The planning of an experiment is a crucial point in research, and also in clinical trials development.
A researcher might have the right intuition, but he can only prove it through good experimental planning. Analogously, in clinical trials, a new effective drug can be experimentally proved to be useful through an adequate and well developed scientific protocol.
As shown in the Introduction, it is a fact that approximately 40% of phase III trials fail. This is due to many reasons, as explained in Sections I.2 and I.3, and in some cases to errors in the experimental plan. In particular, wrong assumptions on the effect size can be postulated, which imply a too small sample size and a consequent too low probability of success. Moreover, some phase III trials regarding effective and useful drugs are not launched because phase II data did not show significant results just thanks to bad chance.
Here, focus is placed on one point of the experimental planning of the phase III trial, and that is the computation of the sample size.
It is a common habit for the information collected during previous phases of the research to be used for planning the following phases (see Guidance for Industry, ICH-E9, 1998). In this context, phase II results and data are used for planning phase III trials.
First, phase III is run only when phase II results are, in some sense, “good enough”. Rationalizing this concept, some launch criteria for phase III based on phase II data ...