Power analysis for cluster randomized trials with continuous coprimary endpoints

Area of Science:

• Biostatistics
• Clinical Trials
• Health Services Research

Background:

• Pragmatic trials frequently use cluster randomization (CRTs) for logistical or scientific reasons.
• Coprimary endpoints are common in CRTs but often overlooked in sample size and power calculations.
• Existing power analysis methods for CRTs primarily address binary endpoints, leaving a gap for continuous outcomes.

Purpose of the Study:

• To develop methods for sample size and power calculations for CRTs with multiple continuous coprimary endpoints.
• To provide a statistical framework for handling complex correlation structures within clusters.
• To extend existing methods to accommodate unequal cluster sizes.

Main Methods:

• Derivation of the closed-form joint distribution for K treatment effect estimators using a multivariate linear mixed model (MLMM).
• Characterization of the relationship between statistical power and various intraclass correlation coefficients.
• Approximation of the joint distribution under unequal cluster sizes using cluster size mean and coefficient of variation.

Main Results:

• The proposed method accurately predicts statistical power for CRTs with continuous coprimary endpoints.
• The derived methods account for multiple intraclass correlation types within the MLMM framework.
• Simulations demonstrate good agreement between predicted and empirical power, even with estimated MLMM parameters.

Conclusions:

• The developed methods provide a robust approach for sample size and power determination in CRTs with continuous coprimary endpoints.
• This work addresses a critical need in pragmatic trial design, enhancing the statistical rigor of health care intervention studies.
• The approach is applicable to real-world CRTs and facilitates more reliable evaluation of interventions.