Anesthetics alter electrocortical activity in a dose-dependent manner, and it is suggested that electroencephalographic (EEG) activity can provide a reliable basis for a surrogate measurement of hypnosis.1 An EEG monitor, the Bispectral Index (BIS), has proven useful to titrate hypnotic drugs.2,3 Because it is a single composite measure monitored continuously, it has been used in controlled studies to automatically guide propofol administration.4–7 Conversely, automated delivery systems for opioids have seldom been described using EEG activity.8
There are no specific measures to quantify analgesia directly. Nevertheless, in the interaction between opioid-hypnotic synergy and loss of consciousness, when the dose of analgesic administered is sufficient to inhibit autonomic response to noxious stimuli, then the required hypnotic concentration is only that needed to achieve loss of consciousness.9,10 Moreover, a noxious stimulus such as laryngoscopy11 or periosteal pressure to the tibia12 may cause electrocortical activation with an increase in BIS values, which indirectly reflects the analgesic state or the level of antinociception. Finally, an excitatory EEG arousal reaction resulting from nociceptive stimulation can provide a rational basis for analgesia administration. It can then be hypothesized that BIS could also be the controlled variable for opioid administration.
We have developed a closed-loop controller using a proportional-integral-derivative algorithm, which guides IV coadministration of propofol and remifentanil solely by BIS monitoring during induction and maintenance of general anesthesia. Specifically, we tested the hypothesis that our closed-loop controller would increase the time spent with BIS values within predetermined boundaries (i.e., the interval 40–60) and limit BIS oscillation as compared with manual administration.
With the approval of the Ethics Committee, the French Regulatory Office (Afssaps), and written informed consent, 196 patients scheduled for elective surgery requiring general anesthesia or combined regional/general anesthesia expected to last >30 minutes and requiring tracheal intubation were enrolled at 4 different hospitals: Hôpital Foch (Suresnes), Centre Hospitalier Victor Dupouy (Argenteuil), Centre Hospitalo-Universitaire of Besançon and of Angers. All investigators were clinically experienced in the use of target-controlled infusion (TCI) of propofol and remifentanil and in BIS monitoring, and all had received 2 days' training on the closed-loop controller in Hôpital Foch. Patients were aged 18 to 90 years with ASA physical status I to IV. Exclusion criteria included psychiatric illness, supraspinal neurological disorders, cranial neurosurgical procedures, and patients equipped with a pacemaker.
Patients were randomly assigned to dual closed-loop (dual-loop group) or manual (manual group) TCI of propofol and remifentanil. Treatments were segregated into blocks of 10 at each participating center, and randomization was determined using a random number generator. Assignments were kept in sequentially numbered opaque envelopes until just before surgery.
On arrival in the operating room, a dedicated IV cannula was placed, and routine monitoring commenced including temperature and monitoring of neuromuscular function at the adductor pollicis. The BIS electrode (Zipprep; Bispectral Index, Covidien, Mansfield, MA) was positioned on the patient's forehead and connected to either an A-2000 XP (version 3.11) BIS monitor (Covidien) or a BIS module (Datex-Ohmeda™ S/5™, Helsinki, Finland).
All patients received total IV anesthesia in TCI mode using the population pharmacokinetic sets of Schnider et al.13 and Minto et al.14 for propofol and remifentanil, respectively. Infusion Toolbox 95® version 4.11 software15 implemented in a personal computer served as a platform: (1) calculation of effect-site concentrations of propofol and remifentanil; (2) display of these concentrations in real time on the screen of the personal computer; (3) user interface to enter the patient's demographic data (sex, age, weight, and height) and to modify these concentrations; (4) steering of the infusion pumps, one for propofol and the other for remifentanil (Alaris Medical, Hampshire, UK); and (5) collection of the data every 5 seconds (BIS, calculated site-effect drug concentrations, hemodynamic data when an AS/5 Datex-Ohmeda S/5™ monitor was used).
In both groups, the induction phase was defined as the time from the start of propofol and remifentanil administration to BIS <60 for 30 seconds and the maintenance phase from this point to the end of propofol and remifentanil administration.
In the manual group, the investigator chose the initial propofol and remifentanil effect-site target concentrations for induction according to his/her clinical judgment. Thereafter, the investigator modified the effect-site target concentrations of both drugs without minimum or maximum concentration limits to maintain BIS at approximately 50 within a range of 40 to 60 to the extent possible. Only manual ventilation via a facemask was performed during the induction phase to avoid noxious stimuli (such as laryngoscopy or tracheal intubation). Throughout the procedure, signs of inadequate analgesia (tachycardia, hypertension, sweating, flushing, movement, or swallowing) or hypnosis (BIS value outside of the range 40 to 60) were treated by increasing remifentanil or propofol at the discretion of the attending physician.
In the dual-loop group, the investigator chose the initial propofol effect-site target concentration according to his/her clinical judgment and the controller fixed the first remifentanil concentration according to the following rules. If the initial propofol target was <2.5 μg · mL−1, then the initial remifentanil target was 4 ng · mL−1; if the initial propofol target was between 2.5 and 2.9 μg · mL−1, then the initial remifentanil target was 5 ng · mL−1; if the initial propofol concentration was >3 μg · mL−1, then the initial remifentanil target was 6 ng · mL−1. Thereafter, modifications of propofol and remifentanil effect-site target concentrations were decided by the controller with lower and upper limits for propofol (1.3 and 5 μg · mL−1) and for remifentanil (3 and 12 ng · mL−1). As in the manual group, only manual ventilation via a facemask was permitted during the induction phase. Throughout the procedure, the investigator could adjust the targets if necessary or switch between closed-loop and manual control. Detailed description of the controller is provided in Figure 1, the Appendix, and in Table 1.
Other than administration of the study drugs, patient management was based on current standards of care. No specific recommendations were given for the treatment of hemodynamic abnormalities and for the use of a neuromuscular blocking drug. Nitrous oxide was not used. Morphine was given at the discretion of the physician 45 minutes before the presumed end of surgery.
In both groups, propofol and remifentanil were stopped simultaneously upon completion of surgery, and if necessary, a neuromuscular blockade reversing drug was given. Tracheal extubation was performed when all of the following were present: patient responsive and cooperative, SpO2 >95% with an oxygen inspiratory fraction <50%, T4/T1 ratio >90%, no hemodynamic instability (continuous infusion of vasoactive drugs), and temperature >36.0°C.
The primary outcome was the global score (GS),5 which characterized the overall performance of the controller including the percentage of adequate anesthesia, defined as BIS between 40 and 60 and the oscillation of the BIS determined by the median absolute performance error (MDAPE) and the wobble.16 The controller performances were calculated according to following equations.
- The performance error, or PE, calculated as the difference between actual and desired values (set point):
- The bias or median performance error (MDPE):
- The inaccuracy or MDAPE:
- The wobble, which measures the intraindividual variability in PE:
- where i = subject number, j = jth (one) measurement of observation period, n = total number of measurements during the observation period.
The GS was calculated according to the following equation5:
Excellent performance is characterized by a low MDAPE and wobble and high percentage of BIS value in the range 40 and 60, thus a low GS.
Secondary outcomes included the percentage of adequate anesthesia (defined as BIS between 40 and 60), overshoot (BIS <40) and undershoot (BIS >60) periods, occurrence of suppression ratio (SR) defined as SR >10% lasting at least a minute, and parameters of Varvel et al.16 (PE, MDPE, MDAPE, and wobble). Secondary outcomes also included the following clinical data: drug consumption, number of somatic events (i.e., movements, grimacing), time to tracheal extubation (i.e., time from discontinuation of propofol and remifentanil infusions until extubation), and recall of intraoperative events as determined by a standardized interview performed in the postanesthesia care unit and on the second or third postoperative day.17
This study is registered with ClinicalTrials.gov, number NCT00392158.
In a previous study, the GS was 50 ± 62 (mean ± SD) using manual propofol and remifentanil infusions and 21 ± 8 using closed-loop control of propofol and manual remifentanil infusion5; therefore, we expected an improvement of >50% using the closed-loop system. We thus estimated that 144 patients (72 per group) would provide a 95% power for a 2-sided α error of 0.01. We planned to recruit 200 patients under the assumption that some would be excluded for various reasons.
Categorical variables, expressed as numbers and frequencies, were compared by means of χ2 test or Fisher exact test as appropriate. Continuous variables were described as mean ± SD and median and interquartile range and compared using the Mann-Whitney U test. Comparison of serial measurements was performed with repeated-measures analysis of variance (with Bonferroni correction), and post hoc analyses were performed with nonparametric tests. Time from discontinuation of propofol and remifentanil infusions until tracheal extubation was compared using a Kaplan-Meier survival analysis followed by log-rank test. Probability values <0.05 were considered statistically significant. Data analysis was performed using SPSS® version 11.0 (SPSS, Inc., Chicago, IL).
Of the 200 patients who were approached, 196 patients were enrolled in the study, and 98 were randomized to each group. However, 15 patients in the dual-loop group and 14 patients in the manual group were excluded from analysis because of neurological disorders, BIS artifact, recording system failure, and other reasons (Fig. 2). There were thus 83 patients in the dual-loop group and 84 in the manual group available for analysis.
The groups were well balanced with respect to demographic, morphometric, and treatment characteristics. More than one-third of the study population received preoperative cardiovascular treatments and/or had a major surgical procedure (Table 2). Study participants had cardiac bypass, thoracic, vascular, urologic, orthopedic, gynecologic, or otolaryngologic surgery.
This study was conducted at 4 different sites, involving 17 anesthesiologists and 22 nurse anesthetists. No significant site differences or group-by-site interaction were found for any end points. Results were therefore pooled from all sites for analysis.
The closed-loop system maintained anesthesia for 312 hours. During this time, 3843 propofol and 4981 remifentanil target modifications were made automatically. The loop was opened in 2 cases; both were manual remifentanil modifications related to an episode of hypertension during laparoscopic procedures. Data for these patients were analyzed in the dual-loop group.
BIS values and calculated effect-site concentrations of propofol and remifentanil from induction to discontinuation of infusion are shown for dual-loop and manual groups in Figure 3.
Induction phase duration was significantly shorter in the dual-loop group than in the manual group (289 ± 122 vs 345 ± 166 seconds, P = 0.01). The initial effect-site concentrations and the amount of propofol were similar between the 2 groups whereas these values were larger for remifentanil in the dual-loop group (Table 3). There were more propofol and remifentanil modifications in the dual-loop group than in the manual group (Table 3). Overshoot, undershoot, and occurrence of SR were all less frequent in the dual-loop group during induction (Table 3).
During maintenance, the mean GS was 26 ± 11 in the dual-loop group versus 43 ± 40 in the manual group (P = 0.0001, Table 4, Fig. 4). The amount of time that BIS was maintained between 40 and 60 was significantly longer in the dual-loop group compared with the manual group (Table 4 and Fig. 5). MDAPE was significantly lower in the dual-loop group whereas wobble was similar between groups (Table 4). Overshoot, undershoot, and occurrence of SR were all less frequent in the dual-loop group (Table 4). The dose and the mean calculated effect-site concentration of propofol were similar in both groups whereas both values were larger for remifentanil in the dual-loop group (Table 4). More propofol and remifentanil modifications were made in the dual-loop than in the manual group, and the adjustments were smaller (Table 4 and Fig. 3). The Feedforward term (Appendix) was triggered 2 ± 2, 2 (1–3) times per hour for propofol and 5 ± 2, 5 (4–6) times per hour for remifentanil (P < 0.0001).
Heart rate, systolic blood pressure, and BIS values were similar between groups before and after painful stimuli (Fig. 6). This analysis was based on 33 patients in the manual group and 31 in the dual-loop group for laryngoscopy, and on 44 patients in the manual group and 49 in the dual-loop group for surgical incision because electronic recording of hemodynamic data and notes of these events were available only for these patients. No patient received combined general/regional anesthesia.
The mean time to tracheal extubation was shorter in the dual-loop than the manual group (Table 4 and Fig. 7). This analysis was based on 70 patients in the manual group and 72 in the dual-loop group because the remaining patients remained intubated because of hypothermia (n = 8), residual neuromuscular block (n = 5), or planned extubation in the intensive care unit (n = 12).
A subgroup of patients corresponds to those who received combined general and regional anesthesia. Patients in the dual-loop group had a nonsignificant decrease in remifentanil consumption (0.22 ± 0.7, 0.20 [0.16–0.27] vs 0.19 ± 0.7, 0.20 [0.16–0.23] μg · kg−1 · min−1, general anesthesia versus combined general/regional anesthesia, P = 0.52). Patients in the manual group had a significant decrease in remifentanil consumption (0.17 ± 0.7, 0.16 [0.11–0.21] vs 0.12 ± 0.6, 0.10 [0.07–0.16] μg · kg−1 · min−1, general anesthesia versus combined general/regional anesthesia, P = 0.003).
Use of neuromuscular blocking drugs, ephedrine treatment, antihypertensive therapy, blood loss, Ringer solution infusion, somatic events, and morphine dose were similar in the 2 groups (Table 4). No cases of awareness with recall were reported.
Our results indicate that automated coadministration of propofol and remifentanil using BIS for the controller allows an improvement in GS: increase in time spent with BIS values within predetermined boundaries (i.e., the interval 40–60) and decrease of MDAPE without an improvement of wobble. It permits a decrease in the duration of the induction phase and the time to extubation.
Closed-loop administration of propofol using BIS as the controlled variable during general anesthesia is becoming a reality as demonstrated by the number of successful cases published.4–7,18–25 On the contrary, the choice of the controlled variable remains when discussing closed-loop administration of an opioid. Most investigators propose the use of heart rate and mean arterial blood pressure to control opioid delivery.26–28 However, hemodynamics are not specific for the response to painful stimulation and may be affected by a number of covariates such as blood loss, heart failure, arrhythmia, manipulation of great blood vessels, and a variety of drugs. Schwilden and Stoeckel8 used the median EEG frequency to steer alfentanil administration. Morley et al.4 described closed-loop control of a propofol/ alfentanil mixture using BIS, but showed no clinical advantage between closed-loop and manual control. The use of an isobole controller has been reported for the closed-loop coadministration of a hypnotic and an opioid drug29 and was tested in 1 dog. We have reported 2 cases using the propofol-remifentanil closed-loop controller in a 9-year-old boy requiring emergency lung volume reduction30 and in a patient with gigantism.31 The dual-loop controller successfully administered propofol and remifentanil. Because our controller continuously titrates drug effect against BIS, it compensates for the shortcomings of pharmacokinetic models.
Clinical hypotheses for the development of our controller were that closed-loop control of propofol maintains a continuous stable hypnotic level (BIS between 40 and 60) in the absence of noxious stimuli and that painful intraoperative stimuli provokes cortical activation with consequent increases in BIS.11,12 Description of the controller is detailed in Figure 1, the Appendix, and in Table 1. Briefly, the controller in the current study measures and calculates the error (BISerror), which is the difference between the set point (BIS = 50) and the measured BIS. If the BIS error is different from 0, the controller determines a new propofol and/or remifentanil concentration. The error size determines which drug will be modified: if the BIS error is small, only the remifentanil is changed; if the BIS error is higher, the 2 drug concentrations are changed (Table 1). The minimal interval between 2 consecutive controls is set equal to the time to peak effect of each drug with an additional delay of 60 seconds; this time interval is shorter for remifentanil14 than for propofol,13 thus remifentanil modifications are made more frequently.
The GS was chosen as the primary outcome for evaluation of controller performance. This score has previously been used5,6,22 but it has not been extensively validated, leaving our primary outcome consequently open to criticism. However, a particular goal of automated control is to keep the average value of the controlled variable within defined limits with low oscillations.32 These criteria are summarized by the GS, which weights the inaccuracy or MDAPE and the wobble16 against the percentage of time with BIS values between 40 and 60. The MDAPE represents the precision or the oscillation around the set point (BIS = 50). The wobble measures the intraindividual variability of PE, but the wobble can be low (or excellent) while the BIS values are outside of the range (40–60). Consequently, the wobble cannot alone assess the controller performance and the GS avoids a misinterpretation.5 Finally, GS, MDAPE, percentage of BIS in the range 40 to 60 (Fig. 4), <40, >60, or period of SR are in agreement and demonstrate a better control of the BIS using the closed-loop controller. The controller decreases the episodes of too deep anesthesia (BIS <40) associated with the occurrence of SR (Table 4). The occurrence of SR may be attributable to a variety of factors other than hypnotic drug overdosing such as changes in metabolism and/or perfusion pressure, hypothermia, or hypoxemia. However, the occurrence of an isoelectric EEG increases mortality in critically ill patients33,34 and SR is related to too deep anesthesia, which has been proposed as a cause of increased long-term mortality.35–38
Hemodynamic responses to painful stimuli (laryngoscopy or surgical incision), recorded in only a fraction of patients, were similar with closed-loop and manual control, which suggests that closed-loop control provides acceptable control of heart rate and arterial blood pressure. However, in the absence of specific or objective clinical signs or specific analgesia monitors, we cannot evaluate whether remifentanil administration was optimal.
Our study also shows some limits of our controller. The first limitation is the duration of the induction phase, which is mostly comprised between 3.4 and 6.1 minutes (interquartile range values). Such a duration, obviously too long for low-risk patients, is related to the delay between each new modification of propofol or remifentanil concentration (Appendix). An evolution of our controller could be the decrease of this delay in low-risk patients. The second limitation is remifentanil consumption. Whereas propofol consumption was similar in both groups, consumption of remifentanil was higher in the dual-loop group (Tables 3 and 4, and Fig. 3). Similarly, remifentanil effect-site concentrations were higher in the dual-loop group both during the induction phase and during maintenance. We plan to modify the controller, gain constant and limits of remifentanil, to decrease remifentanil consumption. Comparison of drug consumption or drug concentration with other studies is difficult because of different populations: one-third had major surgery, and 15% had combined general anesthesia and regional block. Finally, high remifentanil consumption can be considered harmful if it increases the occurrence of adverse events such as hemodynamic instability or awareness. Our trial lacks power to determine the incidence of awareness. The third limitation is the number of dropouts (Fig. 2). Some were attributable to errors in the conduct of the study by the investigators, i.e., patients with neurological disorders, dilution, or drug errors. Prolonged artifacts of BIS cannot be attributed to our system unlike recording system failures and computer failure. All of these cases are explained by the use of a prototype device in a conventional personal computer with Windows XP as the directory operating system.
Other points must be discussed. We defined an induction phase as the delay between the beginning of drug infusion and BIS value ≤60 for 30 seconds. We chose such a definition to avoid the risk of human error at the appearance of a clinical sign such as loss of verbal contact and the risk of a modification of the hypnotic state caused by repeated patient stimulation. The period of 30 seconds was arbitrary but it is usual to see the BIS value oscillating around 50 after an initial abrupt decrease. A shorter period would not have been enough. Second, our study was unblinded, multicentered, and had many investigators. The relatively high number of investigators is a strength showing that the controller works in other hands than those of the inventors. However, it is also a weakness because several points of patient care were left to the usual practice of the investigators, making the comparisons between the groups imprecise. This is particularly true when discussing hemodynamic profiles, requirement for fluid or vasoactive drugs, and extubation. Third, we compared our controller against an anesthesiologist-based control; this methodology has been used to study a single BIS-propofol closed-loop controller.5–7,23 Other approaches could be proposed in case of a dual controller to clarify the interaction between drugs: comparison of the propofol-remifentanil closed-loop controller to a single closed loop of propofol with a continuous and fixed infusion at different targets of remifentanil, or with a continuous fixed infusion of propofol at different targets associated with a single remifentanil closed-loop controller guided by BIS. Finally, our aim was to compare the current closed-loop controller as a new method for drug administration, with manual TCI of propofol and remifentanil, a present-day practice.
In conclusion, electrocortical activity given by the BIS monitor allows computerized control of simultaneous propofol and remifentanil administration. Several improvements of the controller are required before trials on large populations to demonstrate that automated control of anesthetic delivery is a useful tool.
Conflict of Interest: Hôpital Foch, N. Liu, T. Chazot, and B. Trillat are patent holders in France for the gain constants and the control algorithm (No. BFF08P669, Institut National de la Propriété Industrielle, France).
Name: Ngai Liu, MD, PhD.
Conflicts of Interest: Dr. Liu is a patent holder in France for the gain constants and the control algorithm (No. BFF08P669, Institut National de la Propriété Industrielle, France).
Name: Thierry Chazot, MD.
Conflicts of Interest: Dr. Chazot is a patent holder in France for the gain constants and the control algorithm (No. BFF08P669, Institut National de la Propriété Industrielle, France).
Name: Sophie Hamada, MD.
Conflicts of Interest: None
Name: Alain Landais, MD.
Conflicts of Interest: None
Name: Nathalie Boichut, MD.
Conflicts of Interest: None
Name: Corinne Dussaussoy, MD.
Conflicts of Interest: None
Name: Bernard Trillat, MSc.
Conflicts of Interest: Dr. Trillat is a patent holder in France for the gain constants and the control algorithm (No. BFF08P669, Institut National de la Propriété Industrielle, France).
Name: Laurent Beydon, MD.
Conflicts of Interest: None
Name: Emmanuel Samain, MD.
Conflicts of Interest: None
Name: Daniel I. Sessler, MD.
Conflicts of Interest: None
Name: Marc Fischler, MD.
Conflicts of Interest: None
1. John ER, Prichep LS. The anesthetic cascade: a theory of how anesthesia suppresses consciousness. Anesthesiology 2005; 102:447–71
2. Gan TJ, Glass PS, Windsor A, Payne F, Rosow C, Sebel P, Manberg P. Bispectral index monitoring allows faster emergence and improved recovery from propofol, alfentanil, and nitrous oxide anesthesia. BIS Utility Study Group. Anesthesiology 1997;87:808–15
3. Johansen JW, Sebel PS. Development and clinical application of electroencephalographic bispectrum monitoring. Anesthesiology 2000;93:1336–44
4. Morley A, Derrick J, Mainland P, Lee BB, Short TG. Closed loop control of anaesthesia: an assessment of the bispectral index as the target of control. Anaesthesia 2000;55:953–9
5. Liu N, Chazot T, Genty A, Landais A, Restoux A, McGee K, Laloe PA, Trillat B, Barvais L, Fischler M. Titration of propofol for anesthetic induction and maintenance guided by the bispectral index: closed-loop versus manual control—a prospective, randomized, multicenter study. Anesthesiology 2006;104:686–95
6. Puri GD, Kumar B, Aveek J. Closed-loop anaesthesia delivery system (CLADS) using bispectral index: a performance assessment study. Anaesth Intensive Care 2007;35:357–62
7. Hemmerling TM, Charabati S, Zaouter C, Minardi C, Mathieu PA. A randomized controlled trial demonstrates that a novel closed-loop propofol system performs better hypnosis control than manual administration. Can J Anaesth 2010;57:725–35
8. Schwilden H, Stoeckel H. Closed-loop feedback controlled administration of alfentanil during alfentanil-nitrous oxide anaesthesia. Br J Anaesth 1993;70:389–93
9. Glass PS. Anesthetic drug interactions: an insight into general anesthesia—its mechanism and dosing strategies. Anesthesiology 1998;88:5–6
10. Bouillon TW, Bruhn J, Radulescu L, Andresen C, Shafer TJ, Cohane C, Shafer SL. Pharmacodynamic interaction between propofol and remifentanil regarding hypnosis, tolerance of laryngoscopy, bispectral index, and electroencephalographic approximate entropy. Anesthesiology 2004;100:1353–72
11. Guignard B, Menigaux C, Dupont X, Fletcher D, Chauvin M. The effect of remifentanil on the bispectral index change and hemodynamic responses after orotracheal intubation. Anesth Analg 2000;90:161–7
12. Iselin-Chaves IA, Flaishon R, Sebel PS, Howell S, Gan TJ, Sigl J, Ginsberg B, Glass PS. The effect of the interaction of propofol and alfentanil on recall, loss of consciousness, and the bispectral index. Anesth Analg 1998;87:949–55
13. Schnider TW, Minto CF, Shafer SL, Gambus PL, Andresen C, Goodale DB, Youngs EJ. The influence of age on propofol pharmacodynamics. Anesthesiology 1999;90:1502–16
14. Minto CF, Schnider TW, Egan TD, Youngs E, Lemmens HJ, Gambus PL, Billard V, Hoke JF, Moore KH, Hermann DJ, Muir KT, Mandema JW, Shafer SL. Influence of age and gender on the pharmacokinetics and pharmacodynamics of remifentanil. I. Model development. Anesthesiology 1997;86:10–23
15. Cantraine FR, Coussaert EJ. The first object oriented monitor for intravenous anesthesia. J Clin Monit Comput 2000;16:3–10
16. Varvel JR, Donoho DL, Shafer SL. Measuring the predictive performance of computer-controlled infusion pumps. J Pharmacokinet Biopharm 1992;20:63–94
17. Leclerc C, Gerard JL, Bricard H. Intraoperative memory: a study of its incidence in general anesthesia in 326 patients [in French]. Ann Fr Anesth Reanim 2001;20:592–9
18. Struys MM, De Smet T, Versichelen LF, Van De Velde S, Van den Broecke R, Mortier EP. Comparison of closed-loop controlled administration of propofol using bispectral index as the controlled variable versus “standard practice” controlled administration. Anesthesiology 2001;95:6–17
19. Absalom AR, Sutcliffe N, Kenny GN. Closed-loop control of anesthesia using bispectral index: performance assessment in patients undergoing major orthopedic surgery under combined general and regional anesthesia. Anesthesiology 2002; 96:67–73
20. Absalom AR, Kenny GN. Closed-loop control of propofol anaesthesia using bispectral index: performance assessment in patients receiving computer-controlled propofol and manually controlled remifentanil infusions for minor surgery. Br J Anaesth 2003;90:737–41
21. Haddad WM, Bailey JM, Hayakawa T, Hovakimyan N. Neural network adaptive output feedback control for intensive care unit sedation and intraoperative anesthesia. IEEE Trans Neural Netw 2007;18:1049–66
22. Liu N, Chazot T, Trillat B, Michel-Cherqui M, Marandon JY, Law-Koune JD, Rives B, Fischler M. Closed-loop control of consciousness during lung transplantation: an observational study. J Cardiothorac Vasc Anesth 2008;22:611–5
23. Agarwal J, Puri GD, Mathew PJ. Comparison of closed loop vs. manual administration of propofol using the bispectral index in cardiac surgery. Acta Anaesthesiol Scand 2009;53:390–7
24. Hegde HV, Puri GD, Kumar B, Behera A. Bi-spectral index guided closed-loop anaesthesia delivery system (CLADS) in pheochromocytoma. J Clin Monit Comput 2009;23:189–96
25. Mendez JA, Torres S, Reboso JA, Reboso H. Adaptive computer control of anesthesia in humans. Comput Methods Biomech Biomed Engin 2009;12:727–34
26. Gentilini A, Schaniel C, Morari M, Bieniok C, Wymann R, Schnider T. A new paradigm for the closed-loop intraoperative administration of analgesics in humans. IEEE Trans Biomed Eng 2002;49:289–99
27. Luginbuhl M, Bieniok C, Leibundgut D, Wymann R, Gentilini A, Schnider TW. Closed-loop control of mean arterial blood pressure during surgery with alfentanil: clinical evaluation of a novel model-based predictive controller. Anesthesiology 2006;105:462–70
28. Hemmerling T, Charabti S. McSleepy™: a completely automatic anesthesia delivery system. Anesthesiology 2009:A460
29. Zhang XS, Roy RJ, Huang JW. Closed-loop system for total intravenous anesthesia by simultaneously administering two anesthetic drugs. IEEE Eng Med Biol Soc 1998;6:3052–5
30. Liu N, Bourgeois E, Chazot T, Murat I, Fischler M. Closed-loop administration of propofol and remifentanil guided by the bispectral index in patient requiring an emergency lung volume reduction. Paediatr Anaesth 2007;17:909–10
31. Declerck A, Liu N, Gaillard S, Chazot T, Laloe PA, Fischler M, Bertherat J. Closed-loop titration of propofol and remifentanil guided by bispectral index in a patient with extreme gigantism. J Clin Anesth 2009;21:542–4
32. O'Hara DA, Bogen DK, Noordergraaf A. The use of computers for controlling the delivery of anesthesia. Anesthesiology 1992;77:563–81
33. Watson PL, Shintani AK, Tyson R, Pandharipande PP, Pun BT, Ely EW. Presence of electroencephalogram burst suppression in sedated, critically ill patients is associated with increased mortality. Crit Care Med 2008;36:3171–7
34. Seder DB, Fraser GL, Robbins T, Libby L, Riker RR. The bispectral index and suppression ratio are very early predictors of neurological outcome during therapeutic hypothermia after cardiac arrest. Intensive Care Med 2010;36:281–8
35. Monk TG, Saini V, Weldon BC, Sigl JC. Anesthetic management and one-year mortality after noncardiac surgery. Anesth Analg 2005;100:4–10
36. Lindholm ML, Traff S, Granath F, Greenwald SD, Ekbom A, Lennmarken C, Sandin RH. Mortality within 2 years after surgery in relation to low intraoperative bispectral index values and preexisting malignant disease. Anesth Analg 2009;108:508–12
37. Leslie K, Myles PS, Forbes A, Chan MT. The effect of bispectral index monitoring on long-term survival in the B-Aware Trial. Anesth Analg 2010;110:816–22
38. Kertai MD, Pal N, Palanca BJ, Lin N, Searleman SA, Zhang L, Burnside BA, Finkel KJ, Avidan MS; B-Unaware Study Group. Association of perioperative risk factors and cumulative duration of low bispectral index with intermediate-term mortality after cardiac surgery in the B-Unaware Trial. Anesthesiology 2010;112:1116–27
In both groups, patients received total IV anesthesia in target-controlled infusion (TCI) mode using the Infusion Toolbox 95® version 4.11 software.15 Infusion Toolbox 95 is programmed in Visual Smalltalk, an object oriented language (VisualAge for Smalltalk® version 5.5; IBM, Armonk, NY). The software allows calculated plasma and effect-site concentrations (Ce) of propofol and remifentanil. The population pharmacokinetic sets of Schnider et al.13 and Minto et al.14 were selected for propofol and remifentanil, respectively. The software steers 2 Asena GH® infusion pumps (Alaris Medical UK Ltd., Basingstoke, Hampshire, UK). The Bispectral Index (BIS) electrodes were positioned on the forehead and connected to either an A-2000 XP (version 3.11) BIS monitor (Covidien, Mansfield, MA) or a BIS Module (Datex-Ohmeda™ S/5™, Helsinki, Finland). A standard personal computer running with Windows 98® or XP® (Microsoft, Redmond, WA) is used to provide a user interface, to store BIS, signal quality index, electromyographic (EMG) activity, and suppression ratio data every 5 seconds and to control communication via an RS232 serial port between the monitor and the 2 infusion pumps.
1. Main Elements of the Controller
The controller has a cascade structure including 2 proportional-integral-derivative (PID) controllers, each one having its own PID algorithm, and an interaction rule (Fig. 1). The controller includes 6 main elements:
- Calculation of the BISerror (difference between the set point of 50 and the actual unfiltered BIS value): it allows the titration until the target level of BIS = 50 is obtained. This parameter is calculated only if the signal quality index given by the monitor is >50%. New drug concentrations are calculated according to the “error” size and sign and to actual concentrations. The controller increases or decreases the drug concentrations according to the “error” sign (positive or negative). The “error” size determines which drug will be modified. If the BISerror is small, only the remifentanil is changed; if the BISerror is higher than a threshold, the 2 drug concentrations are changed. Finally, the 2 agents are continuously modified by the controller. The delay between each modification is shorter and the trigger threshold lower for the remifentanil than for the propofol. Consequently, the controller is more reactive for remifentanil.
- Amplification of the feedback (AFB): a specific AFB has been determined for each drug and for each BIS “error.” A new target is determined by the following equation, which corresponds to an integral controller:
The controller continuously modifies the target concentration until a BISerror = 0 is obtained. The new target is calculated by using the current target, which was determined by the previous BISerror. In fact, the controller sums the instantaneous errors over time (integrating the error every 5 seconds throughout the maintenance phase), giving the accumulated offset that should have been corrected previously. The new target calculated by the use of this accumulated error provides the PID integral action.
The values of the gain constants or K values for propofol and remifentanil are given in Table 1; these values were determined empirically using the simulator included in the Infusion Toolbox 95 software.
- Delay between each new modification of propofol or remifentanil concentration: it is determined by the time necessary for equilibration of the previous effect-site compartment given by the pharmacokinetic models (Fig. 1). For a given patient, the delay given by the pharmacokinetic model is constant. It depends especially on age. For propofol, the delay varies linearly between 96 seconds (20 years old) and 120 seconds (80 years old). For remifentanil, the delay varies nonlinearly between 80 seconds (20 years old) and 151 seconds (80 years old). Remifentanil delays were calculated using the simulator included in the Infusion Toolbox 95 software.15 An additional period of 60 seconds is added systematically during the maintenance phase.
- Feedforward, derivative term of PID algorithm: its 2 components, 1 for propofol and 1 for remifentanil, check the profile every 5 seconds and decide on a concentration correction proportionally to the BISerror. It is activated in 3 circumstances: (i) when the BIS is >62 (propofol component) or 60 (remifentanil component), (ii) when the slope of the BIS increases >15 (propofol component) or 10 (remifentanil component) in 10 seconds, and (iii) when the EMG activity is >37 dB (propofol component) or 35 dB (remifentanil component).
It is deactivated when the EMG is >42 dB for 1 minute, a figure related to an artifact. The derivative action has the priority for the decisions of concentration modifications. Furthermore, when the current propofol concentration is <1.3 μg · mL−1or the current remifentanil concentration is <4 ng · mL−1, default corrections (1.8 μg · mL−1 and 6 ng · mL−1, propofol and remifentanil, respectively) are performed to avoid a too small or not clinically relevant correction. The derivative action is inhibited during induction. Finally, during the maintenance phase, when the BIS or EMG limits are exceeded, then the propofol and remifentanil target concentrations are updated according to AFB.
- Interaction rule between propofol and remifentanil: if the controller increases the remifentanil concentration successively >3 times, then the controller increases the propofol concentration.
- Safety feature: the system automatically maintains the calculated drug concentrations in the case of controller or BIS dysfunction or low signal quality index (SQI <50). Furthermore, minimum and maximum (default values) of concentrations are set at 1 and 5 μg · mL−1 for propofol and at 3 and 12 ng · mL−1 for remifentanil during maintenance. The user can modify these values without limits.
2. Clinical Use of the Controller
All investigators were clinically experienced in the use of BIS to titrate propofol and remifentanil, and all received 2 days' training on the dual closed-loop controller at our center. The user enters the patient's demographic data (sex, age, weight, and height) and the initial propofol target concentration for induction. The controller decides the first remifentanil concentration related to initial propofol concentration: if the initial propofol target is <2.5 μg · mL−1, then the initial remifentanil target is 4 ng · mL−1; if the initial propofol target is between 2.5 and 2.9 μg · mL−1, then the initial remifentanil target is 5 ng · mL−1; if the initial propofol concentration is >3 μg · mL−1, then the initial remifentanil target is 6 ng · mL−1. The derivative action or feedforward is inhibited during induction. After the induction phase (i.e., BIS ≤60 for 30 seconds), the controller switches automatically to the maintenance phase. Throughout the procedure, the user can adjust the targets if necessary or switch between closed-loop and manual control.
3. Summary of the Controller
When starting a procedure, the clinician enters an initial target value of Ce for the propofol Ceprop (I) and the controller sets an initial target value of Ce for the remifentanil Ceremi (I):
- if Ceprop (I) < 2.5 μg · mL−1, then Ceremi (I) = 4.0 ng · mL−1
- if 2.5 μg · mL−1 < Ceprop (I) < 2.9 μg · mL−1, then Ceremi (I) = 5.0 ng · mL−1
- if Ceprop (I) > 3.0 μg · mL−1, then Ceremi (I) = 6.0 ng · mL−1
The following rules are used to update the new concentrations until BISerror = 0 is obtained:
where Ceprop (k) and Ceremi (m) are the new target of propofol or remifentanil, Ceprop (k − 1) and Ceremi (m − 1) are the current values of Ce, and BISerror is the difference between the current BIS value and 50. Table 1 gives the value for Kprop and Kremi as a function of the error BISerror.
A new target was updated after the time necessary for equilibration of the previous effect-site compartment given by the pharmacokinetic models. For a given patient, the delay given by the pharmacokinetic model is constant. It depends especially on age. The propofol delay varies linearly between 96 seconds (20 years old) and 120 seconds (80 years old).13 For remifentanil, the delay varies nonlinearly between 80 seconds (20 years old) and 151 seconds (80 years old).14 Remifentanil delays were calculated using the simulator included in the Infusion Toolbox 95 software.15 An additional period of 60 seconds is added systematically during the maintenance phase.
However, every 5 seconds a new correction is possible while the concentration between plasmatic and effect-site compartment is not in steady state, thanks to the derivative term if:
- BIS >62, then Ceprop (k) is updated.
- BIS >60, then Ceremi (m) is updated.
- BISerror >15, then Ceprop (k) is updated.
- BISerror >10, then Ceremi (m) is updated.
- EMG >37 dB and EMG <42 dB, then Ceprop (k) is updated.
- EMG >35 dB and EMG <42 dB, then Ceremi (m) is updated. If none of the 6 conditions listed above are met, the previous actions or rules 1 and 2 are invoked. During the maintenance phase, which is the period after the induction phase (defined as the time from the start of propofol and remifentanil administration to BIS <60 during 30 seconds) until the end of drug administration, we have added a supplementary delay of 60 seconds between each concentration modification. Moreover, if Ceremi is updated 3 times in a row, then Ceprop is automatically updated once.