Evidence Synthesis: Step 4: Deduplication

Why is deduplication necessary?

Running similar searches in multiple databases will produce duplicate results. This is because subject-specific databases often have overlapping content. A single study should not be counted as more than one piece of evidence in your final synthesis. It is important to track how many duplicates are removed in this step; this number will be part of your PRISMA diagram.

How many duplicates should there be?

This is highly dependent on your research topic and on the databases you use. There is no "typical" amount of duplicates, so do not be concerned about having too many or too few. 

What's the process for deduplicating?

There are two options.

1. After you run each database search, you can upload the citation files to a citation management program and delete duplicates manually. 

• If you choose this option, you will need to keep track of these numbers and update your PRISMA diagram manually. 

Here are instructions from HSL on deduplicating in a citation manager:

• Deduplicating results in EndNote
• Deduplicating results in Zotero
• Deduplicating results in Lean Library Workspace (Sciwheel)

2. After you run each database search, you can upload each citation file to Covidence and it will identify duplicates automatically.

• If you choose this option, Covidence will automatically update a downloadable PRISMA diagram. 
• If you are including conference proceedings in your search, Covidence may mistakenly mark these as duplicates. You can "undo" a duplicate in Covidence if you believe it was marked in error.

Tips for Deduplication

• Zotero and EndNote work better than Lean Library Workspace (formerly Sciwheel) for deduplicating results.
  • Lean Library Workspace automatically removes duplicates when citations are uploaded, which makes it tricky to keep track of the numbers for your PRISMA diagram.
• Sometimes researchers publish multiple articles based on the findings of one study (e.g., a conference paper with partial results and a peer-reviewed journal article with final results).
  • The Cochrane Handbook recommends that only one publication per study is included in an evidence synthesis; otherwise, the findings of that one study may appear inflated and create bias in your review. 
  • You can identify these semi-duplicate publications by examining grant/funding numbers, author names, location or setting of the study, and characteristics of the study. You may wish to include all of the semi-duplicate publications in your review, but they should be combined into one study finding in your final synthesis.
• If you prefer to deduplicate in Covidence, it is still a good idea to store copies of the citation files in a citation manager library or in a folder on your computer.  
• An easy way to keep track of your citation files is to name them with the database and number of results. E.g.: PsycInfo-134.ris 

Mitchell, N., & Ennis, L. A. (2022). Evaluating for Inclusion. In M. J. Foster & S. T. Jewell (Eds.), Piecing Together Systematic Reviews and Other Evidence Syntheses (pp. 173-181). Find@UNC