We are grateful that Hauschke and Pigeot (2005) elaborate a little on our (Koch and Raehmel, 2004) considerations on what may constitute under which circumstances a valid proof of efficacy for a new experimental treatment as compared to a reference and placebo in a so-called "gold standard design". In essence, we have stated that a mandatory requirement should be that (i) the experimental treatment can be shown to be superior to placebo and (ii) the experimental treatment can be shown to be noninferior to the reference treatment.
Cautiously we have stated in addition that situations may exist, where it is clear at the out-set of the trial, that even proven superiority of the experimental treatment over placebo would not be convincing unless at the same time reference has shown superior efficacy as compared to placebo. We have heard that this superiority would provide evidence that the correct patient population has been selected for the current trial. The example of treating patients with diarrhoea with an anti-depressant and with an anti-diarhoeal (and placebo) in a three arm clinical trial, where everybody (except the placebo patients) is happy in the end, clarifies, that in such situations more than simple superiority of the reference treatment over placebo is needed. Others may want to comment on the difficulties in describing respective requirements appropriately which have lead to the definition of terms like "assay sensitivity" and "sensitivity to drug effects" (Committee for Proprietary Medicinal Products, 2000).
In support of our argument that at a minimum the above mentioned claims (i) and (ii) have to be substantiated we have identified situations, where, although a reference exists, placebo may be needed, in addition: main reasons were that reference is a traditional standard with not much or outdated scientific support (e.g. because the co-medication has changed completely for the disease under investigation), or a weak standard (in that it would be difficult to justify a non-inferiority margin in an active controlled two-arm trial). Should in such a situation a new experimental treatment that has shown to be superior to placebo and non-inferior (or even better than reference) be blamed for the fact that reference could not beat placebo?
Early applications investigated also in the three-arm setting only the superiority of the experimental treatment over placebo confirmatorily. Both, from an ethical, as well as from a regulatory point of view, we estimate the role of the second claim high: to our understanding it is unethical to randomize patients on a treatment group which will be later only used for hand-sight arguments about relative efficacy. In the process of drug licensing, where in the end a careful assessment of the merits of a new treatment is performed, it is wise to discuss right at the planning stage of a trial, what needs to be demonstrated with respect to the relative efficacy of test and reference. As clearly relevant losses in efficacy would be difficult to ignore, endless post-hoc discussions about what constitutes an irrelevant loss can only be avoided, if addressed beforehand.