Part 3 · Module 3
Searching Bibliographic Databases
Plan, construct, translate, test and document reproducible searches across bibliographic databases used in health and medical research.
Learning outcomes
After completing this lecture, you should be able to:
- select bibliographic databases according to topic, discipline and review purpose
- construct concept blocks using Boolean operators, phrases, truncation, field searching and subject headings
- translate rather than copy a search strategy across databases
- test a search for sensitivity, specificity and conceptual completeness
- document database searches so that another researcher can understand and reproduce them
Overview
Bibliographic databases organise records of scholarly literature and provide search functions that go beyond a general web search. A defensible health-research search combines appropriate databases, conceptually complete terminology, database-specific syntax and transparent documentation.
No single database covers all health evidence. Database selection depends on the topic, eligible designs, professional disciplines, geographic context and review purpose. A search that is excellent in MEDLINE may still miss nursing, allied-health, psychological, social-science, regional or conference literature indexed elsewhere.
Why this matters
Search errors can change the evidence base before appraisal or analysis begins. Missing studies may create distorted estimates, conceal harms, overlook under-represented populations or produce a false research gap. An excessively broad search can also consume large screening resources and increase inconsistent decisions.
Systematic and scoping reviews require a transparent, reproducible process for identifying evidence. Cochrane’s current guidance treats planning, database searching, supplementary searching, record management and study selection as connected parts of one retrieval process [1]. PRISMA-S provides specific guidance for reporting literature searches [2].
Selecting databases
Database choice should be justified by expected coverage rather than habit.
Common health and multidisciplinary sources
- MEDLINE, commonly searched through PubMed or another interface, is central for biomedical and clinical literature.
- Embase provides broad biomedical and pharmacological coverage and includes substantial conference indexing.
- CINAHL is important for nursing and allied-health literature.
- PsycINFO is relevant to behavioural, psychological and mental-health questions.
- Scopus and Web of Science provide multidisciplinary citation indexes useful for broad retrieval and citation searching.
- Cochrane CENTRAL is important for reports of controlled trials.
- Regional databases may improve retrieval of locally published evidence and non-English-language journals.
PubMed is a search platform that includes MEDLINE records as well as additional biomedical citations; PubMed and MEDLINE should not be treated as perfectly interchangeable names.
Select databases by evidence domain
Ask:
- Which clinical, public-health or professional disciplines study the question?
- Which study designs and publication types are eligible?
- Could important evidence be published in regional or non-medical journals?
- Are conference records, dissertations or trials important?
- Which databases are accessible to the team?
- Does an information specialist recommend additional sources?
A limited search should be described as limited, not presented as comprehensive.
Building concept blocks
Within each concept, combine synonyms with OR. Combine different concepts with AND.
(gestational diabetes OR diabetes in pregnancy OR GDM)
AND
(postpartum OR post-partum OR postnatal OR puerperium)
AND
(glucose test* OR oral glucose tolerance test* OR diabetes screen* OR metabolic follow-up)This is a conceptual illustration, not a database-ready final strategy. Exact syntax, phrase marks, truncation symbols and field tags vary by platform.
Search techniques
Phrase searching
Quotation marks or proximity syntax can keep words together, such as “gestational diabetes”. Phrase searching may improve precision but can miss variants or records where words are separated. Check how the database interprets quotation marks.
Truncation and wildcards
A truncation symbol retrieves multiple word endings. For example, postnatal* may retrieve postnatal and postnatally. Over-truncation can retrieve unrelated words. Wildcards may capture spelling variants, but symbols and limits differ by database.
Field searching
Searches can be limited to titles, abstracts, keywords or other fields. Title-only searching is precise but often too restrictive for evidence synthesis. Title-and-abstract searching is commonly combined with subject headings.
Controlled vocabulary
MEDLINE uses Medical Subject Headings (MeSH), maintained by the US National Library of Medicine; other databases use different thesauri [3]. Subject headings may be “exploded” to include narrower terms or “focused” to indicate a major topic. Searchers should inspect the hierarchy, scope notes, entry terms and indexing date.
Proximity operators
Some platforms allow terms to occur within a specified number of words, which can capture flexible phrases without retrieving every record containing the words anywhere. Proximity syntax must be translated between platforms.
Filters
Validated search filters may help retrieve particular study designs, but unvalidated or overly restrictive filters can miss eligible studies. Database interface filters for “free full text”, language or article type should not be used casually.
Concepts translated into a PubMed-style illustration
illustrative("Diabetes, Gestational"[Mesh] OR gestational diabet*[tiab] OR GDM[tiab])
AND
("Postpartum Period"[Mesh] OR postpartum[tiab] OR post-partum[tiab] OR postnatal[tiab])
AND
("Glucose Tolerance Test"[Mesh] OR glucose test*[tiab] OR oral glucose tolerance test*[tiab] OR diabetes screen*[tiab] OR follow-up test*[tiab])Note: This strategy is illustrative. It must be tested, refined and translated for each database.
Translating across databases
Do not paste one strategy into several interfaces and assume equivalence. Translation requires checking:
- subject headings and hierarchy;
- title and abstract field codes;
- phrase and proximity syntax;
- truncation and wildcard rules;
- line-combination commands;
- limits and publication-type indexing; and
- how the platform maps untagged terms.
A master concept table helps preserve semantic consistency while allowing technical translation. Structured approaches to search development emphasise concept analysis, term harvesting, optimisation and database-specific translation [4].
Testing a search
Search development is iterative. Useful tests include:
Known-item testing
Confirm that several clearly eligible and diverse “seed” studies are retrieved. Failure indicates a term, indexing or syntax problem. Retrieval of known studies does not prove completeness.
Term-contribution testing
Inspect which terms retrieve unique relevant records and which mainly add noise. Do not delete a low-yield term solely because it retrieves few records; it may represent an important subgroup or older terminology.
Concept-block testing
Run blocks separately and in combinations to identify the block that causes unexpected losses. Outcomes, settings and design terms commonly reduce retrieval.
Precision review
Screen a sample of results to understand dominant irrelevant meanings. Precision can be improved through field restrictions, phrase or proximity searching and exclusion of clearly unrelated meanings. Broad NOT statements should be used cautiously because they can remove relevant records.
Peer review and information-specialist input
Complex searches benefit from collaboration with a librarian or information specialist. The PRESS guideline provides a structured approach to peer review of electronic search strategies, including translation of the question, Boolean and proximity operators, subject headings, text words, spelling and syntax [5]. Peer review should occur before the final search where possible.
Sensitivity, precision and search purposeMMed extension
A systematic review that aims to identify all eligible studies usually prioritises sensitivity, accepting substantial screening. A rapid evidence scan may accept a more precise search with an explicit risk of missing evidence. The correct balance is determined by the review purpose, not by a universal target number of records.
Recording the search
For each database, retain:
- database and platform name;
- coverage dates if relevant;
- complete line-by-line strategy;
- all limits and filters;
- date searched;
- number of records retrieved;
- person who ran the search;
- version or update notes; and
- export filename and format.
The platform matters because the same database can behave differently across interfaces. Search documentation should be sufficient for another researcher to understand what was done and to update the search. Current Campbell guidance also emphasises transparent source selection, search translation and supplementary methods across evidence-synthesis contexts [6].
Before running the final database searches
- Confirm the review question and eligibility criteria.
- Finalise the concept and synonym table.
- Select databases based on topic coverage.
- Test known eligible studies.
- Translate syntax and controlled vocabulary for every database.
- Consider peer review of the principal strategy.
- Predefine limits and justify them.
- Create a search log and file-naming convention.
- Save the exact strategy immediately after execution.
Worked example
Repairing the initial postpartum search
Initial search: gestational diabetes AND postpartum glucose test barriers Malaysia
Problems identified: It requires the word “barriers”, restricts geography without justification, lacks synonyms and subject headings, and uses only one database.
Revised plan:
- Define three core concept blocks: gestational diabetes, postpartum period and glucose follow-up.
- Combine free text with relevant subject headings.
- Search MEDLINE, Embase and CINAHL because the question spans clinical, service and nursing literature.
- Search barrier and facilitator information during screening rather than requiring those words.
- Retain Malaysia as a subgroup or context for interpretation, not as a mandatory search concept, unless the review is explicitly Malaysia-only.
- Test the strategy against known eligible studies and have the MEDLINE strategy peer reviewed.
Resulting interpretation: The number of retrieved records now reflects a more defensible search process. It still cannot establish completeness without supplementary searching and documented screening.
Appraising a database search
Database-search appraisal questions
- Are the databases appropriate and sufficient for the evidence domain?
- Is the platform identified for each database?
- Does the strategy combine subject headings and free text?
- Are Boolean grouping, phrases, truncation and fields used correctly?
- Has the strategy been translated rather than copied?
- Are filters and restrictions justified?
- Were known studies or other validation checks used?
- Are the full strategies, dates and result counts reported?
Suggested protocol or reporting wording
Common mistakes
Knowledge check
Why should a MEDLINE search not be pasted unchanged into Embase?
Answer and reasoning
The databases use different controlled vocabularies, field codes, proximity rules and platform syntax. The underlying concepts should remain consistent, but the technical strategy must be translated and tested.
What is the role of OR within a concept block?
Answer and reasoning
OR retrieves records containing any of the linked synonyms or related terms. It broadens retrieval within that concept. AND is then used to intersect different concepts.
A search retrieves a known eligible trial only when the outcome block is removed. What should the team consider?
Answer and reasoning
The outcome may be absent or described differently in searchable fields. The team should consider omitting the outcome block and applying the outcome criterion during screening, while checking the effect on the overall result set.
What information is needed to make a database search reproducible?
Answer and reasoning
At minimum, report the database and platform, complete search strategy, limits, coverage where relevant, search date and result count. Retaining line-by-line strategies and export records supports updating and audit.
Key take-home messages
Key take-home messages
- Database choice should reflect the evidence domain, not habit.
- Combine synonyms with OR and intersect concepts with AND.
- Use free text and database-specific controlled vocabulary together.
- Translate and test strategies separately for each database.
- Save complete strategies, dates, platforms, limits and result counts.
References
- Lefebvre C, Glanville J, Briscoe S, Featherstone R, Littlewood A, Metzendorf M-I, et al. Technical Supplement to Chapter 4: Searching for and selecting studies. In: Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al., editors. Cochrane Handbook for Systematic Reviews of Interventions. Version 6.5. Cochrane; 2024.
- Rethlefsen ML, Kirtley S, Waffenschmidt S, Ayala AP, Moher D, Page MJ, et al. PRISMA-S: an extension to the PRISMA Statement for Reporting Literature Searches in Systematic Reviews. Syst Rev. 2021;10:39. https://doi.org/10.1186/s13643-020-01542-z
- National Library of Medicine. Medical Subject Headings. Bethesda (MD): National Library of Medicine. https://www.nlm.nih.gov/mesh/
- Bramer WM, de Jonge GB, Rethlefsen ML, Mast F, Kleijnen J. A systematic approach to searching: an efficient and complete method to develop literature searches. J Med Libr Assoc. 2018;106(4):531–541. https://doi.org/10.5195/jmla.2018.283
- McGowan J, Sampson M, Salzwedel DM, Cogo E, Foerster V, Lefebvre C. PRESS Peer Review of Electronic Search Strategies: 2015 Guideline Statement. J Clin Epidemiol. 2016;75:40–46. https://doi.org/10.1016/j.jclinepi.2016.01.021
- MacDonald H, Comer C, Foster M, Labelle PR, Marsalis S, Nyhan K, et al. Searching for studies: a guide to information retrieval for Campbell systematic reviews. Campbell Syst Rev. 2024;20(3):e1433. https://doi.org/10.1002/cl2.1433
These materials support learning and research planning. They require adaptation to the study context and do not replace methodological, statistical, ethical, or regulatory consultation.