Bayesian Adaptive Study Design Supported by VICTR Biostatistics Core

The VICTR biostatistics core has a Bayesian adaptive study design component. Under this effort supported by VICTR, we have helped investigators design clinical trials with response adaptivity and flexible stopping rules. These stopping rules include stopping for efficacy, safety, or futility with sample size reestimate. These studies include ones with continuous, binary, and time-to-event endpoints. We can put these flexible designs into the context of the classical fixed-sample size design and the sequential design and help you identify an optimal study design so that you can answer your research question in a speedy way.

If you consider the following, Bayesian adaptive design with flexible stopping rules may help you
  • Early stopping for efficacy, safety, or futility
  • Possible early stopping for a subgroup which is not likely to respond to the therapy
  • Possible early stopping for one or more of the multiple dosing groups
  • Estimation-based studies with sample size reestimate

We have experience/expertise in designing the following types of studies using Bayesian adaptive flexible design:
  • Binary endpoint
    • In an R01 application to evaluate the efficacy of a treatment on blood product transfusion in patients undergoing cardiac surgery, we designed a Bayesian monitoring plan for the binary primary endpoint, blood product transfusion. In Aim 2 of the study, we use Bayesian monitoring of a continuous biomarker to evaluate a personalized weight-based dose of the treatment compared to a fixed-dose regimen. As soon as there is enough evidence on the relative merit of the dosing regimens, we can make informed decision to terminate one dosing regimen. EACA.pdf
    • In another study of evaluating the efficacy of a treatment compared to placebo on ICU death, the primary endpoint also serves as a key safety endpoint, Bayesian monitoring of the binary endpoint is ideally suited to facilitate both efficacy monitoring and safety monitoring. Envelope.pdf
    • A randomized, pacebo controlled, double blind noninferiority study of the TYRX antibacterial envelope alone versus envelope along with intraoperative antibacterial irrigant and postoperative oral antibiotics to prevent cardiac implantable electronic device infections in high-risk patients.
  • Continuous endpoint
    • We designed a Bayesian monitoring plan for a biomarker to identify a gender group that is not likely to respond to the treatment. Thus, informed decision can be made to stop enrollment in this gender group so that research resources can be deviated to the more promising group.
  • Time-to-event endpoint
    • We developed a Bayesian flexible design for a biologic agent in an ongoing Phase II trial in pancreatic cancer. The primary endpoint is recurrence-free survival. We want to have many looks at the data during the study and to have possible study extension when results obtained at the originally planned study termination are equivocal. Our analysis is based on a Bayesian Cox proportional hazards model. Evidence for efficacy is taken to be a posterior probability of efficacy >= 0.95 at any analysis time, where "efficacy" means a true hazard ratio < 1.0. The planned rule for extending the study is a probability of efficacy >= 0.7 at the last pre-planned analysis.

Contact: Chang Yu (chang.yu@Vanderbilt.Edu, Tel: 322-8422)
Team members: Chang Yu, Hui Nian, Li Wang, Frank E Harrell Jr, and Daniel W. Byrne

Topic attachments
I Attachment Action Size Date Who Comment
EACA_20140402.pdfpdf EACA_20140402.pdf manage 142.2 K 13 Jan 2015 - 16:32 HuiNian  
antibacterial_envelope_20141028.pdfpdf antibacterial_envelope_20141028.pdf manage 160.2 K 13 Jan 2015 - 16:49 HuiNian  
Topic revision: r11 - 02 Oct 2015, LiWang

This site is powered by FoswikiCopyright © 2013-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback