This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Rehberg, B. Articles by Baars, J. H. Search for Related Content PubMed PubMed Citation Articles by Rehberg, B. Articles by Baars, J. H.

---

TECHNOLOGY, COMPUTING, AND STIMULATION

Variability Comparison of the Composite Auditory Evoked Potential Index and the Bispectral Index During Propofol-Fentanyl Anesthesia

From the Department of Anesthesiology, Charité- Universitaetsmedizin Berlin, Berlin, Germany.

Address correspondence and reprint requests to Benno Rehberg, MD, Department of Anesthesiology, Charité Campus Mitte, Schumannstr, D-10117 Berlin, Germany. Address e-mail to benno. rehberg{at}charite.de.

Abstract

BACKGROUND: Monitors of hypnotic depth help anesthesiologists to guide the anesthetic. The performance of different monitors depends on several factors, index variability at a steady state of hypnotic depth being one. We compared the recently introduced AAI1.6 with the established bispectral index (BIS), regarding index variability during stable values of propofol effect-site concentration.

METHODS: After ethics committee approval and written informed consent, anesthesia was performed in 40 patients with propofol as the target controlled infusion and fentanyl. Variability of BIS and AAI1.6 was calculated during periods of constant predicted propofol effect compartment concentration and constant levels of surgical stimulation as the median absolute deviation (MAD) from the median value. A variability index was calculated as 1.48*MAD/(threshold – median value), with threshold being the division line between awake and asleep. Threshold crossing time was used to evaluate the performance in predicting return of consciousness.

RESULTS: Variability index, however, was significantly larger for the AAI1.6, despite similar absolute variability measured as MAD. Lightening of anesthesia before recovery could be noticed earlier using the BIS than the AAI1.6, although consciousness was detected with a significantly higher Pk-value by the AAI1.6.

CONCLUSION: Variability in relation to the difference between the median index value during anesthesia and the threshold necessary to detect consciousness with high sensitivity is higher for the AAI1.6 than for the BIS. This, as well as the steeper concentration–response function found for AAI1.6, impairs the performance of the AAI1.6 in predicting imminent return of consciousness during decreasing propofol concentrations. However, it makes AAI1.6 well suited to detect consciousness when it has occurred.

In recent years, a multitude of monitors of hypnotic depth have been developed, and many studies have investigated and compared their performance and clinical usefulness.

For the anesthesiologist using these monitors to guide the anesthetic, two issues are of general importance besides insensitivity to electrical artifacts and ergonomics. First, the index value displayed by the monitor should vary with the true hypnotic depth. Although there is no "gold standard" for measuring hypnotic depth, this can be assessed by comparing the monitor index with subjective sedation scales such as the Observer Assessment of Alertness and Sedation scale1 or by correlation with anesthetic concentrations2,3 at stable levels of stimulation, assuming that hypnotic depth is the balance between concentration-dependent suppression by the anesthetics and the level of surgical stimulation. Second, the monitor index should separate the anesthetic (unconscious) and the awake state with high sensitivity and specificity.

One factor determining performance related to both issues is the stability of the displayed index at stable hypnotic depth, i.e., stable anesthetic concentration and stable stimulation levels. Recently, we have reported that a monitor index may fluctuate at stable hypnotic depth even when a second monitor connected to the same patient displays a stable index.4 Therefore, there appears to be inherent variability in the index generation of different monitors. Previously, variability of different monitors of hypnotic depth in the awake state has been used to calculate "baseline stability,"5 but variability during anesthesia has not been systematically analyzed to compare different monitors.

Observed variability of a monitor index during anesthesia can be due to real changes in hypnotic depth or due to inherent fluctuations in index calculation at stable hypnotic depth. Inherent variability of a monitor index will not only impair the detection of awareness, but particularly weaken the ability to guide the anesthetic in terms of "lighter" or "deeper" anesthesia. For that reason, we propose a variability index as an additional comparator in the assessment of the different monitors of hypnotic depth available.

Here, this variability index was used to compare two indices of depth of hypnosis utilizing different electrophysiological signals: the bispectral index (BIS) using the spontaneous electroencephalogram (EEG), and the recently introduced AAI1.6_1.66 index derived from both the spontaneous EEG and middle-latency auditory evoked potentials (AEP).

On the basis of our previous study, we hypothesized that the AAI1.6 would exhibit a larger variability index than the BIS, leading to a higher incidence of spurious data beyond the threshold of adequate hypnosis and thus falsely indicating "too light" hypnosis.

Higher variability will also impair the detection of trends in hypnotic depth, such as the lightening of hypnotic depth when administration of the anesthetic drug has been stopped. This predictive performance can be measured as the time from the point when the parameter increases above the "noise" created by the variability (i.e., crosses a threshold defined by the variability) until the actual return of consciousness (threshold crossing time). If AAI1.6 displays larger variability, we hypothesized that it should have a lower predictive performance (i.e., threshold crossing time should be shorter).

The study was thus designed to ensure (1) periods of stable "deep" hypnosis during surgery with a minimum chance of awareness to measure variability and (2) a controlled recovery phase with steadily decreasing hypnotic depth to analyze threshold crossing times.

METHODS

Study Design
After IRB approval (Charité Campus Mitte, Berlin, Germany) and written informed consent, 40 patients, with an ASA physical status I–III, scheduled for urological or gynecological surgery were enrolled in the study. Patients suffering from neurological or psychiatric disease, severe cardiac and vascular symptoms, hearing impairment or patients needing a rapid-sequence induction were excluded from the study.

Midazolam (0.1 mg/kg) was administered p.o. as syrup 1 h before surgery according to the standards of the department.

Total IV anesthesia with propofol and fentanyl was performed using a target-controlled infusion for propofol (Base Primea®, Fresenius Medical Care, Brezins, France) and effect-site targeting (Schnider model7). Infusion rates were downloaded directly to computer hard disk, and dosing of fentanyl was recorded exactly for post hoc recalculation of the concentration-time profile. Propofol effect-site concentration was kept at induction between 7 and 10 µg/mL and during surgery in the range of 2.5–4.0 µg/mL to assure unconsciousness with high probability, adjusted to clinical needs within these limits. Clinical needs refer to the apparent sensitivity of the patient to anesthetic effect as measured by standard clinical variables such as heart rate, arterial blood pressure, sweating and tears, and their response to painful stimulation (monitoring of hypnotic depth is not used routinely at our institution). Fentanyl was given as bolus of 0.05–0.15 mg as necessary until 30 min before the anticipated end of surgery, judged by clinical signs. During the last 15 min before the end of surgery, the propofol effect-site concentration should not be increased. Patients were excluded post hoc, if the propofol target concentration was increased during the last 20 min before the actual return of consciousness.

Ventilation was performed either via a laryngeal mask airway or an endotracheal tube. Cisatracurium (0.1–0.15 mg/kg) was used as muscle relaxant only for tracheal intubation so that no neuromuscular blockade was present at the end of surgery.

At the end of surgery, the propofol infusion was terminated. Return of consciousness (termed here "recovery") was recorded by one of the authors (C.R.) when patients were able to open their eyes to verbal command (repeated every 10 s).

Postoperatively (12–24 h after end of surgery), patients were interviewed for intraoperative awareness using a structured interview.8 Questions concerned the last memory before induction and first memory after return of consciousness, memory of events in between, and dreams during anesthesia.

EEG Data Collection
Before anesthesia, self-adhesive electrodes for monitoring BIS (Quattro sensor, Aspect Medical Systems, Newton, MA), and the AAI1.6 (Medicotest "blue point," Istykke, Denmark) were attached to the patient’s forehead. The frontal electrode of the AAI1.6 monitor (Danmeter, Odense, Denmark) was placed cranial of, but tightly adjacent to, the BIS electrode. Previous experience has shown that it makes no difference which electrode is cranial to the other if the distance is small. Impedance was kept below 5 kOhm. In addition, the A-line headphone was placed on the patient’s head according to manufacturer instructions.

Both monitors were placed away from the anesthesiologists’ workplace; anesthesiologists were blinded to the monitor readings.

BIS (BIS XP version 3.1) was recorded in 5 s intervals via serial interface and Microsoft "Hyperterminal" software onto hard disk. The smoothing interval was set to 15 s. The A-line AEP monitor was equipped with a special flash memory to record AAI1.6 values in 1 s intervals. Both BIS and AAI1.6 were later downsampled to 10 s intervals for further analysis using timepoint-centered averaging.

Variability Assessment
The distribution of index values around the average value during a period of stable hypnotic depth is normally distributed for values close to the average value; however, the distribution has longer tails including some outliers especially to the right. To account for this nonideal distribution, we prefer to use the median value and the median absolute deviation from the median value (MAD) to describe the variability

where x_ij is the jth index measurement in the ith patient and median(x_i) is the median of all index measurements in the ith patient at a particular level of hypnotic depth. The MAD is more "robust" than the parameters derived from the squared errors such as standard deviation.9

For better comparability of this variability parameter between different indices of hypnotic depth, it should be normalized. Normalization could be performed by using the median of the index measurements as the denominator, similar to the performance error parameter introduced to evaluate the performance of computer-assisted infusion devices.10 However, the main clinical relevance of a variability parameter would be to show the stability of an index value against crossing the threshold values that separate anesthesia from sedated states just because of index variance. Therefore, normalization of the variability parameter should reflect the relation between the amount of variability and the distance between median value of the index and anesthetic threshold value. By multiplying MAD by the factor of 1.48, the index can additionally be interpreted as a fraction of the standard deviation of the signal (assuming normal distribution). An index value of 0.5 would mean that the threshold is 2 standard deviations "away" from the median value, an index value of 0.25 indicate 3 standard deviations. Thus, we propose the following variability index:

As threshold, we used the upper limits recommended by the respective manufacturers for surgical anesthesia. For the BIS, this is a value of 60, and for the AAI_1.6, a value of 25.

We based our analysis on intraoperative data from intervals (at least 10 min long, one per patient) where (a) predicted propofol effect compartment concentration did not change more than 5%, (b) no fentanyl had been given within the last 15 min, and (c) the level of surgical stimulation was constant (e.g., laparoscopic preparation or skin closure).

To characterize the clinical implication of the variability index, we analyzed in how many patients BIS and AAI1.6 cross the manufacturer recommended threshold during these intervals for longer than 30 s (assuming that anesthesiologists would be alarmed after 30 s and increase anesthetic concentration).

Prediction of Recovery
To analyze the effect of index variability on performance of the monitors in the recovery phase, we performed calculations based on data from the last 20 min of anesthesia before the individual time of recovery (eye opening on verbal command) and 2 min after this time point. Since it has been pointed out that the signal of the AAI1.6 between values of 60 and 100 behaves as random noise,6 we performed all further calculations with both the AAI1.6 restricted to 60 and with the full scale to 100.

The performance in separating the awake state (data points after time of recovery) and the unconscious state (data points before time of recovery) was assessed using the prediction probability Pk, calculated using the EXCEL macros developed by Smith et al.,11 which also allow analysis of differences between Pk-values. Pk-analysis assumes independent data; therefore, data were reduced to data points every 60 s.

The predictive performance of an index of hypnotic depth (e.g., BIS) to anticipate the recovery from anesthesia can be expressed as the time from the point the anesthesiologist detects decreasing depth of hypnosis until the observed return of consciousness. Decreasing depth of hypnosis can be detected when the hypnotic depth index increases above the "noise" level caused by the variability of the index during steady-state anesthesia.

The threshold for detecting decreasing depth of hypnosis can be mathematically expressed using the standard deviation of the data or, alternatively, the MAD. We used the value 1.48*MAD, which would be one standard deviation in normally distributed data. Crossing of a threshold was arbitrarily defined as the increase of the parameter above the threshold after which no further decrease below the threshold for longer than 30 s occurred. The predictive performance of the indices of hypnotic depth to anticipate the recovery from anesthesia was thus measured as the time from crossing these parameter-specific thresholds until the observed return of consciousness (threshold crossing time).

Since threshold crossing times may not only depend on index variability, but also on the steepness of the concentration–response curves, we also calculated concentration–response curves for both AAI1.6 and BIS using pharmacokinetic–pharmacodynamic modeling.

Pharmacokinetic and Pharmacodynamic Modeling
Plasma concentrations of propofol were calculated post hoc, using the STANPUMP program (freely available from the author, S. Shafer, at his website at Stanford university: http://anesthesia.stanford.edu/pkpd/), the Schnider7 parameter set, and the infusion rate profiles downloaded from the infusion pump.

Individual concentration–response functions were fitted to the data on a spreadsheet program (EXCEL, Microsoft, Redmond, WA), using a sigmoid model:

In this model, E₀ is the baseline effect, c_eff is the apparent effect-site concentration, EC₅₀ is the concentration that causes 50% of the maximum effect, and describes the slope of the concentration–response relation. The time lag between changes in calculated plasma concentration and observed effect was modeled by an effect compartment and a first-order rate constant determining the efflux from the effect compartment k_e0.

Data of the induction period until 10 min from start of the propofol infusion were not included in the analysis.

Statistical Analysis
All other statistical comparisons were made using GraphPad Prism (Prism 3.0, Graphpad Software, San Diego, CA). The significance level was set at P = 0.05.

Sample size was determined assuming a medium effect size (Cohen’s d = 0.5), an error of 0.05, and a power of 80%, yielding a necessary total sample size of 27 (G-power calculation program 3.0.3, University of Kiel, Germany). Effect size was estimated based on variability index determination of data from a previous study at low anesthetic concentrations.4 To allow for the possible imprecision due to the difference in study design, a slightly higher actual sample size was used.

RESULTS

Of the 40 patients enrolled in the study, four were excluded post hoc because the propofol concentration was increased during the last 20 min before return of consciousness. The population characteristics of the remaining 36 patients were: 29 women and seven men, age 55 ± 14 yr (mean ± sd), weight 72 ± 15 kg, and height 168 ± 8 cm.

The median propofol effect-site concentration during surgery was 3.2 (2.9–3.5, 25%–75% percentiles) µg/mL, resulting in median AAI1.6 and BIS values of 15 (11–20) and 35 (26–40) during surgery, respectively.

No patient reported intraoperative awareness in the postoperative structured interview.

An example of the variability of the two indices during intervals of constant propofol effect-site concentration and constant level of stimulation is shown in Figure 1. Absolute levels of variability calculated as the MAD (Fig. 2A) were not significantly different between the two indices (Wilcoxon matched-pairs test). The variability index normalized to the difference between manufacturer-recommended thresholds and median values, however, was significantly smaller for the BIS (Fig. 2B, Wilcoxon matched-pairs test, P < 0.001).

View larger version (7K): [in this window] [in a new window]   Figure 1. Example of the variability over time of the bispectral index (BIS) (A) and AAI1.6 (B) during periods of constant levels of surgical stimulation and almost constant effect-site concentrations of propofol. BIS and AAI1.6 data are from the same patient. The median value of the respective index (26.2 for the BIS and 18.0 for AAI1.6) during the analysis period is shown as the thin horizontal line. Thick lines denote the threshold values recommended by the manufacturers below which awareness is unlikely. In this example, the median absolute deviation from the median value (MAD) is 0.8 for the BIS and 3 for the AAI1.6. The proposed variability index relates this number to the difference between the thin and the thick lines (threshold – medianxi), thus emphasizing the clinical significance of the variability. In this case, the variability index is 0.03 for the BIS and 0.6 for the AAI1.6 (more than 10-fold larger), denoting that the AAI1.6-threshold is almost within one standard deviation of the current AAI1.6 value. Using the AAI1.6, an anesthesiologist would be tempted to increase anesthetic concentration due to the variability (which leads to intermittent values above the threshold), although the median AAI1.6 of 18 indicates rather deep hypnosis. The BIS of the same patient gives the same information of deep hypnosis, but with minimal variability. It should be kept in mind that the true hypnotic depth was not measured in this study.

View larger version (13K): [in this window] [in a new window]   Figure 2. Variability parameters of the bispectral index (BIS) and AAI1.6 for individual patients during periods of stable anesthetic concentrations and constant surgical stimulation (example shown in Fig. 1). Median absolute deviation (MAD) from the median index value is shown in A and the variability index calculated as 1.48*MAD/(threshold – median) is shown in B. Filled diamonds indicate the interpatient median.

In 8 of the 36 patients analyzed, AAI1.6 was above the manufacturer recommended range for intraoperative anesthesia of 25 for longer than 30 s during the measurement interval of stable deep hypnosis. In none of the patients was BIS above 60 for longer than 30 s during the measurement interval.

After the end of surgery, the AAI1.6 remained at a low level for most patients until about 2 min before recovery (eye opening on verbal command), when a sudden increase to values above 60 occurred in most patients. In contrast, the BIS started to increase far earlier in advance of recovery. The propofol concentration during the last 20 min before surgery monotonically decreased. The median duration from stop of the propofol infusion at the end of surgery until recovery was 16 (11–22) min.

The AAI1.6 separated the data points after recovery (eye opening on verbal command) from those in the last 20 min before recovery with a prediction probability of 0.944 (se 0.005). This Pk-value was significantly higher (paired t-test calculated from the jackknife analysis) than the prediction probability of the BIS (0.909, se 0.007).

The "threshold crossing times," i.e., the time differences between the crossing of the variability threshold value (threshold = median + (1.48*MAD), which indicates that "decreasing depth of hypnosis" becomes apparent for the anesthesiologist, and the individual return of consciousness is shown in Figure 3.

View larger version (12K): [in this window] [in a new window]   Figure 3. Performance of bispectral index (BIS) and AAI1.6 in identifying decreasing hypnotic depth after stopping the propofol infusion, measured as the threshold crossing time. A, Time course of BIS (upward triangles) and AAI1.6 (crosshairs) during the last 20 min of anesthesia before return of consciousness (dotted vertical line) to illustrate the calculation of the threshold crossing time. Propofol infusion was stopped at 35 min (thin vertical line), and median BIS/AAI1.6 values were calculated for the preceding 10 min interval of stable anesthetic concentration (thin horizontal lines). Thresholds for BIS/AAI1.6 (thick horizontal lines) were calculated as median + 1.48*MAD, denoting the point where BIS/AAI1.6 deviate more than one standard deviation from the median value. Threshold crossing time is calculated as the time difference between time of threshold crossing (dashed vertical lines) and return of consciousness (dotted vertical line). B, Individual threshold crossing times. Open circles are data from individual patients, black diamonds indicate the median values.

Threshold crossing times were longer for the BIS (Wilcoxon matched pairs test), which implies that recovery can be anticipated earlier with the BIS than with AAI1.6. Median threshold crossing times (with 25%–75% percentiles) were 480(190–719) s for the AAI1.6 and 740(480–940) s for the BIS.

Calculated concentration–response functions for BIS and AAI1.6, which may also influence threshold crossing times, are shown in Figure 4.

View larger version (24K): [in this window] [in a new window]   Figure 4. Concentration–response functions of (A) the bispectral index (BIS) and (B) the AAI1.6. Effect-site concentrations were calculated with individual fits for each patient and for AAI and BIS separately. Thin lines are the fits of a negative sigmoid Emax model to the data shown. Successful fitting of the concentration–response data was possible in 32 patients for BIS data and 28 patients for the AAI data.

Using the Pk statistic to describe the nonlinear correlation of the monitor value and the propofol effect-site concentration (absolute values were used), the BIS (Pk = 0.731, se = 0.001) correlated with propofol effect-site concentration only slightly, but significantly better than the AAI1.6 (Pk = 0.711, se = 0.002). Since it had been suggested by Vereecke et al.6 that the AAI1.6 correlates better with propofol effect-site concentration when the upper limit of the scale is reduced to 60, we also included an analysis of this AAI1.6₆₀. However, the Pk-values for prediction of propofol effect-site concentration as well as for predicting return of consciousness were identical for both AAI scales.

The median parameters of the individual concentration– response fits to the data are given in Table 1. A significant difference was found between the slope parameter for BIS and both AAI1.6 variants (Kruskal– Wallis-test, Dunn’s post test), being smaller for the BIS. For EC₅₀ and k_e0 no significant differences were found.

DISCUSSION

In this study, we analyzed the magnitude and the consequences of signal variability of two indices of hypnotic depth, BIS and AAI1.6, during stable levels of propofol-fentanyl anesthesia. We had chosen rather high concentrations of propofol to ensure that variability was not due to intermittent fluctuations in the state of consciousness. During most measurement periods, suppression of the EEG was just above the level where burst suppression occurs.

Although absolute variability, measured as the MAD from the median value, is not significantly different between BIS and AAI1.6, variability is higher for the AAI1.6 when compared with the difference between median index values during anesthesia and the threshold recommended by the manufacturer. The threshold, however, is necessary to detect consciousness with high sensitivity.

In 8 of the 36, patients the higher variability of the AAI1.6 led to AAI values above 25 for longer than 30 s during intervals of otherwise stable deep hypnosis. This would have probably caused the anesthesiologist to increase anesthetic concentration. Since there is no gold standard to assess the true level of consciousness, we cannot exclude that these increases were caused by changes in the underlying state of consciousness not affecting the BIS. However, because of the otherwise deep level of anesthesia, this appears to be unlikely. Similar effects may also lead to possible false decisions in the opposite direction, although we did not measure this due to the already deep baseline level of hypnosis.

The higher variability of the AAI versus propofol effect-site concentration compared with BIS has been reported in the literature for the previous version of the AAI,12 which relied entirely on the AEP signal, and also for another AEP index developed by Doi et al.13

It has been suggested that monitors using AEP perform better in detecting awareness than those using the EEG,14 but one study has shown that a combination of both signals may yield the best performance.15 The AAI1.6 includes, apart from the AEP, β-ratio and EEG burst suppression rate; however, the EEG parameters are mainly used when the signal-to-noise ratio of the AEP is low.6

A higher relative variability should impair the performance of an index in helping the anesthesiologist decide whether depth of hypnosis is becoming deeper or lighter. Here, we studied the course of the monitor indices from the time of the end of the propofol infusion until patient recovery. The impeding recovery of the patients, measured as crossing a threshold one standard deviation above the median value in the preceding 10 min of anesthesia, became apparent earlier when using the BIS than with the AAI1.6. Variability of the signal, however, may not be the only explanation for this result, since the AAI1.6 also displays concentration–response curves with a higher steepness coefficient, which imply less change of the AAI1.6 at higher concentrations of propofol.

The calculation of concentration–response functions in our study is limited by the fact that we had to exclude data from the induction period, because rapid induction alters propofol kinetics7 and monitor indices may not correlate well with concentration at sudden changes in hypnotic depth because of time delays of index presentation.16 However, our data are complementary to those of Vereecke et al.6 and Jensen et al.,17 who obtained data only during slowly increasing propofol concentrations. Interestingly, the concentration–response function for the AAI1.6 in our study was significantly steeper than for the BIS. This difference in the results of Vereecke et al. may have been due to the effect of residual fentanyl concentrations present at the end of anesthesia, different levels of nociceptive input, or differences between loss of consciousness and return of consciousness.

We did not attempt to elucidate the origin of the index variability. In our study design, dosing of propofol and fentanyl was based on subjective criteria, which increases interindividual variability. However, median BIS was below 40 for almost all of the patients, indicating a stable deep level of hypnosis.

Variability of the BIS may correlate with the level of nociceptive input,18 and the same may be true for the AAI1.6. Although we analyzed only periods of apparently stable and low levels of surgical stimulation, such as laparoscopic preparation or skin closure, minor variations cannot be excluded, and individual patients might react differently to surgical stimulation. Variability may be different at different levels of surgical stimulation and/or different levels of opioid concentration.

Another potential source of the variability are muscle potentials. Since we did not control for muscle relaxation, we cannot exclude this possibility.

A certain variability may also be inherent to index calculation, since simultaneous recordings from two BIS monitors attached to the same patient can display differences.19 There has been no study with simultaneous recordings of the AAI1.6 from the same patient.

In contrast to the BIS, which was developed by combination of different subparameters using a database approach to correlate with the concentration of propofol, the AAI1.6 is directly based on physiological variables (albeit also in combination). At least the AEP analysis used by this monitor was developed to react rapidly to changes in consciousness20 with the ability to extract information for the AEP within 2–6 s,6 possibly at the cost of stability of the signal. When monitors of hypnotic depth are used for the surveillance of unconsciousness, this is a worthwhile goal. Monitor indices with higher inherent variability, however, will be less suited for the development of closed-loop systems, and will be less suited to help the anesthesiologist decide how deep anesthesia, in terms of propofol effect-site concentration, actually is.

In summary, we have shown that different indices of hypnotic depth can display quite different variability when variability is viewed in relation to the index range below the consciousness threshold.

ACKNOWLEDGMENTS

We thank all participating colleagues and nurses as well as the recruited patients for their support. The target controlled infusion system was kindly supplied by Fresenius-Kabi, Bad Homburg, Germany, and the A-line monitor by Danmeter, Odense, Denmark.

Footnotes

Accepted for publication January 31, 2008.

REFERENCES

1. 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 "anesthetic depth" during coadministration of propofol and remifentanil. Anesthesiology 2003;98:621–7[Web of Science][Medline]
2. Billard V, Gambus PL, Chamoun N, Stanski DR, Shafer SL. A comparison of spectral edge, delta power, and bispectral index as EEG measures of alfentanil, propofol, and midazolam drug effect. Clin Pharmacol Ther 1997;61:45–58[Web of Science][Medline]
3. Vanluchene AL, Vereecke H, Thas O, Mortier EP, Shafer SL, Struys MM. Spectral entropy as an electroencephalographic measure of anesthetic drug effect: a comparison with bispectral index and processed midlatency auditory evoked response. Anesthesiology 2004;101:34–42[Web of Science][Medline]
4. Hadzidiakos D, Petersen S, Baars J, Herold K, Rehberg B. Comparison of a new composite index based on midlatency auditory evoked potentials and electroencephalographic parameters with bispectral index (BIS) during moderate propofol sedation. Eur J Anaesthesiol 2006;1–6
5. Bruhn J, Bouillon TW, Hoeft A, Shafer SL. Artifact robustness, inter- and intraindividual baseline stability, and rational EEG parameter selection. Anesthesiology 2002;96:54–9[Web of Science][Medline]
6. Vereecke HE, Vasquez PM, Jensen EW, Thas O, Vandenbroecke R, Mortier EP, Struys MM. New composite index based on midlatency auditory evoked potential and electroencephalographic parameters to optimize correlation with propofol effect site concentration: comparison with bispectral index and solitary used fast extracting auditory evoked potential index. Anesthesiology 2005;103:500–7[Web of Science][Medline]
7. Schnider TW, Minto CF, Gambus PL, Andresen C, Goodale DB, Shafer SL, Youngs EJ. The influence of method of administration and covariates on the pharmacokinetics of propofol in adult volunteers. Anesthesiology 1998;88:1170–82[Web of Science][Medline]
8. Brice DD, Hetherington RR, Utting JE. A simple study of awareness and dreaming during anaesthesia. Br J Anaesth 1970;42:535–42[Abstract/Free Full Text]
9. Huber PJ. Robust statistics. Hoboken, New Jersey: Wiley-Interscience, 2004
10. Varvel JR, Donoho DL, Shafer SL. Measuring the predictive performance of computer-controlled infusion pumps. J Pharmacokinet Biopharm 1992;20:63–94[Web of Science][Medline]
11. Smith WD, Dutton RC, Smith NT. Measuring the performance of anesthetic depth indicators. Anesthesiology 1996;84:38–51[Web of Science][Medline]
12. Manyam SC, Gupta DK, Johnson KB, White JL, Pace NL, Westenskow DR, Egan TD. When is a bispectral index of 60 too low?: rational processed electroencephalographic targets are dependent on the sedative-opioid ratio. Anesthesiology 2007; 106:472–83[Web of Science][Medline]
13. Doi M, Gajraj RJ, Mantzaridis H, Kenny GN. Relationship between calculated blood concentration of propofol and electrophysiological variables during emergence from anaesthesia: 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]
14. Drummond JC. Monitoring depth of anesthesia: with emphasis on the application of the bispectral index and the middle latency auditory evoked response to the prevention of recall. Anesthesiology 2000;93:876–82[Web of Science][Medline]
15. Schneider G, Hollweck R, Ningler M, Stockmanns G, Kochs EF. Detection of consciousness by electroencephalogram and auditory evoked potentials. Anesthesiology 2005;103:934–43[Web of Science][Medline]
16. Pilge S, Zanner R, Schneider G, Blum J, Kreuzer M, Kochs EF. Time delay of index calculation: analysis of cerebral state, bispectral, and narcotrend indices. Anesthesiology 2006;104: 488–94[Web of Science][Medline]
17. Jensen EW, Litvan H, Revuelta M, Rodriguez BE, Caminal P, Martinez P, Vereecke H, Struys MM. Cerebral state index during propofol anesthesia: a comparison with the bispectral index and the A-line ARX index. Anesthesiology 2006;105:28–36[Web of Science][Medline]
18. Greenwald SD, Rosow C. BIS and EMG variability increase before somatic responses during surgery. Anesthesiology 2006; 105:A1027
19. Niedhart DJ, Kaiser HA, Jacobsohn E, Hantler CB, Evers AS, Avidan MS. Intrapatient reproducibility of the BISxp monitor. Anesthesiology 2006;104:242–8[Web of Science][Medline]
20. Jensen EW, Lindholm P, Henneberg SW. Autoregressive modeling with exogenous input of middle-latency auditory-evoked potentials to measure rapid changes in depth of anesthesia. Methods Inf Med 1996;35:256–60[Web of Science][Medline]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Rehberg, B. Articles by Baars, J. H. Search for Related Content PubMed PubMed Citation Articles by Rehberg, B. Articles by Baars, J. H.