Abstracts

Rationale: Inpatient video-EEG monitoring (LTM) is commonly used for diagnosis of patients who present with recurrent unprovoked spells, for seizure classification, and for presurgical evaluation of patients with medically refractory epilepsy. During LTM, a number of typical events are captured. Some LTM events establish a diagnosis (diagnostic events) while others only serve to confirm this diagnosis (confirmatory events). Some events are unclear (indeterminate) while others are irrelevant (false alarms). How many events do we need to see to be confident in the diagnosis and how long will that take? Methods: To begin to address these questions we retrospectively analyzed video-EEG data from all patients (1478) admitted for LTM between June 1, 1999 and June 1, 2006 at the University of Washington Regional Epilepsy Center. We determined the types of events (diagnostic, confirmatory, indeterminate or irrelevant) and at the times they occurred relative to admission. Each event was further described as non-epileptic, generalized, focal, or epilepsy undefined if the classification of focal or generalized could not be made. For each focal seizure, the site of localization was determined or, if impossible, the seizure was described as non-localized. Results: 606 male (41%) and 872 female (59%) patients were analyzed. There were 1825 LTMs in this time-period as some patients were monitored multiple times. The average number of monitoring days was 6.5 ± 0.1 (median 5). There were 14,214 total events of which 10.8% were diagnostic, 63.1% confirmatory, 23.4% indeterminate, and 2.8% irrelevant (false alarm). Each patient had, on average 9.6 ± 0.6 events (median 5) per LTM admission. 15.4% of the events were non-epileptic, 7.5% were generalized, and 50.7% were focal seizures. Of the focal seizures, 79.7% were localized and 20.3% could not be localized. The first event was usually diagnostic although it was indeterminate or irrelevant 37.9% of the time. The first diagnostic event occurred on average 28 ± 1 hours from admission. The second event was diagnostic 12.8% of the time and occurred on average 57.4 ±5 hours from admission. The third event was diagnostic 7.8% of the time and occurred on average 65.6 ±9 hours from admission. Event four was diagnostic 6.0% of the time and occurred on average 73 ± 5 hours from admission. Events became more frequent as monitoring continued. 95% of the diagnostic events occurred within the first 9 events captured.Conclusions: Long-term video-EEG monitoring is an essential diagnostic tool to confirm diagnosis, classify seizure types, and screen surgical candidates. This study is a first attempt to describe the number of events and the length of time necessary to reach a diagnosis. Most of the diagnostic information is contained within the first 9 events.