Related Experiment Videos
Application of Bayesian statistics to decision making during a clinical trial
L S Freedman1, D J Spiegelhalter
1Division of Cancer Prevention and Control, National Cancer Institute, Bethesda, MD 20892.
Statistics in Medicine
|January 15, 1992
Summary
Bayesian methods offer a robust approach for monitoring advanced colorectal carcinoma trials. This analysis highlights how prior distributions influence stopping rules, impacting trial efficiency and outcomes.
Area of Science:
- Clinical Trials
- Biostatistics
- Oncology
Background:
- Advanced colorectal carcinoma treatment trials require rigorous monitoring and analysis.
- Traditional statistical methods may not fully capture evolving data during a trial.
- Bayesian approaches offer a flexible framework for adaptive trial designs.
Purpose of the Study:
- To apply Bayesian methods for monitoring and analyzing a treatment trial in advanced colorectal carcinoma.
- To investigate the impact of prior distributions on trial stopping rules.
- To compare Bayesian stopping rules with traditional group sequential boundaries.
Main Methods:
- Utilized Bayesian statistical methods for trial monitoring.
- Employed a truncated normal prior distribution with a probability mass at zero difference.
- Developed a stopping rule based on posterior distributions and equivalence ranges.
Main Results:
- The Bayesian stopping rule produced an upper boundary similar to the Pocock group sequential boundary.
- The stopping rule's sensitivity to the prior distribution's mass at zero was demonstrated.
- This indicates Bayesian methods can provide effective trial monitoring.
Conclusions:
- Bayesian methods are applicable and effective for monitoring advanced colorectal carcinoma trials.
- The choice of prior distribution significantly influences the Bayesian stopping rule.
- This approach offers a valuable alternative for adaptive clinical trial design and analysis.