|
Activity Number:
|
231
- SPEED: SPAAC SESSION I
|
|
Type:
|
Topic-Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2021 : 10:00 AM to 11:50 AM
|
|
Sponsor:
|
Biopharmaceutical Section
|
|
Abstract #318189
|
|
|
Title:
|
The Value of a Two-Armed Bayesian Response Adaptive Randomization Trial Comparing Early Preterm Birth Rates in High Versus Low Dose of DHA: Reassuring Results
|
|
Author(s):
|
Byron Gajewski* and Susan E Carlson and Alexandra R Brown and Dinesh Pal Mudaranthakam and Elizabeth H Kerling and Christina J Valentine
|
|
Companies:
|
University of Kansas Medical Center and Department of Dietetics and Nutrition and Department of Biostatistics & Data Science and Department of Biostatistics & Data Science and Department of Dietetics and Nutrition and Department of Obstetrics and Gynecology
|
|
Keywords:
|
Group sequential designs;
beta-binomial;
mixture normal;
drift;
prediction
|
|
Abstract:
|
Adaptive trials have prespecified decision rules in order to be flexible but valid. Researchers follow these decision rules to gain trust in their conduction. This talk describes the value of a two-armed Bayesian adaptive design using response adaptive randomization (RAR) to investigate early preterm birth rates of high versus low dose of DHA during pregnancy. The trial was designed to have power over 90% and type I error of 5%. The pandemic toward the last part of enrollment forced the team to halt enrollment at 1100 subjects rather than the original 1355. The primary analysis resulted in a posterior probability of efficacy signal to be .81. Bayesian posterior predictive distributions showed that had we finished the trial a 0.01 probability of reaching the success criteria (>.99). It would have been unlikely to reach that success criteria, and given the burden of the pandemic, it was a good idea to stop. The Bayesian approach is valuable - inference despite an unplanned ending to the trial. We also found that the trial did not have a drift effect that can harm RAR trials. We conclude that the RAR value in two-armed studies is quite high and believe that their use in the future.
|
Authors who are presenting talks have a * after their name.
