Electroencephalographic Characteristics of Emergence from
Propofol/Sufentanil Total Intravenous Anesthesia
Henri S. Traast, MD, and Cor J. Kalkman, MD, PhD
Department of Anesthesiology, Academic Medical Center, University of Amsterdam, The Netherlands
We recorded the electroencephalogram (EEG) in 16 pa- value during the infusion (P > 0.005). MF, F95, and the
tients during propofol/sufentanil total intravenous an- two frequency band power ratios increased during
esthesia to determine whether EEG changes might pre- emergence (P > 0.05). Of the individual spectral vari-
dict imminent awakening during emergence. Changes ables, only a 50% decrease in absolute CYpower was
in absolute and relative power in four frequency bands, more than 90% sensitive and specific in predicting eye
median frequency (MF), 95th percentile frequency opening. We conclude that, although pronounced EEG
(F95), and two frequency band power ratios (P/a and changes occur during emergence from propofoll
((Y + /3)/S) were quantified. One minute before eye sufentanil anesthesia, the EEG does not reliably predict
opening, absolute power in the 6 and (Ybands had de- eye opening.
creased to 49% (25%-73%) and 42% (25%-58%) of the (Anesth Analg 1995;81:366-71)
sufentanil TIVA. In addition, sensitivity and specific-
ropofol, as part of a total intravenous anesthetic
(TIVA) technique, combines fast recovery with ity of these changes in predicting imminent awaken-
minimal hangover effects and a reduced inci- ing were computed.
dence of postanesthetic nausea and vomiting (l-3).
However, as a result of interpatient variability in
propofol kinetics, inadequate levels of anesthesia may Methods
occasionally occur. Routinely used clinical signs may
Sixteen patients (nine female), ASA physical status I,
not always reliably detect inadequate levels of anes-
aged 19-47 yr, undergoing general and orthopedic
thesia (4). Use of muscle relaxants and the relative
surgical procedures of 60-120 min duration, gave in-
hypotension and bradycardia induced by the combi- formed consent to participate in the study, which was
nation of propofol and an opioid may obscure hemo- approved by the Human Studies Committee of our
dynamic and motor signs of inadequate anesthesia. A hospital.
monitoring technique that could provide additional Oral diazepam 10 mg was given 1 h before arrival in
information for the detection of inadequate levels of the operation room. Anesthesia was induced with
propofol anesthesia is desirable. Several approaches propofol 2 mg/kg, and sufentanil 1 pg/kg intrave-
for assessingthe level of anesthesia have been studied, nously. Anesthesia was maintained with propofol 10
including the electroencephalogram (EEG) (5-7), fron- mg * kg-i * h-’ for the first 10 min, 8 mg * kg-’ . h-i
talis muscle, electromyogram (81, and auditory evoked for the next 10 min, and 6 mg * kg-’ . h-’ for the
potentials (9-111, but none of these methods has be- remainder of the procedure (12). Tracheal intubation
come a generally accepted standard. was facilitated by administration of vecuronium 0.1
The present study was designed to characterize the mg/kg, and muscle relaxation was maintained with
EEG changes during emergence from propofol/ intravenous increments of vecuronium l-2 mg as nec-
essary. Ventilation was controlled with a mixture of
oxygen in air (Fio, 40%) to maintain an end-tidal Pco,
of 30-35 mm Hg. Electrocardiogram, noninvasive
Supported by a grant from Zeneca Farma, Ridderkerk, The Neth-
erlands. blood pressure, pulse-oximetry, end-tidal Pco~, Fio,,
This study was presented in part at the American Society of airway pressures, and tidal and minute volumes were
Anesthesiologists annual meeting, October 1991, San Francisco, CA. monitored during the procedure.
Accepted for publication February 24, 1995. If clinical signs of inadequate analgesia were
Address correspondence and reprint requests to Cor J. Kalkman,
MD, PhD, Department of Anesthesiology, Academic Medical Cen- present, as defined by an increase in arterial blood
ter, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. pressure and/or heart rate of more than 20% from
01995 by the International Anesthesia Research Society
366 An&h Analg 1995;81:366-71 0003-2999/95/$5.00
ANESTH ANALG TRAAST AND KALKMAN 367
1995;81:366-71 EEG AND EMERGENCE FROM PROPOFOL/SUFENTANIL ANESTHESIA
baseline, lacrimation, or sweating, a bolus of sufen- The sensitivity and specificity of the EEG descrip-
tanil 10 pg was administered. In addition, a bolus of tors in predicting eye opening were calculated using a
propofol 40 mg was administered if the patient threshold of 5 Hz for MF as described by Schwilden et
moved. The propofol infusion was stopped 5 min after al. (6,16), and a threshold of 14 Hz for F95 as described
completion of the surgical procedure, and residual by Rampil and Matte0 (5). For the other variables, a
neuromuscular block reversed if necessary. Patients value that gave the best combination of sensitivity and
were repeatedly asked to open their eyes at 30-s inter- specificity as determined by a post hoc review of the
vals, and eye opening marked the return of conscious- data was used as a threshold. A threshold of 1.4 was
ness. Stimulating the patient in any other way was used for 6 ratio, and a threshold of 0.5 for BA ratio. For
avoided. absolute 6, (Y, and total power, an amplitude decrease
Propofol plasma levels during emergence were de- to 50% or less of the value at the time of discontinua-
termined in six patients from venous blood samples, tion of propofol was used as threshold. A relative
drawn from a cannula in the arm opposite to the power of 10% or more was used as threshold for
infusion, derived at 5 min intervals from the time of relative p power. Sensitivity was calculated using data
stopping the infusion until eye opening. Propofol on the time points 1 and 3 min prior to eye opening.
plasma concentrations were determined by high- Specificity was defined as the percentage of patients
that, according to the chosen threshold value, was
performance liquid chromatography with fluores-
correctly predicted not to open the eyes within 5 min.
cence detection (13).
Specificity was calculated from EEG spectral variables
After skin preparation, five silver/silver chloride
obtained 5 min after discontinuation of propofol using
cup electrodes were placed according to the interna-
data from patients who were more than 5 min from
tional lo-20 system at F3, F4, Al, A2, and FpZ
eye opening (10 patients).
(ground) (14,15). Electrode impedances were 2000
ohms or less. Two channels of EEG (F3-Al, F4-A2)
were acquired simultaneously using the SM-200 am- Results
plifiers of a Nicolet Pathfinder II (Nicolet Biomedical
Instruments, Madison, WI) and were recorded from The average time between stopping the infusion of
induction until eye opening on a digital tape recorder propofol and eye opening was 12.8 min (9.7-15.8).
(EDR 8000, Earth Data Ltd., Southampton, UK) with a Heart rate decreased from 64 bpm (57-72) to 56 bpm
sample rate of 256 Hz. Band pass filters were set at 0.5 (52-60) (P < 0.05) 1 min before eye opening when
and 30 Hz, and amplifier sensitivity at 200 pV. compared with the moment of stopping the infusion
Power spectral analysis was performed off-line with (15 patients). Mean arterial pressure did not change
an epoch length of 4 s using the fast Fourier frequency during emergence.
analysis software of the Nicolet Pathfinder II. Abso- Reversal of neuromuscular block was needed in one
lute power in four frequency bands, 6 (0.5-3.0 Hz), f3 patient, who received 1 mg of neostigmine and 0.2 mg
of glycopyrrolate before stopping the propofol infu-
(3.25-8.0 Hz), (Y (8.25-13.0 Hz), p (13.25-30.0 Hz), and
sion. The hemodynamic data of this patient were ex-
total power were analyzed. In addition, the following
cluded from analysis.
frequency descriptors were analyzed: median fre-
Propofol plasma concentration measured in six pa-
quency (MF) and 95th percentile frequency (F95).
tients was 3.2 pg/mL (2.2-4.2) at the end of surgery
The ratio of power in the p range to power in the (Y
immediately before discontinuation of propofol, and
range (BA ratio), and the ratio of power in the 8-20 Hz 1.4 pg/mL (1.0-1.9) at the time of eye openin
range to power in the 6 range [6 ratio (15>1 were also -5. T!y
average sufentanil dose used was 0.54 pg * kg *h
To quantify EEG changes during emergence, data Raw EEG traces during the various stages of the
derived from the following 1-min intervals (15 4-s surgical procedure are shown in Figure 1. During
epochs, excluding artifacts) were compared: 1 min steady state propofol/sufentanil TIVA, the EEG was
before stopping the propofol infusion, 5 min after characterized by a bimodal distribution of power with
stopping the propofol infusion, 3 min before eye open- predominance of activity in the 6 band, 51% (43%-
ing, and 1 min before eye opening. 59%) of total power, and the (Y band, 29% (23%-35%)
Data were analyzed by a repeated-measures analy- of total power. Large amplitude variability was ob-
sis of variance and presented as mean and 95% con- served between patients. Figure 2 shows changes in
fidence interval. Absolute EEG power was log- EEG spectral variables during emergence in one pa-
transformed before statistical analysis. Differences tient. EEG spectral variables at the four time points
between means were compared using t-tests with cor- during emergence are shown in Table 1. The most
rection for multiple comparisons and a P value of consistent EEG changes during emergence were pro-
~0.05 was considered significant. nounced decreases in power in the 6 and (Y bands. The
368 TRAAST AND KALKMAN ANESTH ANALG
EEG AND EMERGENCE FROM PROPOFOL/SUFENTANIL ANESTHESIA 1995;81.366-71
Delta Power propofol stop eyes open
stopping propofol infusion
- Alpha Power
1 min prior to eye opening
’ 1 set ’
Figure 1. Raw electroencephalogram (EEG) traces in one patient,
during various stages of the procedure. After induction, the EEG
shows an increase in amplitude and a bimodal distribution of power
with predominance of power in the 6 and the a! band. One minute 7-
prior to eye opening a decrease in slow wave activity is apparent.
- delta-ratio -
decrease in 6 power preceded the decrease in 01 power. 3- BA-ratio -
An increase in p power was observed 5 min after
discontinuation of the propofol infusion. With the ex-
ception of relative /3 power, that increased from 5% -15 -10 -5 0 5 IO 15
(2%-8%) of total power at the moment of stopping the
propofol infusion to 11% (5%-17%) 5 min later, the
power distribution of the spectrum showed only mi- Figure 2. Electroencephalogram (EEG) spectral variables during
nor changes. emergence in one patient. After discontinuation of the propofol
infusion a decrease in 6 power, followed by a decrease in cy power
MF increased from 4.0 Hz (2.7-5.3) at propofol stop becomes apparent. Five minutes after discontinuation of propofol
to 7.2 Hz (5.4-9.0) 1 min before eye opening, and F95 there is an increase in p power, thereafter p power decreases with
from 12.9 Hz (12-14) to 17.2 Hz (15-20) (P < 0.05). BA the overall decrease in EEG amplitude. Median frequency (MF),
95th percentile frequency (F95), and the two frequency band power
ratio increased from 0.3 (0.1-0.4) at the time of stop-
ratios increase during emergence. S ratio = (a + PI/S power; BA
ping the infusion, to 1.2 (0.4-2.0) 3 min, and 1.4 (0.6- ratio = p/a power.
2.2) 1 min before eye opening (P < 0.05). There was
considerable interpatient variability in 6 ratio. The
largest increase in 8 ratio occurred in the first 5 min EEG descriptor (90%). However, sensitivity was only
after stopping the propofol infusion from 1.0 (0.5-1.5) 63%.
at propofol stop to 1.8 (0.7-3.0). The 6 ratio increased
to 2.0 (0.8-3.3) 1 min before eye opening (P < 0.05). Discussion
Table 2 shows sensitivity and specificity of the var-
ious threshold values. A 50% decrease in absolute a! The results of our study indicate that there are pro-
power was the most sensitive (94%) and specific nounced changes in the EEG during emergence from
(100%) in predicting eye opening. A decrease in abso- propofol/sufentanil TIVA. The initial EEG changes
lute 6 power and increase in MF were moderately after discontinuation of propofol were a decrease in
sensitive at the same time point (88%). Using a thresh- slow wave activity, followed several minutes later by
old value of 0.5, the BA ratio was a moderately specific a decrease in 01activity. The decrease of 6 power was
ANESTH ANALG TRAAST AND KALKMAN 369
1995;81:366-71 EEG AND EMERGENCE FROM PROPOFOL/SUFENTANIL ANESTHESIA
Table 1. Electroencephalographic Spectral Variables During Emergence from Propofol/Sufentanil Total
Propofol stop Eyes open Eyes open
Propofol stop + 5 min - 3 min - 1 min
Total power (pV2) 388 (238-632) 295 (187465)* 171 (108-274)* 159 (98-259)*
6 power (pV*) 194 (110-340) 113 (62-205)+ 72 (43-118)” 69 (41-116)*
Rel. 6 power (%) 51 (43-59) 42 (34-50)* 45 (37-53)* 46 (36-57)
0 power (pV*) 52 (33-81) 47 (31-71) 26 (1642)* 21 (1240)*
Rel. 8 power (%) 15 (12-18) 16 (14-19) 17 (13-21) 16 (12-20)
a power (r*V2) 100 (61-165) 86 (53-141) 38 (20-72)* 23 (9-61)*
Rel. a! power (%) 29 (23-35) 31 (25-37) 25 (18-33) 21 (15-27)*
P power (PV~) 14 (7-28) 19 (lo-38Y 18 (9-33) 21 (1141)
Rel. p power (%) 5 (2-8) 11 (5-171% 13 (8-18Y+ 16 (11-22)”
Median frequency (Hz) 4.0 (2.7-5.3) 6.1 (4.6-7.7)* 5.8 (4.4-7.2)* 7.2 (5/l--9.0)*
95th percentile frequency (Hz) 12.9 (12-14) 14.0 (13-16)* 15.2 (13-17)* 17.2 (15-20)*
BA ratio 0.3 (0.1-0.4) 0.5 (0.1-0.9) 1.2 (0.4-2.0)* 1.4 (0.6-2.2)*
6 ratio 1.0 (0.5-l .5) 1.8 (0.7-3.0)* 1.9 (l.O-2.8Y 2.0 (0.8-3.3Y+
Data are mean (95% confidence limits).
Rel. = relative power, % of total power; BA ratio = P/CI power; 6 ratio = (01 + @/S power.
* P < 0.05 versus propofol stop.
Table 2. Sensitivity and Specificity of Electroencephalogram (EEG) Spectral Variables in the Prediction of Eye Opening
Eye opening, Eye opening, (%): propofol
EEG spectral variable Threshold 3 min 1 min stop + 5 min
Absolute 6 power 50% decrease 50 88 60
Absolute (Y power 50% decrease 63 94 100
Total power 50% decrease 56 81 90
Relative /3 power 10% increase 56 75 80
Median frequency 5.0 Hz (6,16) 62 88 60
95th percentile frequency 14 Hz (5) 60 73 78
BA ratio 0.5 50 63 90
6 ratio 1.4 64 57 78
BA ratio = /~/IY power; 6 ratio = (01 + 0)/S.
reflected in an increase of both MF and F95. Com- in /3 power during emergence is related to a change
parable EEG changes have been reported during in benzodiazepine concentration.
emergence from isoflurane/nitrous oxide anesthesia Although the present study demonstrates a consis-
(7,15). Both authors observed a decrease in amplitude tent pattern of decreasing absolute power and an in-
and a shift toward faster frequencies after disconti- crease in relative amount of fast activity in most pa-
nuation of isoflurane. However, during emergence tients, the sensitivity and specificity of the individual
from fentanyl/nitrous oxide anesthesia only minimal spectral descriptors was limited. An exception was
changes in 6 ratio, MF, and F95 were observed, and absolute (Ypower; that proved the best predictor of eye
they were only apparent in the final minute before eye opening with a sensitivity of 94% and a specificity of
opening (15). 100%.
The transient increase in absolute p power ob- Because of the difficulty in interpreting EEG ampli-
served in our study 5 min after stopping the propo- tude changes, most authors have emphasized fre-
fol infusion may be related to the high amplitude p quency variables rather than power variables to meas-
activity observed during propofol sedation (17-19). ure anesthetic depth. Schwilden et al. (6,16) reported
The increase in /3 activity might then be considered that the median frequency showed the greatest degree
to mark the transition from hypnotic propofol con- of discrimination of all spectral variables. A MF lower
centrations to the sedative concentration range. Ben- than 5 Hz was always associated with unconscious-
zodiazepines in sedative doses are known to in- ness. Although in our study MF changed from 4.0 Hz
crease /3 activity in awake patients (20). The oral to 7.2 Hz during emergence, a MF threshold of 5 Hz
diazepam premeditation might have contributed to did not reliably predict eye opening.
the amount of absolute p activity during surgery. Similarly, the spectral edge frequency has been used
However, it is not likely that the transient increase as an indicator of anesthetic depth in several studies
370 TRAAST AND KALKMAN ANESTH ANALG
EEG AND EMERGENCE FROM l’ROPOFOL/SUFENTANIL ANESTHESIA 1995;81:366-71
(5,7,14,21,22). Rampil and Matte0 (5) reported that a exclude that some of our patients were “light.” The
spectral edge frequency of less than 14 Hz was asso- value of the frequency variables MF and F95 at the
ciated with a minimal hemodynamic response to in- time of stopping the propofol infusion was not rele-
tubation after induction of anesthesia with thiopental. vant to the calculation of sensitivity at 1 and 3 min
In our study F95 changed from 12.9 to 17.2 Hz during before eye opening. For the calculation of sensitivity of
emergence, yet a threshold of 14 Hz was not very absolute band powers, we used the power changes
sensitive in predicting eye opening. One possible ex- relative to the time of stopping the propofol infusion.
planation might be that with progressively lower EEG Here low power at the time of stopping the propofol
amplitude the contribution of random noise increased infusion (an “awake type” EEG) might decrease the
the variability of the percentile frequencies. This might apparent sensitivity of the variable.
also explain the observation that in several patients In conclusion, the present data demonstrate that
MF and F95 increased initially but decreased in the discontinuation of propofol administration as part of a
final minutes before eye opening. Patient movement TIVA technique is accompanied by minor hemody-
and frowning may contaminate the EEG with electro- namic changes, decreases in power in the (Y and 6
myogram activity, especially with frontal placement band, and a shift toward higher frequencies. The de-
of electrodes. This can result in an apparent increase in crease in amplitude is more prominent than the shift
p activity. However, no movement or bucking was toward faster frequencies, and results in a gradual
observed in any patient until the time of eye opening. shift of MF and F95 toward higher frequencies after
Long et al. (15) suggested that the ratio between fast discontinuation of propofol. The sensitivity and spec-
((Y + /3) and slow (6) activity might be an indicator of ificity of the individual spectral variables in predicting
anesthetic depth and found an increase in 6 ratio to be actual eye opening was limited, with the exception of
more abrupt and of larger magnitude than an increase absolute a! power. Nonetheless, the data suggest that
in median frequency or spectral edge during emer- the EEG may be of use during propofol TIVA in
gence from isoflurane anesthesia. In our study the alerting the anesthesiologist to an inadvertently de-
maximum increase in 6 ratio occurred in the first 5 min creasing propofol concentration.
after stopping the propofol infusion. Thereafter, only a
slight increase was observed until eye opening. This We are grateful to Dr. John C. Drummond MD, Professor of Anes-
suggests that the decrease in 6 activity is an early thesiology, University of California, San Diego, for critically review-
indicator of a diminishing propofol concentration. ing the manuscript and providing us with his valuable comments.
Because power in the 6 and the CY band decreased in
parallel, a ratio of power in the p band divided by
power in the (Y band (BA ratio) seemed a reasonable References
approach to detect changes in the relative contribution
1. Korttila K, Nuotto EJ, Lichtor JL, et al. Clinical recovery and
of the relevant frequency bands during emergence psychomotor function after brief anesthesia with propofol or
from propofol/sufentanil anesthesia. The magnitude thiopental. Anesthesiology 1992;76:676-81.
of change in the BA ratio was more pronounced than 2. Nightingale JL, Lewis IH. Recovery from day-case anaesthesia:
comparison of total iv. anaesthesia using-propofol with an
that of MF and F95 in the period before eye opening. inhalational techniaue. Br I Anaesth 1992;68:356-9.
A threshold value of 0.5 was highly specific (94%), 3. Watcha M, Simeonk White I’, Stevens J.’ Effect of propofol on
but the sensitivity at 3 and 1 min before eye opening the incidence of postoperative vomiting after strabismus sur-
gery in pediatric outpatients. Anesthesiology 1991;75:204-9.
was only 50% and 63%, probably as a result of the
4. Russell IF. Midazolam-alfentanil: an anaesthetic? An investiga-
increasing contribution of noise with decreasing EEG tion using the isolated forearm technique. Br J Anaesth 1993;70:
Although we used a standard propofol infusion 5. Rampil I, Matte0 R. Changes in EEG spectral edge frequency
correlate with the hemodynamic response to laryngoscopy and
regimen, it is not likely that all patients were at a intubation. Anesthesiology 1987;67:139-42.
similar “level of anesthesia” at the time of stopping 6. Schwilden H, Stoeckel H, Schuttler J. Closed-loop feedback
the infusion. Interindividual differences in propofol control of propofol anaesthesia by quantitative EEG analysis in
humans. Anaesthesist 1989;38:180-8.
pharmacokinetics may have resulted in considerable
7. Drummond JC, Brann CA, Perkins DE, Wolfe DE. A comparison
differences in propofol plasma levels, whereas differ- of median frequency, spectral edge frequency, a frequency band
ences in pharmacodynamics might have been respon- power ratio, total power, and dominance shift in the determi-
sible for variability in EEG responses, even in patients nation of depth of anesthesia. Acta Anaesthesiol Stand 1991;35:
who had similar plasma levels of propofol. Four pa- 8. Herregods L, Rolly G, Mortier E, et al. EEG and SEMG moni-
tients had a MF higher than 5.0 Hz at the time of toring during induction and maintenance of anesthesia with
stopping the propofol infusion. These patients had full propofol. Int J Clin Monit Comput 1989;6:67-73.
9. Newton DE, Thornton C, Creagh BP, Dore CJ. Early cortical
recovery of neuromuscular transmission, but did not
auditory evoked response in anaesthesia: comparison of the
move during skin closure. There appears to be no gold effects of nitrous oxide and isoflurane. Br J Anaesth 1989;62:
standard for “adequate anesthesia,” and we cannot 61-5.
ANESTH ANALG TRAAST AND KALKMAN 371
1995;81:366-71 EEG AND EMERGENCE FROM PROPOFOL/SUFENTANIL ANESTHESIA
10. Thornton C, Konieczko K, Jones JG, et al. Effect of surgical 17. Drummond JC, Iragui MV, Alksne JF, Kalkman CJ. Masking of
stimulation on the auditory evoked response. Br J Anaesth epileptiform activity by propofol during seizure surgery. Anes-
1988;60:372-8. thesiology 1992;76:652-4.
11. Schwender D, Keller I, Klasing S, Madler C. Middle-latencv 18. Veselis R, Reinsel R, Marino I’, Wronski M. EEG power spec-
auditory evoked potentials durmg high-dose opioid analgesia. trum changes during propofol sedation. Anesthesiology 1991;
Anaesthesist 1990:39:299-305. 75:A 181.
12. Roberts FL, Dixon J, Lewis GT, et al. Induction and maintenance 19. Seifert HA, Blouin RT, Conard PF, Gross JB. Sedative doses of
of propofol anaesthesia. A manual infusion scheme. Anaesthe- propofol increase beta activity of the processed electroenceph-
alogram. Anesth Analg 1993;76:976-8.
13. Plummer GF. Improved method for the determination of propo-
20. Entholzner E, Schneck HJ, Hargasser S, et al. Electroencephalo-
fol in blood by high-performance liquid chromatography with
graphic demonstration of central nervous system effects of dif-
fluorescence detection. J Chromatogr (Biomed Appl) 1987;421:
171-6. ferent premeditation regimens. Anaesthesist 1994;43:431-40.
14. White PF, Boyle WA. Relationship between hemodynamic and 21. Stone DJ, DiFazio CA. Anesthetic action of opiates: correlations
electroencephalographic changes during general. anesthesia. of lipid solubility and spectral edge. Anesth Analg 1988;67:
Anesthesiologv 1989;70:177-81. 663-6.
15. Long CW, Shah NK; Loughlin C, et al. A comparison of EEG 22. Withington I’. Assessment of power spectral edge for monitor-
determinants of near-awakening from isoflurane and fentanyl ing depth of anaesthesia using low methohexitone infusion. Int
anesthesia. Spectral edge, median power frequency, and delta J Clin Monit Comput 1988;3:117-22.
ratio. Anesth Analg 1989;69:169-73.
16. Schwilden H, Stoeckel H. Quantitative EEG analysis during
anaesthesia with isoflurane in nitrous oxide at 1.3 and 1.5 MAC.
Br J Anaesth 1987;59:738-45.