Skip to Main Content
XClose

Library Services

Home

UCL LIBRARY SERVICES

Systematic reviews

Context of database searching

Databases structure research citations into fields to describe the citation, for example, Title, Abstract, Journal, Year, Author Keywords, Database Keywords or Descriptors. Database searching involves using terms and phrases to search the fields in order to find citations that are likely to be relevant to a research question.

Generally, in Boolean database searching, the user specifies the exact combination of terms to search the database. Terms and phrases are combined with AND, OR, NOT or proximity operators. This is different to an Internet search engine which interprets what it is thinks the user is looking for. This means the choice of search terms must be carefully planned and tested.

Many topic-specific databases such as PsycINFO (pyschology), ERIC (education), ASSIA (social science), PubMed (health) have a thesaurus, or controlled vocabulary to enables the citations in the database to be organised and found by the user. Each citation is described by the terms from the thesaurus, and can be very useful for a systematic search.  They are sometimes called Descriptors.

A good systematic search will comprise searching of both thesaurus terms and terms in the title and abstract fields and other appropriate fields that may be available.  As the thesaurus is different for each database, each database search is adapted to make use of the appropriate thesaurus. Some databases such as the Citation Indexes in the Web of Science don't have a thesaurus, the Keywords Plus field is automatically generated from the citation and reference.

Find out more:

Overview of steps in developing a systematic database search

Steps to developing a systematic database search

Developing a systematic database search

Begin planning the database search by setting out the distinct concepts that you would expect to be present in the research you wish to find.  

Some familiarisation with the topic is essential. It can help to find relevant citations though Internet searching and database searching.  Look which concepts are present in each citation, and the terms used to describe the concepts, the database descriptors and author keywords that are used.

Use the Help sections in each database to learn the specific functions available for that database.

Checking the database search

Look at a selection of search results to check and refine the search. Be aware of how the results are displayed in the database.

Where a database defaults to displaying by relevance, this can be changed to display by publication year in order to see the less relevant items that are located by the search

  • Consider if the search is revealing new terms that would be useful to use in the search. 
  • Consider how the search terms is identifying any unwanted items or concepts (i.e. generating "noise")

One way of checking the search is to see differences between the results of the controlled vocabulary and free-text searches for each search concept.  

How to develop a systematic database search

Sensitive focussed searching

A sensitive search will search the freetext of specific database fields (e.g. title, abstract, possibly journal fields) and the controlled vocabulary fields. The more terms used in the free-text fields, the more sensitive the search and the higher the yield of citations. 

Proximity searching (NEAR and adjacency operators) and phrase searching are sometimes used to make the free-text terms more precise.  (see the Help section of a database for assistance with this)

A precise (or specific) search is more focused and will miss more studies. This may be appropriate for highly-focused questions or rapid searches. Precise searches often employ fewer freetext terms.  

Some precise searches rely on controlled vocabulary terms or major controlled vocabulary terms, though this approach only works if there is a suitable controlled vocabulary to describe each concept. Be aware that not every concept will be described by a controlled term all of the time. So if the search strategy has three concepts, it is possible that a relevant research citation may have controlled terms for only one or two concepts.

In large databases where many controlled terms are assigned to each research citation, major controlled terms are sometimes useful for finding citations where the focus of the article is upon a particular concept rather than the concept being peripheral or tangential to the main thrust of the article. Try this for one concept at a time rather than all of them together.  

Minimising bias

A search should aim to find an unbiased sample of the research literature.  There are many potential sources of bias in the process, such as choice of search terms and the choice of databases used.

Some bias can be mitigated by becoming familiar with the variety of language used to describe key concepts, and finding out how relevant research in that topic is published. This includes consulting overviews or handbooks on a topic, obtaining feedback from others, or looking at the approaches used for similar systematic reviews.

Look at searches published in systematic reviews that are produced from review-producing organisations e.g. Campbell Collaboration, Cochrane, EPPI-Centre (links are in the Identifying studies section).

Sources of published search filters can be useful (see box on the right). 

Developing a search strategy process is iterative, and it can easily become overwhelming. Create a table that lists each of terms for the concepts, alternative words and spellings.

Searching is a skilled activity and many systematic reviews draw on the expertise of information specialists and librarians.