Classification of epileptic seizures: a comparison of two systems

Extensive experience with video-EEG seizure monitoring, notably in the setting of epilepsy surgery programs, has exposed the limited value of the current International Classification of Epileptic Seizures (ICES) for providing relevant localizing information. To overcome this limitation, a Semiologic Seizure Classification (SSC) has recently been proposed. This study aimed to assess and to compare the usefulness and reliability of both systems in the setting of a tertiary epilepsy center.

Three epileptologists independently reviewed video-taped seizures, randomly selected from the archive of the Epilepsy Monitoring Unit. They were blinded to the EEG findings and final diagnosis and classified all seizures according to both classifications.

One hundred thirty-eight seizures from 60 patients (age range, 2-59 years) were reviewed (maximum, three seizures per patient). Fifty-five seizures from 20 patients were recorded in the setting of presurgical evaluations, and the remainder as part of regular diagnostic evaluations. The average interobserver agreement was higher for SSC (63.3%, kappa = 0. 56) than for ICES (38.6%, kappa = 0.41). Some categories of SSC, such as hypermotor or automotor, had the best interobserver agreement, and were strongly correlated with the anatomic localization of the seizures (frontal and temporal lobe, respectively). All reviewers agreed that SSC provided a better description of the seizures than did ICES, in 60% of the patients.

SSC provides a more comprehensive picture of epileptic seizures than does ICES, notably in patients with localized epilepsy syndromes, and appears to be very useful and reliable, particularly in the setting of specialized epilepsy centers.