Reduced EEG montage has a high accuracy in the post cardiac arrest setting

Abstract

ObjectiveTo study if comatose cardiac arrest patients can be assessed with a reduced number of EEG electrodes. Methods110 routine EEGs from 67 consecutive patients, including both hypothermic and normothermic EEGs were retrospectively assessed by three blinded EEG-experts using two different electrode montages. A standard 19-electrode-montage was compared to the reduced version of the same EEGs, down-sampled to six electrodes (F3, T3, P3, F4, T4, P4). We used intra-rater and inter-observer statistics to assess the reliability of the reduced montage for background features and discharges. ResultsThe reduced montage had almost perfect performance for background continuity (κ 0.80–0.88), including identification of highly malignant backgrounds (burst-suppression/suppression) (κ 0.85–0.94) and benign backgrounds (continuous/nearly continuous) (κ 0.85–0.91). We found substantial performance for identifying rhythmic/periodic discharges (κ 0.79–0.86). The reduced montage had high accuracy for assessment of both highly malignant (sensitivity 91–95%, specificity 94–99%) and benign (sensitivity 89–98%, specificity 91–96%) backgrounds, and periodic/rhythmic patterns (sensitivity 79–100%, specificity 89–99%), compared to the full montage. The inter-observer variability was not increased by the reduced montage. ConclusionReduced EEG had high performance for classifying important background and discharge patterns in this post cardiac arrest cohort. SignificanceOur results support the use of reduced EEG-montage for monitoring comatose cardiac arrest patients.