Spectral Entropy Predicts Auditory Recall in Volunteers

Abstract

From a patient's perspective, intraoperative amnesia is an essential component of general anesthesia. Without specific strategies to reduce recall, its incidence is approximately 0.2% in the general surgical population and may be higher in certain subpopulations. We sought to test the validity for predicting recall of a new spectral entropy-based clinical electroencephalogram monitor. We studied 16 volunteers in an unblinded crossover design to assess the correlation of entropy values with behavioral end points during sedation with either propofol or sevoflurane. The end points we considered included word recall, and motor response to verbal command. We also examined the stimulatory effect of verbal commands on electroencephalogram entropy. Logistic regression, receiver operating characteristics, and prediction probability were used to analyze the data. Both State Entropy and Response Entropy were closely correlated with both behavioral end points. The prediction probability of these parameters under a variety of conditions ranged from 0.85 to 0.96. Verbal command to move increased entropy in a dose and drug-dependent fashion. Entropy parameters in this group of young, healthy volunteers appear to be reliable predictors of recall. These results justify extending these studies to additional anesthetics and to surgical patients.