The Reliability of Computer-Processed EEG in the Determination of ECT Seizure Duration

The reliability of single-channel analog EEG and two-channel, computer-processed EEG (cEEG) in determining seizure duration during electroconvulsive therapy (ECT) was studied in 144 consecutive individual treatment sessions of 14 hospitalized patients. Seizure durations determined by post hoc, blind readings of data generated by each method were compared. These estimates of seizure duration were also compared to those determined by the "cuff" method. We found that under our study conditions, the cEEG method was more reliable than the EEG method in two tests of reliability: the cEEG method had fewer readings with a discrepancy of 10 s or more between readers, and had a higher degree of correlation between readers. There was no difference, however, between the cEEG and EEG methods when comparing mean differences and the mean absolute difference between readers within the methods. We also found that the EEG and cEEG methods detected seizure durations that were substantially longer than those detected by the "cuff" method. We conclude that under our study conditions, cEEG was more reliable than analog EEG in certain measures of reliability.