ABSTRACT

Faculty of Industrial Engineering and Management, Technion – Israel Institute of Technology and Department of Statistics, University of Pennsylvania

9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 9.2 Inference Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203

9.2.1 Minimizing Covariance Matrices . . . . . . . . . . . . . . . . . . . . . . . . 203 9.2.2 Maximizing Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

9.3 Multi-Objective Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 9.4 Implementation-Binary Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209

9.4.1 Two Treatments and One Control . . . . . . . . . . . . . . . . . . . . . . 209 9.4.2 Active Control Equivalence Trials . . . . . . . . . . . . . . . . . . . . . . . 211 9.4.3 The Tradeoff between Inference and Treatment . . . . . . . . 214

9.5 Implementation-Normal Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 9.6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218

Clinical trials have typically several goals, which can be divided into two types. The first type concerns the future patients that could potentially benefit from the treatments under consideration, and the second pertains to the health of the patients in the trial. For the former, it is important to correctly identify the best treatment, while the latter emphasizes the ethical requirements of the trial. This conflict is called “individual versus collective ethics” in Baldi Antognini and Giovagnoli [4], and “treatment versus experimentation dilemma” in Bartroff and Lai [5] and Azriel, Mandel and Rinott [1].