Validation of a clinical decision instrument for emergent neuroimaging after a seizure: Let’s image malignancy, intracranial hemorrhage, and trauma (LIMIT)

Abstract

Background: Given the many causes of seizures, emergency physicians often utilize brain computed tomography (CT) to evaluate for intracranial pathology. However, a CT exposes patients to 100 times more radiation than a chest radiograph. Previously, we developed a four-item clinical decision instrument (CDI) to determine which patients with status epilepticus (SE) do not require emergent neuroimaging. In this study, we seek to prospectively validate our CDI in patients with a history of seizures with both SE and generalized tonic–clonic seizures. Methods: This was a prospective observational study of 1,739 consecutive patients who were recruited from two urban hospitals in Philadelphia, Pennsylvania. All patients, 18 years and older, who presented with a chief complaint of seizure and had emergency neuroimaging performed were eligible for inclusion. Patients were excluded from analysis if this was a first-time seizure, had a ventriculoperitoneal shunt, or had focal neurologic deficits. Results: A total of 376 patients were in the final analysis. Of the 376 patients, 10 patients (3%) had positive CTs. Nine of the 10 of the patients were identified by our CDI, resulting in a negative predictive value (NPV) of 99.5%. On secondary analysis, we refined our CDI from four to three criteria: 1) history of intracranial hemorrhage (ICH), 2) active malignancy, and 3) trauma. These criteria also had a NPV of >99% when applied to patients in SE. Conclusion: The validation of our CDI showed improved NPV when compared to the derivation set. Use of the criteria of history of ICH, active malignancy, and trauma could have reduced the use of emergent neuroimaging in our cohort by up to 49%. This CDI should be validated in a larger subset of patients and in multiple centers prior to widespread adoption.