Objective: Therapeutic drug monitoring (TDM) analysis software is often used in clinical settings to estimate the dosage for vancomycin (VCM). In this study, the new TOWA-TDM analysis software was evaluated and compared with an existing analysis software based on predictability and Bayesian convergence.
Methods: The abilities of two TDM analysis software, TOWA-TDM ver.1.0.0 and “Vancomycin MEEK TDM analysis software ver. 3.2” (MEEK), were compared retrospectively. Twenty patients who received VCM therapy were randomly chosen, and clinical data were obtained from their medical records. The VCM trough concentrations were estimated from the clinical data and compared with the observed trough concentrations. Predictive performance was evaluated by calculating the mean prediction error (ME), mean absolute prediction error (MAE), and root mean square error (RMSE). Moreover, the pharmacokinetic parameters of each patient were estimated via Bayesian analysis using the observed trough level and then compared between the two software.
Results: In the initial dosage design of VCM using population pharmacokinetic (PPK) parameters, the predictability (MAE and RMSE) of the VCM trough levels showed no significant difference between the two software. However, the predicted trough values calculated using the TOWA-TDM were significantly lower than those calculated using MEEK. In Bayesian analysis, the estimated individual pharmacokinetic parameters differed between the software because of differences in the PPK equations and parameter restrictions employed; however, good convergence property to the observed trough values was shown by both the software.
Conclusions: This study indicates that TOWA-TDM has equivalent predictability and convergence (MAE and RMSE) with those of the existing software, MEEK, in the initial dosage design of VCM and Bayesian analysis. However, ME in the initial dosage design showed a significant difference between the two software owing to the difference in predicting renal function.
