Highlights
- A Bayesian re-analysis of the RELAx trial reveals a 75% to 78% probability that a lower PEEP strategy (0–5 cm H2O) is superior to a higher PEEP strategy (8 cm H2O) for increasing ventilator-free days in patients without ARDS.
- The probability of clinical benefit with lower PEEP exceeds 90% in specific subgroups, including patients not admitted for cardiac arrest and those intubated for reasons other than respiratory failure.
- In patients with acute brain injury (ABI), higher PEEP is independently associated with increased odds of extubation failure and ICU mortality, reinforcing the need for restrictive PEEP strategies in this population.
- While personalized PEEP titration (e.g., EIT or transpulmonary pressure-guided) shows promise in ARDS, evidence suggests that for the general non-ARDS ICU population, a lower, more restrictive PEEP approach is likely safer and more effective.
Background
Positive end-expiratory pressure (PEEP) is a fundamental component of mechanical ventilation, traditionally utilized to prevent alveolar collapse and improve oxygenation. In patients with Acute Respiratory Distress Syndrome (ARDS), higher PEEP levels are often required to maintain lung recruitment. However, the optimal PEEP strategy for the majority of intensive care unit (ICU) patients—those who do not meet ARDS criteria—remains a subject of intense clinical debate. Historically, a “standard” PEEP of 5 to 8 cm H2O has been applied across most populations as a preventive measure against atelectasis.
The REstricted vs. Liberal positive end-expiratory pressure in patients without ARDS (RELAx) trial was designed to address this uncertainty. The initial frequentist analysis of this multicenter randomized clinical trial concluded that a lower PEEP strategy was non-inferior to a higher PEEP strategy regarding ventilator-free days at day 28 (VFD-28). However, the binary nature of frequentist statistics (significant vs. non-significant) often obscures the nuance of treatment effects. The recent Bayesian re-analysis (Caroli et al., 2026) provides a more granular assessment of the probability of benefit, harm, or equivalence, offering clinicians a sophisticated framework for bedside decision-making.
Key Content
The Bayesian Re-analysis of the RELAx Trial
The post hoc Bayesian re-analysis of the RELAx data (PMID: 41914828) included 980 patients randomized to either the lowest possible PEEP (0–5 cm H2O) or a standard higher PEEP (8 cm H2O). Unlike the original frequentist approach, this analysis calculated the posterior probability of superiority for the lower PEEP group across various clinical outcomes.
Primary Outcome: Ventilator-Free Days (VFD-28)
The analysis found an odds ratio (OR) of 1.08 (95% credible interval, 0.87–1.35) for higher VFD-28 with lower PEEP. Crucially, the probability that the lower PEEP strategy was superior to the higher PEEP strategy ranged from 75% to 78% across different prior distributions (neutral, optimistic, or pessimistic). This suggests that while non-inferiority was the original conclusion, the weight of evidence leans toward a meaningful clinical advantage for the restrictive approach.
Secondary Outcomes: Mortality and Duration
For 28-day mortality, the probability of benefit for the lower PEEP strategy was high, ranging from 72% to 89%. Conversely, the probability of benefit regarding the total duration of ventilation was lower (11% to 28%), suggesting that while lower PEEP may improve survival and ventilator-free time, it does not necessarily shorten the absolute duration of the initial ventilation bout for all patients.
Heterogeneity of Treatment Effect: Subgroup Insights
The Bayesian framework excelled in identifying Heterogeneity of Treatment Effect (HTE). The probability of benefit for the lower PEEP strategy exceeded 90% in patients admitted for reasons other than cardiac arrest and those intubated for reasons other than respiratory failure. This suggests that in “healthier” lungs (e.g., surgical recovery or neurological monitoring), higher PEEP (8 cm H2O) may be unnecessary and potentially deleterious.
Evidence in Acute Brain Injury (ABI)
A separate secondary analysis of the ENIO study (PMID: 41061337) and the PROLABI trial (PMID: 39288368) reinforces the risks of higher PEEP in non-ARDS patients with brain injury. In 1,154 ABI patients, higher median PEEP was independently associated with an OR of 1.13 for extubation failure. Furthermore, the PROLABI trial demonstrated that a lung-protective strategy utilizing higher PEEP in ABI patients did not reduce mortality or ARDS incidence and was associated with a higher composite risk of death and ventilator dependency (61.5% vs 45.3%). This suggests that in the neuro-critical care setting, the hemodynamic and intracranial pressure implications of PEEP may outweigh its respiratory benefits.
Contrast with ARDS Management: Personalized Titration
While the evidence for non-ARDS patients shifts toward lower PEEP, management of established ARDS is evolving toward personalized titration. Studies such as the trial on PEEP-guided Electrical Impedance Tomography (EIT) (PMID: 41318662) show that EIT-guided titration can improve oxygenation and static compliance compared to standard low PEEP/FiO2 tables. Similarly, PEEP titration based on transpulmonary driving pressure (PMID: 39158781) has been shown to reduce mechanical ventilation duration and enhance 60-day survival in ARDS patients.
These divergent trends—restrictive PEEP for non-ARDS and personalized, often higher PEEP for ARDS—highlight the importance of accurate phenotype identification. As shown in the Bayesian analysis of the ALVEOLI and ART trials (PMID: 38768777), higher PEEP was actually harmful in one specific ARDS subphenotype (Subphenotype A), with a 94–97% probability of harm, emphasizing that even within lung injury cohorts, “liberal” PEEP is not universally beneficial.
Expert Commentary
The shift from a “standard” PEEP of 8 cm H2O to a “restrictive” PEEP of 0–5 cm H2O in non-ARDS patients represents a significant change in the philosophy of protective ventilation. The Bayesian re-analysis of RELAx provides the necessary statistical confidence for clinicians to adopt this change. The mechanistic rationale is clear: in patients without significant alveolar collapse, higher PEEP may lead to alveolar overdistension, increased mechanical power, and right ventricular strain without providing additional recruitment.
The finding that patients with acute brain injury or those intubated for non-respiratory reasons benefit most from lower PEEP is particularly relevant. In these patients, the lungs are often physiologically normal, and the adverse effects of PEEP on venous return and cerebral venous outflow are more clinically significant than any theoretical prevention of atelectasis. However, clinicians must remain vigilant; a PEEP of 0–5 cm H2O requires careful monitoring for progressive derecruitment, especially in obese patients or those with high intra-abdominal pressure.
A remaining controversy is the role of driving pressure. While driving pressure-limiting strategies are a cornerstone of ARDS management, their feasibility in non-ARDS patients is more limited because these patients already exhibit relatively low driving pressures. The focus in non-ARDS ventilation should likely remain on minimizing unnecessary PEEP and ensuring rapid weaning.
Conclusion
The evidence synthesized from the RELAx Bayesian re-analysis and supporting literature indicates that a lower PEEP strategy (0–5 cm H2O) is likely superior to a liberal PEEP strategy (8 cm H2O) for the majority of ICU patients without ARDS. The high probability of benefit (75–78% overall, and >90% in specific subgroups) suggests that a restrictive PEEP approach should be the default for patients intubated for neurological or surgical reasons. Future research should focus on real-time monitoring tools, such as EIT or automated P0.1 measurements, to identify the exact moment a patient transitions from needing PEEP for recruitment to being at risk for PEEP-induced injury.
References
- Caroli A, et al. Effect of a Lower Vs. Higher Positive End-Expiratory Pressure Strategy on Clinically Relevant Outcomes in ICU Patients Without Acute Respiratory Distress Syndrome: Bayesian Re-analysis of the RELAx Randomized Clinical Trial. Crit Care Med. 2026. PMID: 41914828.
- Algera AG, et al. Comparison of the effect of a lower versus a higher PEEP strategy on clinically relevant outcomes… statistical re-analysis plan. Crit Care Sci. 2025. PMID: 41416916.
- Robba C, et al. Associations of positive end-expiratory pressure (PEEP) with extubation failure… in patients with acute brain injury. J Crit Care. 2026. PMID: 41061337.
- Duarte AG, et al. Lung-Protective Mechanical Ventilation in Patients with Severe Acute Brain Injury: A Multicenter Randomized Clinical Trial (PROLABI). Am J Respir Crit Care Med. 2024. PMID: 39288368.
- He H, et al. The impact of PEEP-guided electrical impedance tomography on oxygenation and respiratory mechanics in moderate-to-severe ARDS. Sci Rep. 2025. PMID: 41318662.
- Golan E, et al. Differential Effect of Positive End-Expiratory Pressure Strategies in Patients With ARDS: A Bayesian Analysis of Clinical Subphenotypes. Chest. 2024. PMID: 38768777.
