- Published
- Updated
When comparing a trial arm with people from an external source, follow-up should begin at comparable clinical moments. Similar age or diagnosis distributions do not automatically remove systematic differences in when people become eligible for observation.
Define the question and assumptions
Map eligibility assessment, treatment initiation and outcome-follow-up start separately in each source. Selection using future information in one group can introduce bias. Explain how each concept is observed in different record systems; similar date-field names are not sufficient evidence that the underlying clinical moments or inclusion processes are equivalent.
Hypothetical study example
In a hypothetical study, the trial arm starts follow-up at randomisation while external controls start at diagnosis, substantially earlier. The team examines selection and events during that interval. It explains why making numerical follow-up lengths equal does not by itself resolve the mismatch in clinical time origins.
Design review checks
- Define eligibility, exposure and follow-up origins explicitly for each comparison group.
- Identify inclusion conditions derived from records occurring after the intended origin.
- Test classification of pre-origin events using representative individual timelines.
Record the decision and its limits
An accepted design reproducibly shows who enters the comparison and at what moment. If a comparable origin cannot be established, carry that limitation into interpretation. Do not assume statistical matching automatically repairs a systematically different starting point. Assess external-control suitability alongside source quality, clinical context and the potential bias that remains.
Frequently asked questions
- Does equal follow-up length align the starting point?
- No. Clinical origins and the inclusion process must also be comparable.
- Do identical date-field names establish identical definitions?
- No. Verify the events and recording processes behind each field.