Abstract:
|
The first time we run a clinical trial should never be the actual time we run a clinical trial. Even for standard designs, we can simulate trials under a variety of safety, efficacy, accrual, and dropout scenarios to visualize how the trial will behave and what observed data will translate into success or futility. This exercise is extremely informative to collaborating clinicians, stakeholders, IRBs, and DMCs. By visualizing possible results and discussing with collaborators we can modify trial design features and optimize the trial before we ever execute it. With adaptive trials, the role of simulation is even more important to understand these innovative trial designs. In particular standard statistical methods are generally based on asymptotic theory and may not be ideally suited rare disease trials where sample sizes are small. Furthermore classical methods make no effort to optimize treatment of patients within the trial. Simulating a variety of potential designs allows us to explore how effective standard methods are and whether other innovative designs may offer advantages for patients and/or statistical power.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.