PRISMA 2020 Flow Diagram Explained Box by Box: How to Report Study Selection in Your Systematic Review

PRISMA 2020 asks authors to describe the results of the search and selection process, from the number of records identified to the number of studies included, ideally using a flow diagram (item 16a) [2]. PRISMA 2020 replaced the 2009 statement; its 27-item checklist comes with revised flow diagrams for original and updated reviews [2], and the explanation-and-elaboration paper explains each item with exemplars [3].

Four templates are available: a new or updated review, each with databases and registers only, or with databases, registers and other sources [1]. Pick the one that matches how you searched and keep it for the whole project. For who completes each box and when, see our PRISMA 2020 flowchart.

The diagram counts three different things, and mixing them is the most common source of wrong numbers [2]:

• Record: the title or abstract (or both) of a report indexed in a database or website.
• Report: a document that gives information about a study, such as a journal article, a registry entry or a clinical study report.
• Study: the investigation itself. One study can have several reports, so reports of included studies can outnumber studies included.

Table 1 lists every box in the template for databases, registers and other sources, where its number comes from in your own records, and the mistake reviewers most often find in it. The same structure applies whatever the evidence synthesis method.

Each box should equal the box above it minus what left the pathway (Table 2). Run these checks on the final counts, then compare the diagram with the text, the included-studies table and the search log. Any mismatch is easy for a reviewer to spot.

The figures in Table 3 are invented to show the arithmetic; they do not come from a real review.

If the reason counts sum to 47 instead of 48, one report is unaccounted for. Finding that gap before submission is the point of the exercise.

Use the new-review templates for a first review. Use the updated-review templates when updating an earlier one, so that studies carried forward from the previous version are shown separately from those found in the new search. Say in the methods which version the numbers belong to, and keep the previous diagram in your records so the two can be reconciled.

• Full-text exclusions without exclusion reasons. PRISMA 2020 also asks you to cite studies that might appear to meet the inclusion criteria but were excluded, and explain why (item 16b) [2].
• Pooled counts. Give separate counts for each database and register; pooled totals hide which sources contributed.
• No other-methods column. Websites, organisations and citation searching need their own path.
• Unnamed automation tools. If a tool removed records before screening, name it and say how it was used.
• Reports counted as studies. The final two boxes count different things [2].

Editors and peer reviewers compare the diagram with the methods and results. They look for totals that reconcile, a count for each database, reasons for full-text exclusions and a clear link to the search report. PRISMA-S item 15 asks for the total records from each database and other source, so the first row of your diagram should match your search log [4]. Duplicate removal is PRISMA-S item 16.

Count the German Clinical Trials Register (DRKS), the WHO primary register for Germany run by the BfArM, as a register in the first box. German-language agency websites belong in the other-methods column, with dates in the search log. Our guide to grey literature lists the German sources worth searching.