The use of glim and the bootstrap in assessing a clinical trial of two drugs

摘要

AbstractAn approach is described for estimating the dose of a new drug which is equipotent to an established dose of an old drug. The approach is basically that of the parallel‐line assay but it can allow for concomitant variables and, by exploiting the facilities available in the statistical computer package GLIM (generalized linear interactive modelling), the approach can be applied when the residuals conform to one of a number of distributions and, with suitable safeguards, to continuous, discrete and even ‘scored’ responses. In some circumstances, it is necessary to obtain confidence limits by Efron's ‘bootstrap’ technique. The method is illustrated with results from a trial of two premedicant drugs in