One of the things that we should always remember in neonatology is that the main goal of any supportive therapy is to stop it. The main goal of blood pressure support should be to achieve hemodynamic stability in order to be able to stop the blood pressure support. The main goal of intubation is extubation. The main goal of CPAP is to stop CPAP, and so on.

Which makes it remarkable that there are so few studies about how best to wean from support, including respiratory support.

We are slowly accumulating some information about how to wean from respiratory support, and this new trial supports previous studies showing that cycling time-off CPAP is not helpful.  The comparisons studied were just stopping the CPAP from 5 cm H2O compared to cycling the baby on and off CPAP while gradually increasing the time off (3 on 3 off for 48 hours, then 3 on 6 off for 48h).  Which is quite similar to the study by Todd et al that I previously mentioned on this blog. This new trial enrolled only 56 babies, the sample size was determined by hypothesizing an enormous difference in success rate, 33% with sudden weaning to 66% with cycling.  Success of weaning was defined as the absence of persistent tachypnea, marked retractions or apneic episodes on room air with no ventilator support, and need for supplemental oxygen for 7 days.

This study was negative (by which I mean there was no difference in outcomes that could not have easily been ascribed to chance). Clearly, the trial is underpowered, it is presented as a pilot trial, which I think is OK as long as the authors recognize the limitations of the small sample and report the confidence intervals. I don’t see in this report the confidence intervals for the primary outcome variable, I think that should be routine, small negative trials (in fact all trials positive and negative of any size) should report the confidence intervals. That way a reader can tell how large an effect could have been missed.

For this study there were 13 successes in the immediate weaning, and 12 in the cycling group. The difference between those proportions is 0.036, with confidence interval of -0.23 to +0.29. Meaning that the real difference in success of weaning, consistent with these results with 95% confidence, lies somewhere between a 29% better success of weaning with the sudden approach and a 23% better success with cycling.

This is not difficult to calculate; I used an on-line calculator, and it took me less than 2 minutes. Reviewers should require that such information be presented, as it is the only way to interpret the meaning of a result like the one from this study.