Comparison of predictive accuracy of teicoplanin concentration using creatinine clearance and glomerular filtration rate estimated by serum creatinine or cystatin C.

We compared the predictive accuracy of TEIC concentrations (TEIC_conc) calculated using either serum cystatin C (CysC) or serum creatinine (SCr) and the population mean method using the mean population parameter of TEIC_conc for Japan. We also compared the predicted TEIC_conc to measured TEIC_conc. Creatinine clearance (CLCr) predicted using the Cockcroft-Gault (C&G) equation with SCr was 45.23 mL/min (interquartile range [IQR]: 32.12-58.28), and the glomerular filtration rate (GFR) predicted using the Hoek equation with CysC was 45.23 mL/min (IQR: 35.40-53.79). The root mean-squared prediction error (IQR) based on CLCr predicted using the C&G equation with SCr was 6.88 (3.80-9.96) μg/mL, and that based on GFR predicted using the Hoek equation with CysC was 6.72 (3.77-9.68) μg/mL. Predicted TEIC_conc did not differ significantly between the two methods. The predictive accuracy of the TEIC_conc using the Hoek equation with CysC was similar to that of CLCr using the C&G equation with SCr. These findings suggest that the predictive accuracy of the TEIC_conc using CLCr based on the G&G equation and SCr might be sufficient for the initial dose adjustment of TEIC. Given that we were unable to confirm that CysC is the optimal method for predicting TEIC_conc, the expensive measurement of CysC might not be necessary.