We prospectively identified ARDS patients at the LDS Hospital in Salt Lake City, UT. From May 1987 to December 1990, we identified ARDS patients by the presence of ah of the following conditions: (1) acute onset of lung injury requiring endotracheal intubation and mechanical ventilation; (2) alveolar-arterial oxygen pressure difference (P(A0a)O2) of £ 0.2; (3) pulmonary capillary wedge pressure (Pw) of £ 15 mm Hg or no evidence of left atrial hypertension; (4) total static thoracic compliance of £ 50 mol/cm H2O; (5) the presence of bilateral chest radiograph infiltrates; and (6) appropriate risk for ARDS.

From February 1994 to March 1996, we identified ARDS patients by the presence of all of the following: (1) acute onset of lung injury requiring endotracheal intubation and mechanical ventilation; (2) Pao2/fraction of inspired oxygen (Fio2) ratio (P/F) £ 150 mm Hg; (2) Pw of £ 18 mm Hg or no clinical evidence of left atrial hypertension; (3) diffuse chest radiograph infiltrates in three of four quadrants; and (4) appropriate risk for ARDS.

From March 1996 to March 1999, we identified ARDS patients by the presence of ah of the following: (1) acute onset of lung injury requiring endotracheal intubation and mechanical ventilation; (2) P/F of £ 173 mm Hg (equivalent to 200 mm Hg at sea level); (3) bilateral chest radiograph infiltrates; (4) Pw of £ 18 mm Hg or no clinical evidence of left atrial hypertension; and (5) appropriate risk for ARDS.

We assigned one primary risk factor for ARDS to each patient at ARDS identification (ie, pneumonia, trauma, sepsis, and “other”). Other causes included massive transfusion, aspiration, shock, pancreatitis, drug overdose, and unknown. The electronic, hospital-wide database for these identified patients was extensive and comprehensive, allowing us to probe the database in detail.

We used the onset of ARDS as the temporal reference point for organ failure. We determined organ failure using the moderate dysfunction criteria from the Brussels organ failure scoring system for each patient for each day prior to the onset of ARDS and each day after its onset. We scored each organ failure daily as present or absent. An organ failure on any day prior to or after the onset of ARDS established organ failure as being present for the period prior to or after the onset of ARDS. We evaluated sepsis using the severe dysfunction criteria of Montgomery et al for each day prior to the onset of ARDS and each day after its onset. The definition for sepsis was identical for both time periods.