Screening is the stage of your review where you decide which papers to include and which to exclude from your review.
You do this by looking at the title and abstract, then the full text of the papers, and comparing with your inclusion and exclusion criteria.
You should expect to screen all the papers yourself, and you may also get another student or your supervisor to screen some or all of the papers well.
Screening can be time-consuming if you have a lot of search results to screen - you should budget approximately one minute per paper, so 60 papers per hour.
Oxford has a subscription to the systematic review software, Covidence, and we highly recommend you use it for your review.
Set up an account with your Oxford Single Sign On (https://app.covidence.org/sign_in/sso) and click Start a new review. Give it a name and some details, click Create Review.
At this stage, you will have run your searches in different databases and exported and saved the results.
Because the contents of databases overlap with each other, your results will contain duplicate copies of the same article.
Covidence identifies the most obvious duplicates and removes them for you, saving you time while screening.
To do this, use the Import screen to add each of the files you saved from the databases. Covidence will automatically deduplicate your results.
You may find extra duplicates while you are screening - there is a button to flag them as duplicates at that point.
Note: Sometimes a study may have multiple articles associated with it - perhaps the patients were assessed at different times for follow-up, or the study had several rounds of recruitment, or an article was published as a pre-print and then in a journal. These different records are not true duplicates (exact copies of the same publication details) and you shouldn't remove them.
For published systematic reviews, each paper needs to be screened by at least two people. For master's projects, it is common for students to screen all the papers themselves against the inclusion and exclusion criteria. Check your course handbook and speak to your supervisor about what is expected of you. They may suggest pairing up with another student on your course to screen each other’s papers, or your supervisor may offer to screen a certain percentage of your papers.
There are two stages of screening which Covidence will guide you through. First, look at the title and abstract of each paper and see if it can be immediately excluded based on your inclusion and exclusion criteria. A large number of your results can usually be excluded at this stage.
Once the title and abstract screening is complete, you will move on to the full text screening. Covidence will look for openly available copies of the articles you have selected, but you may need to check SOLO for subscription access PDFs.
When you exclude a paper at this stage, you need to record a reason (based on your exclusion criteria) - you can do this in Covidence.
If more than one person is screening, you are likely to have some "conflicts" - where you have made different decisions. You may want to try screening a few records each and check that you both understand the criteria and generally agree. You can resolve conflicts by having a discussion between screeners until you reach a consensus, or having an additional person decide.
If you are struggling to find the last few PDFs for your full-text screening, you can contact the library team by emailing hcl-enquiries@bodleian.ox.ac.uk
If you have more than 10 articles you can't find, please put the publication details into a spreadsheet and email it.
The reference checking team will advise you how to access the PDFs - if they are older or more obscure articles, you may need to place interlibrary requests for them.
Make sure you check SOLO for the article before contacting the library to save yourself time. You can also check PubMed and Google.
Once you have a set of papers which you want to include in your review ("seed references"), you can use citation searching to find articles which cite them or are cited by them. This is known as citation searching, citation chaining, citation tracking or snowballing.
You typically perform citation chaining after you've completed your main database searches and have a list of initial studies that meet your criteria. Start with your "seed references" - the studies you already found that look relevant. Remember to keep good notes of what you did and what you found.
This involves looking at the reference lists of the studies you've already included in your review. By doing this, you can find older, foundational research that your "seed references" built upon.
This involves identifying newer articles that have cited the studies you've already included in your review. This helps you discover more recent research that may have been influenced by or responded to your initial set of studies.
Traditional searches in databases might miss some important studies. Citation chaining helps you catch as much relevant evidence as possible, making your review more comprehensive and reliable. It's particularly useful for research questions that are complex and where the search terms are broad or ambiguous.
You can use a free tool called Citation Chaser to do citation searching on your included papers - just put in the DOIs or PubMed IDs of the papers and download two .ris files, one for backwards citations (the References tab) and one for forwards citations (the Citations tab).
You can then go back to Covidence and upload the .ris files using the Import button (choose citation searching as the source). Covidence will deduplicate the results again so that you don't screen papers you have already seen.
Then, go through the same process of title and abstract and full text screening on these additional references, to see if there are any new papers you want to include.