Worth Adapting? Outcome?Adaptive Randomization ? Hype or Hope?
- 2011-11-22 (Tue.), 14:00 PM
- Recreation Hall, 2F, Institute of Statistical Science
- Prof. J. Jack Lee
- The University of Texas MD Anderson Cancer Center Houston, Texas
Abstract
Randomization eliminates the treatment assignment bias in clinical trials and can reduce the confounding effect of both known and unknown prognostic factors. Equal randomization (ER) has been widely applied in clinical trials with the focus on collective ethics to maximize statistical power. On the other hand, outcome adaptive randomization (AR) emphasized on individual ethics by assigning more patients to the more effective treatments based on the interim data to increase the overall success during the trial. Typically, AR has better overall success than ER, but increases the average sample size. AR can have substantial benefit over ER when the efficacy difference between treatments is large. In a more complex setting with three arms, AR can yield both higher overall response as well as lower average sample size. The differences between AR and ER quickly diminish with early stopping for efficacy and futility. The choice of AR or ER depends on the focus of the trial and the utility function of the clinical trialists. With more predictive markers being developed, AR has the advantage to steer patients to better treatments, hence, provides more benefit to patients enrolled in the trial. AR can be highly useful in the co?development of effective treatments and companion diagnostics for predictive markers. The use of AR will be illustrated through a recently completed lung cancer BATTLE trial.