Propofol Blood Concentration and the Predict Suppression of Learning During Propofol/Epidural in Volunteers

Kate Leslie, MBBS, FANZCA, Daniel I. Sessler, MD, Marc Schroeder, BA, and Kristin Walters, BA Thermoregulation Research Laboratory, Department of Anesthesia, University of California, San Francisco, California

Propofol is often used for sedation during regional an- reference, C,; ground, mastoid). Propofol caused con- esthesia. We tested the hypothesis that propofol blood centration-related impairment of learning. The propo- concentration, the Bispectral Index and the 95% spectral fol blood concentration suppressing learning by 50% edge frequency predict suppression of learning during was 0.66 + 0.1 pg/mL. The Bispectral Index value when propofol/epidural anesthesia in volunteers. In addi- learning was suppressed by 50% was 91 2 1. In contrast, tion, we tested the hypothesis that the Bispectral Index the 95% spectral edge frequency did not correlate well is linearly related to propofol blood concentration. with learning. The Bispectral Index decreased linearly Fourteen healthy, male volunteers were studied on as propofol blood concentration increased (Bispectral three randomly ordered days: no propofol, target Index = -7.4 * [propofol] + 90; ? = 0.47, n = 278). propofol blood concentration 1 pg/mL, and target There was no significant correlation between the 95% propofol blood concentration 2 pg/mL. Each day, epi- spectral edge frequency and propofol concentration. In dural anesthesia (-Tll level) was induced using 2% order to suppress learning, propofol blood concentra- 2-chloroprocaine. Propofol was infused by a computer- tions reported to produce amnesia may be targeted. Al- controlled pump, and propofol concentration meas- ternatively, the Bispectral Index may be used to predict ured in central venous blood. We administered a Trivial anesthetic effect during propofol sedation. Pursuit@-type question task on all 3 days. The electroen- cephalogram was monitored continuously (Fe,, FP2; (Anesth Analg 1995;81:1269-74)

earning is a relatively long-lasting adaptive behav- Alternatively, electroencephalography (EEG), which ioral change occurring as a result of experience. offers real-time of central nervous system L Learning may be either explicit (accompanied by activity, might predict learning during anesthesia. This conscious recall), or implicit (occurring without conscious would avoid the problem of predicting actual blood recall). Both types of learning during anesthesia may concentrations of a drug from those targeted. Recently, a have untoward physical and psychologic sequelae (1). new EEG indicator, the Bispectral Index, was proposed Propofol impairs learning and produces sedation in as a measure of anesthetic effect (5). This index quantifies low doses. It is advocated for sedation during regional the nonlinear relationships between EEG component anesthesia (2,3). Although in sufficient concentrations it waves, as well as analyzing their frequency and ampli- may produce respiratory and cardiovascular compro- tude. The Bispectral Index can predict movement after mise (4) as well as unnecessarily deep hypnosis. Inade- painful stimulation during anesthesia (5,6), but it is not quate concentrations may be associated with intra- known whether it predicts suppression of learning. We operative recall. More appropriate propofol blood therefore tested the hypothesis that propofol blood con- concentrations can be targeted if the concentration-effect centration, the Bispectral Index, and the 95% spectral relationship for propofol and suppression of learning is edge frequency predict suppression of learning during better defined. propofol/epidural anesthesia in volunteers. The EEG usually exhibits a biphasic response to in- creasing anesthetic concentration; that is, there is activa- This study was supported in part by the Joseph Drown Founda- tion, followed by slowing of the EEG. This phenomenon tion and National Institutes of Health Grant GM49670. has confounded attempts to predict anesthetic effect us- Accepted for publication July 14, 1995. ing the EEG, as one EEG variable value may be associ- Address correspondence to Daniel I. Sessler, MD, Thermoregula- tion Research Laboratory, University of California, San Francisco, ated with two different anesthetic concentrations (7,8). San Francisco, CA 94143-0648. The Bispectral Index (version 3.0) incorporates features

01995 by the International Anesthesia Research Society 0003.2999/95/$5.00 Anesth Analg 1995;81:1269-74 1269 1270 LESLIE ET AL. ANESTH ANALG PROPOFOL, EEG, AND LEARNING 1995;81:1269-74

designed to overcome this problem. We therefore tested (13). Combined data from hypothermic and normother- the hypothesis that the Bispectral Index is linearly re- mic volunteers were used to program the pump, using lated to propofol blood concentration. the method of Plasma Drug Efflux (14). Oxygen (4 L/min via nasal cannulae) was administered as nec- Methods essary to maintain oxygen saturations >95%. Fifteen minutes after commencement of the infu- With approval from the Committee on Human Re- sion, we administered the learning tasks. Headphones search at the University of California, San Francisco, were placed over the volunteers’ ears, and we con- we studied 14 healthy male volunteers. None reported firmed that output volume was adequate. The volun- taking prescription medications or illegal drugs. The teers were instructed to listen carefully to the tape. A volunteers’ height was 181 rt 7 cm (mean 5 SD), total Trivial Pursuit@-type question task, of 3 min duration, body mass 75 -+ 11 kg, lean body mass 60 + 7 kg, and was administered at all three test concentrations. age 28 + 5 yr. Volunteers were studied on three randomly ordered Learning Task days: control (no propofol), a target propofol blood concentration of 1 pg/mL, and a target propofol blood The Trivial Pursuit@-type questions were based on a concentration of 2 pg/mL. Volunteers were blinded to learning task devised by Chortkoff et al. (15). They the target concentration. The propofol blood concen- conducted a preliminary test in 100 students to estab- tration preventing movement after a painful stimulus lish a pool of Trivial Pursuit@-type questions of appro- in 50% of subjects (Cp50) is >8 pg/mL (9,lO). After priate difficulty and interest (e.g., “How tall was the completion of memory testing each day, volunteers world’s tallest hairdo?“). We selected three sets of 10 participated in a study evaluating the thermoregula- questions and presented them to our volunteers in tory effects of propofol (11). Epidural anesthesia was random order on the three study days. During anes- induced, allowing volunteers’ core temperatures to be thesia, volunteers were presented with answers to six manipulated without altering their sentient skin questions, repeated three times. After recovery from temperatures. anesthesia, the 10 questions were presented in a five- option, multiple-choice format. Protocol Measurements The volunteers fasted for 8 h before arriving at the laboratory; during studies, they rested supine on a The EEG was recorded continuously during each standard operating-room table. The room lights were study. Data analyzed were collected when the volun- dimmed and extraneous noise was minimized. teers’ eyes were closed. After preparation of the skin An L2-3 epidural catheter and right internal jugular with OmniprepTM (D.O. Weaver and Co., Aurora, CO), vein catheter were inserted using standard techniques. gold cup electrodes were filled with conductive paste An antecubital vein in the nondominant arm was can- (EC2TM; Grass Instrument Co., Quincy, MA) and at- nulated for fluid and drug administration. Routine tached to the skin with collodion adhesive. The elec- anesthesia monitors were attached and were used to trodes were positioned at FP1 and FpZ, with the refer- monitor standard anesthetic variables, including ence electrode at C, and the ground electrode at the blood pressure, heart rate, and oxygen saturation. mastoid. A microprocessor-based, four-channel mon- Induction of epidural anesthesia was preceded by itor (B500; Aspect Medical Systems, Inc., Natick, MA) 1 h of cutaneous warming to prevent redistribution was used to collect raw EEG data which were re- hypothermia (12), and we infused 15 mL/kg lactated corded on digital tape for off-line analysis (version 3.0) Ringer’s solution. Epidural anesthesia then was in- by Aspect Medical Systems, Inc. duced using 2% 2-chloroprocaine. A test dose of 3 mL Raw EEG data were inspected for artifact by a with epinephrine l:lOO,OOO was followed by slow ad- blinded observer, and spurious data deleted. Both the ministration of 20 mL 2% 2-chloroprocaine without power spectrum and the Bispectral Index were calcu- epinephrine, to produce a dermatomal level of sensory lated continuously. Version 3.0 of the Bispectral Index block near Tll. Subsequently, the sensory level was combines improved artifact detection algorithms to maintained with an infusion of 2% 2-chloroprocaine correctly identify awake artifacts (eye blinks and administered at a rate approximately 20 mL/h. movement, and baseline wander), and improved dy- Propofol was infused using a pump (Ohmeda 9000; namic range in the light sedation range, and was Ohmeda, Steeton, England) controlled by a computer applied prospectively to our data. programmed to target propofol blood concentrations of Core temperature was measured at the tympanic 1 and 2 pg/mL. The pharmacokinetic data were derived membrane (Man-a-Therm@; Mallinckrodt Anesthesi- from a previous study of propofol pharmacokinetics ology Products, Inc., St. Louis, MO). The aural probe during mild hypothermia in young healthy volunteers was inserted until felt by the volunteer next to the ANESTH ANALG LESLIE ET AL. 1271 199531 1269-74 PROPOFOL, EEG, AND LEARNING

membrane; appropriate placement was confirmed by gently rubbing the wire. The probe then was taped in position and the canal occluded with cotton wool. 75 0 0 Only data collected when the volunteers’ core temper- atures were 35.0-38.O”C were used in the analysis. Correct Central venous blood was sampled for meas- Responses50 0 urement of propofol blood concentration just prior to (%I memory testing, 5 min later, and at 15-min intervals thereafter. Three-milliliter samples were stored in he- 0 0 0 parinized tubes at 4°C for up to 10 wk (propofol blood concentrations decrease less than 0.2% per week at 4”C), and later analyzed using a high-pressure liquid chromatography assay, modified from the method of Plummer (16). This assay is linear to at least 20 pg/mL [propof bg / ml) and has a detection limit of 0.025 pg/mL and a coef- Figure 1. Propofol blood concentration (pg/mL) versus correct re- ficient of variation of 4.1% at 2 Fg/mL. sponses to Trivial Pursuit@-type questions corrected for control (%). The propofol blood concentration suppressing learning by 50% was 0.66 t- 0.1 pg/mL (nonlinear regression line with 95% confidence Data Analysis limits shown). All questions were answered correctly by all volun- teers when no propofol was given. The percentage of control Trivial Pursuit@-type ques- tions answered correctly over the 3 days was calcu- lated for each volunteer, and then averaged for the poor EEG signal quality and artifact. Variation in group. This quantified the volunteers’ ability to guess propofol concentration over the time of memory test- the correct answer. The number of presentedquestions ing was 0.47 ? 0.48 pg/mL. correct then was corrected for control: The results Propofol caused concentration-related impairment were thus negative when the volunteers’ answers to of learning, as measured by the Trivial Pursuit@-type presented questions were worse than control. question task. Control questions were answered with a frequency a little greater than random guessing % correct = (~28%). In order to successfully fit the nonlinear re- (% presented questions correct gression function, all negative results were set at zero. - % control questions correct) The Cp50 for propofol-induced suppression of learn- (100 - % control questions correct) ’ loo ing was 0.66 + 0.1 pg/mL (Figure 1). The Bispectral Index decreased as learning was suppressed. When Negative results were set at zero. Nonlinear regression the volunteers’ responses to 50% of questions were was used to fit a sigmoid equation to the data and correct, the Bispectral Index was 91 -t 1 (Figure 2). In generate Cp50 values for suppression of learning by contrast, the 95% spectral edge frequency correlated propofol (JMP; SAS Institute, Cary, NC). The same poorly with suppression of learning (Figure 3). It was analysis was used to compare the Bispectral Index and not possible to fit a sigmoid curve to these data. 95% spectral edge frequency with suppression of The Bispectral Index was related to propofol blood learning. concentration by the e uation: Bispectral Index = Linear regression, with residual analysis, was used to - 7.4 * [propofol] + 90 (Y9 = 0.47, P = 0.0001, y1 = 278) correlate propofol blood concentration and Bispectral (Figure 4). In contrast, there was no significant relation Index or 95% spectral edge frequency. Step-wise linear between the 95% spectral edge frequency and propo- regression was used to define the influence of propofol fol blood concentration (? = 0.094) (Figure 5). concentration and core temperature on the Bispectral Core temperatures between 35°C and 38°C had an Index. Results are presented as mean Ilt SD, unless spec- insignificant effect on the Bispectral index (Bispectral ified otherwise; P < 0.05 was considered statistically Index = 45 * [propofol] + 7.7 * core temperature - significant. 1.4 * ([propofoll * core temperature) - 193). Results Discussion Induction and maintenance of propofol sedation was smooth in all cases; none of the volunteers required Propofol produced a concentration-related suppres- airway support. Epidural-induced dermatomal level sion of learning during epidural anesthesia in volun- for loss of temperature discrimination was Tll ? 1 teers, and therefore may be suitable for suppressing (range T9-12). One volunteer did not attend his third memory of perioperative events. The Bispectral Index study day, and another’s data was rejected because of correlates with both suppression of learning and 1272 LESLIE ET AL. ANESTH ANALG PROPOFOL, EEG, AND LEARNING 1995,81 1269-74

80 Correct Responses 50 Bispectral (%) Index 60 25

0 1, 60 70 80 90 100 0 2 4 6 8 Bispectral Index [propof (&ml) Figure 2. Bispectral Index versus correct responses to Trivial Figure 4. Propofol blood concentration (&mL) versus Bispectral Pursuit@-type questions corrected for control (%). The Bispectral Index (linear regression line with 95% confidence limits shown). The Index value when learning was suppressed by 50% was 91 k 1 BisnectralI Index was linearlv related to nronofol blood concentra- (nonlinear regression line with 95% confidence limits shown). tion (Bispectral Index = 7:4 . [propofolj +bO; r2 = 0.47, n = 278).

100 moooomo 00

OcmJ 75 25

Correct 0 95% SEF 2. Responses 50 (H-4 (%) 000 15 25

0 000003 10 o- -- 0 10 20 30 0 i 4 6 8 95% SEF (Hz) [propofoll bg / ml> Figure 3. 95% spectral edge frequency (SEF) (Hz) versus correct responses to Trivial PursuitVype questions corrected for control Figure 5. Propofol blood concentration (pg/mL) versus 95% spec- (D/o). There was no relationship between the 95% spectral edge tral edge frequency (SEF) (Hz) (linear regression lme with 95% frequency and suppression of learning. confidence limits shown). There was no relationship between propofol blood concentration and 95% spectral edge frequency (Y’ = 0.094). propofol blood concentration and thus may be suit- able for monitoring anesthetic effect during propofol Our data support reports that learning is sup- sedation. pressed at propofol blood concentrations insufficient Several other studies have investigated the memory to produce unresponsiveness (17,201. The Cp50-awake effects of propofol. Veselis et al. (17) administered of propofol (blood concentration preventing response constant-rate infusions to volunteers, achieving to verbal command in 50% of subjects) is -3 Fg/mL propofol blood concentrations of 0.86 +- 0.45 pg/mL when administered alone during steady-state condi- (maximum: 1.10 2 0.39 pg/mL). Although the volun- tions (9,211. The eyelash reflex is lost at similar propo- teers’ responses varied substantially, propofol blood fol concentrations (22). Therefore, loss of responsive- concentration predicted impairment of immediate re- ness, or of the eyelash reflex, will ensure absence of call and delayed recognition of test material. Other recall, albeit at the cost of greater respiratory depres- studies, in which no propofol blood concentrations sion. However, unparalyzed patients will still be able were reported, confirmed a dose-dependent effect on to move in response to painful stimulation, as the delayed recall, apparently due to impaired storage Cp50 (minimum alveolar anesthetic concentration an- and consolidation of new material, rather than inade- alog) of propofol is >8 pg/mL (9,lO). quate retrieval (18,191. Our study confirms and ex- To ensure that propofol blood concentrations tends these findings by establishing the concentration- equaled effect-site (brain) concentrations, constant effect relationship for propofol over a clinically blood propofol concentrations were required during relevant range, during steady-state conditions. administration of the learning tasks. We attempted to ANESTH ANALG LESLIE ET AL. 1273 1995;81:1269-74 PROPOFOL, EEG, AND LEARNING

minimize variation in drug concentration by including a biphasic response to increasing propofol concentra- volunteers of similar age and body morphology, and tion (29), whereas the Bispectral Index does not. using lean body mass to calculate the dose (23). Phar- We have demonstrated that the Bispectral Index macokinetic data used to program the infusion were (version 3.0) is monotonically related to propofol derived from a similar group of volunteers (13). We blood concentration over the sedative range. In con- believe that the constancy of propofol concentration trast, it is not possible to predict propofol concentra- we achieved was adequate to allow definition of the tion using the 95% spectral edge frequency. The memory effects of propofol. Hysteresis between blood Bispectral Index therefore may be used to measure and effect-site concentrations, was minimized by al- anesthetic effect during propofol sedation. The reli- lowing at least 15 min (Xhree times the half-time for ability of the Bispectral Index as a measure of the equilibration with the effect compartment*) for equil- anesthetic effect of other drugs awaits further study. ibration before testing. Regression analysis assumed independent data. Re- Pharmacodynamic variation was more difficult to peated testing in each volunteer, although economical, control. Learning may be facilitated by autonomic and may result in intravolunteer correlation that may vio- endocrine responses (24). We tried to control ambient late this assumption. In the absence of a comparable light and noise, and minimize distractions in the lab- statistic with which to analyze these data, we have oratory, but inevitably some volunteers were more accepted, as have others (30), the potential error intro- aroused than others. The volunteers’ motivation to duced into our results. Despite these problems, the remember answers likely also was influenced by dif- functions derived appear to adequately describe our ferences in personality and experience, particularly data. familiarity with medical environments. As periopera- In conclusion, propofol caused a concentration- tive events are even more stressful, caution should be dependent impairment of learning. Propofol is thus used when extrapolating our results to the clinical suitable for conscious sedation during regional anes- thesia. The Bispectral Index was linearly related setting. The Trivial Pursuit@-type question task was easy to propofol blood concentration, and predicted suppres- sion of learning. use for testing learning during anesthesia. Complete learning of test material in all volunteers when no propofol was given confirms that this material was We would like to acknowledge the advice of Jeff Sigl, PhD. We easy to learn and retain (25). However, the task was thank Warren D. Smith, PhD, for assistance in statistical analyses; not sufficiently sensitive to allow the definition of Ben S. Chortkoff, MD, for advice about memory testing during propofol effect, as each incorrect answer decreased the anesthesia; Andrew R. Bjorksten, PhD, for devising the propofol infusion protocol and performing the propofol assays; and Charles percentage of questions answered correctly by 23%. Hackman, FANZCA, (assisted by an Australian and New Zealand Using more questions, however, may not have im- College of Anaesthetists Research Project Grant), who programmed proved our results as there is a limit to the amount of the drug-infusion pump. We would like to thank Ohmeda, Inc. for information that a volunteer can remember (1). loan of a Modulus@ CD-integrated anesthesia machine and infusion As an alternative to targeting specific anesthetic pump, and Hermes Systems, Inc, for loan of an IdaCareTM automatic concentrations, a monitor of pharmacodynamic effect record-keeping system. may be applied. The EEG is suitable as it offers non- invasive, real-time monitoring of central nervous sys- tem activity. However, previous attempts to correlate References EEG changes during anesthesia with memory have met with mixed success (17,26-28). Use of drug com- 1. Ghoneim M, Block R. Learning and consciousness during gen- binations and inadequate control of drug concentra- eral anesthesia. 1992;76:279-305. 2. Knell PJW, McKean JF. An investigation of the pharmacokinetic tion in some of these studies may have increased the profile of propofol (‘Diprivan’) after administration for induc- variability observed in the EEG measures. tion and maintenance of anaesthesia by repeat bolus doses in In this study, we compared the 95% spectral edge patients having spinal anaesthesic block. Postgrad Med J 1985; 61(Suppl3):60 -1. frequency, an example of a conventional processed- 3. MacKenzie N, Grant I. Propofol for intravenous sedation. An- EEG variable, with the Bispectral Index, and demon- aesthesia 1987;423-6. strated that the Bispectral Index was superior to the 4. Sebel I’S, Lowdon JD. Propofol: a new intravenous anesthetic. Anesthesiology 1989;71:260-77. 95% spectral edge frequency in predicting suppression 5. Kearse L, Manberg I’, Chamoun N, et al. Bispectral analysis of of learning in our volunteers. We postulate that this is the electroencephalogram correlates with patient movement to because conventional processed-EEG variables exhibit skin incision during propofol/ anesthesia. Anes- thesiology 1994;81:1365-70. 6. Vernon J, Lang E, Sebel I’, et al. Prediction of movement using bispec- *Schuttler J, Schwilden H, Stoekel H. Pharmacokinetic-dynamic tral electroencephalographic analysis during propofol/alfentanil modeling of Diprivan [abstract]. Anesthesiology 1986;65:A549. or isoflurane/alfentanil anesthesia. Anesth Analg 1995;80:780-5. 1274 LESLIE ET AL. ANESTH ANALG PROPOFOL, EEG, AND LEARNING 1995;81:1269-74

7. Buhrer M, Maitre I’, Hung 0, et al. Thiopental pharmacody- 19. Polster M, Gray I’, O’Sullivan G, et al. Comparison of the namics. 1. Defining the pseudo-steady state serum concentra- sedative and amnesic effects of and propofol. Br J tion EEG relationship. Anesthesiology 1992;77:226-36. Anaesth 1993;70:612-6. 8. Hung 0, Varvel J, Shafer S, et al. Thiopental pharmacodynam- 20. Rupreht J. Awareness with amnesia during total intravenous its. 2. Quantitation of clinical and electroencephalographic anaesthesia with propofol [letter]. Anaesthesia 1989;44:1005. depth of anesthesia. Anesthesiology 1992;77:237-44. 21. Chortkoff B, Eger EI, Crankshaw D, et al. Concentrations of 9. Smith C, McEwan A, Jhavieri R, et al. Interaction of fentanyl on desflurane and propofol that suppress response to command in the Cp50 of propofol for loss of consciousness and skin incision. human volunteers. Anesth Analg 1995;81:737-43. Anesthesiology 1994;81:820-8. 22. Vuyk J, Engbers FH, Lemmens HJ, et al. Pharmacodynamics of 10. Davidson J, Macleod A, Howie J, et al. Effective concentration 50 propofol in female patients. Anesthesiology 1992;77:3-9. for propofol with and without 67% nitrous oxide. Acta Anaes- thesiol Stand 1993;37:458-64. 23. Leslie K, Crankshaw DP. Lean tissue mass is a useful predictor 11. Leslie K, Sessler D, Bjorksten A, et al. Propofol causes a dose- of induction dose requirements for propofol. Anaesth Intensive dependent decrease in the thermoregulatory threshold for va- Care 1991;19:57-60. soconstriction but has little effect on sweating. Anesthesiology 24. Weinberger N, Gold I’, Sternberg D. Epinephrine enables pav- 1994;81:353-60. lovian fear conditionin under anaesthesia. Science 1984;223: 12. Hynson J, Sessler D, Moayeri A, et al. The effects of preinduc- 605-6. tion warming on temperature and blood pressure during 25. Dwyer R, Bennett H, Eger E, et al. Isoflurane anesthesia pre- propofol/nitrous oxide anesthesia. Anesthesiology 1993;79: vents unconscious learning. Anesth Analg 1992;75:107-12. 219-28. 26. Levy W. Power spectrum correlates of changes in consciousness 13. Leslie K, Sessler D, Bjorksten A, et al. Mild hypothermia alters during anesthetic induction with enflurane. Anesthesiology propofol pharmacokinetics and increases the duration of action 1986;64:688-93. of atracurium. Anesth Analg 1995;80:1007-14. 27. Veselis R, Reinsel R, Alagesan R, et al. The EEG as a monitor of 14. Crankshaw D, Boyd D, Bjorksten AR. Plasma Drug Efflux-a midazolam amnesia: changes in power and topographu as a new approach to optimization of drug infusion for constant function of amnesic state. Anesthesiology 1991;74:866-874. blood concentration of thiopental and . Anesthesi- 28. Dwyer R, Rampil I, Eger EI, et al. The electroencephalogram ology 1987;67:32-41. does not predict depth of isoflurane anesthesia. Anesthesiology 15. Chortkoff B, Bennett H, Eger E. Does nitrous oxide antagonize 1994;81:403-9. isoflurane-induced suppression of learning? Anesthesiology 29. Seifert H, Blouin R, Conard I’, et al. Sedative doses of propofol 1993;79:724-32. increase beta activity of the processed electroencephalogram. 16. Plummer G. Improved method for the determination of propo- fol in blood by high performance liquid chromatography with Anesth Analg 1993;76:976-8. fluorescence detection. J Chromatogr 1987;421:171-6. 30. Chortkoff B, Bennett H, Eger E. Subanesthetic concentrations 17. Veselis R, Reinsel R, Wronski M, et al. EEG and memory effects of isoflurane suppress learning as defined by the category- of low-dose infusions of propofol. Br J Anaesth 1992;69:246-54. example task. Anesthesiology 1993;79:16-22. 18. Zacny JP, Lichtor JL, Coalson DW, et al. Subjective and psy- chomotor effects of subanaesthetic doses of propofol in healthy volunteers. Anesthesiology 1992;76:696-702.