Activity Number:
|
30
- Statistical Considerations in Adaptive Designs
|
Type:
|
Contributed
|
Date/Time:
|
Sunday, August 7, 2022 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Biopharmaceutical Section
|
Abstract #323535
|
|
Title:
|
Bayesian Empirical Balancing Calibration for Addressing Nonconcurrent Controls in Adaptive Platform Trials
|
Author(s):
|
Evan Kwiatkowski* and Ying Yuan and Ruitao Lin
|
Companies:
|
Rice University and the University of Texas MD Anderson Cancer Center and MD Anderson
|
Keywords:
|
empirical likelihood;
hybrid designs;
external controls;
real-world data
|
Abstract:
|
A fundamental issue in analyzing platform trials is the degree to which the distribution of covariates is similar between current and nonconcurrent subjects. To ensure that the information from nonconcurrent subjects can be consistently used, we propose a Bayesian empirical balancing calibration approach to balance the covariate distributions between concurrent and nonconcurrent subjects. Built upon the empirical likelihood approach, we directly match the covariates between concurrent and nonconcurrent controls to any number of moments. A prominent feature of the proposed design is that it does not require specification of either a propensity score or an outcome regression model, which greatly enhances the robustness of the statistical inference and avoids the issue of model misspecification. We show results from simulation studies under fixed-sample or group-sequential platform trials with different assumptions on the consistency of concurrent and noncurrent subjects. These simulation studies show the advantage of the proposed method in accounting for model misspecification arising from the incompatibility between the concurrent and nonconcurrent subjects.
|
Authors who are presenting talks have a * after their name.