You need a plan to start a literature/systematic review. This plan is a Protocol/Proposal.
1: Is your literature/systematic review novel or duplicate?
To find out, you should run the ‘scoping searches’ or a ‘scoping review’. It means you should search for systematic reviews or reviews trying to answer the same question as yours.
* a scoping review is a type of review through which the researchers find our about the scope and amount of the literature related to a topic or a research question. Scoping search is very useful before any research or review because it tells use if your research/review is novel.
1.1: If there is no similar systematic review
Then you can start writing your protocol/proposal.
1.2: If there is a similar systematic review
Then either you stop your project or you check the existing systematic review to see what is done and how. You might realize that the existing systematic review is old and requires updating, then you could run an update systematic review. Also the existing review might not assess the aspects of the research questions that you want. Then you could run another systematic review focusing on a certain aspect of the review.
1.3: If there are many similar systematic reviews
Then you can conduct a ‘systematic review of systematic reviews’. It means that you could write a protocol/proposal for your research question but one of your inclusion criteria for the studies will be being a systematic review.
2: Where to search for existing systematic reviews?
PubMed is and the most used medical resource and since it includes MEDLINE you can find out if anyone has already published a similar systematic review.
2.2: Cochrane Library
Its full reviews are already visible in PubMed so searching PubMed might be enough. But if you could search Cochrane Library, you could search for published protocols, not visible in PubMed, and other reviews as well. [Example of published Cochrane systematic review protocol]
This is the database of protocols/proposals of published, unpublished, ongoing, ceased systematic reviews. If it helps, you should register your systematic review protocol in this register as well. [Example of PROSPERO Record]
This source is searching different types of studies but displays the systematic reviews in a separate category.
2.5: Google Scholar and Google
To broaden your search, you could search for systematic reviews in Google Scholar or Google.
2.6: Set email alerts
Since your project might take between one month and 3 years to finish, there might be new systematic reviews becoming available since your search date. To keep updated, it is better you sent an alert in each one of the searched resources so that if a new systematic review in your topic gets published then you will receive an email alert from the database.
3: How to write a systematic review protocol/proposal
3.1: Research question
Write your question down and elaborate its concept and their scientific definition and meaning. Be as detailed as possible.
The most common structure for clinical questions is PICOS
3.2: Eligibility criteria
What are your inclusion and exclusion criteria for the studies in your review?
3.3: Search methods
What are the databases that you are going to search?
What are the search terms that you have collected?
What is the search strategy for each database?
When will you search?
Are you going to update the search before the end of project?
3.4: Screening methods
Two members of the team must screen title and abstracts of the search results against the eligibility criteria. While screening the full texts, again, two members of the team should repeat the same process. If there are any disagreements then either they discuss them among themselves or ask the third member of the team for consensus.
Are you using a tool to screen the search results?
*A systematic review team must have at least two members. This is to reduce the human errors and subjectivity bias. This double-checking and triple-checking continues all over the systematic review.
The data should be extracted and checked by two members of the team.
What types of data and meta-data will you need? Characteristics of studies? Outcomes data? Risk of bias and quality of studies? Qualitative data?
Have you designed a data extraction form? [See additional file as an example of data extraction form for randomized controlled trials]
Are you using a default tool or program for data extraction and management?
3.6: Data analysis methods
Are you going to analyze the data using a certain statistical test or program?
What are the types of analysis?
How are you going to deal with the missing data or the same data reported in different formats?
3.7: PROSPERO registration number
When you finished your protocol, register it in PROSPERO and record the registration number. You could report this registration number in any publication or report of your systematic review and keep the PROSPERO record updated.
Doing so you let the other researchers now that you are doing this systematic review so that they won’t repeat your work or they might contact you to get involved or get updates.