Incorporating pragmatic features into power analysis for cluster randomized trials with a count outcome

Area of Science:

• Biostatistics
• Clinical Trials
• Health Services Research

Background:

• Cluster randomized trials (CRTs) are vital for pragmatic trials, aiming for broad applicability.
• Existing power analyses often simplify real-world clinical settings, potentially affecting generalizability.
• Incorporating pragmatic features into sample size calculations for CRTs is crucial.

Purpose of the Study:

• To develop a power analysis method for CRTs with count outcomes that directly integrates pragmatic features.
• To provide a closed-form sample size formula for enhanced implementation in pragmatic trials.
• To assess the impact of pragmatic features on sample size requirements.

Main Methods:

• The proposed method utilizes generalized estimating equations (GEE) for power analysis.
• It accounts for arbitrary randomization ratios, overdispersion, variable cluster sizes, and unequal follow-up durations.
• An efficient Jackknife algorithm is presented to address small cluster number variance estimation issues.

Main Results:

• The sample size formula derived from GEE is in a closed form, simplifying application.
• Theoretical exploration quantifies the influence of various pragmatic features on sample size.
• Simulation studies and a real clinical trial application validate the proposed method's performance.

Conclusions:

• The developed method offers a practical approach to sample size determination for CRTs in pragmatic settings.
• Directly incorporating pragmatic features improves the accuracy and relevance of power calculations.
• This facilitates more robust and generalizable findings from pragmatic clinical trials.