Systematic reviews and other evidence synthesis projects

A few ways to find search terms:

• Run a simple search using terms you already know. Look for additional terms in their controlled vocabulary terms, title, and abstract.
• Run a simple search in PubMed using terms you already know. Look in Details in Advanced search to find what MeSH and free-text terms Automatic Term Mapping applied.
• Find published reviews on similar topics, and read methods sections which may contain exact search strategies used. Full search strategies often are provided in appendices of Cochrane reviews.
• Do you already have some articles you know should be included? Use Yale MeSH Analyzer to examine and compare their MeSH terms for ones you should include. For more information, visit this video about using the analyzer from the Cushing/Whitney Medical Library.
• Browse the MeSH database for terms to use in PubMed.
• Check the entry terms for your MeSH terms for good free-text terms
• Find the equivalent controlled vocabulary terms in another database and check their entry terms too.
• Not sure how to phrase what you're looking for? Try typing it into NLM's LitSense, which uses a form of machine learning to identify semantically similar terms.

A comprehensive search string will usually contain a mix of controlled vocabulary terms and free-text terms (sometimes called keywords) to describe the concepts.

You can use your concept table to record your terms for each database.

A controlled vocabulary is a list or database of subject terms in which all terms or phrases representing a concept are brought together. Often there is a term or phrase designated as the preferred or authorized term to be used for that concept. Controlled vocabulary terms are assigned to citations to describe the content of each article. These are terms applied by an algorithm or human indexer for the database, not by the article’s author.

Many databases assign controlled vocabulary terms to citations, but the terms and format are often specific to each database. For example, the controlled vocabulary system searchable via PubMed is MeSH, or Medical Subject Headings. More information on searching with MeSH can be found here.

Here are some examples of how terms and format may vary between databases.

Controlled vocabulary terms are often in a hierarchical "tree" with broader and narrower terms, and may appear in more than one tree. In the example below, "Epidemiology"[mh] is a narrower term to "Public Health"[mh] and is a broader term to "Molecular Epidemiology"[mh].

Terms may have features for broadening or narrowing their scope. They can be "exploded," which means that all the narrower controlled vocabulary terms are also included. To try it out, compare "foot"[Mesh] to "foot"[Mesh:noexp] in PubMed. There may be an option to add subheading, which restricts the search to that aspect of the term. For example, "Foot/injuries"[Mesh]. Review a database's instructions for controlled vocabulary to determine how best to use it in your search. You can also meet with a librarian to discuss it.

Searching with controlled vocabulary can improve the relevancy of search results. Not all citations are indexed with controlled vocabulary terms, however, so it is important to combine controlled vocabulary searches with text word searches. For example, combining controlled vocabulary and text words in PubMed would look like this: ("Enterocolitis, Necrotizing"[Mesh] or necrotizing enterocolitis[tw] OR NEC[tw])

Problematic language

Terminology used to describe race and ethnicity have evolved over time, and continues to evolve. This variability in language can make searching comprehensively for literature on race/ethnicity difficult. If you are performing a comprehensive literature search (e.g., systematic review, scoping review, historical perspective), you must include outdated terminology in your search strategy or check to see if old terminology has been included in the databases’ subject headings (e.g., illegal aliens). And other times, you will need to use terms that were common from the preferred date range of the research. This may be frustrating and discouraging.

What to do about it?

• "Addressing antiquated, non-standard, exclusionary, and potentially offensive terms in evidence syntheses and systematic searches" by Townsend W, Anderson P, Capellari E, et al. has recommendations for how to explain the inclusion of such terms while keeping them out of the body of the paper.
• Researchers who wish to further expand their knowledge can visit "Language matters: Handling tough terms in systematic searching," a webinar and collection of resources from Taubman Library.

Inclusive language

Although the language in articles from the past can't be changed, databases can and do update their controlled vocabulary, with the updated controlled vocabulary term being applied to citations previously tagged with the old term. If you notice a controlled vocabulary term that has fallen out of date, you can send the database provider feedback about it. You can also ask a librarian for assistance if it isn't clear where to send it.

An article may use terms other than the ones most familiar to you for a variety of reasons, including lack of consensus around best term and differences between disciplines. In some cases, groups of information professionals have developed "hedges" or filters--reusable search strings--to comprehensively search for articles relating to a group. Here are a few:

• MLA Latinx Caucus Latinx/Hispanic US Population Search Hedges
• Indigenous peoples whose lands are within the region of the United States
  • University of Alberta filters
  • ASTED(3S)/CHLA Indigenous people search filter [2018]
  • note: in PubMed, AIAN is more effective than AI/AN
• LGBTQIA+
  • Identifying Lesbian, Gay, Bisexual, and Transgender Search Terminology: A Systematic Review of Health Systematic Reviews (scroll to end of results section for filter)
  • LGBTQIA+  filter by Andy Hickner (written for Ovid Medline)
• Immigrants
  • Meeting a need: development and validation of PubMed search filters for immigrant populations (see also this article about the string)

Please note that with the exception of the MLA Latinx Caucus hedge and the Wafford immigrant population hedge, these hedges are not being updated with newly-recognized terminology or controlled vocabulary terms. They are a good place to start but should be evaluated and adjusted based on current knowledge.

• Yale's Library has a series of short video tutorials on searches for systematic reviews: https://library.medicine.yale.edu/tutorials/subjects/systematic-searches
• UW's Database Searching resource guides, for details on using specific databases including links to tutorials: https://guides.lib.uw.edu/hsl/database-searching

Search hedges are search strategies or filters that have been developed by expert searchers. Some have gone through a validation process to assess their sensitivity and precision. They can be incorporated into your search to save time and improve efficiency. Search hedges, also know as filters, are not the same as the general filters found in various databases. Search hedges should be used carefully as they are database specific.

• ISSG Search Filters Resource (InterTASC Information Specialists' Sub-Group)
  Collaborative venture to identify, assess and test search filters designed to retrieve research by study design or focus.
• CADTH's Database Search Filters
  These filters have been extensively tested by the Canadian Agency for Drugs and Technologies in Health Information Services Team and are routinely used in CADTH projects; however, they have not been validated against a gold standard set of articles
• McMaster University Hedges Project search filters
• Evidence-based hedges for Medline, PsycInfo, and Embase
• Flinders Filters
  Evidence-based PubMed filters
• University of Alberta filters
  Search strings for socioeconomic conditions, settings, and a wide variety of other topics
• PubMed Search filters
  For clinical queries, health services research, comparative effectiveness research, and more.
• PubMed Developing Countries Search (from UNC)
  Phrases plus countries according to World Bank low- and middle-income economies, 2019.
• PubMed Search Strategies Blog
  User-developed search strings for PubMed. Not validated.