When a peer reviewer opens a systematic review, the PRISMA flow diagram is often the first thing they examine, because a figure that does not reconcile signals a review that was not conducted carefully. Reviewers run a predictable set of checks: they verify the numbers subtract cleanly from top to bottom, confirm the full-text exclusion reasons sum correctly, and cross-match the figure against the numbers stated in the methods and the abstract. Knowing exactly what they look for lets you fix problems before submission, and you can rebuild a corrected figure in seconds in our free PRISMA 2020 flow diagram generator. This guide lists the checks in the order reviewers tend to apply them.
The diagram is scrutinised first precisely because it is falsifiable. Prose can be vague, but a flow chart either adds up or it does not, which makes it the fastest way for a reviewer to gauge rigour.
The Reconciliation Check Comes First
The single most common reviewer action is to read the diagram top to bottom and check that every number is accounted for. Records identified minus duplicates removed must equal records screened. Records screened minus records excluded must equal reports sought. Reports sought minus reports not retrieved must equal reports assessed. Reports assessed minus full-text exclusions must equal studies included.
If any link in that chain fails to subtract cleanly, the reviewer flags it immediately, and a single broken subtraction casts doubt on the whole figure. The only place a count is allowed to diverge is the distinction between studies and reports at the inclusion stage, because one study can be described in several reports. Our guide to what goes in each PRISMA box walks through the arithmetic stage by stage so every chain reconciles before a reviewer ever sees it.
The Exclusion-Reason Sum
The second check targets the eligibility phase, where PRISMA requires a reason for every excluded full text. Reviewers add up the exclusion-reason counts and confirm the total equals the number of reports excluded at that stage. If you assessed 132 reports and included 22, the reasons must account for all 110 exclusions, for example wrong population, wrong intervention, wrong outcome, and wrong study design, each with its own count.
A mismatch here is one of the most common PRISMA diagram mistakes, and it is conspicuous because the reviewer only has to add a short list. Vague or overlapping reasons draw the same scrutiny, so each category should be specific and tied to your protocol.
Cross-Matching the Figure Against the Text
Reviewers do not read the diagram in isolation. They compare it against three other places the same numbers appear:
- The abstract, which usually states how many studies were included.
- The methods, which describes the databases searched and the screening process.
- Any summary tables, such as characteristics of included studies, whose row count should match the included number.
When the figure says 18 included studies but the abstract says 19 and the characteristics table has 17 rows, the reviewer cannot tell which number is correct, and the manuscript loses credibility fast. Keeping every surface synchronised is essential, and it is worth a final pass dedicated solely to confirming the included count is identical everywhere.
The Label and Template Check
Beyond the numbers, reviewers confirm the figure uses current PRISMA 2020 structure and labels. They look for the correct phase headings, an "other sources" or registers column when the methods mention citation searching or trial registries, and the updated-review boxes when the review is an update of a prior version. A diagram that still uses the older 2009 layout, or that omits a column for a search method described in the text, signals that the reporting standard was not followed. Choosing the right starting layout avoids this entirely, as our overview of the PRISMA 2020 checklist items explains alongside the rest of the reporting requirements.
Reviewers also notice when a figure looks degraded in the proof. A diagram pasted in as a low-resolution screenshot blurs at print size and undermines an otherwise clean review, so it is worth exporting the corrected figure in a journal-ready vector format such as SVG or PDF, which stays sharp at any scale and is what most journals prefer for line-art figures.
The Comprehensiveness Judgement
Finally, reviewers form a judgement about whether the search was thorough enough to support the conclusions. They look at how many databases the identification box reports, whether grey literature or citation searching appears, and whether the screened volume is plausible for the topic. A review claiming a definitive answer from a single database and a thin screen will be questioned regardless of how cleanly the figure reconciles. The diagram is the evidence of thoroughness, so it should show the full breadth of the search you actually ran.
A Pre-Submission Checklist
Before you submit, run the reviewer's checks yourself:
- Confirm every vertical subtraction reconciles from identification to inclusion.
- Add the full-text exclusion reasons and confirm they equal the total excluded.
- Match the included count across the figure, abstract, methods, and tables.
- Verify the template, labels, and any other-sources column reflect your actual search.
- Sanity-check that the search breadth supports the strength of your conclusions.
Passing these five checks yourself is the most reliable way to keep the flow diagram from becoming the reason a reviewer recommends revision.
Frequently Asked Questions
What do peer reviewers look at first in a systematic review?
Many reviewers examine the PRISMA flow diagram first, because it is the fastest falsifiable check on rigour. They verify the numbers reconcile top to bottom, the exclusion reasons sum correctly, and the figure matches the abstract, methods, and tables.
Do the numbers in a PRISMA flow diagram have to add up?
Yes. Each stage must reconcile with the next: records identified minus duplicates equals records screened, and so on down to inclusion. The only intentional difference is that the number of reports of included studies can exceed the number of included studies.
What is the most common reason a flow diagram is flagged?
The most frequent flags are a vertical chain that does not subtract cleanly and full-text exclusion reasons that do not sum to the total excluded. Mismatches between the figure and the abstract or characteristics table are close behind.
How do I make sure my diagram matches the rest of the paper?
Do a dedicated pass that compares the included-studies count in the figure against the abstract, the methods text, and any characteristics table. The number must be identical in every location, and the databases listed in the figure must match those described in the methods.
Does using an outdated PRISMA template matter to reviewers?
Yes. Reviewers expect the current PRISMA 2020 structure, with correct phase labels and an other-sources or registers column when the methods describe such searching. An older 2009 layout or a missing column suggests the reporting standard was not followed.