Premium Writing, Design & Advisory · Dubai, UAE

Blog / Academic Support

Academic Support

PubMed Search Strategy for a Literature Review: MeSH, Keywords and a Reproducible Log

A strong literature search is a documented method, not a single search box entry. Build concepts, test terms, read PubMed’s search details and preserve the exact strategy you used.

Conceptual research question, PubMed search terms and evidence records

Quick answer: how do you build a PubMed search?

Translate the research question into a small number of searchable concepts. For each concept, list relevant text words, spelling variants and Medical Subject Headings (MeSH) where appropriate. Combine alternatives within a concept with OR, then combine concepts with AND. Run the search in PubMed, inspect the Search Details and results for known relevant papers, adjust transparently, and record the exact query, search date, result count and any limits. The National Library of Medicine’s PubMed User Guide documents the Advanced Search Builder, field tags, Search Details and history.

This guide is for researchers preparing a health or biomedical literature review. PubMed is valuable but may not cover every discipline or every source needed for a systematic review. Follow your protocol, university instructions and, for formal systematic reviews, specialist librarian advice where available. Labeeb can help structure and document the review while you retain authorship and eligibility decisions.

Begin with concepts, not a paragraph

Suppose a fictional review asks: “What studies examine telehealth follow-up for adults with type 2 diabetes in Gulf countries?” This is an illustration of search construction, not a claim that the search retrieves a complete evidence set. Identify the core concepts—condition, intervention and geography—then decide whether requiring every concept in the first query is too restrictive. A narrow geography term can miss a paper whose abstract does not name its country, so test a broad condition-and-intervention search before adding location terms.

Concept Candidate text words What to verify
Type 2 diabetes type 2 diabetes; T2DM Relevant MeSH heading and spelling variants
Telehealth follow-up telehealth; telemedicine; remote consultation Whether each term captures the intended intervention
GCC setting United Arab Emirates; Saudi Arabia; Qatar; Kuwait; Oman; Bahrain Whether affiliation or abstract fields consistently describe study location

Do not paste every synonym found online. Test whether each candidate adds relevant studies. In a scoping search, broader recall may be appropriate; a tightly defined review may use different limits. Keep a note of terms tested and rejected so the final method is explainable.

Use MeSH and text words for different jobs

MeSH terms are controlled subject headings assigned to indexed MEDLINE records. They can bring together papers that use different author wording. Text words search what authors wrote in titles and abstracts and help find newer or not-yet-indexed records. Neither route replaces the other in a carefully planned search. Verify a proposed heading in the NLM MeSH Browser, inspect its scope note and entry terms, and check whether the heading actually reflects the concept. The PubMed guide explains how field tags and automatic term mapping affect results.

Quoted phrases, truncation and field tags can alter PubMed’s automatic mapping. Use them deliberately and inspect Search Details rather than assuming PubMed searched exactly what you typed. For example, a text-word field such as [tiab] narrows to title and abstract; a MeSH field identifies the controlled heading. The precise query should be checked in PubMed before it becomes a reported search strategy.

Build the Boolean logic visibly

OR adds alternatives

Within one concept, group terms such as telehealth OR telemedicine. Brackets keep the relationship clear.

AND combines concepts

Intersect the condition and intervention groups. Adding more concepts can reduce retrieval, sometimes too much.

NOT needs restraint

Exclusions can remove relevant records unexpectedly. Prefer eligibility screening when the exclusion is nuanced.

An illustrative starting structure is (diabetes text words OR verified MeSH term) AND (telehealth text words OR verified MeSH term). This is a logic diagram, not a finished PubMed query: replace placeholders with verified tags and terms, run it, then save the exact final syntax from PubMed. A search strategy for one review question cannot simply be copied into another.

PubMed’s Advanced Search History lets you build concept sets separately and combine them. Search Details show how PubMed translated a query and flag some syntax problems. Use that feedback before relying on a result count. The NLM instructions describe both functions.

Test the search against evidence, not just a count

A huge result set may contain little relevant evidence; a small set may be missing key studies. Identify a few known relevant articles from your preliminary reading and ask whether the search retrieves them. If not, investigate which concept or field restriction excluded them. Review several early and later results for relevance and record why you changed terms. Search-date and result-count changes do not, by themselves, show improved quality.

Beware of adding publication-date, language, article-type or age filters simply because the initial result list feels long. Every limit needs a reason tied to eligibility criteria; filters can have imperfect sensitivity. Formal reviews may require more than PubMed, including other bibliographic databases and supplementary methods. The PRISMA-S reporting extension covers transparent documentation of literature searches.

Keep a reproducible search log

Record Why it matters Example entry
Database/platform Interfaces differ PubMed, NLM interface
Date and time zone Results change Search run on the actual date
Exact query Enables replication Copy the final syntax, including tags
Limits Shows restrictions List every filter or state none
Retrieved count Supports flow reconciliation Record the count shown at execution
Changes and reason Separates testing from final search Added synonym after a known article was missed

Export references and preserve the file or citation-manager record used for screening. If multiple databases are searched, keep the strategy for each database rather than saying only that “keywords were searched.” Our literature review search-log checklist can help organize those records. For screening decisions and flow counts, use a separate workflow; the search log does not replace an eligibility protocol.

What should the methods section report?

  1. State the review question and eligibility scope that guided the search.
  2. Name the databases and interfaces, the final search dates and any updates.
  3. Give the complete strategy for each database in the manuscript or appendix, with all field tags and limits.
  4. Explain any language, date or study-design restrictions.
  5. Describe how records were exported, deduplicated and screened, without inventing counts.
  6. If a librarian or information specialist reviewed the strategy, acknowledge the role accurately.

For a systematic review, PRISMA 2020 and PRISMA-S are reporting references, not substitutes for a sound search design. Requirements differ for narrative, scoping and systematic reviews; use the standard appropriate to the work.

Common mistakes and better decisions

One natural-language query

Better: identify concepts, synonyms and field choices, then inspect PubMed’s translation.

MeSH only

Better: consider both indexed headings and text words so newer records are not silently excluded.

Filters without rationale

Better: connect every restriction to eligibility and test what it removes.

A carefully documented search can still miss research. Explain scope and limitations truthfully. If your project concerns nonmedical business or education evidence, PubMed may be a supporting database rather than the primary one.

Sources and next step

Illustrative search concepts above must be adapted and tested for your own question. No specific retrieval count or study result is claimed.

Need a search strategy you can explain and defend?

Share your question, eligibility criteria, databases and current search log. Labeeb can scope ethical Literature Review support and help you document a transparent method while you retain authorship and screening decisions.

Discuss Literature Review support