Systematic Reviews & Other Evidence Synthesis

Writing a successful search strategy for an evidence synthesis project requires more time and thought than your average search for information. Although some search techniques, such as Boolean searching, work the same in most if not all databases, there are subtle differences between databases that make their search functionality a little different. Therefore, you will need to "translate" your search from one database to the next. More information about this process is provided in the tab for Step 7: Translate the Search Strategy.

A librarian can assist you with the basics of a search strategy, such as locating information on each database's advanced search functionality. However, writing, testing, and refining the search is part of the authorship process for an evidence synthesis project.

Boolean Searching

Using Boolean logic is an important component of writing a search strategy for databases. As shown below:

• Connecting search terms with AND narrows the search—e.g., children AND exercise.
• Connecting search terms with OR broadens the search—e.g., (children OR adolescents) AND (exercise OR diet).
• Connecting search terms with NOT excludes terms—e.g., exercise NOT diet—even if your preferred term is also in a source.

Using Index or Subject Terms in a Search

If the database you're using for a search includes a controlled vocabulary or thesaurus, you can leverage this in your search strategy. For example, PubMed uses Medical Subject Headings (MeSH) to index articles by subject. Combining these subject terms with keywords helps ensure that you are likely finding everything available on your topic in your search.

Other Search Tips

Some other commonly used search techniques might come in handy in what can become a complex database search. Use with caution, however, unless you're certain these techniques don't end up eliminating potential sources from your search or making your search too broad. While you want your search to be inclusive, an overly broad search will often become overwhelming. It's necessary to try to find the right balance between a search's sensitivity and precision (see below).

• Using an asterisk (*) at the root of a word finds all forms of that word—e.g., (child* OR adolescen*) AND (exercise* OR diet*). This process is known as truncation.
• Using parentheses around terms will ensure that all terms will be searched together as a set—e.g., (child* OR infant* OR adolescen* OR teen*).
• Using quotation marks around a phrase searches for that exact phrase—e.g. "young adults".
• Most databases allow you to set a proximity limiter, which looks for your search terms within a certain distance from each other, such as within 5 words of each other. For example, in ProQuest databases like APA PsycINFO, connecting search terms with NEAR/# (e.g., "cognitive NEAR/5 therapy") will look for where those words appear with a maximum of 5 intervening words. Note: The format of these limiters often varies between database providers, so check the database help file or ask a librarian when you are translating your search for other databases.

Sensitivity and Precision

As mentioned, ideally you will find a balance between your search's sensitivity (i.e., find all the things) and precision (i.e., find the specific things I'm looking for) that leaves you with a manageable number of search results that are reasonably on target for your evidence synthesis. Here are a few tips for maximizing sensitivity and precision.

To increase your search's sensitivity, you can:

• Include more alternative search terms for each concept (connected with OR).
• Include index or subject terms in your search.
• Search in additional databases or other sources.

To increase your search's precision, you can:

• Refine your research question to be as specific as possible (e.g., limiting population by gender or age if relevant) and reflecting that in the search strategy.
• Restricting your search to title and abstract rather than full text. Note: This is a common practice in many systematic reviews.
• Limit by date range if this is relevant for your systematic review.