Peer Review a Search Strategy Before Running the Final Search

Review the executable strategy

A prose description of a search is not enough for peer review. Supply the exact line-by-line strategy, database and platform, planned coverage dates, review question and known relevant studies. The reviewer needs to inspect what the system will execute, including parentheses, adjacency syntax and field restrictions.

Cochrane’s search chapter recommends involvement of information specialists and discusses peer review of strategies. The review should happen early enough to correct the search before screening effort makes changes expensive.

Check meaning before mechanics

Ask whether each required concept is necessary for retrieval. Requiring an outcome term can miss eligible studies whose titles and abstracts describe the intervention but omit that outcome. Inspect synonyms, controlled vocabulary and spelling variants within each concept. Terms combined with OR usually broaden that concept; concepts combined with AND require intersection.

Then inspect database-specific syntax. A strategy copied from another platform may use unsupported field labels or proximity operators. Check brackets, line references, truncation and limits by executing the relevant parts and inspecting translated queries or representative records.

Use a hypothetical logic test

Suppose concept A retrieves 1,000 records and concept B retrieves 800, with 300 shared records. A OR B retrieves 1,500 unique records, while A AND B retrieves 300. If a misplaced operator changes a planned synonym block from OR to AND, it can remove records before the clinical reviewer sees them.

A known-item set can reveal such errors. If eight previously identified eligible studies should be indexed in the database but only six are retrieved, investigate the two misses. Do not simply add their exact titles and declare the strategy validated; determine which missing terminology or field restriction caused the failure.

Record actions and residual limits

Maintain a review table with issue, proposed change, decision, rationale and resulting strategy version. Some suggestions may conflict with the intended question or database functionality; record why they were not adopted. Rerun affected lines after changes and preserve both versions.

A search that retrieves every known item may still miss unknown studies, and a peer review does not establish perfect sensitivity. Its value is a systematic check of avoidable errors and coverage assumptions. The final package should contain the executable strategy, reviewer feedback, resolved decisions and search dates so that later updates begin from an understood method rather than an unexplained block of search text.

Source verification: 6 November 2026. Examples are hypothetical.