Variability in Early Do Not Attempt Resuscitation Orders Among Patients With Serious Traumatic Brain Injury

Abstract

The objective was describe the use of early do not attempt resuscitation (DNAR) orders in patients with serious traumatic brain injury (TBI) and its association with outcomes. This was a retrospective cohort study of all serious TBI patients admitted through the emergency department (ED) to acute care hospitals in California between 2002 and 2010 using ED International Classification of Diseases, 9th revision (ICD-9), admitting diagnosis codes specifying intracranial hemorrhage. DNAR placement within 24 hours of admission was the primary variable of interest. Outcomes included neurosurgical procedures and in-hospital mortality. Hospital and patient characteristics were analyzed using descriptive statistics and multivariable generalized estimation equation regression models to account for hospital-level clustering. Of all 76,962 patients with serious TBI, 71,275 were admitted at 141 hospitals that each cared for at least 10 serious TBI patients annually and formed the primary sample. Early DNAR orders were placed in 7.5% of patients (range = 0 to 36.1% by hospital). Early DNAR use varied by trauma designation: Level I, 4.0% (95% confidence interval [CI] = 3.8% to 4.4%); Level II, 6.7% (95% CI = 6.5% to 7.1%); Level III, 9.7% (95% CI = 8.4% to 11.3%); and nontrauma hospitals, 10.8% (95% CI = 10.6% to 11.3%). Early DNAR was also less likely in teaching hospitals (9.3% vs. 4.3%). These results persisted after accounting for age, year, and hospital-level clustering. In-hospital mortality (39.4% vs. 8.7%) and neurosurgical interventions (14.5% vs. 19.7%) also differed for patients with versus without early DNAR orders. Patients 65 years of age and older constituted 87.7% of those with early DNAR orders; our findings remained qualitatively unchanged when restricted to older adults. Use of early DNAR orders among patients with serious TBI is highly variable by individual hospital and hospital type, suggesting substantial practice variation. Associations with fewer surgical intervention and higher mortality suggest that such practice variation may be contributing to differences in TBI outcomes, particularly among older adults.