I see in the methods of review articles and meth-analyses (PRISMA flow diagram) that authors usually claim that they excluded numerous duplicated articles that had been found in several databases. However, the numbers (sometimes more that 1000 articles) are extremely high which make it impossible to procced the process manually. Is there any was to assess the duplicated articles to expedite the process?
Such endeavor is not impossible manually, especially if it is well-planned and coordinated across members of a systematic review team. However, doing so is made substantially easier through the use of reference management softwares, particularly the proprietary EndNote and the purchased version of Mendeley. You can find published studies optimizing the utility of these softwares through addition/modification especially for the former. Freewares are also available, such as Zotero, but by experience they are less reliable for this particular task (they remain extremely useful for standard reference manager work like citations, though).
It is not impossible, however, I would recommend using several applications for that. Simple Google Sheets or Excel could be used to deduplicate all articles you have found from your queries. For paid service, which I am using right now, you can use DistillerSR to ease up the process. Reference manager can also be used to speed up the process.
R Programming may be useful to screen duplicates
revtools: revtools: Tools to support reviews and evidence synthesis in revtools: Tools to Support Evidence Synthesis
https://rdrr.io/cran/revtools/man/revtools.html
In addition to what others have said, you could remove duplicates in OVID. Reference management softwares and COVIDENCE could do much of the work as well. You will need to remove a few manually as others noted.
Endnote works very well for this, and allows syncing between users so you are all working in the same updated database.
Just to update this given more recent developments, I highly highly recommend using revtools for R for deduplication. It is far more accurate and sensitive, particularly the fuzzy logic option, than reference managers and is completely automated.
Generally the easiest way to do it is to combine the reference lists from the various databases in a reference manager into a single RIS file and then deduplicate that in revtools. This then outputs a new deduplicated single RIS file that you can use for screening abstracts.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis