A new electroencephalogram monitor, the Narcotrend, was developed to measure anesthetic depth. The authors compared the Narcotrend, the Bispectral Index, and classic electroencephalographic and hemodynamic parameters during anesthesia with propofol and remifentanil.
The authors investigated 25 patients undergoing laminectomy at different anesthetic states: awake, steady state anesthesia, first reaction during emergence, and extubation. Narcotrend value; BIS; relative power (percent) in delta, theta, alpha, and beta; median frequency; spectral edge frequency; and hemodynamic parameters were recorded simultaneously. The ability of the classic and processed electroencephalographic and hemodynamic parameters to predict the clinically relevant anesthetic states of awake, steady state anesthesia, first reaction, and extubation was tested using prediction probability.
Only the Narcotrend was able to differentiate between awake versus steady state anesthesia and steady state anesthesia versus first reaction/extubation with a prediction probability value of more than 0.90.
Modern electroencephalographic parameters, especially Narcotrend, are more reliable indicators for the clinical assessment of anesthetic states than classic parameters.
ASSESSMENT of the depth of anesthesia is of profound interest for anesthesiologists. However, measurement of depth of anesthesia is still an unsolved problem because there is no clear definition of what depth of anesthesia means.
The Narcotrend®, a new monitor displaying a derived electroencephalographic parameter (Narcotrend®; MonitorTechnik, Bad Bramstedt, Germany), automatically classifies the resting electroencephalogram into stages defined by Kugler during the 1980s. 1,2Narcotic stages deescalate from the awake state (stage A) to isoelectric electroencephalogram using 14 distinct gradations (table 1). 1Adequate depth of anesthesia is indicated by D0, D1, D2, E0, and E1, followed by F0 and F1, indicating burst suppression and isoelectric electroencephalogram, respectively. In a recent study, the Narcotrend® was compared with the Bispectral Index® (BIS®; Aspect Medical Systems, Newton, MA). Decreasing Narcotrend® stages during anesthesia were accompanied by decreasing BIS® values. 3However, no validation of the predictive probability of these monitors to differentiate clinically relevant endpoints of anesthesia is available so far.
The goal of the current study was to compare the Narcotrend®, BIS®, and classic electroencephalographic and hemodynamic parameters by analyzing the accuracy of each parameter to distinguish between different states of anesthesia, such as awake, steady state anesthesia, and emergence/extubation. Our hypotheses are that there is no difference in performance between the Narcotrend® and the BIS® in differentiation of anesthetic states, and both monitors would be better than classic power spectral electroencephalographic or hemodynamic parameters in this issue.
Materials and Methods
After institutional review board approval (Ärztekammer Hamburg, Hamburg, Germany) and written informed consent, 25 elective patients were included in the study. Selection criteria were age between 18 and 75 yr, American Society of Anesthesiologists physical status classification I or II, and laminectomy surgery. No patient with any medication interacting with the central nervous or cardiopulmonary system was included in the study to avoid influences on the electroencephalographic and hemodynamic parameters.
After premedication with 7.5 mg oral midazolam (30 min before induction), anesthesia was induced with 0.7 μg · kg−1· min−1remifentanil continued after 2 min with additional target-controlled infusion (TCI) of 5.0 μg/ml propofol over 10 min. Rocuronium bromide (0.6 mg/kg) was used to facilitate tracheal intubation. After intubation, anesthesia was maintained with 0.3 μg · kg−1· min−1remifentanil and 3.0 μg/ml propofol (Diprifusor, Graseby 3500; Graseby Medical Limited, Hertfordshire, Watford, United Kingdom). After the end of surgery, remifentanil infusion was stopped, and 10 min later, propofol was reduced step by step (0.2 μg/ml) every 3 min. Extubation criteria were sufficient spontaneous breathing and spontaneous eye opening. We defined different anesthetic states: awake, steady state anesthesia, first reaction, and extubation (table 2). Comparisons were performed for awake versus steady state anesthesia, steady state anesthesia versus first reaction, and steady state anesthesia versus extubation.
The electroencephalogram was registered by seven adhesive silver-silver chloride gel-filled electrocardiogram electrodes (Blue-Sensor; Medicotest, Olstykke, Denmark) on carefully prepared skin (Arbo-Prep; Tyco Healthcare, Neustadt, Germany). Electrode placement was performed according to the instructions of the manufacturers of BIS® (two-channel reference, At1-Fpz and At2-Fpz, ground Fp2) and Narcotrend® (one-channel bipolar at the hairless skin of the forehead). Electrode impedance was kept below 5 kΩ. BIS® (version 3.3), relative power in δ (%δ: 0.5–3.75 Hz), θ (%θ: 4.0–7.75 Hz), α (%α: 8.0–13.5 Hz), β (%β: 13.75–30.0 Hz), spectral edge frequency, and median frequency were recorded by the Aspect A-1000 monitor (Aspect Medical Systems). 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. Narcotrend® stages (version 2.0 AF/F) were registered by the Narcotrend® electroencephalographic system (MonitorTechnik). All data were stored on disk.
Hemodynamic parameters such as heart rate, noninvasive mean arterial pressure (MAP), and oxygen saturation were measured and registered at every point of measurement (Solar 8000; Marquette Hellige Medical Systems, Milwaukee, WI). End-expiratory carbon dioxide concentrations were maintained between 35–40 mmHg during the whole observation time.
Statistical Analysis
For all parameters, the 95%, 90%, 75%, 50%, 25%, 10%, and 5% percentiles were calculated for the investigated states. The accuracy to distinguish between the anesthetic states awake versus steady state anesthesia, steady state anesthesia versus first reaction, and steady state anesthesia versus extubation were analyzed with the prediction probability (PK). PKwas calculated for all parameters using a custom spreadsheet macro, PKMACRO, as described by Smith et al. 4The jackknife method was used to compute the SE of the estimate. Comparisons of the PKvalues were performed with the PKDMACRO. 4A value of PKof 0.5 means that the parameter predicts the states not better than a 50:50 chance. A value of PKof 1.0 means that the parameter predicts the states correctly 100% of the time. A value less than 0.5 means that discordance is more likely than concordance. To enable comparison of PK, we used 1 − PKwhen the PKvalue was less than 0.5. 4Logistic regression was used to analyze the probability for first reaction in comparison with steady state anesthesia for the Narcotrend® and BIS®. Correlations between the Narcotrend® and BIS® were evaluated by nonparametric Spearman rank correlation coefficient and linear regression analysis for all data pairs from awake through extubation. Statistical analysis were performed using the SPSS package (version 9; SPSS, Chicago, IL), PKMACRO, and PKDMAKRO. 4
Results
Data evaluation was performed in 25 patients (12 female, 13 male; age, 51 ± 13 [SD] yr; weight, 73 ± 13 kg; height, 171 ± 10 cm) with almost artifact-free signal registration. Length of laminectomy was 98 ± 34 min without unusual perturbations such as blood loss or hypothermia. Patients were extubated at a propofol TCI of 1.5 ± 0.2 μg/ml without complications.
Because the information of separation measures (such as the Wilcoxon test) is limited for the evaluation of a depth of anesthesia parameter, 4we investigated PKvalues for the comparison of anesthetic states (fig. 1). Only the Narcotrend® was able to differentiate between awake versus steady state anesthesia, steady state anesthesia versus first reaction, and steady state anesthesia versus extubation with a PKvalue of more than 0.90 (table 3and fig. 2). Logistic regression analysis indicated a 95% probability that the patients would show a first reaction during emergence from propofol anesthesia for the Narcotrend® stages D0/C2 (fig. 3). The BIS® showed high PKvalues for awake versus steady state anesthesia and steady state anesthesia versus extubation (PK> 0.90), whereas the PKvalue for steady state anesthesia versus first reaction was 0.79 (table 3).
Because of the different algorithms of the Narcotrend® and BIS® to evaluate the spontaneous electroencephalogram, we were interested in the relation of both parameters. A high correlation coefficient (r =−0.85, P < 0.001) of both was found (fig. 4). Linear regression was able to predict BIS® by Narcotrend® (BIS = Narcotrend®× (−5.397 + 99.318); P < 0.001) and Narcotrend® by BIS® (Narcotrend®= BIS × (−0.149) + 16.478, P < 0.001). The results of the linear regression for Narcotrend® are shown in table 4.
Discussion
In the current study, we confirmed the advantage of modern over classic electroencephalographic parameters. The classic electroencephalographic parameters were unsatisfying to distinguish between the awake and anesthetized state, but they provided useful information during emergence from anesthesia. Interestingly, also hemodynamic parameters, especially MAP, seemed to be a reliable parameter to differentiate between the investigated states in healthy patients during propofol and remifentanil anesthesia.
The measurement of depth of anesthesia is still an unsolved problem because there is no definition of exactly what depth of anesthesia means. However, a depth of anesthesia monitor would be of enormous clinical and experimental interest to increase the safety of the patient and to reduce costs by avoiding drug overdosing. A gold standard for the measurement of depth of anesthesia is missing. One strategy is to investigate the relation of a potential parameter to anesthetic effect-site concentrations. However, it is well known that sensitivity to anesthetics is associated with interindividual variations. 5Another strategy is to compare clinically measured depth of anesthesia with a parameter such as the Observer's Assessment of Alertness and Sedation Score. 6In this case, the interindividually sensitivity of anesthetics could be minimized.
The present study was performed using a propofol and remifentanil anesthetic, often used in our hospital. We were aware that anesthetic depth could be different between the patients (interindividual), especially during steady state anesthesia. With the detection of first reaction and the extubation state, we also investigated individual and not drug-dosing regimen-dependent anesthetic states. It should be mentioned that we did not measure propofol blood/plasma concentrations but used a calculated professional pharmacokinetic model, the Diprifusor system. 7Previous studies with the Diprifusor system have shown variations between measured and calculated concentrations. The bias of systemic overprediction and underprediction of the measured plasma concentrations (median performance error) has been reported to be −7%, 816%, 9and 21%. 10However, because the real plasma concentrations were not analyzed in the current study, potential relevant differences from the TCI data remain unclear.
In the current study, only the Narcotrend® was able to distinguish all investigated states of anesthesia accurately. To our knowledge, no other investigations of the ability of the Narcotrend® to predict anesthetic state have been conducted. Data on the Narcotrend® in the literature are very rare. 3,11In a previous study, our group found significant increases in Narcotrend® stages during emergence from propofol when TCI doses of propofol were reduced. 11Also, a significant correlation between the Narcotrend® and TCI of propofol was found. However, a high correlation does not necessarily mean a good distinction between anesthetic states such as steady state anesthesia versus first reaction. Kreuer et al. 3investigated the relation between the Narcotrend® and BIS® during anesthesia with propofol and remifentanil with time intervals of 1 min during induction and emergence and every 5 min during maintenance of anesthesia by descriptive analysis. When BIS® values were 64–40, indicating general anesthesia, 12in 94% of the cases, the Narcotrend® stage was D or E. The authors concluded that increasing depth of anesthesia as indicated by the BIS® is accompanied by corresponding effects as displayed by the Narcotrend®. 3These findings are in concordance with results from the current study using the statistical method of linear regression to analyze the relation of Narcotrend® and BIS®. A Narcotrend® stage of D or E was highly significantly correlated with BIS® values between 56 and 35. Interestingly, figure 4show a cluster of points well off the regression line, with BIS® values greater than 80 and Narcotrend® values between C2 and E0. This cluster is due to the data pairs during induction of anesthesia. Narcotrend® was associated with a faster decrease than BIS®. The evaluation of the spontaneous electroencephalogram by different mathematic strategies (Narcotrend® and BIS®) resulted in similar information during anesthesia with propofol and remifentanil.
The BIS® is the most evaluated parameter derived from the spontaneous electroencephalogram. In the current study, the BIS® showed a perfect division of the awake versus steady state anesthesia states. However, BIS® was a poor detector of the states of steady state anesthesia versus first reaction during emergence. One possibility for the poor detection could be the well-known delay of BIS® during the fast changes of propofol emergence. Baker et al. 13observed that BIS® most accurately reflects the level of consciousness of the patient approximately 60 s previously. This seems to be realistic because the total update delay of BIS® is approximately 30 s. 14This would also explain the results from the current study with the high PKvalue for steady state anesthesia versus extubation. Extubation was executed more than 60 s after the first reaction period in all cases. However, we were not able to use the newest version of BIS® (version 3.3 instead of version xp) in the current study, which may restrict our results in this issue. We used the Aspect A-1000 monitor because of the possibility to calculate the classic electroencephalographic parameters. Moreover, the delay of BIS® as described by Baker was also performed for the new Aspect-2000 monitor.
The value of hemodynamics to assess depth of anesthesia is still controversial. 15–19MAP can be only an indirect parameter to estimate the hypnotic effects. Changes in MAP are mediated by the cardiodepressive side effects of propofol and remifentanil. MAP was therefore far more likely to predict increasing and decreasing concentrations of propofol and remifentanil rather than any particular stages of depth of anesthesia. However, our study underlined the experience of many anesthesiologists to estimate the drug-dosing regimen by hemodynamics.
The median frequency, representative of the classic electroencephalographic parameters, had been reported to be a useful tool to measure depth of anesthesia. Median frequency was used by Schwilden et al. 20,21to confirm a closed-loop feedback system to control propofol and methohexital anesthesia. In contrast, other studies showed that classic electroencephalographic parameters did not correlate with depth of anesthesia. 22–24In the current study, we found limitations of the classic electroencephalographic parameters to monitor depth of anesthesia. Our findings confirm that it seems to be difficult to assess depth of anesthesia using only classic electroencephalographic parameters. Otherwise, the classic parameters, especially spectral edge frequency and %β, mediated important information during emergence from anesthesia. Further studies should evaluate whether these data are useful to create a subparameter improving Narcotrend® or BIS® during emergence from anesthesia.
In the current study, we used the PKMACRO to estimates the PKand SE based on the given data of the 25 patients. It should be mentioned that these estimates are not the true values of PKand SE, but the estimates can be used for statistical hypothesis testing. 4For example, the estimates of PK= 1.0 for awake versus steady state anesthesia in the current study may be estimates of PK< 1.0 if more patients are investigated.
The results of the current study underline our hypothesis that the Narcotrend® and BIS® show nearly similar changes during the investigated periods of anesthesia. The Narcotrend® was superior to BIS® to detect emergence from anesthesia, most probably because of the known time delay of 60 s for BIS®. Both are superior to classic electroencephalographic parameters in monitoring and differentiating different states of anesthesia.