Conservative oxygen after cardiac arrest did not improve survival in a large trial

After a cardiac arrest, the brain that survives is the outcome that matters, and clinicians have long suspected that how much oxygen a ventilator delivers in the hours afterwards might shape it. Too little oxygen starves an injured brain; too much may generate reactive oxygen species that injure it further. The theory has produced a decade of small, inconclusive trials. LOGICAL is the large one, and it found no difference [s1].

What the trial tested

LOGICAL randomly assigned unresponsive adults receiving mechanical ventilation in the intensive care unit after cardiac arrest to conservative or liberal oxygen therapy [s1]. In both groups the default lower limit for arterial oxygen saturation measured by pulse oximetry was 90% [s1].

The two protocols diverged at the top. In the conservative-oxygen group, the upper-limit alarm for oxygen saturation was set at 95%, and the fraction of inspired oxygen was decreased to 0.21 — room air — provided saturation stayed above the lower limit [s1]. In the liberal-oxygen group, there were no measures limiting the upper saturation, and the minimum fraction of inspired oxygen permitted during mechanical ventilation was 0.3 [s1].

The primary outcome was survival with a favourable functional outcome at 180 days, assessed with the Extended Glasgow Outcome Scale (GOS-E), which runs from 1 (death) to 8 (upper good recovery) [s1]. A favourable outcome was defined as a GOS-E level of 5 — lower moderate disability — or higher [s1]. It is a patient-centred endpoint, not a laboratory surrogate: it asks whether the person can live with some independence, not merely whether a number on a monitor improved.

What happened

A total of 1,840 patients were recruited from 53 ICUs in Australia, New Zealand and Ireland, with 882 assigned to conservative oxygen therapy and 958 to liberal oxygen therapy [s1].

A favourable 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 [s1]. The relative risk was 0.97 (95% confidence interval, 0.87 to 1.09), with P = 0.65 [s1]. No adverse events were reported [s1].

There is no signal to interpret here. The point estimate sits almost exactly on one, the confidence interval is symmetrical around it, and the two survival percentages differ by 1.5 percentage points in favour of the group that received more oxygen. The authors' conclusion is flat: among unresponsive adults undergoing mechanical ventilation after a cardiac arrest, the percentage who survived with a favourable functional outcome was not higher with conservative oxygen therapy than with liberal oxygen therapy [s1].

Why a null result here is worth reporting

A trial that changes no practice can still be the most useful one published in a season, because it closes a question that had been kept open by underpowered studies and mechanistic plausibility. The worry about oxygen toxicity after arrest was reasonable, and it had generated enough smaller trials to sustain uncertainty. LOGICAL is large enough — 1,840 patients across three countries — to say that if a clinically meaningful difference exists at this level of titration, this trial did not find it [s1].

It also matters who ran it. LOGICAL was conducted by the LOGICAL Investigators and the Australian and New Zealand Intensive Care Society Clinical Trials Group, and funded by the Health Research Council of New Zealand and others, with support from the National Health and Medical Research Council [s1]. It was registered before enrolment on the Australian New Zealand Clinical Trials Registry as ACTRN12621000518864 [s2]. This is publicly and charitably funded, investigator-led critical-care research, without a product to sell — the kind of trial whose null result carries no commercial reason to be reframed as a near miss.

Limits

The comparison was between two saturation targets, not between oxygen and no oxygen: both groups shared a lower saturation limit of 90%, so LOGICAL tested how tightly to cap oxygen, not whether to give it [s1]. A more aggressive conservative strategy, or one applied earlier in the prehospital phase before ICU admission, was not studied [s1]. The trial enrolled adults who were unresponsive after resuscitation, and its findings speak to that population rather than to every patient who arrests [s1].

The endpoint was assessed at 180 days; longer-horizon differences in cognition or function are beyond what this analysis reports [s1].

What to watch

Whether guideline groups now fold LOGICAL into their oxygen-titration recommendations for post-arrest care, and whether attention shifts from the ICU ceiling to the earlier links in the chain — the minutes around resuscitation itself, where oxygen exposure is harder to measure and harder to randomise. For the specific question LOGICAL asked, the answer is settled for now: capping oxygen more tightly did not help these patients survive with a better outcome [s1].

This article describes trial results. It is not medical advice, and nothing here should be used to guide the care of any individual patient.

Sources

• Conservative Oxygen for Unresponsive Patients after Cardiac Arrest, New England Journal of Medicine, 10 June 2026
• LOGICAL trial registration (ACTRN12621000518864), Australian New Zealand Clinical Trials Registry