Monitoring depth of anaesthesia is widely used; its clinical evaluation, however, is still controversial. Bispectral index (BIS) is an approved monitor of anaesthetic effect . The exact algorithm of BIS has not been published, but the concept for electroencephalography (EEG) processing to calculate BIS was described by Rampil . A sufficient level of anaesthesia is achieved with BIS values ranging from 40 to 60 .
Entropy describes the irregularity, complexity or unpredictability of a signal. A completely regular signal has an entropy value of zero, the more irregular the signal becomes, the more entropy increases. Recently, spectral entropy of the EEG was introduced as a monitor of anaesthetic depth (S/5™ Entropy Module; Datex-Ohmeda, Helsinki, Finland). The algorithm was published in detail by Viertio-Oja and colleagues . Two different entropy parameters are calculated: state entropy (SE) is computed over the EEG dominant frequency range from 0.8 to 32 Hz primarily reflecting the cortical state. Response entropy (RE) includes both EEG and electromyographic (EMG) components, it is calculated over a frequency range of 0.8-47 Hz. The difference between RE and SE serves as an indicator for EMG activity from the upper facial muscles. According to Viertio-Oja and colleagues , SE is a stable indicator of the effect of hypnotics on the cortex, while RE reacts fast to changes with muscle activation, for example noxious stimuli. The two-parameter entropy might be superior to single number indicators of depth of hypnosis in differentiating between nociception (RE) and hypnotic state (SE) [3,4]. Entropy also reduces EEG processing to a simple number. RE ranges from 100 (awake) to 0 (suppression state of EEG) and SE from 91 to 0 . Decreasing values indicate deepening levels of hypnosis.
The primary objective of this study was to compare two different monitors of the depth of anaesthesia, BIS and spectral entropy in patients undergoing coronary artery bypass grafting (CABG). The secondary objective was the adequacy and quality of anaesthesia.
Patients and methods
Sixty-six patients undergoing first-time elective CABG using cardiopulmonary bypass (CPB) were enrolled in this study. Inclusion criteria were good or only slightly reduced left ventricular function (ejection fraction >40%; left ventricular end-diastolic pressure <15 mmHg) and age below 80 yr. Patients with valvular disease, former CABG, misuse of alcohol or drugs, severe hepatic or renal insufficiency were excluded. The study was approved by the Institutional Ethics Committee (Landesärztekammer Rheinland-Pfalz, Mainz, Germany, No. 837.284.04 (4442)) and all patients gave written informed consent.
All patients were premedicated orally with 1-2 mg of flunitrazepam 1 h before anaesthesia. The patients were prospectively randomized into two groups. In Group BIS 50 (n = 33), the main target was to achieve a BIS of 45-55. Simultaneously, the spectral entropy parameters RE and SE were measured. Induction was performed using a bolus of midazolam (0.07 mg kg−1) and sufentanil (1 μg kg−1). Tracheal intubation was facilitated by pancuronium (0.1 mg kg−1). After intubation, a continuous infusion of sufentanil (0.5-1.5 μg kg−1 h−1) was started. The lungs of all patients were ventilated to normocapnia with oxygen in air (FiO2 0.5). Additional pancuronium (0.03 mg kg−1) was given when necessary. In Group BIS 40 (n = 33), the main target was to achieve a level of BIS of 35-44. Simultaneously, RE and SE were measured. Inductions were performed using a bolus of midazolam (0.1 mg kg−1) and sufentanil (1.5 μg kg−1). Tracheal intubation was facilitated by pancuronium (0.1 mg kg−1). After intubation, a continuous infusion of sufentanil (1.5-2 μg kg−1 h−1) was started. Ventilation and neuromuscular blocking drugs were used as in the other group. In both groups, an additional bolus of midazolam (BIS 50, 0.03-0.07 mg kg−1; BIS 40, 0.05-0.1 mg kg−1) and sufentanil (BIS 50, 0.5-1 μg kg−1; BIS 40, 1-2 μg kg−1) were given if BIS increased above the upper limit of each group for more than 60 s. If BIS did not decrease within the next 60 s below the upper limit of each group, propofol could be given as rescue medication. The choice of the BIS ranges from 45 to 55 and from 35 to 44 and the dosage of the anaesthetic drugs have been previously considered .
Bispectral index (BIS XP, Version 4.0) was measured after skin preparation with antiseptic alcohol and slight rubbing using an S/5™ BIS module (Datex-Ohmeda, Helsinki, Finland) and BIS-quatro Sensor™ (Aspect medical Systems, Natick, MA, USA). BIS was calculated with a smoothing rate of 30 s. When electrode impedance exceeded 10 kΩ, the electrode was replaced and skin preparation was repeated. Electrode impedance was tested repeatedly according to the software protocol of the module. Signal quality was continuously monitored and values of BIS were only recorded when signal quality index was >75. Spectral entropy parameters were measured using an S/5™ entropy module and Entropy Sensor (Datex-Ohmeda, Helsinki, Finland), electrode impedance was kept <7.5 kΩ. The BIS or Entropy sensor were placed on the forehead according to the manufacturer's guidelines. The position on the right or left side of the forehead was decided by flipping a coin. BIS and entropy values were manually recorded by an independent physician who was not involved in the care of the patients. Neuromuscular blockade was assessed by a train-of-four (TOF) stimulation of the ulnar nerve of the left hand (S/5™ NeuroMuscular Transmission Module; Datex-Ohmeda, Helsinki, Finland).
The following data points were defined: T0, awake before induction of anaesthesia; T1, at steady state after induction of anaesthesia; T2, after sternotomy; T3, 30 min after start of cardiopulmonary bypass; T4, 5 min after CPB; T5, at the end of surgery. Arterial plasma levels of cortisol, epinephrine and norepinephrine were measured as stress markers at T0, T2 and T5 using standard laboratory techniques. CPB was performed using mild hypothermia (core temperature 32.5-33.5°C), alpha stat and non-pulsatile flow (2.4 L min−1 m−2). Mean arterial pressure (MAP) was adjusted to 50-80 mmHg using vasopressors (norepinephrine) or vasodilatators (nitroglycerin).
When MAP decreased to less than 60 mmHg and right and left ventricular filling pressures were below 12 mmHg, colloids (hydroxyethylstarch, MW 130 000 Da) were infused. When cardiac index (CI) decreased to less than 2.0 L min−1 m−2 despite adequate volume loading, a continuous infusion of dobutamine (2 μg kg−1 min−1) was started. The dose of dobutamine was increased until CI was greater (2.5 L min−1 m−2). Norepinephrine was used during and after CPB when MAP was <60 mmHg (<50 mmHg during CPB) and systemic vascular resistance (SVR) < 850 dyn s−1 cm−5. Nitroglycerin was used when MAP was >90 mmHg, SVR >1200 dyn s−1 cm−5 and BIS was in the intended range.
After surgery, all patients were transferred to the intensive care unit (ICU). Controlled mechanical ventilation was continued in the ICU. FiO2 and ventilation patterns were adjusted to keep PaO2 between 80 and 120 mmHg and PaCO2 between 38 and 45 mmHg. The tracheal tube was removed when no major blood loss occurred and haemodynamic and respiratory parameters remained stable for at least 30 min. Time from arrival in the ICU until extubation was documented. No fast track procedures were performed.
On the third day after operation, all patients were visited and they were asked to answer a standardized questionnaire (Appendix 1) to measure explicit intraoperative recall . If the patient was unable to answer the questionnaire, the event was noted as having neurological impairment.
A difference of greater than 25% between BIS and SE measurements during anaesthesia was considered clinically significant. For a power of 0.8 and an α value of 0.05, a sample size of 30 patients in each group was calculated to be appropriate.
Data are presented as mean ± standard deviation (SD) unless otherwise stated. The assumption of normality was checked using the Kolmogorov-Smirnov test. Continuous, normally distributed data were compared using two-factorial analysis of variance (ANOVA) for repeated measurements. For significant findings, post hoct-test was applied. When multiple comparisons were made, Bonferroni's test was applied. Continuous, non-normally distributed data were compared using the Wilcoxon signed rank sum test. Binominal data were compared using χ2 analysis and Fisher's exact test. Pearson's correlation coefficients, simple linear regression analyses and forward step-wise regression analyses were used to compare associations between continuous variables. P < 0.05 was considered statistically significant. MedCalc 4.30 software (MedCalc Software, Mariakerke, Belgium) was used.
The groups were similar in age, weight and duration of operation (Table 1). All 66 were ASA III, median New York Heart Association (NYHA) classification was II.
When the premedicated patients were awake before induction of anaesthesia (T0), BIS was 83 ± 7, RE 93 ± 5 and SE 85 ± 4 in Group BIS 50, and BIS was 82 ± 7, RE 92 ± 6 and SE 83 ± 11 in Group BIS 40. In both groups, BIS was within the intended range (BIS 50: 45-55, BIS 40: 35-44) (Fig. 1) during anaesthesia (T1-T5). During anaesthesia, median RE was in a range of 20-26 and significant differences (P < 0.05) between the two groups were found only after induction of anaesthesia (T1) (Fig. 2). SE also fell below the values of BIS and we were not able to distinguish different levels of anaesthesia (T2-T5) (Fig. 3). In both groups, a strong correlation of BIS and RE and of BIS and SE was found for all values (T0-T5) (Table 2). At T0 when the premedicated patients were awake, BIS correlated with SE in both groups. During anaesthesia (T1-T5), BIS correlated with both entropy parameters in Group BIS 40. No correlation of BIS and entropy parameters was found in Group BIS 50 during anaesthesia (T1-T5). The entropy parameters SE and RE were always significantly correlated (Table 2, Figs 4-6). Complete neuromuscular blockade during anaesthesia (T1-T5) was confirmed in all patients (BIS 50, mean TOF ratio <0.01; BIS 40, mean TOF ratio <0.01; n.s.).
Patients in Group BIS 40 received significantly (P < 0.05) more sufentanil (BIS 40, 704 ± 181 μg; BIS 50, 490 ± 107 μg) and midazolam (BIS 40, 18.5 ± 6.1 mg; BIS 50, 15.6 ± 3.8). In Group BIS 40, 21 patients needed propofol as rescue medication and in Group BIS 50, 13 patients (P < 0.05). The amount of propofol was significantly (P < 0.05) higher in Group BIS 40 (BIS 40, 3.5 ± 3.6 mg kg−1; BIS 50, 1.1 ± 1.8 mg kg−1).
Time to extubation was not significantly (P> 0.05) prolonged in Group BIS 40 (15.0 ± 8.7 h) compared to Group BIS 50 (12.7 ± 4.6 h). The number of patients needing prolonged respiratory support (intubation >16 h) did not differ between the two groups (BIS 40, 12 patients; BIS 50, nine patients).
Mild confusion was seen in three patients of Group BIS 40 and in one patient of Group BIS 50. Confusion in all patients was transient and on the third day after operation all patients were able to answer the questionnaire. In the postoperative interview, no significant differences in patient satisfaction and memory (Table 3) were present. In Group BIS 40, one patient reported an intraoperative dream but in neither group was there any sign of explicit intraoperative memory during anaesthesia.
Significantly, more patients (P < 0.05) in Group BIS 40 needed inotropic support with dobutamine (BIS 40, 79%; BIS 50, 52%) to increase CI >2.0 L min−1 m−2 after CPB. The need for norepinephrine during and after CPB did not differ between the two groups (BIS 40, 58%; BIS 50, 67%). Nitroglycerin was significantly (P < 0.05) more often used in Group BIS 50 (BIS 40, 52%; BIS 50, 67%). No patient suffered from a low output syndrome.
The plasma levels of cortisol, epinephrine and norepinephrine are presented in Table 4. Epinephrine and norepinephrine were only analysed before CPB (T0 and T2), as several patients received catecholamines after CPB. Cortisol and norepinephrine increased significantly during anaesthesia in Group BIS 50 while epinephrine did not. In Group BIS 40, no differences from baseline value were found.
In patients undergoing CABG, no relationship was found between BIS levels as well as SE and RE levels at two different stages of a sufentanil-midazolam anaesthetic. Spectral entropy parameters were significantly lower than BIS values and could not be used to differentiate these two levels of anaesthesia. Two questions are of utmost interest:
- Why did BIS measure different levels of anaesthesia while RE and SE did not?
- Did we really achieve two different levels of anaesthesia as indicated by BIS, or was there no difference as indicated by RE and SE?
A dose-response relationship has been found for BIS in several studies using inhalational anaesthesia [7-9]. Changes of end-tidal desflurane concentration between 3 and 9 vol% during deep anaesthesia were adequately represented by BIS in 16 from 21 patients . BIS is dose dependently influenced by opioids and hypnotics [11-13]. Anaesthesia has previously been guided by BIS in 62 patients undergoing CABG . In our study, a deeper BIS-level was achieved by increasing the dose of midazolam and sufentanil, which has also been found in a series of 110 patients undergoing CABG . The dose-dependent behaviour of BIS might be a result of its empirical derivation from a prospectively collected database of EEG and behavioural scales . The multivariate model of BIS accounts for the non-linear stages of EEG activity by allowing different subparameters to dominate the result, i.e. the displayed BIS value .
A clear dose-response relationship for entropy has been found for inhalational anaesthesia [7-9]. For example, SE and RE were found to be useful EEG measures of sevoflurane drug effect , and Shannon entropy and approximate entropy correlated significantly with the desflurane effect compartment concentration [8,9]. In these studies an end-tidal concentration of 0.5-1.6 minimum alveolar concentration of desflurane was used. The ability of SE, RE and BIS to distinguish between the anaesthesia steps awake vs. loss of response, awake vs. anaesthesia, anaesthesia vs. first reaction and anaesthesia vs. extubation was analysed in 20 patients during minor gynaecologic surgery . All patients were anaesthetized with increasing doses of propofol and a fixed dose of remifentanil. The authors concluded that SE, RE and BIS revealed similar information about the level of sedation and distinguished between different stages of anaesthesia. Different levels of anaesthesia were not studied . In 60 patients, anaesthetized with varying doses of propofol and remifentanil, BIS, SE and RE decreased with increasing effect site target concentration of propofol  but BIS decreased more smoothly than SE and RE at deeper levels of sedation. At a depth of hypnosis sufficient for surgery indicated by BIS between 40 and 60, a linear relation was found to RE ranging between 45 and 70  in patients anaesthetized with thiopental, desflurane and nitrous oxide. According to Vakkuri and colleagues , two-parameter entropy should be superior to single number indicators of depth of hypnosis in differentiating between nociception (RE) and lightening hypnotic state (SE). However, Ellerkmann and colleagues  failed to detect any differentiation between nociception (RE) and hypnotic state (SE). The two parameters did not provide any additional information as in the study we are reporting (Fig. 6, Table 2c). A ceiling effect of sufentanil might explain this unexpected result. At a plasma concentration greater than 0.5 ng mL−1, sufentanil exhibited a ceiling effect . We did not measure the plasma concentrations of sufentanil. The poor correlation of SE and RE to different levels of anaesthesia sufficient for surgical painful stimuli might also be explained by experimental work from Sleigh and colleagues . They described spectral entropy as a function of the intensity of synaptic input. At loss of responsiveness, there is an abrupt decrease of spectral entropy because of prolonged inhibitory post-synaptic potentials. However, spectral entropy remains relatively constant with further deepening of anaesthesia. This hypothesis was proven in anaesthetized sheep, when spectral entropy changed minimally with further increase of propofol concentration . The algorithm to calculate spectral entropy used by Sleigh and colleagues, however, differs from the algorithm used by Datex-Ohmeda [3,17].
But did the levels of anaesthesia actually differ in the present study? Monitoring of BIS in patients undergoing cardiac surgery resulted in a reduction of the hypnotic drugs used [5,18]. The cost-saving effect of the reduction in anaesthetic medication was less than the costs for the BIS electrode . Interestingly, the reduction of propofol, or midazolam and sufentanil as guided by BIS, prevented the release of cortisol, epinephrine and norepinephrine [5,18]. In contrast to these results, low dose fentanyl resulted in an increased urinary cortisol excretion and an increased rate of myocardial infarction compared to moderate doses of fentanyl and remifentanil  in 77 CABG patients. In the present study, anaesthesia at a BIS level of 35-44 blunted the release of cortisol, epinephrine and norepinephrine completely. In our patients anaesthetized at a BIS of 45-55, a significant increase of cortisol and norepinephrine was seen. Myocardial ischaemia was not studied. The increased level of endogenous catecholamines may be responsible for the decreased need for dobutamine in the more lightly anaesthetized patients. As the predictive value of BIS and approximate entropy for painful stimuli is low , lightening anaesthesia to the upper limits of BIS, SE or RE can endanger the coronary patient to suffer from awareness, hypertension and myocardial ischaemia.
There is a major limitation of the study presented. The study was not designed to compare BIS and spectral entropy for the prevention of intraoperative awareness. It was designed and powered to compare differences in the values of BIS and spectral entropy. The low incidence of intraoperative awareness of 0.3-2.3% in patients undergoing cardiac surgery [6,18,21] would require a much larger study population to detect any differences in postoperative recall. Nine hundred and twenty-nine patients undergoing cardiac surgery were interviewed after operation for possible intraoperative awareness . The incidence of definite awareness with recall was 0.5%, and the incidence of possible recall was 2.3%. For patients with awareness and recall, a lower dose of midazolam was used . The B-Aware trial included 2463 patients to show that BIS guided anaesthesia reduced the incidence of intraoperative awareness . In the study presented, there was no hint of explicit intraoperative memory; implicit memory was not tested.
In answer to the two questions, we conclude that varying the depth of anaesthesia in patients undergoing CABG resulted in two different levels of anaesthesia indicated by a reduction of anaesthetic medication, a reduction of inotropic support and a partially blunted stress response. Monitoring of BIS allowed us to differentiate these two different levels of anaesthesia. The empirical derivation from a prospectively collected database of BIS  might explain the dose-dependent behaviour in patients anaesthetized with sufentanil and midazolam. No relationship was found between BIS levels and SE and RE. RE and SE were not able to detect different levels of this type of anaesthesia. Spectral entropy as a logarithmic measure of the rate of synaptic interaction might be a sensitive monitor of the transition from consciousness to unconsciousness rather than a progressive indicator of anaesthetic drug effect .
The study was completely financed by the Department of Anaesthesiology and Intensive Care Medicine of the Klinikum der Stadt Ludwigshafen, Germany. No further financial support has been received.
1. Johansen JW, Sebel PS. Development and clinical application of elctroencephalographic bispectrum monitoring. Anesthesiology
2. Rampil IJ. A primer for EEG signal processing in anesthesia. Anesthesiology
3. Viertio-Oja H, Maja V, Sarkela M et al.
Description of the Entropy™ algorithm as applied in the Datex-Ohmeda S/5™ entropy module. Acta Anaesthesiol Scand
4. Vakkuri A, Yli-Hankala A, Talja P et al.
Time-frequency balanced spectral entropy as a measure of anaesthetic drug effect in central nervous system during sevoflurane, propofol, and thiopental anaesthesia. Acta Anaesthesiol Scand
5. Lehmann A, Karzau J, Boldt J, Thaler E, Lang J, Isgro F. Bispectral index-guided anesthesia in patients undergoing aortocoronary bypass grafting. Anesth Analg
6. Dowd NP, Cheng DC, Karski JM, Wong DT, Munro JA, Sandler AN. Intraoperative awareness in fast-track cardiac anesthesia. Anesthesiology
7. Ellerkmann RK, Liermann VM, Alves TM et al.
Spectral entropy and bispectral index as measures of the electroencephalographic effects of sevoflurane. Anesthesiology
8. Bruhn J, Lehmann LE, Ropcke H, Bouillon TW, Hoeft A. Shannon entropy applied to the measurement of the electroencephalographic effects of desflurane. Anesthesiology
9. Bruhn J, Ropcke H, Hoeft A. Approximate entropy as an electroencephalographic measure of anesthetic drug effect during desflurane anaesthesia. Anesthesiology
10. Kreuer S, Bruhn J, Larsen R, Bauer C, Wilhelm W. Comparison of BIS and AAI as measures of anaesthetic drug effect during desflurane-remifentanil anaesthesia. Acta Anaesthesiol Scand
11. Mi WD, Sakai T, Kudo T, Kudo M, Matsuki A. Performance of bispectral index and auditory evoked potential monitors in detecting loss of consciousness during anaesthetic induction with propofol with and without fentanyl. Eur J Anaesthesiol
12. Koitabashi T, Johansen JW, Sebel PS. Remifentanil dose/electroencephalogram bispectral response during combined propofol/regional anesthesia. Anesth Analg
13. Forestier F, Hirschi M, Rouget P et al.
Propofol and sufentanil titration with the bispectral index to provide anesthesia for coronary artery surgery. Anesthesiology
14. Schmidt GN, Bischoff P, Standl T, Hellstern A, Teuber O, Schulte Esch J. Comparative evaluation of the Datex-Ohmeda S/5 entropy module and the bispectral index monitor during propofol-remifentanil anesthesia. Anesthesiology
15. Vanluchene AL, Struys MM, Heyse BE, Mortier EP. Spectral entropy measurement of patient responsiveness during propofol and remifentanil. A comparison with the bispectral index. Br J Anaesth
16. Brunner MD, Braithwaite P, Jhaveri R et al.
MAC reduction of isoflurane by sufentanil. Br J Anaesth
17. Sleigh JW, Steyn-Ross DA, Steyn-Ross ML, Grant C, Ludbrook G. Cortical entropy changes with general anaesthesia: theory and experiment. Physiol Meas
18. Bauer M, Wilhelm W, Kraemer T et al.
Impact of bispectral index monitoring on stress response and propofol consumption in patients undergoing coronary artery bypass surgery. Anesthesiology
19. Myles PS, Hunt JO, Fletcher H et al.
Remifentanil, fentanyl, and cardiac surgery: a double-blinded, randomized, controlled trial of costs and outcomes. Anesth Analg
20. Bruhn J, Bouillon TW, Radulescu L, Hoeft A, Bertaccini E, Shafer SL. Correlation of approximate entropy, bispectral index, and spectral edge frequency 95 (SEF95) with clinical signs of ‘anaesthetic depth’ during coadministration of propofol and remifentanil. Anesthesiology
21. Ranta SO, Herranen P, Hynynen M. Patients' conscious recollections from cardiac anesthesia. J Cardiothorac Vasc Anesth
22. Myles PS, Leslie K, McNeil J, Forbes A, Chan MT. Bispectral index monitoring to prevent awareness during anaesthesia: the B-Aware randomised controlled trial. Lancet
Appendix 1 Standardized questionnaire
On the third day after operation all patients were visited and asked to answer the seven following questions:
- Were you satisfied with the anaesthesia you received?
- (Satisfaction was evaluated using a scoring system ranging from 1 to 6 (1, best; 6, worst)).
- What was your last memory before the operation?
- What was your first memory after the operation?
- What of your anaesthesia was very pleasant?
- What of your anaesthesia did you not like at all?
- If you need anaesthesia again, would you like the type of anaesthesia you had, or do you prefer another type of anaesthesia?
- Did you dream during anaesthesia?