Skip to Main Content

Systematic review guide

A step by step guide to doing a systematic review

Scope existing literature

Before progressing further into your systematic review, you should conduct scoping searches. These searches will provide you with an overview of some of the literature on your topic and give an insight into the potential number of relevant studies. They will also help to identify other reviews on the same subject or if there are gaps in the currently available research. This information will indicate if your own review is feasible and justified.

Scoping and scanning the literature available will also ensure further understanding of the area you are researching. This will enable you to better plan your review by allowing you to potentially further focus your research question, start to develop a thorough search strategy for your final systematic searches later in the review process, as well as giving you insight into what kind of data is available (homogenous or not) and how it could be synthesised. 

Is there enough research available for you to be able to analyse and synthesise? If not, you may need to broaden or change the focus of your question.

More commonly, has too much research been written about this topic? if your search is already retrieving very large numbers of relevant results, then you will need to focus your question further. 

Finding prospective, in progress and completed systematic reviews

Before starting wider scoping searches you will need to identify any similar or identical reviews to yours. These websites are a good way of seeing what has already been published and what reviews have been registered and are underway. 

  • Epistemonikos - database of systematic reviews, systematic reviews found also list their included studies 
  • PROSPERO – prospective register of systematic reviews in healthcare
  • Cochrane Database of Systematic Reviews – database for systematic reviews (including protocols) in healthcare conducted by the Cochrane Collaboration

Can you justify your review if there is too much overlap with an existing review? Especially if not enough original research has been published since that was written or there are not enough improvements to be made to that original review?

Scoping searches

The main point of scoping searches is to get an idea of what types of studies are out there (and how many), which helps you understand how (and if) you should conduct your review going forward. A bonus is that they also help you see useful terminology, spellings and alternative terms to inform your later systematic search strategies.

While starting a search can be daunting it can be easier if you break it into steps. Start with what you already have or can easily find: 

  • You are likely to already know of key papers or been recommended them by other review team members or a supervisor. Look through their references, or check who has cited them, to get ideas of related papers. 

  • Quick searches on Google Scholar 

  • Using Artificial Intelligence (AI) products such as Elicit, Semantic Scholar or Research Rabbit could help to to quickly produce a pool of relevant literature. Be aware that AI cannot, and should not, replace doing your own scoping searches in relevant databases.  

These key papers and quick gathering exercises may help form your “gold set” or "test set" (explained further in our Test Your Search Results page). They will also inform your preliminary scoping searches, as now you’ve found a selection of what you are looking for you can “reverse engineer” a search. Quickly scan these papers to see both common or repeating and less common or alternative ways that concepts are described. This will be the basis of your database searches. 

Doing the searches

A search in a couple of the most relevant databases for your topic should be sufficient to achieve the aims of scoping. Preliminary searches can be conducted in databases like Ovid MEDLINE and EMBASE, or a more specialist focused database like PsycINFO, CINAHL (Nursing and Allied Health) or ERIC (education) if your question has a discipline or profession specific element. SCOPUS could give you more general STEM and cross-disciplinary results.

Your searches in these databases should be quite targeted and specific, rather than the complex, sensitive and exhaustive searches you will do later in the review. However, using some advanced searching techniques will ensure you identify a selection of the more relevant literature available. 

Your searches should: 

  • Have a mix of the most relevant and commonly used keywords and terms for a concept

  • Include precise Subject Headings from the database being searched (if the database has them)

  • Boolean search operators OR and AND to construct the search with result sets for each concept

  • Be conducted in one database at a time 

You may wish to play with language, date or study type limits to get a better picture of the available literature. 

Below is an example of a targeted, specific (rather than sensitive) search for MEDLINE looking for articles on the effects of fathers providing skin-to-skin care in NICUs.  The search concepts are fathers AND NICUs AND skin-to-skin care
Ovid MEDLINE(R) ALL  
1    exp Fathers/    11292  
2    (father or fathers).mp.   54663  
3    1 or 2    54663  
4    Intensive Care Units, Neonatal/    18749  
5    (neonatal intensive care unit* or NICU*).mp.    30559  
6    4 or 5    37007  
7    Kangaroo-Mother Care Method/    802  
8    (kangaroo care or skin-to-skin).mp.     8244  
9    7 or 8    8602  
10    3 and 6 and 9    54 

For the purposes of identiying key papers and getting a representative sample of very relevant papers, 54 results (from final search line 10) is very manageable to quickly scan through. However, given the small number of results found you might wish to widen the search slightly, for instance by adding in more synonyms and broader terms as keyword and subject heading searches e.g “dad or dads or parent or parents” as well as “father or fathers). This will make the search a bit more "sensitive" and increase recall - bringing back more papers - some of which are potentially relevant but not present in the 54 results.
Remember that searching and scoping will be an interative process and you should reserve some time to create and test with various search strategies to find the information you need at this stage.