Conservative Oxygen for Unresponsive Patients after Cardiac Arrest

The LOGICAL Investigators and the Australian and New Zealand Intensive Care Society Clinical Trials Group. NEJM. 2026. doi: 10.1056/NEJMoa2513814

Clinical Question

• In patients who are unresponsive after cardiac arrest, does limiting oxygen exposure to achieve acceptable oxygenation increase the likelihood of surviving with a favourable functional outcome

Background

• Brain damage and unfavourable functional outcomes is a common outcome after cardiac arrest
• Injury may result from both the hypoxia during the arrest period and the re-perfusion injury post return of spontaneous circulation (ROSC)
• Animal models suggest that hyperoxia contributes to neuronal death from overwhelming antioxidant defences
• Prior evidence was either pre-hospital (demonstrating potential harm from hypoxia), or has been conflicting in nature

Design

• Investigator-initiated, parallel-group, open-label, assessor-blinded randomised control trial
• Nested trial within the Mega-ROX registry trial
• Independent data and safety monitoring committee and reviewed the primary outcome interim analysis at the first 500 patients
• Randomised via online, centralised system
• Adaptive randomisation to increase the probability of assignment to the regimen associated with the lower in-hospital risk of death (assignment from 1.05:1 to 1:1.05)
• Sample size of 1400 patients required
  • 90% power to detect an 8.5% absolute difference between treatment groups assuming a control event rate of 32%, and allowing for 4% loss to follow up
  • Based on data from ICU Rox trial and a post hoc analysis of patients with suspected hypoxaemic ischaemic encephalopathy within this trial
  • Sample size increased to 1840 to account for loss to follow up, protocol deviations and a higher than expected proportion of patients with PEA/asystole
• Primary outcome was favourable functional outcome measured by the extended Glasgow Outcome Scale (GOS-E) at 180 was post randomisation
  • GOS-E 1-4 (death to upper severe disability) considered unfavourable
  • GOS-E 5-8 (lower moderate disability) considered favourable

Setting

• Enrolment of patients in ICU’s across Australia (39), New Zealand (13) and Ireland (1)
• Enrolment commenced September 2021 and concluded June 2024

Population

• Inclusion:
  • Adults (>18 years of age) receiving mechanical ventilation in the ICU following cardiac arrest
  • Ischemic encephalopathy was suspected
    • Unable to follow verbal commands after ROSC
    • Clinical concern for possible brain damage
  • Enrolled within 12 hours of fulfilment of enrolment criteria
• Exclusion:
  • Previously enrolled
  • Enrolment was not considered to be in their best interests
  • Death was deemed to be imminent or inevitable

• A total of 1840 patients were enrolled in the trial, 19 patients withdrew consent, leaving 1821 patients
  • 873 assigned to the conservative-oxygen group
  • 948 assigned to the liberal-oxygen group with intention to treat analysis
• Protocol deviations affecting trial eligibility occurred in 76 patients (4.2%)
• Primary outcome data were available for 1709 patients (819 in the conservative-oxygen group [93.8%] and 890 in the liberal-oxygen group [93.9%])
• Baseline characteristics were similar between the conservative and liberal groups
  • Age: 60.1 vs 60.7 years old
  • Sex (Male): 72.1% vs 70.8%
  • Similar coexisting medical conditions
  • Location of cardiac arrest
    • Out of hospital: 74.1% vs 73.6%
    • In hospital: 25.3% vs 25.6%
  • Bystander CPR: 78.2% vs 77.8%
  • First rhythm:
    • Shockable (VF/VT/automatic external defibrillator shockable) 51.0% vs 52.8%
    • Asystole: 16.3% vs 14.1%
  • Time to defibrillation: 11.2  vs 11.7  mins
  • Time to ROSC: 26  vs 25 mins
  • Cause of cardiac arrest
    • Medical: 88.4% vs 86.8%
    • Other: 11.6% vs 13.2%
  • Median FiO2: 0.52  vs 0.51
  • Median time from ROSC to randomisation: 7.3 hours vs 7.0 hours
  • Median time from ICU admission to randomisation: 3.0 hours vs 3.1 hours

Intervention

• Conservative oxygen group
  • Low oxygen saturation alarm limit of 90%
  • High oxygen saturation alarm limit of 95%
  • If oxygen above lower limit FiO2 decreased to 0.21 or supplemental oxygen discontinued if extubated

Control

• Liberal oxygen group
  • Minimum FiO2 requirements of 0.3
  • No upper limit oxygen saturation alarm

Management common to both groups

• Both groups had lower oxygen saturation alarm of 90%
• If an arterial blood gas demonstrated a PaO2 of <60mmHg, then the FiO2 could be increased regardless of the SpO2 at the clinical discretion
• Special circumstances where higher FiO2 was permitted despite the oxygen saturations (e.g. suctioning, bronchoscopy, transport, tracheostomy insertion, preparation for extubation)
• Randomisation oxygen targets were continued for 90 days or discharge from the ICU, whichever was sooner

Intervention Fidelity:

• Comparing conservative vs liberal groups
• Median % hours per patient with SpO2 ≥97%: 21.2 vs 53.0
• ≥1  PaO2 > 100 mmHg: 57.7% vs 78.4% (RR 0.74, 95% CI 0.68 – 0.80)
• Median % hours per patient with SpO2 < 88%: 0 vs 0
• ≥1 PaO2 < 60 mmHg: 43.4 vs 27.5% (RR 1.57, 95% CI 1.39 – 1.78)
• Median % hours with FiO2 = 0.21: 38.8 vs 0%
• 26.3% patients in conservative group had an FiO2 > 0.21 and not reduced when SpO2 95% or more
• 8.4% of patients in liberal group used an FiO2 < 0.3

Outcome

• Primary outcome:
• Favourable functional outcome byGOS-E at 180:
• 313/819 (38.2%) in conservative vs 353/890 (39.7%) in liberal group
  • Adjusted absolute risk difference, −0.9 % (95% CI: −5.5 to 3.7); relative risk, 0.97 (95% CI: 0.87 to 1.09)
• No difference in any pre-specified subgroup (cause of arrest, first rhythm, location, time to from ROSC to randomisation, time from ICU admission to randomisation)

• Secondary outcomes:
• All comparing conservative vs liberal:
  • Alive at 180 days, n (%): 48.0% vs 49.7%
  • Quality of life (EuroQol Group 5-Dimension 5-Level questionnaire) at 180 days in survivors: Conservative: 0.92  vs 0.92
  • Cognition (Montreal Cognitive Assessment) at 180 days in survivors: Conservative: 18 vs 19
  • No difference in duration of mechanical ventilation, ICU and hospital LOS
  • Discharge directly home, n (%): 32.8% vs 34.0%

Authors’ Conclusions

• Conservative oxygen therapy did not result in a higher likelihood of a favourable neurological outcome compared to liberal oxygen therapy

Strengths

• Large randomised trial in an often difficult to study cohort of patients
• Achieved degree separation of oxygen with conservative and liberal, acknowledging usual challenges with trials of oxygen targets and that the provided FiO2 can not be lower than 0.21
• Increased generalisability given broad range of cardiac arrest locations, cardiac arrest types and multiple sites of enrolments with pragmatic design and limited exclusion criteria
• Blinded outcome assessor
• Clinically and patient centred outcomes; GOS-E, quality of life, cognition and discharge directly home
• Pre-published SAP

Weaknesses

• Whilst separation in groups was achieved, this was modest
• More patients assigned to the liberal oxygen therapy group despite adaptive randomisation not favouring this at any point
• Moderate number of protocol violations, however result robust on exclusion of patients with deviations in a sensitivity analysis
• Enrolment occurred in Australia, New Zealand and Ireland – all well resourced countries with similar health care systems, may limit generalisability to other health care settings
• No evaluation of adherence to neuroprognostication criteria; although use of neuroprognostication tests similar between groups
• Some of the therapeutic window could have been lost given the median time to enrolment from ROSC was 7.3 vs 7 hours although no subgroup effect noted when analysing by time to randomisation

The Bottom Line

• This large, investigator-initiated, parallel-group, open label, assessor-blinded randomised control trial provides strong evidence that targeting conservative oxygenation (SpO2 90-95%, minimal FiO2, PaO2 >60mmHg) is not associated with an improved 180 day neurological outcome
• Future aims might explore the time to enrolment from ROSC and if this represents a vulnerable period, acknowledging the difficulties in the field for achieving these targets

External Links

• article Conservative Oxygen for Unresponsive Patients after Cardiac Arrest
• further reading A logical interpretation of the data on optimal oxygen therapy targets for unresponsive adults in ICU after cardiac arrest
• further reading EXACT Trial
• further reading BOX Trial

Metadata

Summary author: Jake Reeve
Summary date: 6th July 2026
Peer-review editor: George Walker

Picture by: Stephen Andrews/Pexels