Strata Academy

Grey Literature & Thesis Searching for Systematic Reviews

Conference abstracts, theses, regulatory documents, preprints, and when grey literature search changes review conclusions

• Pre-specify grey sources in the protocol — do not add them after seeing database results.
• Trial registries and conference abstracts often surface unpublished or negative findings.
• Thesis searching is labour-intensive; justify scope based on review question.
• Report grey literature sources and yields transparently in PRISMA Item 7.

1. What counts as grey literature

Grey literature is research output not controlled by commercial publishers: dissertations, conference proceedings, technical reports, regulatory submissions, policy documents, and organisational white papers. It sits outside standard bibliographic databases like MEDLINE.

The term is imprecise but methodologically important. Positive results publish in journals; null or inconvenient findings often remain in conference posters, student theses, or registry records. Ignoring grey sources can inflate apparent treatment effects in meta-analyses.

Grey literature is not synonymous with low quality. A well-conducted MSc thesis may contain more detail than a 300-word conference abstract. Appraise each record on its merits using the same eligibility and risk-of-bias criteria as published trials.

Preprints (manuscripts posted before peer review) occupy a grey zone. They may later publish as journal articles — deduplicate carefully. Preprint searching is increasingly relevant in fast-moving fields but requires explicit protocol justification.

• Theses and dissertations (undergraduate, MSc, PhD)
• Conference abstracts and poster presentations
• Regulatory documents (FDA, EMA, MHRA submissions where accessible)
• Government and NGO reports
• Trial registry records with results posted
• Preprints (medRxiv, bioRxiv) — pre-specify handling of updates

2. Why grey literature matters for bias

Publication bias occurs when studies with statistically significant or favourable results are more likely to appear in indexed journals. Funnel plot asymmetry in meta-analysis is one signal; searching grey sources is one mitigation.

Cochrane Handbook Chapter 4 notes that including grey literature may increase the number of included studies and shift pooled estimates toward null. The direction of change is unpredictable — which is precisely why comprehensive searching matters.

For student reviews, supervisors often ask whether you addressed unpublished work. A documented grey literature plan, even if yield is modest, demonstrates awareness of synthesis limitations.

Note: Searching grey literature without screening and appraisal standards invites inclusion of poorly reported abstracts. Apply the same eligibility criteria and risk-of-bias tools as for journal articles.

3. Where to search

Trial registries are the highest-yield grey source for intervention reviews. ClinicalTrials.gov, ISRCTN, EU Clinical Trials Register, and WHO ICTRP should be searched with terms parallel to your database strategy. Compare registry-recorded outcomes to published reports for selective reporting.

Conference proceedings: Embase indexes many conference abstracts; additionally search conference websites or proceedings repositories for specialty meetings relevant to your topic (e.g. ESC, ASCO, EULAR).

Thesis repositories: ProQuest Dissertations & Theses Global, EThOS (British Library), and institutional repositories. OpenGrey aggregates European grey literature. Google Scholar is supplementary — not reproducible as a sole source — but useful for locating reports and theses.

Regulatory and industry sources: FDA Drugs@FDA, EMA EPAR documents, and Clinical Study Reports where accessible (often limited for students). Justify scope in the protocol.

4. Building a grey literature search strategy

Mirror database search concepts where platforms support keyword search. Trial registries accept simpler strings than Ovid MEDLINE — often population and intervention terms without MeSH.

For thesis searching, use ProQuest advanced search with publication type filter 'Dissertation/Thesis' and concept blocks aligned to PICO. EThOS provides free access to many UK theses.

Set pragmatic limits. Exhaustive handsearching of every conference in a field is rarely feasible for student timelines. Pre-specify which conferences, date ranges, and repositories you will search; acknowledge what was not searched as a limitation.

Document search dates and export counts. Grey literature yields are often small but methodologically significant — one unpublished RCT can change a meta-analysis conclusion.

• Pre-specify grey sources in PROSPERO before searching
• Use consistent keywords adapted from main database blocks
• Screen thesis titles and abstracts with same dual-review process
• Retrieve full text for conference abstracts where data may exist
• Contact authors when abstracts report outcomes without effect sizes

Tip: Pair grey literature search with reference and citation searching of included studies — complementary methods, not substitutes.

5. Appraising and including grey records

Conference abstracts frequently lack methods detail. Apply the same exclusion criteria: if design, population, or intervention cannot be verified, exclude with reason 'insufficient information' rather than assuming quality.

Link duplicate reports: a conference abstract may correspond to a later journal publication or registry record. Use trial registry IDs and author names to merge records before extraction.

Risk-of-bias assessment applies equally. ROB 2 for randomised trials in theses; ROBINS-I for observational grey reports. Missing methods domains should be rated unclear or high risk, not assumed low risk because the source is unpublished.

In meta-analysis, consider sensitivity analyses excluding grey literature records to assess robustness. Report whether pooled estimates change materially.

6. Reporting grey literature in PRISMA

PRISMA 2020 Item 7 lists all information sources searched. PRISMA-S provides structured fields for supplementary searches including grey literature repositories and trial registries.

Report the number of records identified from each grey source separately before deduplication. Note how many proceeded to full-text assessment and how many were included.

If no grey literature was searched, state this explicitly as a limitation — do not imply comprehensive searching. AMSTAR 2 Item 4 asks whether the search included grey literature databases and trial registries.

In the discussion, interpret how missing unpublished data might affect conclusions. GRADE certainty may be downgraded for publication bias when funnel plot asymmetry or registry-discrepancy suggests selective reporting.

Frequently asked questions

Is grey literature searching mandatory for all systematic reviews?

Cochrane and AMSTAR 2 expect trial registry searching for intervention reviews and recommend grey literature where feasible. Scope depends on the question — a review of well-reported RCTs in major journals still benefits from registry searching; exhaustive thesis handsearching may be optional if justified.

Can I include conference abstracts in meta-analysis?

Only if they provide sufficient data for effect size extraction and meet eligibility criteria. Many abstracts lack variance measures — contact authors or exclude with documented reason. Sensitivity analyses excluding abstract-only data are good practice.

How do I find UK PhD theses?

EThOS (British Library) provides free access to many UK doctoral theses. University institutional repositories and ProQuest Dissertations & Theses Global are additional sources. Search with the same population and intervention keywords as your main strategy.

Should I search Google Scholar?

Google Scholar is useful for locating specific reports but is not reproducible enough to serve as a primary search source. Use it supplementarily to find theses or government documents identified through other means, and document any additional records found.

Interactive walkthroughs and quizzes load when JavaScript is enabled — the checklist and tables above are fully readable without it.