PRISMA Flow Diagram

The PRISMA flow diagram is a standardized graphical representation that documents the number of records identified, screened, assessed for eligibility, and included or excluded at each stage of a systematic review. It provides a transparent, visual summary of the study selection process, allowing readers to understand how the final set of included studies was determined. The PRISMA 2020 update revised the flow diagram to accommodate searches from databases, registers, and other sources separately.

The study selection process in a systematic review involves numerous decisions, from initial search results through screening and eligibility assessment. Without clear documentation, readers cannot evaluate whether the review team cast a sufficiently broad search net, applied eligibility criteria consistently, or might have missed relevant studies. The flow diagram serves as a concise accountability tool that makes the entire selection process transparent in a single figure. PRISMA 2020 asks authors to describe the results of the search and selection process, ideally using a flow diagram, and readers and reviewers can use the diagram to check the selection numbers.

The PRISMA 2020 flow diagram is structured in three stages: identification, screening and included (the 2009 diagram had a separate eligibility stage, which 2020 folds into screening). The identification stage reports the number of records identified from each source category (databases, registers, and other sources such as citation searching, hand searching, and expert consultation) and, for the database and register searches, the records removed before screening: duplicates, records marked as ineligible by automation tools, and records removed for other reasons. The screening stage documents how many records were screened by title and abstract and how many were excluded, then how many full text reports were sought for retrieval, how many could not be retrieved, how many were assessed for eligibility, and how many were excluded after full text review, with the reasons for exclusion and a count for each. The included stage reports the final number of studies included in the review and the number of reports of those studies, since one study may produce multiple publications. The 2020 diagram no longer has a box for studies included in a quantitative synthesis, because a review typically includes several syntheses, each with its own number of studies.

Researchers should ensure that the numbers reported at each stage are internally consistent: the records entering each stage minus those excluded should equal the records progressing to the next stage. The 2020 template includes separate flows for studies identified through databases and registers versus other methods, acknowledging that non database sources (reference lists, expert recommendations, grey literature) follow a different path: their records go straight to reports sought for retrieval, without a title and abstract screening box or a duplicates box.

The flow diagram can be created using templates available from the PRISMA website, or generated using software tools. Many systematic review management platforms produce PRISMA flow diagrams automatically from the screening data.

Include a PRISMA flow diagram in a systematic review or a meta analysis based on a systematic search; PRISMA 2020 recommends one for reporting the search and selection results. It should also accompany scoping reviews (using the PRISMA-ScR extension template) and other evidence syntheses that involve structured study selection. The diagram should be completed after the study selection process is finalized and updated if any changes occur during revision. It is not needed for narrative reviews or commentaries that do not follow a systematic search and selection process.

A common mistake is presenting numbers that do not add up across stages, creating confusion about where records were gained or lost. Another frequent error is providing only a single generic exclusion reason at the full text stage (such as "did not meet inclusion criteria") instead of specific reasons with counts (such as "wrong population, n = 12; wrong intervention, n = 8; wrong outcome, n = 5"). Some researchers use outdated flow diagram templates that do not match the PRISMA 2020 structure. Finally, failing to account for multiple reports from the same study can inflate the apparent number of included studies, and the diagram should distinguish between reports and studies at the inclusion stage.

A systematic review of cognitive behavioral therapy for chronic pain identifies 2,847 records from database searches. After removing 612 duplicates, 2,235 records are screened by title and abstract, and 2,095 are excluded. Of 140 full text reports sought, 8 cannot be retrieved. The remaining 132 reports are assessed for eligibility; 80 are excluded (36 wrong comparator, 24 wrong outcome measure, 12 wrong population, 8 conference abstracts with insufficient data), leaving 52 reports. In a separate column, citation searching and expert recommendations identify 20 records, all of which are sought as reports; all 20 are retrieved and assessed, and 14 are excluded (6 wrong comparator, 4 wrong outcome measure, 2 wrong population, 2 conference abstracts with insufficient data), leaving 6 reports. The final inclusion is 58 reports representing 51 distinct studies. The flow diagram presents these numbers in the standardized PRISMA 2020 template; the 38 studies that provide sufficient data for meta analysis are reported with that synthesis rather than in the diagram.