Predictors of Recording an Event During Prolonged Inpatient Video Electroencephalogram Monitoring in Children

摘要

BACKGROUND: Distinguishing between seizures and nonepileptic events is a key challenge in pediatric neurology. The diagnostic gold standard is prolonged inpatient video electroencephalogram monitoring. However, little is known about preadmission characteristics that are predictive of recording an event during such monitoring. METHODS: This is a retrospective chart review of children undergoing prolonged inpatient video electroencephalogram monitoring between 2009 and 2012 at a tertiary referral center for the purpose of distinguishing between seizures and nonepileptic events. Demographic information, medical history, event characteristics, and inpatient monitoring course were abstracted. RESULTS: Two-hundred thirteen children were identified. The median recording duration was 25 hours (interquartile range 22.4-48.5), and median time to event of interest (among those with an event recorded) was 4.5 hours (interquartile range 1.4-18.8). An event of interest was recorded in 66% of patients. At the event level, 20% of recorded events were associated with an electroencephalogram correlate, which refers to a change in the pattern seen on the electroencephalogram during a seizure. At the patient level, 112 (79.4%) with events recorded had only nonepileptic events recorded, 25 (17.7%) had only seizures recorded, and 4 (2.8%) had both recorded. Recording an event was predicted by the presence of intellectual disability (P = 0.001), greater preadmission event frequency (P < 0.001), and shorter latency since most recent event (P < 0.001). CONCLUSIONS: Prolonged inpatient electroencephalogram monitoring captured an event of interest in two-thirds of patients, with most of these events captured within less than four and a half hours of recording onset. Several factors predict a greater yield with prolonged inpatient video electroencephalogram monitoring-including event frequency, latency since the most recent event, and the presence of intellectual disability-and can be used to counsel patients regarding this study for the purpose of event capture in the context of shared decision making.