Your first step is to devise a focused, clear research question which your review will address. Most people use a framework to help them devise this question.
You may wish to do some "trial run" scoping searches of a database to find out how much has been written, and what limits you should apply. Web of Science (Core Collection) is good for this, as it includes articles from most disciplines; it is also available in most institutions, so if you are working with non-Trinity colleagues, they should have access as well.
It's not unusual for the research question to change, if it looks like the amount of results will be too many or too few. For postgraduate level, a few thousand results in Web of Science might be the roughly appropriate amount.
There are several different frameworks you can use to help structure your research and ensure you have clear parameters for your search. The most commonly used one used for health-related reviews is the PICO framework:
|Review title||The effect of blueberries on cognition and mood: a systematic review of human intervention trials|
|Population||Individuals of all ages, without regard to gender, race or ethnicity.|
|Intervention||Supplementation with blueberries, relevant blueberry products or extracts from blueberries. This may include freeze-dried blueberries, blueberry concentrate, or blueberry juice.|
|Comparator||Placebo or control groups.|
|Outcome||Changes in cognitive function based on cognitive screening measures (such as Mini mental state examination, Montreal Cognitive Assessment), neuropsychological interview, informant/carer responses to assessment tools or changes in mood. Secondary outcomes include: changes in biochemical levels in biological fluids. Of particular interest are inflammatory markers, cardiovascular disease risk factors, and markers of gastrointestinal health.|
This example is extracted from: PROSPERO 2018 CRD42018100888.
Another framework may be more suitable depending on your review topic. Here are some other options:
Now you have framed your question, you should develop a protocol - a statement of intent on what you plan to do in the systematic review. You won’t have your findings yet, but you know what you are (likely) to do... a draft version with some of the details planned out. It isn’t set in stone – what you do can change when you actually do the systematic review; but this is noted in the SR.
Publishing a protocol also deters others from doing exactly the same review, even if you haven’t published yet!
A protocol outlines the study methodology, including: