Abstracts

Rationale: ESES is an electrographic pattern in which interictal epileptiform activity is augmented by the transition to sleep, with near-continuous lateralized or bilateral epileptiform discharges during the nonrapid eye movement (NREM) sleep state. ESES was first described as the presence of 1.5-3 Hz epileptiform discharges occupying at least 85% of the NREM electroencephalogram tracing. The classic measure used to identify ESES is the spike-wave index (SWI) expressed as a percentage, although there is heterogeneity among authors how this is quantified. In 2017, Weber et al. proposed analyzing the SWI of the first 100 seconds of NREM sleep. Using fourteen long-term monitoring (LTM) studies, they found the SWI from the first 100 seconds of NREM sleep compared favorably with the SWI from the first NREM sleep cycle. The current study aims to measure the reliability of the SWI of the first 100 seconds of sleep as a tool for the diagnosis of ESES

Methods: 140 studies from 60 unique patients met the inclusion criteria for review. Two neurophysiologists blinded to the initial LTM report calculated the SWI of the first 100 seconds of stage II NREM sleep. This was compared to the SWI of the first 5 minutes of NREM sleep and the cumulative SWI of three 5-minute bins of sleep sampled throughout the first NREM sleep cycle. Agreement between the three SWI methods were analyzed using several statistical tools and methods.

Results: Using an SWI of 50% as a diagnostic cut-off, 57% of records had a diagnosis of ESES based on the first 100 seconds of sleep and 54% of records had a diagnosis based on the cumulative SWI of the three bins. This results in a diagnostic accuracy of 92%, sensitivity of 96%, and specificity of 88%. Positive predictive values using the first 100 seconds of sleep, compared to 3 combined bins, was 90% and the negative predictive value was 95%. The type 1 error rate was 12% and the type 2 error rate was 4%. Cohen’s kappa was 0.84, showing strong agreement between the first 100 seconds and 3 combined bins. SWI values in the 0-19% or 81-100% range had perfect diagnostic accuracy when comparing the first 100 seconds to all three bins. Accuracy dropped in the SWI range of 20-80%, resulting in a kappa value of 0.53 showing only moderate agreement between these two methods.

Conclusions: This analysis confirmed the diagnostic accuracy and moderate agreement between calculations using the SWI of the first 100 seconds of sleep and the cumulative total of three 5-minute samples. The 100 second method was particularly accurate in records with SWI values in the 0-19% or 81-100% range. Therefore, an SWI of 0-19% excludes the diagnosis of ESES, while patients with SWIs of 81-100% in the first 100 seconds of sleep may begin treatment in the appropriate clinical context. SWIs in the 20-80% range warrant consideration of a larger sleep sample for definitive diagnosis. Further study will assess treatment implications and pursue validation of the SWI of the 100-second method during routine EEG as a reliable scoring method for ESES.

Funding: Please list any funding that was received in support of this abstract.: This work was supported with internal funding from the Nationwide Children's Hospital Division of Pediatric Neurology (Grant 45141).