Publication

• Title: Effect of titrating positive end-expiratory pressure (PEEP) with an esophageal pressure–guided strategy vs an empirical high PEEP-FIO2 strategy on death and days free from mechanical ventilation among patients with acute respiratory distress syndrome: a randomised clinical trial
• Acronym: EPVent-2
• Year: 2019
• Journal published in: JAMA
• Citation: Beitler JR, Sarge T, Banner-Goodspeed VM, Gong MN, Cook D, Novack V, Loring SH, Talmor D; EPVent-2 Study Group. Effect of titrating positive end-expiratory pressure (PEEP) with an esophageal pressure–guided strategy vs an empirical high PEEP-FIO2 strategy on death and days free from mechanical ventilation among patients with acute respiratory distress syndrome: a randomised clinical trial. JAMA. 2019;321(9):846-857.

Context & Rationale

• Background

  • Low tidal volume ventilation improves survival in ARDS, but the clinical benefit (if any) of titrating PEEP remains uncertain.
  • Airway pressures alone can be misleading because pleural pressure (and chest wall/abdominal mechanics) varies widely in critical illness; identical airway pressures can translate to very different transpulmonary pressures (lung stress).
  • Earlier physiological and single-centre randomised data (EPVent, 2008) suggested oesophageal pressure–guided PEEP might improve oxygenation and compliance, with a signal towards improved survival, but the evidence base was small and potentially fragile.
  • “Higher PEEP” trials and meta-analyses suggested possible benefit in more hypoxaemic ARDS subgroups, but did not establish an optimal individualised method for PEEP selection.
• Research Question/Hypothesis

  • Does an oesophageal pressure–guided strategy to titrate PEEP (targeting slightly positive end-expiratory transpulmonary pressure, while limiting end-inspiratory transpulmonary pressure) improve patient-centred outcomes compared with an empirical high PEEP-FIO2 table in moderate-to-severe ARDS?
  • Is an invasive physiological “precision” approach superior to a simple, widely available oxygenation-based approach when both arms use lung-protective ventilation?
• Why This Matters

  • If effective, transpulmonary pressure–guided PEEP could provide a scalable physiological framework for individualising PEEP (particularly in patients with high pleural pressures), potentially reducing ventilator-induced lung injury without excessive overdistension or haemodynamic compromise.
  • If ineffective, routine oesophageal manometry for PEEP titration would represent unnecessary invasiveness, cost, and complexity compared with simpler PEEP-FIO2 strategies.

Design & Methods

• Research Question: In adults with moderate-to-severe ARDS early in their disease course, does titrating PEEP using oesophageal pressure (transpulmonary pressure–guided) improve a ranked composite outcome of death and ventilator-free days compared with an empirical high PEEP-FIO2 strategy?
• Study Type: Multicentre, randomised, phase II, investigator-initiated trial conducted in ICU settings at 14 centres in the United States and Canada; open-label clinical management with prespecified physiological protocols.
• Population:
  • Setting: Mechanically ventilated ICU patients at 14 academic centres (US/Canada) with ARDS early after meeting criteria.
  • Inclusion criteria (key): Adult patients (age threshold per protocol); ARDS defined by bilateral infiltrates/oedema not fully explained by cardiac failure; PaO2/FiO2 ≤200 while mechanically ventilated; enrolment targeted within 36 hours of meeting ARDS criteria; requirement for volume-controlled ventilation and deep sedation to permit standardised oesophageal pressure measurements.
  • Exclusion criteria (key): Contraindication to oesophageal balloon placement; inability to tolerate protocolised ventilation; prior “rescue” strategies before randomisation (e.g., ECMO/high-frequency oscillation in some cases); advanced comorbid conditions limiting full supportive care; pregnancy; other protocol-specified exclusions.
• Intervention:
  • Oesophageal pressure–guided PEEP strategy using a balloon catheter positioned in the mid-thoracic oesophagus, with formal measurement technique and calibration requirements.
  • After stabilisation and a standardised recruitment manoeuvre (35 cm H2O for 30 seconds), PEEP and FiO2 were selected from a prespecified table to target end-expiratory transpulmonary pressure (PLexp) in an allowed range (0 to 6 cm H2O) while maintaining lung-protective constraints.
  • Safety constraints included maintaining low tidal volume ventilation and limiting airway plateau pressure; end-inspiratory transpulmonary pressure targets/limits were incorporated into the titration algorithm.
  • Protocol applied for the first 7 study days (or until liberation from ventilation), with daily reassessment and titration steps. ¹
• Comparison:
  • Empirical high PEEP-FIO2 strategy: PEEP and FiO2 selected from a prespecified high PEEP-FIO2 table (oxygenation-targeted).
  • Same core lung-protective ventilation approach as the intervention arm (low tidal volume, airway pressure limits), with protocolised rescue therapy criteria.
  • Oesophageal balloon catheter placement was performed in both groups, but oesophageal measurements were not used to set PEEP in the control arm (reducing instrumentation-related performance bias while preserving treatment separation at the level of decision-making).
• Blinding: Not blinded for bedside clinicians (PEEP titration strategies were inherently visible); prespecified outcomes were largely objective (mortality; ventilator-free days), limiting detection bias for primary endpoints.
• Statistics: Power calculation targeted 200 participants to detect a probabilistic index of 0.59 (vs null 0.50) for a ranked composite outcome (death and ventilator-free days) with 85% power at a 2-sided α=0.05; primary analysis was intention-to-treat using the probabilistic index framework (rank-based composite approach). ¹
• Follow-Up Period: Primary composite assessed through day 28 (ventilator-free days) with mortality follow-up to day 60 and additional longer-term outcomes reported to 1 year.

Key Results

This trial was not stopped early. The planned sample size (200 randomised) was enrolled and primary follow-up through day 28 was completed.

Outcome	PES-guided PEEP	Empirical high PEEP-FIO2	Effect	p value / 95% CI	Notes
Primary outcome (ranked composite: death + ventilator-free days through day 28)	49.6% (probability of more favourable outcome)	Reference	Probabilistic index 0.496	95% CI 0.417 to 0.575; P=0.92	Death ranked worst; survivors ranked by ventilator-free days; no between-group difference.
28-day mortality	33/102 (32.4%)	30/98 (30.6%)	Risk difference +1.7%	95% CI −11.1% to +14.6%; P=0.88	Confidence interval compatible with clinically important benefit or harm.
Ventilator-free days through day 28 (deaths counted as 0)	Median 15.5 (IQR 0 to 23)	Median 17.5 (IQR 0 to 24)	Median difference 0 days	95% CI 0 to 0; P=0.93	Component of ranked primary outcome.
Days free from mechanical ventilation among survivors (through day 28)	Median 22 (IQR 15 to 24)	Median 21 (IQR 16.5 to 24)	Median difference 0 days	95% CI −1 to +2; P=0.85	Prespecified secondary endpoint.
Rescue therapy use	4/102 (3.9%)	12/98 (12.2%)	Risk difference −8.3%	95% CI −15.8% to −0.8%; P=0.04	Only prespecified secondary clinical outcome significantly different.
Acute kidney injury requiring renal replacement therapy	21/100 (21.0%)	32/96 (33.3%)	Risk difference −12.3%	95% CI −24.7% to 0.0%; P=0.056	Borderline statistical signal; not a prespecified primary endpoint.
Gross barotrauma	6/102 (5.9%)	5/98 (5.1%)	Risk difference +0.8%	95% CI −5.5% to +7.1%; P>0.99	No excess barotrauma detected with PES-guided approach.
60-day mortality	38/101 (37.6%)	37/98 (37.8%)	Risk difference −0.2%	95% CI −13.2% to +12.9%; P=0.98	No signal of medium-term survival benefit.
1-year mortality	44/100 (44.0%)	44/96 (45.8%)	Risk difference −1.8%	95% CI −16.0% to +12.4%; P=0.82	Long-term survival similar.
Barthel Index ≥95 among 1-year survivors (survey respondents)	26/41 (63.4%)	28/34 (82.4%)	Not reported	P=0.08	Functional outcome subset; response among survivors 69% vs 67%.
SF-12 Physical Component Score among 1-year survivors (survey respondents)	Median 38 (IQR 25 to 50)	Median 42 (IQR 30 to 54)	Not reported	P=0.64	No signal of improved long-term physical health-related quality of life.
SF-12 Mental Component Score among 1-year survivors (survey respondents)	Median 54 (IQR 43 to 57)	Median 55 (IQR 45 to 58)	Not reported	P=0.96	Long-term mental health-related quality of life similar.

• Despite a physiologically motivated intervention, the primary ranked composite outcome was essentially null (probabilistic index 0.496; 95% CI 0.417 to 0.575; P=0.92), with no meaningful signal for improved ventilator liberation or survival.
• Between-group separation in delivered ventilator settings and intermediate physiological variables was small at the group level (e.g., protocol day 1 PEEP 17.6 vs 17.3 cm H2O; end-expiratory transpulmonary pressure 0.7 vs 1.3 cm H2O; driving pressure 11.5 vs 10.8 cm H2O).
• Rescue therapy was less frequent with PES-guided PEEP (3.9% vs 12.2%; risk difference −8.3%; 95% CI −15.8% to −0.8%; P=0.04), but other prespecified secondary clinical endpoints were not different.

Internal Validity

• Randomisation and allocation:
  • Randomised 1:1 with site-stratified block randomisation using a web-based system (with backup envelopes described in the protocol). ¹
  • Trial conduct was overseen by an independent data and safety monitoring board.
• Dropout/exclusions:
  • 200 randomised; 2 withdrew consent before protocol initiation; analysis set included 102 (PES-guided) and 98 (control).
  • Day-28 primary follow-up was complete for the randomised cohort.
  • 1-year functional outcomes were limited by survey response among survivors (e.g., Barthel completion 69% vs 67%; SF-12 completion 69% vs 65%).
• Performance/detection bias:
  • Open-label bedside management was unavoidable (PEEP selection strategy could not be masked).
  • Primary outcome components (mortality; ventilator-free days) are objective and less vulnerable to outcome-assessment bias, though ventilator liberation can be influenced by clinician behaviour and co-interventions.
• Protocol adherence:
  • High adherence to protocol-specified PEEP targets during the first week (e.g., protocol day 1: within ±2 cm H2O in 95.8% vs 99.0%; days 1–7 overall: within ±2 cm H2O in 93.3% vs 94.6%).
  • Standard lung-protective ventilation principles were applied in both groups (low tidal volumes with airway pressure limits).
• Baseline characteristics:
  • Groups were broadly similar: mean age 56 vs 55 years; 47% vs 45% female; mean APACHE II 27 in both; baseline PaO2/FiO2 128 vs 120.
  • ARDS severity predominantly moderate-to-severe; most patients had pneumonia/aspiration or sepsis as precipitants.
• Separation of the variable of interest:
  • Delivered PEEP (protocol day 1): 17.6 ± 4.4 vs 17.3 ± 3.8 cm H2O (difference 0.4; 95% CI −0.7 to 1.4; P=0.52).
  • End-expiratory transpulmonary pressure (protocol day 1): 0.7 ± 3.4 vs 1.3 ± 3.1 cm H2O (difference −0.6; 95% CI −2.2 to 0.8; P=0.38).
  • Driving pressure (protocol day 1): 11.5 ± 3.4 vs 10.8 ± 3.1 cm H2O (difference 0.7; 95% CI −0.2 to 1.7; P=0.15).
  • Group-level trend over days 1–7: Mixed-effects model group coefficient for PEEP +1.26 cm H2O (SE 0.73; P=0.08) and for end-expiratory transpulmonary pressure +0.53 cm H2O (SE 0.55; P=0.33), suggesting modest between-group separation overall.
  • Potential for individual-level differences: At randomisation, a post hoc comparison suggested the assigned PEEP/FiO2 tables would have differed for 53.6% of patients, but this did not translate into sustained large group-level separation.
• Timing:
  • Enrolment targeted early after ARDS onset (within 36 hours), aligning with the biological rationale for preventing early atelectrauma/VILI.
• Dose and mechanism fidelity:
  • The intervention allowed higher maximum PEEP (up to 36 cm H2O) than the control table (maximum 24 cm H2O), but average PEEP exposure was similar, limiting “dose” separation.
  • The intervention mixed oxygenation and mechanics constraints (allowed combinations of FiO2 and transpulmonary pressure targets), which may have constrained purely mechanics-driven optimisation.
• Outcome assessment and statistical rigour:
  • Ranked composite outcome (death + ventilator-free days) reduces the conceptual problem of treating death as equivalent to prolonged ventilation in traditional ventilator-free day analyses.
  • Power assumptions were ambitious (large mortality and ventilator-free day improvements), yielding wide confidence intervals for key clinical outcomes.

Conclusion on Internal Validity: Overall, internal validity appears moderate to strong due to robust randomisation, high protocol adherence, and objective primary endpoints; however, open-label management and relatively limited between-group separation in delivered PEEP/transpulmonary pressures plausibly reduced the trial’s ability to detect a true physiological effect.

External Validity

• Population representativeness:
  • Patients resembled a typical academic-centre cohort with moderate-to-severe ARDS (mean age mid-50s; APACHE II ~27; PaO2/FiO2 ~120–130).
  • Exclusions (e.g., inability to tolerate oesophageal balloon placement; need for immediate rescue therapies; limitations of life-sustaining therapy) may reduce applicability to the sickest or most complex cases.
• Applicability:
  • Requires specific expertise, equipment, and training for oesophageal manometry technique and troubleshooting; many ICUs do not routinely have this capability.
  • Comparator was an empirical high PEEP-FIO2 approach; translation to ICUs using lower PEEP strategies (or more heterogeneous usual care) may differ.
  • Centres were predominantly academic; generalisability to resource-limited settings is constrained by instrumentation and staffing requirements.

Conclusion on External Validity: Findings are reasonably generalisable to moderate-to-severe ARDS managed in well-resourced ICUs, but are less generalisable to settings without oesophageal manometry capability and to practice environments where high PEEP-FIO2 strategies are not routinely used.

Strengths & Limitations

• Strengths:
  • Multicentre randomised design across 14 centres with prespecified physiological protocols.
  • Standardised low tidal volume ventilation in both groups, isolating the PEEP selection strategy as the key contrast.
  • High protocol adherence for prescribed PEEP targets during the intervention window.
  • Objective primary endpoint (ranked composite of death and ventilator-free days) with follow-up to 60 days and additional 1-year outcomes.
  • Instrumentation in both groups (oesophageal balloon placement) reduced differential co-intervention related to monitoring.
• Limitations:
  • Sample size (n=200) and effect-size assumptions produced wide confidence intervals for clinically important outcomes, limiting precision for modest treatment effects.
  • Open-label ventilation management can influence extubation practices and co-interventions (though mortality is less susceptible).
  • Delivered PEEP and transpulmonary pressures were similar at the group level, suggesting limited separation of the physiological exposure of interest.
  • Complexity and technical dependency of oesophageal manometry (placement, calibration, interpretation) limits scalability and may introduce variability across centres.
  • Long-term functional outcomes were available only for a subset of survivors (survey response limitations), constraining interpretation of recovery endpoints.

Interpretation & Why It Matters

• Clinical implication

  • In moderate-to-severe ARDS managed with lung-protective ventilation and a high PEEP-FIO2 approach, routine oesophageal pressure–guided PEEP titration did not improve survival or ventilator-free outcomes compared with a simpler empirical strategy.
• Mechanistic interpretation

  • The absence of benefit is biologically coherent given minimal between-group differences in delivered PEEP and intermediate physiological outcomes (plateau pressure, driving pressure, PaO2/FiO2), implying limited divergence in lung stress exposure.
• Where it may still matter

  • Transpulmonary pressure measurements may remain useful as an adjunct in selected patients with suspected high pleural pressures (e.g., obesity/abdominal hypertension) or complex chest wall mechanics, but EPVent-2 does not support a “default” invasive approach for most patients.

Controversies & Subsequent Evidence

• Interpretation of a null result given limited physiological separation:
  • The accompanying editorial emphasised that between-group differences in PEEP and intermediate physiological outcomes were minor, making a downstream clinical benefit biologically less plausible; it framed EPVent-2 as suggesting similar outcomes between a simple high PEEP-FIO2 approach and a more invasive, costly strategy for most moderate-to-severe ARDS patients. ²
• Power and effect-size assumptions:
  • The editorial highlighted that sample size assumptions required a very large clinical effect (approximately a 10% absolute mortality reduction plus improved ventilator-free days), leading to wide confidence intervals for mortality that remain compatible with clinically relevant benefit or harm. ²
• Technical challenges of oesophageal pressure measurement:
  • Accurate interpretation of oesophageal pressure as a pleural pressure surrogate requires careful attention to balloon filling volume, oesophageal wall mechanics, and measurement artefacts; expert frameworks describe when the measurement is most interpretable and how it should be used clinically. ³
  • In vivo calibration strategies have been proposed to improve reliability of oesophageal pressure measurements in mechanically ventilated patients, underscoring that the “test” is technique dependent and not a plug-and-play monitor. ⁴
• How EPVent-2 sits within the PEEP evidence base:
  • Prior higher-vs-lower PEEP RCTs were largely neutral on mortality, with subgroup and IPD meta-analytic signals that higher PEEP may benefit more hypoxaemic ARDS; EPVent-2 tested two “higher PEEP” strategies (physiology-guided vs empirical high PEEP-FIO2), finding no advantage of the invasive approach in this context.
  • EPVent-2 also provides a cautionary counterpoint to earlier small single-centre trials stopped early, where treatment effects can be exaggerated; this methodological concern has been systematically documented in critical care trials. ⁵

Summary

• EPVent-2 randomised 200 patients with moderate-to-severe ARDS to oesophageal pressure–guided PEEP vs an empirical high PEEP-FIO2 table, with lung-protective ventilation in both arms.
• The primary ranked composite outcome (death + ventilator-free days through day 28) was null (probabilistic index 0.496; 95% CI 0.417 to 0.575; P=0.92).
• Mortality at 28 days (32.4% vs 30.6%), 60 days (37.6% vs 37.8%), and 1 year (44.0% vs 45.8%) did not differ meaningfully between groups.
• Between-group differences in delivered PEEP and intermediate physiology were small at the group level, limiting mechanistic separation of lung stress as a target.
• Rescue therapy use was lower with PES-guided PEEP (3.9% vs 12.2%), but most other secondary outcomes (including long-term functional outcomes among respondents) were similar.

Overall Takeaway

EPVent-2 is a landmark “physiology vs pragmatism” trial in ARDS ventilation: despite a plausible mechanistic target (transpulmonary pressure), an oesophageal pressure–guided PEEP strategy did not improve a ranked composite of death and ventilator-free days compared with an empirical high PEEP-FIO2 strategy. The trial underscores that complex monitoring must meaningfully separate lung stress exposures to plausibly change downstream outcomes, and it supports prioritising scalable, evidence-based lung-protective strategies over routine invasive PEEP titration in most moderate-to-severe ARDS patients.

Bibliography

• 1Fish E, Novack V, Banner-Goodspeed VM, Sarge T, Loring S, Talmor D. The Esophageal Pressure-Guided Ventilation 2 (EPVent2) trial protocol: a multicentre, randomised clinical trial of mechanical ventilation guided by transpulmonary pressure. BMJ Open. 2014;4(9):e006356. Link
• 2Cavalcanti AB, Amato MBP, Serpa-Neto A. The elusive search for “best PEEP” and whether esophageal pressure monitoring helps. JAMA. Published online February 18, 2019. Link
• 3Mauri T, Yoshida T, Bellani G, et al; PLeUral pressure working Group (PLUG—Acute Respiratory Failure section of the European Society of Intensive Care Medicine). Esophageal and transpulmonary pressure in the clinical setting: meaning, usefulness and perspectives. Intensive Care Med. 2016;42(9):1360-1373. Link
• 4Mojoli F, Iotti GA, Torriglia F, et al. In vivo calibration of esophageal pressure in the mechanically ventilated patient makes measurements reliable. Crit Care. 2016;20:98. Link
• 5Montori VM, Devereaux PJ, Adhikari NKJ, et al. Randomized trials stopped early for benefit: a systematic review. JAMA. 2005;294(17):2203-2209. Link