BISPECTRAL INDEX PREDICTS RATE OF RECOVERY FOLLOWING CORONARY ARTERY BYPASS GRAFT (CABG) SURGERY

Abstract

Introduction. The bispectral (BIS) index successfully predicts the return of patient responsiveness during recovery in non-cardiac surgery [1]. Our objective was to examine the predictive relationships among a series of BIS-related variables and time-to-responsiveness following CABG surgery. Unlike many earlier studies, we examined the relative clinical utility of tracking high and low BIS states. Methods. In 37 patients undergoing uncomplicated first-time CABG surgery by a single surgeon, we retrospectively reviewed the digitized EEG recording for the final 30 minutes of each procedure. The BIS from successive 5-second epochs was used to classify the EEG hypnotic state [2] as 1) below the lower BIS threshold (i.e. loBIS(x) or ultrahypnosis), 2) above the higher BIS threshold (i.e. hiBIS(x) or subhypnosis), and 3) between the two thresholds (i.e., clinically adequate hypnosis). The aggregate minutes spent in each BIS state (i.e., loBIS(x)-min) were automatically calculated from the digitized EEG recordings. Each record was re-examined 27 times, using moving average intervals of 18, 36 or 72 epochs, loBIS thresholds of (x) = 30, 35 or 40 and hiBIS thresholds of (x) = 65, 70 or 75. Patient recovery was measured from discontinuation of anesthetic administration until appropriate response to voice command (checked at 5 min intervals). The relative impact of varying the loBIS and hiBIS thresholds and the moving average interval was then asessed. Results. Significant (P<.05) linear relationships were observed between loBIS-min and recovery time for all 9 combinations of loBIS and moving average interval. Any combination could be used to successfully predict recovery time from aggregate loBIS-min. In contrast, none of the 9 regressions of recovery time on hiBIS-min were significantly different from zero. Variations in the loBIS threshold had a much larger effect on regression parameters than variation in the moving average interval. Regression parameters were insensitive to variations in the moving average interval (i.e., any of the moving average intervals could be used). In contrast, the regression parameters were five times as sensitive to changes in the loBIS threshold (i.e., loBIS(40) vs. loBIS(30). Although all loBIS thresholds predicted recovery time, loBIS(40) was the best discriminant, since it had the largest number of loBIS-min above zero. A scatter plot of recovery time vs. loBIS(40)-min (incorporating hiBIS(65) & 72 epoch averaging) revealed three patient groupings: <5, 5-15 and >15 aggregate BIS-minutes. Minutes to recovery in the three groupings were 45 +/- 31 (mean +/- sd; n=21), 56 +/- 56 (n=6) and 91 +/- 49 (n=10, respectively. Thus, during the final 30 minutes of the case, those patients with <5 aggregate loBIS(40)-min recovered twice as fast (P<.005) as those with >15 loBIS(40)-min. Discussion. These results are consistent with the earlier BIS studies in non-cardiac surgery [1]. Moving averages of loBIS values obtained during the final portion of CABG surgery were able to predict return of patient responsiveness. The process was robust, since it was quite insensitive to the degree of averaging. The concept of loBIS-min provided a simple numeric with which to characterize episodes of ultrahypnosis. It enabled objective quantification of the clinical and economic consequences of producing a very deep hypnotic state. Therefore, loBIS-min may aid in the development of the most efficacious fast-track anesthetic algorithms.