Comparability of Narcotrend™ index and bispectral index during propofol anaesthesia

Abstract

The dimensionless Narcotrend (NCT) index (MonitorTechnik, Germany, version 4.0), from 100 (awake) to 0, is a new index based on electroencephalogram pattern recognition. Transferring guidelines for titrating the Bispectral Index (BIS, Aspect Medical Systems, USA, version XP) to the NCT index depends on their comparability. We compared the relationship between BIS and NCT values during propofol anaesthesia. Eighteen adult patients about to have radical prostatectomy were investigated. An epidural catheter was placed in the lumbar space and electrodes for BIS and NCT were applied as recommended by the manufacturers. After i.v. fentanyl 0.1 mg, anaesthesia was induced with a propofol infusion. After intubation, patients received bupivacaine 0.5% 15 ml via the epidural catheter. Forty-five minutes after induction, the propofol concentration was increased to substantial burst suppression pattern and then decreased. This was done twice in each patient, and BIS and Narcotrend values were recorded at intervals of 5 s. The efficacy of NCT and BIS in predicting consciousness vs unconsciousness was evaluated using the prediction probability (P(K)). We collected 38 629 artefact-free data pairs of BIS and NCT values from the respective 5-s epochs. Because of artefacts, another 5008 epochs had been excluded from data analysis (3855 epochs for the NCT index alone, 245 epochs for the BIS alone and 908 epochs for both indices). Mean (sd) values in awake patients were 94 (6) for Narcotrend and 91 (8) for BIS. With loss of the eyelash reflex, both values were significantly reduced, to 72 (9) for NCT (P<0.001) and to 77 (11) for the BIS index (P<0.001). The PK value for loss of eyelash reflex was similar for BIS (0.95) and NCT (0.93). Decreasing BIS values coincided with decreasing NCT values. A sigmoid model [NCT index=52.8+26.8/(1+exp(-(BIS-78.3)/4.8))(0.4); r=0.52] described the correlation between BIS and NCT index in a BIS range between 100 and 50. For BIS values lower than 50, a second sigmoid model with a correlation of r=0.83 was applied [NCT index=6.6+45.3/(1+exp(-(BIS-29.8)/2.4)) (0.6) r=0.83]. The relationship between burst suppression ratio (BSR) and NCT index was best described by the following sigmoid model: NCT index=265/(1+exp((-BSR+108)/-49); r=0.73. We found a sufficient correlation between BIS and NCT index, but deviations from the line of identity in some ranges require attention. Therefore, a simple 1:1 transfer from BIS to NCT values is not adequate. Our results might serve as a blueprint for the rational translation of BIS into NCT values.