Abstracts

Rationale: Commercially available pre-fabricated EEG templates with a limited number of electrodes offer an opportunity to hasten diagnosis of nonconvulsive seizures; however, their accuracy to detect seizures is unknown. In the present study we examined the accuracy of two electronically-configured limited montage EEG setups for the detection of seizures in hospitalized patients with suspected nonconvulsive seizures. Methods: Thirty previously interpreted EEG segments with seizures and 43 segments without evidence of electrographic seizure activity acquired with standard 21 electrode 10-20 arrays were rendered in a “full” bipolar montage, “neonatal” montage with double electrode distances and “hatband” circumferential montage with 19, 9 and 10 electrodes, respectively. Participants proficient in reading EEG reviewed the clips and determined whether seizures were present. Seizure was defined as rhythmic sharp activity or spike and wave discharges with evolution in frequency, location or morphology. The patterns with pseudo-periodic discharges at 2.5 Hz along the ictal-interictal continuum were excluded. The reviewers also determined if the EEG data provided in the clips with limited number of electrodes were sufficient to make a decision on escalation of clinical care. Results: Three board-certified epileptologists and 2 neurophysiologists completed the survey. The sensitivities to detect seizures using the full rendered, “neonatal” and “hatband” montages were 0.77, 0.85 and 0.68, respectively. The specificities for each montage were 1.0, 0.99 and 0.99, respectively. Inter-rater reliability in diagnosis of seizure was substantial for all three montages (i.e. 0.75, 0.77 and 0.74, respectively). When working with the “neonatal” and “hatband” montages the areas under the receiver operating characteristic curve were 0.92 and 0.84, respectively which suggests that these methods were excellent and good, respectively in distinguishing seizures from non-seizure patterns as defined by standard clinical EEG. Following the review of the seizure-containing clips in “neonatal” and “hatband” montage, participants chose to appropriately escalate care in all cases (100%, CI = 98%-100%) and 94% (CI= 89%-98%) of the time, respectively. Conclusions: The accuracy of the limited electrode configuration in the “neonatal” and “hatband” montages for the detection of seizures was high. The sensitivity of the “neonatal” montage EEG in detecting seizures was superior to that of a “hatband” montage. Funding: No funding