How to Search Grey Literature for a Systematic Review
Start with a coverage rationale, not a universal source list. Identify which unpublished, non-commercial, regulatory, organizational, repository, register, thesis, or conference evidence could meet the review’s eligibility criteria; select sources that plausibly contain it; then record the exact query or browse route, date, searcher, limits, depth, stopping rule, yield, retained records, storage, update, and reproducibility limitations.
Grey literature is a practical label for evidence distributed outside conventional commercial journal publishing. Boundaries vary across disciplines and organizations, so source selection has to follow the review question. Grey-literature searching may reduce some publication and dissemination biases, especially when eligible studies or reports don’t reliably reach journals, but it can’t eliminate bias or guarantee comprehensiveness.12
Begin by sketching the publication pathway. Ask who produces the evidence, where it must be registered or submitted, which organizations commission it, where reports are archived, and how findings circulate when journal publication is delayed or absent. This reasoning is stronger than listing sources first because it exposes both coverage and gaps.
Decide whether grey-literature searching is warranted
The need is stronger when eligible evidence is commonly registered before publication, commissioned by governments or organizations, submitted to regulators, reported in theses or conferences, produced for policy decisions, or delayed or selectively published. It can also matter in emerging fields, rare conditions, adverse effects, implementation research, public health, education, international development, and topics with strong commercial or political incentives.
That doesn’t mean every review must search every grey source. Search burden can be substantial, interfaces can be weak, and some sources may add little relevant evidence. Record both selected and rejected source categories with reasons.
Build a conditional source map
| Source category | When it may matter | What to document |
|---|---|---|
| Trials and study registers | Intervention studies, ongoing or unpublished work, prospective records, and results postings. | Register, coverage, fields searched, query, status filters, dates, record IDs, and links between registrations and reports. |
| Institutional or subject repositories | Preprints, reports, datasets, working papers, and institution-produced research. | Repository, collection, metadata fields, query or browse path, date range, capture method, and duplicate handling. |
| Theses and dissertations | Topics where graduate research may contain unique methods, data, or null findings. | Database or repository, degree/geography coverage, query, filters, full-text access, and report-study linkage. |
| Government, regulatory, and HTA sources | Policy, safety, approvals, submissions, surveillance, assessments, and commissioned research. | Agency, jurisdiction, document type, website sections, search/browse route, version, and archived copy. |
| Organizations and professional bodies | Guidelines, technical reports, programme evaluations, standards, and sector-specific evidence. | Organization-selection rationale, site route, terms, folders, contacts, and document version. |
| Conference outputs | Recent or incompletely published studies and disciplines with important meeting dissemination. | Meeting, years, abstract source, query/browse method, duplicate publication links, and update plan. |
| Search engines | Cross-site discovery, hard-to-find reports, and identification of organizations or terminology. | Engine, exact terms, date/time, location or account context if relevant, ranking depth, stopping rule, and captured URLs. |
| Expert or organization contacts | Known networks, incomplete records, unpublished studies, missing reports, or clarification. | Who, why selected, request, date, response, follow-up, permissions, and how supplied material was handled. |
Adapt the search to the source
Don’t paste a long bibliographic-database strategy into a simple website search box and assume it ran. Many sites ignore Boolean operators, truncate the query, search only a title field, use opaque ranking, or have no stable history. Start with the few concepts most likely to discriminate the topic, test synonyms individually, use available filters cautiously, and supplement search boxes with structured browsing.
Document search engines without overstating reproducibility
Google and Google Scholar can help discover reports and terminology, but ranking, personalization, location, index changes, result caps, and limited export functions complicate reproducibility. Google Scholar’s broad coverage and citation links can be useful, yet its grey-literature role has to be understood alongside these constraints.3
Before searching, define a screen depth such as the first 100 results or first 10 pages. Record the exact query, date and time, browser/account state if material, location context if material, filters, number screened, number captured, and reason for stopping. Preserve URLs or screenshots rather than claiming that another researcher will see the same ranking.
Use a screen-depth rule and a stopping rule
A screen-depth rule states how much of a source will be reviewed: a number of results, pages, folders, years, document types, or site sections. A stopping rule explains when work ends. The strongest approach defines a maximum depth in advance and, when appropriate, adds a topic-specific rule such as no new potentially eligible reports across a stated number of consecutive pages.
Example: screen the first 100 ranked results and every report in the source’s named policy collection from 2020 onward; stop only after the predefined depth is reached. Record whether the final two pages produced any new potentially eligible records.
This makes the work auditable. It doesn’t show that results below the cutoff were irrelevant.
Separate yield, unique records, and retained records
A website may display an estimated result count that changes as the user navigates. Record that displayed yield if useful, but distinguish it from the number actually screened, captured, exported, deduplicated, and retained for formal screening. Link registrations, conference abstracts, reports, preprints, and journal articles that describe the same underlying study rather than counting them as independent studies.
Preserve evidence and project files
For each source, name the export file, capture folder, screenshot, URL list, citation file, or web archive. Record access failures and unavailable full text. If content changes, retain the version used in the review where lawful and practical. Store contact requests and responses with appropriate permissions and privacy protections.
Contacting organizations and experts
Use a standardized request that identifies the review, eligibility boundary, types of material sought, date range, response deadline, and preferred metadata. Record how contacts were selected, not just whom the team happened to know. Follow up consistently and report non-response. Materials supplied by experts still require eligibility assessment, duplicate linkage, and quality appraisal.
Plan updates for unstable sources
Registers, repositories, agency sites, and organizational pages can change between the original search and publication. Record an update date or trigger, rerun the same sources where feasible, document syntax or interface changes, and report new records separately. If a source has disappeared or changed beyond replication, say so and preserve the evidence trail available.
Report methods and limitations together
PRISMA-S asks authors to report information sources, search strategies, dates, limits, supplementary methods, peer review, records, deduplication, and updates.4 For grey sources, a useful report includes source identity, coverage rationale, exact query or browse route, date, limits, screen depth, stopping rule, yield, capture method, contact process, update status, and non-reproducible features.
Don’t use the number of sources searched as a proxy for quality. Explain which evidence pathways were covered and what remained inaccessible. Recent research continues to identify time, discoverability, interface, documentation, and access barriers in grey-literature searching; those barriers should be planned for and disclosed rather than hidden.5
Grey source does not mean low-quality study
Publication status and methodological quality are separate. A journal article can be poorly conducted, and a government report or thesis can be rigorous. Apply the review’s eligibility criteria and appropriate risk-of-bias or quality-appraisal method to the evidence itself. Don’t exclude or downgrade a study solely because it’s grey literature.
Treat browsing as a reproducible route where possible
Some sources are best searched through a hierarchy rather than a query. A regulator may organize documents by product, indication, decision, and year; an organization may divide evidence across programme pages, publication series, and archives. Record each step in the route and the boundaries applied at that step. If a page contains a fixed list, note the list title and version or access date. If links open dynamic filters, record the selected values. Capture stable URLs and lawful snapshots where practical.
The route should allow a second researcher to repeat the intent even if the exact interface changes. Where replication is impossible, say precisely why: ranking changed, history was unavailable, content required an account, a page was removed, results were personalized, or only part of the screened set could be exported. “Not fully reproducible” is useful metadata when supported by a clear explanation.
Link reports that describe the same study
Grey searching often retrieves multiple records for one underlying study: a register entry, conference abstract, thesis, funder report, regulatory submission, preprint, and journal article. Preserve every report needed for eligibility and data collection, but link them under a study identifier before interpreting counts. The number of records retrieved is not the number of studies identified. Record how links were established and retain uncertainty when identity is not clear.
Plan the capture before searching
Decide where exports, PDFs, screenshots, URL lists, contact responses, and notes will be stored. Use filenames that connect each artifact to the source, date, searcher, and log row. For sources without export, record the minimum metadata required to relocate each potentially eligible record and preserve the page address. A search can’t be audited if its evidence trail exists only in browser tabs or personal email.
Close the search with a coverage statement
A useful conclusion describes the evidence pathways covered, the sources and methods used, and the main residual gaps. It doesn’t substitute the number of sources for quality. State which jurisdictions, organizations, years, languages, document types, or interfaces remained inaccessible; whether contacts responded; how dynamic ranking constrained the method; and when the sources were last checked. This gives readers a defensible basis for judging the search and gives the review team a concrete update plan.
References
- Paez A. Gray literature: An important resource in systematic reviews. Journal of Evidence-Based Medicine. 2017;10(3):233–240. https://doi.org/10.1111/jebm.12266.
- Adams J, et al. Searching and synthesising ‘grey literature’ and ‘grey information’ in public health. Systematic Reviews. 2016;5:164. https://doi.org/10.1186/s13643-016-0337-y.
- Haddaway NR, et al. The role of Google Scholar in evidence reviews and its applicability to grey literature searching. PLoS ONE. 2015;10(9):e0138237. https://doi.org/10.1371/journal.pone.0138237.
- Rethlefsen ML, et al. PRISMA-S. Systematic Reviews. 2021;10:39. https://doi.org/10.1186/s13643-020-01542-z.
- O’Sullivan J, Dawson S, Higgins JPT. Barriers to searching for grey literature in systematic reviews: A survey of authors and editors. Research Synthesis Methods. First View; published online 12 May 2026:1–13. https://doi.org/10.1017/rsm.2026.10087.
- Lefebvre C, et al. Searching for and selecting studies. Cochrane Handbook, version 6.5. 2024. Official chapter.
