How to Search for Grey Literature Without Lowering Your Standards

Grey literature is where a focused evidence question often goes to be tested. Registers, agency reports, and institutional publications can carry outcome data that never appears in a journal article, but they arrive without the packaging that makes a peer-reviewed study easy to appraise. The way to keep your standards intact is not to exclude these sources by default. It is to define what counts as eligible before you search, document where each record came from, and label its publication status in the evidence record so that readers can see exactly what kind of document is being cited.

This article sets out a complete sequence for that work, using a single running example: adults with a specified chronic condition, and the outcome of condition-specific hospital readmission within 30 days. The example is deliberately narrow so that eligibility, provenance, and applicability can be judged against a fixed population and outcome rather than against a vague topic. The process produces a defensible evidence record for research, guideline development, or scholarly use.

The Cochrane Handbook's guidance on searching for and selecting studies begins with planning searches from the eligibility criteria. That ordering matters for grey literature because the volume of non-journal material is large and loosely indexed. If you begin with a search engine, you will spend your time triaging whatever surfaces rather than retrieving what your question requires.

Write the eligibility criteria first, in the same structure you would use for a database search. For the readmission example, that means specifying the population (adults with the named chronic condition), the outcome (condition-specific readmission within 30 days), and the study designs you will accept. Then add a separate grey-literature eligibility layer that answers three questions:

• Which source types are eligible? For instance, trial registers, health technology assessment reports, national audit reports, or institutional research reports.
• Which publishers or institutions are eligible? Named agencies, registries, or institutions whose methods you can inspect.
• What minimum documentation must a record carry? A date, an authoring body, and enough methodological description to judge how the outcome was measured.

Deciding these in advance is what prevents the common failure mode in which a document is admitted because it is the only thing available on a sub-question. If a source type is not eligible, its absence from your evidence record is a planned decision rather than an oversight.

Cochrane recommends using multiple sources rather than relying on a single channel. For grey literature, translate that into an explicit source map with categories, so that coverage can be described later.

Keep the map alongside the eligibility criteria. When you later describe your search, the map becomes the structure of that description, and any category you did not search is visible rather than implied.

Cochrane's guidance is to combine free-text terms with subject headings. Grey literature complicates this because institutional repositories rarely apply the subject headings used in biomedical databases. Two practical responses follow.

First, run the structured search in the bibliographic databases where subject headings exist, and use that search to identify the concepts and named reports that then drive grey-literature retrieval. PubMed supports field tags, phrase searching, filters, and Advanced Search, and its search details show how terms were translated and combined. Reviewing those details is a diagnostic step: it tells you whether a concept was actually searched or silently dropped.

Second, for institutional sources, build a free-text string from the population and outcome terms plus the names of the agencies on your source map. Record the exact string and the date it was run for each source, because institutional sites change without notice.

Discovery tools help you find candidate documents; they do not settle eligibility. WisPaper's scholarly search interface is a place to find academic papers relevant to your question, and the public Scholar Search entry point explains the same finding function for readers new to the platform. Use either to assemble the journal and preprint layer that will anchor your grey-literature chasing.

Google Scholar's documented features are useful at this stage for a specific reason: searching by author and title, date sorting and date limits, Cited by and Related articles, citation export, and locating alternative versions of a work all help you trace a report from a citation in a journal article back to the document itself. Citation export also gives you a machine-readable record to check later.

Diagnostic checkpoint: for every candidate you keep, you should be able to name the page where you found it, the page where the document itself lives, and the authoring body. If any of those three is missing, the record is not yet eligible.

Provenance is the difference between grey literature and an unattributed document. Apply the same checks to every candidate so that your record is consistent.

1. Authoring body. Identify the institution or agency responsible, not merely the website that hosts the file.
2. Date and version. Record the publication or release date and note whether a later version exists.
3. Method description. Confirm the document states how the population was defined and how the outcome was ascertained.
4. Identifier. Capture a report number, register identifier, or persistent link where one exists.
5. Retrieval trail. Note how you found it — database search, citation chasing, or institutional site — and the date.

Where a document fails the method-description check, the appropriate action is to record it as ineligible with the reason, not to infer the methods. That single discipline keeps the evidence record honest when the document is later cited by someone else.

Publication status is a field, not an afterthought. Give every included record a status label and define the labels before you start so that two reviewers apply them the same way.

Two further fields belong beside the status label. The first is linked reports: Cochrane recommends linking multiple reports of one study, so a register entry, a conference abstract, and a journal article describing the same study should be grouped under one study identifier rather than counted as three separate bodies of evidence. The second is retrieval date, which lets a reader reconstruct what was available when you searched.

These four judgements are often collapsed into a single impression of quality. Keep them apart in the record, because they answer different questions.

• Evidence level describes the design and role of the document in your synthesis — for example, whether it is a registered study with results, a report of a completed study, or a summary without primary data.
• Applicability asks whether the population, setting, and outcome definition match your question. A report on readmission in a different care setting may be relevant but not directly applicable.
• Indirectness concerns how far the evidence departs from your question — a different but related population, a differently defined outcome window, or a comparator that does not match.
• Study limitations are the internal problems of the document itself: unclear outcome ascertainment, incomplete reporting, or an unstated analysis approach.

Recording these separately means a reader can see that a document is directly applicable yet internally limited, or well conducted yet indirect. That distinction is lost when a single quality score stands in for all four.

Suppose your search of a trial register returns a registered study in adults with the named chronic condition that lists 30-day readmission as an outcome but posts no results. Separately, an institutional report from a named agency describes a completed evaluation in a similar population, with a stated method for identifying readmissions.

In the evidence record, the register entry receives the status registered study, no results, an evidence level reflecting that no outcome data are available, and a note that it is directly applicable to your population and outcome. The institutional report receives the status institutional or agency report, an evidence level reflecting a completed study, and a limitations note describing whatever the report does not state about outcome ascertainment. If a journal article later cites the report, link the two under one study identifier rather than treating them as independent findings.

This is the point at which a library becomes useful rather than decorative. Saving both records in My Library keeps the register entry and the report together with their status labels, so that the grouping survives the writing stage. When you need to interrogate the saved set — for example, to check what the report says about how readmission was defined — Library QA lets you ask questions based on papers uploaded to My Library or added from search results.

Cochrane recommends rerunning searches before publication. For grey literature this step matters more than for journal databases, because registers update and institutional pages are revised. Rerun each source on your map, record the new date, and add any newly eligible documents with their status labels. Then close the record: state which source categories were searched, which were not, and the date of the final run.

If your work involves a longer synthesis, the Scholar Agent area is designed for extended research tasks with literature review use cases, which is where a documented grey-literature set feeds into drafting. Before any citation list leaves your hands, run the references through TrueCite, which checks BibTeX against real academic databases to flag hallucinated references. That check is a final provenance gate on the citation layer, complementing the provenance checks you applied to each document.

Set these before searching so that decisions are not made under time pressure.

Stop and record as ineligible when a document has no identifiable authoring body, no date, or no method description sufficient to judge the outcome. Stop and escalate to a second reviewer when two reviewers disagree on applicability or on the status label, or when a document is the sole source for a sub-question and its limitations are unclear. Escalate to your methods lead when a source category on your map turns out to be inaccessible, because the resulting gap must be described rather than quietly absorbed. Do not proceed to synthesis while any included record lacks a status label and a retrieval date.