PRISMA — Preferred Reporting Items for Systematic reviews and Meta-Analyses — is the guideline for how you report a systematic review (one that collects and appraises the literature exhaustively, by a procedure fixed in advance) or a meta-analysis (one that combines the results of several studies statistically).
It does not tell you how to run the review. It tells you what has to appear on the page afterwards. The current version is PRISMA 2020, published in the BMJ in 2021.
Why the reporting format is worth fixing
What a systematic review is worth rests on its method, not its conclusion. Where did you search, how many records came back, how many did you drop and for what reason? If that is on the page, a reader can walk the same path and check it. If it is not, all a reader can do is decide whether to believe you.
PRISMA is the list of things you have to report for that path to be walkable. The first version, in 2009, grew out of the 1996 QUOROM statement, which covered the reporting of meta-analyses of randomised controlled trials.
The three pieces of PRISMA 2020
| Piece | What it covers |
|---|---|
| Checklist | 27 items to report in the body, grouped into seven sections, several with sub-items |
| Abstract checklist | 12 items to report in the abstract |
| Flow diagram | One figure carrying the counts from search to inclusion |
There is also an expanded checklist: the same 27 items, each with an explanation of what to report and an example of it being reported well.
What goes in each tier of the flow diagram
The diagram has three tiers, top to bottom: identification, screening, included.
Identification takes the counts per place searched. Records found in databases and registers (public listings of planned and ongoing studies) are counted separately from records found by other methods — websites, organisations, citation searching. This tier also carries the records removed before screening, split three ways: duplicates, records a tool flagged as ineligible, and records removed for other reasons.
Screening takes the narrowing. Records screened, records excluded, reports sought for retrieval, reports not retrieved, reports assessed for eligibility, and reports excluded. That last one needs a breakdown by reason. A total on its own leaves the reader unable to tell what you passed over.
Included takes the number of studies included in the review, and the number of reports of those studies.
Records, reports and studies are counted separately
This is the distinction people most often get wrong.
| Term | What one of them is |
|---|---|
| Record | One search hit. The same paper returned by two databases is two records |
| Report | One document about a study — a journal article, but also a preprint, a conference abstract, a registry entry, a clinical study report, a thesis, an unpublished manuscript, a government report |
| Study | One piece of research. A single study can be described in several reports |
The reason the three are kept apart is that one study can produce several reports, and one report can surface as several records. Twelve included studies might be described across eighteen reports, which is why the last tier asks for both numbers.
Four templates are provided: new review or updated review, crossed with searches of databases and registers only or searches that also used other methods. Pick the one that matches your review. Grey boxes in the template are filled in only when they apply; when they do not, delete them from the figure.
What changed in 2020
PRISMA 2020 replaces PRISMA 2009, and reflects eleven years of change in how studies are identified, selected, appraised and synthesised.
- Every checklist item was rewritten, and the expanded checklist was added alongside it.
- The abstract checklist, first published in 2013 as PRISMA for Abstracts, was revised to match the 2020 version.
- The flow diagram was redrawn so that one figure works for both new and updated reviews.
The count is still 27 items, but they are not the same 27. Do not reuse a 2009 checklist.
Choosing an extension
PRISMA 2020 was written with reviews of interventions in mind. For anything else there is an extension, and the PRISMA website lists seventeen of them. Four come up most often:
| Extension | When you need it |
|---|---|
| PRISMA-P | Reporting the protocol, before the review itself |
| PRISMA-ScR | Reporting a scoping review, which maps a field rather than answering one question |
| PRISMA-S | Reporting the search itself in full detail |
| PRISMA for Abstracts | Reporting the abstract |
Others cover diagnostic test accuracy, network meta-analyses, individual participant data and health equity, among other subjects. Check the website before you start writing, rather than after.
Doing the collecting part in one place
The first tier of the diagram asks where you looked and what you found there. In some fields PubMed alone will not do: you end up in preprint servers, conference proceedings and institutional repositories as well. Every extra place you search is another line in that tier.
Litlas is built for that part of the job. One query runs across 16 scholarly sources (OpenAlex, Crossref, DataCite, OpenCitations, ROR/ORCID, arXiv, DBLP, OpenReview, ACL Anthology, PubMed, PMC Open Access, DOAJ, CORE, OpenAIRE Graph, Unpaywall and Common Crawl), and results filter by venue, author or year. What you keep goes into a board for the review with your own notes and tags attached, and exports as BibTeX, RIS or CSL-JSON. Opening the citation graph on any paper shows what it cited and who has cited it since, which turns up work no keyword would have reached.
The free plan needs no card and works right away.
For the screening log and for drawing the diagram itself, pair it with a tool built for that stage. What Litlas covers is the step before: finding the literature and getting it into one place.
Three things to remember
- PRISMA is a reporting guideline. It does not prescribe how to run the review, and it does not score its quality.
- The flow diagram counts records, reports and studies separately. Because one study can have several reports, the two totals in the last tier will not match.
- Excluded reports need a breakdown by reason. A bare total is not something a reader can follow.
Bibliography
- Page, M. J., J. E. McKenzie, P. M. Bossuyt, et al. “The PRISMA 2020 Statement: An Updated Guideline for Reporting Systematic Reviews.” BMJ 372 (2021): n71. https://doi.org/10.1136/bmj.n71
- PRISMA. “PRISMA 2020 Flow Diagram.” https://www.prisma-statement.org/prisma-2020-flow-diagram. Accessed 21 August 2026.
- PRISMA. “PRISMA Extensions.” https://www.prisma-statement.org/extensions. Accessed 21 August 2026.
- Uegaki, H., Y. Kaneko, K. Tsutani, T. Nakayama, and H. Orikasa. “Commentary and Japanese Translation of ‘The PRISMA 2020 Statement: An Updated Guideline for Reporting Systematic Reviews’” (in Japanese). Japanese Pharmacology and Therapeutics 49, no. 6 (2021): 831–842. https://www.lifescience.co.jp/yk/jpt_online/topics/j20210831.pdf