How do I resolve a conflict between a trial's published and registry outcomes?

Resolve the conflict by comparing the outcome definition, timepoint, and analysis in the paper against the registry entry, then cite each source for the outcome it actually reports. ClinicalTrials.gov's data element definitions describe structured registration and results fields covering outcomes and reported results, which gives you a field-by-field basis for the comparison.

1. Align the outcome measure. Match the paper's primary and secondary outcomes to the registry's registered outcome fields. Record whether the measure, instrument, or definition differs; a changed definition is a different outcome, so note it as a discrepancy rather than a match.
2. Align the timepoint. Compare the registered assessment timepoint with the one reported in the paper. A different timepoint changes what the result describes, so cite the source that reports the timepoint you are using.
3. Align the analysis. Check whether the paper's analysis population and effect estimate correspond to what the registry entry describes. Where they diverge, treat the two results as separate findings.
4. Decide the citation per outcome. For each outcome, cite the paper when it reports that outcome with a matching definition and timepoint; cite the registry entry when the paper omits the registered outcome or reports it differently. State which source you used and why in your methods or evidence table.
5. Carry the decision into certainty. GRADE rates certainty at the outcome level using explicit criteria, so a discrepancy that leaves an outcome unresolved lowers your confidence in that outcome's estimate. Cochrane Handbook Chapter 15 guidance on confidence intervals, exact P values, and absolute versus relative effects supports reporting the estimate with its precision and noting the discrepancy alongside it.

Document the comparison in a short table: outcome, registered definition and timepoint, published definition and timepoint, analysis used, source cited, and the certainty judgement for that outcome. This keeps each decision traceable when the review is updated or checked.