Closed-loop control of propofol anaesthesia using bispectral indexTM: performance assessment in patients receiving computer-controlled propofol and manually controlled remifentanil infusions for minor surgery{dagger}

A. R. Absalom and G. N. C. Kenny

University Department of Anaesthesia, Glasgow Royal Infirmary, Glasgow, UK

{ddagger}LMA® is the property of Intavent Limited.

Accepted for publication: February 19, 2003


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Background. In a previous study we used the bispectral index (BIS)TM for automatic control of propofol anaesthesia, using a proportional-integral-differential control algorithm. As control was less than optimal in some patients, we revised the constants of the control algorithm. The aim of the current study was to measure the performance of the revised system in patients undergoing minor surgery under propofol and remifentanil anaesthesia.

Methods. Twenty adult patients scheduled for body surface surgery were enrolled. Anaesthesia was manually induced with target-controlled infusions (TCI) of propofol and remifentanil. After the start of surgery, when anaesthesia was clinically adequate, automatic control of the propofol TCI was commenced using the revised closed-loop system. For patients 11–20, effect-site steering was also incorporated into the closed-loop control algorithm. Adequacy of anaesthesia during closed-loop control was assessed clinically, and by calculating the median performance error (MDPE), the median absolute performance error (MDAPE) and the mean offset of the control variable.

Results. The system provided adequate operating conditions and stable cardiovascular values in all patients during closed-loop control. The mean MDPE and MDAPE were –0.42% and 5.63%, respectively. Mean offset of the BISTM from setpoint was –0.2. No patients reported awareness or recall of intraoperative events.

Conclusions. The system was able to provide clinically adequate anaesthesia in all patients, with better accuracy of control than in the previous study. There was a tendency for more accurate control in those patients in whom the control algorithm incorporated effect-site steering.

Br J Anaesth 2003; 90: 737–41

Keywords: anaesthesia, general; anaesthetics i.v., propofol; monitoring, bispectral index


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
We have produced a closed-loop anaesthesia system that uses the bispectral index (BIS)TM as the control variable to automatically control the target blood concentration of a propofol target-controlled infusion (TCI) system. In a previous study, the system was tested in patients undergoing hip and knee surgery under combined general and regional anaesthesia. Although it was able to control anaesthesia, control was less than optimal.1 In an attempt to improve its performance, the system has been revised as described below. The revised system has also been used for closed-loop control of sedation.2 The aim of the current study was to assess the performance of the revised system in a group of patients undergoing minor surgery, with analgesia provided by a remifentanil TCI (target concentration 4 ng ml–1).


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Description of closed-loop anaesthesia system
The software and hardware comprising the closed-loop anaesthesia system have been described previously.1 However, some changes have been made to the system since then. The current system used the same BISTM monitor, but with a newer version of the BISTM software (version 3.3, Aspect Medical Systems, Newton, USA). Small adjustments were made to the gain constants used in the control algorithm—the proportional constant was decreased from 0.25 to 0.2, and the integral constant was increased from 0.5 to 0.75. In all other respects the control algorithm for the first 10 patients was the same as that previously published.

For the second group of 10 patients, the control algorithm was altered to enable ‘effect-site steering’ in an attempt to reduce the problems of oscillation seen in the previous study. With effect-site steering, the system adjusts the target blood concentration taking into account the estimated effect-site concentration. With the original version of the algorithm used in the first 10 patients, if anaesthesia was inadequate (BISTM>setpoint), the system increased the target concentration above the previous target blood concentration, regardless of the estimated effect-site concentration. If the effect-site concentration was above the blood concentration, then until the blood concentration had surpassed it, the effect-site concentration would decrease, causing even more inadequate levels of anaesthesia. Under the same conditions in the second group of 10 patients, the system immediately increased the target blood concentration to the effect-site concentration, with further increases after that as necessary. This avoids the inappropriate reduction in effect-site concentration mentioned above, and results in a faster rise in this concentration when anaesthesia is inadequate. The opposite applies if the BISTM is below the setpoint, and the blood propofol concentration above that at the effect site. Under these circumstances in the second group of 10 patients, the target blood concentration would immediately be set to below that at the effect-site, in order to reduce the effect-site concentration as quickly as possible.

Clinical protocol
After obtaining local research ethics committee approval and written informed consent, 20 adults presenting for body surface surgery were enrolled. To be included, patients had to be of ASA status I or II, and aged between 18 and 80 yr. Exclusion criteria included BMI greater than 30 kg m–2, a history of neurological disease and use of psychoactive medication. No sedative premedication was used. Two anaesthetists were involved with each case – one was in charge of the clinical management of the patient, while the other took care of the research equipment and manually recorded the BISTM, physiological data and the blood and effect-site propofol concentrations every 5 min.

Anaesthesia was induced in the operating theatre. A 20-gauge cannula was inserted into a vein on the dorsum of the hand, and the patient was connected to the BISTM monitor using four Zipprep electrodes (Aspect MS, Newton, USA) in the standard montage. While other monitors were being connected, the closed-loop anaesthesia system was started in manual mode, with the target propofol concentration set at 1–2 µg ml–1 to provide a degree of anxiolysis. Routine physiological monitoring (pulse oximetry, ECG, non-invasive arterial pressure) was started and baseline values recorded while the patient breathed 100% oxygen. The target remifentanil concentration was then set at 2 ng ml–1 and the target propofol concentration at 4 µg ml–1 in younger patients and at 2.5 µg ml–1 in the elderly. The target propofol concentration was increased by 0.5 µg ml–1 every 30 s until the patient had lost consciousness and was able to tolerate insertion of the laryngeal mask airway (LMA{ddagger}). Once the LMA had been inserted, the target remifentanil concentration was increased to 4 ng ml–1 and the lungs were mechanically ventilated with oxygen 40% in air via a circle breathing system. The target propofol concentration was reduced to approximately 0.5 µg ml–1 above the estimated effect-site concentration during LMA insertion.

Once surgery had started, a note was made of the BISTM value at which the level of anaesthesia was clinically adequate (no patient movement, haemodynamic stability, absence of signs of autonomic activation). Automatic control of the propofol TCI was initiated using this BISTM value as the setpoint. The target remifentanil concentration was left unchanged at 4 ng ml–1. When the surgeon began the final skin sutures, the closed-loop system was switched to manual mode and the target propofol and remifentanil concentrations were set to zero. Patients remained in the operating room until they had regained consciousness, the LMA had been removed and they had correctly stated their date of birth.

The time, BISTM and estimated blood and effect-site propofol concentrations during the following events were recorded manually: loss of consciousness (eyelash reflex), just before intubation, just before skin incision, start of closed-loop control, end of surgery, eye opening, response to command and patient able to state date of birth.

Data analysis
Physiological data are presented as mean (SD) and time intervals as median (range). Performance of the system was assessed using the measures recommended by Varvel and colleagues:3 median prediction error (MDPE), median absolute prediction error (MDAPE), wobble and the mean offset. These variables were calculated for each patient and then summarized as mean (95% confidence interval [CI]). MDPE and MDAPE are measures of bias and precision, respectively; wobble is a measure of the intra-individual variability in performance error; offset is the difference between the measured value and the setpoint of the control variable. The proportion of time (during automatic control) that the BISTM was within 5, 10 and 15% of the setpoint was also calculated.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Twenty patients (12 females and 8 males) aged 30–72 yr (mean 43 yr) were enrolled. Their mean weight and height were 68 (SD 11) kg and 169 (9) cm, respectively. Median BISTM and calculated blood and effect-site propofol concentrations at key clinical endpoints before and after the period of automatic control of anaesthesia are summarized in Table 1. Median duration of automatic control of anaesthesia was 27.5 (12–86) min. During this time the BISTM was within 15% of the setpoint 85% of the time (Table 2), and cardiovascular and respiratory variables were stable. There was a minor degree of oscillation in only one patient, in whom the BISTM cycled between 45 and 62, while the blood propofol concentration cycled between 2.2 and 3.5 µg ml–1. Despite this, the heart rate and arterial pressure remained stable and the patient did not move in response to surgical stimuli.


View this table:
[in this window]
[in a new window]
 
Table 1 BISTM values and propofol concentrations at key events. Data are median (range)
 

View this table:
[in this window]
[in a new window]
 
Table 2 Control of BISTM as a percentage of total closed-loop anaesthesia time
 
The median BISTM chosen as the setpoint was 50 (40–65). Performance of the system in individual patients is summarized in Table 3. Overall performance variables were: MDPE –0.42% (95% CI –1.41 to 0.57%); MDAPE 5.63% (4.51–6.75%); wobble 5.16% (3.83–6.5%) and mean offset –0.24 (–0.83 to 0.35%). Performance was slightly better in the patients in whom effect-site steering was used (cases 11–20). In these patients the mean MDAPE, MDAPE, median wobble and mean offset were –1.39% (–2.75 to –0.02%), 4.66% (3.39–5.94%), 4.39% (3.08–5.70%) and –0.70 (–1.37 to –0.04%), respectively. This compared with 0.55% (–0.83 to 1.93%), 6.60% (4.73–8.47%), 5.94% (3.42–8.46%) and 0.23 (–0.78 to 1.24%), respectively, in the patients in whom simple blood concentration targeting was used (cases 1–10). These differences were not statistically significant. Figure 1 shows the offset values for all patients during feedback control of anaesthesia.


View this table:
[in this window]
[in a new window]
 
Table 3 BISTM setpoint and performance parameters for individual patients. MDPE, median prediction error; MDAPE, median absolute prediction error
 


View larger version (23K):
[in this window]
[in a new window]
 
Fig 1 Individual offset values (measured BISTM – setpoint) during automatic control of anaesthesia.

 
The median time interval between end of closed-loop control and eye opening was 8 (4–18) min, and the median interval between end of surgery and eye opening was 7 (3–12) min. Operating conditions were satisfactory in all patients except for one, who moved for a brief period after the remifentanil infusion had been inadvertently switched off. All patients were satisfied with their anaesthetic, and there was no evidence of explicit recall.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
We have studied the performance of a BISTM-guided, computerized anaesthesia system in 20 patients undergoing body surface surgery, and found that it was able to keep the control variable (BISTM) within satisfactory limits in terms of bias, precision and wobble. Haemodynamic variables were stable during automatic control of anaesthesia, and with the exception of one patient (where human error resulted in the patient moving for a short period), operating conditions were satisfactory in all patients.

This is the second study during general anaesthesia of the performance of this feedback system that automatically delivers propofol by TCI, using the BISTM as control variable. In the first study, the patients (n=10) were older (mean age 67 yr) and were undergoing hip or knee replacements under computer-controlled propofol anaesthesia supplemented by regional blockade.1 Overall control was satisfactory in that study (MDPE, MDAPE and wobble were 2.2%, 8.0% and 7.3%, respectively), but in three cases there was oscillation of the BISTM around the setpoint, with concomitant cycling of the blood and effect-site propofol concentrations. Also, a third patient regained consciousness briefly when the hip was manipulated vigorously, causing the surgical stimulus to increase suddenly. To reduce the likelihood of these problems, we altered the gain constants of the control algorithm and have incorporated effect-site steering; these alterations were associated with an improvement in control performance and a reduced incidence of oscillation. However, it would be unwise to conclude firmly that the revised algorithm is better, as it was tested in a younger and healthier patient group under different surgical and anaesthetic conditions.

As far as we are aware, there are no published, agreed limits for acceptability of the bias and precision of control systems in human studies. However, control performance in this study compares very favourably with that reported by others who used similar methods of assessing performance. Using the BISTM as the control variable, an adaptive control algorithm and a propofol TCI, Struys and colleagues4 reported a MDPE of –6.6%, a MDAPE of 7.7% and median wobble of 5.9%. Morley and colleagues5 used the BISTM (target of 50) and a proportional-integral-differential (PID) control algorithm to control the rate of infusion of a mixture of propofol and alfentanil, and reported a median absolute unweighted residual of 6.8. This equates to a MDAPE of 13.6%. Finally, an adaptive model-based feedback controller has been used to control muscle relaxation using administration of various drugs; MDPE, MDAPE, wobble and offset all varied between –0.3% and 1.9%.6

Kenny and Mantzaridis7 evaluated the performance of an auditory evoked potential (AEP)-based closed-loop anaesthesia system in 100 patients by calculating the proportion of time that the measured AEP was within 5%, 10% and 15% of the target AEP value. They reported figures of 65%, 90% and 99%, respectively – somewhat better than the corresponding figures found in the current study (Table 2). The control algorithm used in the AEP system was identical to that used in our first study of BISTM-guided propofol anaesthesia,1 but differs from that used in the current study only in respect of changes to the gain constants and the addition of effect-site steering. The AEP system also used a propofol TCI as the control actuator. Although the figures from the two studies are not directly comparable, they may reflect some of the differences between the two control variables. The AEP, by definition, continuously measures the EEG response to an auditory stimulus, whereas the BISTM measures phase and frequency relationships among component frequencies in the spontaneous surface EEG. This may explain why spontaneous-EEG-based variables are no better than chance alone at predicting a response to noxious stimulus, whereas the AEP index is able to predict a response better than chance.811 It may also explain the greater fluctuation of the control variable, possibly in response to varying levels of stimulus, when the BISTM is the control variable compared with the AEP as control variable.

At the start of automatic control of the propofol infusion, we judged the adequacy of anaesthetic depth according to some of the traditional clinical signs of anaesthetic depth: lack of patient movement, haemodynamic stability and absence of signs of autonomic activation. Although these signs are the main measures by which many anaesthetists judge the adequacy of anaesthesia in their day-to-day anaesthetic practice, they are not particularly sensitive or specific predictors of the presence or lack of awareness. Thus it is hardly surprising that a broad range of BISTM values was associated with these conditions at the time of skin incision (29–72) and a broad range of values used as the setpoint for automatic control of anaesthesia (40–65). Aspect Medical Systems recommends that the BISTM is maintained between 40 and 60 in order to prevent awareness. Although it can be argued that the setpoint was thus too high in two patients (62 and 65), and that higher setpoints may be associated with less stability of the control variable, the data from our study do not show this.

Effect-site targeting has the potential for making a feedback system respond more quickly and accurately than blood concentration targeting. In the current study the trend was for effect-site targeting to be associated with slightly worse bias but improved precision and reduced wobble but these changes did not reach statistical significance. A study with larger subject groups is required to definitively test with adequate power if effect-site steering is superior.

In conclusion, a closed-loop computer system using the BISTM as the control variable provided satisfactory control of propofol anaesthesia in patients having minor body surface surgery, with analgesia provided by a fixed blood concentration of remifentanil.


    Acknowledgements
 
The authors wish to thank Aspect International (Leiden, The Netherlands), who donated the EEG electrodes and loaned the authors the A-2000 monitor used in the study.


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
1 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[CrossRef][ISI][Medline]

2 Leslie K, Absalom A, Kenny GN. Closed-loop control of sedation for colonoscopy using the Bispectral Index. Anaesthesia 2002; 57: 693–7[ISI][Medline]

3 Varvel JR, Donoho DL, Shafer SL. Measuring the predictive performance of computer-controlled infusion pumps. J Pharmacokinet Biopharm 1992; 20: 63–94[ISI][Medline]

4 Struys MRF, De Smet T, Versichelen LFM, et al. Comparison of closed-loop controlled administration of propofol using Bispectral Index as the controlled variable versus ‘standard practice’ controlled administration. Anesthesiology 2001; 95: 7–17

5 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[CrossRef][ISI][Medline]

6 Kansanaho M, Olkkola KT. Performance assessment of an adaptive model-based feedback controller: comparison between atracurium, mivacurium, rocuronium and vecuronium. Int J Clin Monit Comput 1996; 13: 217–24

7 Kenny GN, Mantzaridis H. Closed-loop control of propofol anaesthesia. Br J Anaesth 1999; 83: 223–8[Abstract/Free Full Text]

8 Doi M, Gajraj RJ, Mantzaridis H, Kenny GN. Prediction of movement at laryngeal mask airway insertion: comparison of auditory evoked potential index, bispectral index, spectral edge frequency and median frequency. Br J Anaesth 1999; 82: 203–7[Abstract/Free Full Text]

9 Kurita T, Doi M, Katoh T, et al. Auditory evoked potential index predicts the depth of sedation and movement in response to skin incision during sevoflurane anesthesia. Anesthesiology 2001; 95: 364–70[CrossRef][ISI][Medline]

10 Katoh T, Suzuki A, Ikeda K. Electroencephalographic derivatives as a tool for predicting the depth of sedation and anesthesia induced by sevoflurane. Anesthesiology 1998; 88: 642–50[ISI][Medline]

11 Kochs E, Kalkman CJ, Thornton C, et al. Middle latency auditory evoked responses and electroencephalographic derived variables do not predict movement to noxious stimulation during 1 minimum alveolar anesthetic concentration isoflurane/nitrous oxide anesthesia. Anesth Analg 1999; 88: 1412–17[Abstract/Free Full Text]