The REST Trial

Effect of Lower Tidal Volume Ventilation Facilitated by Extracorporeal Carbon Dioxide Removal vs Standard Care Ventilation on 90-Day Mortality in Patients With Acute Hypoxemic Respiratory Failure – The REST trial

McNamee et al. JAMA August 31, 2021; doi:10.1001/jama.2021.13374

Clinical Question

  • In mechanically ventilated patients with acute hypoxaemic respiratory failure, does lowering tidal volume using extracorporeal carbon dioxide removal, compared with conventional low tidal volume ventilation, improve day 90 all-cause mortality?


  • Hypoxic respiratory failure requiring mechanical ventilation is common in the ICU, and is associated with high morbidity, mortality and health care costs
  • Studies of ARDS have shown that strategies that reduce tidal volumes cause a reduction in ventilator induced lung injury and improve patient outcomes
  • Current international guidelines recommend protective lung ventilation strategies which include targeting tidal volumes (Vt) ≤6mls/kg PBW and plateau pressures (Ppl) of <30cmH20
  • However, protective lung ventilation is associated with respiratory acidosis and other interventions which may adversely impact patient care and outcomes
  • Extracorporeal carbon dioxide removal (ECOR) is an extracorporeal device that removes CO2 from the blood via an external membrane and circuit (but unlike ECMO is not able to add oxygen in sufficient amounts)
  • Several studies have shown that it is feasible to use ECOR to remove significant amounts of CO2 to facilitate very low tidal volume mechanical ventilation strategies (SUPERNOVA 2019, XTRAVENT 2013)
  • However to date there have been no large studies that assess ECOR with patient centred outcomes


  • Pragmatic, multi centre, open label, randomized controlled clinical trial
  • 1:1 randomisation, using a computer-generated schedule with variable block sizes stratified by recruitment centre
  • All sites were given an education package to standardise ECOR and trial processes
  • Primary statistical analysis using an intention to treat chi squared analysis, and a secondary analysis using log binomial regression adjusted for age, SOFA score and baseline PaO2/FiO2
  • Power calculation: 1120 patients provided a 90% power to detect an absolute difference of 9% in 90 day mortality, assuming a mortality rate of 41%
    • Rationale was based on a 9% ARR that occurred in the ARDSNet trial which randomised patients tidal volumes to either 12 or 6ml/Kg PBW
  • Original plan was for an interim analysis at 560 patients, however this was moved forwards to 412 patients after the trial was paused to investigate an intracranial haemorrhage in the intervention arm
    • At this time point, the DSMB performed a conditional power analysis and found that ongoing recruitment was unlikely to show benefit
    • There were no stopping rules, but the analysis suggested only a 44% chance of finding a difference between groups, therefore they recommended stopping the trial for feasibility and futility reasons
    • Of note, it was not stopped specifically due to safety concerns


  • 51 adult general ICUs within the NHS in the UK
  • Recruitment from May 2016 to December 2019


  • Inclusion:
    • Adult patients undergoing invasive mechanical ventilation
    • Potentially reversible cause of acute hypoxic respiratory failure
    • <48 hours since onset of hypoxaemia
    • Moderate to severe respiratory failure with PaO2:FiO2 ratio of <150mmHg with ≥5cmH20 PEEP
  • Exclusion:
    • Mechanical ventilation >7 days
    • Clinicians able to reduce tidal volumes ≤3mL/kg
    • Any contraindication to systemic anticoagulation with heparin
    • Untreated pulmonary embolism, pleural effusion or pneumothorax
    • Acute respiratory failure fully explained by LVF or fluid overload
  • Screening
    • 7071 patients screened, 412 (6%) were recruited into the study
      • The two most common reasons for exclusion were:
        • “Other” (26%) which included any reason not listed in the protocol, most commonly this was rapid change in clinical state (improvement or deterioration)
        • CI to anticoagulation (21%)
  • Included population 
  • Generally good balance of baseline variables with minor differences that would not have impacted the primary outcome (ECOR vs standard care)
    • Median age:  60.2 vs 61.8 years old
    • ARDS at baseline:  59% vs 63%
    • APACHE II: 19 vs 20
    • SOFA: 10 vs 10
    • PF ratio: 118 vs 115 mmHg
    • Driving pressure: 15 vs 16 cm H2O
    • Median Vt: 6.3 vs. 6.4 ml/kg PBW
    • PEEP: 10 vs 10 cm H2O
    • PaCO2: 53.8 vs 54.6 mmHg
    • pH: 7.3 vs 7.3
    • Prone positioning: 11% vs 11%
    • Neuromuscular blocking drugs: 51% vs 49%


  • A dual lumen catheter was percutaneously inserted into the patient
  • Commencement of ECOR and initiation with systemic heparin anticoagulation
  • Blood flow was maximised (typically 350-450mL/min) and sweep gas removal was turned up to 10Lmin
  • Mechanical ventilator was adjusted to target tidal volumes of 3ml/kg predicted body weight
  • ECOR was continued for ≥48 hours, then weaned according to the trial protocol
    • Maximum duration of ECOR use was 7 days
  • Dose of intervention
    • 185/202 actually received ECOR, for a mean duration of 4 (+/- 2) days
    • Reasons for cessation of ECOR:
      • 7 days post randomisation in 18%, safety concerns in 8%, and  requirement for ECMO in 7%


  • Guideline based mechanical ventilation, including targeting 6ml/Kg predicted body weight and PEEP based on ARDSNet tables

Management common to both groups

  • Patients in both groups could receive neuromuscular-blocking drugs, prone positioning, or referral for consideration of extracorporeal membrane oxygenation (ECMO), as per UK guidelines


  • Primary outcome
  • No difference in primary outcome of day 90 all cause mortality
    • 41.5% in the lower tidal volume ventilation with extracorporeal carbon dioxide removal group vs 39.5% in the standard care group
      • Risk Ratio, 1.05 [95% CI,0.83-1.33]; difference, 2.0% [95% CI, −7.6% to 11.5%]; p = .68)
    • This was unchanged after adjusting for age, SOFA score and baseline PaO2/Fio2, and unchanged after looking at per protocol analysis, and after excluding the first 2 patients initiated on ECOR at each site.
  • Secondary outcomes:
  • Comparing ECOR vs. standard care group:
    • Lower tidal volumes on day 2-3
      • Day 2: 4.5 mls/kg PBW vs 6.5mls/kg PBW
        • mean difference -2.0 [95% CI -2.3 – -1.7]
      • Day 3: 4.4mls/kg PBW vs 6.7 mls/kg PBW
        • mean difference -2.3 mls/kg PBW [95% CI -2.7 – -2.0]
    • Less ventilator free days (to day 28):
      • 7.1 (8.8) vs 9.2 (9.3) p=0.02
    • No difference in 28d mortality, ICU or hospital length of stay
  •  Other physiological parameters and use of adjunctive therapies
  • For clarity only day with largest difference has been given (all data available in eTable 4 and 5)
    • Lower tidal volumes day 2 – 7, although no difference days 0 and 1
    • Lower driving pressure day 2-5, although no difference days 0 and 1
      • Day 2: 12.1 cm H20 vs 15.3 cm H20
        • mean difference -3.2 [95% CI -4.4.- -2.1]
    • Less prone positioning days 1 and 2
      • Day 1: 5.5% vs 13.9%, mean difference -8.4 [-14.0 – -2.8]
    • Lower PaO2:FiO2 values days 2 and 3
      • Day 3: 147.9 vs 167, mean difference -19.1 [-30.1 – -8.2]
    • Higher respiratory rates days 2-4
      • Day 3: 27.8 vs 24.4, mean difference 3.41 [2.19 – 4.63]
    • Higher use of mandatory modes of MV days 2-7
      • Day 4: 70.0% vs 41.4%, point difference 28.5% [19.0 – 38.1]
    • Higher use of neuromuscular blockade days 2-7
      • Day 4: 42.1% vs. 21.5%, point difference 20.6% [11.4 – 29.7]
  • Adverse Events: 
    • Higher rates of adverse events: 168 (52% of patients) vs 61 (23% of patients)
      • 65 of these felt to be related to study intervention
      • Higher rates of ICH: 10 vs 2
        • 5 were thought related to the intervention and 3 which resulted in death
    • Higher rates of infectious complications (7 vs 1)

Authors’ Conclusions

  • In patients requiring mechanical ventilation for acute hypoxemic respiratory failure, lower tidal volume ventilation facilitated by extracorporeal carbon dioxide removal, compared with standard care, did not result in a reduction in mortality at 90 days. However, due to early termination, the study may have been underpowered to detect a clinically important difference


  • Large, multicentre, RCT led by an experienced clinical trials group
  • Important clinical question with potentially large financial implications for critical care around the world
  • Reasonable attempts to reduce bias (central randomisation, allocation concealment, unbiased primary outcome, pre-published protocol and statistical analysis plan)
  • Good baseline balance suggesting effective randomisation
  • Good adherence to the ECOR intervention (92% received it)
  • No industry funding
  • Minimal missing data (1.7% missing data from primary outcome)


  • Although the inclusion criteria were broad, only 6% of all those screened were entered into the study – suggesting a significant selection bias and lack of generalisability to other populations
  • Although there was good group separation of tidal volume across days 2-7, the dose of the intervention fell short of the 3ml/kg target
    • In addition, many other clinically important  variables were not separated e.g. plateau pressure, respiratory rate and CO2 level
  • Both arms had low rates of proning (11%), not representing contemporary standard care, although the mortality rate in the control arm accurately reflected current UK mortality (41%)
  • Patients, clinicians and researchers were not blinded resulting in moderate risk of performance bias
    • Use of an unbiased primary endpoint of mortality reduces this risk
  • The trial was stopped early due to futility and feasibility issues, contributed to by:
    • a lower than expected recruitment rate
    • the requirement of a very large (?realistic) reduction in mortality as set out in the power calculation (ARR 9%)
  • Therefore, the trial as it stands was underpowered and could have missed a true difference (the point estimate favoured the control arm)
  • Many sites had little experience of ECOR prior to the trial, despite there being a known volume outcome relationship with the use of ECLS
    • However the sensitivity analysis which excluded the first 2 patients from sites did not demonstrate a significant learning curve
    • Only 4/51 sites had > 10 patients with ECOR in the trial
  • The study investigated a complex bundle of interventions, including the effects of ECOR, anticoagulation AND a target of 3ml/kg. Targeting such a low tidal volume resulted in a lower pH and higher CO2 levels in the intervention arm, and additional interventions such as more mandatory mechanical ventilation, more NMB and a higher respiratory rate. Separating out and interpreting each component of the trial is therefore challenging

The Bottom Line

  • The data presented in this study suggest that while achieving lower tidal volumes using ECOR was possible, any potential benefit on the lung was offset by the increased requirement of mandatory mechanical ventilation and NMB, and adverse events including serious complications such as intracranial haemorrhage
  • Given the safety concerns outlined above, the strategy outlined in this study should not be used in patients outside further clinical trials
  • Future studies that harness the potential benefits of ECOR without increasing the risk of other complications are needed to progress this technology

External Links


Summary author: Aidan Burrell @aidanburrell
Summary date: 13 September 2021
Peer-review editor: George Walker

Picture by: iStock


One comment

  • Another great review, well done.

    Interesting that despite the use of ECCOR, they couldn’t achieve their target Vt of 3ml/kg.

    Seems unlikely to be broadly implemented without better identification of subgroups that might benefit

Leave a Reply

Your email address will not be published. Required fields are marked *

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