Different analytical concepts were introduced to quantify the changes of the electroencephalogram. The Datex-Ohmeda S/5 Entropy Module (Datex-Ohmeda Division, Instrumentarium Corp., Helsinki, Finland) was the first commercial monitor based on the entropy generating two indices, the state entropy (SE) and the response entropy (RE). The aim of the current study was to compare the accuracy of SE and RE with the Bispectral Index(R) monitor (BIS(R); Aspect Medical Systems, Newton, MA) during propofol-remifentanil anesthesia.
The authors investigated 20 female patients during minor gynecologic surgery. SE, RE, BIS, mean arterial blood pressure, heart rate, and sedation level were recorded every 20 s during stepwise increase (target-controlled infusion, 0.5 microg/ml) of propofol until the patients lost response. Five minutes after loss of response, remifentanil infusion (0.4 microg . kg(-1) . min(-1)) was started. Spearman correlation coefficient and prediction probability were calculated for sedation levels with SE, RE, BIS, mean arterial blood pressure, and heart rate. The ability of the investigated parameters to distinguish between the anesthesia steps awake versus loss of response, awake versus anesthesia, anesthesia versus first reaction, and anesthesia versus extubation was analyzed with the prediction probability.
SE correlates best with sedation levels, but no significant differences of the prediction probability values among SE, RE, and BIS were found. The prediction probability for all investigated steps of anesthesia did not show significant differences among SE, RE, and BIS. SE, RE, and BIS were superior to mean arterial blood pressure and heart rate.
SE, RE, and BIS revealed similar information about the level of sedation and allowed the authors to distinguish between different steps of anesthesia. Both monitors provided useful additional information for the anesthesiologist.
ELECTROENCEPHALOGRAPHY has been introduced in anesthesiology to monitor the hypnotic state of patients. Different analytical concepts have been presented to quantify the changes of the electroencephalogram during the past decades. A big advantage was the fast-Fourier transformation, which allowed calculation of the linear information of the electroencephalography. In recent years, higher-order statistics have been used to describe changes of the electroencephalography. One method is bispectral analysis, which is based on the correlation of the phase between different frequency components of the electroencephalogram.1In 1992, the Bispectral Index® monitor (BIS®; Aspect Medical Systems, Newton, MA), a commercial electroencephalographic monitor system, became available, based on the bispectral analysis. Another concept to quantify electroencephalography is the calculation of the entropy, describing the irregularity, complexity, or unpredictability characteristics of a signal. A signal in which sequential values are generated by a random number of generators has high entropy. A regular and well-ordered signal has an entropy value of zero. Bruhn et al .2,3demonstrated the usefulness of the Shannon entropy and the approximate entropy to describe the electroencephalographic changes during desflurane anesthesia. The Datex-Ohmeda S/5 Entropy Module (Datex-Ohmeda Division, Instrumentarium Corp., Helsinki, Finland) is the first commercial monitor based on the spectral entropy.4The Datex-Ohmeda S/5 Entropy Module generates two indices, the state entropy (SE) and the response entropy (RE). The SE is computed over the frequency range from 0.8 to 32 Hz, the RE is calculated over the frequency range from 0.8 to 47 Hz. According to the manufacturer, the RE includes additional information about the electromyographic activity (activity higher than 32 Hz) of the face muscles.4The SE (range from 91 to 0) and the RE (range from 100 to 0) are normalized in such a way that the RE becomes equal to the SE when there is no electromyographic activity. In a recent study, Vakkuri et al .5found a faster reaction of RE during emergence than SE and BIS.
The aim of the current study was to give a first impression of the new Datex-Ohmeda S/5 Entropy Module during propofol–remifentanil anesthesia. In detail, we were interested in whether the Datex-Ohmeda S/5 Entropy Module is as reliable or better or worse than the BIS® monitor in reflecting changes during increasing propofol and remifentanil infusions and sedation levels. Moreover, we were interested in the ability to differentiate between the anesthesia steps awake versus loss of response, awake versus anesthesia, anesthesia versus first reaction, and anesthesia versus extubation and the nuances of anesthetic steps loss of response versus first reaction. In addition, we analyzed the correlation of SE, RE, and BIS from awake throughout extubation during anesthesia with propofol and remifentanil.
Materials and Methods
After institutional review board (Ethik-Kommission der Arztekammer Hamburg, Hamburg, Germany) approval and written informed consent, 20 elective patients were included in the study. Selection criteria were age 18 to 75 yr, American Society of Anesthesiologists physical status classification I or II, and minor gynecologic surgery.
After premedication with 7.5 mg oral midazolam (30 min before induction), anesthesia was induced with a 0.5-μg/ml target-controlled infusion (Diprifusor, Graseby 3500; Graseby Medical Limited, Hertfordshire, Watford, United Kingdom) of propofol followed by a stepwise increase (0.5 μg/ml) until the patient lost response. Every propofol concentration level was maintained over 1 min. Loss of response was detected by the Modified Observer’s Assessment of Alertness/Sedation Scale (MOAAS; table 1).6Loss of response was defined by MOAAS values less than 1. MOAAS was tested every 20 s until each target-controlled concentration was reached.
After loss of response, 0.4 μg · kg−1· min−1remifentanil (Graseby 3400; Graseby Medical Limited) was started. Five minutes after the start of remifentanil infusion, cisatracurium (0.1 mg/kg) was given to facilitate tracheal intubation. Anesthesia was maintained with the target-controlled concentration obtained for loss of response added with 0.3 μg · kg−1· min−1remifentanil until skin incision. After skin incision, anesthesia was maintained by the anesthesiologist according to clinical signs. At the end of surgery, propofol and remifentanil infusions were stopped simultaneously. MOAAS was tested every 20 s until extubation. Extubation criteria were sufficient spontaneous breathing and spontaneous eye opening.
We defined five different anesthetic steps: awake, loss of response, steady state anesthesia (target-controlled concentration obtained for loss of response added with 0.3 μg · kg−1· min−1remifentanil) before surgical incision, first reaction (any reaction such as movement, coughing, or eye opening during emergence after stop of propofol–remifentanil infusion; MOAAS was tested every 20 s), and extubation. Comparisons were performed for the anesthesia steps awake versus loss of response, awake versus anesthesia, anesthesia versus first reaction, and anesthesia versus extubation. Moreover, we were interested in the differentiation between the anesthesia steps anesthesia loss of response versus first reaction.
The electroencephalogram was registered for BIS with BIS® sensor XP electrodes (Aspect Medical Systems) over the temporal-frontal area of the forehead. The BIS (version 4.0) was calculated by the BIS® module (Datex-Ohmeda Division, Instrumentarium Corp.). Electrode impedance was kept below 10 kΩ. The BIS was calculated with a smoothing rate of 30 s. The Entropy Sensor (Datex-Ohmeda Division, Instrumentarium Corp.) was placed over the temporal-frontal area of the forehead after a careful preparation of the skin. The position of the BIS® sensor XP or the Entropy Sensor over the left or right forehead was decided by flipping a coin. After the sensor application, an impedance check was performed automatically. Electrode impedance was also kept below 7.5 kΩ. SE and RE were calculated by the Datex-Ohmeda S/5 Entropy Module.
Heart rate (HR), noninvasive mean arterial blood pressure (MAP), and oxygen saturation were measured and registered at every point of measurement (AS/3; Datex-Ohmeda Division, Instrumentarium Corp.). End-expiratory carbon dioxide concentrations (ETco2) were maintained between 35 and 40 mmHg during the whole observation time. All data (SE, RE, BIS, HR, and MAP) were recorded on-line at a 5-s data sampling rate and stored by the Datex-Ohmeda S/5 Collect software (Datex-Ohmeda Division, Instrumentarium Corp.).
For SE, RE, BIS, MAP, and HR, the 95, 90, 75, 50, 25, 10, and 5% percentiles were calculated for every change in the MOAAS, at 1, 2, 3, 4, and 5 min after the start of remifentanil infusion, and for the investigated steps (one value for each patient).
Comparison of MOAAS with SE, RE, BIS, MAP, and HR was performed with the Spearman correlation coefficient. Moreover, correlations between SE, RE, and BIS were evaluated by nonparametric Spearman rank correlation coefficient for all data pairs from awake through extubation.
Comparison of MOAAS scores (first value after reaching a new MOAAS level for every patient) with SE, RE, BIS, MAP, and HR was performed with prediction probability (PK).7The accuracy to distinguish between the anesthetic steps awake versus loss of response, awake versus anesthesia, anesthesia versus first reaction, and anesthesia versus extubation were analyzed with PK. Therefore, we were comparing one value per patient for SE, RE, BIS, MAP, and HR for every step. Moreover, we analyzed PKfor loss of response versus first reaction. PKwas calculated using the PKMACRO as described by Smith et al .7The jackknife method was used to compute the standard error of the estimate. A value of PK= 0.5 means that the parameter predicts the steps not better than a 50:50 chance. A value of PK= 1.0 means that the parameter predicts the steps correctly 100% of the time. To compare the parameters to distinguish all steps, we calculated
All steps were weighed equally. A parameter that distinguishes all steps of anesthesia perfectly with PK= 1.0 results in PKall= 1.0 * 1.0 * 1.0 * 1.0 = 1.0, and a parameter predicting all steps with PK= 0.5 results in PKall= 0.5 * 0.5 * 0.5 * 0.5 = 0.0625. Comparisons of the PKvalues were performed with the PKDMACRO.7
A power calculation, based on differences in PK, was performed to estimate the sample size. We considered that a difference of less than 0.05 in PKwould not be of clinical importance. The t statistic was then calculated as the quotient between the chosen difference in clinical importance and the standard error of the estimate. With these characteristics and testing with a statistical significance of P < 0.01, 20 patients should be included in the study.8
For SE and BIS, the logistic regression was calculated for the probability of loss of response, anesthesia, first reaction, and extubation. The Hill coefficients were calculated to describe the steepness of those relations.9
Changes of RE, BIS, MAP, and HR after the start of remifentanil infusion were tested with the Friedman test for repeated measurements. In case of significant “overall” effects, changes were evaluated in detail a posteriori by the Wilcoxon test. The Bonferroni correction was performed to account for multiple testing.
Statistical analyses were performed using the SPSS package (version 9; SPSS, Chicago, IL), SigmaPlot (version 8.0; SPSS), PKMACRO, and PKDMACRO.7A P value less than 0.05 was considered statistically significant.
Data evaluation was performed in 20 female patients (age, 51 ± 14 [SD] years; weight, 71 ± 16 kg; height, 166 ± 6 cm). The duration of surgery was 68 ± 36 min, without unusual perturbations such as blood loss or hypothermia. Patients were extubated after the stop of propofol–remifentanil infusion and at a calculated propofol plasma concentration of 1.2 ± 0.3 μg/ml, without any complications. In no patient did the electrode strips have to be replaced because the electrode impedance was above 10 kΩ for BIS or 7.5 kΩ for SE and RE.
SE, RE, and BIS decreased after induction of anesthesia and increased during emergence (fig. 1). Increasing sedation (from MOAAS = 5 to MOAAS = 0) was associated with decreases for median SE from 89 to 51, median RE from 97 to 55, and median BIS from 94 to 58 as shown in figure 2. SE correlated best with MOAAS (r = 0.82, PK= 0.89), but differences of PKvalues were not statistically significant (table 2). Correlation coefficients and PKvalues for the sedation levels were higher for SE, RE, and BIS than for MAP and HR (table 2).
The start of remifentanil infusion resulted in statistically significant decreases for SE, RE, BIS, MAP, and HR (fig. 3and table 3). The first effects were found for MAP and HR (second minute), followed by BIS (third minute) and SE and RE (fourth minute).
Changes of SE, RE, BIS, MAP, and HR during the investigated steps are shown in figure 4. Only BIS was able to distinguish each evaluated step with PK> 0.90 (table 4), resulting in PKall= 0.9. SE and RE showed a high prediction probability (PK> 0.9) for the steps awake versus loss of response, awake versus anesthesia, and anesthesia versus extubation (table 4) and a lower prediction probability (PK> 0.8) for the step anesthesia versus first reaction. PKvalues for SE, RE, and BIS were without any significant differences (P < 0.05). The 95% possibility for loss of response was 80 for BIS and 81 for SE. The 95% possibility for the first reaction was 72 for BIS and 79 for SE (fig. 5and Table 5). PKallwas higher for BIS, SE, RE than for MAP and HR. However, SE, RE and BIS were not able to distinguish accurately between loss of response versus . first reaction (Table 4and fig. 6).
Because of the different algorithms of entropy (SE and RE) and BIS, we were interested in the correlation of both parameters. A high correlation coefficient (SE: r =−0.83, P < 0.001; RE: r =−0.84, P < 0.001) of both was found (fig. 7). BIS values between 100 and 85 indicating awake patients10were indicated in 88% of data pairs by SE values between 91 to 80. BIS values of 65–40 (recommended for general anesthesia)10were associated in 84% with SE values of 59 to 30. Details are given in figure 8.
An accepted standard for the measurement of depth of anesthesia is still missing. In the current study, we used three ways to estimate the status of the Datex-Ohmeda S/5 Entropy Module, a new designed electroencephalographic system, to monitor depth of anesthesia.4First, we tested the Datex-Ohmeda S/5 Entropy Module in comparison with different sedation levels. Second, we were interested in changes of the Datex-Ohmeda S/5 Entropy Module after the application of remifentanil. Third, we investigated the ability of the Datex-Ohmeda S/5 Entropy Module to differentiate between different steps of anesthesia.
One strategy to test a depth-of-anesthesia monitor is comparison with patient’s clinically observed level of sedation.8Therefore, we studied SE, RE, BIS, MAP, and HR during different sedation levels. We used the MOAAS because it provides a good correlation with different degrees of sedation and has been tested in many other studies.8In the current study, decreases in sedation level were associated with decreases in SE, RE, and BIS. In contrast, changes in MAP and HR did not correlate well with changes in MOAAS. SE correlated best with MOAAS, but PKvalues were without any statistically significant difference among SE, RE, and BIS.
Interestingly, in the current study, SE, RE, BIS, MAP, and HR decreased statistically significantly after the start of remifentanil infusion. No investigations regarding the effects of remifentanil on SE and RE have been published until now. The reaction of BIS on remifentanil infusion has been described controversially. Guignard et al .11obtained similar BIS values during administration of 4.0 μg/ml propofol effect site concentration combined with remifentanil effect site concentrations of 0, 2, 4, 8, and 16 ng/ml before laryngoscopy, whereas quantitative BIS responses to noxious stimuli (laryngoscopy and intubation) were attenuated in relation to increasing remifentanil applications. These findings indicate that BIS is not a reliable indicator for prediction of responses to painful events, even if BIS changes are related to analgesic components, quantitatively. Explaining the unsatisfying indication of the analgesic power by the BIS, Guignard et al .11hypothesized that these components are mediated in subcortical brain structures and at the level of spinal cord, which cannot be detected by electroencephalographic registration from the surface of the scalp. In contrast, the current study showed that BIS changed significantly from a median of 49 to a median of 37 in response to remifentanil infusion. Besides a pronounced analgesic effect, all opioids also provide dosage-related sedative effects. Without propofol, remifentanil infusion resulted in high-amplitude electroencephalographic slowing and delta waves. This might be the reason for the correlation of increasing remifentanil dosages (0.01–0.1 μg · kg−1· min−1) with decreasing BIS values during lower plasma concentrations of propofol (2 μg/ml).12The responses of BIS to remifentanil might not be simply remifentanil dosage related. Moreover, BIS also depends on the existence of the hypnotic power from propofol, quantitatively.
A depth-of-anesthesia monitor should distinguish accurately between different steps of anesthesia such as awake, loss of response, and first reaction during emergence. Therefore, not only statistically significant differences should be recommended. It must be demanded that, for an accurate index, values indicating different steps should not overlap.13In the current study, we used the prediction probability (PK), presented by Smith et al .,7to detect the accuracy of the parameters to distinguish between the investigated steps of anesthesia. A PKvalue of 0.5 means that the parameter distinguishes between the two steps by chance (50:50). A PKvalue of 1.0 indicates a correct distinction with no overlap. In the current study, the awake patients were represented by median SE values of 89 (median RE = 97), whereas the median SE for the state of anesthesia was 38 (median RE = 40). SE and RE showed a high prediction probability (PK> 0.9) for the steps awake versus loss of response, awake versus anesthesia, and anesthesia versus extubation. The PKvalues of SE and RE were lower (PK< 0.9) to distinguish between anesthesia versus first reaction. One reason might be that the step first reaction (e.g ., coughing or moving spontaneously or after stimulation with MOAAS) is not necessarily related to the cortex activity but rather to subcortical activation. However, for the anesthesiologist, it is important to detect and predict the first reaction. The PKvalues of SE and RE were similar and with no significant differences compared with BIS. The BIS is the most evaluated parameter derived from the spontaneous electroencephalogram, but its status as a depth of anesthesia monitor is still controversial.13,14In a recent study, we underlined the potency of BIS to distinguish between patients who were awake and anesthetized with propofol and remifentanil. For the 25 patients observed, no overlap in the data were found between the awake and anesthetized states.15During emergence, BIS was not able to distinguish the first reaction during conditions without any external stimulation from the anesthetized state. We hypothesized that this was due to the well-known delay of BIS of 60 s.16Interestingly, in the current study, BIS was able to distinguish between all investigated conditions, including steady state anesthesia versus first reaction, accurately (PK> 0.9). The reason for the better differentiation in the current study may be due to stimuli (MOAAS) during the emergence period. In the current study, we found a limitation of the Datex-Ohmeda Entropy Module and the BIS® monitor. SE, RE, and BIS were not able to distinguish between the nuances of anesthesia when patients only lost response to MOAAS versus the first reaction during emergence. In accord with our data, for example, an SE value of 80 is not able to predict whether the patient shows a response or not.
The SE, based on electroencephalography, ranges in values from 91 to 0, and the RE, based on electroencephalography and electromyography, ranges in values from 100 to 0.4The advantage of RE should be the possibility of a faster reaction to changes such as arousal.4Vakkuri et al .5demonstrated that RE indicates emergence from anesthesia 11 s earlier than SE and 12.4 s earlier than BIS. In the current study, no advantage of RE over SE was found. Differences between SE and RE to distinguish between the investigated steps of anesthesia were without any statistically significant effect. In the study of Vakkuri et al .,5the patients were tested only for loss of consciousness, defined as inability to respond to verbal command (“squeeze my hand”), and recovery of consciousness was determined to have taken place when the patient again responded purposefully to verbal command (“squeeze my hand”). Therefore, the testing scales in these studies were different.
In summary, SE, RE, and BIS were similar and reliable indices to detect different sedation levels and to distinguish between the investigated steps of anesthesia. We also showed that RE, SE, and BIS did not differentiate between different nuances of anesthesia (loss of response vs . first reaction). However, SE, RE, and BIS were superior to MAP and HR to distinguish between different steps of anesthesia with propofol and remifentanil and provided additional information for the anesthesiologist.