抄録
Five population parameter sets for intermittent intravenous drip infusion of vancomycin (VCM) were evaluated retrospectively to assess the effect of those on the predictive performance of the Bayesian method employing two measurements of one hour after the end of infusion and within 15 minutes of dose initiation (trough) in predicting plasma VCM concentration. Population mean values studied were taken from the literature of Moellering et al.(I), Matzke et al.(II), Uaamnuichai et al.(III), Grasela et al.(IV), and Burton et al.(V).
Root mean squared errors (RMSEs) as a composite measure of bias and precision were 8.16, 8. 38, 8.42, 7.95 and 8. 00μg/ml in predicting the one-hour plasma concentration after the end of infusion, and were 5.04, 5. 53, 5. 32, 5. 18 and 5. 83 μg/ml in predicting the trough plasmaconcentration for the method using Is, Ills, IVs and Vs, respectively. The mean prediction error (ME) as a measure of bias demonstrated that the predictive performance of the Bayesian method in predicting trough plasma concentration underpredicted the measured concentration significantly.
On reevaluation of Grasela's parameters in patients with different degrees of renal function, the predictive performances of the Bayesian method were more precise than those available by population parameter. The predictive performance of the Bayesian method employing two measurements using Grasela's parameters is concluded to be clinically acceptable.
