A long break from blogging this summer, with a trip to Vietnam, and other great experiences over the last few weeks. I will put up some of my Vietnamese wildlife photos on a page very soon (mostly birds).
I thought I would start with a series of short posts about recent publications, before getting back into my usual prolonged dissection of studies.
To start off, a study that I already discussed after the PAS meeting has now been published in full, that is the randomized trial designed to protect noses during CPAP. Imbulana DI, et al. A Randomized Controlled Trial of a Barrier Dressing to Reduce Nasal Injury in Preterm Infants Receiving Binasal Noninvasive Respiratory Support. The Journal of pediatrics. 2018;201:34-9 e3. A clinically important intervention with a clearly positive result, hydrocolloid dressings should be the standard of care unless someone can develop something better..
Secondly I discussed the Family Integrated Care study by Karel O’Brien and colleagues previously, and noted that I couldn’t understand the primary outcome, that it seemed to have been miscalculated. The paper stated that the babies had a z-score for their body weight that had changed by about +1.5 in each group, but the data didn’t seem to make sense as the babies appeared to have dropped off the growth centiles a little, over the 4 weeks of the study period, and I thought they should have had somewhat negative z-score changes. The authors have now published a correction to their article, the correction notice states that the primary outcome was actually a z-score change of −0·071 [SD 0·42] vs −0·155 [0·42]; p<0·0002. The adjusted difference in z-scores between groups was also corrected to a difference of 0·095 (95% CI 0·0–0·14), p<0·0001. I now don’t understand how a difference between groups, for which the 95% confidence intervals reach 0.0, can be significant at p<0.0001; even if there is a rounding issue, and the 95% confidence limit is in reality a little more than 0.0, it seems unlikely that this could translate into a p-value which is so tiny. Anyway, the on-line version has apparently been corrected (I have asked for a new copy of the pdf from the inter-library loan system) and the abstract, which now appears in PubMed because the journal is now indexed, has also been corrected, you can follow the link above to the entry in PubMed. What the new result means of course is that both groups of babies grew less than the Fenton percentile curves, and lost about 1 tenth of a standard deviation of their weight in comparison to standardized growth percentiles, losing less in the FICare group than the controls.