PRISMA Diagram: Build the Systematic Review Flow

A PRISMA diagram is a single flowchart that shows how many studies you found, screened, excluded and finally included in a systematic review, and why records dropped out at each step. It turns the messy reality of searching, de-duplicating and reading through hundreds of papers into an auditable trail that a reader can follow from top to bottom. If someone wants to check whether your review was thorough and reproducible, the PRISMA diagram is usually the first thing they look at.

PRISMA stands for Preferred Reporting Items for Systematic Reviews and Meta-Analyses, an international reporting guideline developed to reduce selective and incomplete reporting in evidence synthesis [1]. The flow diagram is one component of that guideline, sitting alongside a checklist of items you are expected to report in the text. The current version, PRISMA 2020, updated the original 2009 statement to reflect changes in how reviews are now conducted, including automated searching and the use of registers [2].

The practical value is honesty. A well built diagram makes it hard to quietly discard inconvenient studies, because every excluded full text has to be counted and given a reason. That is why journals, editors and peer reviewers increasingly treat the figure as mandatory rather than optional, and why the Cochrane Handbook and most methodological guidance point authors towards it [3].

For most Australian researchers, the diagram is not something you draw once at the end. It is built as you go, updated each time you run or rerun a search, and finalised only when screening is complete. Getting the structure right early saves considerable pain later, and this page walks through what a PRISMA diagram contains, the different templates available, how to complete one accurately, and where the common mistakes hide. If you want the full end-to-end walkthrough of the standard, our PRISMA guide covers it in depth.

Key takeaways

• A PRISMA flow diagram is a standardised chart that traces every record through a systematic review, from initial identification to final inclusion, along with the reasons studies were removed at each stage.
• The current standard is the PRISMA 2020 template, which replaced the 2009 version and added a separate box for records found through methods other than database searching.
• Numbers must reconcile across the diagram: records excluded plus records carried forward should always account for the records entering each phase, and mismatches are the most common reporting error.
• Two main templates exist, one for brand new reviews and one for updated reviews, with variants depending on whether you searched databases only or also registers and other sources.
• Free generators and reference managers can build the figure quickly, but the reviewer remains responsible for the accuracy of every count.

What a PRISMA Flow Diagram Is

The PRISMA flow diagram is a vertical flowchart divided into phases: identification, screening, and inclusion. Each phase is a box or set of boxes carrying a count, with arrows showing records moving downward and exclusion boxes branching off to the side. Its job is descriptive, not analytical. It does not tell the reader whether the included studies were any good, only how you arrived at the final set. The related concept people often confuse it with is the risk-of-bias summary, which assesses quality rather than flow.

It also differs from the reporting checklist. The checklist is a list of text items you must report; the diagram is the numeric audit trail. Both belong to the same PRISMA 2020 statement but do different work.

Prisma Diagram Compared

The table below sets out how the PRISMA diagram relates to the tools it sits beside.

In short, the diagram answers what got in and what did not, while the other tools answer what the evidence means [2].

Components and Structure of a PRISMA Flow Diagram

A PRISMA diagram is built from three linked phases. The identification phase records how many entries you retrieved from each database and any other source, then how many duplicates you removed before screening began. The screening phase splits into two steps: title and abstract screening, where records are excluded in bulk, and full-text assessment, where each excluded report must be tagged with a specific reason such as wrong population, wrong study design or no relevant outcome. The inclusion phase gives the final count of studies, and where relevant the number of reports linked to those studies, since one study can generate several papers.

The defining feature of the PRISMA flow is that the arithmetic has to balance. The records entering a phase must equal the records excluded plus the records passing through. A number that appears without explanation is the single clearest sign the figure was assembled carelessly [4].

The 2020 layout also separates records found through database and register searching from those found through other methods, such as citation chasing, contacting authors or checking reference lists. Keeping these streams distinct matters because they carry different risks of bias, and readers need to see which studies came from a structured search versus an opportunistic one [2].

The Purpose of Using a PRISMA Flow Diagram

The purpose of a PRISMA diagram is reproducibility. Another researcher should be able to read the figure, rerun your search strategy, and arrive at a comparable set of studies. Without that trail, a review is just an opinion with citations attached. The diagram forces you to account for every record, which is why it curbs the temptation to quietly leave out studies that complicate the conclusion [1].

It also serves the reader who needs to judge quality at a glance. A review that screened 40 records and included 38 looks very different from one that screened 4,000 and included 12, and the PRISMA flow makes that visible immediately. Editors and peer reviewers use it as a quick check on whether the search was broad enough and whether exclusions were handled transparently.

There is a practical Australian dimension too. Health Technology Assessment submissions and many funding bodies expect PRISMA-compliant reporting, and grant reviewers increasingly treat a missing or unbalanced diagram as a red flag. Beyond publication, the figure protects you: when a decision is challenged years later, the diagram is the record of exactly what you did and why. Building it properly is inseparable from doing the review properly, which is why it is embedded in reporting guidance rather than bolted on afterwards [3].

Types of PRISMA Flow Diagrams

There are two official PRISMA 2020 diagram formats, and choosing correctly matters because they are not interchangeable. The first is the diagram for new reviews, used when you are conducting a systematic review from scratch. The second is the diagram for updated reviews, which adds boxes to distinguish records identified in the previous version from records found in the new search, so readers can see what the update actually added.

Within each of those, there are two variants depending on your sources. One variant is for reviews that searched databases and registers only. The other adds a parallel column for records identified through other methods, such as reference list checking, citation searching, contacting study authors or searching organisational websites. If any of your included studies came from outside your formal database search, you must use the version with the extra column, because collapsing everything into one stream hides where studies came from.

A common error is using the 2009 diagram out of habit. The 2009 version predates the register and other-methods distinction and is no longer the current standard, so a 2009 layout in a recent review signals that the reporting is out of date [2]. Matching the template to your actual method is the first step in placing your own review correctly.

How to Build and Complete a PRISMA Flow Diagram

Build the PRISMA flow diagram alongside the review, not after it. Start by recording the raw number of records retrieved from each database separately, and note the date of every search so the figure can be reproduced. Export everything into a reference manager, then de-duplicate and record how many duplicates were removed before screening. This first count is where most errors originate, so keep the exports.

Move to title and abstract screening and log how many records you excluded. At full-text stage, retrieve each remaining report and, for every one you exclude, record a specific reason. Vague labels such as not relevant will not survive peer review; use categories tied to your eligibility criteria, such as ineligible population, wrong comparator or conference abstract only. Screening two reviewers independently and reconciling disagreements strengthens this stage; our practical guide on how to screen studies for a systematic review sets out the workflow.

Finally, enter the count of studies included in the review and, if reports and studies differ, both figures. Before you submit, add every number in each phase and confirm the totals reconcile. If records in do not equal records excluded plus records forward, the diagram is wrong. Keeping a live audit log as you screen means the figure assembles itself, rather than forcing you to reconstruct counts from memory at the end [4].

Best Practices for Avoiding Common Reporting Errors

The recurring failures in PRISMA reporting fall into a few classes, and each has a preventable cause. Arithmetic mismatches are the most frequent: the numbers entering a phase do not equal the exclusions plus the records carried forward. Reconcile the totals in every box before submission, ideally with a spreadsheet that recalculates automatically as counts change.

The second class is missing exclusion reasons at full-text stage. Every full text you reject must carry a specific, eligibility-linked reason, and the sum of those reasons must match the total excluded. Recording reasons at the moment of exclusion, rather than reconstructing them later, prevents this.

A third class is the wrong template, most often a 2009 diagram used for a recent review, or the database-only version used when studies actually came from citation chasing. Match the template to your real method and review type.

A fourth class is conflating reports and studies. Several papers can describe one study, and counting them as separate studies inflates your final number. State clearly how many studies and how many reports you included.

Finally, keep the search dates and per-database counts, since a reader cannot reproduce a diagram that only gives a combined total. Independent double screening and a documented audit trail address most of these errors at source rather than at proof stage [3].

When to Seek Expert Help With Study Flow Diagrams

Most reviewers can complete a study flow diagram without specialist support, but some situations genuinely warrant expert help. If your numbers will not reconcile no matter how many times you check them, the problem is usually upstream in de-duplication or in double counting reports as studies, and an information specialist or research librarian can trace it faster than repeated manual recounts.

Complex reviews are the clearest case for early advice. Living systematic reviews, reviews with multiple linked publications per study, network meta-analyses, and reviews drawing on registers and unpublished data all strain the standard diagram, and getting the structure agreed before screening saves rework. Most Australian universities provide systematic review librarians who will advise on search strategy and reporting at no cost, and using them early is far cheaper than fixing a rejected manuscript.

Seek help too when a review is heading for a high-stakes destination, such as a guideline, a Health Technology Assessment or a regulatory submission, where reporting standards are audited closely. Peer review comments that repeatedly flag the flow diagram are a signal that the underlying method needs review, not just the figure. If you are still deciding whether your project is a review or an analysis at all, our explainer on systematic review versus meta-analysis can help you place it before you start drawing.

Tools and Generators for Creating Flow Diagrams

Several tools can produce a compliant flow diagram, and the right one depends on how much of your workflow is already digital. Free web generators let you type your counts into the official PRISMA 2020 templates and export a publication-ready figure, which suits smaller reviews done largely by hand. Reference managers and dedicated review platforms go further, tracking records through de-duplication and screening automatically so the figure populates from your actual data rather than numbers you retype.

The advantage of a platform that carries counts through from search to inclusion is that the arithmetic reconciles by design, which removes the single most common reporting error. Systematicly builds the diagram from your screening decisions as you work, so the counts stay in step with what you have actually done. Our tools comparison sets out how the options differ on automation, export formats and register support.

Whatever you choose, the tool does not absolve you of responsibility. A generator will happily produce a beautifully formatted diagram from numbers that do not add up, because it cannot see your source exports. Check every count against your reference manager before you export, and keep the underlying data in case a reviewer asks. Automation reduces transcription error; it does not replace the reviewer's judgement about what those numbers mean [4].

The Role of a Study Selection Diagram in Systematic Reviews

Within a systematic review, the study selection diagram is the bridge between your search and your findings. Everything upstream, the protocol, the search strategy and the eligibility criteria, feeds into it, and everything downstream, the data extraction, risk-of-bias assessment and synthesis, flows out of the studies it identifies. It is the point where the review's methods become visible as a single set of numbers.

Its role is also evidentiary. Because the diagram documents exactly which studies were considered and rejected, it is what allows a reader to trust that the conclusions rest on the whole relevant literature rather than a convenient subset [1]. That is why it sits at the front of the results section in most reviews and why it is scrutinised so closely in peer review.

Increasingly, screening is assisted by automation, and validation work shows machine-assisted abstract screening can reach high sensitivity while cutting workload, though it still feeds into the same transparent selection diagram as this validation study reports. The technology changes how records are sorted; it does not change the requirement to account for every one of them.

Summary

A PRISMA diagram is the audit trail of a systematic review: it shows how many records you found, how many you screened, why you excluded studies, and how many you finally included. Its value lies in reconciled arithmetic and specific exclusion reasons, so choosing the correct PRISMA 2020 template for your review type and sources matters as much as drawing it neatly. Build it as you screen rather than reconstructing it afterwards, lean on your institutional librarian for complex reviews, and check every count against your source data before submission. Done well, the figure lets any reader reproduce your work, which is the whole point of evidence synthesis. The Systematicly platform can keep those counts in step as you go.

Frequently asked questions

What is a PRISMA diagram?

A PRISMA diagram is a flowchart that maps how records move through a systematic review, from the total identified in your searches, through title, abstract and full-text screening, to the studies finally included, recording the number excluded at each stage and the reasons for full-text exclusions.

What are the 4 steps of PRISMA?

The four phases are identification, where you record all records retrieved and duplicates removed; screening, where records are excluded on title and abstract; eligibility, where full texts are assessed and each exclusion is given a specific reason; and inclusion, where the final count of studies entering the review is reported.

How do I make a PRISMA diagram?

Record your per-database search counts and de-duplication figure, then log exclusions at title, abstract and full-text stages with reasons for each full text rejected. Enter these into an official PRISMA 2020 template using a free generator or a review platform, and confirm every phase reconciles before exporting.

What does PRISMA stand for?

PRISMA stands for Preferred Reporting Items for Systematic Reviews and Meta-Analyses. It is an international reporting guideline, updated to the PRISMA 2020 statement, that sets out the minimum items authors should report so that reviews are transparent, complete and reproducible.