PRISMA 2020: reporting a systematic review and its flow

Stetleaf Academy

Reviewed by Alex Juusela, MD, MPH, FACOG, Maternal-Fetal Medicine Physician · Last reviewed 2026-11-05

A systematic review needs to explain not just what it found but how the evidence was identified, selected and synthesized. PRISMA 2020 is a reporting framework for making that process visible. This introduction is for authors organizing a review manuscript and its selection-flow account. It is not a manual for conducting a review or a replacement for methodological support.

Why it matters

The PRISMA 2020 statement provides updated guidance for reporting systematic reviews [1]. Its explanation and elaboration gives further guidance and examples [2]. A systematic review can require PRISMA reporting even when no meta-analysis is possible. The absence of a pooled estimate does not remove the need to explain eligibility, searching, selection and synthesis.

Clear reporting lets readers follow the route from the question to the included evidence. It does not establish that the search found every relevant study, that selection decisions were sound or that the evidence is free from bias. A checklist and a flow account are supporting records of the work, not assurances about its conclusions.

Key ideas

Match the scope before starting

Describe the review question and eligibility criteria before presenting the findings. Explain the populations, interventions or exposures, comparators, outcomes and study designs as applicable. Consider whether the review type requires additional or different reporting guidance. Do not call a scoping review a systematic review of intervention effects simply because both use structured searching.

Keep a dated record of the intended methods and any changes. Explain where the protocol or registration can be found, where applicable, and describe relevant amendments [1,2]. A reader should be able to distinguish decisions made before seeing the results from decisions made later.

Describe the search as work that can be traced

Name the information sources, their platforms where relevant and when they were searched. Provide the search strategies and limits with enough detail to understand what was done. State any language or date restrictions rather than leaving readers to infer them. Keep the final search documentation separate from early exploratory searches so that the report describes the search actually used.

A database name alone does not explain a search. Nor does a statement that references were checked describe which reports were checked or how that process contributed studies. Preserve the logs while searching; trying to reconstruct every action at submission invites inconsistencies.

Separate records, reports and studies

A record is an entry found through a source such as a database. A report is a document describing research. A study is the underlying investigation and may have several reports. This distinction matters when counting what reached each stage of selection [1,2]. Two articles about the same trial do not automatically represent two separate trials.

Document duplicate removal and how reports were linked to studies. Keep screening decisions at the appropriate level. A report that was not retrieved is different from a retrieved report excluded after assessment. Do not combine those groups simply to make the flow account shorter.

Make selection counts reconcile

Use the flow structure appropriate to the review and the sources used. Explain how many records were identified, removed before screening, screened and excluded, then how many reports were sought, retrieved and assessed. Give reasons for exclusions at the report-assessment stage and show the included reports and studies [1,2].

This article provides a text-only teaching example, not a reproduced PRISMA diagram. Obtain the relevant diagram through the guideline's materials and check its conditions of use before adding an adapted graphic.

Report extraction, appraisal and synthesis

Explain how data were collected, which outcomes were sought and how disagreements or unclear information were handled. Describe the methods used to assess risk of bias and how those assessments informed interpretation. Report the synthesis methods, including how studies were grouped and how differences between them were considered [1,2].

If meta-analysis was not performed, describe the actual alternative rather than saying only that findings were reviewed narratively. If it was performed, make the analysis choices and uncertainty visible. Keep the number of studies included in the review distinct from the number contributing to a particular synthesis.

Interpret the body of evidence carefully

Distinguish limitations of the included evidence from limitations of the review process. Missing results, uncertain measurements and weaknesses in study methods raise different issues from restricted searching or unavailable reports. Explain the implications without presenting a pooled estimate as a resolution of every limitation. A useful discussion returns to the review question and the evidence that could actually address it.

Worked example

Illustrative example (invented). Not a real study.

An invented review examines appointment reminders. All numbers below are teaching counts, not search results. The team uses only database searching in this simplified scenario.

Identify records. The searches return 600 records. After removing 100 duplicate records, 500 remain for screening.
Screen. Title and abstract screening excludes 440 records. The team seeks 60 reports for retrieval.
Track retrieval. Five reports cannot be retrieved. The remaining 55 are assessed against the eligibility criteria.
Record exclusions. Forty reports are excluded: 20 for an ineligible population, 12 for an ineligible intervention and eight for an ineligible design. Each excluded report receives one primary reason in this example.
Link reports to studies. Fifteen reports remain, describing 12 studies. Three studies each have two reports; nine each have one. The review does not count all 15 reports as separate studies.
Check the synthesis denominator. If only some studies contribute to an outcome, the corresponding results section names that subset rather than implying all 12 contributed.

The arithmetic checks selection reporting only: 55 assessed minus 40 excluded leaves 15 reports. It says nothing about whether the review methods or included evidence support a particular conclusion.

Common pitfalls

Equating papers and studies. Link multiple reports before counting investigations.
Hiding retrieval failures among exclusions. Keep the stages distinguishable.
Describing only the database names. Preserve strategies, limits and dates.
Reporting only a pooled result. Explain selection, appraisal and synthesis methods too.
Treating the flow account as evidence of completeness. Counts document a process; they do not prove that nothing was missed.
Ignoring reviews without meta-analysis. Report their selection and synthesis methods explicitly.

Quick self-check, in our own words

I stated the question and eligibility criteria.
I documented sources, search strategies and search dates.
I distinguished records, reports and studies.
My selection counts reconcile and exclusions have reasons.
I explained extraction, risk-of-bias assessment and synthesis methods.
I distinguished review limitations from limitations of the evidence.
I checked the manuscript against the full reporting guidance.

References

Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. doi:10.1136/bmj.n71. PMID: 33782057.
Page MJ, Moher D, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviews. BMJ. 2021;372:n160. doi:10.1136/bmj.n160. PMID: 33781993.