How to keep a reproducible literature-search log
Document your database, platform, full query, dates and deduplication. Use a worked planning record and a blank search-log template.
AcademicFix Editorial Team · Sources reviewed:
- Where?
- Database + platform
- How?
- Exact query run
- When?
- Search date + version
- What scope?
- Limits + reasons
Record how you reached the results, as well as the results themselves.
Go to the search-log template“We searched MEDLINE” does not show exactly what was run. Keep the source, platform, full strategy, date and limits together. This guide helps students and researchers document searches, including those for narrative literature sections. Keeping a log does not make a review systematic.
On this page
Open a record before you search
PRISMA-S is a 16-item reporting extension for systematic-review searches. It is not a search-quality score or a complete conduct manual. PRISMA-S explanation.
Record both database and platform, such as MEDLINE — Ovid. Platforms can differ in syntax and search fields. Keep the executed strategy, rather than reconstructing keywords later. PRISMA-S items 1 and 8.
The six groups below are AcademicFix’s working arrangement, not the official checklist. Give each source and update a row; link long queries to plain-text files.
| Record group | What to retain | A practical check |
|---|---|---|
| Identity | Short search ID, version and searcher | Avoid two different files both called the final version. |
| Source | Database or other source, plus platform or URL | Identify each database when several were searched together. |
| Strategy | Exact executed query, or the linked text file | Keep field codes, parentheses and line combinations. |
| Time and limits | Execution date; publication period, language and other limits | Distinguish no limit from an unfilled field. |
| Output | Displayed results, exported record count and output file | Explain any difference between the counts. |
| Next operation | Deduplication batch, change reason and previous version | Show which export entered the combined pool. |
Enter observed counts only. For an unexecuted query, leave counts blank and mark it as draft; zero would suggest an observed result.
A fictional record with no invented results
This is neither a research result nor a tested search strategy. It demonstrates the record structure only. Filled-in headings do not show that a query is suitable or complete.
| Field | Teaching example |
|---|---|
| Search ID / status | SEARCH-01 / not executed |
| Topic | Planned search about burnout among nurses |
| Database / platform | MEDLINE / Ovid |
| Strategy version | v01 — awaiting review |
| Full query | Not yet prepared; an executable strategy will be added separately |
| Execution date | Blank — no search has been run |
| Planned publication-year limit | None planned; verify the applied setting after execution |
| Results / exported records | Blank — no observed counts |
| Check note | Terms, field codes and scope to be reviewed with an information specialist |
The topic label is not a finished query. Develop and review your strategy, then record the actual execution. Cochrane recommends ongoing documentation and early information-specialist involvement. This health-review example does not prescribe one database for every discipline. Cochrane §§4.2.1 and 4.5.
Keep an update separate from the old record
Retain previous and updated strategies, with changes and reasons. Keep the execution date separate from a publication-period limit such as 2015–2025; if only the execution date changed, record that too. PRISMA-S items 12–13.
Suggested names: SEARCH-01_v01_query.txt and SEARCH-01_v01_records.ris, linked from the log. Give the next run a separate v02 record. Name any added term and explain why. This is our naming suggestion, not a mandatory standard.
Separate duplicate records from reports of one study
One article retrieved twice may have two records. A study’s protocol, results article and follow-up report can be different documents. Similar titles or shared authors alone do not justify deletion. PRISMA 2020 distinguishes records, reports and studies. PRISMA 2020 glossary.
In our proposed processing log, retain the software/version, matching rule and notes on ambiguous pairs checked manually. Preserve source exports and trace which records were combined. Link different reports of a study because a secondary report may contain useful design or outcome information. Cochrane: managing and linking reports.
For a worked example of this distinction, see the report-to-study mapping guide.
Turn the working record into a methods account
Connect the concise methods description to supplementary files containing the full strategies. PRISMA 2020 recommends reporting strategies for all searched databases, registers and websites, rather than only one. PRISMA 2020, search-reporting update.
Fill-in reporting outline: “We searched [source] through [platform] on [actual date]. The full strategy as executed is provided in [supplement]. We recorded [limits and their reasons]. The update changed [details]. Duplicate records were handled using [software/version and decision process].”
Replace every bracket with an actual procedure. A later rerun can return a different count as databases change; a log does not demonstrate that every relevant study was found. PRISMA-S item 15 and scope.
Add the log to your working files
Download the blank search-log CSV. It is an AcademicFix working template, not an official PRISMA-S form. Import it as UTF-8 if characters display incorrectly. The count and execution-date cells are intentionally empty.
Compare retrieved studies in a literature matrix. Organise document sections with the Word research-template builder; it does not validate search strategies.
Before returning to your study
- Can each search be traced to its source, platform and execution date?
- Can you open the complete executed query, rather than only a keyword list?
- Can reported counts be traced back to actual export files?
- Have changes, reasons and previous versions been retained?
- Have duplicate records been distinguished from different reports of the same study?
- Have private team notes and access credentials been removed from material you plan to share?
Frequently asked questions
Is a screenshot of the search screen enough?
A screenshot can support the record, but it may hide query lines or limits. Keep a text copy where possible. Cochrane §4.5.
Does completing PRISMA-S certify my search quality?
No. It helps describe the search; it does not certify its quality or completeness. Source selection and methodological assessment remain separate tasks. PRISMA-S.
Is a search log the same as a literature matrix?
The log traces how sources were retrieved. A literature matrix compares their research questions, methods and findings. Link the two rather than replacing one with the other.
Sources and scope
Sources are linked beside the methodological claims they support. The example tables are teaching illustrations, not findings from a real study.
- Rethlefsen et al. (2021) — PRISMA-S explanation
- Cochrane Handbook, Chapter 4 (2025) — Searching for and selecting studies
- Page et al. (2021) — PRISMA 2020 statement
A shorter introduction appears in our Turkish Instagram post.
Continue reading
- Map multiple reports to their underlying studies
- Compare retrieved studies in a literature matrix
- Build the section structure of your research document