Use of limited laboratory-acquired therapeutic drug monitoring data in determining individualized drug requirements.

Routinely acquired therapeutic drug monitoring (TDM) data from 220 patients were used to estimate patient-specific pharmacokinetic parameters for the following drugs: aminoglycoside antibiotics (gentamicin and tobramycin), digoxin, theophylline, carbamazepine, procainamide, phenobarbital, and quinidine. A microcomputer based set of algorithms operating on two relatively unconstrained TDM values estimated pharmacokinetic parameters with which future TDM levels were forecast. Mean prediction errors (mpe) and root mean squared errors (rmse) were used as measures of predictive performance. Values of mpe deviated from zero by less than one microgram per l for digoxin and by less than one microgram per l for all other drugs. Values of rmse were also small when viewed in the context of the respective therapeutic plasma concentration ranges.