Evidence Synthesis Methods Interest Group

  • 1.  Testing and reporting an updated search

    Posted Mar 04, 2025 12:44 PM

    Hi all,

     

    I initially did a search for a scoping review in 2021. The PI wasn't able to complete the project at that time, but is interested in updating the search and completing the project in 2025. In this case we had a relatively low yield for our search and we're all interested in increasing the precision of our search, likely from reducing the number of search terms from the least defined concept.

     

    I wondered if anyone had experience or had read others' methods for updating *and* revising a search. What considerations are there for revising a search after conducting it the first time? We have an idea to test the search with reduced search terms, ensuring that the included studies from the initial search would still be included. Have you done this before and do you have any methods you can share?

     

    Looking forward to hearing about your experience and seeing any relevant citations!

     

    Thanks in advance,


    Michael Peper

    Head of the Center for Faculty Initiatives

    University of Kansas Libraries




  • 2.  RE: Testing and reporting an updated search

    Posted Mar 05, 2025 12:35 PM

    Hi Michael,

    Great question and I'd be very curious to hear other's thoughts on this. I have a similar situation, though in this case, its an update of a search for a review that is ongoing (and will eventually be a living review). We really want to optimize the search as much as possible as the first time around it was very sensitive also with low yield.

    One thing I'm planning to do is a sort of text analysis, identifying any search terms that occurred zero times in our potentially relevant studies and multiple times in excluded studies, and thus simply removing those from the search string. 

    A couple of R packages (with Shiny apps, so you don't need to know R) are available to help with this sort of thing. One is CiteSource which can help determine what sources were most useful (you may be able to exclude certain sources that were searched based on whether they contributed anything useful to the review) and SearchBuildR (more of a text and keyword analysis tool). Note that CiteSource is still a bit in development, and there is a current issue with the Shiny that I'm trying to figure out, which will hopefully be fixed soon.

    Looking forward to additional responses on this topic!

    Sarah



    ------------------------------
    Sarah Young
    Liaison Librarian
    Carnegie Mellon University
    ------------------------------