Steppedwedge

A stepped-wedge design is a type of longitudinal cluster randomized trial in which groups or clusters switch from a control condition to an intervention in a staged, sequential order over multiple time periods. All clusters begin the study under control, and at predetermined steps, one or more clusters cross over to the intervention until every cluster has received it. Data are collected from all clusters at each time point, allowing within-cluster and between-cluster comparisons across time.

This design is often used when withholding the intervention for the entire study is unethical or impractical, or when a phased rollout aligns with available resources, logistics, or policy implementation plans. It also accommodates situations where a comprehensive, contemporaneous rollout is not feasible, yet evaluation is still needed.

Analysis typically employs longitudinal, cluster-adjusted models (such as mixed effects or generalized estimating equations) that account for clustering and for time trends. The design enables estimation of the intervention effect while controlling for secular trends, since outcomes are observed in both control and intervention states across time.

Advantages include ethical appeal, practical rollout, and the ability to use each cluster as its own control over time. Limitations involve complex sample size calculations, potential confounding from time-related factors, longer study duration, and possible contamination between clusters. Variants exist, with different numbers of steps or staggered rollout schedules.

Stepped-wedge trials are commonly used in implementation research and health services studies to evaluate policy changes, new protocols, or quality-improvement interventions.