Optimal adaptive promising zone designs

Area of Science:

• Biostatistics
• Clinical Trial Design

Background:

• Adaptive group sequential trials often have initial sample sizes insufficient for detecting the smallest clinically meaningful treatment effect.
• Securing upfront funding for the maximum possible sample size is challenging for trial sponsors.

Purpose of the Study:

• To develop optimal decision rules for sample size re-estimation in two-stage adaptive group sequential clinical trials.
• To address the challenge of adequately powering trials to detect clinically meaningful treatment effects while managing upfront resource commitments.

Main Methods:

• Investigated promising zone designs that optimize both unconditional and conditional power.
• Proposed a Bayesian approach for sample size re-estimation when direct reliance on treatment effect parameters is not preferred.
• Integrated out the unknown treatment effect using posterior distributions in the Bayesian option.

Main Results:

• Developed optimal decision rules for sample size re-estimation in adaptive trials.
• Demonstrated methods to enhance the probability of trial success through sample size adjustments within promising zones.
• Showcased a Bayesian alternative for sample size adjustments, accommodating reluctance to rely solely on specific treatment effect parameters.

Conclusions:

• Optimal decision rules can improve the efficiency and success probability of adaptive clinical trials.
• Promising zone designs offer a structured approach to sample size re-estimation, balancing statistical power and practical constraints.
• The proposed Bayesian method provides a flexible alternative for sample size adjustments in adaptive trials.