Target-controlled infusion (TCI) systems can control the concentration in the plasma or at the site of drug effect. A TCI system that targets the effect site should be able to accurately predict the time course of drug effect. The authors tested this by comparing the performance of three control algorithms: plasmacontrol TCI versus two algorithms for effect-site control TCI.


One-hundred twenty healthy women patients received propofol via TCI for 12-min at a target concentration of 5.4 microg/ml. In all three groups, the plasma concentrations were computed using pharmacokinetics previously reported. In group I, the TCI device controlled the plasma concentration. In groups II and III, the TCI device controlled the effect-site concentration. In group II, the effect site was computed using a half-life for plasma effect-site equilibration (t1/2k(eo)) of 3.5 min. In group III, plasma effect-site equilibration rate constant (k(eo)) was computed to yield a time to peak effect of 1.6 min after bolus injection, yielding a t1/2keo of 34 s. the time course of propofol was measured using the bispectral index. Blood pressure, ventilation, and time of loss of consciousness were measured.


The time course of propofol drug effect, as measured by the bispectral index, was best predicted in group III. Targeting the effect-site concentration shortened the time to loss of consciousness compared with the targeting plasma concentration without causing hypotension. The incidence of apnea was less in group III than in group II.


Effect compartment-controlled TCI can be safely applied in clinical practice. A biophase model combining the Marsh kinetics and a time to peak effect of 1.6 min accurately predicted the time course of propofol drug effect.

TARGET-CONTROLLED infusion (TCI) devices for propofol incorporate an internal model of propofol pharmacokinetics to rapidly achieve and maintain a constant drug concentration in the plasma 1,2or at the site of drug effect. 3,4When a plasma concentration is targeted, an attempt is made to achieve a square wave in the plasma. The effect-site equilibrates with the half-life of ln2/plasma effect-site equilibration rate constant (ke0). In turn, the target effect compartment concentration is approached slowly. In contrast, when the effect site is targeted, the plasma compartment must be overdosed initially to drive the drug into the effect site (fig. 1). The performance of TCI systems that target plasma drug concentration has been reported extensively. 5–16Recently, Wakeling et al.  17reported that targeting the effect-site concentration for propofol resulted in a more rapid loss of consciousness (LOC) without an increase in the risk of hypotension. The purpose of this study was to compare three different TCI control algorithms: plasma control (as currently implemented in the Diprifusor device developed by AstraZeneca, Manchester, England; effect-site control based on a ke0of 0.20 min−1, as reported by Billard et al.18; and effect-site control based on a time to peak effect of 1.6 min, as recently proposed by Schnider et al.  19The time course of propofol drug effect was measured using the bispectral index (BIS). 20–23 

Clinical Protocol

After Institutional Ethics Committee approval, informed consent was obtained from 120 women patients, American Society of Anesthesiologists physical status I and II, aged 18–60 yr, and scheduled for ambulatory gynecologic surgery. Exclusion criteria included weight less than 70% or more than 130% of ideal body weight; neurologic disorder; and use of psychoactive medication, including alcohol. Patients were allocated randomly to one of the three groups. In each group, the plasma pharmacokinetic model used was that reported by Marsh. 24The Marsh model was chosen because of the good performance reported for this model by Coetzee et al.  25More importantly, the Marsh pharmacokinetic model is incorporated into the Diprifusor propofol infusion device marketed in Europe, Australia, and Asia. The Diprifusor is the only commercially available TCI device and, therefore, evaluation of control algorithms based on the Marsh pharmacokinetic model represent potential commercial growth for the Diprifusor device.

Each patient received a 12-min infusion of propofol via  the TCI device to a target level of 5.4 μg/ml. This was selected based on a report by Wakeling et al.  17that a target amount of 5.4 μg/ml provided LOC in all volunteers and was able to distinguish plasma control from effect-site control. Previous studies by Smith 26and Gepts 27suggested that a propofol concentration of 5.4 μg/ml is associated with LOC in 95% of subjects.

In group I, the TCI device targeted the propofol concentration in the plasma, as presently implemented in the Diprifusor. An example of the anticipated propofol plasma and effect-site concentrations from this infusion method is shown in the top graph of figure 1and is based on the pharmacokinetics reported by Marsh et al. , 24and the propofol ke0of 0.20 min−1as reported by Billard et al.  18In groups II and III, the TCI device targeted the propofol concentration at the site of drug effect using the strategy proposed by Shafer and Gregg 3and the mathematical implementation reported by Jacobs and Williams. 4In group II, the ke0for plasma-effect site equilibration was 0.20 min−1, as reported by Billard et al.  18This value of ke0,when combined with the propofol pharmacokinetics reported by Marsh, predicts a time to peak effect of 4.5 min after bolus injection. The middle graph of figure 1shows an example of the expected plasma and effect-site concentrations for subjects in group II. For subjects in group III, we calculated the ke0that predicted a time to peak effect 1.6 min after bolus injection, as reported by Schnider et al.  19The calculation is an adaptation of the time-to-peak-effect algorithm reported by Shafer and Gregg 3and implemented in the computer programs STANPUMP (written by S. Shafer, M.D., PAVMC, Palo Alto, CA) and RUGLOOP (written by T. De Smet, M.Sc., and M. Struys, M.D., Ph.D., Ghent University Hospital, Gent, Belgium) ‡‡. In brief, ke0is found using nonlinear regression to iteratively search out the value of ke0that predicts the peak effect-site concentration at the desired time. The lower graph of figure 1shows an example of the expected plasma and effect-site target concentrations for subjects in group III. Because subjects in group II were predicted to have much slower blood–brain equilibration than that of subjects in group III, the control algorithm called for larger doses in group II than in group III, resulting in the larger overshoot in plasma concentration seen in figure 1.

No premedication was given. Before anesthesia, an 18-gauge catheter was inserted in a large forearm vein for fluid and drug administration. The propofol infusion was connected as close as possible to the intravenous catheter to minimize dead space. Patients did not receive a loading dose of intravenous fluid before the propofol infusion and received 100–200 ml Ringer’s lactate during the propofol infusion. No other drugs were administered during the 12 min of propofol administration. Patients received oxygen via  face mask. The presence of apnea was recorded. If necessary, oxygen saturation as measured by pulse oximetry (SpO2< 90%) manual breathing support using a circle system with 100% oxygen was applied. Loss of consciousness  was defined as failure to respond to verbal command and was evaluated every 5 s.

Propofol concentrations were not measured. Rather, the analysis was based on the predicted plasma propofol concentration calculated from the Marsh pharmacokinetics (CpCALC) and on the predicted effect-site propofol concentrations calculated from the Marsh pharmacokinetics and the estimates of ke0(CeCALC). 28Three observations were made at the moment of LOC: time, BIS, and CeCALC(groups II and III only). We also calculated the propofol induction dose as the amount of propofol delivered to the patients up to the time of LOC.

Heart rate, end-tidal carbon dioxide, and SpO2were measured every 10 s using the Datex AS3 (Datex, Helsinki, Finland). Apnea  was defined as any interval in carbon dioxide exhalation exceeding 10 s. Blood pressure was measured every 30 s using the Datex monitor. Electroencephalographic BIS was measured every 10 s using an Aspect A-1000 EEG (Aspect Medical Systems, Natick, MA) monitor, version 3.2. Artifacts caused by poor signal quality were automatically detected and excluded from further analysis. All physiologic data were recorded by the RUGLOOP program (described below) and stored on hard disk.

Propofol was administered via  a Graseby 3500 syringe pump (SIMS Graseby Ltd., Herts, England). The pump was controlled by RUGLOOP, a program written by the authors (T. D. S. and M. S.) in Visual C++ (Microsoft, Redmond, WA) for Windows 95/NT operating system (Microsoft). This TCI program can also be used as data management system. RUGLOOP incorporates algorithms to target the plasma 2and the site of drug effect. 3The algorithms in RUGLOOP, including those to target the effect site using ke0(group II) and time to peak effect (group III) are adapted from STANPUMP.

Evaluation of the Pharmacokinetics–Pharmacodynamics Model

For all patients, the calculated plasma concentration, calculated effect-site concentration, infusion rate, and cumulative dose of propofol were recorded every 10 s and stored on hard disk. The performance of the systems was evaluated using the following measures, as illustrated in figure 1:

  • • tpeak= observed time necessary for reaching maximal drug effect (lowest BIS), taken within 2 min of LOC. Patients receiving propofol targeting the effect site (groups II and III) should have a faster onset of drug effect than patients receiving propofol targeting the plasma (group I). Therefore, this was our primary outcome measure for comparison of group I with groups II and III.

  • • teq= calculated time necessary for equilibration between CpCALC(the plasma propofol concentration) and CeCALC(the effect-site propofol concentration). This was not calculated for group I, in which CeCALCshould only approach CpCALCasymptotically and, thus, teqis infinite.

  • • terror= teq− tpeak. Using a value of ke0calculated to give the correct time to peak effect should more accurately reflect the observed time course of drug effect than using a ke0taken from the literature. If the effect-site model is accurate, the calculated time to equilibration should be similar to the observed time of peak effect. Therefore, terrorwas our primary outcome measurement for comparison of group II with group III.

  • • CpMAX= calculated maximum plasma propofol concentration.

  • • BISpeak= observed BIS at tpeak.

  • • tMinMAP= observed time to lowest arterial blood pressure (the “onset of side effect”).

Statistical Analyses

Data were presented as the mean ± SD or as the median (range). Differences between the groups for the primary outcome measure were determined using a Wilcoxon rank sum test, as was the comparison in time to lowest blood pressure between groups II and III. Differences for secondary measures were assessed using Student two-tailed t  tests after confirming that the data were normally distributed, except for the change in blood pressure, which was analyzed using analyses of variance for repeated measures. Significance level was set at 5%. The incidence of apnea was analyzed using the chi-square test.

Graphical Analysis

For visual comparison of the time course of drug effect, we graphed the propofol plasma concentration, BIS, and mean arterial pressures over time for all subjects. For subjects in groups II and III, we also graphed the effect-site concentrations over time.

Population demographics for the three groups are shown in table 1. There were no significant differences among the three groups. No patients were excluded from the analysis. All data captured by the recording system were included in the analysis.

The observations made at the time of LOC are shown in table 2. As expected, patients lost consciousness more slowly when the TCI device targeted the plasma (90 s; range, 44–601 s) than when the device targeted the effect site (68 s [range, 45–104 s] and 71 s [range, 43–110 s] for groups II and III, respectively). The largest doses of propofol were administered in patients in group II. Patients in group III received a similar induction dose as the plasma-controlled TCI patients (group I). Patients in groups II and III lost consciousness at similar values of BIS. However, the calculated effect-site concentration in group II was only 1.8 ± 0.7 μg/ml, significantly less than the calculated effect-site concentration of 4.7 ± 0.6 in group III.

The measures to evaluate the performance of the three infusion algorithms can be found in table 3. Onset was significantly faster when the effect site was targeted (116 ± 21 and 120 ± 21 s in groups II and III, respectively) than when the plasma concentration was targeted (218 ± 86 s). The error in predicting the time of peak effect was 207 ± 73 s in group II and 9 ± 15 s in group III, a difference that was highly significant (P < 0.0001). Group II was associated with the largest overshoot in plasma propofol concentration (14.2 ± 0.4 μg/ml) and the lowest BIS (27 ± 11), both significantly different from groups I and III. The peak cardiovascular depression shown in table 3occurred significantly later than the peak electroencephalography (EEG) depression in groups II and III (P < 0.05).

Figures 2, 3, and 4show the time course of propofol concentration, BIS, and mean arterial pressure in all patients. The vertical bars on the top show the times of LOC. There was considerably more variability in group I than in groups II or III in the time to LOC. Group II was characterized by a larger plasma overshoot and a more precipitous decrease in BIS and blood pressure than was observed in groups I or III. In group III, the decrease in BIS was as rapid as in group II, but not as large. With the exception of the more rapid change in blood pressure observed in group II, the blood pressure response was similar in all three groups. There were no significant changes in heart rate from baseline in any of the groups.

Significantly more patients experienced apnea in group II (34 or 40 patients) than in groups I (9 of 40 patients) or III (17 of 40 patients). Eight patients in group II required manual ventilation. In the other groups, spontaneous ventilation resumed before manual ventilation was necessary.

The purpose of this study was to compare three methods of propofol administration via  TCI. The first method reproduced the performance of the commercially available Diprifusor device, which was designed to control the plasma propofol concentration. The second two methods evaluated the performance of a device that targeted the concentration in the effect site rather than in the plasma. Our results confirmed the findings published by Wakeling et al.  17When a TCI device targets the effect site, onset of drug effect is hastened without adverse hemodynamic consequences. The peak cardiovascular depression occurred significantly later than did the peak EEG depression in groups II and III. As a result, targeting the effect site does not increase toxicity because the toxicity has a slower equilibration than does the desired effect. Our results confirm prospectively those by Kazama et al.,  29who found that the effect of propofol on BIS occurs more rapidly than its effect on systolic blood pressure. In addition, there was less variability and, hence, potentially greater predictability in the time to LOC when the effect site was targeted.

There are several ways that the equilibration delay between the plasma and the site of drug effect can be added to the pharmacokinetic model. The most straightforward approach is to introduce an effect-site model using the value of ke0from a previously published report. This is the approach taken in group II. As pointed out by Gentry et al.,  30the value of ke0is highly influenced by the pharmacokinetic model; therefore, it may be unwise to mix the ke0from one study with the pharmacokinetics from a different study. It is possible to directly observe the time to peak effect; therefore, this is a “model-independent” descriptor of blood–brain equilibration. The time to peak effect is experimentally verifiable by giving a bolus and using an appropriately sensitive measure of drug effect. 19Time to peak effect can be used to establish the appropriate value of ke0for use with any pharmacokinetic model or, indeed, any representation of the bolus response (e.g. , a set of time vs.  concentration data points), provided that a submaximal effect is elicited and that the time of peak effect can be observed precisely. Schnider et al.  19reported a time to peak propofol effect of 1.6 min, based on close inspection of the EEG waveform in 48 subjects. We calculated that a ke0of 1.21 min−1would produce a peak effect-site concentration of 1.6 min when using the Marsh pharmacokinetics. The resulting value of ke0was prospectively tested in group III. Group III performed better than did group II in this study. The algorithm in group III more accurately predicted the time of peak EEG effect. It also provided an anesthetic induction with less drug, less of an overshoot with BIS, less apnea, and a more gradual decrease in blood pressure than in group II. These results show that the clinical outcome is dependent on the value of ke0. This also validates the use of the time-to-peak-effect concept as a pharmacodynamic parameter.

Because the BIS is a continuous measure of propofol drug effect, we chose to use it as the primary measure of the time course of drug effect rather than a more clinically oriented measure, such as the time of LOC. Doi et al.  20demonstrated a correlation (r2= 0.55) between BIS and calculated blood concentrations of propofol, and others confirmed these results. 21–23The BIS is calculated using a 30-s rolling window and, thus, lags behind the current status of the patient by approximately 15 s. This may account for the observation in table 2of lower BIS values for LOC for subjects in group I than those in group II and group III, and for the somewhat longer time to peak BIS effect in group III than the predicted time to peak effect.

The equilibration delay between the plasma and the site of drug effect is a physiologic reality. Incorporating this delay into a TCI device increases the complexity of the pharmacokinetic model used to control the infusion. However, if the delay is ignored, the model relating dose to drug effect is fundamentally incorrect. A TCI device programmed with an incorrect model cannot be expected to produce the desired time course of drug effect.

In conclusion, we demonstrated that a TCI device that controlled the concentration at the site of drug effect more accurately produced the desired time course of drug effect than did a device that only controlled plasma drug concentration. We also demonstrated that the choice of plasma effect-site equilibration delay makes a difference in the performance of the device. It is not appropriate to use a documented value of ke0in an infusion device without consideration of the corresponding pharmacokinetic model. Future studies will be needed to determine the combined pharmacokinetic–pharmacodynamic model that best predicts the time course of propofol drug effect.

The authors thank Dr. Thomas W. Schnider, M.D., Ph.D., Universität Bern, Institut für Anästhesiologie und Intensivbehandlung, Bern, Switzerland, for assistance during the preparation of this manuscript.

Schwilden H: A general method for calculating the dosage scheme in linear pharmacokinetics. Eur J Clin Pharmacol 1981; 20:379–86
Bailey JM, Shafer SL: A simple analytical solution to the three-compartment pharmacokinetic model suitable for computer-controlled infusion pumps. IEEE Trans Biomed Eng 1991; 38:522–5
Shafer SL, Gregg KM: Algorithms to rapidly achieve and maintain stable drug concentrations at the site of drug effect with a computer-controlled infusion pump. J Pharmacokinet Biopharm 1992; 20:147–69
Jacobs JR, Williams EA: Algorithm to control “effect compartment” drug concentrations in pharmacokinetic model-driven drug delivery. IEEE Trans Biomed Eng 1993; 40:993–9
Alvis JM, Reves JG, Govier AV, Menkhaus PG, Henling CE, Spain JA, Bradley E: Computer-assisted continuous infusions of fentanyl during cardiac anesthesia: comparison with a manual method. A NESTHESIOLOGY 1985; 63:41–9
Ausems ME, Vuyk J, Hug CC Jr, Stanski DR: Comparison of a computer-assisted infusion versus intermittent bolus administration of alfentanil as a supplement to nitrous oxide for lower abdominal surgery. A NESTHESIOLOGY 1988; 68:851–61
Shafer SL, Varvel JR, Aziz N, Scott JC: Pharmacokinetics of fentanyl administered by computer-controlled infusion pump. A NESTHESIOLOGY 1990; 73:1091–102
Glass PS, Jacobs JR, Smith LR, Ginsberg B, Quill TJ, Bai SA, Reves JG: Pharmacokinetic model-driven infusion of fentanyl: Assessment of accuracy. A NESTHESIOLOGY 1990; 73:1082–90
Raemer DB, Buschman A, Varvel JR, Philip BK, Johnson MD, Stein DA, Shafer SL: The prospective use of population pharmacokinetics in a computer-driven infusion system for alfentanil [published erratum appears in A NESTHESIOLOGY 1990 Oct;73(4): 798]. A NESTHESIOLOGY 1990; 73:66–72
Varvel JR, Donoho DL, Shafer SL: Measuring the predictive performance of computer-controlled infusion pumps. J Pharmacokinet Biopharm 1992; 20:63–94
Theil DR, Stanley TED, White WD, Goodman DK, Glass PS, Bai SA, Jacobs JR, Reves JG: Midazolam and fentanyl continuous infusion anesthesia for cardiac surgery: A comparison of computer-assisted versus manual infusion systems. J Cardiothorac Vasc Anesth 1993; 7:300–6
Dyck JB, Maze M, Haack C, Azarnoff DL, Vuorilehto L, Shafer SL: Computer-controlled infusion of intravenous dexmedetomidine hydrochloride in adult human volunteers. A NESTHESIOLOGY 1993; 78:821–8
Fiset P, Mathers L, Engstrom R, Fitzgerald D, Brand SC, Hsu F, Shafer SL: Pharmacokinetics of computer-controlled alfentanil administration in children undergoing cardiac surgery. A NESTHESIOLOGY 1995; 83:944–55
Vuyk J, Engbers FH, Burm AG, Vletter AA, Bovill JG: Performance of computer-controlled infusion of propofol: An evaluation of five pharmacokinetic parameter sets. Anesth Analg 1995; 81:1275–82
Schnider TW, Gaeta R, Brose W, Minto CF, Gregg KM, Shafer SL: Derivation and cross-validation of pharmacokinetic parameters for computer-controlled infusion of lidocaine in pain therapy. A NESTHESIOLOGY 1996; 84:1043–50
Veselis RA, Glass P, Dnistrian A, Reinsel R: Performance of computer-assisted continuous infusion at low concentrations of intravenous sedatives. Anesth Analg 1997; 84:1049–57
Wakeling HG, Zimmerman JB, Howell S, Glass PSA: Targeting effect compartment or central compartment concentration for propofol. A NESTHESIOLOGY 1999; 90:92–7
Billard V, Gambus PL, Chamoun N, Stanski DR, Shafer SL: A comparison of spectral edge, delta power, and bispectral index as EEG measures of alfentanil, propofol, and midazolam drug effect. Clin Pharmacol Ther 1997; 61:45–58
Schnider TW, Minto CF, Shafer SL, Gambus PL, Andresen C, Goodale DB, Youngs EJ: The influence of age on propofol pharmacodynamics. A NESTHESIOLOGY 1999; 90:1502–16
Doi M, Gajraj RJ, Mantzaridis H, Kenny GN: Relationship between calculated blood concentration of propofol and electrophysiological variables during emergence from anaesthesia: Comparison of bispectral index, spectral edge frequency, median frequency and auditory evoked potential index. Br J Anaesth 1997; 78:180–4
Kearse LA Jr, Rosow C, Zaslavsky A, Connors P, Dershwitz M, Denman W: Bispectral analysis of the electroencephalogram predicts conscious processing of information during propofol sedation and hypnosis. A NESTHESIOLOGY 1998; 88:25–34
Struys M, Versichelen L, Byttebier G, Mortier E, Moerman A, Rolly G: Clinical usefulness of the bispectral index for titrating propofol target effect-site concentration. Anaesthesia 1998; 53:4–12
Glass PS, Bloom M, Kearse L, Rosow C, Sebel P, Manberg P: Bispectral analysis measures sedation and memory effects of propofol, midazolam, isoflurane, and alfentanil in healthy volunteers. A NESTHESIOLOGY 1997; 86:836–47
Marsh B, White M, Morton N, Kenny GN: Pharmacokinetic model driven infusion of propofol in children. Br J Anaesth 1991; 67:41–8
Coetzee JF, Glen JB, Wium CA, Boshoff L: Pharmacokinetic model selection for target controlled infusions of propofol. Assessment of three parameter sets. A NESTHESIOLOGY 1995; 82:1328–45
Smith C, McEwan AI, Jhaveri R, Wilkinson M, Goodman D, Smith LR, Canada AT, Glass PS: The interaction of fentanyl on the Cp50 of propofol for loss of consciousness and skin incision. A NESTHESIOLOGY 1994; 81:820–8
Gepts E, Camu F, Cockshott ID, Douglas EJ: Disposition of propofol administered as constant rate intravenous infusions in humans. Anesth Analg 1987; 66:1256–63
Glass PS, Glen JB, Kenny GN, Schuttler J, Shafer SL: Nomenclature for computer-assisted infusion devices (letter). A NESTHESIOLOGY 1997; 86:1430–1
Kazama T, Ikeda K, Morita K, Kikura M, Doi M, Ikeda T, Kurita T, Nakajima Y: Comparison of the effect-site ke0of propofol for blood pressure and EEG bispectral index in elderly and younger patients. A NESTHESIOLOGY 1999; 90:1517–27
Gentry WB, Krejcie TC, Henthorn TK, Shanks CA, Howard KA, Gupta DK, Avram MJ: Effect of infusion rate on thiopental dose-response relationships. Assessment of a pharmacokinetic-pharmacodynamic model. A NESTHESIOLOGY 1994; 81:316–24