Systematic Review and Bayesian Meta-Analysis: Higher Protein Delivery May Increase Mortality in Critically Ill Patients

Background: Nutritional guidelines often recommend higher protein doses (approximately 1.2–2.0 g/kg/d) to mitigate muscle loss in critically ill patients. However, recent multicenter trials have raised concerns that elevated protein targets might increase mortality and adversely affect patient-centered outcomes. This study applied a Bayesian approach to synthesize current evidence regarding the effect of higher versus lower protein delivery on mortality, infections, mechanical ventilation duration, and health-related quality of life in critically ill adults.

Objective: To estimate the probability of beneficial or harmful effects of increased protein delivery on clinically important outcomes, with emphasis on quantifying the likelihood of mortality benefit versus risk.

Methods: A systematic review and Bayesian meta-analysis were conducted according to a preregistered protocol (PROSPERO CRD42024546387) and PRISMA 2020 guidelines. Twenty-two randomized controlled trials comparing higher (mean 1.5 g/kg/d) versus lower (mean 0.9 g/kg/d) protein delivery in adult ICU patients were included, ensuring similar energy intake in both groups. A hierarchical random-effects Bayesian model was applied, using vague priors to estimate relative risks for mortality and infections, mean differences for ventilator days, and standardized mean differences for quality of life.

Results: A total of 4,164 patients were analyzed. The posterior probability that higher protein intake increases mortality was 56.4%, compared with a 43.6% probability of any mortality benefit. Probabilities for a clinically relevant (≥5%) mortality decrease were low (22.9%), while the probability of at least a 5% increase reached 32.4%. Infections were slightly more likely with higher protein, although the likelihood of a major detrimental effect remained modest. The probability of a clinically meaningful difference in ventilator days was negligible, suggesting near equivalence for that endpoint. Conversely, quality of life might be negatively impacted by higher protein dosing, although few trials measured this outcome.

Conclusions: Under a Bayesian framework, current evidence suggests that high protein delivery in critically ill patients might pose a meaningful risk of increased mortality. Although a beneficial effect cannot be fully excluded, its probability appears comparatively small. These findings challenge the longstanding assumption that more protein universally translates to better outcomes.

Implications for Practice: Clinicians should exercise caution when aiming for higher protein targets. Individual patient characteristics, such as severity of illness, renal function, and underlying comorbidities, may modulate outcomes. The data support considering a personalized protein prescription rather than routinely pushing intake beyond conventional targets.

Study Strengths and Limitations: Strengths include a robust Bayesian analysis that evaluates probabilities of both benefit and harm across multiple thresholds, as well as the inclusion of recently published large trials. Limitations involve heterogeneity in protein dosing strategies, potential publication bias (indicated by Egger’s test), and limited data on quality of life.

Future Research: Ongoing trials, such as TARGET Protein and REPLENISH, will provide valuable insights into optimal protein dosing, particularly in specific subgroups. Further investigation should explore mechanistic underpinnings of how high protein intake could adversely affect recovery in critically ill patients.

Reference: Heuts S, Lee ZY, Lew CCH, et al. Higher Versus Lower Protein Delivery in Critically Ill Patients: A Systematic Review and Bayesian Meta-Analysis. Critical Care Medicine. December 27, 2024. DOI: http://doi.org/10.1097/CCM.0000000000006562