24/09/2025
How to Fill Out the PRISMA Flow Diagram for Your Systematic Review
Systematic reviews are essential for synthesizing evidence in research, but transparency in reporting is key to their credibility. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines provide a standardized way to report these reviews, and the PRISMA flow diagram is a core component. Updated in 2020, this diagram visually maps the flow of information from initial record identification to the final inclusion of studies. It helps readers understand how studies were selected and why others were excluded.
In this detailed guide, I'll walk you through how to create and fill out the PRISMA 2020 flow diagram step by step. This is based on the official PRISMA recommendations and practical tutorials from academic libraries. Whether you're conducting a new or updated systematic review, and regardless of your search sources (databases only or including registers and other methods), this process ensures your diagram is accurate and complete.
Why Use the PRISMA Flow Diagram?
The diagram depicts the phases of a systematic review: identification, screening, eligibility, and inclusion. It includes numbers of records at each stage and reasons for exclusions, promoting reproducibility and reducing bias. PRISMA 2020 introduces templates for different scenarios, such as reviews searching only databases and registers versus those incorporating additional sources like citation searching or expert consultations. Failing to include a properly filled diagram can lead to your review being rejected by journals, so getting it right is crucial.
Choosing the Right Template
Before starting, download the appropriate editable template from the official PRISMA website (prisma-statement.org). There are four main versions:
For new systematic reviews searching databases and registers only.
For new systematic reviews including other sources (e.g., websites, organization reports, or citation searching).
For updated systematic reviews searching databases and registers only.
For updated systematic reviews including other sources.
Updated reviews include additional boxes for prior studies from previous versions of the review. If your review involves automation tools (e.g., for duplicate removal or screening), note their use in relevant boxes. Always modify grey boxes (optional elements) as needed—remove them if they're not applicable.
Step-by-Step Guide to Filling the Diagram
The diagram is divided into sections with boxes for numbers and arrows showing the flow. Use tools like citation managers (e.g., EndNote, Zotero) or review software (e.g., Covidence) to track your numbers accurately. Here's how to complete it, adapted from a practical university guide.
Step 1: Preparation
Gather your search strategy details, including databases (e.g., PubMed, Embase), registers (e.g., ClinicalTrials.gov), and any other sources.
Open your chosen template and note your research question to ensure relevance during screening.
No boxes to fill yet: This is setup. Document your full search strategy elsewhere in your review (PRISMA Item 7).
Step 2: Identification – Records Identified from Searches
Run searches in each database/register individually, applying filters (e.g., date ranges, language).
Tally the results from each source.
What to fill in:
In the top-left box ("Identification of studies via databases and registers" or similar, depending on template): Enter the total number of records identified. Break it down, e.g., "Databases (n= total from all databases)" and "Registers (n= total from registers)."
If using the "other sources" template, add a parallel branch for records from websites, citations, etc.
Example: If PubMed yields 500 records and Embase 300, enter "Databases (n=800)" with optional per-database notes.
Tip: Include all records before deduplication. For updated reviews, add boxes for "Previous studies" and "Records identified from new searches."
Step 3: Identification – Remove Duplicates and Other Pre-Screening Exclusions
Export results to a citation manager and use its deduplication feature. If using automation tools for initial filtering (e.g., excluding irrelevant topics), count those here too.
What to fill in:
In the box below identification ("Records removed before screening"): Enter the number of duplicates removed, plus any automation exclusions or other reasons (e.g., "Duplicate records removed (n=150)").
Example: If 800 records minus 150 duplicates = 650 remaining.
Tip: Be transparent about automation to avoid bias claims. If no duplicates, enter 0.
Step 4: Screening – Records Screened
This is the number of unique records after Step 3, ready for title/abstract review.
What to fill in:
In the "Records screened" box: Enter the calculated number (e.g., 650).
Tip: Two reviewers should independently screen for reliability (report inter-rater agreement in your methods).
Step 5: Screening – Records Excluded at Title/Abstract Level
Review titles and abstracts against your inclusion/exclusion criteria (e.g., study design, population, outcomes).
Exclude irrelevant ones.
What to fill in:
In the "Records excluded" box: Enter the number excluded (e.g., 400). Optionally, list broad reasons (e.g., "Not relevant to topic (n=250), Wrong study type (n=150)"), though detailed reasons aren't mandatory here.
Example: 650 screened minus 400 excluded = 250 proceeding.
Tip: Err on the side of inclusion if uncertain—full-text review comes next.
Step 6: Eligibility – Reports Sought for Retrieval
These are the records advancing to full-text review.
What to fill in:
In the "Reports sought for retrieval" box: Enter the number from Step 5 minus exclusions (e.g., 250).
Tip: "Reports" can include journal articles, preprints, abstracts, or other documents.
Step 7: Eligibility – Reports Not Retrieved
Attempt to access full texts (use interlibrary loans if needed). Count any you can't obtain.
What to fill in:
In the "Reports not retrieved" box: Enter the number (e.g., 10). Reasons might include paywalls or unavailable archives.
Example: 250 sought minus 10 not retrieved = 240 assessed.
Tip: Minimize this by using institutional resources; report efforts in your methods.
Step 8: Eligibility – Reports Assessed for Eligibility
Review full texts for final inclusion.
What to fill in:
In the "Reports assessed for eligibility" box: Enter the number (e.g., 240).
Tip: Again, use dual reviewers and resolve disagreements via discussion or a third party.
Step 9: Eligibility – Reports Excluded at Full-Text Level
Exclude those not meeting criteria.
What to fill in:
In the "Reports excluded" box: Enter the total excluded (e.g., 200), with specific reasons and counts (e.g., "Wrong population (n=80), Insufficient data (n=60), Duplicate study (n=60)"). Count each report only once, even if multiple reasons apply.
Example: 240 assessed minus 200 excluded = 40 included.
Tip: Reasons must be detailed here for transparency—this is a PRISMA requirement.
Step 10: Inclusion – Studies Included in the Review
These are the final studies for synthesis.
What to fill in:
In the bottom boxes: "Studies included in qualitative synthesis (n=40)" and, if applicable, "Studies included in quantitative synthesis (meta-analysis) (n=30)".
If your review has no meta-analysis, remove or grey out the quantitative box.
Tip: If studies report multiple outcomes, count them as one study but note in your results.
Variations for Different Types of Reviews
Updated Reviews: Add branches for "Studies included in previous version" and integrate new searches. The total included combines prior and new studies.
Reviews with Other Sources: Include a separate identification branch for non-database sources (e.g., "Records identified from citation searching (n=50)").
Scoping Reviews or Other Variants: PRISMA extensions (e.g., PRISMA-ScR) adapt the diagram; check specific guidelines.
Automation Tools: Note their use in exclusion boxes to reflect modern workflows.
Common Tips and Best Practices
Accuracy: Double-check numbers—they must add up logically.
Software Help: Use Covidence or Rayyan for automated tracking.
Customization: Edit the template in Word or similar; ensure the diagram is readable in your manuscript.
Reporting in Your Paper: Include the diagram as a figure (e.g., Figure 1) and reference it in the results section (PRISMA Item 16a).
Common Pitfalls: Don't forget exclusions from other sources; always provide reasons at full-text stage; avoid inflating numbers by miscounting duplicates. AuthorLoom