KEYNOTE 3: Optimising Group Sequential and Adaptive Designs: Where Frequentist meets Bayes
SPEAKERS: Prof. Christopher Jennison and Bruce Turnbull
MODERATOR:Ivan F Chan
Abstract:
In designing a group sequential clinical trial to test the treatment effect on a primary endpoint, one may search for an optimal design subject to type I error and power requirements. We shall illustrate how the solution to this frequentist problem involves solving a Bayes sequential decision problem. Conversely, if one starts with a Bayesian formulation but calibrates the design to guarantee a type I error condition, the result should match an optimized frequentist procedure. Underlying this equivalence are theorems which prove that the class of admissible frequentist designs is the same as the class of Bayes optimal designs. From a practical perspective, this relationship is helpful in computing optimal designs. At a philosophical level, we see that, when implemented well, the two approaches to inference should produce the same results.
A more complex trial can have multiple treatments, multiple endpoints, or subgroups of patients, so multiple hypotheses may be tested. Examples include seamless Phase 2-3 trials with treatment selection and adaptive enrichment trials. In such cases, control of the family-wise error rate is usually required. Although the relation between optimal frequentist and Bayesian designs is not so straightforward here, there are design strategies that combine the two approaches. One may start with a Bayesian design and calibrate features of this to achieve a frequentist error rate. Alternatively, one can define a class of frequentist designs that control the family-wise error rate and use Bayesian methods to optimize within this class.
Instructors’ Biography:
Christopher Jennison is Professor of Statistics at the University of Bath, UK. He was awarded his PhD from Cornell University for research into the sequential analysis of clinical trials and has continued to work in this area for the past 25 years. He has published extensively on group sequential methods and adaptive designs. His book with Professor Bruce Turnbull, “Group Sequential Methods with Applications to Clinical Trials”, is a standard text on this topic and is widely used by practising statisticians. Professor Jennison’s research is informed by experience of clinical trial analysis at the Dana Farber Cancer Institute, Boston and a broad range of consultancy with Medical Research institutes and Pharmaceutical companies in Europe, America and Asia. He has made numerous presentations at international conferences, in which he sets out to describe novel statistical methodology and its application to the design and analysis of clinical trials.
Bruce Turnbull received the B.S. from Cambridge University in 1967 and the Ph.D. from Cornell University in 1971. After serving on the faculty at Stanford University and at the University of Oxford, he joined Cornell University in 1976, where he is currently Professor in the School of Operations Research and Information Engineering where he previously served as Acting Director.
From 2000–2002, he was founding Chairman of the newly formed Department of Statistical Science and currently also holds a professorial appointment there. In 1979 he was awarded the Snedecor Memorial Award by the American Statistical Association in recognition of his research.
He has authored over 130 publications and is the co-author of a book on statistical procedures for monitoring clinical trials. He has been a consultant to many organizations, including the Oak Ridge National Laboratory, Institute for Energy Analysis; and various pharmaceutical companies. Turnbull has served on the Board of Directors of the National Institute of Statistical Sciences, and on the Expert Review Panel for the National Toxicology Program Board of Scientific Counselors. He is on the Data and Safety Monitoring Committees for several major national and international clinical trials in the areas of cancer, heart disease, pulmonary disease and of AIDS sponsored by the National Institutes of Health and by the Veterans Administration. He has served on the editorial board of a number of statistical journals and is currently editor of the Chapman and Hall book series on biostatistics. He is a Fellow of the American Statistical Association and of the Royal Statistical Society.
Professor Turnbull also has emeritus status in the School of Operations Research and Information Engineering in Cornell’s College of Engineering.