Highly efficient stepped wedge designs for clusters of unequal size

Area of Science:

• Biostatistics
• Clinical Trials Methodology
• Epidemiology

Background:

• Stepped wedge designs (SWD) are cluster randomized trials comparing treatments across time periods and sequences.
• Existing literature often overlooks the impact of varying cluster sizes in SWD analysis.
• Efficient design is crucial for accurate treatment effect estimation in SWD.

Purpose of the Study:

• To develop methods for efficient stepped wedge designs considering unequal cluster sizes.
• To analyze the impact of cluster and individual proportions on treatment effect variance.
• To evaluate the properties of cluster-balanced designs in SWD.

Main Methods:

• Utilized an approximation to the variance of the treatment effect.
• Expressed variance in terms of proportions of clusters and individuals per sequence.
• Illustrated methods with SWDs in sexually transmitted diseases and renal replacement therapy.

Main Results:

• Identified key proportions of clusters and individuals for efficient SWD.
• Demonstrated that cluster-balanced designs possess excellent statistical and practical properties.
• Provided practical application suggestions for efficient SWD.

Conclusions:

• Efficient stepped wedge designs can be achieved even with unequal cluster sizes.
• Cluster-balanced designs are recommended for their robustness and practicality.
• Methods are extendable to closed-cohort SWD, enhancing applicability.