Related Experiment Videos
Exact group-sequential designs for clinical trials with randomized play-the-winner allocation.
Nigel Stallard1, William F Rosenberger
1Medical and Pharmaceutical Statistics Research Unit, The University of Reading, P.O. Box 240, Earley Gate, Reading RG6 6FN, UK.
Statistics in Medicine
|February 12, 2002
Summary
Sequential and adaptive designs reduce inferior treatments in clinical trials. This study uses exact distributions for small samples, finding adaptive allocation may need larger sample sizes and potentially allocate more patients to inferior treatments.
Area of Science:
- Biostatistics
- Clinical Trial Design
- Medical Research Methodology
Background:
- Sequential designs and adaptive treatment allocation minimize patient exposure to inferior treatments in clinical trials.
- Standard sequential designs combined with adaptive allocation are effective in large samples where test statistics are asymptotically normal.
- Small sample sizes pose challenges for adaptive allocation due to potential violations of planned error rate constraints when assuming normality.
Purpose of the Study:
- To develop sequential stopping rules with precise properties for adaptive allocation in small samples.
- To address the limitations of assuming normality in small sample adaptive clinical trials.
- To evaluate the performance of adaptive allocation using exact distributions under the randomized play-the-winner rule.
Main Methods:
- Constructing sequential stopping rules based on the exact distribution of test statistics.
- Utilizing the randomized play-the-winner (RPW) rule for adaptive treatment allocation.
- Comparing adaptive allocation designs with traditional equal allocation trials.
Main Results:
- Adaptive allocation designs, when analyzed using exact distributions, may require larger total sample sizes to achieve desired statistical power compared to traditional equal allocation.
- The expected number of patients allocated to the inferior treatment in adaptive designs can be higher than anticipated, depending on the actual success rates of the treatments.
- The study highlights the importance of considering exact distributions for accurate error rate control in small sample adaptive clinical trials.
Conclusions:
- Sequential designs and adaptive allocation are valuable tools in clinical trials, but their application, especially in small samples, requires careful consideration of statistical properties.
- The randomized play-the-winner rule, when analyzed with exact distributions, provides a robust method for adaptive allocation in small samples.
- Findings underscore the need for precise sample size and allocation planning in adaptive clinical trials to ensure both efficacy and patient safety.