How do you design randomised trials for smaller populations? A framework.

Area of Science:

• Clinical Trial Design
• Biostatistics
• Medical Research Methodology

Background:

• Randomized phase III trials are crucial for evaluating new treatments but often require large sample sizes.
• Recruitment challenges can prevent achieving ideal sample sizes within practical timeframes.
• Existing frequentist approaches need adaptation for feasibility in resource-limited settings.

Purpose of the Study:

• To present an ordered framework for designing feasible randomized trials when ideal sample sizes cannot be met.
• To provide practical strategies for adapting trial parameters to increase achievable participant numbers and reduce sample size targets.
• To ensure high-quality evidence generation for treatment evaluation despite recruitment constraints.

Main Methods:

• Proposes a three-step framework for trial design adaptation.
• Step 1: Maximize recruitment potential (extend collaborations, broaden eligibility, increase time).
• Step 2: Optimize design parameters (research arm, outcomes, power, effect size).
• Step 3: Implement advanced statistical strategies (one-sided tests, adjusted alpha, covariates, re-randomization, external data) if feasibility remains an issue.
• Illustrates application with a worked example from the Euramos-1 trial.

Main Results:

• The framework offers a systematic approach to modify trial designs for feasibility.
• Specific alterations can increase achievable participant numbers and potentially reduce required sample sizes.
• The proposed methods, when applied judiciously, can still yield a good evidence base for treatment evaluation.

Conclusions:

• This framework enables the appropriate evaluation of treatments when large-scale phase III trials are not feasible.
• It addresses the pressing need for high-quality randomized data, even for less common diseases or challenging recruitment scenarios.
• The approach balances feasibility with the generation of robust evidence for clinical decision-making.