- Design
- Multicentre, unblinded, randomised trial (SAVE-O2 AI)
- Population
- 300 acutely ill adults on supplemental oxygen at four US hospitals
- Primary outcome
- Proportion of time with SpO2 90 to 96% over 72 hours
- Effect
- 85% vs 63%; adjusted difference 21 points (95% CI 18 to 25)
SAVE-O2 AI randomised 300 adults admitted with acute respiratory illness, trauma, burns or acute surgical conditions who needed oxygen. Half had oxygen flow adjusted continuously by a closed-loop device aiming for 90 to 96%; half had usual manual titration by staff.
The device kept patients in range far more of the time. Most of the gain came from avoiding unnecessary oxygen: usual care left patients above 96% for nearly a third of the time. Hypoxaemia was also reduced, though it was uncommon in both groups.
The trial did not look at clinical outcomes. Its value for the ward physician is the size of the hyperoxaemia gap under ordinary care, which a prescribed target range and a daily weaning check can narrow without any device.
- Under usual care, patients spent 29% of the time above 96% saturation; the device cut that to 9%.
- Time in the 90 to 96% target range rose from 63% to 85% with automated titration.
- Time below 88% fell modestly, from 3.6% to 2.0%.
- Results held across light, medium and dark skin tones on the Monk scale.
- Automated titration devices are not widely available; the lesson for now is closer manual weaning.
Why it matters
Over-oxygenation, not just hypoxaemia, is the everyday failure on medical wards, and it goes unnoticed because the number looks reassuring.
Don't overread it
The trial measured time in a saturation range, not mortality or length of stay.
The statistics, in plain English
A 21-point difference with an interval of 18 to 25 is large and precise. Time in range is a process measure; whether it changes outcomes needs a larger trial.
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.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for internal medicine, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free