Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
In their article, Thia et al attempted to address the potential benefit of nasal continuous positive airway pressure (NCPAP) compared with the currently practiced standard treatment in the management of infants with bronchiolitis.1 After the established use of NCPAP in neonates, it is promising to see its value being assessed for the use in infants for one of the commonest causes of respiratory compromise. By restricting their outcome measures to demonstrable and achievable partial pressure of capillary carbon dioxide (pCO2) changes the authors have done justice to the aim of the study.
However, a number of clinical and methodological concerns need to be further reflected upon before the authors’ conclusions are applied to standard practice.
The duration of the illness prior to the commencement of treatment in the two groups is not mentioned (even though a comparison was made to the non-included group).
“Comparison of NCPAP with standard treatment”, as stated in their aim, is not happening using their applied methodology. Instead they are comparing the staged introduction of NCPAP with standard treatment during varying phases of bronchiolitis. By not having a group who received standard treatment only and no NCPAP, authors have lost a great opportunity to truly demonstrate the clinical value of NCPAP. Their cross-over design has perhaps influenced the disease process in both groups.
The range of pCO2 from 6 to 12 kPa is very wide in the context of respiratory compromise. The potential value of an intervention could be negligible clinically if a 1 kPa decrease (as chosen by the authors) is happening from 7 to 6 kPa, whereas the same rate of fall could be crucial from 11 to 10 kPa or 10 to 9 kPa especially if the clinical picture also corresponds. The authors’ aim to study the moderately severe bronchiolitis patient group is negated by their inclusion criteria, and this is further reflected in the standard deviation (SD) of pCO2 between the two study groups.
The authors’ justification of not including the level of hypoxia, hypoxaemia, apnoeas and worsening respiratory distress seems to move the study further away from potential clinical usefulness.
Supplemental oxygen used in the study groups to give an oxygen saturation (SaO2) above 92% could be any amount. Perhaps those on NCPAP would have required less fraction of inspired oxygen (FiO2) to maintain SaO2 above 92%. Even when agreeing with the authors’ reasons for not taking FiO2 into account, the complete lack of even a mention of partial pressure of oxygen (PaO2) anywhere in the article puzzles me.
Carefully selected secondary outcome measures give methodological and statistical credibility to the article. A bronchiolitis severity rating or scoring2 could have added clinical validity as well.
Block randomization of four at a time and “blinding” only those who recruited the cases leaves reasonable room for bias. Perhaps there is a reason, but it is not evident in the article. Permuted block randomization is an appropriate tool in this clinical setting; however, a stratified analysis with blocks as strata would have properly controlled for type I error and hence provided the optimal power to detect a possible treatment effect.
The power of numbers is demonstrated in a table depicting the expected fall of 1 kPa in the NCPAP group. The authors did not consider it unreasonable to add −1.35 and +0.5 to get a difference of −1.85 of pCO2, thus getting the p value. The reality is that whatever illness modification was achieved by the initial NCPAP, it drastically changed when standard treatment was commenced. Ironically, it could be argued that the safety and efficacy of standard treatment with oxygen and fluid management, as performed during the 12 h of standard treatment, resulted in the pCO2 actually decreasing by 0.53 (not climbing).
Our 3-year review of 21 infants with bronchiolitis managed with NCPAP, as an add-on treatment when clinically indicated, showed that the maximum sustained reduction of pCO2 and improvement in PaCO2 was in those with chest x-rays that showed changes of collapse/consolidation rather than just uniform hyperinflation.3 Physiologically, such patients would also be more comparable to those with neonatal respiratory distress syndrome (RDS) and adult respiratory distress syndrome (ARDS). Were chest x-rays performed or analysed in the Thia et al study?
One mode of action of NCPAP suggested by the authors on the basis of a previous publication4 that analysed a different patient population was that it decreased functional residual capacity, which is physiologically not plausible. Perhaps it is increasing the functional residual capacity as it does in neonatal RDS. This mode of action was reflected in the subgroup of our patient population with volume loss shown on chest x-rays.
As bronchiolitis is a common infantile morbidity with a relative sparsity of evidence-based treatments, a multi-centre study as suggested by the authors needs to be taken seriously by the paediatric community. I hope that my combing of the article to bring out clinical concerns for an open discussion will be positively received. Authors’ chose the path of “power of numbers” (1 kPa), perhaps an approach of a “number of powers” (standard treatment only arm) would have had the potential to convert this work to a landmark paediatric respiratory study.
Competing interests: None.