OBJECTIVES:

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement, first published in 2009 by Moher et al., has been widely adopted by researchers for reporting of systematic literature reviews (SLRs). Since then, research methods and relevant technologies for SLRs, i.e., artificial intelligence (AI), have also evolved, becoming a key component of evidence to support clinical decision-making and guideline development. As such, the PRISMA statement was updated in 2020 to reflect these advances and ensure more comprehensive, detailed reporting of SLR methodology and findings to better facilitate replicability. This study aimed to assess the uptake of the PRISMA 2020 statement in recently published SLRs and determine whether AI-assisted screening is being reported and how it is being used.

METHODS:

Searches were performed in Embase and MEDLINE (via OvidSP) for SLRs published in 2022 using a combination of multi-purpose (.mp) terms including “systematic review”, “PRISMA”, and “artificial intelligence”. Records were reviewed and assessed for reporting of methodological details related to the use of AI for title/abstract screening and whether the authors stated they followed the PRISMA 2020 statement.

RESULTS: In a preliminary review of 68 SLRs, 37 (54%) did not follow the PRISMA 2020 statement. Of the 31 SLRs that did, 18 cited the new guidance by Page et al. 2021, 19 used the updated flow diagram, and 2 provided only a PRISMA 2020 checklist. Three used automation tools to exclude studies before screening and 3 stated screening was conducted without use of AI tools.

CONCLUSIONS:

Since its publication in March 2021, the updated PRISMA statement has not been consistently adopted by systematic reviewers. AI software or automation tools are not widely used in SLRs, or reporting lacks such transparency. Further review of the published literature is needed to obtain a better understanding of how the updated guidance is being applied.