Unequal cluster sizes in stepped-wedge cluster randomised trials: a systematic review

Area of Science:

• Biostatistics
• Clinical Trials Methodology
• Epidemiology

Background:

• Stepped-wedge cluster randomised trials (SW-CRT) are increasingly used in various settings.
• Variability in cluster sizes is a common characteristic of these trials.
• Inconsistent accounting for cluster size variability can affect trial outcomes.

Purpose of the Study:

• To assess the extent of cluster size variation in SW-CRTs.
• To determine if this variability is adequately addressed in sample size calculations and analyses.
• To highlight reporting gaps in SW-CRT methodology.

Main Methods:

• Systematic review of SW-CRTs published up to March 2016.
• Analysis of cluster size variability using the coefficient of variation (CV).
• Comparison of assumed vs. actual cluster sizes and assessment of analytical methods used.

Main Results:

• 48% of SW-CRTs acknowledged cluster size variation, but only 13% incorporated it into sample size calculations.
• 69% used appropriate analytical methods for variable cluster sizes.
• Median CV was 0.41; actual cluster sizes deviated significantly (29%-480%) from assumed sizes.

Conclusions:

• Cluster size variability is prevalent in SW-CRTs and often inadequately handled in sample size planning.
• Suboptimal reporting of SW-CRTs necessitates improved methodological transparency.
• Further research is needed on the impact of unequal cluster sizes on statistical power, and appropriate methods should be adopted.