Developing a decision instrument to guide computed tomographic imaging of blunt head injury patients

BACKGROUND: Computed tomographic (CT) head scanning of blunt trauma patients is expensive, delays care, and necessitates radiation exposure, while detecting intracranial injuries in a minority of patients. Clinical characteristics may be able reliably identify patients who do not have intracranial injuries and consequently, do no require imaging. METHODS: Physicians assessed blunt trauma patients undergoing imaging for the presence or absence of specific criteria. Recursive partitioning was used to identify criteria that predict intracranial injuries with high sensitivity. RESULTS: Intracranial injuries were found in 917 of 13,728 enrolled patients (6.7%). Injuries were rare among patients under age 65 who had no evidence of skull fracture, scalp hematoma, neurologic deficit, abnormal alertness, abnormal behavior, coagulopathy, or persistent vomiting. These characteristics would have identified 901 injury cases (sensitivity 98.3% [CI: 97.2-99.0]), while classifying 1,752 patients (12.8%) as "low risk." CONCLUSIONS: Clinical characteristics can reliably identify patients who are unlikely to have intracranial injuries and who do not require CT imaging.