Bayesian methods to compare dose levels with placebo in a small n, sequential, multiple assignment, randomized trial.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacology

Background:

• Clinical trials for rare diseases face challenges due to small patient populations.
• Sequential, multiple assignment, randomized trial designs offer a potential solution.

Purpose of the Study:

• To propose and evaluate a new snSMART design for rare disease drug trials.
• To compare a Bayesian approach with standard methods for analyzing snSMART data.

Main Methods:

• A novel snSMART design involving two stages of randomization and treatment assignment.
• A Bayesian statistical approach for analyzing data from both stages, borrowing information between stages.
• Comparison with standard methods using only stage 1 data and a generalized estimating equation-based log-linear Poisson model.

Main Results:

• The proposed Bayesian method demonstrated smaller root-mean-square error and narrower 95% credible interval widths compared to standard methods.
• Utilizing data from both stages of the snSMART design is advantageous for primary efficacy analysis.

Conclusions:

• The novel snSMART design and Bayesian analysis are effective for rare disease drug development.
• This approach can support drug registration for rare diseases by providing robust efficacy data.