Purpose: Requiring PaO2/FiO2 ratio (PF) to define ARDS may bias towards children with cardiovascular dysfunction and hypoxemia. We sought to evaluate (1) the Berlin definition of ARDS in children using PF; (2) the effect of substituting SpO2/FiO2(SF) ratio; (3) differences between patients with and without arterial blood gases; and (4) the ability of SpO2 and PaO2 indices to discriminate ICU mortality.
Methods: Single center retrospective review (3/2009-4/2013) of mechanically ventilated (MV) children. Initial values for PF, SF, oxygenation index (OI), and oxygen saturation index (OSI) after intubation and average values on day 1 of MV were analyzed against ICU mortality, subgrouped by Berlin severity categories.
Results: Of the 1,833 children included, 129 met Berlin PF ARDS criteria (33 % mortality); 312 met Berlin SF ARDS criteria (22 % mortality). Children with a PaO2 on day 1 of MV had higher mortality and severity of illness, were older, and had more vasoactive-inotropic infusions (p < 0.001). SF could be calculated for 1,201 children (AUC for ICU mortality 0.821), OSI for 1,034 (0.793), PF for 695 (0.706), and OI for 673 (0.739). Average SF on day 1 discriminated mortality better than PF (p = 0.003).
Conclusions: Berlin PF criteria for ARDS identified less than half of the children with ARDS, favoring those with cardiovascular dysfunction. SF or OSI discriminate ICU mortality as well as PF and OI, double the number of children available for risk stratification, and should be considered for severity of illness scores and included in a pediatric-specific definition of ARDS. Multicenter validation is required.