Medical Subject Headings (MeSH) constitute one of the most extensively developed controlled vocabularies in biomedical science. As a practising scientific curator, I use MeSH mapping daily to translate the natural language of published research into standardised, searchable terminology. This guide explains the principles, structure, and practical application of MeSH in biomedical literature indexing.
The Architecture of MeSH
MeSH is a hierarchically structured thesaurus developed and maintained by the United States National Library of Medicine (NLM). Its hierarchical organisation, termed the MeSH Tree Structure, arranges subject headings from broad conceptual categories at the top of each tree to increasingly specific subcategories at lower levels. The hierarchy currently encompasses over 30,000 descriptors organised across 16 broad categories including anatomy, organisms, diseases, chemicals and drugs, analytical techniques, and health care.
Each MeSH descriptor may carry one or more subheadings, also termed qualifiers, which specify the conceptual aspect of a subject being discussed. For example, the descriptor Staphylococcus aureus may be combined with the subheading drug effects, pathogenicity, or genetics depending on the specific focus of the indexed article. This descriptor-subheading combination system enables precise and multidimensional characterisation of complex biomedical literature.
Why Controlled Vocabulary Matters
The fundamental problem that MeSH and similar controlled vocabularies address is terminological inconsistency in scientific literature. Authors may refer to the same concept using multiple different terms — myocardial infarction, heart attack, cardiac infarction, and coronary thrombosis may all appear in different publications to describe the same clinical entity. Without a controlled vocabulary that maps all these variants to a single standardised heading, literature searches would fail to retrieve a significant proportion of relevant publications.
Controlled vocabulary indexing ensures that a researcher searching for publications on myocardial infarction retrieves all indexed articles on that topic regardless of the specific terminology used by individual authors. This recall completeness is particularly critical in systematic reviews and meta-analyses, where failure to retrieve relevant literature directly compromises the validity of research conclusions.
MeSH Major and Minor Headings
A critical distinction in MeSH indexing practice is the designation of major versus minor headings. Major MeSH headings identify the primary subjects of an article — the concepts that the article is fundamentally about. Minor headings identify secondary or incidental subjects that are mentioned but do not constitute the central focus of the publication.
In database searching, restricting retrieval to major MeSH headings substantially increases search precision by filtering out articles in which the subject of interest is only peripherally mentioned. Conversely, including minor headings increases recall at the cost of precision. Professional indexers must exercise careful scientific judgement when determining which subjects warrant major heading designation, requiring a thorough reading and comprehensive understanding of each indexed article.
EMTREE vs MeSH: Key Differences
While MeSH is the controlled vocabulary used by PubMed and MEDLINE, EMBASE employs its own proprietary thesaurus known as EMTREE. EMTREE contains over 80,000 preferred terms, making it considerably more granular than MeSH, particularly in the domains of pharmacology and drug terminology. EMTREE maps drug names to their preferred terms with considerably greater specificity, distinguishing between drug classes, individual compounds, metabolites, and combination products at a level of detail that MeSH does not achieve.
For systematic reviewers searching both PubMed and EMBASE, developing parallel search strategies using both MeSH and EMTREE is essential. Studies comparing database coverage have consistently demonstrated that EMBASE and MEDLINE have substantial non-overlapping content, particularly for pharmacological and European literature, making dual-database searching a methodological necessity for comprehensive systematic reviews.
Practical MeSH Mapping Workflow
Effective MeSH mapping requires a systematic approach to each indexed article. The workflow I follow in professional practice begins with a comprehensive reading of the full article text — not merely the abstract — to identify all scientifically significant concepts addressed by the authors. Each identified concept is then mapped to the most specific applicable MeSH descriptor, using the MeSH Browser provided by NLM to navigate the hierarchy and identify appropriate terms.
When no single MeSH descriptor adequately captures a specific concept, curators must select the most closely related available term and note the limitation. This situation arises frequently with newly described conditions, emerging pathogens, or novel therapeutic approaches that have not yet been incorporated into the controlled vocabulary. In such cases, supplementary concept records in MeSH or free-text keywords serve as interim solutions pending vocabulary update.
Key Takeaways
- MeSH contains over 30,000 descriptors organised in a hierarchical tree structure across 16 broad categories
- Controlled vocabulary overcomes terminological inconsistency in scientific literature to ensure comprehensive retrieval
- Major MeSH headings identify primary article subjects; minor headings identify secondary or incidental topics
- EMTREE is more granular than MeSH particularly in pharmacology with over 80,000 preferred terms
- Effective MeSH mapping requires full article reading and systematic concept identification before descriptor selection