Stepped-wedge time-adjusted analysis

Separate intervention effects from trends

Calendar time and intervention exposure are associated by design. Model time explicitly, justify dependence assumptions and assess sensitivity to exposure-duration and time-trend specifications.

Flow diagram

Stepped-wedge time-adjusted analysisF1.S1Define primary effect modelF1.S2Model calendar time explicitlyF1.S3Model multilevel dependenceF1.S4Assess exposure-duration effectsF1.S5Plan inference and sensitivityF1.N1Stepped-wedge analysis planContrastTime adjustmentCorrelationExposuredurationSensitivityPrimary effectAlternativetrendDuration effectMissing periodsF1.N2Assumptions and rationaleF1.N3Method and protocol evidenceF1.S0 Start flowStepped-wedge time-adjusted analysisF1.S999 End flowEnd flow

Steps

Define primary effect model

Choose an outcome-appropriate model for the sampling structure.

Model calendar time explicitly

Prespecify period effects or justified trends to separate rollout from secular changes.

Model multilevel dependence

Account for cluster-period and repeated-person correlation as applicable.

Assess exposure-duration effects

Specify constant or time-varying treatment effects and transition handling.

Plan inference and sensitivity

Address few clusters, missing periods, trend flexibility and rollout deviations.

Stepped-wedge time-adjusted analysis

End flow

Tables and supporting information

Stepped-wedge analysis plan

Stepped-wedge analysis planContrastTime adjustmentCorrelationExposuredurationSensitivityPrimary effectAlternativetrendDuration effectMissing periods

Keep planned models separate from completed estimates.

Assumptions and rationale

Ignoring calendar time can confound treatment with rollout; model adequacy still requires investigation.

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.

Sources