PaperfyPAPERFYHome
Start freeLog in
Paperfy Guides
PRISMA flow diagramSystematic review5 minFor anyone writing a PRISMA-compliant review

The PRISMA Flow Diagram:Count From Day One, Don't Draw It at the End

You sit down to write the paper, start on the PRISMA flow diagram — and the numbers don't add up. You can't remember the search hit counts. You can't reconstruct the breakdown of exclusion reasons. That struggle has nothing to do with how to draw the diagram. A PRISMA flow diagram isn't a drawing you make at the end; it's a container for numbers you've been counting all along. This article is about how to count so the diagram never becomes a problem.

Published: 2026-07-22

Turn a paper PDF into a 3-minute radio show

No registration required. Try a paper PDF you want to read.

Try without signing up

In this article

The flow diagram is a flow meter for papersThe numbers the diagram needsCounting from day one, in practiceThe hard part: breaking down full-text exclusion reasonsBeside the diagram sits a checklistGetting familiar with each paper before you decideFinal check: does everything sum top to bottom?Start with Paperfy

The flow diagram is a flow meter for papers

The PRISMA flow diagram shows how many records entered at the search stage, how many were removed at each step, and how many studies remained at the end.

Readers and reviewers use it to gauge the transparency of your review at a glance. A diagram whose numbers don't connect undermines the credibility of the whole review — which is exactly why it deserves care.

The numbers the diagram needs

The general flow: records identified per database → records after deduplication → records screened at title/abstract, with the number excluded → reports assessed at full text, with exclusions broken down by reason → studies included.

The exact box structure depends on the version, such as PRISMA 2020. When you write, always check your target journal's requirements and the latest official PRISMA template. Here we stick to the general idea.

Counting from day one, in practice

It's simple: on the day you run each search, record the hit count per database. On the day you deduplicate, record the count after deduplication. At each screening stage, record how many you excluded.

Just writing down the numbers on the day you do each step turns the flow diagram into a fill-in-the-blanks exercise at write-up. Numbers reconstructed after the fact, by contrast, will always fail to reconcile somewhere.

The hard part: breaking down full-text exclusion reasons

The most time-consuming piece of the diagram is the breakdown of exclusions at the full-text stage. You need not just the total, but a count for every reason.

The fix is to define your exclusion-reason categories in advance (ineligible population, different intervention, outcome not reported) and pick one from the list each time you exclude a paper. Recording the reason on the spot, as you make each decision, means the breakdown is already done when you need it.

Beside the diagram sits a checklist

PRISMA is more than the flow diagram — it also includes a checklist of reporting items. Skim it once during the design phase of your review so it doesn't ambush you at write-up.

Which items apply, and how strictly, depends on the review type and where you're submitting. Confirm with your supervisor or team.

Getting familiar with each paper before you decide

At the full-text stage, you're making decisions that feed directly into the flow diagram. Each exclusion needs a reason; each inclusion needs confidence.

Before you read the full PDF in detail, listen to the paper as a short radio show first — Paperfy turns any research PDF into a roughly 3-minute two-voice conversation covering background, methods, key findings, and limitations. You can try one paper on the landing page without creating an account. Then read the PDF yourself for the figures, numbers, and final judgment. That two-step rhythm keeps decisions grounded and reasons clear.

Final check: does everything sum top to bottom?

Once the diagram is filled in, verify each stage: records remaining + records excluded = records from the stage above.

If even one line doesn't balance, there's a gap in the record somewhere. The moment this check passes is the moment your review's record is complete.

Key takeaways

  • Hit counts per database were recorded on the day of each search.
  • The post-deduplication count was preserved.
  • Full-text exclusions can be counted by reason category.
  • The sums were verified at every stage of the diagram.

Turn a paper PDF into a 3-minute radio show

No registration required. Try a paper PDF you want to read.

Try without signing up

Systematic review

Explore This Topic

Read the full cluster for this workflow, from first steps to practical use.

Browse all guides

Read Next

These guides continue the same workflow.

PICO extraction with AIPICO Extraction With AI: Verify Against the Methods, Don't Just LabelAI paper screeningUsing AI for Paper Screening: Let It Set the Order, Not Make the Callfull-text reviewFull-Text Review: Check What the Abstract Didn't Tell You