Why Every Systematic Review Needs a PRISMA Flow Diagram
The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram is a mandatory component of systematic review manuscripts. It visually tracks how studies flow through identification, screening, eligibility assessment, and final inclusion.
Without a proper PRISMA flow diagram, your manuscript will be desk-rejected by most medical journals. The EQUATOR Network lists PRISMA as the standard reporting guideline, and editors check for it before sending papers to peer review.
Key fact: A 2023 cross-sectional study found that 68% of systematic reviews published in top medical journals had at least one error in their PRISMA flow diagram. The most common: numbers that don't add up between stages.
PRISMA 2020 vs. 2009: What Changed?
PRISMA 2020 (Page et al., BMJ 2021) is a major update from the 2009 version. If you're submitting a manuscript in 2024 or later, you must use the 2020 version.
| Feature | PRISMA 2009 | PRISMA 2020 |
| Flow diagram structure | Simple four-tier | More detailed; new "Previous studies" source |
| Database reporting | Merged total | Individual counts per database |
| Exclusion reasons | Optional | Mandatory with individual counts at full-text stage |
| Automation tools | Not addressed | New "excluded by automation tools" category |
| Registration | Recommended | Strongly recommended (PROSPERO, etc.) |
| Checklist items | 27 items | 27 items (content substantially updated) |
Common rejection reason: Submitting a PRISMA 2009 flow diagram to a journal that requires the 2020 version. Since 2023, virtually all major medical journals require PRISMA 2020.
PRISMA 2020 Flow Diagram Template
Here is the standard PRISMA 2020 flow diagram structure. Each box must be filled with the correct count from your systematic review process:
+===================================================================+
| IDENTIFICATION |
| |
| +-------------------------+ +------------------------------+ |
| | Database records | | Other source records | |
| | (n = _____) | | (n = _____) | |
| | | | - Citation searching (n=__) | |
| | PubMed: n = ____ | | - Hand searching (n=__) | |
| | Embase: n = ____ | | - Expert referral (n=__) | |
| | Cochrane: n = ____ | | | |
| +------------+-------------+ +-------------+----------------+ |
| | | |
| v v |
| +-------------------------+ +------------------------------+ |
| | After deduplication | | After deduplication | |
| | (n = _____) | | (n = _____) | |
| +------------+-------------+ +-------------+----------------+ |
+===============|====================================|===============+
| |
v |
+===================================================================+
| SCREENING |
| |
| +-------------------------+ +-------------------------------+ |
| | Title/abstract screened | | Records excluded | |
| | (n = _____) |->| (n = _____) | |
| +------------+-------------+ | - By automation (n=__) | |
| | | - Other reasons (n=__) | |
| v +-------------------------------+ |
| +-------------------------+ +-------------------------------+ |
| | Full-text assessed | | Full-text excluded + reasons | |
| | (n = _____) |->| (n = _____) | |
| +------------+-------------+ | - Wrong population (n=__) | |
| | | - Wrong intervention (n=__) | |
| | | - Wrong outcome (n=__) | |
| | | - Wrong study design (n=__) | |
| | | - Other (n=__) | |
| | +-------------------------------+ |
+===============|===================================================+
|
v
+===================================================================+
| INCLUDED |
| |
| +---------------------------------------------------------+ |
| | Studies in qualitative synthesis (systematic review) | |
| | (n = _____) | |
| +----------------------------+----------------------------+ |
| | |
| v |
| +---------------------------------------------------------+ |
| | Studies in quantitative synthesis (meta-analysis) | |
| | (n = _____) | |
| +---------------------------------------------------------+ |
+===================================================================+
Number verification formula: Total records - Duplicates = After deduplication. After deduplication - Title/abstract excluded = Full-text assessed. Full-text assessed - Full-text excluded = Included. If these don't add up, reviewers will flag it immediately.
Box-by-Box Filling Guide
Each box in the PRISMA 2020 flow diagram requires specific information. Here's how to fill each one correctly:
| Box | What to Report | Common Mistake |
| Database records | Report counts per database (PubMed, Embase, Cochrane separately) | Only reporting the merged total |
| Other source records | Citation searching (snowballing), hand searching, expert referrals | Omitting citation searching sources |
| After deduplication | Total records minus duplicates | Not reporting the deduplication count |
| Title/abstract screened | Number entering initial screening | Merging deduplication and screening steps |
| Records excluded | Initial screening exclusions. PRISMA 2020 adds: report "excluded by automation tools" separately | Not reporting automation exclusions |
| Full-text assessed | Number entering full-text review | — |
| Full-text excluded + reasons | Must list each exclusion reason with its count | Only reporting total exclusions without reasons |
| Qualitative synthesis | Studies included in the systematic review | Not distinguishing from quantitative count |
| Quantitative synthesis | Studies included in meta-analysis (may be fewer than qualitative) | Omitting this box entirely |
How to Create a PRISMA Flow Diagram with MetaReview (Free)
MetaReview is a free online meta-analysis tool that includes a built-in PRISMA 2020 flow diagram generator. Here's how to use it:
1
Open MetaReview — Go to
MetaReview in your browser. No download or account needed.
2
Search PubMed (optional) — Use the built-in PubMed search to find studies. The number of search results automatically populates the Identification section of your PRISMA diagram.
3
Screen studies (optional) — Use PICO keyword scoring or AI-powered screening (Llama 3.1). Screening decisions automatically fill the Screening section numbers.
4
Go to the PRISMA tab — Click the "PRISMA" tab in the navigation bar to open the flow diagram editor.
5
Enter or verify numbers — Fill in any boxes that weren't auto-populated. Adjust numbers as needed for your final manuscript.
6
Preview in real-time — The diagram updates instantly as you type. Check that all numbers are consistent across stages.
7
Export as SVG or PNG — Click the export button to download. SVG for vector quality (editable in Illustrator/Inkscape); PNG for direct manuscript insertion (2x retina resolution).
Bonus: MetaReview also exports your PRISMA data as a structured DOCX table, ready to paste into your manuscript's Methods section. The PRISMA checklist compliance is built into the workflow.
PRISMA Flow Diagram Tools Compared
Several tools can create PRISMA flow diagrams. Here's how they compare:
| Feature | MetaReview | prisma-statement.org | RevMan | Word Template |
| Price | Free | Free | Paid | Free |
| PRISMA 2020 | Yes | Yes | 2009 only | Varies |
| Auto-fill from search | Yes | No | No | No |
| SVG export | Yes | No | No | No |
| PNG export | Yes (2x retina) | Yes | Limited | No |
| DOCX table | Yes | No | No | Manual |
| Integrated meta-analysis | Yes (120+ features) | No | Yes | No |
| AI screening | Yes | No | No | No |
| Browser-based | Yes | Yes | No | No |
PRISMA 2020 Checklist: 27 Items (6 Most Commonly Missed)
The PRISMA 2020 checklist has 27 items. Items marked OFTEN MISSED are the ones reviewers most frequently flag:
1 Title — Include "systematic review" and/or "meta-analysis" in the title
2 Abstract — Structured abstract (background, objectives, methods, results, conclusions)
3 Rationale — Why this review was needed
4 Objectives — Clear PICO question
5 Eligibility criteria — Detailed inclusion/exclusion criteria OFTEN MISSED
6 Information sources — All databases, registries, websites searched
7 Search strategy — Full search strategy for at least one database OFTEN MISSED
8 Selection process — Screening method (how many reviewers? independent?)
9 Data collection — Data extraction methods and form
10 Data items — Specific data extracted from each study
11 Risk of bias — Assessment tool used and process
12 Effect measures — Effect size type and rationale OFTEN MISSED
13 Synthesis methods — Statistical model, heterogeneity assessment, subgroup analysis plan
14 Reporting bias assessment — Funnel plot and statistical test methods
15 Certainty assessment — GRADE method (if applicable) OFTEN MISSED
16 Study selection — PRISMA flow diagram results
17 Study characteristics — Summary table of included studies
18 Risk of bias results — Assessment results for each study
19 Individual results — Effect size data for each study
20 Synthesis results — Pooled effect size, CI, heterogeneity
21 Reporting bias results — Funnel plot and Egger's test results
22 Certainty of evidence — GRADE results (if applicable) OFTEN MISSED
23 Discussion — Interpretation, limitations, comparison with existing evidence
24 Registration — PROSPERO number and protocol changes OFTEN MISSED
25 Data availability — Data sharing statement
26 Funding — Funding sources
27 Competing interests — Conflict of interest disclosure
6 Most Common PRISMA-Related Rejection Reasons
Based on editorial feedback patterns across major medical journals, these are the PRISMA issues that most frequently lead to rejection or major revision requests:
- Using the 2009 template — Most journals now require PRISMA 2020. A 2009-style flow diagram signals outdated methodology to reviewers.
- Missing full-text exclusion reasons — PRISMA 2020 explicitly requires listing each reason (wrong population, wrong intervention, wrong outcome, wrong design) with individual counts. A single "excluded (n=42)" is insufficient.
- Numbers don't add up — If "screened" minus "excluded" doesn't equal "full-text assessed," reviewers will immediately request clarification. This often indicates errors in the screening process itself.
- Missing complete search strategy — You must report the full search string for at least one database. "We searched PubMed using relevant keywords" is not acceptable.
- No PRISMA checklist submitted — Many journals require the completed 27-item checklist as a supplementary file at submission. Don't forget to fill and attach it.
- No PROSPERO registration — While not strictly mandatory, an increasing number of journals treat pre-registration as a submission requirement. Register at PROSPERO before starting your review.
Pre-submission checklist: (1) Using PRISMA 2020? (2) Numbers add up? (3) Full-text exclusion reasons listed? (4) Complete search strategy included? (5) PRISMA checklist attached? (6) PROSPERO number included?
Expert Tips for a Publication-Ready PRISMA Diagram
Tip 1: Run the Numbers Check Before Submission
Create a simple spreadsheet that tracks your flow: Total → Deduplicated → Screened → Full-text → Included. At every stage, "kept + excluded = previous stage total." If this doesn't hold, find and fix the discrepancy before submitting.
Tip 2: Be Specific About Exclusion Reasons
Don't use vague categories like "not relevant" or "doesn't meet criteria." Map your exclusion reasons to PICO components: wrong Population, wrong Intervention, wrong Comparator, wrong Outcome, or wrong study design. This makes your diagram both informative and PRISMA 2020-compliant.
Tip 3: Document Each Database Separately
PRISMA 2020 requires individual database counts. Keep a log during your search: "PubMed: 234, Embase: 189, Cochrane: 56, Web of Science: 312." This avoids the retroactive guessing that leads to errors.
Tip 4: Include the "Previous Studies" Source
PRISMA 2020 adds a right-side pathway for studies found through citation searching, hand searching reference lists, or expert recommendations. Even if this number is zero, include it to show you considered these sources.
Tip 5: Export as SVG for Best Quality
SVG (Scalable Vector Graphics) produces crisp, infinitely scalable diagrams perfect for journal production. If your journal requires raster images, export as PNG at 2x resolution (MetaReview does this automatically). Avoid screenshot-based submissions.
PRISMA Extensions You Should Know
Beyond the standard PRISMA 2020 statement, several extensions exist for specific review types:
| Extension | Use When | Key Difference from Standard |
| PRISMA-S (Search) | Reporting search strategies in detail | 16-item checklist specifically for search methodology |
| PRISMA-P (Protocols) | Publishing your review protocol | 17-item checklist for systematic review protocols |
| PRISMA-NMA | Network meta-analysis | Additional items for indirect comparisons and network geometry |
| PRISMA-DTA | Diagnostic test accuracy reviews | Modified items for sensitivity, specificity, and ROC analyses |
| PRISMA-IPD | Individual participant data meta-analysis | Additional items for IPD collection, quality, and missing data |
| PRISMA-ScR | Scoping reviews | 22-item checklist adapted for scoping review methodology |
Which to use? Most researchers need only the standard PRISMA 2020 statement. Use an extension only if your review type specifically matches one above.
Stay Updated
Get notified about new features and meta-analysis tips. No spam, unsubscribe anytime.
We respect your privacy. Unsubscribe anytime.