Neurological Prognostication Using Raw EEG Patterns and Spectrograms of Frontal EEG in Cardiac Arrest Patients

Abstract

Purpose: We investigated which raw EEG and spectrogram patterns in frontal EEG predict poor neurological outcomes in patients with hypoxic ischemic encephalopathy after cardiac arrest.

Methods: This multicenter, prospective, observational study included 52 patients with anoxic brain injury after cardiac arrest. Raw EEGs and spectrograms (color density spectral arrays) measured with hardwired frontal EEG monitoring were used to predict poor prognosis. Neurological variables upon admission, raw EEG patterns, including highly malignant and malignant EEG patterns, and changes in frequency and amplitude from color density spectral arrays were investigated.

Results: All patients exhibiting highly malignant EEG patterns died, and malignant EEG patterns were significant predictors of poor prognosis as the area under the receiver operating characteristic curve was 0.83 to 0.86. Irregular high-voltage waves in the high-frequency beta band in continuous background EEGs were associated with poor prognosis ( P = 0.022). Malignant EEG patterns including high-voltage and high-frequency beta waves were significantly stronger predictors of poor prognosis than the absence of ventricular fibrillation and pupil reflex, delayed length of anoxic time, and lower Glasgow coma scale score (odds ratio, 9; P = 0.035). Compared with prognostication using malignant EEG patterns alone, the area under the receiver operating characteristic curve of results incorporating high-voltage and high-frequency beta waves was 0.84 (vs. 0.83) at day 1, 0.88 (vs. 0.85) at day 2, 0.92 (vs. 0.86) at day 3, and 0.99 (vs. 0.86) at day 4.

Conclusions: Frontal EEG monitoring is useful for predicting poor neurological outcomes. Brain function monitoring using both raw EEG patterns and color density spectral arrays is more helpful for predicting poor prognosis than raw EEG alone.