When literature indexes were published in print (like the Index Medicus for medical journal literature), subject headings were the only topic-based access points to journal articles. You identified what your subject was called (your "heading") using a specialized thesaurus for the controlled vocabulary (the Medical Subject Headings for the medical literature). You looked up your heading in the index (a book or set of books), and the relevant articles were listed under that heading.

Connecting Synonyms

With computer-based literature searching, retrieving journal article records using keywords opened access to the literature via almost any term in the title or abstract (depending on how the system is built). However, retrieving articles by keyword means that you must search with every variation of how authors describe a concept to retrieve all the relevant literature or use Boolean techniques such as truncation (e.g., child* to retrieve child, childhood, children, etc.) to retrieve variations. For example, searching for all COVID-19 articles by keyword would mean that you need to search for the terms COVID-19, SARS-CoV-2 Infection, 2019 Novel Coronavirus Infection, 2019 nCoV Infection, Coronavirus Disease 2019, etc.

But if we add uniform terminology, like Medical Subject Headings, for each significant concept to a record, you need only search the standard term (e.g., COVID-19) to retrieve the relevant results. In addition, because the thesaurus contains a wide variety of synonyms, the system can map your search term, which may not be the preferred term, to MeSH and retrieve the relevant results. This is part of the power of a controlled vocabulary (see Module 1).

Relating Concepts

Another essential role of MeSH in literature retrieval is to define relationships between concepts so that related concepts are grouped together and can be searched together. In particular, concepts that are more specific can be automatically included in a search. For example, by grouping all Arbovirus Infections (African Horse Sickness, Bluetongue, Dengue, Yellow Fever, etc.) together under the general term "Arbovirus Infections," a literature search system can take your search of arbovirus infections and automatically include terms for all the specific types of arbovirus infections (which you may not even be aware of) in your search. In PubMed, we call this use of the MeSH hierarchy to improve your search "Automatic Explosion" (see Module 1).

Automating Indexing & Enhancing Retrieval

In the past, MeSH was applied to MEDLINE records by human indexers who were subject matter experts. They would skim the entire journal article, focusing on aspects of interest like research methodology and the characteristics of the individuals being studied (e.g., species, ages, sex, etc.). Over the last several decades, human indexers were increasingly aided by automated indexing techniques. Beginning in the Spring of 2022, MEDLINE records have MeSH terminology applied completely automatically using the title and abstract of the publication as supplied by the publisher. As we undergo this transformation in how MEDLINE/PubMed is constructed, NLM and other interested researchers are looking carefully at ways to refine, improve, and make more efficient access to the biomedical literature. Consider setting up an automated e-mail alert to research in this area by running this search and selecting "Create alert" in PubMed.

There are other uses of MeSH, and much more to know about standard terminologies used in health information systems and crosswalks between them. These topics are out of scope for this course, but for more information, see the Medical Subject Headings home page and the course, A Bird’s Eye View of Health Data Standards.