Conservative Oxygen Therapy during Mechanical Ventilation in the ICU

The ICU-ROX investigators and the Australian and New Zealand Intensive Care Society Clinical Trials Group (ANZICS-CTG). NEJM;October 14 2019 DOI:10.1056/NEJMoa1903297

Clinical Question

  • In mechanically ventilated ICU patients does a conservative use of oxygen therapy compared to usual oxygen therapy affect the number of ventilator-free-days?


  • Oxygen therapy has traditionally been thought of as a harmless adjunct to patient care and is often given without much thought. Hypoxemia is dangerous but by the same token hyperoxia can cause cellular damage via oxidative stress, absorption atelectasis, acute lung injury, central nervous system toxicity, reduced cardiac output and cerebral and coronary vasoconstriction
  • In the laboratory setting there is some data demonstrating that exposure to oxygen above 21% causes the production of oxygen free radicals. The clinical significance of this is something that can be explored in randomised controlled trials
  • In the pursuit to, above all, do no harm, it is important to question what is the optimal amount of oxygen our intensive care patients should receive
  • AVOID demonstrated no benefit in giving oxygen to normoxic patients with STEMI undergoing PCI. DETO2X showed no difference in one year mortality between room air and oxygen therapy in normoxic patients with suspected AMI. OXYGEN-ICU found a striking mortality benefit in those critically ill patients given a conservative oxygen regimen and several stroke trials have suggested a conservative oxygen strategy is non-inferior (the largest being the SO2S trial) to the liberal strategy
  • The IOTA meta-analysis suggested that in a heterogenous group of ICU patients hyperoxia increased mortality, although there is a very heavy weighting in this meta-analysis on the OXYGEN-ICU trial which had several problems with methodology


  • Randomised controlled trial
  • Multi centre study
  • 1:1 computer generated allocation concealment
  • Single-blinding; clinicians weren’t blinded. Outcome assessors were blinded
  • Based on a power of 90%, a sample size of 1000 patients would provide the ability to detect an absolute between group difference of 2.6 ventilator-free days at day 28 with a 2-sided type 1 error rate of 0.05
  • Intention-to-treat analysis


  • 21 ICUs in Australia and New Zealand
  • Patients enrolled from September 2015 to May 2018


  • Inclusion: Adults. Ventilated and expected to be intubated until at least the day after tomorrow. Randomised <2 hours after initiation of ventilation. Patients or their surrogate provided written informed consent
  • Exclusion:
    • Hyperoxia indicated, eg carbon monoxide poisoning, hyperbaric O2
    • Avoidance of hyperoxia indicated, eg COPD, paraquat poisoning, previous bleomycin therapy, chronic hypercapnic respiratory failure
    • Pregnant
    • Death inevitable
    • Long-term ventilated patients
  • 1000 patients enrolled, but 35 withdrew consent leaving 965 for the analysis
    • 479 patients in conservative group, 481 in usual oxygen group
    • The baseline demographics were well matched between groups
  • Subgroups included: sepsis, acute brain injury, hypoxic ischaemic encephalopathy, surgical or non-surgical, paO2:FiO2 ratio <300mmHg


  • Conservative oxygen group; if SaO2 97% or >97%, the FiO2 was reduced by 10% every 5 minutes until SaO2<97%. If the high SaO2 alarm sounded this was viewed as an emergency and the treating team had to adjust the dose of O2 immediately
    • If SaO2 <90% the FiO2 was increased
    • The lowest FiO2 possible was used to achieve the acceptable SaO2 range of 91-96%


  • Usual oxygen therapy; no upper limit prescribed for SaO2
    • Any SaO2 > 90% is allowed
    • An FiO2 <0.3 is discouraged

Management common to both groups

  • Other ICU management was left to the discretion of the treating clinicians


  • Primary outcome:The number of ventilator-free days from randomisation to day 28. No difference: conservative group vs usual group 21.3 days vs 22.1 days
    • Total number of days of unassisted breathing (not invasively ventilated through ETT or tracheostomy tube and free from NIV). Patients receiving HFNP were not considered to be ventilated. If a patient died they were considered to have no ventilator-free days
  • Secondary outcome:conservative vs usual. No differences
    • Day 90 34.7% vs 32.5% [OR 1.1 (0.84-1.44)]
    • Day 180 mortality 35.7% vs 34.5% [OR 1.05 (0.81-1.37)]
    • Day 180 Cognitive function (TICS score). 30.6 vs 30.4, p = 0.72
    • Day 180 Health-related QoL (EQ-5D-5L). The conservative group had less problems with mobility (p=0.006) and personal care (p=0.0008).
    • Employment status compared to baseline (68.8% vs 61.1%, p =0.24)
    • The subgroup of patients with ischaemic hypoxic encephalopathy were found to benefit from the conservative oxygen strategy. At Day 180 43% vs 59% had died: (RR 0.73 95% CI 0.54-0.99). The conservative oxygen group also had better neurological outcomes.
    • Patients who had acute brain pathologies other than acute ischaemic hypoxic encephalopathy appeared to do better with usual care oxygen (detected in a post-hoc analysis)
    • Patients with sepsis also did better in the usual care oxygen group (Day 180 death 25.9% vs 20.1%)
    • Mean, highest and lowest PEEP used between groups was very similar

Authors’ Conclusions

  • During the first 28 days in the ICU, conservative oxygen therapy, as compared with usual oxygen therapy, did not significantly affect the number of ventilator-free days


  • This is the largest trial conducted in mechanically ventilated ICU patients exploring this question of conservative vs usual oxygen therapy
  • Given oxygen is ubiquitously administered in ICU patients, there is little known of the optimal dose and so this trial asks a very important question
  • The trial has ticked the boxes: allocation concealment, intention-to-treat analysis and complete follow-up
  • There was a separation in the groups as far as therapies received. The conservative group spent a median duration of 29 hours with FiO2 0.21 vs 1 hour in the usual group. The conservative group had less time with SaO2>97%: 27 hours vs 49 hours.
  • The PaO2 level achieved in the conservative and usual care group were similar to the previously published OXYGEN-ICU trial, yet that trial (smaller sample size, only 2/3 of patients were mechanically ventilated, patients were not as sick and the trial stopped early) suggested a mortality benefit for the conservative group. The internal validity of this trial is much better.
  • Ventilator-free days is a clinically important endpoint


  • Clinicians were unblinded. This may bias the attending physician to make decisions regarding management that impacts the outcome, particularly the decision to extubate the patient.
  • The subgroups: hypoxic-ischaemic encephalopathy, sepsis, acute brain pathology showing a difference in mortality benefit can be seen as hypothesis generating and can be the basis for further research
  • The usual oxygen group in this trial had relatively low doses of oxygen and this trial does not preclude harm with more liberal oxygen therapy which is still seen in some ICUs across the world. A greater separation in the groups could have been achieved by being even more conservative in oxygen delivery in the intervention arm or to systematically target hyperoxia in the usual group (although this is perhaps ethically questionable)
  • The results of this trial may not be applicable to centres that routinely use higher FiO2 for usual care in ventilated patients
  • It is possible that this trial was not large enough to detect a clinical difference: the authors are proposing a trial with a sample size of 40000 (#MEGAROXtrial) to definitively answer the question. This trial will have a novel adaptive design to maximise the information received, with power to detect important differences in the subgroups suggested by this trial to have outcomes determined by FiO2
  • Ventilator-free days may not be a sensitive enough end-point to assess the potential harm of oxygen and perhaps the optimal oxygen concentration for ICU patients may differ depending on the patient’s presenting illness

The Bottom Line

  • This trial does not change my practice, but is reassuring and hypothesis generating
  • I will continue to use the minimal FiO2 to keep my patients SaO2> 90%
  • I will avoid hyperoxaemia

External Links


Summary author: Celia Bradford
Summary date: 31st October 2019
Peer-review editor: Steve Mathieu


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.