Journal of Clinical Monitoring and Computing (2010) 24:283–288 DOI: 10.1007/s10877-010-9246-0
Springer 2010
SNAP II VERSUS BIS VISTA MONITOR COMPARISON DURING GENERAL ANESTHESIA
Hrelec C, Puente E, Bergese S, Dzwonczyk R. SNAP II versus BIS VISTA monitor comparison during general anesthesia. J Clin Monit Comput 2010; 24:283–288
Candace Hrelec, BS, Erika Puente, MD, Sergio Bergese, MD and Roger Dzwonczyk, MS
ABSTRACT. Introduction. Effectively monitoring the level of consciousness during general anesthesia is clinically beneficial to both the patient and the physician. An electroencephalogram (EEG)-based level-of-consciousness monitor can help minimize intraoperative awareness as well as the effects of over-sedation. In this study, we compared the SNAP II (Stryker Instruments, Kalamazoo, MI USA) and BIS VISTA (Aspect Medical Systems, Newton, MA USA) monitors’ primary metrics (SI and BIS, respectively) in terms of correlation, agreement and responsiveness to return to preoperative baseline in surgical cases involving general anesthesia. Methods. With institutional approval and written informed consent, 33 patients received general anesthesia with isoflurane while undergoing abdominal surgery. We attached both the SNAP II and BIS VISTA electrodes to each patient. We collected data from each monitor simultaneously and continuously, beginning just prior to induction and ending after extubation. Each monitor’s levelof-consciousness index is a unit less metric that ranges from 0 to 100, with 100 indicating full consciousness. We performed a Bland–Altman and parameter difference analyses on the data. We calculated the time it took for each monitor to return to preoperative baseline level following cessation of anesthesia. We established an equivalence between the two indices over their entire range for our particular clinical scenario. Result. The indices were correlated (r = 0.736, P < 0.0001, N = 3,706 data point pairs). There was an overall difference between the two indices (median = 16.0, 25th/75th%ile = 10.0/21.1) with BIS lower than SI. A 40-60 BIS range (the typical target range during general anesthesia) was approximately equivalent to a 54–74 SI range. In all 33 subjects, SI reached baseline before BIS at the end of the case (median = 3.3 min, 25th/75th%ile = 1.6 min/ 8.2 min versus median = 8.9 min, 25th/75th = 3.7 min/ 14.5 min, P = 0.0200), even though both metrics were equal at the beginning of the case. Discussion. Although the SI and BIS both can assess a patient’s level of consciousness and are correlated, they are not in agreement with each other numerically and therefore are not interchangeable. It is difficult to assess each monitor’s true responsiveness to acute changes in consciousness level from our study design. The differences between the metrics we observed in this study are most likely due to differences in signal processing methodologies, EEG frequencies employed and signal filtering utilized in the monitors. KEY WORDS. BIS VISTA, SNAP II, EEG monitoring, level of consciousness.
From the Department of Anesthesiology, The Ohio State University, 410 West 10th Avenue, Columbus, OH 43210, USA. Received 27 April 2010. Accepted for publication 5 July 2010. Address correspondence to R. Dzwonczyk, Department of Anesthesiology, The Ohio State University, 410 West 10th Avenue, Columbus, OH 43210, USA. E-mail:
[email protected]
INTRODUCTION
Monitoring consciousness during general anesthesia is beneficial to both the physician and the patient. Effective
284
Journal of Clinical Monitoring and Computing
brain activity monitoring during surgery can prevent episodes of under-sedation and intraoperative awareness [1–3]. Close brain monitoring during general anesthesia allows the anesthetist to dose patients appropriately according to consciousness parameters in order to prevent over-sedation and its side effects, thereby decreasing recovery time from anesthesia [2]. Monitoring the level of consciousness during general anesthesia with EEG-based monitors has become the standard of care in anesthesia practice and has proven to be cost-effective [4–6]. There are several devices on the market that measure level of consciousness during anesthesia, using parameters derived from the electroencephalogram (EEG) signal. The BIS monitor (Aspect Medical Systems, Newton, MA USA) is the most familiar of these monitors and has been discussed in the literature for many years; it has proven to correlate well with level of consciousness including both the loss and subsequent recovery. The BIS VISTA, the newest version of the monitor, derives its bispectral index (BIS) metric from bispectal analysis of the EEG signal. The SNAP II monitor (Stryker Instruments, Kalamazoo, MI USA) is a newer product on the market that combines spectral analysis of selected EEG frequency bands to derive its SNAP index (SI). The SNAP II algorithm avoids including frequency bands that are thought to be contaminated by muscle activity and environmental noise signals [7–12]. Both devices generate a dimensionless index (BIS and SI) that ranges from 0 to 100, with 100 representing complete consciousness and 0 representing an isoelectric EEG. Using the BIS monitor, a patient under general anesthesia is typically represented in the 40–60 range. Recent studies have indicated that the SNAP II monitor is also a useful indicator of consciousness [13–15]. Wong et al. [7] compared the SNAP II with the BIS XP, an earlier monitor model, in a study using sevoflurane and nitrous oxide. They concluded that the monitors are correlated, but the SNAP II may be a better indicator for measuring acute return to awareness than the BIS. In this study, we compared the SI with the BIS, derived from the new BIS VISTA, in terms of correlation, agreement, sensitivity and responsiveness to return to preoperative baseline in surgical cases involving general anesthesia.
MATERIALS AND METHODS
Brief technical description of the monitors The BIS VISTA and SNAP II are designed to monitor hypnotic state or level of brain activity in patients undergoing anesthesia and surgery. These machines are
considered level-of-consciousness monitors. The devices provide an indication of depth of anesthesia during surgery. The devices extract data from the ongoing EEG signal and derive an empirical dimensionless metric that ranges from 0 to 100, where 100 indicates a fully awake/ fully conscious brain state. Each device uses a proprietary algorithm to generate its metric. The BIS VISTA uses bispectral analysis signal processing as the basic tool to generate the BIS. Unlike power spectrum analysis, bispectral analysis retains information about the phase of the various EEG components in the frequency domain and is perhaps better in analyzing nonlinear systems [16]. The SNAP II derives the SI from frequency analysis of both the high frequency (80–420 Hz) and low frequency (0.1–18 Hz) components of the EEG signal. The SNAP II monitor specifically ignores frequencies in the 40–80 Hz range, which are predominated by muscle activity and other signals that interfere with the EEG signal.
Monitor configuration and data format Both monitors generate their primary index internally at a per-second rate. The SNAP II monitor provides essentially no user control over the SI data that are displayed on the screen or exported from the monitor. For instance, there are no user-controllable filters available to smooth the data. The SNAP II outputs the time-stamped SI to a compact flash memory card at a per-second rate. We used this data for our study. The BIS VISTA monitor, on the other hand, allows the user to smooth the BIS value displayed on the screen and exported from the monitor with one of three moving-average filters. In this study, we used the 15 s-wide moving-average filters to smooth the BIS. The BIS VISTA monitor exports live time-stamped BIS data at a per-second rate and historical data at a perminute rate to a universal serial bus flash drive. Each minute’s historical BIS data is the overall average of the smoothed per-second data over the previous minute. We used the historical per-minute data for this study.
Experimental procedure This study had institutional review board approval. All patients who enrolled in this study gave informed written consent to participate and signed a Healthcare Insurance Portability and Accountability Act form. The BIS VISTA and SNAP II monitors are approved for patient use by the United States Food and Drug Administration and had been safety tested by our hospital’s clinical engineering department before use. All patients who participated were 18–85 years old, ASA I-II and scheduled for abdominal surgery under general
Hrelec et al.: SNAP II versus BIS VISTA
anesthesia with isoflurane. The surgeries included open or laparoscopic procedures, inpatient or outpatient procedures, and procedures that were anticipated to be less than 4 h in duration. Patients failing to meet all inclusion criteria, patients who did not sign the informed consent, and patients with evidence of recent trauma, active infection, neurological disorder, seizure disorder, dementia, currently taking psychoactive medications as part of routine care, or with a history of drug or alcohol abuse were excluded from participating in the study. We attached the EEG electrodes of each monitor to the patient’s forehead using the procedure specified by the manufacturers. We randomized on which side of the forehead we placed the electrodes so that each monitor had an equal chance of monitoring from the left or right side of the forehead. The patients acted as their own control. Because the BIS monitor is a standard-of-care in our hospital, it was the only one of the two EEG monitors visible to the clinical staff in the operating room during the surgery. We synchronized the real-time clocks of the monitors to each other prior to the beginning of each case. We recorded SI and BIS data continuously and simultaneously from just prior to induction until just past patient awakening and extubation at the end of the surgical procedure. The periodic electrode impedance testing feature on each monitor was disabled to avoid possible interference with the measurements. We tested electrode impedance at the start and end of each case to ensure the integrity of the electrode contact with the skin. We marked the time of the following surgical/anesthesia events during the case: • • • • • • • • • •
Mask placed on patient; 1 min baseline recording; Induction; Intubation; Isoflurane on; Maintenance time (10 min post induction); Incision; Close following surgical procedure; Discontinue isoflurane; First purposeful response (mOAAS = 3; mRamsey = 3); • Extubation. Following baseline BIS and SI measurements, the anesthesiologist administered midazolam 0.02 mg/kg for pre-medication. Anesthesia was induced with propofol 2 mcg/kg, fentanyl 100 mcg, and succinylcholine 2 mg/ kg. The exact induction time was specified as the time propofol was administered. The subject was then intubated and maintenance of anesthesia was controlled with isoflurane with delivery stabilized to the desired end tidal
285
level. Maintenance time was defined as 10 min post propofol administration. Vital signs were controlled throughout the surgery with esmolol, nicardipine, phenylephrine, and glycopyrrolate. Esmolol and glycopyrrolate were administered if the heart rate increased over 30% or decreased under 30% of baseline, respectively. Nicardipine and phenylephrine were administered if blood pressure increased over 30% or decreased under 30% of baseline, respectively. Neuromuscular blockade was maintained with a reversible agent, vecuronium or rocuronium, with the agent of choice left to the discretion of the anesthesiologist.
Data processing and analysis The time-stamped BIS was used as exported from the BIS VISTA post-operatively, in the smoothed per-minute format, in all analyses in this study. The time-stamped SI was exported from the SNAP II post-operatively in persecond format. For the Bland–Altman [17] and difference analyses, we transformed SI into a per-minute measurement format by averaging each minute’s SI values (60 measurements) and aligning this averaged SI value time wise with the BIS. For the sensitivity analysis, we determined the time it took for each metric to return to its baseline value after anesthesia was suspended. This analysis was performed with the data as exported post-operatively from the monitors. We defined the baseline window for each metric as the variance around the central value of the metric measured at baseline prior to anesthetizing the patients. We determined the time it took each metric to rise above the lower edge of the baseline window at the end of the clinical procedure. We performed an ANOVA analysis to compare the measurements made by each monitor at each defined time point. For this analysis, significance was set a priori at alpha = 0.05. In order to establish an equivalence between SI and BIS, we calculated the difference between the two metrics over incremental ranges of BIS, specifically 0–10, 11–20, 21–30…91–100, as well the difference between the two metrics over the 40–60 BIS range, the typical BIS target range for patients under general anesthesia.
RESULTS
Index comparison Thirty-three patients completed the study. Our data were not normally distributed. We therefore described our data
Journal of Clinical Monitoring and Computing
as medians and percentiles, and analyzed our data with nonparametric statistical tests. The Bland–Altman analysis is shown in Figures 1 and 2. SI and BIS were correlated (r = 0.736, P < 0.0001, N = 3,706 data point pairs) but were not in agreement with each other, with BIS lower in value, on overall average, than SI (median = 16.0, 25th/75th %tile = 10.0/ 21.0). Only 7.4% of the BIS measurements were higher than the corresponding SI measurements at any time during the cases. At the peak of the difference distribution (Figure 3), 7.8% of the 3,706 SI measurements were 36.1% higher in numeric value than the corresponding BIS measurements. Figure 4 shows a graph of SI and BIS measurements versus event time. The data was not normally distributed. Using ANOVA on ranks and Dunn’s post hoc pair-wise testing, we found no statistical difference between SI and BIS at the defined events.
Bland-Altman Analysis SNAP II v BIS VISTA
80
60
40
Difference
286
20
0
-20
-40
-60
N = 3,706 Cases = 33
Difference v Mean
20
40
Sensitivity analysis: return to baseline
Bland-Altman Analysis SNAP II v BIS VISTA
80
SNAP II
80
100
Fig. 2. Bland-Altman analysis graph of SI-BIS difference versus SI-BIS mean.
Difference Analysis SNAP II v BIS VISTA Percent of SI Measurements
There was no statistical difference between SI and BIS at baseline (SI: median = 92.7, 10th/25th/75th%tile = 86.1/88.4/96.2 versus BIS: median = 94.0, 10th/25th/ 75th%tile = 86.2/91.2/95.2, P = 0.517). For the sensitivity analysis, we defined the lower edge of the baseline
100
60
Mean
60
SI > BIS 88.6% of Mesurements
SI < BIS 7.4% of Measurements
10
8
6
4
2 N = 3,706 Cases = 33
0 -100
-75
-50
-25
0
25
50
75
100
SI Measurement Difference From BIS (%)
40 r = 0.736 p < 0.0001 N = 3,706 Cases = 33
20 SNAP II v BIS XP Regression 95% Prediction Interval Unity Line
20
40
60
BIS VISTA Fig. 1. Bland-Altman analysis graph of SI versus BIS.
80
100
Fig. 3. Difference analysis of SI versus BIS. 88.6% of the SI measurements were greater than the corresponding BIS measurements. The distribution was centered around a +36.1% difference between SI and BIS.
window for each metric as the 10th%ile of each metric’s median value (SI = 86.1; BIS = 86.2) (Figure 5). Under this criterion, the SI reached its baseline window before the BIS following cessation of anesthesia (mean = 5.0 min ± 4.2 min versus 11.3 min ± 5.9 min, P < 0.001).
Hrelec et al.: SNAP II versus BIS VISTA
range. This analysis indicated that a 54–74 SI range is equivalent to a 40–60 BIS range (SI-BIS difference: median = 14.3, 25th/75th %ile = 7.9/19.2) under the clinical conditions of this experiment.
SNAP II v BIS VISTA
100 BIS Lower Edge Baseline = 86.2 SI Lower Edge Baseline = 86.1
80
DISCUSSION
60
40
20 Median 25th/75th %ile
Extubation
First Response
Close
Incision
Maintenance
ISO On
Intubation
No Pairwise Differences
Induction
Baseline
Mask On
SNAP II BIS VISTA
ISO off
Level of Consciousness
120
Event Time (Relative) Fig. 4. BIS and SI at each defined time point in the surgical procedure. There was no statistical difference between BIS and SI at any time point.
SI v BIS Equivalence
70
Indice Difference (SI - BIS)
287
Difference in the 40-60 Range BIS Ranges
60 50 40 30 20 10 0 -10
91-100
81-90
71-80
61-70
51-60
41-50
31-40
21-30
11-20
0-10
Median 25th/75th%ile
BIS Range Fig. 5. Equivalence between SI and BIS over the 0–100 range (light gray bars). The 40–60 BIS range (dark gray bar), the typical target range for anesthetized patients, is equivalent to 54–74 SI range.
Equivalence Figure 5 discloses the equivalence relationship between SI and BIS over the BIS range of values. With the exception of the 0–10 BIS range, the difference between SI and BIS was smaller at higher metric values than at lower values. BIS was essentially always lower than SI over the entire
We have demonstrated that BIS, as derived by the new BIS VISTA monitor, and SI, derived by the SNAP II, are correlated with each other but are not directly interchangeable. SI is virtually always higher in value than the corresponding BIS despite being equivalent at the beginning of each case prior to a patient’s exposure to general anesthesia. The data we collected here have allowed us to establish a general equivalence between the two level-of-consciousness metrics over their entire 0–100 range under our particular clinical conditions (surgery and drug regimen). Of particular importance, we elucidated that the typical 40–60 BIS target range is equivalent to a 54–74 SI range. This difference should not be interpreted as a universal alignment factor between the two monitors, as one might do to align two blood pressure monitors. Although the two indices are both derived from the same EEG signal, the monitors use different signal processing strategies on different parts of the spectra and aspects of the signal, and only both report their indices on a 0–100 scale by convention. They are not measuring exactly the same thing, even though they are both marketed as level-of-consciousness monitors. SI returns to its baseline window faster than BIS. From Figure 4 we see that, on average, SI enters its baseline window before the patient is awake and able to respond to stimulus, whereas BIS is still below the baseline window at first response. Wong et al. [7] made this same observation in their study comparing SNAP II with the earlier BIS XP monitor. The BIS VISTA monitor may utilize or process certain segments of the EEG frequency spectrum in a way that maintains the effects of sedation or anesthesia for a longer period in comparison to the SNAP II. At anesthesia cessation, both metrics appear to rise in a similar fashion and at a similar rate. Since BIS is lower in value than SI at the end of anesthesia, it must increase numerically more than SI to reach the baseline window. This simple fact may be another reason why SI returns to baseline before BIS. Our data set did not allow us to elucidate which monitor responds faster to acute unintentional awareness, since we used historical rather than real-time measurements. Although both monitors display data in real time on a per-second rate, the BIS VISTA monitor uses an adjustable moving-average filter to smooth the displayed BIS. Any filtering done to the SI is unadjustable. The BIS
288
Journal of Clinical Monitoring and Computing
VISTA’s moving average process acts like a low-pass filter that will smooth the data over time and attenuate rapid changes in the BIS [18]. The BIS VISTA offers three levels of filtering: 10, 15 and 30 s moving average filtering. Even the 10 s filter—the filter with the highest cutoff frequency (lowest filtering effect)—will, to some degree, attenuate acute changes in the BIS, but would be the best setting to use to detect acute awareness. To answer the question of which monitor is more sensitive to acute awareness, we must compare real-time monitor data available to the anesthesiologist during a case, not averaged historical data.
CONCLUSION
The SNAP II SI and BIS VISTA BIS metrics both indicate level of consciousness and are correlated. The two metrics are not numerically interchangeable.
This study was funded by the Department Anesthesiology.
REFERENCES 1. Monk TG, Saini V, Weldon BC, Sigl JC. Anesthetic management and one-year mortality after noncardiac surgery. Anesth Analg 2005; 100: 4–10. 2. Sebel PS, Lang E, Rampil IJ, White PF, Cork R, Jopling M, Smith NT, Glass PS, Manberg P. A multicenter study of bispectral electroencephalogram analysis for monitoring anesthetic effect. Anesth Analg 1997; 84: 891–899. 3. Joint Commission on Accreditation of Healthcare Organizations. Preventing, and managing the impact of, anesthesia awareness. Sentinel Event Alert 2004; 32. 4. White PF, Tang J, Ma H. Is the patient state analyzer with the PSArray2 a cost-effective alternative to the bispectral index monitor during the perioperative period? Anesth Analg 2004; 99: 1429–1435.
5. Rampil IJ. A primer for EEG signal processing in anesthesia. Anesthesiology 1998; 89: 980–1002. 6. Dressler O, Schneider G, Stockmanns G, Kochs EF. Awareness and the EEG power spectrum: analysis of frequencies. Br J Anaesth 2004; 93: 806–809. 7. Wong C, Fragen R, Fitzgerald P, McCarthy R. A comparison of the SNAP II and BIS XP indices during sevoflurane and nitrous oxide anesthesia at 1 and 1.5 MAC and at awakening. Br J Anaesth 2006; 97: 181–186. 8. Bruhn J, Boullion TW, Schafer SL. Electromyographic activity falsely elevates the bispectral index. Anesthesiology 2000; 92: 1485–1487. 9. Wong CA, Fragen RJ, Fitzgerald PC, McCarthy RJ. The association between propofol-induced loss of consciousness and the SNAP index. Anesth Analg 2005; 100: 141–148. 10. Sing HC, Kautz MA, Thorne DR, Hall SW, Redmond DP, Johnson DE, Warren K, Bailey J, Russo MB. High-frequency EEG as a measure of cognitive function capacity: a preliminary report. Aviat Space Environ Med 2005; 76: 114–135. 11. Draguhn A, Traub RD, Schmitz D, Jeffreys JG. Electrical coupling underlies high-frequency oscillations in the hippocampus in vitro. Nature 1998; 394: 189–192. 12. Vanhatalo S, Volpio J, Kaila K. Full-band EEG (fbEEG): a new standard for clinical electroencephalography. Clin EEG Neurosci 2005; 36: 311–317. 13. Schmidt GN, Standl T, Lankenau G, Hellstern A, Hipp C, Bischoff P. SNAP-index and bispectral index during induction of anaesthesia with propofol and remifentanil. Anesthesiol Intensivmed Notfallmed Schmerzther 2004; 39: 286–291. 14. Schmidt GN, Bischoff P, Standl T, Lankenau G, Hellstern A, Hipp C, Schulte am Esch J. SNAP index and bispectral index during different states of propofol/remifentanil anaesthesia. Anaesthesia 2005; 60: 228–234. 15. Ruiz-Gimeno P, Soro M, Pe´rez-Solaz A, Carrau M, Belda FJ, Jover JL, Aguilar G. Comparison of the EEG-based SNAP index and the bispectral (BIS) index during sevoflurane-nitrous oxide anaesthesia. J Clin Monit Comput 2005; 19: 383–389. 16. Sigl JC, Chamoun NG. An introduction to bispectral analysis for the electroencephalogram. J Clin Monit 1994; 10: 392–404. 17. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307–310. 18. Stanley WD. Realization and frequency response of discretetime systems. In digital signal processing (1st ed.). 5 Chap, Reston Publishing Company, Inc.: Reston, VA, 1975.
