Responsible Literature Searching Guide

Database Structure and Features

Databases – including the Library’s online catalog, as well as PubMed/Ovid MEDLINE, Ovid Embase, CINAHL, APA PsycINFO, Scopus, and Web of Science – are electronic filing systems of retrievable citation records. The databases may include books, journals, journal articles, book chapters, multimedia, theses, dissertations, and meeting abstracts. As part of identifying which sources to search, it is important to know what resources are included or excluded in the databases. For example, some databases do not include conference proceedings, book chapters, or book reviews.

Librarian indexers review materials and identify and assign relevant information to searchable fields. While the organization of the citation record may vary across the different databases, the record typically includes three parts:

• Bibliographic information – i.e., author, title, journal (or source) information
• Database producer or vendor information – i.e., subject headings or controlled vocabularies, keywords, descriptors (e.g., publication type, language, age, etc.), abstract, publisher, place of publication
• Processing information – data used to control the display of information or construction of the database (i.e., search history, entry dates, links to full text, etc.)

Most databases include a basic and advanced search interface.

• The basic interface is similar to a Google search where the searcher simply types in a string of keywords. The database’s search algorithm returns citations in seconds.
• The advanced search interface provides the searcher with the flexibility to use the database’s automatic mapping software (if available), search by keyword or natural language and by subject heading, and combine search results with Boolean operators (AND, OR, or NOT).

Understanding the features of the database’s search interface(s) – default search fields (i.e., fields that are automatically searched when keywords or phrases are typed into the search box); available searchable fields; differences between subject headings, keywords, and search techniques (phrase searching, truncation, wildcards) – and knowing when to use one or more of these features will result in a more systematic and effective search. Regardless of the interface, the searcher needs to determine whether the results answer the question.

(Health Sciences Library McMaster University, 2019; Jankowski, 2008; Patrick & Munro, 2004; U.S. National Library of Medicine, 2019; Walker & Janes, 1999; Wessel, 2019)

Database Search Approaches

Building blocks, pearl growing, and searching by parts of a citation are search methods which can be used in combination or separately. The default search interfaces of the Library’s databases vary between advanced (i.e., Ovid MEDLINE, Ovid Embase, APA PsycINFO, CINAHL) and basic (i.e., PubMed, Scopus, Web of Science). Personal preferences, database features, and search questions will influence the selected search method(s).

Building Blocks

1. Refer to the concepts, related keywords, and subject headings identified in the PICO(M) question. These words are the building blocks: use the words as the starting point.
2. Search each concept or keyword separately. If available, use the database’s automatic term mapping feature.
3. Combine the building blocks with the appropriate logical connectors (i.e., Boolean operators) to create a set of citations to review.

Searching one concept at a time – rather than searching multiple concepts simultaneously – reduces connector errors. Additionally, this approach allows the searcher to recombine concepts or try new combinations without having to retype the search string.

Pearl Growing

The pearl growing technique is an excellent way to begin a search to either identify subject headings or extend a search when few results are retrieved.

1. Begin the search with a single known reference (e.g., cited in a specific journal article) or conduct a cursory search to identify a relevant citation.
2. Examine the known reference or relevant citation (i.e., the pearl) for subject headings, author-assigned keywords, recurring words, etc., which can be used to improve the search.
3. Use the terms identified in step 2 to retrieve additional citations. Search each concept or keyword separately. Continue the process.

The "Similar Articles/Cited By" features in PubMed or "Find Similar/Find Citing Articles" features in Ovid MEDLINE are examples of pearl growing algorithms in the database.

Why is cited reference searching important? Scientific research is built on the works of others. The "Cited By/Find Citing Articles" or equivalent feature in a database leads to additional relevant and more recently published references that may be examined for suggestions of other search terms. For more information, review the database’s Help section.

(Booth, 2012; Health Sciences Library McMaster University, 2019; Jankowski, 2008; Welch Medical Library, 2019)

Searching by Parts of a Citation

To search by parts of a citation, the searcher applies their basic understanding of the database’s structure and the citation record components. Using specific fields (i.e., author, title, abstract, journal title, institution, etc.) increases the precision since the scope of the search is restricted by the selected field(s). While the available fields vary by database, search fields can be restricted by using dropdown menus (when available) or field tag qualifiers. Search each field separately. As databases and their interfaces are continually updated, review the database's Help section.

Select field tags for PubMed and Ovid Medline follow.

More Information

Refer to Appendix 3 - Database Comparison Chart to review select field tags for commonly searched databases.

(Jankowski, 2008; Patrick & Munro, 2004; U.S. National Library of Medicine, 2019; Walker & Janes, 1999; Wessel, 2019)

Boolean Logic

The next step is to use connectors to link the results generated from the building block, pearl growing, field tag, or other search methods. Boolean logic (named after the mathematician George Boole) is a system of showing relationships between sets. Commonly used terms are OR, AND, and NOT; these operators often must be capitalized when combining search sets manually.

The shaded areas in the following table illustrate the search results retrieved using each operator and how they differ from each other.

(Jankowski, 2008; Walker & Janes, 1999; Welch Medical Library, 2019)

Order of Precedence

When using Boolean operators, word order in search statements is important. If more than one Boolean operator is used in a search statement, the database algorithm interprets the search according to the database’s designated order of precedence of Boolean operators (i.e., the order in which connectors are processed when entered into the database search interface) or from left to right. As the order of precedence varies, it is important to review the database's Help section.

For example, a database may process the search query flu OR vaccination AND elderly in the following ways:

• Order of precedence: if the database's designated order of Boolean operators is AND, OR, NOT, then the search would retrieve (1) records containing vaccination AND elderly and then (2) any records on flu.
• From left to right: if the database processes in the order in which the terms are entered, then the search would retrieve records (1) containing either flu OR vaccination in (2) the elderly.

To minimize incorrect processing order, terms can be nested using parentheses, just as in algebra. This approach forces a specific processing order because combinations within the parentheses are completed first.

Using the search query (flu OR vaccination) AND elderly as an example, the database would first retrieve records containing flu OR vaccination. Those results would then be searched (i.e., ANDed) for any records on elderly.

Note: "Order of precedence" errors can be more easily avoided by searching each concept, search term, keyword, or field one at a time and then applying the appropriate Boolean operator. This approach allows the searcher to revise the search by trying different combinations and keep track of the results.

(Jankowski, 2008)

More Information

See Appendix 3 - Database Comparison Chart to view a chart that summarizes Order of Precedence by commonly searched databases.  This tool is helpful when translating a search from one database to another database.

(Elsevier, 2019; Jankowski, 2008; Patrick & Munro, 2004; U.S. National Library of Medicine, 2019; Welch Medical Library, 2019; Wolters Kluwer, 2019)

Proximity and Adjacency

The database's search algorithm may also recognize proximity or positional operators. These operators look for requested terms within certain distances of one another. The searcher can designate word order or how close together search terms should be. The operators may be words or letters combined with numerals to designate the nearness of the two words. Often, these operators must be capitalized.

• Adjacency or Near – finds words within x words of one another. For example, sleep N3 stress finds results that have a maximum of three words between them; results may include stress alters sleep; sleep apnea causes stress; etc.
• Next or Within or Precedes – finds words within x words of one another in the exact order entered. For example, sleep W2 stress finds results that have a maximum of two words between them, where sleep is always the first word; results may include sleep disturbances and stress; sleep problems with stress; etc.

Use of proximity operators are useful to refine a search, especially in a database which does not have a controlled vocabulary, i.e., Scopus or Web of Science. 

How to Build a Proximity Search in PubMed

To create a proximity search in PubMed, enter terms using the following format:

"search terms"[field:~N]

• Search terms = Two or more words enclosed in double quotes.
• Field = The search field tag for the [Title] or [Title/Abstract] fields.
• N = The maximum number of words that may appear between your search terms.

For example, to search PubMed for citations where the terms "hip" and "pain" appear with no more than two words between them in the Title/Abstract search field, try the search:

"hip pain"[Title/Abstract:~2]

Search results may include hip pain, hip-related pain, hip joint pain, hip/groin pain, hip biomechanics and pain, pain after total hip arthroplasty, pain in right hip, and more.

See the PubMed User Guide and view the PubMed Proximity Searching tutorial for more examples and information about proximity searching in PubMed.

Tutorials

More Information

Refer to Appendix 3 - Database Comparison Chart to review proximity operators by commonly searched databases.  This tool is helpful when translating a search from one database to another database.

(Gray, 2012; Jankowski, 2008; Welch Medical Library, 2019)

References

Booth, A., Papaioannou, D., Sutton, A. (2012). Systematic Approaches to a Successful Literature Review. London: Sage Publications Ltd.

Gray, J. R., Grove, S.K., and Burns, N. (2012). The practice of nursing research: Appraisal, synthesis, and generation of evidence. London: Elsevier Health Sciences.

Health Sciences Library McMaster University. (2019). The Researcher's Toolkit: The Research Cycle. Retrieved from https://hslmcmaster.libguides.com/research-toolkit

Jankowski, T. A. (2008). The Medical Library Association Essential Guide to Becoming an Expert Searcher: Proven Techniques, Strategies, and Tips for Finding Health Information. New York: Neal-Schuman Publishers.

Patrick, L. J., & Munro, S. (2004). The literature review: demystifying the literature search. Diabetes Educ, 30(1), 30-34, 36-38. doi:10.1177/014572170403000106

U.S. National Library of Medicine. (2019, November 15, 2019). PubMed User Guide. Retrieved from https://pubmed.ncbi.nlm.nih.gov/help/

Walker, G., & Janes, J. (1999). Online Retrieval. A Dialogue of Theory and Practice (Second ed.). Englewood, Colorado: Libraries Unlimited.

Welch Medical Library, J. H. U., School of Medicine. (2019). Expert Searching. Retrieved from https://browse.welch.jhmi.edu/searching

Wessel, C. B. (2019). Responsible Literature Searching for Research: A Self-Paced Interactive Program. Retrieved from https://cme.hs.pitt.edu/ISER/app/learner/loadModule?moduleId=8381&dev=false