Abstracts

Rationale:
Video-EEG monitoring is the gold standard to classify transient neurological symptoms as epileptic seizures, to identify localizing information of seizure semiology and to document timing and frequency of seizure occurrence. Due to resource limitations, technical and patient-related constraints, the duration of in-patient video-monitoring is, however, limited to days or few weeks. We here assessed the diagnostic yield, i.e. documentation of habitual seizures, during monitoring of patients with ascertained epilepsy at two major European epilepsy centers.
Method:
Electronic documentation of video-EEG monitoring in patients with focal epilepsy monitored in the period of 2005-2019 (Freiburg, UKF) and 2005-2020 (King´s College Hospital London, KCH) was retrospectively reviewed with regard to its yield in documenting habitual seizures. At KCH, electronic reports of video-EEG monitoring were retrieved and extracted with COGSTACK, a platform for medical records. A rule-based natural language processing approach was used to automatically extract information related to seizure presentation during the video-EEG monitoring from the reports. The model was manually validated.  A logistic regression was used to assess the relationship between duration of the recording and seizure occurrence. A subanalysis using Fisher´s exact test was furthermore performed in Freiburg patients on the role of drug reduction during the monitoring period for occurrence of any seizures, and focal and focal to bilateral tonic-clonic seizure types
Results:
Out of 2622 patients monitored at UKF and 2919 patients monitored in KCH, 55.4% and 41.6%, respectively, had seizures documented at the respective sites with mean durations of 4.9 days at UKF and 4.1 days at KCH. At either institution,  a longer duration of recordings was associated with a significantly higher yield in recording a seizure. Drug reduction performed in 71% of patients at UKF increased chances to record seizures but also the risk for the occurrence of bilateral tonic-clonic seizures (all p< 0.001). 
Conclusion:
With the time restrictions of in-patient video-EEG recording, only in 41-55% of patients could habitual seizures be documented,  despite antiepileptic drug reduction associated with an increased risk for occurrence of bilateral tonic-clonic seizures was performed in more than two thirds of patients. The duration of video/scalp EEG monitoring only moderately influenced this yield, leaving a considerable percentage of patient in whom alternative methodologies to document seizure types and seizure frequency are needed.
Funding:
:The project was supported by the European project RADAR-CNS.