The Effect of Propofol on Thermal Pain Perception

Michael A. Fro lich, MD, MS, Donald D. Price, PhD, Michael E. Robinson, PhD,
Jonathan J. Shuster, PhD, Douglas W. Theriaque, MS, and Marc W. Heft, DMD, PhD
From the Department of Anesthesiology, University of Alabama at Birmingham, Birmingham, Alabama, the Department
of Oral and Maxillofacial Surgery and Diagnostic Sciences, University of Florida College of Dentistry, and Department of
Clinical Health and Psychology, University of Florida College of Health Related Professions, Gainesville, Florida
We studiedthe effect of propofol, a widelyusedsedative-
hypnotic drug, on pain perception. Eighteen subjects re-
ceived propofol in two sedative concentrations that were
balanced and randomized in order. Painful (45C, 47C,
and 49C) stimulation temperatures were presented in
randomorder, andnonpainful 31Cstimuli were pre-
sented on alternate trials. We used a target-controlled
infusion and chose effect site concentrations of 0.5
g/mL for mild sedation and 1.0 g/mL for moder-
ate sedation. Using a visual analog scale, subjects
rated both pain intensity and unpleasantness higher
when sedated with propofol. The average pain inten-
sity was 28/100 for placebo, 35/100 for mild, and 40/
100 for moderate sedation. Pain unpleasantness was
23/100 for placebo, 29/100 for mild, and 33/100 for
moderate sedation. This effect was unexpected and
may be explained by a difference of subjective pain
experience by a patient and the perceived level of an-
algesia by a health care provider in sedated patients.
This finding calls further attention to the need for ad-
equate analgesia in patients sedated with propofol.
(Anesth Analg 2005;100:4816)
T
he use of propofol as an IV sedative for moni-
tored anesthesia care is widespread. It is used in
the emergency room as a sedative for many short
but painful procedures such as anoscopy, fracture re-
duction, abscess incision, or cervical dilation and cu-
rettage (1,2). It is used extensively for sedation of
tracheally intubated patients in the intensive care unit
(3). Other applications include various nonsurgical
diagnostic procedures, such as gastrointestinal endos-
copy and bronchoscopy, which require sedation (4).
Although the sedative- pharmacokinetic profile of
propofol is well described, there is only sparse and
somewhat conflicting literature on the effect of propo-
fol on pain perception.
A discussion in the anesthesiology literature illus-
trates that this is not just an academic question (5,6).
Authors discussed whether inadequate analgesia, of-
ten manifested as restlessness in the sedated patient,
should be treated with supplemental bolus doses of
propofol or analgesic medication. The widespread use
of propofol for many medical and surgical procedures
and for sedation in the intensive care environment
underscores the clinical relevance of this question. We
therefore decided to assess the effect of propofol se-
dation on pain perception using a well established
model that is reliable, reproducible, and accurately pre-
dictive of changes and intensities of clinical pain (7).
Methods
After approval by the institutional review board 18
healthy subjects were recruited by public advertise-
ment. Each subject was scheduled for a screening ap-
pointment. After informed consent was obtained, sub-
ject history, physical examination, and a complete
blood count and chemistry panel were obtained. Fe-
male subjects were scheduled for a urine pregnancy
test on the day of the study. The study procedure was
explained in detail to volunteers. Several thermal pain
stimuli were given and subjects were asked to rate the
pain intensity and unpleasantness on a visual analog
scale (VAS). Volunteers were included if their medical
screening was unremarkable and they were able to
follow study instructions. Exclusion criteria were any
medical condition that could affect the study proce-
dure or potentially put the subject at risk. Examples of
exclusion conditions are obesity, sleep apnea, moder-
ate or severe bronchial asthma, and cardiovascular
problems such as hypertension. Subjects taking any
Supplemental material available at anesthesia-analgesia.org.
Supported, in part, by funds of National Institutes of Health K-30
and GCRC MO1-RR00082 at the University of Florida.
Accepted for publication July 26, 2004.
Address correspondence and reprint requests to Michael A.
Fro lich, MD, MS, Department of Anesthesiology, 619 South 19th
Street, Birmingham, AL 35249-6810. Address e-mail to froelich@
uab.edu.
DOI: 10.1213/01.ANE.0000142125.61206.7A
2005 by the International Anesthesia Research Society
0003-2999/05 Anesth Analg 2005;100:4816 481
drugs or substances that would alter their pain per-
ception were not allowed to participate in this study.
Ultimately, 7 female and 11 male subjects were en-
rolled. None of the subjects was obese (weight range,
43 to 90 kg), their height ranged from 156 to 193 cm,
and their age ranged from 19 to 28 yr.
Stimulus temperatures were 45C, 47C, and 49C,
temperature levels that have been shown to activate
A and C fibers. After each of the thermal pain stimuli,
presented in random order, subjects were asked to
rate sensations with a mechanical slide algometer.
Subjects were instructed to rate the perceived pain
intensity and unpleasantness using visual analog
scales (VAS) whose end points are no pain sensa-
tion, most intense pain sensation imaginable, and
not at all unpleasant, most unpleasant imaginable
(7). We used the TSA-II Neuro Sensory Analyzer (Me-
doc, Ltd., Ramat-Yishai, Israel) for the administration
of thermal pain stimuli. This device consists of a com-
puter driven heat exchanger that regulates the tem-
perature of a water circuit. The circuit perfuses a 30
30 mm thermode, which was attached to the forearm.
The TSA-II Neuro Sensory Analyzer is controlled us-
ing a personal computer-based stimulation software
that we programmed to a staircase test algorithm.
Four, 4 s square wave painful target temperatures
(45C, 47C, and 49C) alternating with a neutral tem-
perature (31C) were programmed. The ramp-up tem-
perature increase was 10C/s. The three target tem-
peratures were randomly assigned within each testing
period and were maintained for 4 s, followed by a 30 s
interval during which pain ratings took place. This
cycle of 3 painful stimuli, each followed by the subject
pain rating, was performed 3 times under each study
condition (placebo, mild and moderate propofol drug
levels), providing a total of 27 measurements for each
subject. After each thermal stimulus, the skin contact
site was moved by 3 cm within the upper section of
the forearm in a systematic fashion to avoid redness of
the skin and stimulus habituation.
Subjects underwent a three period crossover study
involving randomization to drug (mild or moderate
propofol) or placebo. Within each of these 3 treatment
periods, 3 temperature stimuli (45C, 47C, 49C) were
randomly administered in a set of 3 replicates (nine
measurements per period). All subjects received both
levels of drug as well as placebo. Figure 1 shows a
graphical representation of a stimulation sequence for
one subject. Four replicates of a 3 3 Greco-Latin
squares design were used to achieve approximate bal-
ance to both drug and temperature assignments, and
SAS 6.1 software (Cary, NC) was used to perform the
randomization.
Subjects were blinded with respect to their treat-
ment. The IV line leading up to the IV catheter was
covered to avoid any visual clues that would help
subjects to learn about their treatment. During the
placebo condition, the propofol target-controlled infu-
sion was set to zero and thus did not infuse. The
crystalloid carrier infusion was maintained. Also, the
IV pole equipped with both carrier and propofol in-
fusion was placed behind the subject and visually
separated by a curtain. Two research nurses were
involved; the nurse responsible for the administration
of thermal pain stimuli and the recording of VAS
ratings, for obvious reasons, had access to the thermal
testing sequence. Another nurse, who recorded vital
signs and withdrew blood at the appropriate times,
was not informed about the testing sequence.
Controlling for psychomotor impairment is impor-
tant when the VAS for pain rating is used by sedated
subjects. At the beginning of each testing sequence,
subjects were asked to perform a task intended to test
cognitive and psychomotor ability. Rather than rating
a painful thermal stimulus, subjects were given a ran-
dom number using the VAS scale, which has no gra-
dation visible to subjects; however, the flip side of the
rating scale has a gradation that allows the study
administrator to obtain an exact reading of the rating.
For this test, subjects were told that the VAS scale
ranged from 0 to 100 mm. Random numbers were
limited to a range of 20 and 80 mm, thus avoiding the
extremes of the scale that might be more difficult to
use. The difference of the intended number rating by
subjects and the actual number (visible to the study
administrator) was calculated for each propofol con-
dition. If the difference of intended versus actual num-
ber rating was significantly larger during the sedation
conditions compared to the placebo (control), subjects
psychomotor ability would be considered impaired.
The infusion of propofol followed the three-
compartment target site infusion model adjusted for
age proposed by Schnider et al. (8). This model is
included in the software STANPUMP by Steven L.
Shafer, MD. This personal computer-based program
was used to drive a Graseby 3400 Infusion pump
(Graseby Medical Limited, Watford, United Kingdom).
Figure 1. Example of thermal pain testing sequence. Sample testing
sequence for one subject. At each propofol condition painful ther-
mal stimuli were applied to the right forearm (R) alternating with
the left forearm (L). The sequence of the three painful stimulation
temperatures is shown in the lower section of the figure.
482 PAIN MEDICINE FRO LICH ET AL. ANESTH ANALG
PROPOFOL AND PAIN PERCEPTION 2005;100:4816
The effect site concentrations were 0.5 g/mL and
1.0 g/mL. These doses are in the lower therapeutic
range and will produce mild and moderate sedation.
Based on propofols short context sensitive half-life, a
time frame of 20 min was deemed sufficient to establish
a new plateau effect site concentration for all possible
drug concentration changes in this study. Thus, the
propofol infusion was set to deliver one of the three
propofol conditions (no infusion, mild and moderate
sedation). Twenty minutes after each rate adjustment
thermal testing is performed (Fig. 1). Blood samples to
determine propofol levels were obtained before each
pain testing sequence. Samples were analyzed using liq-
uid chromatography-mass spectroscopy (9).
All subjects were monitored according to the stan-
dards of the American Society of Anesthesiologists
using pulse oximetry, electrocardiogram (ECG), and
noninvasive blood pressure in addition to inspection
of subjects breathing and circulation (10).
Sample size calculations were performed for the
primary outcome related to VAS pain intensity rat-
ings, as baseline data were available in the literature.
These calculations assume a moderate within-subject
correlation of 0.75, 0.05, and two-tailed tests. We
calculated that 12 subjects would need to be tested to
detect an 8 mm difference in VAS pain intensity rat-
ings at 80% power. We decided to enroll a total of 18
patients to compensate for potential missing data, but
we were able to gather a complete set of data.
The methodology used was a repeated-measures
parametric analysis of variance, which is described in
detail as follows. The dependent variables for all anal-
yses were intensity and mean unpleasantness VAS
ratings. For the treatment comparisons, we calculated
each patients global mean score for each treatment,
yielding 54 dependent observations (18 subjects 3
treatments). This averaging process improved the
power as compared to analyses that do not do this
averaging. To limit the errors of multiple testing, we
first conducted randomized block analyses for three-
way treatment differences, with the subject used as the
block. For each of the dependent variables, if the three
treatment F-statistic was nonsignificant at P 0.05, all
statistical comparisons would cease. If significant,
pairwise comparisons of treatments and secondary
3-group analysis within each of the individual tem-
peratures would be done. This secondary 3-way anal-
ysis further determined if pairwise treatment compar-
isons within temperature would be necessary. Other
analyses, which are viewed as diagnostic, were also
completed. We looked at the interaction between tem-
perature and experimental group, asking if the mean
difference between treatments depended upon tem-
perature. No significant interactions were uncovered.
Finally, as a diagnostic, we studied the role of the
order of the treatments to see if the analysis might
need to take that into account, but no significant order
effect was uncovered.
Results
We found overall statistically significant differences in
pain intensity and unpleasantness ratings among
study conditions (Tables 1 and 2 and Fig. 2). Pain
intensity was rated significantly higher under mild (P
0.01) and moderate (P 0.001) sedation when com-
pared with the placebo condition. In a similar fashion,
pain unpleasantness was rated significantly higher
under mild (P 0.003) or moderate sedation (P
0.001) when compared with the placebo condition.
Mean VAS scores increase by 12 points on a scale
ranging from 0 to 100 for intensity and by 10 points for
unpleasantness when comparing placebo and moder-
ate sedation.
Subjects general cognitive ability under propofol
sedation was not impaired. The difference between
numbers given to subjects and numerical values of
VAS ratings of subjects did not significantly increase
with sedation (see also number rating test in Meth-
ods). Mean differences between presented 0100 num-
bers and 0100 VAS ratings were 4.3 (3.3 sd) for the
Table 1. Analysis of Variance Tables
Source dF
Sum of
Squares
Mean
Square F Value
Partial
Eta Squared
Dependent Variable: VAS Intensity Rating
Subject 17 13892 817 15.94 (P .001) 0.89
Sedation Group 2 1263 631 12.32 (P .001) 0.42
Error 34 1743 51
Dependent Variable: VAS Unpleasant Rating
Subject 17 12946 762 17.85 (P .001) 0.90
Sedation Group 2 913 457 10.70 (P .001) 0.39
Model 19 13859 729 17.1
Error 34 1450 43
Analysis of variance tables are based on a univariate general linear model with pain ratings as dependent variable and sedation group and subjects as
independent variable. Effect sizes are displayed as partial eta squared; 42% (intensity) and 39% (unpleasantness) of variations on pain ratings can be explained
by propofol sedation.
ANESTH ANALG PAIN MEDICINE FRO LICH ET AL. 483
2005;100:4816 PROPOFOL AND PAIN PERCEPTION
placebo condition, 4.7 (2.9 sd) for mild sedation, and
5.6 (3.5 sd) for moderate sedation. These mean differ-
ences were very small and did not differ statistically
(analysis of variance, P 0.48).
The finding of increased pain ratings with propofol
sedation were confirmed by our secondary analysis
that was based on the correlation of pain VAS ratings
and plasma propofol levels (random effects analysis).
Propofol plasma levels showed a significant positive
association with pain ratings (Table 3).
Discussion
We have demonstrated that propofol, at an effect site
concentration of 0.5 g/mL and 1.0 g/mL, increases
both pain intensity and pain unpleasantness. Most
anesthesiologists are more accustomed to linear infu-
sion rates; the sedative doses chosen in this study
would approximately equal linear infusion rates of 25
and 50 g kg
1
min
1
. These concentrations of
propofol are used for sedation during regional anes-
thesia, uncomfortable diagnostic procedures, such as
endoscopies, reposition of a dislocated joint, or inter-
ventional procedures in radiology where sedation is
desired (2).
Some reports describing the effects of propofol on
pain perception in humans were published in 1995
(11). The change in thermal pain detection thresholds
was studied in healthy patients scheduled for ortho-
pedic surgery under epidural anesthesia who received
thiopentone or propofol. Thresholds did not increase
significantly, either for a small (0.5 mg/kg bolus and
1 mg kg
1
h
1
infusion) or large (0.5 mg/kg bolus
and 5 mg kg
1
h
1
infusion) dosage range in 15
subjects who received propofol. Anker-Mller et al.
(12) found significant increases in pain detection
thresholds associated with argon laser stimulation af-
ter an IV bolus of 0.25 mg/kg propofol. These results
have led some physicians to believe that painful con-
ditions could be treated with small bolus doses of
propofol (5), a strategy that has been frowned upon by
many other clinicians who maintain that propofol
should not be used to treat inadequate analgesia (6).
Indeed, there is evidence that points toward hyperal-
gesic effects of propofol (13,14). In a study involving
12 volunteers, Petersen-Felix et al. (14) demonstrated
that propofol in subhypnotic doses has hyperalgesic
effects on mechanical pressure pain, whereas pain
perception in response to electrical or heat stimulation
appeared unaltered.
The finding that propofol increases pain perception
is somewhat counterintuitive. Patients sedated with
propofol appear less responsive to painful stimuli to
observers and one might even assume analgesic prop-
erties of the sedative drug. Yet, based on this study,
we have to recognize that subjective pain experience
may in fact be increased. Hyperalgesic effects of
propofol may not be well recognized if patients are
simply not able to recall a painful and/or unpleasant
procedure. The hyperalgesic effects may well be clin-
ically significant if they are systematically present in
large numbers of patients. This underscores the im-
portance of assuring appropriate analgesia if propofol
sedation is chosen as adjuvant medication and cer-
tainly argues against the treatment of pain with addi-
tional doses of propofol as proposed by some clini-
cians (5).
Table 2. Summary of Pairwise Comparisons of Visual
Analog Scale Ratings
Pairwise Comparison of Means Averaged over
Temperature
Variable No. Mean sd Pr t
Intensity
ModeratePlacebo 18 11.8 11.4 .001
MildPlacebo 18 6.8 10.1 0.011
ModerateMild 18 5 8.6 0.026
Unpleasantness
ModeratePlacebo 18 10 9.91 .001
MildPlacebo 18 5.9 7.2 0.0028
ModerateMild 18 4.1 10.3 0.11
Data presented here are mean and sd of pain ratings and the level of
significance for their pairwise comparison.
Figure 2. Visual analogue scale (VAS) pain ratings comparing pla-
cebo to mild and moderate sedation. The bar graphs displays VAS
Ratings as means sd. Significantly different ratings are labeled
with
a
P 0.011,
b
P 0.001,
c
P 0.026,
x
P 0.003,
y
P 0.001.
Table 3. Pain Scores and Propofol Levels
Treatment
Quartiles
(25, 50, 75)
Mean
(sd)
P value
(two-sided)
Intensity 0.5, 16, 20 9.2 (39.9) 0.034
Unpleasantness 0.1, 12, 21 11.5 (28.8) 0.016
Tabulated are fitted mean pain score change per unit dose increase.
484 PAIN MEDICINE FRO LICH ET AL. ANESTH ANALG
PROPOFOL AND PAIN PERCEPTION 2005;100:4816
Additional evidence for hyperalgesic properties of
sedative-hypnotic medications is provided by animal
studies. Ewen et al. (13) describe the hyperalgesic
effects of barbiturates and propofol in the rat, thus
indicating that hyperalgesia may be a property of
different anesthetics when administered in subhyp-
notic doses. A proposed mechanism for this effect is
the modulation of the central gamma aminobutyric
acid (GABA)
A
ionophore by drugs such as barbitu-
rates (15) This modulation of GABA
A
receptors may
well be the pathophysiologic mechanism for a central
sensitization of noxious stimuli by propofol. This fac-
tor might play a major role for the development of
pain after medical procedures performed with propo-
fol sedation and may help to explain why sedated
critically ill patients report frequent unpleasant
events, which they thought had taken place before
they regained consciousness (16). In fact, more than
half of the patients in the intensive care unit actively
recall pain (17). Similar considerations apply for many
endoscopic procedures, procedures in interventional
radiology, the emergency room, and some office-
based surgical procedures performed under local an-
esthesia with sedation.
The proposed mechanism leading to central mech-
anism of noxious stimuli by propofol may also be
present at larger, hypnotic doses of propofol. Al-
though pain is considered the unpleasant sensory and
emotional experience associated with actual or poten-
tial tissue damage, there may be important implica-
tions of enhanced processing of noxious stimuli dur-
ing propofol anesthesia as a state during which the
subjective experience of such a stimulus is presumably
suppressed. The study of this hypothesis might be an
interesting area of future research.
A potential limitation of our study is the variation in
sedation levels among subjects. However, the prospec-
tive randomized and balanced treatment assignment
should have helped to minimize these differences as
well as minor pharmacokinetic differences in drug
washout among subjects. We were also concerned
about oversedation that might affect the subjects abil-
ity to perform pain rating. Unfortunately, some inves-
tigators do not attempt to address this potential con-
founding variable (12). We have used the number
rating test to compare our subjects ability to use the
mechanical slide algometer under different sedation
conditions. We believe that this test, although not well
established, has reasonable face validity.
We observed a significant variability in subjects
pain rating when exposed to identical stimulation
temperatures. Based on this finding, one might argue
that stimuli should be normalized using pain thresh-
old measurements. However, there are important ad-
vantages of using unadjusted suprathreshold pain;
first, normal variability in pain sensitivity is captured,
and second, suprathreshold pain is the more relevant
to assess clinical pain.
A potential confounding factor for the assessment of
pain in patients receiving propofol is its ability to
cause burning on injection. Fortunately, none of our
subjects reported propofol injection pain, a fact that
we attributed to the careful selection of the injection
site at a large caliber vein. We also changed forearms
two times and received consistent rating within sub-
jects. Thus, pain rating appeared to be unaffected by
the possible local irritant effects of propofol.
The reason for the conflicting results in studies eval-
uating the effect of propofol on pain perception may,
at least in part, be attributed to study design and
methods. We therefore chose an experimental model
of pain and rating scale that have been extensively
used in both clinical and experimental contexts (7,18
20). In particular, the combined use of mechanical
VAS and contact heat-induced pain has been shown to
provide ratio scale measures of pain and internally
consistent measures of experimental and clinical pain
when both forms of pain are rated by pain patients
(7,18,20). This method is also predictive of changes in
clinical pain intensity. For example, conventional an-
algesic treatments such as opioid administration have
been shown to produce similar magnitudes of pain
reduction in both clinical and this form of experimen-
tal pain (20).
In summary, our findings indicate that propofol in
mild to moderate sedative doses increases pain inten-
sity and unpleasantness. This finding calls attention to
the need for adequate analgesia in sedated patients
and stimulates the continuing discussion about the
pharmacologic profile of anesthetic drugs and their
mechanism of action. Further research is necessary to
determine the effects of larger, hypnotic doses on pain
experience.
References
1. Havel CJ Jr., Strait RT, Hennes H. A clinical trial of propofol vs
midazolam for procedural sedation in a pediatric emergency
department. Acad Emerg Med 1999;6:98997.
2. Swanson ER, Seaberg DC, Mathias S. The use of propofol for
sedation in the emergency department. Acad Emerg Med 1996;
3:2348.
3. Chamorro C, de Latorre FJ, Montero A, et al. Comparative study
of propofol versus midazolam in the sedation of critically ill
patients: results of a prospective, randomized, multicenter trial.
Crit Care Med 1996;24:9329.
4. Carlsson U, Grattidge P. Sedation for upper gastrointestinal
endoscopy: a comparative study of propofol and midazolam.
Endoscopy 1995;27:2403.
5. Lau WC, Green CR, Faerber GJ, et al. Determination of the
effective therapeutic dose of intrathecal sufentanil for extracor-
poreal shock wave lithotripsy. Anesth Analg 1999;89:88992.
6. TerRiet MF, Jacobs JS, Lewis MC, DeSouza GJ. Propofol and
analgesia. Anesth Analg 2000;90:1455.
7. Price DD, Long S, Harkins SW. A comparison of pain meas-
urement characteristics of mechanical visual analogue and sim-
ple numerical rating scales of pain. Pain 1994;56:21726.
ANESTH ANALG PAIN MEDICINE FRO LICH ET AL. 485
2005;100:4816 PROPOFOL AND PAIN PERCEPTION
8. Schnider TW, Minto CF, Gambus PL, et al. The influence of
method of administration and covariates on the pharmacokinet-
ics of propofol in adult volunteers. Anesthesiology 1998;88:
117082.
9. Lim CK, Lord G. Current developments in LC-MS for pharma-
ceutical analysis. Biol Pharm Bull 2002;25:54757.
10. Basic anesthetic monitoring. Standards, guidelines and state-
ments. Park Ridge, IL: American Society of Anesthesiologists,
2003.
11. Wilder-Smith OH, Kolletzki M, Wilder-Smith CH. Sedation
with intravenous infusions of propofol or thiopentone: effects
on pain perception. Anaesthesia 1995;50:21822.
12. Anker-Moller E, Spangsberg N, Arendt-Nielsen L, et al. Sub-
hypnotic doses of thiopentone and propofol cause analgesia to
experimentally induced acute pain. Br J Anaesth 1991;66:18558.
13. Ewen A, Archer DP, Samanani N, Roth SH. Hyperalgesia dur-
ing sedation: effects of barbiturates in the rat. Can J Anaesth
1995;42:53240.
14. Petersen-Felix S, Arendt-Nielsen L, Bak P, et al. Psychophysical
and electrophysiological responses to experimental pain may be
influenced by sedation: comparison of the effects of a hypnotic
(propofol) and an analgesic (alfentanil). Br J Anaesth 1996;77:
16571.
15. Staley K. Enhancement of the excitatory actions of GABA by
barbiturates and benzodiazepines. Neurosci Lett 1992;146:
1057.
16. Rundshagen I, Schnabel K, Wegner C, am Esch S. Incidence of
recall, nightmares, and hallucinations during analgosedation in
intensive care. Intensive Care Med 2002;28:3843.
17. Puntillo KA. Pain experiences of intensive care unit patients.
Heart Lung 1990;19:52633.
18. Price DD, Harkins SW. Combined use of experimental pain and
visual analogue scales in providing standardized measurement
of clinical pain. Clin J Pain 1987;3:18.
19. Price DD, Harkins SW, Baker C. Sensory-affective relationships
among different types of clinical and experimental pain. Pain
1987;28:297300.
20. Price DD, Harkins SW, Rafii A, Price CA. Simultaneous com-
parison of fentanyls analgesic effects on experimental and clin-
ical pain. Pain 1986;24:197203.
486 PAIN MEDICINE FRO LICH ET AL. ANESTH ANALG
PROPOFOL AND PAIN PERCEPTION 2005;100:4816