Search filters and hedges: Finding and using validated search strings

Zahra Premji

What are search hedges

Search hedges, search filters, validated search strings, search blocks, etc. all refer to the same thing.

• A search string that can be copied and pasted, is created for a specific database, and is designed to (focus and) retrieve literature on a specific concept (typically either a topic, study design, geographic region, population characteristic, and so on).

​

• May be validated or not validated
• May be designed for sensitivity or specificity

​

Prompt question

What about database filters? Are they the same or different?

Let’s discuss

Where to find them

• InterTASC Search Filter Resource, CRD: http://www.york.ac.uk/inst/crd/intertasc/
• Harvard Countway Library - Search filters by study design: https://guides.library.harvard.edu/c.php?g=309982&p=2070466
• Hedges Project, Health Information Research Unit, McMaster University: http://hiru.mcmaster.ca/hiru/HIRU_Hedges_home.aspx
• Scottish Intercollegiate Guidelines Network: https://www.sign.ac.uk/what-we-do/methodology/search-filters/
• Cochrane Handbook technical supplement: https://training.cochrane.org/technical-supplement-chapter-4-searching-and-selecting-studies-v63
• Strings Attached: CADTH's Database Search Filters: https://searchfilters.cadth.ca/
• Flinders Filters: https://www.flinders.edu.au/research-centre-palliative-care-death-dying/partnerships-and-projects/flinders-filters
• Ovid expert searches: https://tools.ovid.com/ovidtools/expertsearches.html
• University of Alberta Search Filters: https://guides.library.ualberta.ca/c.php?g=342568&p=5103517

Where to find them (part 2)

In the scholarly literature using some combination of terms such as (validation or development or performance or evaluation) AND (search) AND (filter or hedge)

allintitle:("validation"|"validated"|"development"|”evaluation”|”performance”|"sensitivity")("search"|"searching")("filter"|"filters"|"hedge"|"hedges")

How are search filters developed and validated

Four main steps in search filter development (not necessarily in order)

1. Selecting search terms and search fields
1. objectively (using a derivation set) or subjectively (based on expert knowledge)
2. Identification of a ‘gold standard’ set
3. Evaluation of the search filter
4. Validation of the filter
• Ideally against a separate set of records. (i.e. derivation set and validation set should be different)�

​

From: Jenkins, M. (2004). Evaluation of methodological search filters - a review. Health Information and Libraries Journal, 21, 148-163.

Terminology/formulas in search filter validation

Sensitivity = a / ( a + c)

​

Specificity = d / (b + d)

​

Precision = a / (a + b)

​

Definitions and table from: Jenkins M. (2004). Evaluation of methodological search filters - a review. Health Information and Libraries Journal, 21, 148-163.

Using a filter

1. Testing the filter
1. Check to see that controlled vocabulary and search fields are still the same
2. Test it against your search and look at what is being removed (using the NOT function)
2. Crediting the filter appropriately
• If you use a filter, make sure to cite it
• If you are adapting a filter, make sure to cite it and specify that it was adapted in the search methodology write-up
1. If you remove a term from a validated search filter, it can affect its performance (esp its sensitivity)
2. If you add a term, it will affect its precision (less important)

Considerations for using a search filter/hedge

• Who was the filter designed for? Was it designed for sensitivity? (what is its sensitivity)
• Is it validated? How was validation done
• When was it created?
• Could terminology have changed since?
• Could controlled vocabulary have changed since?
• Could the search fields have changed since?
• Was it designed for a specific discipline? Could discipline-specific terminology be missed?
• Was it designed for a specific database? What do I need to consider when translating to another database?

​

Prompt question: What is the quality of abstracts published in your discipline, and how does this impact search filter development?

Final discussion questions

1. What questions do you have?
2. How common are search filters in your disciplines?
3. If you use a search filter for your primary/master database search, do you have to use it in all subsequent database searches?
4. What is the difference between a search strategy block from a random systematic review and a search hedge?
5. What other issues or challenges can you think of about the development or use of search filters in your work?

Resources and readings about search filters

Glanville, J. Search Filters: What are they good for. https://youtu.be/BAdYZPF57-g

Wilczynski, N. L., Lokker, C., McKibbon, K. A., Hobson, N., & Haynes, R. B. (2016). Limits of search filter development. Journal of the Medical Library Association : JMLA, 104(1), 42–46. https://doi.org/10.3163/1536-5050.104.1.006

Jenkins, M. (2004). Evaluation of methodological search filters - a review. Health Information and Libraries Journal, 21, 148-163.

Glanville, J., Bayliss, S., Booth, A., Dundar, Y., Fernandes, H., Fleeman, N.D., Foster, L., Fraser, C., Fry-Smith, A., Golder, S., Lefebvre, C., Miller, C., Paisley, S., Payne, L., Price, A., & Welch, K. (2008). So many filters, so little time: The development of a Search Filter Appraisal Checklist. Journal of the Medical Library Association, 96(4), 356-61.