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

Beitler JR et al. for the EPVent-2 Study Group. JAMA February 2019; DOI:10.1001/jama.2019.0555

Clinical Question

  • Does titrating positive end-expiratory pressure (PEEP) with the use of an esophageal balloon to estimate pleural pressure improve outcomes compared with an empirical high PEEP –FiO2 strategy in patients with moderate – severe acute respiratory distress syndrome (ARDS)


  • Despite decades of research, the clinical benefits of higher PEEP and the best method to titrate PEEP for patients with ARDS remain unclear https://www.atsjournals.org/doi/full/10.1164/rccm.201610-2035CI
  • Plateau pressure (Pplt) measured during an inspiratory hold is typically used to approximate alveolar distending pressure
  • The true distending pressure of a collapsible structure (like an alveolus) is defined by its transmural pressure (pressure inside minus pressure outside). For the lung, this pressure is termed the transpulmonary pressure (PL) and is defined by airway pressure minus pleural pressure
  • For many mechanically ventilated patients, the contribution of pleural pressure to PL is minimal and Pplt serves as a reasonable approximation of alveolar stress. However, in patients with chest wall deformities, obesity, significant ascites, etc., Pplt may be an unreliable surrogate for PL
  • Esophageal manometry using an esophageal balloon provides an approximation of pleural pressure through the measurement of esophageal pressure (PES) and may provide a mechanism to tailor PEEP to a patient’s lung mechanics
  • This approach was studied in a previous single-center trial (the EPVent study) which was stopped early due to a significantly higher PaO2/FiO2 in the intervention arm (https://www.nejm.org/doi/full/10.1056/NEJMoa0708638). However, the use of a low PEEP strategy in the control group may have biased the trial towards a positive result


  • Multicenter prospective randomized phase II clinical trial
  • Blinding not performed given the nature of the intervention
  • Randomization performed using a random sorting algorithm through central web-based software using a balanced randomization scheme. The maximum allowable deviation was 6.5%
  • Power calculation
    • Investigators estimated a sample of 200 patients would be required to provide 85% power to detect a significant difference in the primary ranked composite outcome with a 2-sided alpha of 0.05. This calculation assumed the following
      • 28-day mortality of 30% in the control arm and 20% in the intervention arm
      • Proportion of patients with 0 days free from mechanical ventilation of 10% in control arm and 15% in intervention arm
      • Remaining non-zero values were assumed to be normally distributed with a mean (SD) of 13 (6.5) and 15 (6.5) days free from mechanical ventilation in the control arm and the intervention arm respectively
    • Analysis was performed using an intention-to-treat approach


  • 14 hospitals across the United States and Canada
  • October 2012 to September 2017


  • Inclusion
    • ≥ 16 years of age
    • Moderate or severe ARDS as defined by the Berlin conference criteria
    • Duration of ARDS ≤ 36 hours
  • Exclusion
    • Mechanical ventilation > 96 hours
    • Recent treatment for bleeding varices, stricture, hematemesis, esophageal trauma, recent esophageal surgery, or other contraindications for nasogastric tube placement
    • Platelet count < 5,000 cells/µL or INR > 4
    • History of lung or liver transplantation
    • Elevated intracranial pressure
    • Active air leak from the lung (e.g., bronchopleural fistula, pneumothorax, pneumomediastinum)
    • Participation in other interventional trials for ARDS or sepsis within the past 30 days
    • Neuromuscular disease that impairs the ability to ventilate spontaneously including high cervical spine injuries, amyotrophic lateral sclerosis, Guillain-Barre syndrome, and myasthenia gravis
    • Severe chronic liver disease (Child-Pugh score 12)
    • Patients not committed to full support
    • Treating clinician refusal or unwillingness to commit to controlled ventilation for at least 24 hours
    • Inability to obtain informed consent
    • Use of rescue therapies prior to enrollment (e.g., nitric oxide, extracorporeal membrane oxygenation, prone positioning, high frequency oscillation)
  • 727 patients eligible, 366 approached for consent, 202 randomized (102 in the intervention arm and 100 in the control arm). 2 patients in the control arm withdrew consent
  • Primary analysis includes 102 patients in the intervention arm and 98 in the control arm
  • 4 patients lost to follow-up (2 in each arm)
  • Baseline characteristics were well matched
    • Actual body weight: ≈80 kg
    • SOFA score: 11
    • ARDS risk factor: sepsis (86%), pneumonia (75%)
    • Tidal volume (Vt): 6.2 mL/kg PBW
    • Pplt: ≈27 cmH2O
    • Set PEEP: ≈ 13 mmHg
    • Respiratory system compliance: ≈ 31 mL/cmH2O
    • Neuromuscular blockade: 33%
    • Vasopressors or inotropes: 57%


  • PES – guided PEEP
    • PEEP adjusted at least daily to maintain an end-expiratory PL 0 – 6 cmH2O ensuring PEEP was never significantly more or less than pleural pressure estimated by PES
    • A PL – FiO2 table was provided and clinicians were instructed to target the lowest PL – FiO2 combination that maintained a PaO2 55 – 80 mmHg or SpO2 88 – 93%
    • Vt could be decreased to as low as 4 mL/kg PBW if end-inspiratory PL > 20 cmH2O
    • Vt could be increased up to 8 mL/kg PBW for severe acidemia or dyspnea as long as inspiratory PL ≤ 20 cmH2O
    • Once end-expiratory PL was 0 with an FiO2 ≤ 5 for 24 hours, the patient was transitioned to a weaning protocol without further PL measurement


  • Empirical PEEP – FiO2 group
    • PEEP adjusted according to the high PEEP table used in the control group from the OSCILLATE trial
    • PL values not provided to treating team
    • Vt could be decreased to as low as 4 mL/kg PBW for Pplt > 35 cmH2O
    • Vt could be increased up to 8 mL/kg PBW for severe acidemia or dyspnea as long as inspiratory Pplt ≤ 35 cmH2O

Management common to both groups

  • All patients received a single recruitment maneuver (breath hold at 35 cmH2O for 30 seconds) at the start of the trial
  • PL measured at end-inspiration and end-expiration at least daily with waveforms sent to a central repository for quality control
  • Sedation, neuromuscular blockade, and resuscitation were guided by the treating team
  • Additional sedation and neuromuscular blockade could be administered to facilitate PL measurements
  • The protocol was followed until day 28, liberation from mechanical ventilation, protocol failure (refractory hypoxemia/acidemia), withdrawal for safety reasons, withdrawn consent, discharge, or death


  • Primary outcome: ranked composite outcome of death and days free from mechanical ventilation at day 28. In this score, death was weighted more heavily than days free from mechanical ventilation. This is reported as a probabilistic index – the estimated probability that a patient from one trial group will have a higher score (more favorable outcome) than an individual randomly selected from the other group
    • Intervention 49.6% [95% CI, 41.7% – 57.5%] vs control 50.4% [95% CI, 42.5% – 58.3%]
    • P value = 0.92
  •  Impact of the intervention
    • Protocol adherence (defined as observed PEEP within 2 cmH2O of protocol-specified PEEP) through the first 7 days
      • 3% (intervention) vs 94.6% (control)
    • PEEP, end-inspiratory PL, Pplt, transpulmonary driving pressure, and PaO2/FiO2 not significantly different between the two groups over the first 7 days
    • Proportion of patients in whom PEEP changed by at least 5 cmH2O on initiating the protocol settings
      • 6% (intervention) vs 35.7% (control), P = 0.88
    • Mean (SD) PEEP and FiO2 the first day on protocol
      • Intervention: 17 (6) cmH2O and 0.56 (0.15)
      • Control: 16 (4) cmH2O and 0.51 (0.17)
      • P = 0.28 for PEEP and 0.048 for FiO2
    • Number of patients with at least 1 day with a PEEP > 24 cmH2O
      • 12 (intervention) vs 0 (control)
  • Secondary outcome: intervention vs control
    • 28-day mortality
      • 4% vs 30.6% (absolute difference, 1.7 [95% CI, -11.1 – 14.6], P = 0.88)
    • 60-day mortality
      • 6% vs 37.8% (absolute difference, -0.1 [95% CI, -13.6 – 13.3], P > 0.99)
    • 1-year mortality
      • 0% vs 45.8% (absolute difference, -1.8 [95% CI, -15.8 – 12.1], P = 0.89)
    • Ventilator-free days through day 28
      • 5 vs 17.5 days (absolute difference, 0 [95% CI, 0 – 0], P = 0.93)
    • ICU length of stay through day 28
      • 10 vs 9.5 days (absolute difference, 1 [95% CI, -1 – 3], P = 0.25)
    • Hospital length of stay through day 28
      • 16 vs 15 days (absolute difference, 0 [95% CI, -1 – 3], P = 0.58)
    • Hospital length of stay through day 60
      • 16 vs 15 days (absolute difference, 1 [95% CI, -2 – 4], P = 0.47)
    • Protocol failure requiring rescue therapy
      • 9% vs 12.2% (absolute difference, -8.3 [95% CI, -15.8 – 0.8], P = 0.04)
    • Functional status among survivors at 1 year
      • No significant difference in the number of respondents with a score ≥ 95 on the Barthel index or scores between groups on the Short Form-12
  • Safety endpoints
    • No significant difference in shock-free days, need for renal replacement therapy in the first 28 days, pneumothorax, bronchopleural fistula, or barotrauma
  • Post-hoc analysis
    • In 192 patients in whom data was available, initial on-protocol PEEP would have differed by > 2 cmH2O in 103 cases (53.6%) if the patient was assigned to the other treatment group

Authors’ Conclusions

  • In patients with moderate to severe ARDS, PEEP guided by PES did not result in a difference in death and days free from mechanical ventilation compared with empirical high PEEP – FiO2


  • Addresses a clinically relevant question
  • The largest trial to date on the subject
  • Baseline characteristics were well balanced between the two study arms
  • High adherence to trial protocol
  • High concordance between interpretation of PES measurements at clinical sites and the core lab (only 1.8% of readings differed by more than 3 cmH2O)
  • Few patients lost to follow-up
  • The primary endpoint, while a composite outcome, placed more weight on death than ventilator free days avoiding the issue with many composite outcomes where two outcomes of significantly different clinical importance are weighted equally


  • The empiric PEEP – FiO2 table used was aggressive (at FiO2 0.6, the minimum allowable PEEP was 20 cmH2O). This does not reflect usual care at most institutions. While the investigators did not design the control arm of this trial to reflect usual care, the very high PEEP – FiO2 table used does have implications for the generalizability of the study’s findings to usual practice
  • PES-guided PEEP titration is, in theory, most informative in patients with significantly elevated pleural pressure. This trial enrolled a heterogeneous cohort of patients with moderate to severe ARDS, many of whom likely did not have significant issues with chest wall elastance. It is possible that if the trial targeted patients with abnormal pleural pressure (e.g. significant obesity, ascites, chest wall deformities, etc.), the results may have been different
  • The low utilization of prone positioning (only 4 patients total) is at odds with the strong recommendation for proning in patients with severe ARDS https://www.atsjournals.org/doi/full/10.1164/rccm.201703-0548ST. The PROSEVA trial was published while this trial was recruiting
  • The trial was underpowered for many of its secondary endpoints. As an example, the 95% confidence interval for 28-day mortality includes both significant benefit and significant harm (-11.1% – 14.6%) with PES-guided therapy
  • PES measurements are only obtained at a single point in the mid-esophagus and therefore may be insensitive to regional heterogeneity in pleural pressure
  • Additional sedation and neuromuscular blockade could be administered to facilitate PL It is not clear how frequently this was needed. Trial protocols which require deeper levels of sedation and neuromuscular blockade are likely to be increasingly seen as not reflective of best practice https://www.nejm.org/doi/full/10.1056/NEJMoa1901686

The Bottom Line

  • Routine use of PES-guided PEEP titration appears to offer little benefit over the use of empirical high PEEP – FiO2 in patients with moderate to severe ARDS
  • This trial does not inform whether either strategy is superior to a less aggressive PEEP –FiO2 approach or one which prioritizes the early use of prone positioning
  • The physiologic rationale for PES monitoring and the concept of PL are important to understand. Whether they can be used to advance the care of patients with ARDS remains to be seen

External Links


Summary author: James M. Walter
Summary date: 6th June 2019
Peer-review editor: Steve Mathieu

One comment

  • Timo

    I wonder, ifmthe Numbers (percents) are correct in the secondary outcomes.
    Eq. 28-day mortality: 4 vs 30.6% (p = 0.88)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.