- Published
- Updated
Rescue medication matters both for participant management and interpretation. Reducing its use to a yes-or-no field can lose the timing, reason and relationship to endpoint assessments that are needed to understand the treatment experience observed.
Define the question and assumptions
Define rescue according to the clinical protocol rules. Determine how the target treatment-effect question handles this event and which observations are needed. Using post-rescue values, defining a different outcome or considering a hypothetical scenario do not necessarily answer the same question; justify the decision jointly with clinical and statistical colleagues.
Hypothetical study example
In a hypothetical trial, one arm receives rescue earlier and subsequent symptom measurements change. The team examines timing and the prespecified strategy rather than comparing only final observations without context. Necessary clinical rescue is not delayed or prevented to make the statistical analysis more convenient.
Design review checks
- Link rescue initiation dates and reasons with the relevant endpoint assessments.
- Separate clinical rules governing rescue from analytical classification of the event.
- Test whether data collection forms can implement the planned rescue strategy.
Record the decision and its limits
An accepted design explains which treatment experience it summarises. Do not hide differences in rescue use or automatically exclude every later observation. Plan the main analysis and uncertainty assessments consistently with the target question. Define clinical responsibility and operational authority so participant safety remains central to rescue decisions.
Frequently asked questions
- Must rescue use count as treatment failure?
- That depends on the scientific question and prespecified endpoint strategy.
- Should post-rescue data be deleted?
- No. Preserve source observations and define their analytical role separately.