Assessing the predictive value of the bispectral index vs patient state index on clinical assessment of sedation in postoperative cardiac surgery patients

Purpose: To compare the depth of sedation determined by Ramsay sedation scale (RSS) with electroencephalogram-based bispectral index (BIS) and patient state index (PSI). Materials and Methods: Fifty mechanically ventilated cardiac surgical patients undergoing propofol and morphine sedation were assessed hourly for up to 6 hours or until tracheal extubation using the BIS, PSI, and RSS. Correlation between RSS, BIS, and PSI was determined, as well as the interrater reliability of RSS, BIS, and PSI. kappa statistics was used to further evaluate the agreement between BIS and PSI. Results: There was positive correlation between BIS and PSI values (rho = 0.73 1, P <.001). The average weighted kappa coefficient was .40 between the BIS and PSI, 0.28 between the RSS and BIS, and 0.16 between the RSS and PSI. Intraclass correlation was consistently higher between the BIS and PSI at all time intervals during the study. Logistic regression modeling over study duration showed that the BIS was consistently better at predicting oversedation (area under the curve, 0.92) than the PSI (area under the curve, 0.78). A comparison of BIS and PSI receiver operating characteristic curves showed that the BIS monitor was a better predictor of oversedation compared with the PSI (P = .02). Conclusions: There is significant positive correlation between the BIS and PSI but poor correlation and poor test agreement between the RSS and BIS as well as RSS and PSI. The BIS is a better predictor of oversedation compared with the PSI. There was no significant difference between the BIS and PSI with respect to the prediction of undersedation. (C) 2009 Elsevier Inc. All rights reserved.