Evidence Synthesis
Inspect How PubMed Translated a Search
Check the query the database executed
A research team enters a phrase in PubMed and saves the words typed into the search box. To understand why known papers appeared or disappeared, open Advanced Search and inspect Search Details for that run. Copy the submitted query, its translation, and any messages into separate fields. PubMed can map untagged words to indexed concepts. Phrases, field tags, and other syntax can change that behavior.
Pick a paper that should match the intended concept. Locate the evidence in its title, abstract, or indexing terms. Make one change to the search and compare the actual identifiers retrieved by each version. Record a concrete reason for a gained or lost paper, such as a synonym retrieved through mapping. A difference in totals alone cannot identify that mechanism.
Reconcile a hypothetical comparison
An untagged search returns 80 citations and an explicitly tagged search returns 65. Sixty identifiers occur in both sets. The first query adds 20 unique citations and the second adds five. Looking only at the net difference of 15 would conceal those five additions. Inspect the five alongside a sample of the 20 before deciding which search belongs in the protocol.
Keep the execution date and database name with the translation. Subsequent indexing and software changes can affect a rerun. The saved identifiers let another reviewer distinguish a change in search behavior from a change in the available literature.
Related worksheet: Known-item retrieval test.
Source checked on 5 November 2026.