Variation in the rates of do not resuscitate orders after major trauma and the impact of intensive care unit environment

Background: There is an increased emphasis on benchmarking of trauma mortality outcomes as a measure of quality. Differences in approaches to end-of-life care or perceptions of salvageability might account for some of the variability in outcomes across centers. We postulated that these differences in perceptions or practice might lead to significant variation in the use of do not resuscitate (DNR) orders and sought to identify institutional characteristics associated with their use. Methods: Patients surviving > 24 hours and admitted to an intensive care unit (ICU) in one of 68 centers across the United States were identified from a large prospective cohort study of severely in-jured patients. Independent predictors of a DNR order at both the patient and institutional level were identified using multivariate hierarchical modeling stratified by age < 55 or >= 55. Results: Of 6,765 patients, 7% had a DNR order, of whom 88% died. The proportion of patients in each center with a DNR order ranged from 0% to 57%. Independent patient-level predictors associated with a DNR order were increasing age, preinjury comorbidity burden, severe injury, and organ failure. Institutional predictors of DNR orders differed by age. Care in an open ICU was associated with a DNR order (odds ratio, 1.7; 95% confidence interval, 1.0-3.0) in the elderly, whereas care in a combined medical-surgical ICU (vs. surgical or trauma ICU) was associated with greater likelihood (odds ratio, 2.0; 95% confidence interval, 1.1-4.1) of a DNR order in the young. Conclusions: DNR orders are relatively common in seriously injured trauma patients, and there is significant variability in their use across centers. Given the institutional characteristics independently associated with DNR status, it is likely that both differences in the ethos of end-of-life care and perceptions of salvageability affect decision making.