmful effects of higher protein delivery in individuals with critical illness.
They performed an updated systematic search in 3 databases up to September 4, 2024, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines, with the protocol preregistered in PROSPERO (CRD42024546387). Randomized controlled trials involving adults with critical illness comparing protein doses delivered enterally and/or parenterally with similar energy delivery were included, 2 authors independently extracted data using a predefined worksheet, with mortality as the primary outcome. Posterior probabilities for benefit (relative risk [RR] < 1.00) or harm (RR > 1.00) were estimated alongside thresholds for significant effect sizes. The risk of bias was assessed using the Risk of Bias 2.0 tool, and analyses employed Bayesian hierarchical random-effects models with vague priors.
The results showed 22 randomized trials (n = 4164). The mean protein delivery was 1.5 ± 0.6 g/kg/d in the higher protein group and 0.9 ± 0.4 g/kg/d in the lower protein group. The median RR for mortality was 1.01 (95% credible interval: 0.84–1.16). The posterior probability of mortality benefit from higher protein delivery was 43.6%, while the probability of harm was 56.4%. The probabilities for a 1% (RR < 0.99) and 5% (RR < 0.95) reduction in mortality were 38.7% and 22.9%, respectively. Conversely, the probabilities for a 1% (RR > 1.01) and 5% (RR > 1.05) increase in mortality were 51.5% and 32.4%, respectively.
Investigators concluded the higher protein delivery in patients with critical illness likely increased mortality risk, although a clinically beneficial effect could not be entirely ruled out based on the current data.
