Search Trial Registries Beyond Published Articles

Start with the study question

A registry search answers a different retrieval problem from an article search. The object is a registered study, which may be ongoing, terminated, completed without results, or linked to several reports. Write the eligible population, intervention, comparison and study design before entering terms. Do not require a publication or a posted numerical result as a condition for discovering the record.

Cochrane’s search guidance includes trial registers among sources for identifying studies. Treat each registry as a named source with its own coverage and searchable fields. A record absent from one source is not proof that the study was never registered.

Search and preserve the actual record

Use intervention names, synonyms, development codes and relevant condition terms. Start broad enough to inspect how records describe the intervention. If you restrict recruitment status, geography or dates, record the scientific reason. Completed studies without posted results may be particularly important to follow up, so a results-only filter can defeat the purpose of this search.

Keep the executed query, interface, search date, result count and export. For each potentially eligible record, preserve the registration identifier and version or retrieval date. Extract anticipated versus actual enrollment separately. Record completion status independently from publication status; an apparently finished study may still lack usable outcome data.

Reconcile a hypothetical search

Suppose a registry search retrieves 48 records. Screening identifies 12 potentially eligible studies: seven completed, three recruiting and two terminated. Linked publications are found for five of the seven completed studies. These numbers describe retrieval and status, not treatment effects. The two completed studies without linked publications need follow-up; they must not silently disappear from the evidence map.

If a bibliographic search already found four of those five published studies, the registry contributes one additional published study plus the unresolved and ongoing records. Do not count all 12 as newly discovered studies without checking overlap.

Close the loop

Compare dates, sponsor, sample size, sites and intervention arms before linking records to articles. A shared acronym alone is insufficient. Record uncertainty when identifiers are absent or conflicting. Check whether posted results use the same outcome definition and assessment time as the planned review.

The finished search package should let another reviewer recover the query and explain each record’s current status. Schedule a final status update near the review’s search cutoff, keeping the earlier snapshot. This preserves the difference between a study that was unavailable at the cutoff and one that appeared later.

Source verification: 6 November 2026. Examples are hypothetical.