Derivation of a preliminary clinical decision rule to guide the interhospital transfer of patients with traumatic brain injury using a resource utilization outcome

Abstract

Study objectives: We derive a preliminary clinical decision rule for persons with traumatic brain injury (TBI) to identify the group of patients requiring specialized tertiary care resources within 72 hours of presentation, using variables readily available to the clinician. Trained medical record abstractors collected data on a retrospective cohort of consecutive persons with blunt head injury using a structured data collection instrument from 1999 through 2003. Patients aged 18 to 65 years, transported directly from the scene of injury, evaluated in the emergency department (ED) of a Level I trauma center, and retrospectively identified as having a serious head injury (Abbreviated Injury Scale [AIS] score ≥3) were included in the analysis. Resource-based high therapeutic intensity (HTI) measures warranting specialized trauma care were identified a priori (all within 72 hours after ED arrival), including neurosurgical intervention, exploratory laparotomy, intensive care (ventilator support >48 hours, transfusion ≥6 units of packed RBC, vasopressor support), or death. We used classification and regression tree analysis to derive a preliminary decision rule using 19 covariates (patient demographics, comorbid disease, vital signs, laboratory values, and radiographic findings) and cross-validation techniques to estimate rule sensitivity and specificity. Binomial 95% confidence intervals (CIs) are provided for the derivation sample. Results: Five hundred four consecutive trauma patients were identified as having a head injury, of whom 258 patients (51%) required at least 1 of the HTI measures. Five variables (emergency medical services [EMS] Glasgow Coma Scale [GCS] score, ED GCS, serum bicarbonate level, EMS respiratory rate, and ED respiratory rate) had a sensitivity of 93% and specificity of 42% for identifying patients requiring 1 of the HTI measures (derivation sample: sensitivity 95% [95% CI 92% to 98%] and specificity 52% [95% CI 46% to 58%]). Conclusion: This preliminary decision rule identified most patients with TBI in the derivation sample requiring urgent tertiary trauma care resources and was much more specific than using a standard injury-based criterion for transfer (eg, head AIS score ≥3). The generalizability of this rule for head-injured patients presenting to nontrauma hospitals will need to be evaluated.