_(18492433125).jpg){kind=link}
- Published
- Updated
An assessment outside its window is not identical to missing data. Its availability also does not prove suitability for the targeted time. Separate collection, classification and analysis decisions before choosing which observation to use.
Define the question and assumptions
Prespecify how protocol visits map to analysis visits. Include rules for several observations within one window and for observations outside every window. Continuing assessments needed for clinical follow-up and selecting the primary analysis value are different responsibilities and should remain distinguishable in the records.
Hypothetical study example
In a hypothetical trial, a participant attends late and has another measurement the next day. Both records are retained. The analysis observation is chosen using the planned timing and selection rule rather than because one of the measurements produces a more favourable outcome.
Design review checks
- Keep the original visit label separate from the derived analysis visit.
- Test selection priorities for repeated measurements and equally distant observation dates.
- Inspect the frequency and treatment-arm distribution of out-of-window observations.
Record the decision and its limits
Acceptance examples cover window boundaries, missing dates and overlapping visits. The derivation trail should also explain why observations were not selected. Where necessary, supportive analyses explore the effect of these records; the primary endpoint definition must not silently expand or contract after results are seen.
Frequently asked questions
- Should an out-of-window measurement be deleted?
- No. Preserve the source record and define its analytical role separately.
- Is the closest date always the right rule?
- No. The rule must fit the measurement and scientific objective.