Design: unblinded cluster randomised crossover trial across 26 maternity hospitals
Population: 1,818 neonates born at 32-35 weeks needing respiratory support within 3 minutes of birth
Primary outcome: ongoing respiratory support when leaving the delivery room
Effect: 72.6% with FiO2 0.30 versus 73.3% with 0.21; risk difference -0.83% (95% CI -4.33 to 2.67)

AIROPLANE was a cluster randomised crossover trial across 26 Australian maternity hospitals covering the full range of tertiary, regional, public and private units. All 1,818 babies born at 32 to 35 weeks without major anomalies who needed respiratory support within three minutes of birth were enrolled under a consent waiver. Each site used an initial FiO2 of 0.30 or 0.21 for the first three minutes, then crossed over halfway through recruitment.

The primary outcome, still receiving respiratory support on leaving the delivery room, occurred in 72.6% with 0.30 and 73.3% with 0.21 - a risk difference of 0.83 percentage points (95% CI -4.33 to 2.67). Ten of 12 secondary outcomes were also no different. The two that were favoured 30% oxygen: escalation to higher levels of delivery room support was less common (proportional odds ratio 0.70, 95% CI 0.52 to 0.95), and fewer infants needed endotracheal ventilation beyond the delivery room (6.4% versus 9.3%, risk ratio 0.69, 0.47 to 0.91).

The honest summary is a negative primary outcome with two secondary signals pointing the same way. That is not enough to mandate a change, and the primary endpoint chosen - still on support at the door - may simply be too blunt to separate the groups, since most of these babies are on support either way. For units that already start at 0.21 there is no reason here to move; for units that start at 0.30 there is no evidence of harm.

Neither starting FiO2 is wrong on this evidence - local protocol consistency matters more than the choice
Have a pulse oximeter on and targeted saturations driving titration, which this trial does not replace
Note the population: 32 to 35 weeks, not extremely preterm, where the oxygen question is different
Treat the escalation findings as secondary and unblinded - the trial was open-label by design
Audit how quickly FiO2 is actually titrated after the first three minutes; that is the less-examined variable

Why it matters

It settles a question units currently answer by habit, and settles it as: it does not much matter.

The statistics, in plain English

The primary result's confidence interval runs from 4.33 percentage points in favour of 30% oxygen to 2.67 in favour of air, so the trial excludes any large difference in either direction rather than proving equivalence. Ten of 12 secondary outcomes were null; when two of twelve reach significance, some of that is expected by chance alone, and neither was the endpoint the trial was designed and powered to answer. The design is also unblinded, so decisions to escalate support - the outcome that did differ - were made by clinicians who knew the assigned oxygen.