Publication

• Title: Conservative Oxygen for Unresponsive Patients after Cardiac Arrest
• Acronym: LOGICAL
• Year: 2026
• Journal published in: New England Journal of Medicine
• Citation: The LOGICAL Investigators and the Australian and New Zealand Intensive Care Society Clinical Trials Group. Conservative oxygen for unresponsive patients after cardiac arrest. N Engl J Med. Published online June 10, 2026.

Context & Rationale

• Background

  Hypoxic–ischaemic brain injury after cardiac arrest reflects both the primary ischaemic insult during circulatory arrest and secondary injury after return of spontaneous circulation, including reperfusion injury, disordered cerebral autoregulation, inflammation, mitochondrial dysfunction, fever, seizures, and exposure to potentially injurious extremes of oxygen and carbon dioxide.¹
  
  The biological rationale for conservative oxygen therapy was strong. Animal data suggested that high inspired oxygen after resuscitation worsened neuronal injury, cerebral metabolic dysfunction, and neurological deficit scores compared with lower oxygen exposure.²
  
  Clinical evidence before LOGICAL was uncertain. In the ICU-ROX subgroup of patients with suspected hypoxic–ischaemic encephalopathy, conservative oxygen therapy was associated with lower day-180 mortality, 43% vs 59%, RR 0.73; 95% CI 0.54 to 0.99; P=0.04, and a numerically higher rate of favourable neurological outcome, 45% vs 32%, RR 1.23; 95% CI 0.95 to 1.59; P=0.13. However, this was a subgroup analysis within a broader mechanically ventilated ICU trial, and the adjusted estimate for survival with favourable neurological outcome remained imprecise, adjusted OR 1.85; 95% CI 0.79 to 4.34; P=0.15.³
  
  International resuscitation guidance before LOGICAL continued to identify post-arrest oxygen targets as a priority area of uncertainty, because existing randomised evidence was inconsistent, underpowered for functional neurological outcomes, or applied to different phases of care.⁴
• Research Question/Hypothesis

  LOGICAL tested whether conservative oxygen therapy in the ICU, compared with liberal oxygen therapy, would increase survival with a favourable functional outcome at 180 days in unresponsive adults receiving invasive mechanical ventilation after cardiac arrest.
  
  The hypothesis was that limiting oxygen exposure to the minimum required to maintain acceptable oxygenation would reduce reperfusion-related brain injury and improve patient-centred neurological recovery.
• Why This Matters

  Oxygen therapy is universal after cardiac arrest, inexpensive, immediately modifiable, and biologically plausible as a neuroprotective intervention.
  
  The clinical balance is delicate: hyperoxaemia may worsen oxidative brain injury, whereas excessive oxygen restriction may increase hypoxaemic episodes in a brain already vulnerable to secondary injury.
  
  LOGICAL was designed to answer the practical ICU question: should clinicians actively avoid SpO2 values of 95% or higher and reduce FiO2 toward room air in unresponsive ventilated post-arrest patients, or is a more liberal approach acceptable?

Design & Methods

• Research Question: In adults who were unresponsive after resuscitation from cardiac arrest and receiving invasive mechanical ventilation in the ICU, does conservative oxygen therapy, compared with liberal oxygen therapy, improve survival with a favourable functional outcome at 180 days?
• Study Type: Investigator-initiated, international, multicentre, parallel-group, open-label, assessor-blinded, adaptive randomised clinical trial conducted in 53 ICUs in Australia, New Zealand, and Ireland. The trial was nested within the Mega-ROX trial programme and was endorsed by the Australian and New Zealand Intensive Care Society Clinical Trials Group.⁵
• Population:
  • Adults aged ≥18 years.
  • Receiving invasive mechanical ventilation in the ICU after cardiac arrest.
  • Suspected hypoxic–ischaemic encephalopathy, operationalised as inability to follow verbal commands after return of spontaneous circulation with clinical concern about possible brain injury.
  • Enrolment required within 12 hours after fulfilling eligibility criteria.
  • Major exclusions were previous enrolment, clinician judgement that enrolment was not in the patient’s best interests, imminent and inevitable death, and enrolment more than 12 hours after meeting inclusion criteria.
• Intervention:
  • Conservative oxygen therapy.
  • The default lower SpO2 alarm limit was 90%, although clinicians could reduce this if judged appropriate.
  • Whenever supplemental oxygen was being administered in the ICU, the upper SpO2 alarm was set at 95%.
  • FiO2 was decreased to 0.21 as rapidly as possible if SpO2 remained above the lower acceptable limit.
  • After extubation, supplemental oxygen was discontinued as soon as possible if SpO2 remained above the lower acceptable limit.
  • If an arterial blood gas showed PaO2 <60 mm Hg or SaO2 below the acceptable minimum, FiO2 could be increased irrespective of the pulse oximetry reading.
  • High FiO2 was permitted during transport outside ICU and during procedures such as bronchoscopy, suctioning, tracheostomy, or preparation for extubation.
  • The assigned strategy continued until ICU discharge or 90 days after randomisation, whichever occurred first.
• Comparison:
  • Liberal oxygen therapy.
  • The default lower SpO2 alarm limit was also 90%, with clinician discretion to reduce it.
  • There was no protocol-defined upper SpO2 limit.
  • Use of upper SpO2 alarm limits was prohibited.
  • The minimum FiO2 permitted during invasive mechanical ventilation was 0.30.
  • All other ICU care, including ventilator settings, PEEP, temperature management, neuroprognostication, and circulatory support, was determined by treating clinicians, although reducing PEEP solely to meet oxygen targets was discouraged.
• Blinding: Treating clinicians and bedside staff were not blinded because oxygen titration and alarm limits were visible. Outcome assessors for the Extended Glasgow Outcome Scale, EQ-5D-5L, and MoCA-blind were trained and unaware of trial-group assignment. This makes the primary outcome more robust than process outcomes, but open-label ICU care remains an important limitation.
• Statistics: The original sample-size calculation specified 1400 patients to provide 90% power at a two-sided alpha level of 0.05 to detect an 8.5 percentage-point absolute increase in survival with favourable functional outcome at day 180, assuming a 32% control event rate and allowing 4% loss to follow-up. The sample size was increased during the trial to 1840 patients to account for loss to follow-up, eligibility deviations, and a higher-than-expected proportion of patients with asystole or pulseless electrical activity. Analyses used an intention-to-treat population excluding patients without consent for use of all data. The primary analysis used log-binomial models with random intercepts for site and adjustment for baseline sepsis and non–hypoxic–ischaemic encephalopathy brain injury, with multiple imputation for missing primary outcome data.
• Follow-Up Period: The primary outcome was assessed 180 days after randomisation. Median timing of 180-day follow-up was 189 days. ICU and hospital outcomes were assessed during the index admission, and oxygen exposure metrics were recorded from randomisation until day 7, ICU discharge, or death, whichever occurred first.

Key Results

This trial was not stopped early. A single interim analysis using primary-outcome data from the first 500 patients was reviewed by the data and safety monitoring board, which recommended continuation. No stopping rule for futility was applied.

• LOGICAL achieved clear separation in oxygen exposure: conservative oxygen reduced SpO2 ≥97%, PaO2 >100 mm Hg, and time exposed to FiO2 above room air, but did not improve survival with favourable neurological outcome.
• The conservative strategy increased the proportion of patients with at least one PaO2 <60 mm Hg, 43.4% vs 27.5%, RR 1.57; 95% CI 1.39 to 1.78, without an observed improvement in patient-centred outcomes.
• Subgroup analyses did not identify a credible responder group. For example, the RR for favourable outcome was 0.92 (95% CI 0.71 to 1.21) in patients with non-shockable rhythm, 1.01 (95% CI 0.90 to 1.14) in those with shockable rhythm, 0.89 (95% CI 0.74 to 1.07) when randomised within 2 hours of ICU admission, and 1.04 (95% CI 0.89 to 1.20) when randomised later. Interaction p values were not reported.

Internal Validity

• Randomisation and Allocation: Randomisation used a secure centralised internet-based system with computer-generated sequences. The trial used adaptive randomisation within the Mega-ROX programme, allowing ratios of 1.05:1 in favour of one oxygen strategy or 1:1 depending on accumulating Mega-ROX data. During LOGICAL, the first 557 patients with ischaemic encephalopathy were randomised 1.05:1 in favour of conservative oxygen; subsequent patients were randomised 1:1. Allocation was concealed from clinical staff, investigators, and the management committee.
• Drop out or exclusions: A total of 1840 patients underwent randomisation: 882 to conservative oxygen and 958 to liberal oxygen. Consent for use of all data was not provided or was withdrawn for 19 patients, leaving 1821 in the intention-to-treat population: 873 vs 948. Primary outcome data were available for 1709 patients: 819/873 (93.8%) vs 890/948 (93.9%). Primary outcome data were missing for 112 patients: 56 declined follow-up and 56 could not be contacted. Multiple imputation was used for missing primary outcome data.
• Post-randomisation eligibility deviations: Eligibility protocol deviations occurred in 76/1821 patients (4.2%). In the conservative group, 19 did not meet inclusion criteria, 6 met exclusion criteria, and 1 was enrolled outside the window. In the liberal group, 33 did not meet inclusion criteria, 2 met exclusion criteria, and 15 were enrolled outside the window. Sensitivity analyses restricted to eligible patients were consistent with the primary result: 303/798 (38.0%) vs 336/848 (39.6%); RR 0.98; 95% CI 0.87 to 1.10.
• Performance/Detection Bias: Bedside clinicians were not blinded, creating risk of performance bias in ventilator management, blood gas sampling, neuroprognostication, and decisions to withdraw life-sustaining treatment. This is particularly relevant because neurological death accounted for most deaths. However, the primary functional outcome was assessed by trained assessors who were unaware of trial assignment, reducing detection bias for the main endpoint.
• Protocol Adherence: Assigned treatment was delivered to 876/882 patients in the conservative group and 957/958 in the liberal group. Protocol deviations in oxygen delivery were not trivial: in the conservative group, FiO2 was not reduced when SpO2 was ≥95% and FiO2 was >0.21 in 230/873 patients (26.3%); in the liberal group, FiO2 <0.30 during mechanical ventilation occurred in 80/948 patients (8.4%). These deviations likely reduced treatment contrast but did not abolish physiological separation.
• Baseline Characteristics: Baseline characteristics were broadly balanced. Mean age was 60.1 vs 60.7 years; male sex 72.1% vs 70.8%; out-of-hospital arrest 74.1% vs 73.6%; shockable first monitored rhythm 51.0% vs 52.8%; medical cause 88.4% vs 86.8%; median PaO2 98 vs 98 mm Hg; and mean FiO2 0.52 vs 0.51. Hypertension and diabetes were slightly more common in the conservative group, but not enough to explain the primary result.
• Illness Severity: The trial enrolled a high-risk population with approximately 50% mortality at 180 days. This was an appropriate population for detecting a clinically important neuroprotective effect if one existed.
• Heterogeneity: Clinical heterogeneity was expected: arrests were out-of-hospital and in-hospital, shockable and non-shockable, medical and non-medical. Prespecified and post hoc subgroups did not show a consistent interaction signal. Heterogeneity therefore improves pragmatic relevance but may dilute effects in narrower biological phenotypes.
• Timing: Median time from return of spontaneous circulation to randomisation was 7.3 hours in the conservative group and 7.0 hours in the liberal group; median time from ICU admission to randomisation was 3.0 vs 3.1 hours. This timing is appropriate for an ICU oxygen-target trial but may be too late to test the earliest reperfusion phase. Post hoc analyses among patients randomised within 4 hours of return of spontaneous circulation did not suggest benefit: 47/112 vs 49/123; RR 1.08; 95% CI 0.80 to 1.45.
• Dose: The conservative intervention was assertive: avoid SpO2 ≥95%, reduce FiO2 toward 0.21, and stop supplemental oxygen after extubation when safe. The liberal comparator was also deliberately separated by prohibiting upper SpO2 alarms and requiring FiO2 ≥0.30 during mechanical ventilation. The tested dose of oxygen separation was therefore clinically meaningful, although not extreme.
• Separation of the Variable of Interest: Treatment separation was robust. Median proportion of ICU hours with SpO2 ≥97% was 21.2% vs 53.0%, difference −31.9 percentage points; median hours with SpO2 ≥97% were 16 vs 37, difference −21 hours; at least one PaO2 >100 mm Hg occurred in 57.7% vs 78.4%, RR 0.74; and median proportion of hours with FiO2 0.21 was 38.8% vs 0.0%, difference 38.9 percentage points.
• Key Delivery Aspects: Conservative oxygen achieved the intended reduction in high oxygen exposure but also increased arterial hypoxaemia exposure, with at least one PaO2 <60 mm Hg in 43.4% vs 27.5%. This is central to interpretation: the intervention reduced hyperoxaemia but did not provide neurological benefit and may have narrowed the safety margin.
• Crossover: Formal crossover was not the main issue, but contamination occurred through protocol deviations. Conservative patients sometimes remained at higher FiO2 despite SpO2 ≥95%, and liberal patients sometimes received FiO2 <0.30 while ventilated. These deviations would bias toward smaller observed differences, but the substantial physiological separation argues against inadequate contrast as the sole explanation for the absence of benefit.
• Adjunctive therapy use: Important cointerventions were similar. Targeted temperature management was used in 430/873 (49.3%) vs 476/948 (50.2%). Neuroprognostic testing was broadly comparable, including EEG in 101/873 (11.6%) vs 123/948 (13.0%), CT brain in 177/873 (20.3%) vs 174/948 (18.4%), and MRI brain in 76/873 (8.7%) vs 74/948 (7.8%).
• Outcome Assessment: The primary outcome, GOS-E at 180 days, is clinically meaningful and directly relevant to patients and families. It was assessed by blinded trained assessors using a standardised questionnaire. Secondary quality-of-life and cognition outcomes were also assessed by blinded personnel. However, death and withdrawal of life-sustaining treatment remain vulnerable to open-label care pathways upstream of the blinded 180-day assessment.
• Statistical Rigor: The protocol and SAP were published before enrolment completion.⁵ The primary analysis matched the design, accounted for site and prespecified covariates, used intention-to-treat principles, and handled missing primary outcome data with multiple imputation. Secondary and subgroup analyses were not adjusted for multiplicity and should be interpreted as exploratory.

Conclusion on Internal Validity: Internal validity is strong for the primary question because allocation was concealed, the primary outcome was blinded and patient-centred, follow-up was high, the analysis was prespecified, and oxygen exposure separated clearly. The main limitations are open-label bedside care, missing primary outcome data in 6.2%, eligibility deviations in 4.2%, and protocol deviations that partially diluted separation.

External Validity

• Population Representativeness: LOGICAL enrolled a clinically recognisable ICU population: unresponsive adults receiving invasive mechanical ventilation after cardiac arrest with suspected hypoxic–ischaemic encephalopathy. Most arrests were out-of-hospital, approximately half had an initial shockable rhythm, and approximately 87–88% were of medical cause.
• Representativeness against registry data: Among Australian and New Zealand ICU patients mechanically ventilated during the first 24 hours after cardiac arrest, excluding palliative and organ-donation admissions, 1837/7376 (24.9%) were included in LOGICAL. Compared with the broader Australian and New Zealand ICU post-arrest population, trial participants were similar in age and APACHE II score: age 60.9±15.1 vs 60.5±16.8 years and APACHE II 28.4±8.6 vs 28.5±9.4.
• Geography: The trial was international but heavily Australasian, with sites in Australia and New Zealand contributing nearly all participants and one Irish site contributing a small number. Generalisability is therefore strongest for high-income ICU systems with continuous oximetry, arterial blood gas access, and standardised post-arrest ICU care.
• Applicability: The findings apply best to invasively ventilated adults who remain unresponsive after return of spontaneous circulation and are admitted to an ICU within a timeframe that permits oxygen-target implementation.
• Limited applicability: The results should not be extrapolated directly to prehospital oxygen titration, emergency department-only strategies, awake or non-intubated survivors, paediatric arrest, extracorporeal cardiopulmonary resuscitation populations, profound acute respiratory failure with high oxygen requirements, or settings where pulse oximetry and arterial blood gases are unreliable or unavailable.
• Resource-limited settings: The intervention is inexpensive but monitoring-intensive. Conservative oxygen titration requires reliable continuous pulse oximetry, alarms that are visible and acted upon, arterial blood gas access, and staffing to titrate FiO2 frequently. Its lack of benefit in LOGICAL makes routine implementation difficult to justify where monitoring resources are constrained.

Conclusion on External Validity: External validity is strong for high-resource ICUs managing unresponsive mechanically ventilated adults after cardiac arrest. It is more limited for prehospital care, lower-resource ICUs, patients with major respiratory failure, non-ventilated survivors, and populations excluded or under-represented in the trial.

Strengths & Limitations

• Strengths:
  • Largest dedicated randomised trial of ICU oxygen targets in unresponsive adults after cardiac arrest.
  • International multicentre design across 53 ICUs.
  • Pragmatic bedside intervention addressing a ubiquitous clinical decision.
  • Patient-centred primary outcome at 180 days rather than short-term oxygen or mortality alone.
  • Blinded outcome assessment for GOS-E, quality of life, and cognition.
  • Prespecified protocol and statistical analysis plan published before enrolment completion.⁵
  • Clear separation in oxygen exposure, making the physiological test credible.
  • Broad inclusion of real-world post-arrest ICU patients, including shockable and non-shockable rhythms and both in-hospital and out-of-hospital arrests.
• Limitations:
  • Open-label ICU care creates risk of performance bias, particularly around ventilation management, arterial blood gas frequency, neuroprognostication, and withdrawal of life-sustaining therapy.
  • Primary outcome data were missing in 112/1821 analysed patients, although missingness was balanced and multiple imputation was used.
  • Eligibility deviations occurred in 76/1821 patients.
  • Protocol deviations occurred in both directions and may have attenuated the treatment contrast.
  • Median randomisation approximately 7 hours after return of spontaneous circulation may be too late to test the earliest reperfusion-injury phase.
  • The conservative strategy increased the proportion of patients with at least one PaO2 <60 mm Hg.
  • The trial does not define the optimal SpO2 or PaO2 range; it only tests these two pragmatic oxygen strategies.
  • Generalisability to prehospital and resource-limited settings is limited.

Interpretation & Why It Matters

• Clinical meaning

  In unresponsive ventilated ICU patients after cardiac arrest, a strategy of actively avoiding SpO2 ≥95% and reducing FiO2 toward 0.21 did not improve 180-day survival with favourable functional outcome compared with a liberal strategy that allowed higher saturations and required FiO2 ≥0.30 during mechanical ventilation.
• Biology versus outcome

  LOGICAL reduced high oxygen exposure, but this surrogate did not translate into neurological recovery, survival, quality of life, cognition, ICU stay, or hospital stay.
• Safety margin

  The conservative strategy reduced PaO2 >100 mm Hg but increased PaO2 <60 mm Hg. This reinforces that oxygen avoidance is not automatically safer in the post-arrest brain.
• Practice impact

  LOGICAL does not support routine aggressive conservative oxygen therapy for all unresponsive ventilated post-arrest ICU patients. A more defensible approach is to avoid both hypoxaemia and unnecessary extreme hyperoxaemia, rather than pursue room air or very low oxygen exposure as a neuroprotective treatment.

Controversies & Subsequent Evidence

• Biological plausibility was real, but not sufficient. The two-hit model of post-arrest brain injury and animal hyperoxia data provided a coherent rationale for oxygen restriction.¹² LOGICAL shows that reducing systemic oxygen exposure after ICU admission does not necessarily alter clinically measurable neurological recovery.
• The ICU-ROX signal is not confirmed as a treatment effect. ICU-ROX generated the central hypothesis for LOGICAL, with lower mortality in the suspected hypoxic–ischaemic encephalopathy subgroup.³ LOGICAL was larger, dedicated to this population, had a neurological primary outcome, and did not show benefit. The most reasonable interpretation is that the ICU-ROX subgroup signal was hypothesis-generating rather than definitive.
• Prehospital conservative oxygen is a different question. EXACT tested lower vs higher oxygen saturation targets in the ambulance phase after out-of-hospital cardiac arrest and was stopped early during the COVID-19 pandemic. Survival to hospital discharge was 38.3% with the lower target vs 47.9% with standard care, difference −9.6 percentage points; 95% CI −18.9 to −0.2, and hypoxic episodes before ICU were more frequent, 31.3% vs 16.1%.⁶ LOGICAL avoids some prehospital titration hazards but still found no benefit from ICU conservative oxygen.
• BOX supports a similar message using PaO2 targets. In comatose survivors of out-of-hospital cardiac arrest, BOX compared restrictive PaO2 9–10 kPa with liberal PaO2 13–14 kPa and found death, severe disability, or coma within 90 days in 126/394 (32.0%) vs 134/395 (33.9%); HR 0.95; 95% CI 0.75 to 1.21; P=0.69.⁷ LOGICAL extends this by testing a larger ICU population with a SpO2/FiO2-based strategy and 180-day GOS-E outcome.
• HOT-ICU did not support benefit in the post-arrest subgroup. In a cardiac arrest subgroup of HOT-ICU, 90-day mortality was 96/147 (65.5%) with a lower oxygenation target and 111/185 (60.0%) with a higher oxygenation target, RR 1.09; 95% CI 0.92 to 1.28.⁸ This subgroup was not the same population as LOGICAL, but it weakens the argument for routine lower oxygenation after cardiac arrest.
• PILOT generated a conflicting secondary signal. A secondary analysis of PILOT found survival with favourable outcome in 50/221 (22.6%) patients assigned to lower or intermediate SpO2 targets vs 15/118 (12.7%) assigned to a higher target; absolute risk difference 9.9 percentage points; 95% CI 1.8 to 18.1; P=0.03.⁹ This was a secondary analysis in a heterogeneous mechanically ventilated cohort, whereas LOGICAL was a dedicated post-arrest RCT and therefore carries greater weight for the specific ICU post-arrest question.
• Broader ICU oxygen trials also argue against routine conservative oxygen for all. UK-ROX randomised 16,500 mechanically ventilated ICU adults and found no significant difference in 90-day mortality with conservative oxygen therapy compared with usual oxygen therapy.¹⁰ LOGICAL is consistent with the broader direction of modern ICU oxygen evidence: avoiding extremes is sensible, but protocolised oxygen restriction is not consistently outcome-improving.
• Guidelines will need to incorporate LOGICAL. Before LOGICAL, the 2025 ILCOR consensus identified oxygen and carbon dioxide targets after cardiac arrest as an important knowledge gap.⁴ A post-LOGICAL practice position should be more restrained: avoid hypoxaemia, avoid unnecessary marked hyperoxaemia, and do not adopt a universal ICU target of SpO2 below 95% or FiO2 0.21 as a proven neuroprotective strategy.
• The accompanying high-level interpretation is pragmatic rather than ideological. A 2026 Critical Care and Resuscitation interpretation of the LOGICAL data framed the accumulating trial evidence as supporting careful oxygen titration and avoidance of extremes, not routine conservative oxygen therapy as a default intervention for unresponsive adults after cardiac arrest.¹¹

Summary

• LOGICAL randomised 1840 unresponsive adults receiving invasive mechanical ventilation in ICU after cardiac arrest to conservative or liberal oxygen therapy; 1821 were analysed.
• Conservative oxygen substantially reduced high oxygen exposure: median ICU hours with SpO2 ≥97% were 16 vs 37, and at least one PaO2 >100 mm Hg occurred in 57.7% vs 78.4%.
• The primary outcome was not improved: survival with favourable functional outcome at 180 days was 38.2% vs 39.7%; RR 0.97; 95% CI 0.87 to 1.09; P=0.65.
• Mortality, quality of life, cognition, mechanical ventilation duration, ICU length of stay, hospital length of stay, and discharge directly home were not improved.
• The conservative strategy increased exposure to PaO2 <60 mm Hg, reinforcing that lower oxygen exposure is not automatically safer in the injured post-arrest brain.

Overall Takeaway

LOGICAL is a landmark post-cardiac arrest oxygen-target trial because it tested a biologically plausible, inexpensive, universal ICU intervention at scale using a patient-centred 180-day neurological outcome. It shows that actively targeting conservative oxygen exposure after ICU admission does not improve functional recovery compared with a liberal oxygen strategy, despite achieving clear physiological separation.

Bibliography

• 1.Sekhon MS, Ainslie PN, Griesdale DE. Clinical pathophysiology of hypoxic ischemic brain injury after cardiac arrest: a “two-hit” model. Crit Care. 2017;21:90.
• 2.Pilcher J, Weatherall M, Shirtcliffe P, Bellomo R, Young P, Beasley R. The effect of hyperoxia following cardiac arrest — a systematic review and meta-analysis of animal trials. Resuscitation. 2012;83:417-422.
• 3.Young PJ, Mackle D, Bellomo R, Bailey M, Beasley R, Deane A, et al. Conservative oxygen therapy for mechanically ventilated adults with suspected hypoxic ischaemic encephalopathy. Intensive Care Med. 2020;46:2411-2422.
• 4.Drennan IR, Berg KM, Böttiger BW, Chia YW, Couper K, Crowley C, et al; Advanced Life Support Task Force Collaborators. Advanced life support: 2025 International Liaison Committee on Resuscitation consensus on science with treatment recommendations. Circulation. 2025;152:S72-S115.
• 5.Young PJ, Hodgson CL, Mackle D, Mather AM, Beasley R, Bellomo R, et al. Protocol summary and statistical analysis plan for the low oxygen intervention for cardiac arrest injury limitation (LOGICAL) trial. Crit Care Resusc. 2023;25:140-146.
• 6.Bernard SA, Bray JE, Smith K, Stephenson M, Finn J, Grantham H, et al; EXACT Investigators. Effect of lower vs higher oxygen saturation targets on survival to hospital discharge among patients resuscitated after out-of-hospital cardiac arrest: the EXACT randomized clinical trial. JAMA. 2022;328:1818-1826.
• 7.Schmidt H, Kjaergaard J, Hassager C, Mølstrøm S, Grand J, Borregaard B, et al. Oxygen targets in comatose survivors of cardiac arrest. N Engl J Med. 2022;387:1467-1476.
• 8.Crescioli E, Lass Klitgaard T, Perner A, Lilleholt Schjørring O, Steen Rasmussen B. Lower versus higher oxygenation targets in hypoxaemic ICU patients after cardiac arrest. Resuscitation. 2023;188:109838.
• 9.DeMasi SC, Clark AT, Muhs AL, Han JH, Shipley K, McKinney JJ, et al; Pragmatic Critical Care Research Group. Oxygen saturation targets and neurologic outcomes following cardiac arrest: a secondary analysis of the Pragmatic Investigation of Optimal Oxygen Targets trial. Chest. 2025;168:1131-1140.
• 10.Martin DS, Gould DW, Shahid T, Doidge JC, Cowden A, Sadique Z, et al; UK-ROX Investigators. Conservative oxygen therapy in mechanically ventilated critically ill adult patients: the UK-ROX randomized clinical trial. JAMA. 2025;334:398-408.
• 11.Young PJ. A logical interpretation of the data on optimal oxygen therapy targets for unresponsive adults in ICU after cardiac arrest. Crit Care Resusc. 2026;28:100191.