Optimization of pharmacokinetic monitoring: I. Linear pharmacokinetics

To reach and maintain therapeutic drug concentrations, reliable estimates of the parameters describing the pharmacokinetic behavior of the drug are required. The accuracy and precision of the pharmacokinetic parameters are dependent upon the error associated with the time of specimen collection and the laboratory assay error. In this report, the determinate error and random error are analyzed independently and incorporated in calculations of regimens of two drugs, gentamicin and theophylline. The calculations show that, within a particular concentration range, there are better times for sampling to determine the elimination rate constant and, consequently, the dose. From the generalized, first-order equation for a one-compartment pharmacokinetic model, ln Ct = ln C0 - Kt, where C = plasma drug concentration, t = time, C0 = C at zero time, and the elimination rate constant, K = 0.693/t1/2, the determinate error, epsilon K, in K is epsilon = (1/C0t)epsilon C0 - (In C0/C)t-2 epsilon t, and the random error, assuming a gaussian distribution, is (formula; see text) where S2x is variance of parameter x. Therefore, the variances of the calculated quantity, C0, the measured quantity, C, and the time of sample collection, t, are initially of equal importance. However, with increasing t, the effect of S2t becomes small. In both random and determinate error cases, increased time of sampling after dose reduces the error in K, thus t1/2, as do recognized biological factors. However, increased relative analytical error may obliterate this advantage if analyses are not modified to obtain best quantitation at low, even subtherapeutic concentrations.(ABSTRACT TRUNCATED AT 250 WORDS)