Nearly 80% of systematic reviews published in leading journals now require a PRISMA flow diagram, yet many researchers still struggle with filling one out correctly. Whether you are completing your first systematic review or updating a published one, the PRISMA flow diagram is the single visual that tells reviewers, editors, and readers exactly how you arrived at your final set of included studies. Getting it wrong can delay publication, invite revision requests, or undermine confidence in your entire review. This guide walks you through every phase of the PRISMA flow diagram so you can build one that is accurate, transparent, and ready for submission.
What is a PRISMA flow diagram?
A PRISMA flow diagram is a standardized flowchart that maps the movement of records through each stage of a systematic review or meta-analysis. PRISMA stands for Preferred Reporting Items for Systematic Reviews and Meta-Analyses, and the flow diagram is one of its most recognizable reporting tools. It visually documents how many records were identified, how many were screened, how many were assessed for eligibility, and how many were ultimately included — along with the reasons records were excluded at each stage.
The current standard is the PRISMA 2020 flow diagram, which replaced the original 2009 version. The updated diagram introduced several important changes, including separate reporting for records identified from databases and registers versus other sources like grey literature, citation searching, and organizational websites. It also added boxes for reporting automation tools used during screening and for documenting studies retrieved from previous systematic reviews.
The PRISMA 2020 flow diagram typically appears as the first figure in the results section of a systematic review. Its purpose is threefold: it demonstrates methodological rigor, it allows readers to assess the strengths and potential biases of the review, and it enables other researchers to replicate the search and selection process.
Why every systematic review needs a PRISMA flow diagram
Journal editors and peer reviewers expect a completed PRISMA flow diagram in every systematic review submission. Journals like The BMJ, The Lancet, and PLOS Medicine explicitly require PRISMA compliance. Beyond journal requirements, the diagram serves several critical functions:
Transparency. It shows exactly where records entered and exited the review pipeline, making the selection process fully auditable.
Reproducibility. Other researchers can follow your documented process to verify or update your review.
Bias assessment. Reviewers can identify potential selection bias by examining where and why large numbers of records were excluded.
Completeness check. The diagram forces you to account for every record, which often reveals counting errors before submission.
According to a 2021 study published in The BMJ (Page et al., BMJ 2021;372:n71), the PRISMA 2020 update was developed because evaluations of published systematic reviews found that reporting quality remained substandard, with many reviews failing to document the full record flow. The updated flow diagram addresses these gaps directly.
How to choose the right PRISMA 2020 template
Before filling in any numbers, you need to select the correct template. PRISMA 2020 provides four official templates, available for download from the PRISMA website:
New reviews — databases and registers only. Use this if your search strategy was limited to bibliographic databases (PubMed, Scopus, Web of Science, etc.) and trial registries (ClinicalTrials.gov, WHO ICTRP).
New reviews — databases, registers, and other sources. Use this if you also searched grey literature, organizational websites, consulted experts, checked reference lists of included studies, or used citation tracking tools.
Updated reviews — databases and registers only. Use this when you are updating a previously published systematic review and only searched databases and registers.
Updated reviews — databases, registers, and other sources. Use this for updated reviews that also incorporated additional search methods.
Most systematic reviews that follow a comprehensive search strategy will need the template that includes other sources alongside databases and registers. If you are unsure, choose the more inclusive template — it is better to have empty boxes for sources you did not use than to omit reporting on sources you did.
Step-by-step guide to completing each phase
The PRISMA 2020 flow diagram is organized into four main phases: identification, screening, eligibility, and inclusion. Here is how to complete each one accurately.
Phase 1: Identification — counting what you found
The identification phase documents every record your search strategy retrieved, broken into two columns if you used the template that includes other sources.
Left column — databases and registers:
Record the total number of results retrieved from all databases combined. Best practice is to also note the count per database (for example, PubMed n=1,204, Scopus n=987, Web of Science n=643).
If you searched trial registries, include those counts here as well.
Below that, record the number of duplicate records removed. This should reflect duplicates removed through your reference management tool or systematic review software before screening began.
You may also note records removed by automation tools (such as machine learning classifiers that flagged clearly irrelevant records) or records marked as ineligible by human reviewers before formal screening.
Right column — other sources (if applicable):
Record the number of records identified from websites, organizations, or citation searching.
Record the number of records identified from previous systematic reviews on the same topic that you consulted for potentially relevant studies.
Practical tip: Track your database search results immediately after running each search. Do not wait until the end of the review to try to reconstruct your numbers from memory or fragmented notes. A research project management platform like ScholarDock lets you log search results, store exported records, and maintain a running count alongside your project tasks — so your identification numbers are always documented and accessible to every team member.
Phase 2: Screening — title and abstract review
After removing duplicates and any records eliminated by automation, you are left with a set of unique records to screen. This phase documents the title and abstract screening stage.
Records screened (n=). Enter the total number of unique records that entered title and abstract screening.
Records excluded (n=). Enter the number of records excluded at this stage. You do not need to provide detailed reasons for exclusion during title and abstract screening — a single aggregate number is sufficient.
Title and abstract screening is where the largest volume of records is typically removed. In many systematic reviews, 70% to 90% of records are excluded at this stage because they are clearly outside the scope of the review question, involve the wrong population, or are not primary research.
Best practice for screening accuracy:
Use at least two independent reviewers for screening.
Define your inclusion and exclusion criteria before screening begins and apply them consistently.
Use a pilot screening round on a random sample of 50 to 100 records to calibrate agreement between reviewers.
Resolve disagreements through discussion or a third reviewer rather than simple majority voting.
Phase 3: Eligibility — full-text assessment
Records that passed title and abstract screening move to full-text review. This is where you read each study in full and apply your eligibility criteria rigorously.
Reports sought for retrieval (n=). Enter the number of full-text articles you attempted to locate. This is usually the same as the number of records that passed screening, but it can differ if a single record corresponds to multiple reports or vice versa.
Reports not retrieved (n=). Enter the number of full texts you were unable to obtain despite reasonable efforts (interlibrary loan, contacting authors, institutional access). Document why retrieval failed.
Reports assessed for eligibility (n=). Enter the number of full texts you actually read and evaluated.
Reports excluded, with reasons (n=). Enter the total number of studies excluded at the full-text stage. Unlike title and abstract screening, you must provide specific reasons for each exclusion. Common reasons include wrong study design, wrong population, wrong intervention, wrong comparator, wrong outcome, duplicate publication, or insufficient data.
Important counting rule: Each excluded study should be assigned only one primary reason for exclusion, even if multiple criteria were not met. Most researchers apply the exclusion reason based on a pre-specified hierarchy — for example, wrong study design takes priority over wrong intervention.
Phase 4: Inclusion — your final study set
The bottom of the flow diagram shows the studies that met all eligibility criteria and were included in the review.
Studies included in the review (n=). Enter the final number of unique studies included in your qualitative synthesis.
Reports of included studies (n=). If some studies have multiple publications (for example, a trial protocol, a primary outcome paper, and a follow-up paper), the number of reports may be higher than the number of studies.
If your review includes a meta-analysis, add a separate box showing the number of studies included in quantitative synthesis. Not all included studies may contribute data to the meta-analysis.
Common mistakes to avoid when creating your PRISMA flow diagram
Even experienced researchers make errors in their PRISMA flow diagrams. Here are the most frequent problems that lead to revision requests from peer reviewers:
Numbers that do not add up. The most basic and most common error. Records identified minus records removed minus records excluded at each stage must equal the number of studies included. Check your arithmetic before submission.
Confusing records, reports, and studies. A "record" is a single entry from a database search (one citation). A "report" is a single document (one article or paper). A "study" is a single investigation, which may produce multiple reports. The PRISMA 2020 diagram carefully distinguishes between these — make sure you do too.
Missing exclusion reasons at the full-text stage. Listing "did not meet inclusion criteria" as an exclusion reason is not specific enough. Reviewers expect concrete reasons such as "wrong study design (n=12), wrong population (n=8), wrong intervention (n=5)."
Forgetting to report records from other sources. If you checked reference lists, consulted experts, or searched grey literature, those records must be reported in the right column of the identification phase. Omitting them misrepresents your search comprehensiveness.
Not updating the diagram after changes. If you re-run searches, add databases, or include studies found during peer review, the flow diagram must be updated to reflect the final numbers.
Using the outdated 2009 template. The PRISMA 2009 flow diagram is no longer current. Always use the PRISMA 2020 version, which includes boxes for automation, previous reviews, and the records-reports-studies distinction.
Tools and methods for generating a PRISMA flow diagram
You do not need to build your PRISMA flow diagram from scratch. Several tools can help:
PRISMA Flow Diagram Generator (Shiny App). Available at estech.shinyapps.io/prisma_flowdiagram, this free web tool lets you enter your numbers and generates a publication-ready diagram. It supports both the databases-only and databases-plus-other-sources templates.
Official Word templates. Download editable Word templates directly from the PRISMA website. These are the simplest option if you prefer to fill in numbers manually.
Covidence. If you use Covidence for your systematic review screening, it generates a PRISMA diagram automatically based on your screening decisions. You may need to add database-level counts manually.
Review management platforms. Tools like ScholarDock, a research project and reference management platform, let you track your screening pipeline within the same workspace where you manage references, assign screening tasks to team members, and organize project milestones — making it straightforward to pull accurate numbers for your PRISMA diagram at any point during the review.
How ScholarDock simplifies the systematic review workflow
Building an accurate PRISMA flow diagram depends entirely on keeping meticulous records throughout the review process — and that is where most teams run into trouble. Screening decisions are scattered across spreadsheets, email threads, and disconnected reference managers. Team members lose track of which records have been screened and by whom. Counts drift as records are added or removed without documentation.
ScholarDock solves this by bringing the entire systematic review workflow into one connected workspace. You can import records from multiple database searches, track duplicates, assign screening tasks to individual team members, log inclusion and exclusion decisions with reasons, and monitor progress across every phase — all within a single platform. Because every action is documented in the same place where your references, notes, and project tasks live, generating your PRISMA flow diagram numbers becomes a matter of reviewing your project dashboard rather than piecing together data from five different tools.
For research teams running multiple systematic reviews simultaneously, this connected approach eliminates the coordination overhead that makes PRISMA reporting feel like a burden rather than a natural part of the workflow.
PRISMA flow diagram checklist before submission
Before you submit your systematic review, run through this final checklist to make sure your PRISMA flow diagram is complete and accurate:
You used the correct PRISMA 2020 template for your review type (new or updated) and sources searched
All database and register search counts are reported, ideally broken down by individual source
Duplicate removal is documented with the exact count
Any records removed by automation tools are reported separately
Title and abstract screening numbers are accurate and the excluded count is recorded
Full-text retrieval attempts and failures are documented
Full-text exclusion reasons are specific and listed with counts for each reason
The distinction between records, reports, and studies is maintained throughout the diagram
The final included studies count matches what appears in your results tables
If a meta-analysis was performed, the number of studies contributing quantitative data is shown separately
All numbers add up correctly from top to bottom — no records are unaccounted for
If you searched other sources beyond databases and registers, they appear in the right column of the identification phase
Moving from diagram to publication-ready review
The PRISMA flow diagram is just one piece of the PRISMA 2020 reporting framework. The full PRISMA 2020 checklist contains 27 items across seven sections — covering everything from the title and abstract to the discussion and funding disclosures. Completing the flow diagram accurately is often the hardest part because it requires perfect record-keeping throughout the entire review. But once your diagram is solid, it anchors the rest of your reporting.
If your research team is tired of chasing numbers across disconnected spreadsheets, losing track of screening decisions, and reconstructing PRISMA data from fragmented notes, ScholarDock brings your entire systematic review workflow — references, screening, project tracking, and team collaboration — into one connected workspace. Start organizing your next review from identification to inclusion in a single place.