Problems and pitfalls in estimating average pharmacokinetic parameters.

The problems of obtaining optimal average parameter estimates (APE) from experimental pharmacokinetic data are considered. Four different approaches, three parametric and one non-parametric tests, are compared, using selected individual alcohol concentration data. Pooling the raw data for estimating APE can obscure individual pharmacokinetic characteristics, whereas averaging individual parameter estimates (IPE) exposes unique statistical problems. Furthermore, careful consideration should be given to weighting procedures. The advantages and shortcomings of all four methods are discussed. It is concluded that none can be considered as a universally applicable statistical method in view of the purpose for which the information, derived from a set of data, e.g. an alcohol-kinetic study, is required.