Model-Based Assessment of Alternative Study Designs in Pediatric Trials. Part II: Bayesian Approaches

Area of Science:

• Clinical Pharmacology and Biostatistics
• Pediatric Clinical Trial Design

Background:

• Evaluating fixed-sample Bayesian designs (BD) and Bayesian sequential designs (BSDs) for pediatric trials.
• Utilizing prior information from adult trials, weighted by population similarity.
• Assessing performance metrics including type I/II errors, sample size (SS), trial duration (TD), and estimate precision.

Purpose of the Study:

• To present a pharmacokinetic-pharmacodynamic (PK/PD) simulation framework for comparing pediatric trial designs.
• To evaluate the performance of a fixed-sample BD against two BSDs (non-hierarchical and semi-hierarchical).
• To determine the optimal Bayesian design strategy for pediatric clinical trials.

Main Methods:

• Development of a PK/PD-based clinical trial simulation framework.
• Elicitation and weighting of prior information from adult pharmacokinetic and pharmacodynamic data.
• Simulation of fixed-sample Bayesian and two Bayesian sequential designs (NON-H, SEMI-H).

Main Results:

• No significant performance differences were found between the non-hierarchical (NON-H) and semi-hierarchical (SEMI-H) Bayesian sequential designs.
• Bayesian sequential designs (BSDs) generally required smaller sample sizes (SS) and shorter trial durations (TD) compared to the fixed-sample Bayesian design (BD).
• The fixed-sample Bayesian design (BD) offered higher estimate precision but BSDs could necessitate large sample sizes when significant pediatric-adult differences were anticipated.

Conclusions:

• Bayesian sequential designs (BSDs) are more efficient for pediatric clinical trials, offering reduced sample size and duration.
• Bayesian approaches demonstrate superiority over frequentist methods in pediatric trial design, even with minimal adult prior data.
• The choice between fixed-sample and sequential Bayesian designs depends on the trade-off between estimate precision and adaptive efficiency.