- Published
- Updated
Whether a participant experiences an event at least once and how often it recurs are different questions. Recurrent-event assessment requires explicit rules defining when a new episode starts and how it differs from continuation of an earlier episode.
Define the question and assumptions
Define event duration, recovery, new onset and relationships to planned procedures. Counts per participant, proportions experiencing an event and follow-up-adjusted rates are different measures. If follow-up duration or observation intensity varies, explain why raw counts may not provide a directly comparable measure of burden across groups.
Hypothetical study example
In a hypothetical study, two records from one clinical episode are entered as separate events. Another site follows a prolonged episode within one record. The team applies episode-merging and new-event rules while preserving raw records and showing how analytical events were derived.
Design review checks
- Test the distinction between new and ongoing events using representative source records.
- Assess follow-up time and observation frequency alongside the resulting event counts.
- Address dependence between events within the same participant in the analysis plan.
Record the decision and its limits
Acceptance requires reproducible totals on example participant timelines. Explain separately how death or withdrawal changes the opportunity for further events. The influence of one participant with many events can be examined, but that person should not be excluded simply because their observations change the result.
Frequently asked questions
- Is the proportion with at least one event sufficient?
- If the question concerns burden, recurrence frequency and timing may also be important.
- Should duplicate-looking records be deleted automatically?
- No. Review sources and event definitions, then document the derivation decision.