University of Pennsylvania ninth annual conference on statistical issues in clinical trials: Where are we with adaptive clinical trial designs? (morning panel discussion)

Valerie Durkalski: I really enjoyed Dr Bretz’s analogy of the Swiss Army knife. I am surrounded by a family of Boy Scouts, and they love their Swiss Army knives. The big Boy Scout motto is ‘‘Be prepared,’’ which I think is a perfect analogy for the discussion of adaptive designs. After several years of studying these designs, we all know that the key step, reinforced by Dr Bretz, is the importance of simulations. The dilemma that we still need to address is the infrastructure needed to support the conduct of these simulations. Where is the infrastructure to do these simulation studies? These do not just appear at ‘‘a push of the button.’’ It can take months, sometimes years to figure out the questions we want, or need, to ask. Then, we need to simulate the design(s) to really understand the operating characteristics. There are very few research settings, particularly in academics, where there is a free-standing think tank endowed with the financial coverage, computers, and knowledge to do the more complex simulations. There are basic simulations. For example, if we’re doing a group sequential design to stop early for efficacy and/or futility, we want to look at the probability of stopping early under various stopping rules/ boundaries and so forth, so maybe those are the more basic ones that can be done with minimal infrastructure support. However, when you start to mix response-adaptive randomization, early stopping rules, blinded sample size re-estimation into a design, the number of different scenarios that could be anticipated becomes huge. We need to consider how the infrastructure for trial planning will be supported and funded.