Validation of a quick modeling program generating clearance estimates at steady state for routine therapeutic drug...

Therapeutic drug monitoring (TDM) of chronic treatments is justified for several reasons, including relative over- or underdosage due to variable individual elimination, pharmacokinetic interactions in drug combinations, and noncompliance. In all these circumstances, the prescribing physician is interested in having an estimation of the patient's clearance of the drug, even from one measurement. We compare a validated bayesian program, USC*Pack of Jelliffe, found difficult to use in daily routine, with a "home-made" program. The latter, which is capable of taking data from a clinical database, will generate a graphic simulation of daily plasma drug concentrations together with an estimation of steady-state clearance more rapidly than does USC*Pack. Both programs were run with only one measured plasma level. The patients were 83 children or young adults treated with phenobarbital (PB), carbamazepine (CBZ), and/or Valproic acid (VPA) who were resistant to monotherapy and who were to be sampled two to four times between doses. Drugs were routinely assayed by high-performance liquid chromatography (HPLC). Despite the rough character of Phacile (numeric integration and adjustment of only two of three parameters, without an acknowledged minimization algorithm), the results are comparable to those obtained with USC*Pack for estimating clearance and predicting plasma drug concentrations. Phacile algorithm, although simple, has proven of interest in routine TDM and as an introduction for medical students to the bayesian approach of population pharmacokinetics.