Discussion

Synthesis of the results

The present study showed that (1) the moderate use of accessory muscles, nutritional support and RR ≥ 70/min or < 30/min or apnea are associated (OR=1.5), from virtually no association (OR=1.0) to a significant positive association (OR=2.6) with the LOFR and (2) the new severity scale proposed the HAS and other known factors such as oxygen support appeared to be related to LOS in this work and (3) the ACT+ group were older and heavier and had a less severe clinical condition at admission compared to the ACT- group.

Adequate nutrition

One major difficulty for this work was to define what constitutes adequate infant feeding. The current guidelines agree on the importance of maintaining adequate feeding during an episode of bronchiolitis, but a clear definition of the nutritional monitoring and management is lacking 3,15. We chose to define adequate feeding according to the ratio of food intake to basal metabolic rate (BM), with a ratio ≥0.7 indicating adequate nutrition based on the latest Recommended Dietary Allowances (RDAs) of the French population and the European guidelines on parenteral nutrition 11. Previous studies including children hospitalized for bronchiolitis in general ward and pediatric intensive care unit (PICU) proposed the use of different ratios (namely 0.8 and 0.9) to detect undernourished children based on expert committee 16,17. Furthermore, the WHO recommends exclusive breastfeeding for children up to 6 months of age, and dietary diversification starts at about 6 months of age18,19,20. As these two forms of feeding are not suitable for mL measurement, we did not include 117 breastfed infants (24.4%) and 93 who had started diversification (19.5%), which could have impacted our results.
The HAS proposes to consider children with bronchiolitis feeding difficulties from a cut-off of <50% of the habitual on three consecutive intakes 5. According to this definition, almost all children who had feeding difficulties on admission received feeding support during the hospital stay (namely 31% and 28%). This proportion is much lower than others studies (86% and 82%)13,21 and could be explained by the higher objectivity of this cut-off value, compared to the oral parental reporting of feeding difficulties in these studies.

Characteristics of the study sample

The comorbidities previously identified as predictors of higher LOS are congenital heart disease with shunt, chronic pulmonary disease (such as cystic fibrosis), immune deficiency, neuromuscular disease (polyhandicap, Down syndrome), preterm birth (<36 weeks of GA) and corrected age <2 months 5,13,22. We did not find any association between them and prolonged LOS, possibly because they were not highly prevalent (1.4 to 3.5%), except for prematurity (14.4%). Furthermore, compared to Gajdos et al.13, our lower proportions of infants exposed to parental smoking (7% vs. 27%) or with atopic background (20% vs. 40%) could be due to the absence of systematical reporting in the admission or medical records.

Respiratory distress assessment

The median Silverman-Anderson score was 2/10 at admission in the present study, corresponding to a mild respiratory distress. This could seem quite surprising for a cohort of hospitalized children and contradictory compared to the HAS severity scale, who mostly rated our cohort as moderate to severe bronchiolitis (39.9 and 51.8% respectively). These results can be explained by the important variability of symptoms during the beginning or the bronchiolitis course, called the critical period by Florin et al 6. As most infants are hospitalized during the critical period, the clinical variability over 24 hours is important and may lead to inconsistencies in the clinical evaluation scores. Secondly, inter-observer reliability of the Silverman-Anderson score has been shown to be questionable between different caregivers23.

The new severity scale (latest French guidelines)

To our knowledge, this study is the first to investigate the relationship between this new severity scale proposed by the HAS and the hospitalization course of these children. It appeared that this scale was associated with the LOS in the present study, but could be questioning, especially since the median respiratory distress score was low. It is worth noting that the Silverman-Anderson score only rates the respiratory distress, whereas the severity scale includes more items, such as general condition and feeding. Secondly, it is usually advised to measure the infant’s RR over a complete minute because of the variability of this parameter. However, in the context of a retrospective study, the RR could have also been rated according to the ED’s monitor, which is usually higher and more variable, than the one-minute measure. In addition, assessing the general condition of a children is highly subjective, which could explain the discrepancies observed in the severity ratings. Finally, RR and HR are lower during sleep, which may influence the use of accessory muscles24. Unfortunately, the children state of arousal was not reported in the medical charts investigated.

ACT

In daily hospital practice, physiotherapists usually perceive the indication for ACT in the mildest infants with bronchiolitis. Several expert’s opinion has hypothesized that children with mild or too severe bronchiolitis would not benefit for in-hospital ACT25. However, ACT referral is largely based on the subjectivity of the physician’s prescription and on the physiotherapist’s assessment. In our cohort, 63% of the infants received regular ACT during their hospital stay, without any clinically relevant significant difference between them, and those who did not. These results are consistent with the previous evidence on that topic since the studies that observed a benefit for ACT in children with acute viral bronchiolitis included children with a moderate profile26. Clear and objective indications for ACT referral should be identified, as well as the profiles of the children that would eventually have a positive response to ACT. These two elements seem unavoidable to get rid of a one size fits all approach that seem inappropriate, as well as an approach only based on subjective features.