The NEUROSIS trial that I discussed in my previous post has, among other published trials, most in common with the trial by Nakamura published in 2016. Nakamura T, et al. Early inhaled steroid use in extremely low birthweight infants: a randomised controlled trial. Archives of disease in childhood Fetal and neonatal edition. 2016. That was a trial of very early (<24 hours) inhaled fluticasone given every 12 hours for up to 6 weeks to ELBW infants. Unlike many of the other trials in recent systematic reviews, this was a recent trial with a decent sample size (n=211) and with truly prophylactic administration. The main differences between Nakamura et al and NEUROSIS were that Nakamura only enrolled intubated babies, and the fluticasone (not budesonide as in NEUROSIS) was stopped once the babies were extubated.

Most of the analyses in the Cochrane review that I discussed do not include data from this moderately sized, recent, high quality trial, because the authors chose to define lung injury by the need for oxygen at discharge, rather than at 36 weeks. To me that is a great strength of this study!

The authors of this study found no difference in the combined outcome of death or needing oxygen at discharge (14% with fluticasone, 22% with placebo), but they showed very similar trends to NEUROSIS, with an increase in mortality of 39% with inhaled steroids. That result could, of course, have been due to chance, but it is entirely consistent with the results from NEUROSIS.

As far as I am aware these are the only 2 trials of reasonable size, of prophylactic inhaled steroid use, from the last decade of neonatology, which have reported death before discharge among very immature newborns. If we put their data together, this is what we get: (using Revman 5, random effects model)

A very worrying consistent increase in pre-discharge mortality between the 2 trials, with a lower confidence limit of 1.02.

The numbers needing home oxygen can be calculated from Nakamura (just by subtracting death before discharge from the combined outcome) and that gives the following risk difference, again using the random effects model:

A possible 5% reduction in the need for home oxygen, among survivors to discharge, which may be due to random variation, the 95% confidence intervals include no effect (RD=0).

The Nakamura study reported the long-term neurological and developmental outcomes of their study infants also, and showed no difference in the combined outcome of ‘death or NDI’, numerically these were all slightly higher in the fluticasone group, so presumably identical between groups if you were just to look at the survivors.

The only other trial I can find of inhaled steroids initiated on day one is the pilot trial of Zimmerman et al, enrolling 39 infants in total of < 1301 g birth weight. For some reason this trial is not mentioned in the recent Cochrane review, either as included or excluded, so I guess their search missed it. It was published in October 2000, and is reported as being randomized. In the placebo group there were 2 deaths, and 2 babies transferred out for surgery, for whom no other outcome data are given, in the beclomethasone group there were 3 deaths.  As it is uncertain whether the 2 surgical babies survived, and not clear whether babies needed oxygen at discharge, I have not added their data to the Forest plots above.

I think the 2 other studies are similar enough, in design and in results, to give us some confidence to say that very early inhaled steroids given to extremely immature newborn infants on the first day of life have no clear pulmonary benefit of importance, and may well increase mortality, without an impact of neurological or developmental outcomes among survivors.