Optimal designs for clinical trials with dichotomous responses

Area of Science:

• Clinical trial design
• Biostatistics
• Medical research methodology

Background:

• Traditional randomized controlled trials (RCTs) focus on comparing treatment differences.
• There is a need for clinical trial designs that prioritize maximizing patient outcomes over solely gathering comparative data.

Purpose of the Study:

• To develop and analyze a two-stage clinical trial design aimed at maximizing the total number of successes.
• To optimize treatment allocation in the initial stage for enhanced efficacy in the subsequent stage.

Main Methods:

• The proposed design involves an information-gathering first stage and a treatment-selection second stage.
• Treatment allocation in the first stage balances information gain with effective treatment delivery.
• The length of the first stage can be fixed or optimized based on prior information and patient horizon.

Main Results:

• The design aims to maximize the expected number of successes across both stages of the trial.
• The optimal size of the first stage can be determined as a function of prior information and patient horizon.
• In specific scenarios with one known and one unknown success probability, the optimal first stage size is proportional to the square root of the patient horizon.

Conclusions:

• This two-stage design offers an alternative to traditional RCTs by focusing on maximizing overall patient success.
• The flexibility in determining the first stage length and optimizing treatment allocation enhances the design's adaptability.
• The findings provide a framework for efficient clinical trial design, particularly when patient outcomes are the primary objective.