PRISMA extensions

Why extensions exist

A reporting guideline lists what to put in a report. PRISMA 2020 does that for systematic reviews in general, but reviews differ, and some have features that the main checklist does not cover. A network meta-analysis must describe the geometry of its network and its assessment of inconsistency. A review of diagnostic accuracy must describe the index test and the reference standard. A review of individual participant data must say how the data were obtained and checked. Rather than make the main checklist ever longer, the PRISMA group and others have developed extensions, each adding items for a particular type of review or for a particular part of the report.

The extensions are meant to be used with the main checklist and not in place of it. A reviewer reports against PRISMA 2020 and then against the additional items of the extension that applies. Journals increasingly ask for both. The extensions have been developed by different groups at different times, so their format and their degree of updating vary, and some were written before PRISMA 2020 and may be revised. Authors should look for the current version. The guide on PRISMA 2020 describes the main checklist.

The extensions at a glance

The list is not complete, and others exist or are being developed, for example for complex interventions, for living reviews and for specific fields. The EQUATOR Network keeps a current directory.

PRISMA-P and PRISMA-S: parts of the report

Two extensions apply to any systematic review because they concern a part of the work and not a type of review. PRISMA-P lists what a protocol should contain: the administrative information such as registration and amendments, the introduction with the rationale and objectives, and the methods with eligibility criteria, information sources, search strategy, study records, data items, outcomes, risk-of-bias assessment and data synthesis, and the assessment of meta-bias and confidence in the evidence. It is the basis for the registration record and for protocols published in journals. See protocol development and registration.

PRISMA-S addresses the search, which is often reported poorly. It lists items about the information sources and methods, including the databases and platforms, the multi-database searching, the study registries, online resources and web searches, citation searching, contacting experts, and the use of other reviews; the search strategies, including full strategies, limits and filters, and prior work; and the peer review of the search, the de-duplication and the record of the search. It is used with PRISMA 2020, which already asks for full strategies, and adds detail on what to report. See building a search strategy.

Extensions for review types

PRISMA-ScR

For scoping reviews, with items on the rationale, the protocol, the search, the selection of sources, the charting of data and the synthesis, and noting that appraisal of sources is optional. See scoping review.

PRISMA-DTA

For reviews of diagnostic test accuracy, covering the definition of the target condition and tests, the use of a reference standard, the way accuracy was summarized, and the handling of thresholds and heterogeneity. See diagnostic accuracy meta-analysis.

PRISMA-NMA

For network meta-analysis, with items on the geometry of the network, the treatment nodes, the methods for assessing consistency and the ranking of treatments. See network meta-analysis.

PRISMA-IPD

For reviews that obtain and analyze participant-level data, with items on how data were requested, what proportion was obtained, how it was checked and how it differs from published results.

PRISMA-Harms

For reviews that evaluate adverse effects, covering the definition and classification of harms and the way they were synthesized.

PRISMA-Equity

For reviews in which equity is a concern, covering how factors such as place of residence, race, occupation, gender, religion, education, socioeconomic status and social capital were considered.

Walk-through: a network meta-analysis of trials

An example shows how the pieces fit together. A team plans a network meta-analysis of antidepressant trials. At the protocol stage they use PRISMA-P to structure the protocol and register it. When they design the search they follow PRISMA-S, recording each database, the date, the full strategy and the peer review of the strategy. Screening is reported with the PRISMA flow diagram. For the analysis they follow the PRISMA-NMA items: they present the network graph, describe how drugs and doses were grouped into nodes, report the methods for checking consistency and the results of the tests, and show the league table with the certainty of each estimate. The whole paper is reported against PRISMA 2020, with the extension items integrated. At submission they provide one combined checklist that names each source and points to the page where each item is met. The example is illustrative and does not describe a real project.

How the family has developed

The original statement, then called QUOROM, appeared in 1999 for meta-analyses of randomized trials, and was replaced in 2009 by PRISMA, which widened its scope to systematic reviews with and without meta-analysis. Extensions followed as methods diversified, from networks of treatments to participant-level data, and as the reporting of particular elements, such as harms, equity and searches, was found wanting. PRISMA 2020 updated the main statement in light of new methods and tools, and some extensions have been updated or are being aligned with it. The proliferation has a cost: authors face a long list, and different checklists can overlap. The EQUATOR Network, which maintains the library of reporting guidelines, helps to navigate it, and the PRISMA website tracks the family. Authors should expect further change, and check for updates when starting a review.

What editors and reviewers do with the checklists

Editors use the completed checklists as a screening tool, to see whether a review reports the information needed to judge it, and reviewers use them to structure their comments. A checklist whose page references do not lead to the information is quickly detected. When an extension applies and has not been used, reviewers sometimes ask for it, which causes a revision round that could have been avoided. Some journals publish the checklist with the paper as a supplement. It helps authors to treat the checklists as a planning tool and an honest self-audit: if an item is not met, it is better to add the information or to say openly that it was not done, with the reason, than to leave a gap that a reviewer will find. This approach also shortens peer review.

Using more than one extension

A review can need several. A network meta-analysis whose protocol is registered, and whose search is reported in detail, would use PRISMA 2020 for the main report, PRISMA-NMA for the network, PRISMA-P for the protocol and PRISMA-S for the search. A scoping review of harms might combine PRISMA-ScR with attention to the items on harms. In practice, authors complete the main checklist and add the extension items as an addition, or merge them into one table, noting in the manuscript which were used. Where items of different extensions overlap, they are reported once, with a cross-reference. The aim is that a reader can see where each requirement is met, with no requirement missing, and without repeating the same information.

What is not covered, and other guidelines

PRISMA and its extensions concern the reporting of reviews of existing research. They do not cover the reporting of primary studies, for which other guidelines apply, such as CONSORT for trials and STROBE for observational studies. Meta-analyses of observational studies in epidemiology have their own earlier guideline, MOOSE, which is often used with PRISMA. Qualitative evidence syntheses use guidance such as ENTREQ, and realist and meta-narrative reviews use the RAMESES standards. Overviews of reviews use PRIOR. And for diagnostic and prognostic reviews, the primary studies are expected to follow STARD and the reporting standards for prognostic markers and prediction models, which the review authors can use as markers of reporting quality. See reporting guidelines and MOOSE.

How to choose

1. Identify the type of review

   Is it a standard systematic review, a scoping review, a diagnostic accuracy review, a network meta-analysis, an individual participant data review, or another type?

2. Identify the parts that need specific guidance

   Is there a protocol to report? Is the search to be described in detail? Are harms or equity central?

3. Check the journal's instructions

   Journals name the guidelines they require, and may require a specific checklist file.

4. Find the current versions

   Use the official checklist and the explanation and elaboration paper from the PRISMA website or the EQUATOR Network, since versions are updated.

5. Use the checklists from the start

   Plan the report with them, and check against them at the end.

Common mistakes

• Using an extension without the main PRISMA 2020 checklist.
• Using an old version of a checklist when a newer one exists.
• Choosing the extension that is easiest and not the one that matches the review type.
• Completing a checklist after writing without changing the manuscript.
• Calling a review PRISMA-compliant without saying which checklist or extension was used.
• Forgetting that the primary studies have their own guidelines.

Support

Frequently asked questions