Abstract:
|
It is common to run a clinical trial to compare a new generation device to an approved device and there are abundant historical data available for the approved device. For this purpose, a Bayesian power prior model is used to incorporate historical data on the approved device as a control, this would potentially decrease the sample size needed in the control arm in the new clinical trial. In this case study, the application of such a model to a randomized, controlled, non-inferiority trial in a regulatory setting with the use of a more conservative Bayesian 'conditional borrowing' strategy with multiple cutoffs, an extension from a model with one cutoff is proposed. This approach restricts the scenarios under which the historical data could be borrowed that would bias in favor of the new device. The statistical challenges in this trial design will be presented.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.