Rationale: Current American-European Consensus Conference definitions for acute lung injury (ALI) and acute respiratory distress syndrome (ARDS) are inadequate for inclusion into clinical trials due to the lack of standardization for measuring the oxygenation defect.

Objectives: We questioned whether an early assessment of oxygenation on specific ventilator settings would identify patients with established ARDS (persisting over 24 h).

Methods: At the time of meeting ARDS criteria (Day 0) and 24 hours later (Day 1), arterial blood gases were obtained on standard ventilator settings, Vt 7 ml/kg predicted body weight plus the following positive end-expiratory pressure (PEEP) and Fi(O(2)) settings in sequence: (1) PEEP >or= 5 cm H(2)O and Fi(O(2)) >or= 0.5, (2) PEEP >or= 5 cm H(2)O and Fi(O(2)) 1.0, (3) PEEP >or= 10 cm H(2)O and Fi(O(2))>or=0.5, and (4) PEEP >or= 10 cm H(2)O and Fi(O(2)) 1.0.

Measurements and main results: One hundred seventy patients meeting ARDS criteria (Pa(O(2))/Fi(O(2)) 128 +/- 33 mm Hg) were enrolled. Overall hospital mortality was 34.1%. The standard ventilator settings that best identified patients with established ARDS and predicted differences in intensive care unit (ICU) mortality were PEEP >or= 10 cm H(2)O and Fi(O(2)) >or= 0.5 at Day 1 (P = 0.0001). Only 99 (58.2%) patients continued to meet ARDS criteria (Pa(O(2))/Fi(O(2)), 155.8 +/- 29.8 mm Hg; ICU mortality, 45.5%), whereas 55 patients were reclassified as having ALI (Pa(O(2))/Fi(O(2)), 246.5 +/- 25.6 mm Hg; ICU mortality, 20%) and 16 patients as having acute respiratory failure (Pa(O(2))/Fi(O(2)), 370 +/- 54 mm Hg; ICU mortality, 6.3%) (P = 0.0001) on these settings.

Conclusions: Patients meeting current American-European Consensus Conference ARDS criteria may have highly variable levels of lung injury and outcomes. A systematic method of assessing severity of lung injury is required for enrollment of patients with ARDS into randomized controlled trials. Clinical trial registered with www.clinicaltrials.gov (NCT 00435110).