Evaluation of a Low-risk Mild Traumatic Brain Injury and Intracranial Hemorrhage Emergency Department Observation Protocol.

Abstract

Among emergency physicians, there is wide variation in admitting practices for patients who suffered a mild traumatic brain injury (TBI) with an intracranial hemorrhage (ICH). The purpose of this study was to evaluate the effects of implementing a protocol in the emergency department (ED) observation unit for patients with mild TBI and ICH. This retrospective cohort study was approved by the institutional review board. Study subjects were patients ≥ 18 years of age with an International Classification of Diseases code corresponding to a traumatic ICH and admitted to an ED observation unit (EDOU) of an urban, academic LevelI trauma center between February 1, 2015, and January 31, 2017. Patient data and discharge disposition were abstracted from the electronic health record, and imaging data, from the final neuroradiologist report. To measure kappa, two abstractors independently collected data for presence of neuro deficit from a 10% random sample of the medical charts. Using a multivariable logistic regression model with a propensity score of the probability of placement in the EDOU before and after protocol implementation as a covariate, we sought to determine the pre-post effects of implementing a protocol on the composite outcome of admission to the floor, intensive care unit, or operating room from the EDOU and the proportion of patients with worsening findings on repeat computed tomography (CT) head scan in the EDOU. A total of 379 patients were identified during the study period; 83 were excluded as they were found to have no ICH on chart review. Inter-rater reliability kappa statistic was 0.63 for 30 charts. Among the 296 patients who remained eligible and comprised the study population, 143 were in the preprotocol period and 153 after protocol implementation. The EDOU protocol was associated with an independently statistically significant decreased odds ratio (OR) for admission or worsening ICH on repeat CT scan (OR= 0.45, 95% confidence interval [CI]= 0.25-0.82, p=0.009) in the observation unit. After a stay in the EDOU, 26% (37/143) of patients required an inpatient admission before implementation of the protocol and 13% (20/153) of patients required an inpatient admission after protocol implementation. There was no statistically significant difference in log transformed EDOU length of stay (LOS) between the groups after adjusting for propensity score (p=0.34). While there was no difference in EDOU LOS, implementing a low-risk mild TBI and ICH protocol in the EDOU may decrease the rate of inpatient admissions from the EDOU. A protocol-driven observation unit may help physicians by standardizing eligibility criteria and by providing guidance on management. As the propensity score method limits our ability to create a straightforward predictive model, a future larger study should validate the results.