Avoiding Under- and Overrecruitment in Behavioral Intervention Trials Using Bayesian Sequential Designs: Tutorial

Area of Science:

• Clinical Trials
• Biostatistics
• Health Intervention Research

Background:

• Fixed sample sizes in intervention trials risk under- or overrecruitment, leading to research waste and participant harm.
• A priori determination of effect sizes is a limitation of traditional trial designs.

Purpose of the Study:

• To demonstrate the benefits of Bayesian sequential designs in reducing research waste and participant exposure.
• To illustrate how a Bayesian sequential design could have significantly shortened a digital alcohol intervention trial.

Main Methods:

• Utilizing a Bayesian sequential design for continuous evidence evaluation during a trial.
• Defining target criteria to determine when a scientific question is sufficiently addressed.
• Revisiting a digital alcohol intervention trial data to simulate Bayesian sequential analysis.

Main Results:

• A digital alcohol intervention trial with a fixed sample size of 2129 participants could have concluded with approximately 300 participants using a Bayesian sequential design.
• This approach would have exposed fewer participants to trial procedures, including control conditions.
• Earlier availability of trial evidence is a key advantage.

Conclusions:

• Bayesian sequential designs offer a more efficient and ethical alternative to fixed sample size trials.
• Implementing Bayesian sequential designs can minimize research waste and protect participants from unnecessary harm.
• Earlier trial completion and evidence dissemination are achievable with adaptive designs.