Objective: Although all definitions of acute respiratory distress syndrome use some measure of hypoxemia, neither the Berlin definition nor recently proposed pediatric-specific definitions proposed by the Pediatric Acute Lung Injury Consensus Conference utilizing oxygenation index specify which PaO2/FIO2 or oxygenation index best categorizes lung injury. We aimed to identify variables associated with mortality and ventilator-free days at 28 days in a large cohort of children with acute respiratory distress syndrome.
Design: Prospective, observational, single-center study.
Setting: Tertiary care, university-affiliated PICU.
Patients: Two-hundred eighty-three invasively ventilated children with the Berlin-defined acute respiratory distress syndrome.
Interventions: None.
Measurements and main results: Between July 1, 2011, and June 30, 2014, 283 children had acute respiratory distress syndrome with 37 deaths (13%) at the Children's Hospital of Philadelphia. Neither initial PaO2/FO2 nor oxygenation index at time of meeting acute respiratory distress syndrome criteria discriminated mortality. However, 24 hours after, both PaO2/FIO2 and oxygenation index discriminated mortality (area under receiver operating characteristic curve, 0.68 [0.59-0.77] and 0.66 [0.57-0.75]; p < 0.001). PaO2/FIO2 at 24 hours categorized severity of lung injury, with increasing mortality rates of 5% (PaO2/FIO2, > 300), 8% (PaO2/FIO2, 201-300), 18% (PaO2/FIO2, 101-200), and 37% (PaO2/FIO2, ≤ 100) across worsening Berlin categories. This trend with 24-hour PaO2/FIO2 was seen for ventilator-free days (22, 19, 14, and 0 ventilator-free days across worsening Berlin categories; p < 0.001) and duration of ventilation in survivors (6, 9, 13, and 24 d across categories; p < 0.001). Similar results were obtained with 24-hour oxygenation index.
Conclusions: PaO2/FIO2 and oxygenation index 24 hours after meeting acute respiratory distress syndrome criteria accurately stratified outcomes in children. Initial values were not helpful for prognostication. Definitions of acute respiratory distress syndrome may benefit from addressing timing of oxygenation metrics to stratify disease severity.