A Bayesian graphic method for predicting individual phenytoin dosage schedule.

A simple and practical graphic method based on Bayesian feedback theory for predicting individual phenytoin dosage was evaluated. Compared with the mathematical calculation, the graphic method only needs one pair of steady-state phenytoin concentration-dose data to predict individual phenytoin dosage. The observed drug levels were compared with predicted ones in 12 epileptic out-patients with a correlation coefficient of 0.94. The mean error between the observed and predicted phenytoin levels was 0.62 mg.L-1. The mean kinetic parameters Vm and Km obtained from the graph were 7.01 +/- SD 0.88 mg.kg-1.d-1 and 4.93 +/- 0.91 mg.L-1, respectively.