Anesth Analg 2003;97:139-144
© 2003 International Anesthesia Research Society
TECHNOLOGY, COMPUTING, AND SIMULATION
ARX-Derived Auditory Evoked Potential Index and Bispectral Index During the Induction of Anesthesia with Propofol and Remifentanil
Gunter N. Schmidt, MD,
Petra Bischoff, MD,
Thomas Standl, MD,
Malte Issleib,
Moritz Voigt, and
Jochen Schulte am Esch, MD
Department of Anesthesiology, University Hospital Eppendorf, Hamburg, Germany
Address correspondence and reprint requests to Gunter N. Schmidt, MD, Department of Anesthesiology, University Hospital Eppendorf, Martinistr. 52, 20246 Hamburg, Germany. Address e-mail to guschmid{at}uke.uni-hamburg.de
 |
Abstract
|
|---|
A new commercial auditory evoked potential (AEP) monitor (A-line AEP monitor) was developed to calculate an index (ARX AEP index; AAI) by automatically using the amplitudes and latencies of the AEP. We investigated 30 patients before spine surgery. AAI; bispectral index (BIS); relative (%) 
, 
, 
, and ß; spectral edge frequency; median frequency; mean arterial blood pressure; heart rate; and oxygen saturation were obtained simultaneously during stepwise (1.0 µg/mL) induction of target-controlled propofol concentration until 5.0 µg/mL, followed by an infusion of 0.3 µg · kg-1 · min-1 of remifentanil. Every minute, the patients were asked to squeeze the observer’s hand. Prediction probability (Pk), receiver operating characteristic, and logistic regression were used to calculate the probability to predict the conditions AWAKE, UNCONSCIOUSNESS (first loss of hand squeeze), and steady-state ANESTHESIA (5.0 µg/mL of propofol and 0.3 µg · kg-1 · min-1 of remifentanil). Although a statistically significant difference among the conditions was observed for AAI, BIS, mean arterial blood pressure, median frequency, and %, only AAI and BIS were able to distinguish UNCONSCIOUSNESS versus AWAKE and ANESTHESIA versus AWAKE with better than Pk = 0.90. The modern electroencephalographic variables AAI and BIS were superior to the classic electroencephalographic and hemodynamic variables to distinguish the observed anesthetic conditions.
IMPLICATIONS: The modern electroencephalographic ARX-derived auditory evoked potential index and the bispectral index were superior to the classic electroencephalographic and hemodynamic variables for predicting anesthetic conditions. Variables derived from the auditory evoked potential did not provide an advantage over variables derived from spontaneous electroencephalogram.
|
Introduction
|
|---|
The measurement of depth of anesthesia is of clinical interest for titrating anesthetic drugs and for avoiding patient awareness during anesthesia. However, assessment of depth of anesthesia is still an unsolved problem because there is no clear definition of what "depth of anesthesia" means exactly. In the last decade, different conditions of anesthesia, such as response to skin incision, loss of consciousness, or drug concentrations, have been represented by different variables, such as hemodynamics (1), esophageal contractility (2), pupillary reflex, and skin conductivity (3). The most reliable variables were derived from the electroencephalogram (EEG). In 1989, Schwilden et al. (4) were able to develop a closed-loop feedback system, using the median frequency (MF) from spontaneous EEG to control propofol anesthesia in humans. The bispectral index (BIS), a modern variable derived from the phase coupling of the spontaneous EEG (5), has been investigated in many different studies. High correlation coefficients between BIS and propofol plasma concentrations and clinical criteria of sedation have been observed (6). However, the use of scalp-recorded EEG variables is still controversial (7,8).
In recent studies, middle latency auditory evoked potentials (MAEPs) have been reported to be superior to the spontaneous EEG for discriminating between consciousness and unconsciousness (9). MAEPs are the EEG response 10–50 ms after predefined auditory stimuli. The spontaneous EEG was eliminated by an averaging technique. In clinical practice, it is sometimes difficult to describe the amplitudes and latencies of the cortical response to the auditory stimuli. For this reason, a new commercial auditory evoked potential (AEP) monitor (A-line AEP monitor; Danmeter) was developed. It works with an algorithm to automatically calculate an index by using the changes in the amplitudes and latencies of the MAEPs (10).
The aim of the study was to find out which variable is best to distinguish between awake and loss of consciousness and between awake and stable anesthetic condition by using propofol and remifentanil. Because of findings showing an advantage of the MAEP over the spontaneous EEG (9), we proposed the hypothesis that the ARX AEP index (AAI), derived from MAEP, is superior to BIS, derived from spontaneous EEG, for this purpose.
|
Methods
|
|---|
After IRB approval and written, informed consent, 30 elective neurosurgical patients were included in the study. Selection criteria were age between 18 and 75 yr, ASA physical status I–II, and spinal surgery. Only patients with no hearing deficit were included in the study.
After premedication with 7.5 mg of midazolam orally (30 min before induction), anesthesia was induced with a target-controlled infusion (TCI) of 1.0 µg/mL of propofol over 2 min (Diprifusor; Graseby 3500; SIMS) followed by a stepwise increase of 1.0 µg/mL every 2 min until a TCI concentration of 5.0 µg/mL was reached. After infusion of 5.0 µg/mL of propofol over 2 min, an additional 0.3 µg · kg-1 · min-1 of remifentanil was administered (Fig. 1). Ten minutes after the start of remifentanil, 0.6 mg/kg of rocuronium bromide was administered to facilitate orotracheal intubation. During the entire study, the patients were asked to squeeze the observer’s hand once a minute. In all patients, responses were documented by the same observer. Study evaluation started at baseline (AWAKE) followed by the very first loss of squeezing the observer’s hand (UNCONSCIOUSNESS). Another evaluated condition was performed after orotracheal intubation under steady-state conditions with propofol TCI concentration of 5.0 µg/mL and remifentanil infusion of 0.3 µg · kg-1 · min-1 (ANESTHESIA). Decreases in heart rate (HR) (<45 bpm) or mean arterial blood pressure (MAP) (<60 mm Hg) were treated with atropine or norepinephrine, respectively. Because of the possibility of interaction with the EEG and hemodynamic variables, patients exhibiting this response were excluded from the study.
View larger version (31K):
[in this window]
[in a new window]
|
Figure 1. ARX auditory evoked potential index (AAI) and bispectral index (BIS) during the induction of propofol/remifentanil anesthesia. Example of a 34-yr-old woman before lumbar spinal surgery. AAI (gray line) and BIS (black line) changes before and during stepwise increases of target-controlled infusion of propofol (gray boxes at the bottom) and during additional infusion of remifentanil (white arrow; start at Minute 9). The black arrows mark the investigated conditions AWAKE, UNCONSCIOUSNESS, and ANESTHESIA.
|
|
EEG was measured from five adhesive silver/silver chloride gel-filled ECG electrodes (Blue-Sensor; Medicotest, Denmark) on carefully prepared skin (Arbo-Prep; Tyco Healthcare, Germany). Electrode placement was performed according to the instructions of the manufacturer of Aspect (BIS; two-channel reference [At1-Fpz and At2-Fpz]; ground Fp2). Electrode impedance was kept <5 k
. The signals were bandpass-filtered between 0.5 and 30 Hz. Bispectral and spectral smoothing rates were 30 s. For artifact detection, "slow rate, suppression, motion, and height frequency" were enabled. BIS (Version 3.3); relative (%; 0.5–3.75 Hz), (%; 4.0–7.75 Hz), (%; 8.0–13.5 Hz), and ß (%ß; 13.75–30.0 Hz); spectral edge frequency; and MF were recorded from an A-1000 EEG monitor (Aspect Medical Systems, Newton, MA). For AAI recording, two silver/silver chloride electrodes (Danmeter, Odense, Denmark) were placed on the forehand, and one was placed behind the ear (A-line AEP monitor; Danmeter). Auditory stimuli were applied by earphones providing an intermittent bilateral click (9 Hz; 2-ms duration; 65-dB sound pressure level). Processing time for the AAI is 30 s for the first signal, and there is a total update delay of 6 s. Struys et al. (11) were able to show that the BIS was unaffected with and without auditory stimuli. All EEG variables were monitored on-line and stored on disk. HR, MAP, and oxygen saturation were measured and registered at each time of measurement (Marquette-Hellige Medical Systems).
Differences between the study conditions were investigated by using a Friedman test (analysis of variance) for repeated measures. In case of significant differences in the analysis of variance, changes were evaluated in detail a posteriori by Wilcoxon’s test. A Bonferroni correction was performed to account for the multiple testing. The probability of predicting (Pk) the conditions UNCONSCIOUSNESS versus AWAKE and ANESTHESIA versus AWAKE was calculated for all variables by using the PKMACRO as described by Smith et al. (12). The area under the receiver operating characteristic (ROC) curve was also used to summarize the accuracy of predicting the dichotomous conditions. The ROC curve plots sensitivity against 1-specificity and reflects the discriminating power of each variable. A value of Pk or ROC = 0.5 means that it predicts the condition not better than a 50:50 chance. A value of Pk or ROC = 1.0 means that the variable predicts the condition correctly 100% of the time. A value <0.5 means that discordance is more likely than concordance. To enable comparison of Pk or ROC, we used 1 - Pk (1 - ROC) when the Pk or ROC value was <0.5 (12). Comparisons of the Pk values between the AAI and BIS were performed with the PKDMACRO (12). Logistic regression was used to analyze the relationship of UNCONSCIOUSNESS versus AWAKE and ANESTHESIA versus AWAKE for AAI and BIS. Statistical analysis was performed with the SPSS package (Version 9; SPSS Inc., Chicago, IL) and PKMACRO (12).
|
Results
|
|---|
Data evaluation was performed for 30 patients (15 women and 15 men; age [mean ± SD], 46 ± 14 yr; height, 175 ± 9 cm; weight, 75 ± 11 kg) with almost artifact-free signal registration. None of these patients was treated with drugs because of hypotension or bradycardia.
Statistically significant differences between the conditions UNCONSCIOUSNESS versus AWAKE and ANESTHESIA versus AWAKE were observed for AAI (43 ± 21 versus 78 ± 14 and 14 ± 3 versus 78 ± 14, respectively), BIS (64 ± 12 versus 94 ± 6 and 30 ± 10 versus 94 ± 6, respectively), MAP (92 ± 13 versus 101 ± 16 mm Hg and 69 ± 12 versus 101 ± 16 mm Hg, respectively), MF (8 ± 5 versus 4 ± 4 Hz and 2 ± 1 versus 4 ± 4 Hz, respectively), and % (26 ± 15 versus 10 ± 10% and 16 ± 7 versus 10 ± 10%, respectively; P < 0.05; Figs. 1 and 2). AAI, BIS, and MAP indicated a significant decrease with deeper anesthesia, and % indicated a statistically significant increase. MF changed significantly but biphasically, with an increase at UNCONSCIOUSNESS and a decrease at ANESTHESIA (Fig. 2).
View larger version (25K):
[in this window]
[in a new window]
|
Figure 2. Variables evaluated for selected study conditions: mean arterial blood pressure (MAP), heart rate (HR), ARX auditory evoked potential index (AAI), bispectral index (BIS), spectral edge frequency (SEF), median frequency (MF), and relative (%) power in , , , and ß during the investigated conditions AWAKE (AW), UNCONSCIOUSNESS (UC), and ANESTHESIA (AN). To demonstrate the scatter of the data, 95th, 90th, 75th, 50th, 25th, 10th, and 5th percentiles are represented. *Statistically significant changes versus AWAKE; Bonferroni corrected; P < 0.05.
|
|
Only AAIand BIS were able to distinguish UNCONSCIOUSNESS versus AWAKE and ANESTHESIA versus AWAKE with a Pk >0.90 (Table 1, Fig. 3). No differences between AAI and BIS were identified for the probability of predicting ANESTHESIA versus AWAKE, whereas the distinction for UNCONSCIOUSNESS versus AWAKE was statistically significantly better for BIS (P < 0.001). The 95% probability that the patients’ loss response to hand squeeze was 29 for AAI and 71 for BIS (Fig. 4).
View larger version (17K):
[in this window]
[in a new window]
|
Figure 4. Probability of UNCONSCIOUSNESS as a function of ARX auditory evoked potential index (AAI) and bispectral index (BIS). Logistic probability response curve for the model of AAI (gray) and BIS (black) for the condition UNCONSCIOUSNESS versus AWAKE. Dotted lines indicate the 95% probability of UNCONSCIOUSNESS for both.
|
|
MAP was better to distinguish ANESTHESIA versus AWAKE (Pk = 0.96) than UNCONSCIOUSNESS versus AWAKE (Pk = 0.69). The 95% probability for ANESTHESIA versus AWAKE for MAP was 72 mm Hg.
|
Discussion
|
|---|
The measurement of depth of anesthesia is still an unsolved problem, because there is no exact definition of what "depth of anesthesia" means. Because a "gold standard" for the measurement of depth of anesthesia is lacking, in this study we defined three conditions of anesthesia: when the patients were awake, lost consciousness, and were under anesthesia. We demanded for a useful variable the ability to differentiate reliably among the three conditions. Whereas the states AWAKE and UNCONSCIOUSNESS are clinical conditions, ANESTHESIA is a drug-dosing regimen-dependent anesthetic condition. We were aware that anesthetic depth could be different between patients during this observed condition. Only AAI, BIS, MAP, %, and MF showed statistically significant differences among all three conditions. These five variables are discussed in detail.
Whereas the brainstem AEPs were constant during anesthesia, the MAEPs seemed to have a dose-response relationship with different anesthetics, such as propofol (2), isoflurane (13), halothane, and enflurane (14). Two studies found that the MAEP was superior to the spontaneous EEG for distinguishing consciousness from unconsciousness during anesthesia (9,15). Struys et al. (11) found no advantage for the first commercially available variable derived from AEP over BIS during anesthesia with propofol. Jensen et al. (10) developed an AEP index by using an autoregressive model with exogenous input (ARX), resulting in an AAI. The advantage of AAI is the rapid extraction of the AEP (15 sweeps over 6 seconds) in comparison with the calculation of the moving time average (250–1000 sweeps over 0.5–2 min) (16) and the automatic exploration of the amplitudes and latencies of the MAEP, resulting in an index from 100 (awake) to 0 (deep anesthesia).
In this study, we found that the AAI was superior to the classic EEG and hemodynamic variables in distinguishing UNCONSCIOUSNESS from AWAKE and in differentiating between ANESTHESIA and AWAKE. The AAI values for the awake patients and the patients who lost consciousness were similar to the results reported by Struys et al. (11). Certainly the AAI showed interindividual differences resulting in a large deviation for the values AWAKE and UNCONSCIOUSNESS. Nevertheless, this study illustrates that the AAI is a useful tool for detecting important conditions of anesthesia.
During the last decade, the BIS, partly calculated by a phase-coupling algorithm, was the most frequently investigated EEG variable derived from spontaneous EEG. BIS has been used as an indicator of sedation with thiopental, propofol, isoflurane, desflurane, enflurane, sevoflurane, and midazolam anesthesia (6,17). However, the use of scalp-recorded EEG variables is still controversial (7,8), especially after the use of ketamine (18). One reason for the good distinction in this study may be the small interindividual difference compared with the other investigated variables. The small variance of the data makes it possible to more accurately differentiate between the conditions. Moreover, the BIS values for the three conditions seem to be reliable because of the good agreement with findings in the literature (6,14,19–21).
Hemodynamic variables such as MAP and HR are used in clinical practice to estimate depth of anesthesia (1). In this study, MAP was superior to all classic EEG variables for the probability of predicting ANESTHESIA versus AWAKE, but MAP was poor at distinguishing UNCONSCIOUSNESS versus AWAKE. Nevertheless, MAP is only an indirect variable to estimate the hypnotic effects mediated by the cardiodepressive impression of propofol and remifentanil. Our results highlight the limitation of MAP for a depth of anesthesia monitor.
Interestingly, in this study, statistically significant increases in % were observed with deepening anesthesia. Regarding this issue, we saw a paradoxical effect in this study (22,23), maybe because of an increase in relaxation during a stepwise increase in propofol until loss of consciousness, accompanied by a decrease with deepening anesthesia. The increase in % may be partly due to the cutoff of low-frequency components, which may result in an increase in higher-frequency components (24). Moreover, % showed large interindividual variation. The SD was 100% of the mean. Thus, % is not reliable for monitoring anesthetic depth in clinical practice.
The MF has been reported to be a useful tool for measuring depth of anesthesia (4). Other studies showed that MF did not correlate with depth of anesthesia (25). In this study, the MF changed to a statistically significant degree but biphasically, with increases at UNCONSCIOUSNESS and decreases at ANESTHESIA. We can only speculate about EEG changes during deepening anesthesia, but our findings confirm that it seems to be difficult to assess depth of anesthesia by using only MF (16).
In summary, in this study with propofol and remifentanil anesthesia, only AAI, BIS, MAP, %, and MF showed statistically significant differences between different anesthetic conditions. Only AAI and BIS were reliable for detecting both when a patient lost consciousness and when the patient entered general anesthesia. AAI and BIS were similar for discriminating between ANESTHESIA versus AWAKE, but BIS performed better for discrimination between UNCONSCIOUSNESS versus AWAKE. Thus, we reject our original hypothesis that AAI (derived from AEP) is superior to variables derived from spontaneous EEG for the monitoring of propofol/remifentanil anesthesia.
|
References
|
|---|
- White PF, Boyle WA. Relationship between hemodynamic and electroencephalographic changes during general anesthesia. Anesth Analg 1989; 68: 177–81.[ISI][Medline]
- Thornton C, Konieczko KM, Knight AB, et al. Effect of propofol on the auditory evoked response and esophageal contractility. Br J Anaesth 1989; 63: 411–7.[Abstract/Free Full Text]
- Goddard GF. A pilot study of the changes of skin electrical conductance in patients undergoing general anesthesia and surgery. Anesthesia 1982; 37: 408–15.
- Schwilden H, Stoeckel H, Schüttler J. Closed-loop feedback control of propofol anesthesia by quantitative EEG analysis in humans. Br J Anaesth 1989; 62: 290–6.[Abstract/Free Full Text]
- Rampil IJ. A primer for EEG signal processing in anesthesia. Anesthesiology 1998; 89: 980–1002.[ISI][Medline]
- Katoh T, Suzuki A, Ikeda K. Electroencephalographic derivatives as a tool for predicting the depth of sedation and anesthesia induced by sevoflurane. Anesthesiology 1998; 88: 642–50.[ISI][Medline]
- Leslie K, Sessler DI, Smith WD, et al. Prediction of movement during propofol/ nitrous oxide anesthesia. Anesthesiology 1996; 84: 52–63.[ISI][Medline]
- Glass PS, Bloom M, Kearse L, et al. Bispectral analysis measures sedation and memory effects of propofol, midazolam, isoflurane, and alfentanil in healthy volunteers. Anesthesiology 1997; 86: 836–47.[ISI][Medline]
- Doi M, Gajraj RJ, Mantzarídis Kenny GNC. Relationship between calculated blood concentration of propofol and electrophysiological variables during emergence from anesthesia: comparison of bispectral index, spectral edge frequency, median frequency and auditory evoked potential index. Br J Anaesth 1997; 78: 180–4.[Abstract/Free Full Text]
- Jensen EW, Lindholm P, Hennenberg S. Auto regressive modeling with exogenous input of auditory evoked potentials to produce an on-line depth of anesthesia index. Methods Inf Med 1996; 35: 256–60.[ISI][Medline]
- Struys MM, Jensen EW, Smith W, et al. Performance of the ARX-derived auditory evoked potential index as an indicator of anesthetic depth: a comparison with bispectral index and hemodynamic measures during propofol administration. Anesthesiology 2002; 96: 803–16.[ISI][Medline]
- Smith WD, Dutton RC, Smith NT. Measuring the performance of anesthetic depth indicators. Anesthesiology 1996; 84: 38–51.[ISI][Medline]
- Jensen EW, Nebot A, Caminal P, Hennenberg SW. Identification of causal relations between hemodynamic variables, auditory evoked potentials and isoflurane by means of fuzzy logic. Br J Anaesth 1999; 82: 25–32.[Abstract/Free Full Text]
- Thornton C, Heneghan CPH, James MFM, Jones JG. Effects of halothane or enflurane with controlled ventilation on auditory evoked potentials. Br J Anaesth 1984; 56: 315–23.[Abstract/Free Full Text]
- Gajraj RJ, Doi H, Mantzaridis H, Kenny GNC. Analysis of the EEG bispectrum, auditory evoked potentials and the EEG power spectrum during repeated transitions from consciousness to unconsciousness. Br J Anaesth 1998; 80: 46–52.[Abstract/Free Full Text]
- Jensen EW, Litvan H. Rapid extraction of middle-latency auditory-evoked potentials. Anesthesiology 2001; 94: 718.
- Sebel PS, Bowles SM, Saini V, Chamoun N. EEG bispectrum predicts movement during thiopental-isoflurane anesthesia. J Clin Monit 1995; 11: 83–91.[ISI][Medline]
- Sakai T, Singh W, Kudo T, Matsuki A. The effect of ketamine on clinical endpoints of hypnosis and EEG variables during propofol infusion. Acta Anaesthesiol Scand 1999; 43: 212–6.[ISI][Medline]
- Liu J, Singh H, White PF. Electroencephalographic bispectral index correlates with intraoperative recall and depth of propofol-induced sedation. Anesth Analg 1997; 84: 185–9.[Abstract]
- Høymork SC, Ræder J, Grismo B, Stehen PA. Bispectral index, predicted and measured drug levels of target-controlled infusion of remifentanil and propofol during laparoscopic cholecystectomy and emergence. Acta Anaesthesiol Scand 2000; 44: 1138–44.[ISI][Medline]
- Schraag S, Bothner U, Gajraj R, et al. The performance of electroencephalogram bispectral index and auditory evoked potential index to predict loss of consciousness during propofol infusion. Anesth Analg 1999; 89: 1311–5.[Abstract/Free Full Text]
- Bischoff P, Scharein E, Schmidt G, et al. Topography of clonidine-induced electroencephalographic changes evaluated by principal component analysis. Anesthesiology 2000; 92: 1545–52.[ISI][Medline]
- Kochs E, Bischoff P, Pichelmayer U, Schulte am Esch J. Surgical stimulation induces changes in brain electrical activity during isoflurane/nitrous oxide anesthesia. Anesthesiology 1994; 80: 1026–34.[ISI][Medline]
- Litscher G, Schwarz G. Is there paradoxical arousal reaction in the EEG subdelta range in patients during anesthesia? J Neurosurg Anesthesiol 1999; 11: 49–52.[ISI][Medline]
- Drummond JC, Brann CA, Perkins DE, Wolfe DE. A comparison of median frequency, spectral edge frequency, a frequency band power ratio, total power, and dominance shift in the determination of depth of anesthesia. Acta Anaesthesiol Scand 1991; 35: 693–9.[ISI][Medline]
Accepted for publication February 21, 2003.
This article has been cited by other articles:
|
|
|
|
 
F. Weber, M. Zimmermann, and T. Bein
The Impact of Acoustic Stimulation on the AEP Monitor/2 Derived Composite Auditory Evoked Potential Index Under Awake and Anesthetized Conditions
Anesth. Analg.,
August 1, 2005;
101(2):
435 - 439.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. Dullenkopf, A. Schmitz, G. Lamesic, M. Weiss, and A. Lang
The Influence of Acupressure on the Monitoring of Acoustic Evoked Potentials in Unsedated Adult Volunteers
Anesth. Analg.,
October 1, 2004;
99(4):
1147 - 1151.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
