Performance of Computer-Controlled Infusion of Propofol

Computer-controlled infusion of propofol is used with increasing frequency for the induction and maintenance of anesthesia. The performance of computer-controlled infusion devices is highly dependent on how well the implemented pharmacokinetic parameter set matches the pharmacokinetics of the patient. This study examined the performance of a computer-controlled infusion device when provided with five different pharmacokinetic parameter sets of propofol in female patients. The infusion rate-time data that had been stored on a disk from 19 female patients who had been given propofol by computercontrolled infusion, using the pharmacokinetic parameter set from Gepts et al. (Anesth Analg 1987;66:1256-63), were entered into a computer simulation program to recalculate predicted propofol concentrations that would have been obtained with four other pharmacokinetic parameter (Shafer et al., Anesthesiology 1988;69:348-56; Kirkpatrick et al., Br J Anesth 1988;60:146-50; Cockshott et al., Br J Anesth 1987;59:941P; Tackley et al., Br J Anesth, 1989;62:46-53) sets of propofol, had these been implemented. The performance error (PE) was determined for each measured blood propofol concentration, on the basis of each of the five pharmacokinetic parameter sets. Then, for each of the five pharmacokinetic parameter sets, the performance in the population was determined by the median absolute performance error (MDAPE), the median performance error (MDPE), the wobble (the median absolute deviation of each PE from the MDPE), and the divergence (the percentage change of the absolute PE with time). The MDPE and MDAPE were compared between the parameter sets by the multisample median test. The initially used pharmacokinetic parameter set from Gepts et al. resulted in a MDPE of 24% and MDAPE of 26%. In comparison with this parameter set (Gepts et al.), the computer simulations revealed that the pharmacokinetic parameter set of Kirkpatrick et al. resulted in a significantly worse performance (MDPE, and MDAPE: 106%, P < 0.001), whereas with the three other pharmacokinetic parameter sets the performance did not differ. For all five pharmacokinetic parameters sets the divergence (median and range) in the patients in Group A, who had received a stepwise increasing target propofol concentration, was significantly greater (median 42%; range, 31%-59%) compared to the corresponding divergence in the patients in Group B (median 1%; range -18%-4%; P < 0.05), who had received a single constant target propofol concentration. The PE thus did not increase with time but with increasing target propofol concentration. In conclusion, the pharmacokinetic parameter sets of propofol described by Gepts et al., Shafer et al., Cockshott et al., and Tackley et al. result in an equally clinical acceptable, but not optimal, performance of the computer-controlled infusion of propofol in the type of patients studied above. With all five pharmacokinetic parameter sets, the underprediction of the measured concentration increases with the increasing target concentration. (Anesth Analg 1995;81:1275-82)