Standard clinical trial methodology provides excellent control of type I error, but is much less adequate for specification of the intended use population. Phase III clinical trials often have very broad eligibility criteria but accrued patients rarely represent a random or representative sample of the eligible population. Consequently use of the eligibility criteria as a basis for specification of the intended use population has little statistical basis. Many classically defined diseases have been found to be heterogeneous mixtures of molecularly distinct entities sharing symptomatology but not pathogenesis or responsiveness to treatment. Consequently, standard clinical trial practice often leads to small average treatment effects, large NNT values and substantial over-treatment of the patient population.

Adaptive enrichment designs enable adjustment of the eligibility criteria in a group sequential manner to focus the trial on the kinds of patients who are demonstrating benefit from the test treatment, thereby increasing statistical power and a improving the specification of the intended use population. This presentation will discuss the specification of the intended use population in adaptive enrichment designs.

Simon N, Simon R. Adaptive enrichment designs for clinical trials. Biostatistics, doi:10.1093/biostatistics/kxt010, published online March 21, 2013.