Methods

Ethics statement, study design and population

We conducted a single-center retrospective study that was approved by our institutional ethics committee (n° S21031211000) and has been declared to the Commission Nationale de l’Informatique et des Libertés (CNIL) (MR-004 n° 2221599). According to the French law, formal consent was not required for this type of study. We included children aged <12 months hospitalized for bronchiolitis in Le Havre Hospital between September 2018 to February 2021. Non-inclusion criteria were: an age >12 months, a LOS <24h and multiple hospitalizations.

Data collection and outcomes

Children’s characteristics were collected from our institution medical records (SILLAGE® Medical and Care Information System software). The following reports were consulted, in order: the pediatric emergency report, the hospitalization report and the detailed care provided report. The data collected were as follows:

Statistical analysis

The clinical condition (severe, moderate, mild) as defined in the HAS5 was automatically calculated according to the extracted data, using an interface created to optimize the data collection process in Excel - Office 365® software (version 2004), and based on the aforementioned algorithm. For our main objective, adequate feeding was defined as a percentage of the daily amount of reconstituted milk (mL) >70%; on the basis of the Schofield equation and the latest European nutritional recommendations 10,11. The daily amount of reconstituted milk (mL) given to exclusively bottle-fed children was calculated on the basis of the modified Appert standard 12. This calculation took the average growth curve values (weight and height by sex), the caloric values per 100 mL of reconstituted milk (first and second age) and the caloric adaptations for infants <6 months of age (E-Table 1 ). The LOFR was then defined as the time needed to reach a daily amount of feeding >70% of this reference. The LOFR was only calculated for exclusively bottle-fed children, excluding breastfed (completely or partially) infants and those who had diversified feeding. However, the data collected for these infants were analyzable to investigate the factors associated with the LOS.
The usual procedure for ACT referred in our center is as follows: (1) a physician assesses the child and (2) refers to a specialist physiotherapist for ACT evaluation and (3) provide ACT or not regarding the clinical conditions of the child. The evaluation for ACT is repeated every day during hospital stay. For the present study, ACT referral (ACT+) was defined as the realization of an ACT session at least each 48h of hospitalization. A single evaluation was not rated as an ACT session. To distinguish the characteristics of the children that received ACT sessions during their hospital stay for bronchiolitis, the study sample was dichotomized according to whether at least one session of ACT was provided every 48 hours of hospital stay by a physiotherapist (ACT+) or not (ACT-).
Patients’ characteristics are reported as numbers (and/or percentages) for categorical data; and as means (± Standard Deviation (SD)) or median (InterQuartile Range (IQR)) for continuous data according to the distribution of the variables. The normality of the distribution for each variable was assessed using the Shapiro-Wilk test. For our main objectives, we calculated univariable logistic regressions for the dependent variables LOFR and LOS, including the patients’ characteristics as independent variables. For this purpose, the LOFR was transformed into a binary variable between a short length (<1 day) and a prolonged length (≥ 1 day) based on the median LOFR of our population. Similarly, the LOS was also transformed into a binary variable between a short (< 3 days) and a prolonged (≥ 3 days) length, according to the average LOS for bronchiolitis in France13,14. A multivariate logistic regression model, using the dummy variable technique, was then calculated by including all the variables significantly associated (p≤0.05) with each outcome, to isolate factors independently related to prolonged LOFR and LOS. Odds ratios (OR) and their Confidence Intervals (95%CI) were reported.
The characteristics of the children that received ACT and those who did not were compared using the Mann-Whitney U test or Student’s t test for continuous variables and the Fisher’s exact test or the Chi2 test for categorical variables. A p-value ≤0.05 was considered as significant. All the statistical analyses were performed using GraphPad Prism 9® software, version 9.0.1.