Performance of Software in Predicting Teicoplanin Serum Trough Concentrations Using Two Sets of Population Pharmacokinetics Parameters

The Teicoplanin TDM analysis software provided by Astellas Pharma Inc.was updated to Ver.2.0 and we tested it in comparison with Ver.1.0 to determine whether its predictions based on population pharmacokinetic parameters were useful in the design of initial and maintenance regimens.In addition,since renal function may be overestimated when the serum creatinine concentration is low,we determined whether correction of serum creatinine concentrations improved predictive capability with respect to initial regimens.As regards predictive performance,for initial regimens,the mean prediction error (ME) for Ver.1.0 was greater than zero and the 95% confidence interval did not include zero.These results indicate that concentrations predicted using Ver.1.0 would be significantly higher than the observed concentrations.The ME for Ver.2.0 was smaller and the 95% confidence interval included zero indicating that its predictive performance was superior to that of Ver.1.0.However,the correction of serum creatinine concentrations did not improve predictive performance for initial regimens.Using the Bayesian method,no significant correlation was observed between the observed and predicted concentrations for maintenance regimens for either Ver.1.0 or Ver.2.0 due to the small patient population used.With respect to observed concentrations at blood sampling times from 48 to 72 h from the start of administration,the predictive capabilities of both Ver.1.0 and Ver.2.0 were good.The results of this study indicated that for initial regimens,the method used for Ver.2.0 would be more useful,though the correction of serum creatinine concentrations did not improve predictive performance in this case.