Practical selection criteria for noncontrast cranial computed tomography in patients with head trauma.

STUDY
OBJECTIVE: To study patients with acute trauma retrospectively for clinical predictors of positive cranial computed tomography.
METHODS: We reviewed the medical records and noncontrast computed tomography scans of 373 consecutive head trauma patients presenting to a trauma Level I emergency department. Potential criteria for patient selection were analyzed statistically, using both univariate and multivariate models.
RESULTS: Our data suggest that relying on four variables--positive neurologic examination, intoxication, and a history of amnesia or focal neurologic deficit--as screening criteria for computed tomography would have saved 58.6% of the scans performed on these patients. Together, these criteria had a sensitivity of 90.1% and a negative predictive value of 98.1% for abnormal computed tomography. Four patients with positive scans would not have been detected using this strategy. None of these patients deteriorated clinically or required operative intervention.
CONCLUSION: Our data suggest that it may be possible to effectively screen patients with head trauma for cranial computed tomography using clinical criteria and so reduce the current number of scans performed by more than half. However, a prospective study is required to confirm our results.