For certain drugs and dosage forms (see example (e), p. 72) plasma concentration time-profile data are not suitable for use in assessing equivalence between two formulations. While pharmacodynamic studies can sometimes be an appropriate tool for establishing equivalence (see section 11), in other instances this type of study cannot be performed because of a lack of meaningful and measurable pharmacodynamic parameters; a comparative clinical trial must then be performed in order to demonstrate equivalence between two formulations. In such a clinical trial, the same statistical principles will apply as in bioequivalence studies. The number of patients to be included in the study will depend on the variability of the target parameters and the acceptance range, and is usually much higher than that required in bioequivalence studies.
The methodology to be used in establishing equivalence between pharmaceutical products by means of a clinical trial in patients in which there is a therapeutic end-point has not yet been discussed as extensively as that used in bioequivalence trials. However, the following are important and need to be defined in the protocol:
(a) The target parameters; these are usually relevant clinical end-points from which the intensity and the onset, if applicable and relevant, of the response can be derived.
(b) The size of the acceptance range; this must be defined on a case-by-case basis, taking into consideration the specific clinical conditions, for example the natural course of the disease, the efficacy of available treatments and the chosen target parameter. In contrast to bioequivalence studies (where a conventional acceptance range is used), the size of the acceptance range in clinical trials cannot be based on a general consensus on all the therapeutic classes and indications.
(c) The statistical method used; this is currently the confidence interval approach, the main concern being to rule out the possibility that the test product is inferior to the reference pharmaceutical product by more than the specified amount. A one-sided confidence interval (for efficacy and/or safety) may therefore be appropriate. The confidence intervals can be derived by either parametric or non-parametric methods.
Where appropriate, a placebo leg should be included in the design, and it is sometimes appropriate to include safety end-points in the final comparative assessments.