Home › Demo
What a defensible PRISMA figure looks like
Most PRISMA generators, including the free ones, draw whatever numbers you type. This is what it looks like when the numbers are counted from decisions you actually made and can still be asked about later.
1. Two searches, logged
- Example database A · 2026-03-01 · 7 records
- Example database B · 2026-03-02 · 5 records
2. Twelve records, screened one at a time
- Example record 1 — included at full textIncluded
- Example record 2 — included at full textIncluded
- Example record 3 — excluded, wrong populationExcluded at full textWrong population
- Example record 4 — excluded, wrong populationExcluded at full textWrong population
- Example record 5 — excluded, wrong study designExcluded at full textWrong study design
- Example record 6 — full text could not be obtainedFull text not retrieved
- Example record 7 — excluded at title screenExcluded at title
- Example record 8 — excluded at title screenExcluded at title
- Example record 9 — excluded at title screenExcluded at title
- Example record 10 — duplicate of record 1Duplicate
- Example record 11 — duplicate of record 2Duplicate
- Example record 12 — not screened yetNot screened yet
3. One record, and every decision ever made about it
This is the part a form cannot do. The researcher excluded this record on its title, then reinstated it after reading the abstract. The earlier decision is still on file — reversing a decision writes a new entry, it does not erase the old one.
- imported→Not screened yetimported
- Not screened yet→Excluded at titlescreened on titles
- Excluded at title→Passed title screenreinstated after reading the abstract
- Passed title screen→Includedincluded after full text
4. The figure, counted rather than typed
- 12Records identified
- 2Duplicates removed
- 10Records screened
- 3Excluded at title
- 6Reports sought
- 1Not retrievednot an exclusion
- 5Assessed for eligibility
- 3Excluded at full textwith reasons
- 2Studies included
Exclusion reasons, broken down
PRISMA requires full-text exclusions to be reported with reasons. These are grouped from what was entered against each record, not typed into a box.
- Wrong population 2
- Wrong study design 1
1 record has not been screened yet, so this figure says so. A half-finished review that renders as a finished one with small numbers is the failure this is built to avoid.
Try the parts that need no account
The tools are free and finished. The screening above is the part that needs a workspace.