Restrictive oxygen targets after cardiac arrest showed no clear benefit in pooled trials

Design

Systematic review and meta-analysis of randomised trials with trial sequential analysis

Population

11 trials, 4,136 mechanically ventilated adults after cardiac arrest

Primary outcome

All-cause mortality and unfavourable neurological outcome

Effect

Mortality RR 0.96 (95% CI 0.88 to 1.06); neurological RR 0.96 (95% CI 0.88 to 1.04); hypoxaemia RR 1.65

This systematic review and meta-analysis searched to June 2026 and pooled 11 randomised trials with 4,136 adults who were mechanically ventilated after cardiac arrest. It compared restrictive oxygenation with higher targets.

All-cause mortality was no different (RR 0.96, 95% CI 0.88 to 1.06) and neither was unfavourable neurological outcome (RR 0.96, 95% CI 0.88 to 1.04). Restrictive targets increased the risk of hypoxaemia (RR 1.65) and added about two ventilator-free days (mean difference 2.14). A mortality reduction in the strict hypothermia (33 to 34°C) subgroup was unstable on trial sequential analysis. Under current normothermia protocols the authors judge that the evidence trends toward futility for a restrictive approach.

The usual aim of avoiding both hypoxaemia and sustained hyperoxia stands. This analysis does not support lowering targets deliberately, and does not show harm from moderate oxygen.

• Pooled mortality: RR 0.96 (95% CI 0.88 to 1.06) for restrictive vs higher oxygen targets.
• Unfavourable neurological outcome: RR 0.96 (95% CI 0.88 to 1.04).
• Restrictive targets raised hypoxaemia risk (RR 1.65).
• Avoid both hypoxaemia and prolonged high oxygen, and titrate to a saturation target.
• The subgroup mortality benefit with strict hypothermia was unstable and should not guide practice.

Why it matters

It suggests the push to run lower oxygen levels after cardiac arrest is unlikely to improve survival.

Don't overread it

The trials differed in targets and temperature management, and the subgroup finding was statistically unstable.

The statistics, in plain English

Both confidence intervals include 1.0, so the data are compatible with no difference or a small benefit or harm. Trial sequential analysis adds a safeguard against declaring a result too early, and here it points toward the question being close to settled for normothermia, although the planned number of patients has not been reached.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.