Closed-loop feedback control of methohexital anesthesia by quantitative EEG analysis in humans

A combined pharmacokinetic and pharmacodynamic model of methohexital was used to establish and evaluate feedback control of methohexital anesthesia in 13 volunteers. The median frequency of the EEG power spectrum served as the pharmacodynamic variable constituting feedback. Median frequency values from 2-3 Hz were chosen as the desired EEG level (set-point). In 11 volunteers, the feedback system succeeded in maintaining a satisfactory depth of anesthesia (i.e., unresponsiveness to verbal commands and tactile stimuli). During feedback control, 75% of all measured median frequency values were in the preset range of 2-3 Hz. This distribution of median frequency was obtained by applying random stimulation (six different acoustic and tactile stimuli) to the volunteers approximately every 1.5 min. The decrease of median frequency from baseline to anesthetic values was primarily induced by increasing the fractional power in the frequency band of 0.5-2 Hz from 12.6 +/- 4.5% (mean +/- SD) to 46.0 +/- 2.5%. The median time to recovery (as defined by opening eyes on command) after cessation of the feedback control period was 20.6 min (10.7-44.5 min) when median EEG frequency was 5.2 Hz (4.7-8.4 Hz). The average requirement of methohexital (mean +/- SD) during the 2 h was 1.02 +/- 0.16 g. It is concluded that pharmacokinetic-pharmacodynamic models of intravenous anesthetics established previously may be used to form a suitable background for model-based feedback control of anesthesia by quantitative EEG analysis. This approach gives a possible solution to the problem of adapting pharmacokinetic and pharmacodynamic data to individuals when using population mean data as starting values for drug therapy.