Very often in neonatology (and in life) we have to make a decision based on inadequate data. For example, when resuscitating a very preterm baby you have to decide where to set the starting FiO2, even if there are no data at all (as was the case a few years ago, and we ‘all’ started with 100%).

At least now there are some data, and Ola Saugstad and colleagues have just published a systematic review of the studies that are available. They limited the review to randomized trials of resuscitation of infants less than 33 weeks gestation where a starting FiO2 of 21 to 30% was compared with a starting FiO2 of 60 to 100%.

They found 10 studies with a total of around 650 babies included. A lot of them were relatively low risk babies, and the total mortality was below 10%. Nevertheless the mortality was lower in the low FiO2 groups, 6.5% compared to 9.8% with higher FiO2. This gives a relative risk for death of 0.62, i.e. a 38% reduction in mortality. Now this isn’t statistically significant by usual thresholds, with the upper 95% CI being 1.04. The absolute risk reduction is not that great (as these were on average relatively low risk babies), being 3.4%. The other outcomes that the authors analyzed, bronchopulmonary dysplasia and intraventricular hemorrhage, showed no evidence of being different between groups.

Also, as I’ve mentioned before on this blog, meta-analyses based on several small trials often over-estimate the effectiveness of an intervention, so the issue is far from proven, and we certainly need a larger definitive trial.

In the meantime, the next time I go to the delivery room, I will have to decide where to set the starting FiO2, and, based on this analysis, I’ll continue to start with a lowish FiO2 of probably 30%, and hope that we have more definitive data in the near future.