Can EEG analyses be used to determine sedation levels in ICU?
Xavier Capdevila M.D, Ph.D., J.P. Roustan M.D.
Department of Anaesthesia and Intensive Care Medicine
Lapeyronie University Hospital , Montpellier, France
Sedation is used in more than 50% of patients of intensive care units (ICU) in Europe (1) and in up to 90% in the United States (2). The sedation protocol typically includes the association of a hypnotic drug and continuous IV administration of an opioid (3). In the absence of monitoring of the level of sedation, the dosage of the drugs is determined empirically, exposing the patient to potential overdosing, which can prolong duration of stay in the ICU (4). The monitoring calls for subjective clinical methods based on scales requiring stimulation of patients, ineffective in patients receiving neuromuscular blockers or deeply sedated. Physicians turn to objective instrumental methods, primary among which is the electroencephalogram (EEG). The EEG is a simple, noninvasive technique, but the interpretation of EEG signal is complex. The development of effective mathematical algorithms for signal processing has led to automated EEG interpretation. Fourier transformation is used to calculate the power spectrum: relative power of frequency bands (ß, , , ), 95% spectral edge frequency (SEF95), and 50% spectral edge frequency (SEF50) (5). Administration of anesthetics results in an offset of the power of the signal toward low frequencies (bands and ). The power spectrum can thus be used to monitor this change in the state of the central nervous system during sedation (6). Unfortunately, large interindividual variability and variability of spectral parameters have limited clinical use of spectral analysis of the EEG. Propofol, etomidate and benzodiazepines cause a biphasic change in EEG frequency, with low doses increasing the frequency and higher doses slowing the EEG (7). One frequency value may correspond to two different levels of sedation. The bispectral index (BIS® ) involves a nondimensional index obtained by combination of three EEG subparameters: the degree of burst suppression, the power spectrum, and bispectral analysis of the EEG. The originality of the BIS® lies in this third subparameter. The equation, which combines these subparameters to give the BIS®, was achieved statistically from an extensive bank of EEG records collected from patients subjected to various types of anesthesia. A multivariate regression model was used to establish the best correlation between the BIS values and the clinical level of sedation (8). The Bispectral Index attempts to overcome the EEG biphasic response to provide values between 0 and 100, with conscious sedation dropping the BIS into the 60 to 75 range (9,10). Unfortunately, the BIS® proved to have little usefulness for monitoring the sedation of ICU patients because of a large interindividual variability of BIS® and interference between the BIS® and the electromyogram (EMG) signal (11–15). BIS monitoring in the ICU may be dramatically overestimated because of high muscular activity. The corollary of this overestimation could be a BIS-induced oversedation (15 ). We have recently demonstrated (16 ) that : calculated EEG spectral descriptors exhibited large interindividual variability; bispectral analysis provided the same precision as spectral analysis in distinguishing between the stages of sedation of ICU patients; there was a strong correlation between spectral and bispectral parameters of the EEG, reflecting in some way the redundant nature of the two methods. Correlations between BIS and subjective scales and between BIS values and drug dosage are very low. In special situations such as deep sedation and neuromuscular blockade, in which clinical sedation scales are prone to failure, the bispectral index may help to assess the level of sedation (17). However in Tonner et al study ( 18 ) some of the deeply sedated patients (Ramsay 5 or 6) had BIS readings >80 (BIS, 16%; BIS XP, 13%; p = not significant). The newer algorithm BIS XP did not perform better than the previous version BIS in patients who were mechanically ventilated and sedated . Finally, Leblanc et al ( 19 ) evaluated in nineteen studies the comparison between the BIS and sedation scales, revealing that the BIS trends lower with increasing sedation. The BIS appeared to correlate better when sedation scores were grouped rather than individual values. At present, monitoring sedation with processed EEG parameters cannot generally be recommended
References