Abstracts

Rationale: Interrater agreement for interpreting EEG varies among board certified clinical neurophysiologists. Electroneurodiagnostic technologists increasingly screen continuous EEG recorded in the critical care setting. However, no published data exist regarding this group’s sensitivity for seizures or other important EEG findings in pediatric critical care. We report interrater agreement for pediatric critical care EEG between individual technicians and technicians and neurophysiologists at a level 4 epilepsy center. Methods: Six EEG technologists (3 ABRET certified) and three board certified clinical neurophysiologists reviewed 90 5-minute segments of video-EEG recorded in the pediatric intensive care unit based on a priori sample size calculation. American Clinical Neurophysiology Society terminology for critical care EEG was utilized (Hirsch et. al, J Clin Neurophysiol 2012;30:1-27). The following variables were scored: EEG background; seizures; seizure onset location; focal slowing; interictal epileptiform discharges; interictal epileptiform discharges location; rhythmic/periodic pattern present; rhythmic/periodic pattern type and pattern location.Fleiss’ kappa was calculated for each variable among technologists and neurophysiologists separately and in respective groups. For the two primary outcomes (EEG background and seizures), pairwise agreement was calculated between each pair of reviewers using Cohen’s kappa. Agreement was classified as poor (0.0-0.199), fair (0.2-0.399), moderate (0.4-0.599), good (0.6-0.799), or very good (0.8-1.0). A ‘gold standard’ assessment based on the consensus decision of three neurophysiologists was used to calculate sensitivity and specificity for each rater on the two primary outcomes. All analyses were run in SAS 9.4 (Cary, NC). Results: Pairwise agreement for EEG background category or presence of seizures was better between neurophysiologists than between individual technologists and neurophysiologists or among technologists with variation (Table 1). Group agreement for neurophysiologists was very good for EEG background and seizures; agreement was good for seizure onset location, interictal epileptiform discharge location and pattern location (p=.0001). Group agreement for technologists was moderate for seizures and seizure onset location (p=.0001).Agreement was best for seizures among technologists (k=0.54, p < .0001) and best for EEG background among neurophysiologists (k=0.846, p < .0001). Sensitivity for seizures was 90% for each neurophysiologist, with specificity of 95-100%. Sensitivity for seizures varied from 52-86% among technologists, with specificity of 85-97%. Conclusions: We report variable among technologists and between technologists and neurophysiologists. Agreement for seizure detection was moderate for technologists and very good for neurophysiologists. However, individual technologist’s agreement with neurophysiologists for seizures varied from fair to good. The variable sensitivity and interrater agreement implies a larger degree of inconsistency in technologists when reading pediatric critical care EEG. Funding: None.