Preserve the literature search that generated the evidence set

Preserve the actual execution

PRISMA-S recommends exact-run search records and transparent reporting of changed update syntax. This reporting guidance does not guarantee adequate coverage or identical future results.

Start the record while searching, not while writing the final report. Identify the review question and the working search version. Keep the expression distinct from the plain-language description of what the searcher meant to find. A well-written intention cannot reveal an accidental omission in the executed query.

Record the source service and relevant interface alongside the date. A database name alone may not tell another researcher which implementation accepted the expression. Preserve the observed result count as context, not as a performance target. If a search returns fewer records than expected, do not quietly broaden it and overwrite the original run. Keep the change and its reason visible.

Separate a retrieval decision from a review decision

List each deliberate restriction and the decision it serves. A date boundary may fit the technology period under review; a language boundary may instead reflect available resources. Those are different explanations. State the limitation honestly rather than presenting convenience as evidence that excluded material is irrelevant. Have the information specialist challenge restrictions that could remove important competing explanations.

Keep rejected search experiments apart from the final search set without deleting their rationale. A query that retrieves many irrelevant uses of a business term can still teach the team how the vocabulary behaves. Record the adjustment that addressed the ambiguity. Do not report only the prettiest final expression if the review’s coverage depends on earlier searches whose records remain in the candidate pool.

Choose what an update is allowed to change

Before refreshing the search, decide whether the purpose is to extend the original review or redesign its question. An update may uncover new vocabulary or a changed interface. Record the proposed change before interpreting its effect on the resulting evidence. If the question has materially changed, label that change rather than suggesting that a later count is directly comparable with the first.

Use a small set of already-known relevant records as diagnostic examples where appropriate. A missing example should prompt investigation of indexing, coverage or syntax; its presence does not prove that the search finds everything. Keep this diagnostic exercise separate from selecting which studies support a desired conclusion. The search should expose material for assessment, not manufacture agreement with the draft headline.

Make the record usable by someone else

Give a colleague the question, saved execution and update note without explaining them verbally. Ask them to identify what was searched, what was restricted and which version produced the candidate pool. If they must guess whether a note describes an intention or an executed run, repair the record. This is a practical handoff check, not a demand that a dynamic service reproduce its historical count exactly.

Finish with a short update decision: rerun the accepted strategy, change specified elements with reasons, or reopen the review question. Preserve original and revised executions in the review’s authorized workspace. Do not copy entire articles into the record when a reference is sufficient. The deliverable should make later methodological changes inspectable while leaving study eligibility, extraction and synthesis to their own decisions.

Make the reasoning concrete.

A fictional industry association reviews research about business-document findability. Six months later, its searcher discovers that the original query used a narrower phrase than the methods note describes. The team must decide whether a refresh continues the original method or repairs it.

The association records a methodological correction rather than calling the changed run an identical refresh. It makes no claim that the revised search is exhaustive.

Take the questions into your next brief.

Use these prompts in your own notes. Nothing is submitted from this page.

1. Execution

   What expression actually ran, where and when?

2. Restrictions

   Which limits were deliberate and why?

3. Version

   Which run contributed each candidate set?

4. Change

   Is the refresh an extension or a revised method?

5. Handoff

   Can another researcher identify these decisions without guessing?

Watch for the wrong conclusion.

• Reconstructing the query from memory.
• Overwriting an earlier run with a cleaner expression.
• Treating a result count as proof of coverage.

Sources & scope

Workflows and explicitly fictional adult business examples are editorial analysis. Methodological statements and their applications remain source-derived. Cochrane-specific requirements are not universal marketing mandates. No real research operations or publication are performed.

• PRISMA-S: an extension to the PRISMA Statement for Reporting Literature Searches in Systematic Reviews →Primary technical documentation · Supports: Supports exact-run reporting and transparent search updates, not guaranteed completeness.