Testing the diagnostic accuracy of common questions for seizure diagnosis: Challenges and future directions
Ellen Snyder,Stefan Sillau,Kelly G Knupp,Jacqueline French,Amber Khanna,Marius Birlea,Kavita Nair,Jacob Pellinen,Kelly G. Knupp
DOI: https://doi.org/10.1016/j.yebeh.2024.109686
IF: 3.337
2024-04-01
Epilepsy & Behavior
Abstract:OBJECTIVE: The aim of this study was to evaluate the diagnostic accuracy of common interview questions used to distinguish a diagnosis of epilepsy from seizure mimics including non-epileptic seizures (NES), migraine, and syncope.METHODS: 200 outpatients were recruited with an established diagnosis of focal epilepsy (n = 50), NES (n = 50), migraine (n = 50), and syncope (n = 50). Patients completed an eight-item, yes-or-no online questionnaire about symptoms related to their events. Sensitivity and specificity were calculated. Using a weighted scoring for the questions alone with baseline characteristics, the overall questionnaire was tested for diagnostic accuracy.RESULTS: Of individual questions, the most sensitive one asked if events are sudden in onset (98 % sensitive for epilepsy (95 % CI: 89 %, 100 %)). The least sensitive question asked if events are stereotyped (46 % sensitive for epilepsy (95 % CI: 32 %, 60 %)). Overall, three of the eight questions showed an association with epilepsy as opposed to mimics. These included questions about "sudden onset" (OR 10.76, 95 % CI: (1.66, 449.21) p = 0.0047), "duration < 5 min" (OR 3.34, 95 % CI: (1.62, 6.89), p = 0.0008), and "duration not > 30 min" (OR 4.44, 95 % CI: (1.94, 11.05), p = <0.0001). When individual seizure mimics were compared to epilepsy, differences in responses were most notable between the epilepsy and migraine patients. Syncope and NES were most similar in responses to epilepsy. The overall weighted questionnaire incorporating patient age and sex produced an area under the ROC curve of 0.80 (95 % CI: 0.74, 0.87)).CONCLUSION: In this study, we examined the ability of common interview questions used by physicians to distinguish between epilepsy and prevalent epilepsy mimics, specifically NES, migraines, and syncope. Using a weighted scoring system for questions, and including age and sex, produced a sensitive and specific predictive model for the diagnosis of epilepsy. In contrast to many prior studies which evaluated either a large number of questions or used methods with difficult practical application, our study is unique in that we tested a small number of easy-to-understand "yes" or "no" questions that can be implemented in most clinical settings by non-specialists.
psychiatry,clinical neurology,behavioral sciences