LAUSR

Learning Objectives: The use of non-invasive ventilation (NIV) is increasing and has been associated with reduced intubation rates, shortened ICU length of stays and reduced cost. However, a concern in adults with acute respiratory distress syndrome (ARDS) is the NIV failure rate of almost 50%. The pediatric acute respiratory distress syndrome (PARDS) consensus recommendations include diagnostic criteria based on NIV settings highlighting the growing importance of this modality. However, there are no data comparing outcomes of patients with PARDS who receive NIV to those who do not, prior to the initiation of invasive mechanical ventilation (IMV). We hypothesized that patients who received NIV prior to IMV would have worse outcomes than patients who did not in the Randomized Evaluation of Sedation Titration for Respiratory Failure (RESTORE) trial. Methods: In a secondary analysis of data from the RESTORE trial, we included children from all PARDS categories (at-risk, mild, moderate and severe) who were not admitted with tracheostomies. NIV was defined as BiPAP, HHFNC ≥ 5 L/min of Oxygen flow or CPAP ≥ 5 cm H2O. Initiation of IMV was defined as time of endotracheal intubation. Analyses were performed with risk adjustment for age group, baseline functional performance (POPC > 1) and severity of illness (PRISM III-12 score). Results: Of 2,427 eligible patients, 995 (41%) received NIV prior to IMV. The median duration between start of NIV and intubation was 6.5 hours. Compared to patients who did not receive preintubation NIV, patients who received pre-intubation NIV had lower PRISM III-12 scores (median 6 vs 8, p < 0.001) but were more likely to have a seizure disorder (10% vs 8%, p = 0.04), cancer (11% vs 6%, p < 0.001) or moderate/severe PARDS by the end of the first full day of IMV (68 vs 60%, p = 0.03). In adjusted analyses, patients receiving NIV prior to IMV had longer duration of mechanical ventilation (7.0 vs 6.0 d, p < 0.001), longer PICU (10.8 vs 8.9 d, p < 0.001) and hospital (17 vs 14 d, p < 0.001) lengths of stay and higher 28-day (5% vs 4%, p < 0.001) and 90-day (8% vs 5%, p = 0.01) mortality. Conclusions: In children with PARDS, use of NIV prior to IMV is associated with worse outcomes. These data can be used to inform the design of evaluating optimal timing of initiation of IMV in children with PARDS. * For the RESTORE Study Investigators