A comparison of methods for designing hybrid type 2 cluster-randomized trials with continuous effectiveness and...

Area of Science:

• Health Services Research
• Biostatistics
• Clinical Trial Design

Background:

• Hybrid type 2 studies are increasingly utilized for evaluating both implementation and health outcomes simultaneously.
• These studies frequently employ cluster-randomized trials (CRTs) design.
• Accurate power calculation is crucial for the validity of CRTs with co-primary endpoints.

Purpose of the Study:

• To compare the statistical power of five different design methods for hybrid type 2 cluster-randomized trials.
• To provide practical guidance for selecting the most appropriate power calculation method based on study characteristics.

Main Methods:

• Theoretical comparison of power equations for five methods: p-value adjustment, combined outcomes, single 1-DF test, disjunctive 2-DF test, and conjunctive test.
• Numerical simulations using a novel R package (crt2power) across 45,000 scenarios for CRTs with two continuous co-primary endpoints.
• Evaluation of power under varying conditions, including equal and unequal treatment effects.

Main Results:

• P-value adjustment methods were consistently less powerful than the combined outcomes approach and the single 1-DF test.
• The disjunctive 2-DF test showed less power than the single 1-DF test under specific conditions.
• The disjunctive 2-DF test demonstrated superior power when treatment effects were unequal, whereas the single 1-DF test was more powerful when treatment effects were equal.

Conclusions:

• The choice of power calculation method significantly impacts the efficiency of hybrid type 2 CRTs.
• The findings offer practical recommendations for researchers designing CRTs with co-primary endpoints, guiding the selection of the most powerful statistical approach.
• The study highlights the importance of considering the expected pattern of treatment effects when determining the optimal design method.