Post Graduate Health & Nursing: Protocols & Search Strategies

One of the distinguishing features of a systematic review is the predetermined, transparent and replicable approach that is taken to conduct the research.  Systematic reviews are typically a standalone research and need to be conducted with the same rigour as other primary research.

This approach or method is written up in your protocol and is one of the early steps in your review process as it needs to be done before the review.

Your protocol will outline and justify:

• The search question such as your PICO elements
  • Population
  • Intervention
  • Comparison
  • Outcomes
• Inclusion and exclusion criteria e.g.:
  • Study types (e.g. Randomised control trials)
  • Date range
  • Clinical setting or lens
  • Specifics of the population
• Any potential or mitigation of bias
• The databases searched and the search strategy employed.
• How the information will be grouped for analysis

For further information see the sources linked below.

Have a look to see how other people have written a protocol

Before starting out on your review, one of the best things you can do is look at what other people have done.  In terms of looking at other people's protocols try PROSPERO (linked below) and using the JBI database's Publication Type limiter: Systematic Review protocol.

If you can get something related to your topic, great, it will help confirm the best databases to use and potentially give you some other ideas, however, it won't matter if it is on a different topic, it is more about looking at the types of things that are included in a protocol and how to lay it out.

Staff and students also have access to the JBI SUMMARI software as part of our subscription to JBI, this includes a protocol template.  See the relevant box on the Software & AI page for more information.

PRISMA (which stands for Preferred Reporting Items for Systematic Reviews and Meta-Analysis) is a reporting guideline for systematic reviews and meta-analyses. 

Most systematic reviews will include a PRISMA chart and this guides the steps taken within the search process which enhances the transparency of the review.

Here is an example (taken from the PRISMA 2020 downloadable Flow Diagram) 

This work is licensed under CC BY 4.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/

PRISMA charts differ between reviews, have a look at reviews others have done to get an idea of how they can be utilised.  They are often much simpler than the above example.  Here is one that had been generated from the Haddaway et al. (2022) PRISMA Flow Diagram tool.

Haddaway, N. R., Page, M. J., Pritchard, C. C., & McGuinness, L. A. (2022). PRISMA2020: An R package and shiny app for producing PRISMA 2020-compliant flow diagrams, with interactivity for optimised digital transparency and open synthesis. Campbell Systematic Reviews, 18, e1230. https://doi.org/10.1002/cl2.1230

PICO

One of the things you will see mentioned a lot in the world of systematic reviews, is PICO (particularly in health areas).  This is a framework for structuring a research question.  It can help to ensure that you clearly define the research question, it helps you to organise the concepts into a comprehensive search strategy, and assists with the defining of inclusion and exclusion criteria which in turn improves the consistency and transparency of a review.

PICO stands for:

For our hypothetical research question looking at: Effectiveness of social media interventions on the eating behaviours of teenagers, our PICO might look something like:

For this particular question we don't have a comparator as we aren't comparing two interventions - this is not uncommon, so really in this case, we'd be using a PIO structure.  Sometimes the I might stand for area of Interest. There are other variations on PICO for example:

• PICOT: where T stands for timeframe.  
• PICOC: where C stands for context
• PECO: where E stands for Exposure

Many of you will be doing non-clinical reviews and/or reviews on qualitative topics, in which case, PICO might not fit your needs.  Don't worry, there are other frameworks out there to consider, see the box in this guide called "Alternative Question Frameworks" .

Further Reading

Using individual databases

When conducting a systematic search, it is essential to use individual databases rather than discovery layers (e.g., Library Search) or aggregated databases such as EBSCOhost or ProQuest. Discovery layers search across a range of databases we subscribe to, while aggregated databases can be a combined selection of smaller databases.

This approach is necessary because:

1. Specialised Search Functions: Individual databases offer unique search functions and utilise specific subject headings (e.g., MeSH) that enhance search precision.
2. Replicability: Systematic searches must be replicable. Since different institutions have varying collections of databases and subscriptions, using discovery layers and aggregated databases makes replication challenging.

The databases you'll use will depend on your subject area, but the main databases you'll see used in systematic reviews (that we have access to) are:

The search structure you follow will also depend on the database being used and whether or not it has a controlled vocabulary (subject headings) attached to it (see information on this under the Searching tab of this guide). It is advisable to document the process you use for each database (a table can be a useful way to do this), to ensure you are using the same free-text terms across each search and so you have a record of what you searched and where (it also makes it easier to copy and paste your terms in).

With systematic searches researchers tend to search term or concept separately and then join them using the Boolean operators (AND OR NOT), reasons for this include:

• Precision in retrieval - by searching concepts or terms separately, the results you retrieve are more likely to be relevant and focused on your topic.
• Following structured, systematic methodology helps to ensure transparency and reproducibility in the search process and is part of the requirements for meeting systematic review standards.  This in turn improves rigor and minimises bias.
• For databases that provide a controlled vocabulary (such as MeSH in Medline and CINAHL subject headings in CINAHL) you will want to identify the appropriate terms to search and search them separately from the free-text keywords. This assists with the precision, structure and reproducibility of the search.

You will also need to include (where appropriate) truncation, phrase searching, and proximity operators (these can differ across databases, but generally they allow you to set your search so that certain words need to be near each other in the result, e.g. within 3 words).  

A trick you can use to focus your search is to only search titles and abstracts for your free-text terms, this will of course depend on the needs of your research, but it is an option to consider.

So, what will it look like?

A basic systematic search structure may look like:

1.Concept 1 - Subject heading

2. Concept 1 - Free-text term*

3. Combine these two with OR

4. Concept 2 - Subject heading

5. Concept 2 - Free-text term

6. Combine these two with OR

7.  Use AND to combine searches 3 AND 6

With some databases, such as Medline, you may find searching each term separately beneficial, as Medline will map each term to the appropriate MeSH term.

Scopus is a database that doesn't have a controlled vocabulary attached to it, so you'd follow a similar process, but skip the controlled vocabulary steps.

Remember that this will be one of the final steps of your systematic search, you'll spend a lot of time playing, testing, tweaking and searching your key terms and databases before you run your final searches.

Use the resources in the links below for more information and clarity.  Some of these resources include examples of systematic searches.

*Free-text - is the term used to indicate the words used to capture beyond the controlled vocabulary term. It can be called by a variety of names such as 'natural language'.

Here is a simplified version of what a systematic search in Medline could look like.  Click on the information 'i's to find out more.