Optimal response-adaptive designs for continuous responses in phase III trials

Area of Science:

• Clinical Trials
• Biostatistics
• Experimental Design

Background:

• Optimal response-adaptive designs are increasingly important in Phase III clinical trials.
• Existing optimal designs often lack robust optimality considerations.
• Previous work includes optimal designs for binary (Rosenberger et al., 2001) and continuous responses (Biswas & Mandal, 2004).

Purpose of the Study:

• To evaluate the suitability of the Zhang and Rosenberger (2006) optimal design for normally distributed responses.
• To identify limitations of the Zhang and Rosenberger (2006) design and its applicability to other continuous distributions (e.g., exponential, gamma).
• To propose a unified framework for optimal response-adaptive designs in comparative clinical trials.

Main Methods:

• Critically analyze the Zhang and Rosenberger (2006) optimal design for normal responses.
• Develop and suggest adjustments for optimal designs across various continuous response distributions.
• Formulate a unified framework for optimal response-adaptive designs comparing two treatments.

Main Results:

• The Zhang and Rosenberger (2006) design is demonstrated to be unsuitable for general normal responses.
• The proposed adjustments and unified framework are effective for continuous response distributions.
• The methods are validated using real-world clinical trial data.

Conclusions:

• The Zhang and Rosenberger (2006) design requires modification for optimal use with normal responses.
• A unified approach provides a more robust and adaptable method for response-adaptive designs.
• The findings contribute to more efficient and statistically sound clinical trial designs.