Abstracts

Rationale: Seizures are among the most common neurological events leading to emergency department (ED) evaluations. A majority (~80%) of ED seizure presentations involve patients with known seizure disorders (‘non-index seizures’) as opposed to new (‘index’) seizures. The value of obtaining ED neuroimaging for non-index seizures is not clear. This study was designed to answer 3 questions: (1) what is the rate of ED neuroimaging for non-index seizures? (2) what clinical factors influence the decision to obtain ED neuroimaging? (3) how often does ED neuroimaging result in a change in care? Methods: We searched the ED database of the Oregon Health & Science University (2010-2012) for patients with a primary seizure diagnosis code. We excluded patients: (1) less than 18 years old; (2) with an index seizure; (3) with a non-index seizure but no seizures in >5 years; (4) with another diagnosis (i.e. syncope). The remaining visits qualified for chart review. For each visit we recorded: (a) characteristics of the seizures and the ED examination (seizure type, associated head injury, status or acute repetitive seizures, focal features, etc); (b) whether neuroimaging was obtained at the ED visit, and if so the results; (c) whether there had been previous neuroimaging, and the results if available. For ED visits with neuroimaging, reviewers determined whether the results: (1) changed the probable seizure etiology; (2) indicated a progression of known CNS pathology; (3) resulted in an acute change in patient management. Statistical analyses utilized Fisher’s exact tests, Wilcoxon tests, and logistic regression. Results: Of 795 ED visits, 477 qualified for study. Of these qualifying visits, 185 (39%) had a neuroimaging procedure as part of the ED evaluation. Of these, 160 (86%) had prior neuroimaging results available, with a median interval between scans of 116 days. Clinical factors associated with obtaining neuroimaging (at p < 0.01) included head injury, prolonged alteration of consciousness, neurological exam focal abnormality, focal features to the seizure, and known CNS lesions including tumor, shunt, and traumatic lesions. Of these factors head injury had the highest odds ratio for obtaining ED neuroimaging (point estimate 14; p < 0.001). Of 185 ED scans, 121 (65%) were abnormal, and 97 (52%) revealed a potentially epileptogenic lesion. However, in only 4 cases (2%) the finding changed the probable seizure etiology. In 4 cases (2%) the finding indicated a progression of known tumor. In 6 cases (3%) the results of ED neuroimaging led to a change in care. However, 2 of these 6 were admitted due to a possible small hemorrhage, later determined to be artifact. Conclusions: These single-center results reveal a high rate of ED neuroimaging for non-index seizures (39%). Multiple clinical factors were associated with obtaining neuroimaging, acute head injury having the greatest influence. The yield of neuroimaging was low with 2% resulting in an acute change in care. Refinement of ED neuroimaging protocols for patients with seizures appear warranted, and could result in significant cost and time savings. Funding: No funding sources.