Evidence Synthesis

• Most of our searches start as natural language searches. We want to make our searches more computer-like, transparent, and replicable – every search will need refining. Searching is iterative, and a process of discovery!
• The two elements of searches are precision and sensitivity. A precise search produces narrow results, but may miss some relevant resources. A sensitive search captures more literature, but requires more screening out of irrelevant literature.
• The basic steps to a search strategy are:
  1. brainstorm keywords
  2. expand keywords (through thesauri, subject headings, results
  3. use Boolean to combine your terms
  4. refine your results
• Remember to document each search you try, and the results (how many, relevance to your topic)
• To achieve a better balance of precision and specificity you will need to design sophisticated searches that leverage controlled vocabularies and search syntaxes.

• Use AND to reduce results to only searches that contain both terms – eg, feminism AND climate change
• Use OR to increase results for synonyms/ related concepts – eg, climate change OR global warming
• Use NOT to exclude all results with that word or phrase – be careful! – eg, forestry NOT Canada
• Proximity (in EBSCO) – N# will search within that number of words – eg, feminism N5 “climate change”
• Use * as a wildcard at the end of a word – eg, femini* = feminist, feminism, etc.
• Use “quotation marks” to search for an exact phrase or title, in that order (works in Google Scholar, too!) – eg, “climate change”
• Use brackets () to group concepts – eg, (“climate change” OR “global warming”) AND (femini* OR patriarchy) NOT Canada

• Look for reoccurring keywords and phrases in your search results, and incorporate them into your search. Some databases are indexed in a way that a controlled vocabulary is added. Occasionally these use out of date terms, but you will get more relevant results by using them.
• In most library databases, you can find and search the thesaurus, as well as add subjects (su: ) in the Advanced Search
• Re-use existing search filters, or hedges. Find one in a relevant study, or read more here: https://www.yopl.info/post/simplify-your-database-search-with-filters-also-known-as-hedges . University of Alberta provides a number of health sciences focused pre formatted search strings for MEDLINE, EMBASE, and PsychInfo : https://guides.library.ualberta.ca/health-sciences-search-filters/home

• When you find a good article, look for other relevant articles in the citation list. In some databases these will be linked, otherwise this is a good time to use quotation marks to search the title!
• In Google Scholar and Web of Science there will also be links showing you where the article you are currently reading has been cited (cited by, cited reference). This is also a good way of finding relevant results!
• CitationChaser is a tool that uses the lens.org database to both forwards and backwards “chase” citations: https://estech.shinyapps.io/citationchaser/

Documenting all of your searches, and the decisions behind them, is a very important part of evidence synthesis projects. You will want to use/ adapt a template to keep track of everything you do in terms of searches.

Additional tracking documents:

• Search strategy checklist
• Project management task sheet for evidence synthesis projects
• Search result numbers and deduplication tracking sheet (for use with Covidence).
• PRISMA examples document

A handout on common inclusion/exclusion criteria: https://unimelb.libguides.com/sysrev/inclusion-exclusion-criteria

For structured searches, start with the University of Winnipeg Databases.  However, you may need to expand your searches, depending on methodology, to include grey literature and other sources.

• Resources in Grey Literature: https://www.greynet.org/greysourceindex.html
• Grey Literature: Policy Commons https://policycommons.net/
• Preprints: SSRN https://www.ssrn.com/index.cfm/en/
• Preprints and more: OSF https://osf.io/search/
• Open Access - Dimensions.ai - https://app.dimensions.ai/discover/publication
• Open Access - Open Lens.org - https://www.lens.org/lens/search/scholar/structured
• The Unpaywall Extension: https://unpaywall.org/products/extension  
• Re3data: Subject specific data repositories
• Open DOAR: Directory of Open Access Repositories

While searches in Google Scholar are not replicable, and it cannot be used as a database in certain types of methodologies, it can still be a useful tool for finding relevant literature. Google Scholar also uses some of the same Boolean operators, as well as some different ones. (- instead of NOT, and you can remove specific sites from search results using -site: . Adding a tilde ~ will search for similar terms, and AROUND (#) is used for proximity searching. intitle: can be used to search for words in a title.

Systematized searching requires translations of searches (syntax, terms, etc) across different databases. For tips on how to do so:

• UVic Search translation guide
• Date filters for updating searches
• Cochrane syntax translation guide
• Low and Middle Income Countries search filters
• Geographic filters
• Search filters vs. Search hedges
• Language for addressing antiquated, non-standard, exclusionary, and potentially offensive terms in evidence syntheses and systematic searches

PRISMA Flow Diagrams can be used to show the results of each step of the searching process, and what has been retrieved, deduplicated, included, and excluded, and why.

Flow diagram templates and an app are available at https://www.prisma-statement.org/prisma-2020-flow-diagram

This example of a flow diagram is from Colclough Z, Estrella MJ, Joyce JM, Hanafy S, Babineau J, Colantonio A, et al. (2024) Equity considerations in clinical practice guidelines for traumatic brain injury and the criminal justice system: A systematic review. PLoS Med 21(8): e1004418. https://doi.org/10.1371/journal.pmed.1004418