While the "effect" of a drug can be observed or deduced from observational data, the
concept of "therapeutic efficacy" represents mainly a theoretical construction of
a high degree of abstraction which is inconceivable without reciprocal combination
with other theoretical constructs. The "therapeutic efficacy" of drugs can be investigated
only via clinical-pharmacological or clinical "models". Several examples are given
and discussed against the background of the actual considerations for shortening phase
III studies and extending pharmacoepidemiological phase IV studies for scientific,
practical and economic reasons. Of special relevance is the question whether study
data of phase III allow an extrapolation to the wider patient population which it
is intended to treat. Thus, it is well known that the criteria for representativeness
in the investigated population are rarely achieved in phase III studies. Furthermore,
observations have shown that various intervening moderator variables, such as the
investigated subgroup or the trial setting (e.g. inpatient or respectively outpatient
treatment), might influence therapeutic efficacy and the possibility of generalizing
the results. This again raises the crucial question of clinical relevance of significant
effects. Possible ways of overcoming this unsatisfactory situation are suggested.
