Abstracts

Rationale: Continuous bedside EEG monitoring (cEEG) is the gold standard for identifying nonconvulsive status epilepticus. Prolonged monitoring from 24 hours to 72 hours or more is typically required to evaluate for the presence of seizure activity, which results in significant amount of data for a neurophysiologist to review. Quantitative EEG (qEEG) analysis is developed by using mathematical algorithms to extract parameters from the raw data that summarize the important aspects of the EEG and display the data in a user-friendly format. Although extensively used, the predictability of individual qEEG parameters in clinical practice has never been systematically studied.Methods: All patients who underwent continuous bedside EEG monitoring at Cleveland Clinic from 2009 to 2010 and were identified to have electrographic seizures were retrieved from Cleveland Clinic Epilepsy Center EEG database. The patient s electrophysiologic data was processed with Magic Marker (Persyst Development Corporation, Prescott, Arizona) and QP-160AK EEG Trend Program (Nihon Kohden Corporation, Shinjuku-ku Tokyo, Japan). Raw EEGs were identified for seizure activity using standard criteria. A neurophysiologist then compared quantitative EEG parameters interictally and ictally. The changes were scored if there were any temporal or spatial correlation with raw EEG data.Results: 138 patients were identified and analyzed. The sensitivity of each quantitative EEG parameter in detecting seizure activity was as follow: Magic Marker Rhythmic Run Detection and Display (R²D²) 51.18%, Fast Fourier Transform (FFT) Spectrogram 70.31%, Hemispheric Asymmetry Index 68.50%, Hemispheric Asymmetry Spectrogram 55.12%, Amplitude-integrated EEG (aEEG) tracing 68.42%, and Burst Suppression Ratio 26.56%; QP-160AK EEG Trend Program Density Spectral Array (DSA) 77.12%, DSA Asymmetry 67.80%, FFT Power Asymmetry 64.41%, Regional (C3-P3 and C4-P4) aEEG tracing 60.17%, Alpha/Delta Ratio 54.24%, and Burst Suppression Ratio 55.93%. The presence of periodic lateralized epileptiform discharges (PLEDs) or periodic patterns was not associated with significantly increased or decreased yield of each qEEG parameter. The presence of burst suppression was associated with increased seizure detection using Magic Marker s Asymmetry Index, but not with other qEEG trends (p=0.037555). Magic Marker s R²D² was significantly more likely to miss if seizure activities were confined to one region of the brain (p=0.028414). QP-160AK s Alpha/Delta Ratio was significantly less likely to detect seizures of frontal lobe origin (p=0.023113), but has a tendency to identify temporal lobe-onset seizures (p=0.084794).Conclusions: The sensitivity of each qEEG parameter in detecting seizure activity varies from approximately 50-80%. Often trends are combined on montages to sample various strengths in a single recording. Future work will include determining qEEG trend specificity and correlating this data to specific etiologies.