What are Search Filters?

Search filters (also known as search hedges or search blocks) are pre-formulated search queries that are tailored to capture a concept as comprehensively as possible. They can be convenient and useful if you need help with devising a search strategy or if you search a concept often and would rather not input the same query every time.

For example, if you wish to create a search that explores the Canadian contexts of a given topic, then you may want to include a search query that completely captures the concept of ‘Canada’. The University of Alberta has some pre-fabricated search filters that could be useful. The one called ‘Canada and Provinces’, for example, contains different versions of the search filter for commonly used databases, such as Embase and CINAHL. The filters include subject headings and all grammatical variations of words used to describe Canada, the provinces, and the territories.  In the search filter below, Canad* covers Canada, Canadian, and Canadians).

TRY IT YOURSELF - COPY AND PASTE THE TEXT BELOW INTO THE CINAHL SEARCH BOX:
[mh Canada] or Canad* or "British Columbia" or "Colombie Britannique" or Alberta* or Saskatchewan or Manitoba* or Ontario or Quebec or "Nouveau Brunswick" or "New Brunswick" or "Nova Scotia" or "Nouvelle Ecosse" or "Prince Edward Island" or Newfoundland or Labrador or Nunavut or NWT or "Northwest Territories" or Yukon or Nunavik or Inuvialuit

From filters to retrieve studies related to Canada in CINAHL - University of Alberta

How do they work?

Search filters are very easy to use. First, find a search filter. You can either use a search engine and look up your concept with ‘search filter’ added to the search query, or you can try to find a search filter from an organization or group dedicated to developing and maintaining search filters. Two such groups are:

Once you find a search filter for your concept, ensure that it is designed for the database(s) you want to search. Then, simply copy and paste the search filter into the search box of the database. From there, you can combine that search filter with your other search concepts to develop your final search for your given research question.

How are they built?

Technically speaking, anyone can create a search filter. If you have a concept that you search often and find yourself entering the same query over and over, you could save that query as a text document for later use. That would count as a personal search filter.

Of course, not all search filters are created equal. Methods for devising them have changed and become more sophisticated with time. Initially, search filters were developed by librarians who had expertise in a given subject area. Some later search filters were validated with a gold standard or reference set of articles (i.e. if a search filter was able to pull all the articles within a reference set, then the search filter was likely specific enough for the concept). More recently, statistical and automated methods were developed to validate search filters. Although many methods of validation are now available, not all search filters are validated.

Frazier et al. (2015) explain the reference set validation method in detail. Their goal was to develop a validated search filter for ‘biomarkers for oral squamous cell carcinoma (OSCC)’. The method used in that paper started with developing a search strategy for general prognostic studies on OSCC. They reasoned that all studies on biomarkers for OSCC should be contained within that broader set. Then, at least two researchers independently screened the set of articles based on inclusion and exclusion criteria to produce the final set of articles on biomarkers for OSCC. (This is a high-level overview of their methods; please read the article for greater detail.)

How reliable are they?

Some search filters can be considered high quality if they have been validated. However, most search filters available are unvalidated, including those available from University of Alberta. This doesn’t mean they aren’t useful, but there is no guarantee that they will capture every article on a given topic. They can still be helpful when you only need to find a couple of key articles for a policy, a research assignment, or for patient care.

If you are working on a knowledge synthesis project (e.g. a scoping or systematic review) where comprehensiveness matters, it is best to use a validated search filter. Most validated search filters are published as journal articles and can be found by searching on databases or on websites that compile them (see ‘How do they Work?’, above). Kavanagh et al. (2021) present an example of a validated search filter. The tested search filters and their validity can be found in Table 1 of their article. They also outline exactly how the search filters were tested and validated.

The creation and validation of search filters is a whole field of study, but for most researchers and clinicians the most important thing to understand is where to find them and how to use them. For those working on knowledge synthesis projects, guideline development, or other cases where comprehensiveness is key, high quality validated search filters will be indispensable and convenient. For most other everyday uses, it is not necessary to use the most validated search filter, but rather the search filters that give you relevant results.

If you have questions about finding and using search filters or would like support in searching library resources, email AskLibrary@nshealth.ca or book a one-on-one consultation with a librarian.

Additional Support

Search Filters in Library Services' Searching subject guide.

References

1. Frazier, J. J., Stein, C. D., Tseytlin, E., & Bekhuis, T. (2015). Building a gold standard to construct search filters: A case study with biomarkers for oral cancer. Journal of the Medical Library Association, 103(1), 22–30. https://doi.org/10.3163/1536-5050.103.1.005.

2. Kavanagh, P. L., Frater, F., Navarro, T., LaVita, P., Parrish, R., & Iorio, A. (2021). Optimizing a literature surveillance strategy to retrieve sound overall prognosis and risk assessment model papers. Journal of the American Medical Informatics Association, 28(4), 766–771. https://doi.org/10.1093/jamia/ocaa232.

Vinson Li

Librarian Educator
Yarmouth Regional Hospital, Western Zone