T63. Optimal recording duration of ambulatory EEG

Abstract

Ambulatory electroencephalography (aEEG) is a useful alternative to inpatient video EEG to differentiate epileptic versus nonepileptic attacks (1, 2, 3). It is a cost-effective tool for quantifying seizure activity (4). The diagnostic yield of 24–72 h aEEG has rarely been analyzed (5, 6). Our primary aim is to measure the overall capture rate of aEEG to detect epileptiform discharges or seizures at 1, 2, or 3 days of recording. Our secondary aim is to determine how frequently positive aEEGs leads to a confirmation of diagnosis or change in management at each time interval. We suspect that the yield declines after 2 days of recording. At the Stanford Comprehensive Epilepsy Center, a total of 361 adult aEEG notes were identified and retrospectively analyzed. AEEGs were recorded with 10–20 electrode placement for at least 20 h. If a patient had multiple aEEGs, only the first study was analyzed. We categorized the aEEGs into 3 durations: 1 day (20–30 h), 2 days (30–50 h), and 3 days (50–76 h). For each interval, we determined the proportion of the studies which detected epileptic seizures or discharges. For the secondary aim, we counted positive aEEGs which led to a direct confirmation of diagnosis (i.e., new diagnosis of epilepsy in a patient who was not suspected of having epilepsy or detection of events of interest). We also counted aEEGs which led to direct change in management (i.e., increased or started medications, hospital admission, or changed the presurgical workup). The Cochran-Armitage Linear Trend Test was used to ascertain significant trends. Among 361 consecutive adults who underwent aEEG between 2010–2017, epileptiform discharges or seizures were detected in 101 records (28%). The yield of epileptiform discharges for each time interval was 22%, 17%, and 18% for 1 day (n = 142 studies), 2 days (n = 123), and 3 days (n = 96) respectively. Seizures were detected in 11%, 7%, and 8% for 1 day, 2 days, and 3 days. There was a significantly decreasing trend in detection of newly diagnosed epilepsy (p < 0.02) in longer recordings; 18%, 14%, and 0% for 1 day, 2 days, and 3 days. A positive aEEG directly led to a change in management in 50%, 28%, and 64% of 1 day, 2 days and 3 days, but did not reach statistical significance. Longer recordings were associated with a statistically significant increasing trend in detecting nonepileptic events (p < 0.01) and increasing anti-seizure medications (p < 0.01). They were not associated with starting new anti-seizure medications. This study confirmed that the yield for epileptic discharges or seizures did not increase with longer recordings over 24 h. AEEGs that recorded over 24 h have higher probability of detecting nonepileptic events and are associated with increasing the anti-seizure medication dosages. Limits of the study include the retrospective design and lack of chart review on 260 patients with a negative aEEG.