MeSH Terms and Other Subject Headings: Do You Still Need Keywords?

If MeSH Already Organizes the Literature, Why Search Keywords Too?

You have learned that PubMed and MEDLINE use Medical Subject Headings (MeSH) to organize biomedical literature by subject. Instead of trying to anticipate every word an author might use, you can search a standardized concept and retrieve records indexed under that concept.

That sounds like it should solve one of the hardest problems in literature searching. If heart attack, myocardial infarction, and related expressions can be brought together through standardized indexing, why spend time constructing keyword searches as well?

The reason is that MeSH and other subject headings solve only part of the retrieval problem. They can provide powerful access to indexed concepts, but keywords reach the literature in a different way. Depending exclusively on either one can leave relevant research outside your search.

Yes, You Usually Still Need Keywords

Why Subject Headings and Keywords Find Different Literature

MeSH Is a Controlled Vocabulary, Not Simply a Better List of Keywords

Medical Subject Headings, or MeSH, is the controlled-vocabulary thesaurus maintained by the U.S. National Library of Medicine (NLM). It is used to represent biomedical concepts consistently for indexing and retrieval.

For MEDLINE, MeSH terms describe the subject content of journal citations. The vocabulary is hierarchical, so broader concepts can have more specific concepts beneath them. PubMed provides a MeSH database where searchers can inspect descriptors, entry terms, scope notes, subheadings, hierarchical relationships, and other information before adding a term to a search.

This makes MeSH fundamentally different from ordinary free-text searching. A MeSH search asks, in effect, whether a record has been indexed as being about a standardized concept. A keyword search asks whether specified words or phrases can be found in the fields you choose to search.

If that distinction is unfamiliar, the underlying difference between keywords and subject headings is worth understanding before constructing a complex search.

MeSH Can Bring Different Expressions of a Concept Together

Authors do not necessarily use the vocabulary that searchers expect. A concept may have a technical term, an everyday name, abbreviations, historical terminology, spelling variants, or several expressions used by different research communities.

MeSH helps address this problem by providing standardized conceptual access. NLM explains that applying MeSH allows MEDLINE articles to be indexed uniformly by subject regardless of the particular words used by the author.

Consider breast cancer. Researchers may encounter expressions such as breast cancer, breast tumor, or breast neoplasm. NLM uses this type of example when explaining how MeSH provides consistency across different ways of expressing a concept.

This is a major advantage. Without controlled vocabulary, you may need to anticipate many of the different terms researchers use for the same concept. Subject indexing gives you another route to those records.

But Not Every PubMed Record Is a MEDLINE Record Indexed With MeSH

This is one of the most important reasons not to rely exclusively on MeSH.

PubMed is broader than MEDLINE. NLM explains that PubMed includes MEDLINE citations as well as additional content, including citations from PubMed Central and the NCBI Bookshelf. NLM's MeSH training materials specifically note that only citations included in MEDLINE receive MEDLINE MeSH indexing.

Consequently, a search restricted to MeSH terms is not equivalent to searching all of PubMed for that concept. Keyword searching can provide access to relevant records through their textual fields even when the records are not retrievable through MEDLINE subject indexing.

Recent Records Create Another Reason to Keep Keywords

Indexing and bibliographic availability are not the same process. A citation can become searchable before all forms of subject indexing relevant to a particular search are available. In addition, PubMed contains records outside the MEDLINE subset.

Keywords therefore provide an important route to recent literature because they can search textual information already present in the record rather than depending solely on the presence of a MeSH heading.

The practical implication is straightforward: if finding recent research matters, do not design a PubMed search that depends entirely on MeSH.

New Concepts May Appear Before the Vocabulary Fully Represents Them

Controlled vocabularies are maintained and updated, but research terminology can change quickly. New technologies, diseases, interventions, social phenomena, and methodological concepts may enter scholarly discourse before a dedicated subject heading exists.

MeSH is updated to reflect developments in biomedical knowledge and terminology. Even so, the vocabulary cannot function as a real-time transcript of every expression appearing in newly published research.

Keywords are particularly useful in this situation because you can search emerging terminology as soon as it appears in searchable bibliographic fields. When terminology itself is evolving, you may also need to consider how the concept has changed names over time.

A Subject Heading May Be Broader Than the Concept You Actually Need

Finding a related MeSH term does not necessarily mean you have found an exact representation of your research concept.

Some concepts are too specific, too new, or too context-dependent to have their own descriptor. A relevant article may instead be represented through a broader heading, several headings used together, or a descriptor combined with a subheading.

NLM explains that complex subjects may be represented through combinations of MeSH descriptors or through descriptors combined with qualifiers. This makes MeSH expressive, but it also means that a single heading may not reproduce every conceptual distinction in your research question.

Keywords can preserve language-specific nuances that disappear when a concept is represented at a broader indexing level.

Keywords Also Let You Search the Authors' Own Language

Sometimes the wording itself matters.

Suppose you are interested specifically in literature that describes an intervention as digital detox. A broader controlled-vocabulary concept may retrieve studies about related behaviors, but you may also want records whose authors explicitly use the phrase digital detox.

Free-text searching lets you target that language. Depending on the database and search field, you might search titles, abstracts, author keywords, or combinations of those fields. The choice of where the keyword is searched can therefore matter almost as much as the keyword itself.

PubMed Complicates the Distinction Through Automatic Term Mapping

There is an additional wrinkle: typing an ordinary keyword into PubMed does not necessarily produce a simple literal keyword search.

PubMed uses Automatic Term Mapping (ATM) for many untagged searches. According to PubMed's current help documentation, the system may modify or add terms to improve retrieval, including MeSH terms, spelling variants, singular and plural forms, and synonyms. You can inspect the Search Details in Advanced Search to see how PubMed interpreted a query.

This means an ordinary PubMed query may already involve both textual retrieval and controlled-vocabulary mapping behind the scenes.

Explicit Title/Abstract Searching Behaves Differently

When constructing a reproducible PubMed strategy, researchers often use explicit field tags. The Title/Abstract field tag, [tiab], searches citation titles, abstracts, other abstracts, collection titles, and author keywords as specified in PubMed's documentation. It does not search the MeSH Terms field.

That separation is useful because it lets you deliberately construct complementary search routes. One portion can search a concept through MeSH, while another searches selected free-text expressions through title and abstract-related fields.

A simplified conceptual structure might look like this:

The precise syntax depends on the database. The logic, however, is broadly useful: subject headings and keywords for the same concept are commonly alternatives, not separate requirements.

Using AND Between a MeSH Term and Its Keyword Equivalent Can Be Too Restrictive

Suppose Myocardial Infarction is your MeSH term and heart attack is one of your free-text alternatives. If both represent the same concept, requiring every result to satisfy both routes may defeat the reason for using complementary retrieval methods.

A record indexed with the relevant MeSH term might not contain your chosen phrase in the fields searched. Another relevant record might contain the phrase but lack that MeSH indexing. Requiring both could exclude either record.

This is why understanding the difference between AND and OR matters when combining subject headings with keywords.

Other Databases Have the Same General Problem but Different Vocabularies

MeSH is not the universal vocabulary of health research databases. Embase, for example, uses Emtree, its own hierarchically structured controlled vocabulary. Elsevier notes that Emtree and MeSH overlap substantially but differ in their structure and coverage.

Other disciplinary databases may use their own thesauri or subject-heading systems. Consequently, the underlying principle travels better than the actual search string: combine appropriate controlled-vocabulary retrieval with suitable free-text retrieval where both are useful.

When you move the strategy elsewhere, the subject headings, field codes, proximity operators, truncation rules, and other syntax may need to change. A search that works well in PubMed should therefore be translated for the destination database, not merely pasted into it.

Using Both Does Not Mean Adding Every Term You Can Find

A combined search can become unnecessarily large if every possible synonym, spelling variation, subject heading, narrower concept, acronym, and phrase is added without testing whether it contributes useful retrieval.

The objective is coverage, not maximal string length. A twenty-line concept block is not automatically more rigorous than a six-line one. Database searches, like literature reviews themselves, do not receive extra methodological credit for intimidating typography.

Terms should earn their place by representing the intended concept and contributing plausibly useful retrieval. This becomes particularly important as you decide when to stop adding search terms.

What Happens When You Search a Concept Through Both Routes?

A simplified PubMed concept block might therefore combine the MeSH descriptor with relevant Title/Abstract terms. The exact search should be developed from the research question and tested rather than copied mechanically from this example.

Notice what the strategy is doing. It is not asking MeSH and keywords to confirm each other. It is giving a relevant article more than one legitimate way to enter the result set.

Common Mistakes When Combining MeSH Terms and Keywords

How Should You Decide Whether to Add Keywords to Subject Headings?

For most substantial literature searches, the useful question is not whether you should choose MeSH or keywords. Ask what would be missed if you relied exclusively on one route.

A useful workflow is to search the concept initially, inspect highly relevant records, examine their indexing, consult the controlled vocabulary, collect free-text variants, and test the combined strategy. Relevant articles can teach you both how authors describe the concept and how the database represents it.

If the combined search becomes unmanageable, do not respond by indiscriminately adding more restrictions. Determine which terms are generating noise and which concept is responsible. Otherwise, an attempt to make the search cleaner can quietly change what the search is capable of finding.

Before Finalizing a MeSH and Keyword Search, Check Both Retrieval Routes

Frequently Asked Questions About MeSH Terms and Keywords

MeSH and Keywords Are Complementary, Not Competing Search Methods

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.