Searching for Systematic Reviews & Evidence Synthesis: Searching Databases

Library Services have produced subject guides to help you to find out more about the resources relevant to your subject area and about the tools on offer to help you with your research. Find the guide for your subject area and explore the Databases tab to learn more about key databases for your subject area.

What databases should I search/how many databases should I search?

The Cochrane Library Handbook suggests Medline, Embase, and CENTRAL as a minimum (if you are undertaking a medical intervention-based systematic review). 

As a general guide you should also choose the databases most relevant to the subject area of your systematic review e.g. PEDRO for physiotherapists, CINAHL for nursing/allied health or PsycInfo for psychiatry/psychology type questions. Web of Science or Scopus are also good broad-ranging databases worth investigating to see whether relevant results are retrieved.

How do you know what you are missing/when can I stop searching?

In your search strategy you are aiming to reach a compromise between precision/specificity (the results are on your topic and relevant for you) and sensitivity (where you are as inclusive as possible in order not to miss any useful results). Systematic reviewers often err more towards being as inclusive as possible but this can generate huge numbers of results (particularly if you are searching across multiple databases). To be practical therefore most systematic reviewers will create search strategies that are precise enough for them to able to cope with the number of results that are returned, but use different methods of searching, e.g. both subject headings and keyword searching (in the databases that allow this) and searching multiple databases (where articles may be indexed differently or a very similar search retrieves results in a different way) to ensure that relevant results are not missed.

The Campbell Collaboration's Searching for studies: A guide to information retrieval for Campbell Systematic Reviews discusses the range of databases to be used and says:

"Given the multidisciplinary nature of most social science research questions and the large selection of social science-related databases, searches must be implemented in multiple databases. Terminology (both keywords and controlled vocabulary) will vary across these databases as different disciplines often use different words to mean the same thing."

They provide a comprehensive list of the databases in the social, behavioural, education and health sciences in the Appendix but suggest that the more frequently used subject databases include:

• Applied Social Sciences Index and Abstracts (ASSIA)
• The Cochrane Central Register of Controlled Trials (CENTRAL)
• Education Resources Information Center (ERIC)
• International Bibliography of the Social Sciences (IBSS)
• MEDLINE
• PsycINFO
• Sociological abstracts

You can access these via the Databases A-Z List.

We have created a comprehensive e-learning pathway to support you to develop the advanced searching skills for systematic reviews. To find the course, click the link to KLaSS below and look for Finding Academic Material for Advanced Health Topics. This e-learning module also describes how to book onto our 2 hour webinar LIB261 Search Techniques for Systematic Reviewers.

Library Services provides access to an extensive range of electronic resources to support teaching, learning and research across the Health Faculties. Look at the databases tab of your library subject guide to learn more about what databases might be relevant to your subject area and how to access them.

Some databases, e.g. Medline, will be searched by most systematic reviewers undertaking reviews in the health area, whilst others may be more subject specific, e.g. SciFinder for chemistry/pharmaceutical science or the Nutrition and Food Sciences CABI database for systematic reviews in that area.

Remember that you should always search one database at a time in order to make the most of the advanced features rather than cross-searching multiple databases even if that is offered as an option on a particular platform, e.g. search Embase and Medline as separate databases even though they are both on the OvidSP platform. This is essential for effective searching for systematic reviews.

The debate as to whether Google Scholar should be used in systematic reviews, either as one of the databases searched, or potentially as the only database needed in a search of the medical literature continues. A study in 2013 (Gehanno et al) found that Google Scholar included all original studies which had been included in 29 Cochrane Systematic Reviews (738 original studies). The conclusion drawn by Gehanno et al was that the authors of these systematic reviews could have searched only Google Scholar rather than all the bibliographic databases they did and found the same results. Criticism of this research (Giustini, 2013; Herman, 2015) focuses on the methodology of the Gehanno et al article (where known articles were searched for, as opposed to seeing whether a search strategy could have been developed to locate all the articles, which is how a systematic review would actually have to be carried out) and concern as to how many results would have had to be sifted through in order to retrieve the relevant ones and whether this would ever be feasible. It is worth noting that none of the criticism suggested that Google Scholar shouldn't be searched as an additional source but they do highlight some of the practical searching limits with Google Scholar and the issues that are caused by Google not being explicit about exactly what sources are searched or their ranking methodology. It is still hard to guarantee that a reported search strategy in Google Scholar could be repeated and the same results produced and this ability to replicate the research is of course as important in systematic reviews as it would be in the lab. 

A further study (Bramer et al, 2016) concludes "Despite its vast coverage of the scholarly literature, Google Scholar (GS) is not sufficient to be used on its own as a single database to support SR searching. The reason for this is not low precision in GS searching, which is comparable to traditional databases. More problematic is GS’ low recall capabilities which are related to the viewable 1000 search results only policy of the search engine. Even if Google Scholar was to allow users to browse beyond the first 1000 search results, its overall recall would still be too low to locate all included references to support the systematic review. We conclude similarly that neither Embase nor MEDLINE on its own is sufficient in retrieving all included references for SRs".

If you are interested in reading more about this then the following articles and discussion may be of interest.

• Bramer, W.M., Giustini, D., Kramer, B.M., and Anderson, P. (2013), 'The comparative recall of Google Scholar versus PubMed in identical searches for biomedical systematic reviews: a review of searches used in systematic reviews', Systematic reviews 2, 115.
• Bramer, W.M., Giustini, D. and Kramer, B.M.R. (2016), 'Comparing the coverage, recall, and precision of searches for 120 systematic reviews in Embase, MEDLINE, and Google Scholar: A prospective study', Systematic Reviews, 5(1).
• Gehanno, J.F., Rollin, L., and Darmoni, S. (2013), 'Is the coverage of Google Scholar enough to be used alone for systematic reviews', BMC medical informatics and decision making 13(7).
• Giustini, D., and Boulos, M.N. (2013), 'Google Scholar is not enough to be used alone for systematic reviews', Online journal of public health informatics 5(2), 214.
• Herman, S.W. (2015), 'Google scholar could be used as a stand-alone resource for systematic reviews', Evidence Based Library and Information Practice 10(2), 147-149.

When you search the Cochrane Library you are searching across a number of different databases, including Cochrane systematic reviews, systematic reviews published elsewhere and the Cochrane Central Register of Controlled Trials (CENTRAL). 

"CENTRAL is a highly concentrated source of reports of randomised and quasi-randomised controlled trials. The majority of CENTRAL records are taken from bibliographic databases (mainly MEDLINE and Embase), but records are also derived from other published and unpublished sources. In addition to bibliographic details (author, source, year, etc.) CENTRAL records will often include an abstract (a summary of the article). They do not contain the full text of the article.

Each Cochrane Review Group maintains a collection of reports of controlled trials relevant to its own area of interest. These are called Specialised Registers. Unique content (i.e. records not already identified in MEDLINE, etc.) from these Specialised Registers is also published in CENTRAL. Groups may also collect items that are not relevant to its own field of interest; known as hand search results, which are added to CENTRAL as well." 

Information taken from the About Cochrane Central Register of Controlled Trials (CENTRAL) webpage.

CENTRAL is therefore only going to be a database of interest to you if you are interested in retrieving randomised and quasi-randomised controlled trials (some areas of research may not be suitable for RCTS; whilst some systematic reviewers are specifically excluding RCTs as they are focusing on qualitative research). If you are not sure, it would be worth undertaking a basic search on CENTRAL to see what is retrieved.

Advanced searching using the Search Manager and MeSH Medline subject headings is also available on the Cochrane platform.

If you are NHS staff you can use the NHS Knowledge and Library Hub with your NHS OpenAthens to access Medline, CINAHL, EMBASE, British Nursing Index, etc. See more information on searching the journal literature on the NHS Library Guide.

Access the AMED, BNI, CINAHL, Embase, Emcare, HMIC, Medline and Social Policy and Practice databases via the links on the Hub.

Remember that you should always search one database at a time in order to make the most of the advanced features rather than cross-searching multiple databases even if that is offered as an option on a particular platform, e.g. search Embase and Medline as separate databases even though they are both on the Ovid platform. This is essential for effective searching for systematic reviews.

Find out more about the support offered through Library Workshops and 1-2-1s in our NHS LIbrary Guide including on how to book onto the NHS002: Healthcare databases for Advanced Searchers workshop.

The Medline database is one of the key databases to be searched by anyone undertaking a systematic review in the health, clinical and life sciences fields.

Medline can be searched on a number of different platforms including PubMed, and Ovid (accessible by Kings staff and students, as well as NHS staff) among others. Whilst it is possible to undertake searches for systematic reviews on the PubMed platform you may find it easier on one of the other platforms with Medline where it may be clearer for you to specify exactly what you are searching for and to combine your selected subject headings (MeSH terms) with free-text keyword terms.