Systematic Reviews: Searching

• To ensure comprehensiveness, searches should be carried out in several databases, including but not limited to Medline or PubMed.
• The initial search should be translated for each additional database (for example, subject headings shoud be changed to the controlled vocabulary of the new database).
• Searches should seek high sensitivity, which may result in relatively low precision.
• Too many different search concepts should be avoided, but a wide variety of search terms should be combined with OR within each concept.
• Both free-text(keywords) and subject headings should be used (for example in PubMed/ Medline use Medical Subject Headings (MeSH)).
• Highly sensitive search strategies (filters) can be used to identify types of studies, for example, Randomized controlled trials (see the Filters tab for more information).
• Truly comprehensive searches will include grey literature searching (see box at right), and hand searching of reference lists and/or important journals.

Adapted from Part 2, Chapter 6 of the Cochrane Handbook.

Google Scholar should not be included as a database, but it can be a good source of additional information.

Tips:

• Search Google Scholar after you have done your database searches. 
• Look through the first 100 results.
• Choose only relevant (include) or potentially relevant (can't tell from abstract) sources that you have not already found in the database searches. 
• More tips on formatting a search for Google Scholar are available at the link below:

The Yale MeSH Analyzer can help identify the reason why some known relevant articles are missing in the initial search result set and/or identify additional search terms. You input PMID numbers of up to 20 known articles and the Analyzer will export MeSH headings and author keywords for each article in grid format.

The first link below is to a tutorial on how to use the Analyzer.

The second link is to the Analyzer itself.

US Government Documents

Dissertations and Theses

Trial Registries