An important goal of neuroanesthesia is a prompt, controlled emergence following a neurosurgical procedure. There are neurosurgical cases where emergence may be delayed, especially after periods of deep anesthesia, for example, (1) cerebral protection induced by propofol or barbiturate administration to an isoelectric electroencephalogram (EEG) during temporary clipping of feeding vessels of cerebral aneurysms and (2) microvascular decompression (MVD) for trigeminal neuralgia. Deep anesthesia is standard for the latter procedure where motor-evoked potentials are assessed to prevent cranial nerve injury.

The bispectral index (BIS) monitor can aid emergence in outpatient procedures with respect to both time-to-wakening and length-of-stay in the recovery room.1 However, the manner in which the BIS directs emergence is not highlighted, and a specific BIS number to indicate emergence is not suggested. A correlation between patient arousal and eye opening appeared poor during emergence, with the BIS output usually lower than pre-induction values.2 On occasion, the BIS reading may be artefactually low during emergence. In these circumstances, it has been suggested that the raw EEG be observed to aid emergence.3 In addition, the BIS may correlate poorly with end-tidal desflurane (ETDes) and the awake state.4 Thus, while the BIS can aid management of the depth of anesthesia during maintenance, it may not be ideally suited to direct a facilitated emergence.

Using the EEG to both monitor emergence and aid its progress makes sense. A monitor that could predict emergence with greater precision would be valuable. New EEG monitoring, engineered to potentially provide this information, is now available in the form of the EEGo,5 which processes the standard EEG signal through three-dimensional time-delay plots.6 With this monitor, a cascade of characteristic waveforms occurs from burst suppression to the awake state (Fig. 1a–c). These identifiable changes resemble phase transitions and occur rapidly from one state to the next.7,8 An analogy is the phase transition that occurs when water changes to ice and vice versa. Monitoring these transitions suggests a means to guide depth of hypnosis, which could potentially shorten emergence at end-procedure.

Fig. 1
figure 1

Three characteristic images depicting depth of sedation with the electroencephalogram (EEGo) monitor. These are time-delay plots called attractors. Each figure shows the attractor in two views with the raw EEG signal at the bottom. a Periodic attractor: demonstrates an adequate depth of hypnosis. b Toroidal attractor: this doughnut-shaped attractor is the characteristic pre-emergent form seen prior to awakening. c Chaotic attractor: seen with an awake patient. The vertical bars (each in triplicate) individually highlight the time-delay (8–10 samples at 1200 samples · s−1). The distance between the vertical bars indicates the composite graphical display of the two images displayed above the raw EEG signal. In this situation a small dot is seen in the centre of the rotating box irrespective of orientation with isoelectricity. A video link from the original paper describing the EEGo can be accessed at: http://download.lww.com/wolterskluwer_vitalstream_com/pt/anes/v105n5p927.mov

Full size image

This study was designed to compare the ability of the BIS monitor with that of the EEGo monitor in facilitating prompt emergence from anesthesia, where EEG monitoring is done in specific neurosurgical procedures. We hypothesized that the EEGo would be superior to the BIS monitor, both in assessing depth of hypnosis during these procedures where deep levels of anesthesia are standard and in facilitating a more prompt emergence following these neurosurgical procedures.

Methods

The study was approved by the Biomedical Research Ethics Board at the University of Manitoba and registered at ClinicalTrials.gov with identifier NCT00443807. All adult patients undergoing neurosurgery where intraoperative EEG monitoring is the standard of care at our institution were approached in the Pre-Anesthetic Clinic. Exclusion criteria included patient refusal, a history of asthma requiring routine use of bronchodilators (due to the use of desflurane as the volatile agent), pregnancy, non-elective aneurysm clipping, and the presence of underlying neurological disorders. Patients who agreed to participate provided signed informed consent. Each patient received midazolam 1 milligram for pre-operative anxiolysis prior to placement of a 20-gauge radial artery cannula. All patients had a BIS monitor sensor strip applied, as per manufacturer’s directions, to the contralateral hemisphere from the planned surgery. Standard monitors were applied. Patients were pre-oxygenated and anesthesia induced with remifentanil 1 μg · kg−1 iv, propofol 1.5–2.5 mg · kg−1 iv, lidocaine 1.5 mg · kg−1 iv, and rocuronium 0.6 mg · kg−1 iv. A 7- or 8-mm ID endotracheal tube was inserted by direct laryngoscopy. Post-induction, routine intraoperative monitoring (IOM) was established by the neurophysiologist. Needle electrodes were placed frontally and occipitally with a reference electrode over the inion using the international 10–20 system for bipolar recording of the raw EEG. Appropriate impedance characteristics were obtained, and acceptable signals were confirmed with the neurophysiologist. The analog EEG signal was processed by a Neurotrac EEG Analyzer, and the output signal was digitized by a DAQ Module USB1420 and cabled to a free-standing laptop computer with EEGo software.

Both the BIS Vista® and the EEGo signals were continuously measured for each patient. To aid management, the clinician was given access to only one of the monitors, as randomly allocated by a third party with a coin toss to either EEGo (heads) or BIS (tails). Continuous data acquisition of blood pressure, heart rate, end-tidal CO2, BIS, and ETDes displayed on a Philips Intellivue MP70TM monitor were secured by a second computer running Trendface Solo™ software.

Anesthesia was maintained with desflurane between 0.5 and 1.5 MAC (to maintain mean arterial pressure in the range of 70–100 mmHg—usual dose 1.0 MAC), remifentanil 0.05–0.1 μg · kg−1 · min−1, and propofol up to 50 μg−1 · kg−1 · min−1 (during the period of dissection of vessels about the brainstem with MVD surgery). Incremental doses of propofol (usually 50–100 mg) were administered just prior to temporary clipping of feeding vessels during cerebral aneurysm surgery. Mean arterial pressure was supported at 90–100 mmHg with phenylephrine infusion (0.1–0.5 μg−1 · kg−1 · min−1) during the period of temporary clipping. Recording of the EEGo and BIS signals were continuously obtained during this period of deep hypnosis. Dexamethasone 6 mg iv and granisetron 1 mg iv were administered within 1 hr of emergence for anti-emetic prophylaxis. Morphine 0.1 mg · kg−1 iv was provided during scalp closure for post-operative analgesia.

With scalp closure, the depth of anesthesia was lightened to facilitate emergence. In the BIS group, the patient was lightened to 60 ABU. In the EEGo group, the patient was lightened to near-toroidal attractor (Fig. 1b). Ventilation parameters were adjusted to an end-tidal CO2 of 40–45 mmHg.

Desflurane was discontinued immediately prior to removal of the Mayfield clamp, while a remifentanil infusion of 0.03 μg · kg−1 · min−1 continued until the clamp was removed. Fresh gas flows were then increased to 8 L · min−1 to encourage wash-out of desflurane. Once the patient made respiratory efforts, the ventilation was changed from pressure control to pressure support. The time was noted from when the desflurane was discontinued to when the patient was able to follow commands. The patient was deemed awake once her/his eyes opened to command. Following tracheal extubation, the patient was asked person, place, and date at 15-sec intervals until best answer. The duration of time until the patient became oriented was recorded.

An assessment of quality of emergence, smooth (1), single cough (2), multiple coughs (3), along with an overall impression on a scale of 1–10, was offered by the neurosurgeon. The neurosurgery resident determined a Glasgow Coma Score (GCS) when the patient was admitted to the postanesthesia care unit (PACU). Time until discharge was determined. Twenty-four hours following surgery, a survey of patient satisfaction9 and a query regarding awareness10 were completed by an anesthesiologist blinded to the mode of EEG monitoring.

Statistical analysis

With on-line EEG processing to aid emergence at end-anesthesia, we hypothesized that the EEGo could decrease emergence from 10 to 5 min from last stitch to awake. With an α-value of 0.05, a 1 − β of 0.80, a standard deviation of 5 and 2.5 min, respectively, and an allocation ratio of 1, we required ten patients per EEG monitoring strategy to have sufficient power to demonstrate this difference with a two-sided statistical test. Between-group comparisons of continuous data were made using Student’s unpaired t-test, P ≤ 0.05 considered significant. Using Student’s paired t-test, P ≤ 0.05 considered significant, within-group comparisons of the time course of deepening hypnosis were observed following propofol administration, as correlated from the raw EEG isoelectricity to both processed EEG monitors prior to temporary clipping of feeding vessels. Comparisons of ordinal data were made using the Mann–Whitney U-test, P ≤ 0.05 considered significant. Interval differences are reported with confidence intervals (CI) of ±95%.

Results

Of 23 patients who were screened, one did not meet eligibility requirements as he was receiving multiple anti-psychotic medications. One patient refused to participate. All 21 recruited patients completed the protocol, for which data collection took place between October 10, 2007 and April 1, 2008. Five patients underwent cerebral aneurysm clipping, while 16 had MVD for trigeminal neuralgia. The two surgical groups behaved very differently on emergence. For this reason, the aneurysm patients were not compared with the MVD patients to assess emergence criteria. Demographics and emergence time data for the five cerebral aneurysm patients are shown in Table 1. There was no difference in demographic characteristics between groups for the 16 MVD patients (Table 2).

Table 1 Demographic data and emergence times for patients undergoing cerebral aneurysm clipping
Full size table
Table 2 Demographics for patients undergoing MVD surgery
Full size table

EEGo and BIS monitoring of burst suppression during cerebral aneurysm surgery

Four of the five patients undergoing elective aneurysm clipping had complete recordings with both EEG monitoring modalities during burst suppression with propofol administration prior to placing temporary vessel clips. The time delay to onset of isoelectricity, as identified by the raw EEG between the two processed EEG monitors, is shown in Table 3. A representative time course comparing the two EEG monitors for the period of burst suppression is shown in Fig. 2.

Table 3 Assessment of burst suppression comparing BIS to EEGo monitors
Full size table
Fig. 2
figure 2

A time chart representing the period of cerebral protection for a patient undergoing elective cerebral aneurysm clipping with burst suppression following propofol administration prior to temporary clipping of the feeding vessel. Bispectral index (BIS) Vista® monitor data taken from downloaded information from a BIS-ready Philips MP70 monitor to Trendface Solo™ software, representing second by second BIS readings in arbitrary BIS units (ABU). Electroencephalogram (EEGo) data can be interpreted based on right-sided y-axis as follows: 3 (Periodic attractor) patient is deeply sedated; 4 (Point attractor) periods of isoelectric EEG with burst suppression (see Fig. 1 for further description). The raw EEG horizontal line represents the length of time that burst suppression was noted by this modality

Full size image

EEGo and BIS monitoring of patients undergoing MVD surgery

Depicted in Fig. 3 is the anesthetic course of one patient who demonstrated arousal during MVD surgery. There were no differences between the BIS and EEGo groups in terms of the time to wake-up, time until orientation, total time to orientation, and time spent in the PACU (Table 4). For patients undergoing MVD, there was no difference in the initial GCS (median value 15, in both groups) in the PACU, the surgeon’s overall impression of emergence (median value 2, in both groups), or emergence quality (median value 8/10, in both groups). There was no difference in the total number of patients requiring narcotics in the recovery room. There was no difference between the two groups in terms of overall patient drowsiness, pain, nausea, cognition, satisfaction with emergence, satisfaction with pain management, nausea/vomiting, and assessment of anesthesia care at 24 hr post-surgery. Additionally, none of the patients reported evidence of intraoperative awareness at 24 hr post-surgery.

Fig. 3
figure 3

Time course of an arousal state in a 66-year old female admitted for increasing trigeminal neuralgia (distribution in all three divisions of cranial nerve V). She was receiving carbamazine. The patient also had a history of hypertension managed with hydrochlorothiazide. At approximately 1 hr and 45 min after induction of anesthesia, the neurosurgeon commented that the patient coughed with incision of the dura. The bispectral index (BIS) output was 43 arbitrary BIS units (ABU). A series of coughs followed, causing bulging of the cerebellum through the dural opening. The patient was immediately deepened by increasing the desflurane concentration, by administering 100 mg of propofol and, subsequently, by administering a further 100 mg of propofol. Hemodynamics (mean arterial pressure; MAP in mmHg), end-tidal desflurane (ETDes; % vapour output), BIS Vista® output in ABU, and anesthetic stage based on EEGo output are shown. Note that the EEGo shows evidence of arousal sooner (change from Stage 3 to Stage 2 depth of sedation) than with the BIS monitor. The patient coughed at 13:19 (noted by the asterisk) with incision of the dura. Following propofol administration to deepen the level of sedation, the EEGo indicated an isoelectric EEG (Stage 2 to Stage 4 depth of sedation) earlier than seen with the BIS Vista®

Full size image
Table 4 Emergence times, PACU times for MVD groups
Full size table

Discussion

This study comparing the EEGo monitor to the BIS monitor in assessing emergence following neurosurgery is the first formal comparison of the EEGo monitor to an existing monitor. Based on a 50% decrease in emergence time with the EEGo monitor, we initially hypothesized that we would require 20 patients to show a statistically significant difference between the two groups. Our study was underpowered to show a difference between processed EEG monitors due to a smaller than anticipated difference in emergence time between groups. The 95% CI of −3.7 to 5.3 min indicating the difference between mean wake-up times closely approximated the presumed clinically important difference of 5 min that we set a priori. We excluded the five patients undergoing cerebral aneurysm clipping from our formal comparisons of emergence times. After performing a number of these cases, we discovered that the two surgery groups were not comparable. Compared to those patients undergoing MVD, the aneurysm patients tended to take longer to wake-up and become oriented. At the time this study was designed, there were a larger number of elective cerebral aneurysm surgeries being performed in our institution. Increasingly, more and more of these patients are having their aneurysms coiled in the radiology suite rather than clipped via surgical intervention. This has resulted in a higher proportion of surgical interventions involving more technically complicated aneurysms. These cases may have longer temporary clamp times, longer overall procedure times, and higher total doses of anesthesia. This may have been a contributing factor to the noticeable differences between the MVD and the cerebral aneurysm patients in terms of emergence times, time to orientation, and time in the PACU.

The following assertion about the BIS monitor is on the Aspect Medical Corporation website, http://www.aspectmedical.com/assets/Documents/MonitoringConsciousnessUsingtheBispectralIndexDuringAnesthesia-PocketGuide.pdf: “A BIS value, while extremely responsive, is not instantaneously altered by changes in clinical status. When abrupt changes occur in hypnotic state—for example, during induction or rapid emergence—the BIS value may lag behind the observed clinical change by approximately 5 to 10 seconds.” While administering propofol to induce an isoelectric EEG with temporary clipping of feeding vessels during aneurysm surgery, our study suggests that the abrupt changes observed in the hypnotic state were displayed more slow, in the BIS monitor, by an average of nearly one order of magnitude. Despite continued burst suppression by raw EEG output, evidence of BIS output above 30 ABU was seen in all examples during the periods of temporary arterial occlusion. Kreuer et al. 4 have demonstrated the inability of the BIS monitor to accurately trend deep levels of desflurane anesthesia.

The EEGo monitor has a processing delay of 50 msec prior to display of the time-delayed raw EEG signal.5 The twice delayed raw EEG signal allows characteristic plots of the signal to be displayed in a continuously rotating 3D box. In the four patients studied who had propofol administered for burst suppression, the EEGo indicated changing depth of sedation by a mean difference of 91 sec more quickly than seen with the BIS Vista® monitor (from 12 to 170 sec for the 95% CI; P = 0.035). When compared to the BIS monitor, the EEGo was more prompt in one incidence at indicating an episode of intraoperative arousal (Fig. 3).

The principal finding of this study indicated no difference between the EEGo and the BIS monitors in facilitating wake-up time, time until orientation, or time spent in the PACU. Additionally, there was no difference in patient satisfaction between the two groups. These results were contradictory to our initial hypothesis. We speculate that there may have been reluctance by the attending anesthesiologist to titrate the anesthetic to the light levels described in the protocol while the patient’s head was still in the Mayfield clamp. Additionally, the transition from periodic attractor (demonstrating an adequate depth of anesthesia; Fig. 1a) to toroidal attractor (demonstrating pre-emergence; Fig. 1b) to chaotic attractor (awake state; Fig. 1c) was very rapid. If vigilance in monitoring the EEGo monitor was not maintained, these rapid changes could have been missed. Rapidly eliminated agents, such as desflurane and remifentanil, potentially make differences in emergence times more difficult to demonstrate than would be the case with longer acting anesthetic agents.

The EEGo has limitations as currently configured. Assessing depth of sedation requires moment-by-moment attention to the graphical output to detect the attendant changes (Fig. 1a–c). Off-line data processing can demonstrate a change in fractal dimension, but is not currently available in real-time. Such a capability would permit trending similar to that seen with the BIS display. Another limitation is that a commercial EEGo monitor is not currently available. We are using beta software provided by the developers of the EEGo. Also, a learning curve to identify the characteristic rotational plots is required. A further limitation is that a sensor strip, as with the BIS, is not available at present (needle electrodes provide the best tracings). Despite the inability to demonstrate a difference in emergence times between the two EEG monitors in this study, the EEGo merits further study as a depth of hypnosis monitor. The EEGo may have merit to identify rapid changes in depth of hypnosis. Potential limitations of both processed EEG monitors are discussed. Some combination of the two processed EEG technologies may be potentially beneficial to aid intraoperative anesthetic management.