It is important to plan your research in advance of beginning your systematic search. A pre-established protocol gives structure and direction to your project, helps reduce error and bias, allows the review to be registered, and may be a requirement for publication.
A systematic review protocol is a document that explicitly lays out the objectives and methods for the review in advance. The benefits of taking the time to establish a protocol include:
Completing a protocol that conforms to a certain format may be a required step in the publication process for some reviews, such as systematic reviews published by the Cochrane Database. Other resources provide suggestions that can help you develop a protocol to serve as a guide for your team throughout the review process:
These are some common elements of a systematic review protocol which should be determined before beginning the search process:
Other elements to consider prior to beginning the search process include how records will be managed, the timetable for the review, how results will be synthesized, and how the findings will be disseminated.
It isn't always possible to think of every potential situation when developing your protocol, and as you gain a better understanding of the review question, it may become necessary to make some adjustments to your initial plan. However, you should be careful not to make changes that are influenced by an awareness of how the studies you discover might affect your review's outcomes. This can introduce bias and raise questions about the quality of your review.
Your methods must be consistent throughout the review process, therefore any changes made in the middle of the process may necessitate redoing previous work using the updated methods.
Just as you should document every step in the systematic review process, be sure to clearly document and justify any amendments you make to your protocol.
Once you've completed your protocol, it is a good idea to register it.
Benefits of registering your review protocol:
Where to register: