Emergency department neuroimaging for epileptic seizures

Objective: To determine the frequency and yield of neuroimaging in patients with known seizure disorders presenting to the emergency department (ED) with recurrent (nonindex) seizures. Methods: We reviewed 822 consecutive ED visits for nonindex seizures at the Oregon Health & Science University and the VA Portland Health Care System. For each visit, we abstracted details of the clinical presentation, whether neuroimaging was obtained, the results of neuroimaging, and the results of previous neuroimaging studies, when available. We determined whether ED neuroimaging led to an acute change in patient management (yield). Clinical factors associated with obtaining ED neuroimaging, and with the yield of neuroimaging, were evaluated by multivariate logistic regression. Results: A majority (78%) of ED seizure visits were for nonindex seizures. Neuroimaging was obtained in 381 of 822 nonindex seizure visits (46%). Of these, 11 imaging studies (3%) led to an acute change in patient management, 8 (2%) after excluding false-positive scans. Acute head trauma, prolonged alteration of consciousness, and a focal neurologic examination at presentation were associated with an increased yield of ED neuroimaging. Absent any of these 3 clinical factors the true positive yield of neuroimaging was zero. Significance: ED neuroimaging was performed in nearly half of all patients presenting with nonindex seizures. A more conservative use of ED neuroimaging for nonindex seizures, based on clinical factors at presentation, could decrease imaging frequency with minimal loss of yield.