Related Experiment Videos
A Bayesian decision approach for sample size determination in phase II trials.
1Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, New York, New York 10021, USA. leung@biost.mskcc.org
Biometrics
|March 17, 2001
Summary
This study improves phase II clinical trial design by maximizing the rate of gain, not just expected financial gain. The new Bayesian design significantly reduces the chance of advancing ineffective treatments to phase III trials.
Area of Science:
- Clinical trial design
- Bayesian decision theory
- Pharmaceutical development
Background:
- Stallard's Bayesian approach maximizes expected financial gains in phase II trials.
- This design has a high probability (0.65) of advancing ineffective treatments to phase III.
- Alternative designs controlling false positive errors yield lower expected gains.
Discussion:
- Stallard's design optimizes per-trial gain but not long-term or rate of gain.
- Maximizing the rate of gain is crucial for efficient drug development.
- The proportion of treatments advanced to phase III impacts overall development efficiency.
Key Insights:
- A novel one-stage Bayesian design is proposed to maximize the rate of gain.
- The new design is twice as efficient as Stallard's one-stage design.
- The proposed design reduces the probability of advancing ineffective treatments to 0.12.
Outlook:
- This optimized design enhances the efficiency of clinical development pipelines.
- Improved phase II trial strategies can accelerate the delivery of effective treatments.
- Further research can explore adaptive designs for maximizing gain rates.