Evaluation of three methods for determining initial vancomycin doses.

Dosing methods proposed by Matzke et al., Moellering et al., and Lake and Peterson were used to predict initial doses of vancomycin for serum concentration simulation using a two-compartment open model. Previously reported pharmacokinetic data from 25 of 28 patients were used to simulate steady-state serum vancomycin concentrations for doses derived from the three methods. The Matzke method failed to provide simulated one-hour postinfusion levels less than 30 micrograms/mL in 48 percent and troughs greater than 5 micrograms/mL in more than 88 percent of the patients. The Moellering method succeeded in achieving this goal in 96 percent of the simulated one-hour levels, and in 72 percent of the troughs when a six-hour dosing interval was selected. For the 8 mg/kg Lake-Peterson doses, one-hour levels greater than 30 micrograms/mL occurred in 28 percent of the simulations, and for the 10 mg/kg doses this increased to 40 percent. The 8 mg/kg doses resulted in trough simulations less than 5 micrograms/mL in 28 percent and the 10 mg/kg reduced this to only 20 percent. These simulations indicated that the Moellering nomogram with the six-hour dosing interval was the most successful method for initial dose selection, but early serum concentration monitoring and adjustment of initial empirical and nomogram-derived doses is necessary to assure safe and effective vancomycin serum concentrations.