Authors :
Presenting Author: Nathaniel Blanco, PhD – Eysz
Shifteh Sattar, MD, MBA – Rady Childrens Hospital, San Diego
Donald Phillips, MD – CHOC Comprehensive Epilepsy Center & Neurodiagnostic Laboratories
Danilo Bernardo, MD – University of California, San Francisco
Heidi Munger Clary, MD, MPH – Wake Forest University School of Medicine
Gewalin Aungaroon, MD – Cincinnati Children’s Hospital Medical Center
Nan Lin, MD – Peking Union Medical College Hospital
Kelly Kremer, MD – Cincinnati Children’s Hospital Medical Center
Rachel Kuperman, MD – Eysz
Rationale: Absence seizures are generalized non-motor seizures characterized by impaired consciousness and “blank stare
”. The electrographic correlate is generalized 3 Hz spike and wave discharges with abrupt onset and termination typically lasting more than 3 seconds. Although often regarded as highly homogenous, absence seizures show significant variability in both the presentation of epileptiform activity and clinical symptoms among children, and even within the same child. (
Brain. 2020 Aug; 143(8): 2341–2368). Due to this variability, the FDA requires that three independent expert reviewers assess electroencephalograms (EEGs) for seizures in clinical studies, with seizure presence determined by majority agreement. This study investigates agreement between clinicians in EEG interpretations of absence seizures to develop a clearer, shared definition that can be reliably applied in clinical practice and future research.
Methods: In Study 1, 168 patients referred for staring spells or suspected absence seizures, completed a routine video EEG (VEEG) lasting 40 to 60 minutes across 4 clinical study sites. Each VEEG was reviewed by three board certified epileptologists at each site who were instructed to record the start and stop times of each seizure, defined as generalized 3 Hz spike wave discharges with a clinical correlate lasting more than 3 seconds. These annotations were assessed for agreement between reviewers, defined as two reviewers marking seizures during an overlapping period of time. In Study 2, participants underwent guided hyperventilation (HV) while undergoing VEEG. Three epileptologists annotated seizure start and stop times. Based on Study 1, we set a minimum seizure duration of 7 seconds for inclusion in Study 2. Agreement between reviewers was assessed in the same manner as in Study 1.
Results: In Study 1, there were 740 events marked across all reviewers, corresponding to 212 electroclinical seizures (determined by majority agreement of the 3 reviewers). For each seizure event marked by a reviewer, its length and whether at least one of the other two reviewers marked the same time as a seizure were recorded. Figure 1 shows the proportion of agreement between reviewers by event length. Agreement was low for short events increasing and plateauing at over 90% agreement for events lasting more than 6 seconds. Given the low agreement for shorter events, we aimed to determine if a minimum threshold could be set above which experts reliably agree. Figure 2 shows individual reviewer performance with different thresholds for minimum seizure length. Agreement reached a high level for all reviewers at a threshold of ≥ 7 seconds and stayed consistently high thereafter. Data collection for Study 2 is ongoing, but with 20 participants currently, there is 100% agreement between all three reviewers on seizure events 7 seconds or more.
Conclusions: This research demonstrates that VEEG reviewer agreement is low for events shorter than 6 seconds but consistently high for events of 7 seconds or more. The high consistency among reviewers suggests that only a single reviewer may be necessary if a threshold of 7 seconds is adopted.
Funding: Funded by NIH SBIR R43NS119015 and NIH SBIR R43NS129363