Noninvasive respiratory support (NRS) is commonly trialed in children with acute respiratory distress syndrome (pediatric ARDS) despite limited high-quality evidence of its efficacy. The study aimed to describe patient, disease, and NRS characteristics associated with NRS failure in pediatric ARDS and to compare outcomes between early and delayed transition to invasive ventilation after NRS failure.
This was a retrospective cohort study including patients who met the 2023 PALICC-2 definition for pediatric ARDS and who were initially supported with NRS. The study focused on those whose first respiratory support modality after ARDS diagnosis was noninvasive.
Eligible patients were divided into two groups: NRS success (patients managed with NRS without subsequent endotracheal intubation) and NRS failure (patients initially managed with NRS who later required endotracheal intubation). The NRS failure group was further stratified by timing of intubation: early (≤8 hours from NRS initiation) and late (>8 hours).
A total of 123 subjects met inclusion criteria. Of these, 79 experienced NRS success and 44 experienced NRS failure. The initial median SpO2/FIO2 ratio for the entire cohort was 160, consistent with mild to moderate ARDS; there was no statistically significant difference in initial SpO2/FIO2 between the success and failure groups (P = .12).
Subjects in the NRS failure group were more likely to have genetic comorbidities (P = .009) and had higher PRISM III illness-severity scores (P < .001). ARDS etiologies more commonly associated with NRS failure included sepsis and aspiration.
The overall NRS failure rate reported was 35.8%. In-hospital mortality across the cohort was 4.9%. Patients who experienced NRS failure had significantly longer duration of both pediatric ICU (PICU) stay and overall hospital stay compared with those in whom NRS succeeded (both P < .001).
Among the 44 patients in the NRS failure group, the median time to intubation was 8 hours. The authors compared outcomes between patients intubated early (≤8 hours) and those intubated late (>8 hours).
Early intubation (≤8 h) was associated with a trend toward fewer ventilator days (P = .052) and was associated with significantly shorter PICU stay (P = .03) and shorter overall hospital stay (P = .033) compared with late intubation. These findings suggest that earlier transition to invasive mechanical ventilation after NRS failure may be associated with shorter critical care and hospital length of stay.
Comparisons used chi-squared tests, Fisher exact tests, and Wilcoxon rank-sum tests as appropriate for categorical and nonparametric continuous data. Reported P values for key comparisons included P = .12 for initial SpO2/FIO2 between groups, P = .009 for genetic comorbidities, P < .001 for PRISM III differences and for longer PICU and hospital stays in the failure group, and P values of .052, .03, and .033 for ventilator days, PICU stay, and hospital stay respectively when comparing early versus late intubation within the failure cohort.
In this retrospective cohort of children meeting PALICC-2 pediatric ARDS criteria who were initially managed with noninvasive respiratory support, the NRS failure rate was 35.8% and mortality was low (4.9%). NRS failure correlated with higher illness severity (PRISM III), presence of genetic comorbidities, and ARDS causes such as sepsis and aspiration, and was associated with significantly longer PICU and hospital length of stay. Within the subset who failed NRS, earlier endotracheal intubation (≤8 hours) was associated with shorter PICU and hospital stays and a trend toward fewer ventilator days compared with later intubation (>8 hours).
These results characterize associations between baseline patient features, timing of escalation to invasive ventilation, and resource utilization in pediatric ARDS patients initially trialed on NRS. The abstract does not provide detailed information about specific NRS modalities used, unit protocols for escalation, or adjustment for potential confounders beyond reported variables; those details were not reported in the source abstract.