Systematic Reviews and Evidence Syntheses : Selecting Databases

Databases often specialize in literature from specific disciplines. For example: CINAHL focuses on nursing and allied health literature, while ACM Digital Library focuses on computer engineering literature.

You may have specific journals which are highly relevant to your research question. Therefore, you may want to know which databases index these journals. You should be able to find such information on the journal website, or by checking a list of indexed journals provided by the database. You can find more information on the latter in the final box on this page, “How do I find information about a coverage and/or scope of a database?”

If your research question has a specific country-level or regional focus, your selection should include consideration of whether the databases index significant numbers of journals from specific countries or regions. For example: the biomedical database Embase has significant coverage of European and East Asian journals.

Beyond knowing whether a database indexes a specific journal, you may also want to know whether a database indexes the journal from its inception, or whether it only covers a smaller time span, for example 2015-2019. Just because a database indexes a journal doesn't necessarily mean that it indexes the full breadth of the journal's publication years. You can find more information on the latter in the final box on this page, “How do I find information about a coverage and/or scope of a database?”

Most databases have a small number of non-English language literature. However, if it is important that you capture literature in other languages, you may need to consider additional specialized databases. For example: SciELO for Spanish and Portuguese language scientific literature, or CNKI for Chinese language scientific literature.

Consider: is it important to capture conference proceedings, clinical trial registrations, preprints, theses & dissertations, government documents, white papers, program reports, policy briefs, organizational annual reports, or legislation? Databases differ greatly on the kinds of publication types they include. There are often different databases for specific publication types.

Also may be called a "controlled vocabulary". Essentially a set of 'tags' which are attached to database results, depending on what the item or article is about. These tags help with search and retrieval, making for a more effective search. PubMed's MeSH (Medical Subject Headings) is one notable example).

Examples: AND, OR, NOT. Most databases have the capacity for boolean-style searching.

Sometimes it can be helpful for search for a word within a certain proximity to another word. Not all databases have this capacity.

Searches for evidence synthesis projects can be quite long, with many synonyms for each concept. It can be helpful to be able to easily combine a search for one concept with a second concept, from within the database interface.

Does the database allow you to easily specify in which fields you would like to search? Can you search with both keywords and subject headings? Is the database's search interface easy to work with? All of these are important considerations.

Do you need to search specific fields? For example: author affiliations, journal name, or full text? Not all databases have these capabilities, so it is important to consider this when selecting databases.

Are you able to save your searches within the database, if desired? Can you view your search history? These capabilities can be helpful as you construct your search strategy.

Some databases have built-in duplicate detection, which can be helpful. However, duplicate detection can be accomplished by using citation management software like Zotero, or screening tools like Covidence. 

It is essential that you can export and download search results. 

It is rare but not unheard of for databases to have a limit on the number of results which can be exported in one export, or over a 24 hour period. Download/export limits can complicate the search process. 

It is essential that search results can be exported in a format which can be used in citation management software like Zotero, or screening tools like Covidence: RIS or BibTeX are two of the most common file types for database search results.

Evidence synthesis methodologies places great value on the reproducibility of searches. Most databases use boolean-style searching, which allows for reproducibility. On the other hand: reproducibility is more difficult in the case of search engines like Google, or virtually impossible with AI-assisted search and discovery tools. However, search engines and AI-assisted tools can still be valuable in supplementing traditional databases searches.