Conservative Oxygen for Unresponsive Patients after Cardiac Arrest

Published June 10, 2026
N Engl J Med 2026;395:571-581
DOI: 10.1056/NEJMoa2513814
Abstract
Background
In patients who are unresponsive after resuscitation from cardiac arrest, limiting oxygen exposure to that necessary to achieve acceptable oxygenation may increase the likelihood of survival with a favorable functional outcome.
Methods
We randomly assigned unresponsive adults receiving mechanical ventilation in the intensive care unit (ICU) after cardiac arrest to conservative or liberal oxygen therapy. In the two groups, the default lower limit of arterial oxygen saturation as measured by pulse oximetry (Spo2) was 90%. In the conservative-oxygen group, the alarm for the upper limit of the Spo2 was set at 95%, and the fraction of inspired oxygen (Fio2) was decreased to 0.21 provided that the Spo2 was above the lower limit. In the liberal-oxygen group, there were no measures limiting the upper Spo2, but the minimum Fio2 permitted during mechanical ventilation was 0.3. The primary outcome was survival with a favorable functional outcome at 180 days, assessed with the Extended Glasgow Outcome Scale (GOS-E). Levels on the GOS-E range from 1 (death) to 8 (“upper good recovery”). We defined survival with a favorable functional outcome as a GOS-E level of 5 (“lower moderate disability”) or higher.
Results
A total of 1840 patients were recruited from 53 ICUs in Australia, New Zealand, and Ireland, with 882 assigned to conservative oxygen therapy and 958 assigned to liberal oxygen therapy. A favorable functional outcome at 180 days was observed for 313 of 819 patients (38.2%) in the conservative-oxygen group and 353 of 890 patients (39.7%) in the liberal-oxygen group (relative risk, 0.97; 95% confidence interval, 0.87 to 1.09; P=0.65). No adverse events were reported.
Conclusions
Among unresponsive adults undergoing mechanical ventilation in the ICU after a cardiac arrest, the percentage who survived with a favorable functional outcome was not higher with conservative oxygen therapy than with liberal oxygen therapy. (Funded by the Health Research Council of New Zealand and others; LOGICAL Australian New Zealand Clinical Trials Registry number, ACTRN12621000518864.)
Are you a member of an institution such as a university or hospital?Learn more about Institutional Access
Notes
This article was published on June 10, 2026, at NEJM.org.
A data sharing statement provided by the authors is available with the full text of this article at NEJM.org.
Supported by the Health Research Council of New Zealand, the Alpha Charitable Trust, and the David and Cassie Anderson Medical Charitable Trust administered by the Perpetual Guardian Trust (all in New Zealand); the National Health and Medical Research Council (Australia); and the Irish Critical Care Clinical Trials Network (Ireland).
Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.
Supplementary Material
Information & Authors
Information
Published In
Copyright
Copyright © 2026 Massachusetts Medical Society. All rights reserved.
For personal use only. Any commercial reuse of NEJM Group content requires permission.
History
Published online: June 10, 2026
Published in issue: August 6, 2026
Topics
Authors
Metrics & Citations
Metrics
Altmetrics
Citations
Export citation
Select the format you want to export the citation of this publication.
Cited by
- Oxygen Targets after Cardiac Arrest — A LOGICAL Finding, New England Journal of Medicine, 395, 6, (605-606), (2026)./doi/full/10.1056/NEJMe2609084
- Conservative versus liberal oxygen therapy in hypoxic ischemic encephalopathy following cardiac arrest: a trial-based cost-effectiveness analysis, Resuscitation Plus, (101403), (2026).https://doi.org/10.1016/j.resplu.2026.101403
Loading...

Facts Only

* 1840 patients were recruited from 53 ICUs in Australia, New Zealand, and Ireland.
* 882 patients received conservative oxygen therapy.
* 958 patients received liberal oxygen therapy.
* The default lower limit for $\text{SpO}2$ was 90% across both groups.
* Conservative group: Upper limit $\text{SpO}2$ set at 95%; $\text{FIO}2$ decreased to 0.21 if $\text{SpO}2 > 90\%$.
* Liberal group: No upper $\text{SpO}2$ limit; minimum $\text{FIO}2$ permitted was 0.3.
* Primary outcome: Survival with a favorable functional outcome at 180 days, assessed by $\text{GOS-E}$ score of 5 or higher.
* Conservative group favorable outcome rate: 313 of 819 patients (38.2%).
* Liberal group favorable outcome rate: 353 of 890 patients (39.7%).
* Relative risk for a favorable functional outcome: 0.97; 95% CI: 0.87 to 1.09; $P=0.65$.
* No adverse events were reported.

Executive Summary

Patients unresponsive after cardiac arrest receiving mechanical ventilation in the ICU were randomized into two groups: conservative oxygen therapy and liberal oxygen therapy. The lower limit for arterial oxygen saturation ($\text{SpO}2$) was set at 90% for both groups. In the conservative-oxygen group, the upper limit for $\text{SpO}2$ was 95%, and the fraction of inspired oxygen ($\text{FIO}2$) was restricted to 0.21 when $\text{SpO}2$ remained above the lower limit. In the liberal-oxygen group, there were no upper limits on $\text{SpO}2$, but the minimum $\text{FIO}2$ allowed during ventilation was 0.3. The primary outcome assessed at 180 days was survival with a favorable functional outcome, defined by an Extended Glasgow Outcome Scale ($\text{GOS-E}$) score of 5 or higher (lower moderate disability). Of the patients in the conservative group, 38.2% achieved this outcome, compared to 39.7% in the liberal group. The relative risk for a favorable functional outcome was calculated as 0.97 with a 95% confidence interval of 0.87 to 1.09, and the difference was not statistically significant ($P=0.65$).

Full Take

The finding that the rate of survival with a favorable functional outcome was numerically slightly higher in the liberal-oxygen group (39.7% vs. 38.2%) did not reach statistical significance ($P=0.65$). This suggests that, within this specific patient cohort post-cardiac arrest and mechanical ventilation, the degree to which limiting oxygen exposure provides a decisive advantage over allowing more oxygenation is highly dependent on unmeasured variables or the precise context of the intervention thresholds. The comparison forces an examination of the tension between clinical conservatism (limiting physiological stress) and providing maximal support during acute recovery.
The lack of statistical significance complicates any definitive statement about oxygen therapy protocols in this setting. It implies that while conservative measures are intuitively appealing, they do not offer a statistically demonstrable benefit or detriment to long-term functional outcomes in this specific population tested. This pattern suggests that the risk associated with restricting oxygen—or the potential benefits derived from higher $\text{FIO}2$ levels in the liberal group—are balanced across the observed outcomes.
The implication for cognitive sovereignty lies in recognizing that optimizing interventions requires moving beyond simple numerical comparisons of endpoints. The decision-making process must account for the ethical and practical weight of limiting physiological parameters versus maintaining maximal support, acknowledging that statistical significance does not equate to clinical superiority or inherent moral correctness regarding patient care decisions. Further inquiry is necessary into whether the subtle difference in $\text{FIO}2$ thresholds reflects true biological necessity or merely methodological variance within the trial design itself.

Sentinel — Human

Confidence

This text exhibits the high density of technical, verifiable data and formal academic structure characteristic of original medical research reporting.

Signals Detected

Varied sentence structure and formal academic tone typical of medical journal writing.

Consistent logical flow from background to methods, results, and conclusion; no overt emotional or persuasive framing.

Use of specific methodological details (GOS-E scale, Fio2 values, specific trial numbers) points to source material rather than template matching.

The citation structure, funding acknowledgments, and reference to peer-reviewed journal standards suggest high authenticity.

Human Indicators

Specific statistical reporting (Relative risk, CI, P-value) tied to a formal trial structure.

Detailed description of randomized assignment groups and specific intervention thresholds (95% Spo2 limits, Fio2 values).