Group sequential multi-arm multi-stage trial design with treatment selection

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Statistical Inference

Background:

• Traditional randomized controlled trials (RCTs) are less efficient for comparing multiple treatments.
• Multi-arm multi-stage (MAMS) designs offer improved efficiency but face computational challenges.
• Existing MAMS methods often require significant computational resources for sample size and boundary determination.

Purpose of the Study:

• To develop a computationally efficient group sequential MAMS trial design.
• To provide analytical solutions for futility and efficacy boundaries in MAMS trials.
• To overcome the computational hurdles associated with current MAMS methodologies.

Main Methods:

• Development of a group sequential MAMS design utilizing the sequential conditional probability ratio test.
• Derivation of analytical solutions for futility and efficacy boundaries applicable to any number of stages and arms.
• Comparison with existing methods, specifically those implemented in the R package MAMS by Magirr et al.

Main Results:

• The proposed method provides direct analytical solutions, eliminating complex computations.
• Demonstrated advantages in computational efficiency compared to existing MAMS approaches.
• The design is applicable to an arbitrary number of stages and arms within the MAMS framework.

Conclusions:

• The developed group sequential MAMS design offers a computationally tractable alternative.
• This approach simplifies the implementation of MAMS trials, promoting wider adoption.
• The analytical solutions enhance the practicality and efficiency of adaptive clinical trial designs.