Glasgow Meeting - May 2003

Dr A Absalom, Consultant in anaesthesia and intensive care, Norwich, UK

The two main types of control system are feedforward and feedback. With feedforward control no information from the process being controlled is used to correct the input to the system, whereas with feedback control this happens continuously. Feedback (closed loop) control is thus more suited to complex systems subjected to several influences, where the results of an input are less predictable. The essential components of a closed loop control system are summarised in Figure 1. To control the state of a system or process it is necessary to identify a control variable that represents the state of that process, and the desired value of that variable (setpoint). It is also essential to choose a control actuator that is able to alter the state of the process. A control algorithm determines how to alter the control actuator to reduce the error (setpoint - actual value) of the control variable. The most commonly used control algorithms are "on-off," Proportional, Proportional-Integral or Proportional-Integral-Differential (PID). Recent advances in computer technology and knowledge of control theory have enabled the development of adaptive or self-learning, model-based and neuro-fuzzy control algorithms. Anaesthesia is difficult to define and measure. In recent years interest has focussed on the electroencephalogram (EEG) - some of the variables derived from EEG analysis correlate with anaesthetic depth. EEG variables that have been used to control anaesthesia include the median frequency¹ auditory evoked potential² and Bispectral Index (BISÔ, Aspect Medical Systems, Newton, USA). Hypnotic drugs that have been used to control anaesthesia include isoflurane³ methohexitone⁴ alfentanil⁵ and propofol². Today’s talk will focus on those systems in which the BIS, is used to control general anaesthesia or sedation.

Description of published systems

Five BIS-guided closed loop systems have been described in the literature (Table 1). Six formal studies of the performance of 3 of these systems have been conducted and published. The systems have been tested in a variety of circumstances – different patient groups, different types of surgery, different noxious stimuli. None have been tested in elderly or unfit patients. The most commonly used method of assessing the performance of closed loop systems is that recommended by Varvel.¹⁴ Table 2 summarises the results of the published studies of the performance of closed loop BIS systems.

Discussion

For optimal control a closed loop system requires:

Clearly depth of anaesthesia is controllable – even anaesthetists can do it! Looking at the other requirements, none of the existing closed loop anaesthesia systems is ideal. All contain one or more sources of phase delay in the input to or output from the system, producing the potential for oscillation and drift. All EEG-based parameters require a processing delay, and even with effect-site targeted target-controlled infusion systems it is not possible to effect stepwise changes in the effect-site propofol concentration. Nonetheless BIS-guided closed loop anaesthesia systems work remarkably well despite the fact that some of them use fixed-constant PID algorithms (tuned for population averages) and/or actuators that simply control an infusion rate rather than a blood or effect-site propofol concentration. All systems have provided good haemodynamic stability, reasonable stability of the control variable and acceptable operating conditions. Intuitively one would expect that systems using adaptive control algorithms and effect-site targeted drug infusion systems should perform best. However when tested the performance of the systems if fairly similar. The small numbers of patients studied limits the significance of any comparisons between the results of the different studies. Morley et al and Struys et al have compared the accuracy of computer control with manual (anaesthetist) control, and in both cases found little difference.

References

1. Schwilden H et al. Br J Anaesth 1989; 62: 290-6

2. Kenny G, Mantzaridis H. Br J Anaesth 1999; 83: 223-8

3. Robb HM et al. Br J Anaesth 1991; 66: 358-364

4. Schwilden H et al. Anesthesiology 1987; 67: 341-7

5. Schwilden H, Stoeckel H. Br J Anaesth 1993; 70: 389-393

6. Guignard B et al. Anesthesiology 1998; 89: A1218

7. Mortier E et al. Anaesthesia 1998; 53: 749-754

8. Morley A et al. Anaesthesia 2000; 55: 953-9

9. Sakai T et al. Acta Anaesthesiol Scand 2000; 44: 1007-1010

10. Absalom A et al. Anesthesiology 2002; 96(1):67-73

11. Leslie K et al. Anaesthesia 2002; 57: 690-709

12. Struys MRF et al. Anesthesiology 2001; 95: 6-17

13. Absalom A, Kenny GNC. Br J Anaes 2003; in press

14. Varvel JR et al. J Pharmacokinet Biopharm 1992; 20: 63-94