Skip to Main Content

Systematic Reviews

Search the Evidence

A systematic review attempts to find all the empirical evidence to answer a research question.

  • If possible, consult with a subject specialist librarian.
  • Search strategies should be complete and reproducible.
  • A thorough study of similar systematic reviews helps with source selection, search terms, and search strategy.
  • Usually at least three library databases are searched.
  • Often, "grey literature," hand searching, cited reference searching, and other techniques are added.
  • If possible, run searches on the same day. Save the retrieved references from each database/source separately.
  • Split search into "concept blocks" such as PICO.
  • Combine synonyms and parallel terms within each concept block with OR. Combine separate blocks with AND.
  • Include thesaurus terms and all common keywords used to describe the concept. Use field tags on search terms.
  • Use truncation where appropriate, and "collapse" similar terms to avoid redundancy.
  • If you use subheadings, search them separately from thesaurus terms ("floating subheadings").
  • Begin with one database (i often begin with PubMed/Medline). Reproduce search strategy as closely as possible with others.
  • Usually, all dates are searched, unless you are updating another SR. Note the exact platform, version, and date searched.
  • Record number of results from each search separately and back up your file.
  • Download or copy/paste exact search strategy for each source and save.
  • Double-check for errors. Make sure your sentinel/example articles are retrieved.

Where to Search

Examples below -- usually at least 3 databases are searched. Look at similar SRs in Cochrane, etc. to see what's expected.

  • Google Scholar has poor reproducibility -- some reviewers consider it unacceptable as "database."
  • WKU does not have Embase, Scopus, or Ovid Medline -- consider substituting another resource.
  • Try to reproduce same search strategy in each database, as database features allow.
  • Detailed, sensitive search (may use only 1-3% of results to compile SR).

"Miscellaneous" sources that aren't peer-reviewed articles. May have to adjust search strategy. Think about best gray literature sources for your subject.

  • Preprints
  • Blogs
  • Clinical trials
  • Theses
  • Technical reports
  • Conference proceedings
  • Newsletters
  • TOC of high-yield journals
  • References citing sentinel articles: who has cited your "best" results?

Save and document your search results

Documenting your search:

  • Review PRISMA Checklist 
  • Run all searches the same day, if possible, and record date
  • Search each database separately, and copy or save exact search strategy, line by line.
  • Record database version and platform.
  • Record number of results in each database, before and after deduping
  • Save results of each database/search engine separately (Endnote or Zotero collection, MyNCBI, MyEBSCOhost, etc.)
  • Back up files! Reference files are easily corrupted.

SR references may be fully handled within Endnote, Zotero, or another reference manager. Excel files may also be used.

Increasingly, SRs are managed within Covidence, Rayyan, or other specialized software. They may be moved here from your reference manager, or references may be exported directly to specialized SR software..