- Published
- Updated
A composite endpoint combines several events into one measure. Interpreting an overall difference as a similar effect on every component can be misleading unless the contributors to that difference are examined carefully.
Define the question and assumptions
Assess each component for clinical importance, anticipated frequency and measurement reliability. Define whether the combination counts first events, all events or a specified hierarchy. Component analyses provide context for the overall finding; separate confirmatory component claims raise their own multiplicity considerations and should not be assumed automatically.
Hypothetical study example
In a hypothetical study, a frequent but less severe component accounts for most events. The composite result appears favourable while uncertainty around a rare severe component remains wide. The team shows component distributions and confidence intervals instead of presenting established benefit for every severe event.
Design review checks
- Describe each component and its event ascertainment or adjudication method.
- Check how several events in the same participant contribute to the composite.
- Present component frequencies and effect directions alongside the overall result.
Record the decision and its limits
Accepted reporting keeps the combined definition consistent with component data and explains how frequency differences influence interpretation. Lack of statistical significance for a sparse component is not proof of no effect. Adding or removing components after outcomes are known cannot be justified solely by a more attractive combined result.
Frequently asked questions
- Does a composite establish every component separately?
- No. The overall inference differs from component-specific inferences.
- Must components occur equally often?
- Not necessarily, but frequency differences and their interpretive consequences need justification.