PRISMA Flow Diagram (2020): How to Fill In Every Box

What is a PRISMA flow diagram? ¶

Journals reject manuscripts that still use the 2009 template. Check the box labels before you submit.

A PRISMA flow diagram is a standardized chart that reports the flow of information through the phases of a systematic review. It shows the number of records identified, the number removed before screening, the number screened and excluded, the number of full-text reports assessed and excluded with reasons, and the final number of studies included. It is item 16a of the PRISMA 2020 checklist.

PRISMA stands for Preferred Reporting Items for Systematic reviews and Meta-Analyses. The current version, PRISMA 2020, was published by Page and colleagues in 2021 and replaced the 2009 statement. The official templates are free to download from the PRISMA website in four versions: new or updated reviews, each with or without searches of other sources.

Records, reports and studies: get the terms right ¶

PRISMA 2020 uses three words precisely, and most counting errors come from mixing them up.

One study can have several reports, for example a trial protocol, a main results paper and a follow-up paper. That is why the final box asks for both studies and reports.

The PRISMA 2020 flow diagram boxes, stage by stage ¶

The template for new reviews has two columns. The left column is titled Identification of studies via databases and registers. The optional right column is titled Identification of studies via other methods.

Identification

• Records identified from: Databases (n = ) and Registers (n = ). PRISMA recommends listing the number from each database or register where feasible.
• Records removed before screening: Duplicate records removed (n = ), Records marked as ineligible by automation tools (n = ), Records removed for other reasons (n = ).

Screening

• Records screened (n = ) leading to Records excluded (n = ). If automation tools were used, say how many were excluded by a human and how many by the tool.
• Reports sought for retrieval (n = ) leading to Reports not retrieved (n = ).
• Reports assessed for eligibility (n = ) leading to Reports excluded, listed by reason with a count for each (Reason 1 (n = ), Reason 2 (n = ) and so on).

Included

• Studies included in review (n = ) and Reports of included studies (n = ).

The other-methods column covers records identified from websites, organizations, citation searching and similar sources. It starts at Reports sought for retrieval, then Reports not retrieved, Reports assessed for eligibility and Reports excluded with reasons, and feeds into the same Included box.

The templates for updated reviews add boxes for studies and reports included in the previous version of the review, and a total for the updated review.

How to count records at each stage ¶

1. Add up raw search results. Record the number retrieved from each database and register on the day you ran the search, before deduplication.
2. Remove duplicates. Report the number removed by your reference manager or screening tool. Records screened equals records identified minus all records removed before screening.
3. Screen titles and abstracts. Records excluded is simply records screened minus those you move to full-text review. You do not need to give reasons at this stage.
4. Retrieve full texts. Reports sought for retrieval equals records screened minus records excluded. Count reports you could not obtain.
5. Assess full texts. Reports assessed equals reports sought minus reports not retrieved. Give one main reason per excluded report and make sure the reasons add up to the total excluded.
6. Group reports into studies. Link multiple reports of the same study, then count studies and reports separately.

Worked example: filling in the diagram ¶

Here is a complete set of numbers for a hypothetical review of exercise programs for type 2 diabetes.

Check: 148 minus 122 gives 26 included reports from the main column, and 8 minus 4 gives 4 from other methods.

IDENTIFICATION OF STUDIES VIA DATABASES AND REGISTERS

Identification
  Records identified from:
    Databases (n = )  [list each database: n = ]
    Registers (n = )  [list each register: n = ]
  Records removed before screening:
    Duplicate records removed (n = )
    Records marked as ineligible by automation tools (n = )
    Records removed for other reasons (n = )

Screening
  Records screened (n = )  ->  Records excluded (n = )
  Reports sought for retrieval (n = )  ->  Reports not retrieved (n = )
  Reports assessed for eligibility (n = )  ->  Reports excluded:
    Reason 1 (n = )
    Reason 2 (n = )
    Reason 3 (n = )

IDENTIFICATION OF STUDIES VIA OTHER METHODS
  Records identified from:
    Websites (n = )
    Organisations (n = )
    Citation searching (n = )
  Reports sought for retrieval (n = )  ->  Reports not retrieved (n = )
  Reports assessed for eligibility (n = )  ->  Reports excluded:
    Reason 1 (n = )
    Reason 2 (n = )

Included
  Studies included in review (n = )
  Reports of included studies (n = )

Arithmetic checks:
  Records screened = identified - removed before screening
  Reports sought = records screened - records excluded
  Reports assessed = reports sought - reports not retrieved
  Included reports = reports assessed - reports excluded (both columns)

The PRISMA 2020 checklist at a glance ¶

The flow diagram is one part of the PRISMA guidelines. The PRISMA 2020 checklist has 27 items, several with sub-items, and a separate 12-item checklist for abstracts.

Most journals ask you to upload the completed checklist with page numbers. For the review process itself, see our guide to conducting a systematic review.

PRISMA-ScR and PRISMA-P ¶

PRISMA-ScR is the extension for scoping reviews (Tricco and colleagues, 2018). It has 20 essential items and 2 optional items on critical appraisal. Scoping reviews also report a flow diagram of source selection, usually adapted from the PRISMA template. Our scoping review guide covers the method.

PRISMA-P (Moher and colleagues, 2015) is the checklist for systematic review protocols. It has 17 numbered items (26 including sub-items) and is used before the review starts, often alongside PROSPERO registration. PRISMA-P has no flow diagram, because no studies have been screened yet.

Other extensions exist for network meta-analyses (PRISMA-NMA), diagnostic test accuracy reviews (PRISMA-DTA), individual participant data (PRISMA-IPD) and literature searches (PRISMA-S).

Common PRISMA flow diagram mistakes ¶

• Using the 2009 template. The old boxes such as "Studies included in quantitative synthesis (meta-analysis)" are no longer part of PRISMA 2020.
• Counts that do not add up. Each box should equal the box above it minus the exclusions beside it.
• Mixing records, reports and studies. Reporting 30 "studies" when you have 30 reports of 24 studies.
• Exclusion reasons without numbers, or reasons that sum to more than the total excluded.
• Giving reasons at title and abstract screening. This is not required and clutters the diagram.
• Leaving out registers or other methods that the methods section says you searched.
• Numbers that differ from the abstract and results text. Reviewers check.

Free generators such as the PRISMA2020 Shiny app by Haddaway and colleagues, and the diagrams built into Covidence, reduce layout errors, but you still need to check the arithmetic. If you would like an expert to check your diagram, checklist and methods text together, our systematic review help service does exactly that.