Sample size calculations for stepped wedge trials using design effects are only approximate in some circumstances

Area of Science:

• Biostatistics
• Clinical Trials Methodology

Background:

• Sample size and power calculations are crucial for stepped wedge cluster randomized trials (SW-CRTs).
• Existing methods include exact analytical approaches, design effects, and simulation.
• A recent study highlighted discrepancies between the design effect and analytical methods for SW-CRTs.

Discussion:

• The design effect method for SW-CRTs is technically accurate only when an equal number of clusters transition at each step.
• Differences observed between design effect and analytical methods stem from unequal cluster transitions.
• Unequal cluster transitions violate the assumptions of the standard design effect calculation.

Key Insights:

• The design effect method for stepped wedge cluster randomized trials is only appropriate for balanced designs.
• Analytical methods or simulation may be more suitable for unbalanced stepped wedge designs.
• Accurate sample size estimation requires careful consideration of the specific trial design.

Outlook:

• Further research is needed to refine sample size and power estimation methods for unbalanced SW-CRTs.
• Development of robust analytical or simulation-based approaches for unequal cluster transitions is warranted.
• Guidelines for choosing appropriate sample size methods based on SW-CRT design balance are recommended.