|
Activity Number:
|
579
- 30 Years Journey of Bayesian Adaptive Designs: From Bayesian Monitoring to Dynamic Treatment Regime — in Honor of Dr. Peter Thall’s 70th Birthday
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Thursday, August 6, 2020 : 3:00 PM to 4:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
|
Abstract #311115
|
|
|
Title:
|
Robust Adaptive Incorporation of Historical Control Data in a Randomized Trial of External Cooling to Treat Septic Shock
|
|
Author(s):
|
Thomas Murray* and Peter F. Thall and Frederique ARTHUR Schortgen and Sarah ARTHUR Zohar and Sandrine Katsahian
|
|
Companies:
|
University of Minnesota and M.D. Anderson Cancer Center and Groupe Hospitalier Henri-Mondor and Centre de Recherche des Cordeliers and Hopital Europen Georges-Pompidou
|
|
Keywords:
|
Commensurate prior;
Evidence synthesis;
Non-proportional hazards;
Gaussian Markov Random Field;
Intensive care unit;
Restricted mean survival
|
|
Abstract:
|
I will discuss a randomized controlled trial design to evaluate external cooling for controlling fever in patients with septic shock. The design includes a concurrent external cooling and control arms, and adaptively incorporates historical control arm data. Bayesian group sequential monitoring is carried out using a posterior comparative testing criterion based on the 60-day survival distribution in each concurrent arm. Posterior inference follows from a Bayesian discrete time survival model that facilitates adaptive incorporation of the historical control data through a regression framework with a multivariate spike-and-slab prior distribution on historical bias parameters. The amount of information borrowed from the historical control data will be determined adaptively in a manner that reflects the degree of agreement between historical and concurrent control arm data. Simulation results will be presented that elucidate how the proposed method borrows strength from the historical control data are reported. With little historical control arm bias, the proposed design controls the type I error rate and provides more power than a group sequential log-rank test.
|
Authors who are presenting talks have a * after their name.
