Acute respiratory distress syndrome criteria in trauma patients: why the definitions do not work

The international consensus definitions for acute respiratory distress syndrome (ARDS) have formed the basis for recruitment into randomized, controlled trials and, more recently, standardized the protocols for ventilatory treatment of acute lung injury. Although possibly appropriate for sepsis-induced ARDS, these criteria may not be appropriate for posttraumatic ARDS if the disease patterns are widely divergent. This study tests the hypothesis that standard ARDS criteria applied to the trauma population will capture widely disparate forms of acute lung injury and are too nonspecific to identify a population at risk for prolonged respiratory failure and associated complications.

Patients with and Injury Severity Score > or = 16 ventilated for > 12 hours were prospectively enrolled. Clinical data, including elements of cardiovascular, renal, hepatic, hematologic, neurologic, and pulmonary function, were collected daily. Two hundred fifty-four patients were enrolled over a 36-month period, of whom 70 met the consensus definitions of ARDS. Patients from whom support was withdrawn within 48 hours were excluded. The remaining 61 patients were stratified into two groups on the basis of intubation (n = 12) days.

There was considerable disparity in severity and clinical course. A mild, limited form of ARDS was characterized by earlier onset (group 1, 2 days; group 2, 4 days; p = 0.002), fewer intubation days (7 days vs. 28 days; p < 0.001), and less severe derangements in lung mechanics. A significant difference between the two groups was also seen in systemic inflammatory response syndrome score, incidence of sepsis, and incidence of multiple organ failure.

The criteria for ARDS, when applied to the trauma population, capture a widely disparate group and has poor specificity for identifying patients at risk. Recruitment of trauma patients for ARDS studies or preemptive ventilatory management based solely on these criteria may be ill-advised.