Build a reconciliation table that maps each trial's adverse-event terms to a common taxonomy, then flag definitions that cannot be merged and record whether they will be reported separately or narratively. The table makes each mapping decision visible and auditable before any pooling begins.
Set up the table
Create one row per adverse-event term as each trial defines it, with columns for the trial identifier, the trial's own term and definition, the proposed common-taxonomy category, the mapping status, and the reason for the decision. Pull the trial's term and definition from its registered outcomes and reported results fields, which ClinicalTrials.gov defines as structured registration and results data elements.
Reconcile and decide
- Extract each trial's adverse-event definitions verbatim into the table and mark the source field. Expected result: a complete term inventory with no paraphrasing that could blur differences.
- Map each term to a common-taxonomy category and mark it as merged, separate, or narrative. Expected result: every term carries an explicit status and a short reason.
- Flag terms that cannot be merged because the underlying definitions differ, and record the decision to report them separately or narratively. Expected result: a documented decision trail for each flagged term.
- Carry the merged categories into the synthesis and keep flagged terms visible alongside the pooled results. Expected result: readers can see which outcomes were combined and which were reported on their own.
Why the flags matter
Pooling outcomes that rest on different definitions can distort the combined estimate. Cochrane Handbook Chapter 10 covers effect measures, heterogeneity, and careful interpretation of pooled results; a reconciliation table gives you the record needed to explain which outcomes were pooled and which were kept apart. When a definition is ambiguous, mark the mapping unresolved and check the trial's full results entry before assigning a category.