Reporting Systematic Reviews: PRISMA 2020 and Its Extensions

A review can be conducted well and still be rejected because it is reported badly. Editors and peer reviewers read the methods against a reporting guideline, and a missing item, such as an unstated effect measure or an unexplained exclusion, reads as a missing step in the work itself.

PRISMA 2020 is the standard for systematic reviews of interventions. Its 27-item checklist covers the title and abstract, rationale and objectives, eligibility criteria, information sources and search, selection and data collection, risk of bias, effect measures, synthesis methods, certainty assessment and the presentation of results, and its flow diagram traces records from identification to inclusion. A separate abstract checklist exists because many readers see nothing else.

Extensions cover what the main statement does not. PRISMA-S sets out how to report literature searches in enough detail to reproduce them. PRISMA-P covers protocols. PRISMA-NMA adds items for network geometry, transitivity and inconsistency. PRISMA-ScR covers scoping reviews, and PRIOR covers overviews of reviews. Most journals ask for the completed checklist with page numbers at submission.

The most common gaps are consistent across journals. The full search strategy for every database is missing, the effect measure is not stated, the synthesis methods do not say how heterogeneity was handled, certainty of evidence is not reported, and the registration number appears only in the cover letter. A final pass against the checklist, item by item, catches nearly all of them, and it is cheaper than a revision round.

The notes in this topic go through the items that most often trigger a revision request, show how to build a flow diagram whose numbers reconcile, explain how to present results so tables and forest plots agree with the text, and describe what to include in supplementary material so the review can be audited.