Unenhanced emergency cranial CT: optimizing patient selection with univariate and multivariate analyses

Charts from 1,074 consecutive emergency department patients who underwent cranial computed tomography (CT) were reviewed for predictors of a CT abnormality. Twenty-six clinical variables and the results of neurologic examination were compared with cranial CT findings. Patients with focal neurologic deficit, unresponsiveness, and hypertension had an increased risk of a CT abnormality. Blurred vision, trauma, loss of consciousness, headache, and dizziness were each associated with a lower risk of a CT abnormality. Multivariate analysis showed that only focal neurologic deficit and unresponsiveness effectively helped predict a CT abnormality. In patients with negative neurologic findings, only intoxication and amnesia were associated with greater than 10% positive scans and an increased risk for a CT abnormality. The data indicate that positive neurologic findings coupled with intoxication and amnesia would have helped detect 90.7% of the positive scans and provide an effective initial approximation strategy for selecting patients to undergo CT. Although 15 patients with positive scans (1.4%) would have been missed, this strategy would have yielded a negative predictive value of 97.3% and eliminated 53.9% of the CT scans obtained.