Sample Size Calculation for the ROCI Design

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacometrics

Background:

• Response Over Continuous Intervention (ROCI) designs offer advantages for optimizing continuous treatment interventions.
• ROCI designs are applicable to de-escalation trials aiming to reduce treatment intensity.
• Current sample size calculations for ROCI designs rely on computationally intensive simulations.

Purpose of the Study:

• To propose a normal approximation approach for sample size calculation in ROCI designs.
• To evaluate methods for estimating nuisance parameters crucial for sample size determination.
• To reduce the computational burden associated with sample size estimation in ROCI trials.

Main Methods:

• Developed a normal approximation method for sample size calculation.
• Assessed various techniques for estimating nuisance parameters, considering reliability and computational cost.
• Utilized fractional polynomial regression to model intervention-response curves within the ROCI framework.

Main Results:

• The normal approximation approach is well-founded for ROCI designs.
• This methodology significantly reduces computational demands compared to simulation-based methods.
• Reliable estimation of distribution parameters necessitates methods addressing model selection uncertainty.

Conclusions:

• Normal approximation offers an efficient and valid method for sample size calculation in ROCI trials.
• The proposed framework streamlines sample size estimation, making ROCI designs more accessible.
• The approach is adaptable to ROCI designs employing diverse flexible modeling strategies.