Statistical problems in interpreting comparative bioavailability trials.

The basic purpose of a comparative bioavailability is to determine whether or not a new formulation of a drug is equivalent in its in vivo behavior to some standard formulation. Statistical analysis of the blood or urine concentrations of the drug plays a critical role in this determination, and two major problems are in deciding what to analyse and what statistical methods to use. It is proposed that the analysis usually employed, namely, hypothesis-testing, is largely irrelevant to the problem and should be replaced by an approach based on confidence intervals. Secondly, it is suggested that only those characteristics of the bood or urine-level sequence that have a meaningful relationship to the therapeutic use of the drug, should be analysed.