Stepped-wedge design scenarios and power
Account for time and clustering
Simulate or calculate using the actual rollout schedule and plausible correlation within and across periods. A simple parallel-cluster design effect does not describe stepped-wedge information.
Flow diagram
Steps
Define rollout design
Specify clusters, sequences, periods and observation sampling.
Specify correlation structures
Distinguish within-period cluster, between-period cluster and repeated-person correlation.
Specify time and exposure effects
Record secular trends, transition periods and duration-dependent effects.
Assess loss and imbalance
Model unequal cluster sizes, missing periods and delayed rollout.
Evaluate design operating characteristics
Document simulation or design-compatible calculations and reproducibility.
Stepped-wedge design scenarios and power
End flow
Tables and supporting information
Stepped-wedge power scenarios
Do not use a parallel-cluster design effect as the full power calculation.
Assumptions and rationale
Power depends on the actual rollout and correlation model; more observations alone may not resolve design weaknesses.
Method and protocol evidence
Attach the protocol, methodological sources and versioned software specification supporting these decisions. Record citations and provenance; do not enter participant identifiers.