Literature Searching for Health Research
A literature search is a planned and documented process for identifying evidence relevant to a research question. It establishes what is known, reveals gaps, prevents unnecessary duplication, informs methods and provides the scientific basis of a proposal.
Learning objectives
- Translate a research question into searchable concepts.
- Use keywords, subject headings and Boolean operators.
- Select appropriate online and text-based sources.
- Screen, appraise, organise and document search results.
- Recognise publication bias, predatory sources and common search errors.
Start with a focused question
For intervention questions, use PICOT. For qualitative questions, frameworks such as population, phenomenon of interest and context may be more appropriate. Break the question into two to four central concepts rather than searching the full sentence.
Worked example
Question: Among pregnant women in rural settings, do mobile-phone reminders improve antenatal attendance?
- Population: pregnant women, pregnancy, antenatal clients.
- Intervention: SMS, text message, mobile phone reminder, mHealth.
- Outcome: antenatal attendance, appointment adherence, ANC visits.
- Context: rural health services, low-resource settings.
Keywords and controlled vocabulary
Keywords are words authors may use in titles and abstracts. Controlled vocabulary consists of standard indexing terms, such as MeSH in PubMed. A sensitive search combines synonyms within each concept and then joins concepts.
(pregnant women OR pregnancy OR antenatal) AND (SMS OR "text message" OR "mobile phone" OR mHealth) AND (attendance OR adherence OR "antenatal care")
Boolean and search operators
| Operator | Function | Example |
|---|---|---|
| OR | Combines synonyms and broadens retrieval | adolescent OR teenager |
| AND | Combines different concepts and narrows retrieval | malaria AND pregnancy |
| NOT | Excludes a concept; use cautiously | jaguar NOT car |
| Quotation marks | Searches an exact phrase | “health care workers” |
| Truncation | Retrieves word variants where supported | nurs* retrieves nurse, nurses, nursing |
| Parentheses | Controls search logic | (SMS OR mHealth) AND attendance |
Where to search
Bibliographic databases
PubMed/MEDLINE, Cochrane Library and other discipline-specific databases.
Regional evidence
African Journals Online, institutional repositories and national research databases.
Guidelines and reports
WHO, ministries of health, professional bodies and recognised public-health agencies.
Text-based sources
Current textbooks, reference lists, libraries, theses and official print reports.
Search workflow
- Define the question and eligibility criteria.
- List concepts, synonyms, spelling variants and subject headings.
- Choose databases appropriate to the topic.
- Build and test the search string.
- Review early results and refine terms.
- Export citations to a reference manager and remove duplicates.
- Screen titles and abstracts, then full texts.
- Record reasons for excluding full texts where required.
- Extract and critically appraise relevant evidence.
- Update the search before final submission.
Eligibility criteria
Decide in advance which populations, designs, exposures, outcomes, languages, settings, publication types and dates are relevant. Do not exclude older landmark evidence merely to make screening easier, and do not impose language restrictions without acknowledging possible bias.
Source appraisal
| Domain | Questions |
|---|---|
| Authority | Who produced the source, and are qualifications and affiliations clear? |
| Methods | Are design, sampling, measurement and analysis appropriate? |
| Currency | Is the evidence current enough for the topic? |
| Relevance | Does it match the population, setting and research question? |
| Bias | Are conflicts, selective reporting and limitations addressed? |
| Consistency | How does it compare with the wider body of evidence? |
Grey literature
Government reports, theses, conference proceedings and programme documents may contain valuable local or negative findings not published in journals. Their methods still require critical appraisal. Inclusion can reduce publication bias but may require additional effort to locate and verify.
Scenario: improving a poor search
A student searches Google for “reasons mothers don’t come to hospital,” reads the first three results and concludes that cost is the main barrier. A defensible search would define the population and service, use synonyms across databases, include local reports, document dates and strings, screen using criteria, and compare study quality before drawing conclusions.
Documenting the search
- Database or source name and platform.
- Date searched.
- Complete search string and filters.
- Number of records retrieved.
- Duplicate removal and screening decisions.
- Full texts included and reasons for key exclusions.
Common errors
- Searching only one source or only the first results page.
- Using a single phrase without synonyms.
- Applying filters so narrowly that important studies disappear.
- Treating every published article as high quality.
- Citing secondary summaries when the primary source is available.
- Copying text instead of synthesising evidence.
- Failing to save the search strategy, citations and PDFs systematically.
Practice exercise
- Develop a search question on antibiotic misuse.
- Create concept blocks with at least four synonyms each.
- Write a Boolean search string.
- Define inclusion and exclusion criteria.
- Design a table for documenting the search and screening results.
Return to the Clinical Medicine Year 2 curriculum
References
- National Library of Medicine. PubMed User Guide.
- Cochrane Handbook for Systematic Reviews of Interventions.
- WHO. Health Research Methodology.
Educational content only. Search requirements vary by institution and review type.
