Systematic Reviews

Preliminary database searching

Early searches may influence the future directions of your research. This could be due to a number of factors including a lack of relevant information, or even no information if the research area you are investigating is very new. 

You can use subject databases, Discovery or Google Scholar for basic, broad searches into a particular area of interest.  

It would also be helpful to search Cochrane and Prospero to see what has already been researched and to harvest useful search terms. These databases will also help you check if your exact study has been done before. If your proposed question has already been answered you will need to choose a different question. You may need to register your chosen search with Prospero. Please check with your supervisor about registering your systematic review.

Use the search methods sections and appendices of any relevant reviews to locate useful keywords and databases.

The search methods section below shows which databases the reviewers have searched. The full review is here.

This is an example of part of a structured search on stroke in CINAHL.  Note the variety of terms used. The full review is here.

At the end of the scoping search you should have identified:

• if your research topic is suitable
• which databases will be the most useful
• some relevant keywords
• key papers in your research areas

Identifying Suitable Databases

To find the published literature/research on your topic you will need to decide which databases to search. There is a wide range available, so you might find it helpful to check your subject LibGuide to identify relevant ones for your review topic. Some key resources for health are also listed below. Your research area may cross over into other disciplines, so check the LibGuides for those subjects too.

The above list is not exhaustive and the databases you chose will depend on your subject area.

MEDLINE, EMBASE and CENTRAL (Cochrane Central Register of Controlled Trials, which is part of the Cochrane Library) are generally considered essential, but additional sources will vary according to the review topic and access/budgetary constraints (Lefebvre et al., 2019).

Whilst it is important that you search as broadly as possible, be aware that with a wide range of databases there are disadvantages to choosing too many:

“It is laborious for searchers to translate a search strategy into multiple interfaces and search syntaxes, as field codes and proximity operators differ between interfaces. Differences in thesaurus terms between databases add another significant burden for translation. Furthermore, it is time-consuming for reviewers who have to screen more, and likely irrelevant, titles and abstracts. Lastly, access to databases is often limited and only available on subscription basis." (Bramer et al., 2017, p. 2).

References:

Bramer, W.M. et al. (2017) 'Optimal database combinations for literature searches in systematic reviews: a prospective exploratory study', Systematic Reviews, 6(245), pp. 1-12.

Lefebvre C. et al. (2019) 'Technical supplement to chapter 4: searching for and selecting studies', in Higgins J.P.T. et al. (eds) Cochrane handbook for systematic reviews of interventions version 6. Available at: www.training.cochrane.org/handbook

Designing an effective search strategy

Developing a search strategy can be time-consuming, as your first attempt is seldom the one you will use for your review.  It is often described as an iterative process, involving consideration of a series of ‘test’ searches which are synthesised into the final version.

Start by breaking your question down into the key concepts in order to generate search terms.  This is often done using the PICO structure, or variations of this approach.  PICO is recognised as a good framework for developing a clinical research question. You can find out more about PICO in the Using a Framework section.

For each of the PICO concepts, it is necessary to list as many synonyms and related terms which might have been used in the journal literature.  Major databases have a subject index, which uses a standardised vocabulary.  Your search strategy should include appropriate subject index terms, plus a range of keywords and phrases, so that it finds as many relevant results as possible.  Scan the database index for useful terms and check how they have indexed the key papers you already have.

The index terms used In MEDLINE are known as MeSH headings.  Look at a full record to see which MeSH headings are assigned to that reference:  

You can also ‘harvest’ additional ‘free-text’ keywords from relevant articles you have read and from your ‘scoping’ searches.  Be sure to include all variant spellings (UK and US English) and abbreviations and be aware of differences in uses of terminology over time and in different countries. The example below is taken from CINAHL:

Start your search with the database that your ‘scoping’ searches indicate will yield the greatest number of results.  Work through your list of possible search terms to identify which of the available index terms will be appropriate for your search.  Some index terms will have narrower or more specific terms, which can be incorporated into your search strategy by using the database's ‘explode’ function. You can find out more about how subject index terms work in the Using subject headings section.

Once you have listed all your synonyms for each of your PICO concepts, they can be combined with the appropriate Boolean operators so that your search strategy finds relevant results.

AND narrows down the search - when terms are combined with AND, they must all be present in each of the records found.

OR broadens the search – when terms are combined with OR, the search finds any record that mentions one or more of the terms.  Use OR to aggregate the results of searches using different terms that relate to the same concept.

You can find out more about using Boolean operators in the Using advanced techniques section.

Here’s an example of a search developed for CINAHL:

(Taken from: Thieme  H, Morkisch  N, Mehrholz  J, Pohl  M, Behrens  J, Borgetto  B, Dohle  C. Mirror therapy for improving motor function after stroke. Cochrane Database of Systematic Reviews 2018, Issue 7. Art. No.: CD008449. DOI: 10.1002/14651858.CD008449.pub3.)

Each database interface has its own unique set of commands.  Information about these will be on the database's own help pages. You can use the help sheets from EBSCOhost to see some of the main differences between the EBSCO and OVID platforms.  You can find out more about advanced search techniques in the Using advanced techniques section.

It isn’t always necessary to use all of the PICO elements in the final search strategy.  (See Cochrane Handbook https://training.cochrane.org/handbook/current/chapter-04#section-4-4-2) Some concepts do not translate well to search terms.  They maybe poorly defined and unlikely to be used in journal articles, so tend not to find relevant results when used to search databases.  This most often applies to terms for the outcomes of research.  It can be more effective to do a broader search, without this element, and sift through a larger number of results. Such concepts can form part of the inclusion and exclusion criteria.  You can find out more about inclusion and exclusion criteria in the Developing inclusion & exclusion criteria section.

It might be appropriate to refine the results of your search using a filter.  Recommended search filters are available for several of the major health databases, enabling you to retrieve research that uses a specified study design such as randomised controlled trials.  You can find out more about search filters in the Using search filters section.

After your initial searches you may wish to revisit and modify your search parameters. Write down all the points you need to consider such as:

• What’s your main focus/plan/goal?
• Scope of study.
• Similar studies (same parameters)
• What’s been done before?

Also consider :

• Age of previous studies.
• Gaps in research? (Arid time period or a “trendy” research period).
• The use of meta-analysis? (several studies consulted to produce one over-view or summary of results)

Focusing your question 

TOP TIPS	Use a framework such as PICO to construct your research question
	Be specific and focused about what you want to find
	Choose a single meaningful issue that fills a gap in knowledge

The aim of a systematic review is to answer a research question that fills a gap in knowledge and formulating the question into an answerable and clear format is important at the planning stage. The research question should be focused; if it is too broad you will be overwhelmed with results and never be able to sift through all of them. However, take care not to make your question too focused; you may not be able to undertake the review due to lack of research. 

Using a framework such as PICO/PIO can help you formulate and focus your question by ensuring you include the important elements of Patient, Intervention, Comparison (if appropriate) and Outcome which then creates a structure for your search strategy. See "Using a framework' in the searching page of this guide for help with using frameworks.

The table below gives examples of how you can take a broad question or area of interest and make it into a focused question that can be used for a systematic review.   

​The question should communicate the intent of the systematic review addressing a single issue that is meaningful to researchers and practitioners. The question will guide your search strategy and the gathering of data.

It should include essential details which emphasise the main issues, but not include too many narrow details as this can be confusing - some elements and details will be in your inclusion and exclusion criteria. (see 'Developing inclusion & exclusion criteria' tab) on the this guide. Your question may be broken down into smaller sub questions if necessary. 

The types of research questions that are often asked in health related systematic reviews include: 

Inclusion and Exclusion Criteria

Consider what inclusion and exclusion criteria you will be using before you start your systematic review. These criteria will be used to assess the relevance and quality of the research you find in your searches and will help you to decide whether they will be included in the actual review.

Using a framework such as PICO (see the 'Using a Framework' tab in the Searching page of this guide) can be useful to develop which concepts you want to include in your search and this will form the basis of your inclusion and exclusion criteria. This can help your search strategy become more focused, targeting the papers which will help to answer your review question and excluding others.

Here is an example of a PICO table used in the systematic review Physical fitness training for stroke patients

You can see which studies were included and excluded for the systematic review above here.

• the time period, for example, including papers published within the last 10 years and excluding any older papers.
• the language of the studies, for example, only including those in English and excluding any non-English language material.

There are additional examples of inclusion and exclusion criteria in systematic reviews (outside of Cochrane) available here:

• Fung, C.N., & White, R. (2012) 'Systematic review of the effectiveness of relaxation training for depression', International Journal of Applied Psychology, 2(2), pp. 8-16.
• Viljoen, E. et al. (2017) 'Police officer disability sensitivity training: A systematic review', The Police Journal, 90(2), pp. 143-159. 

For further information on this topic see the additional resources below: