Prospective evaluation of pharmacokinetic model-controlled infusion of propofol in adult patients.

We have tested prospectively the algorithm of White and Kenny for pharmacokinetic model-controlled infusion of propofol in 40 healthy Oriental adults. Anaesthesia consisted of a target-controlled infusion of propofol, 70% nitrous oxide and an infusion of alfentanil. For the first 20 patients studied, median performance error was -5%, median absolute performance error 19%, divergence -9% and wobble 6%. The model was revised to minimize the prediction error of the variable set and tested prospectively in another 20 patients. There was a deterioration in performance of the revised model. Performance statistics for the original model in all 40 patients were: median performance error 2% (range -34 to 69%), median absolute performance error 21% (6-69%), divergence -17% (-92 to 49%) and wobble 7% (2-34%). The algorithm was found to perform adequately in our Oriental patient population.