Aug 25, 2026

PRISMA Flow Diagram: How to Track Study Selection in a Systematic Review

By JournalLabs Research Team

Introduction

A systematic review may identify thousands of search results but include only a small number of eligible studies. Readers need to understand how the review moved from the initial search to the final evidence set.

A PRISMA flow diagram provides this information visually. It records how many sources were identified, removed, screened, retrieved, assessed, excluded, and included.

An accurate diagram cannot be reconstructed reliably from memory at the end of a project. Researchers must track search results and selection decisions throughout the review.

This guide explains what a PRISMA flow diagram contains, how to maintain consistent counts, and how to avoid common reporting errors.

What Is a PRISMA Flow Diagram?

PRISMA stands for Preferred Reporting Items for Systematic Reviews and Meta-Analyses. Its flow diagram describes how evidence moves through the identification and selection process of a systematic review.

PRISMA 2020 provides different templates for new and updated reviews. The appropriate version also depends on whether evidence was identified through databases and registers only or through additional methods such as websites, organizations, reference lists, and citation searches.

The diagram covers four broad stages:

1. Identification: finding records through databases, registers, and other sources.
2. Screening: reviewing titles and abstracts after removing duplicates.
3. Eligibility: retrieving and assessing full-text reports.
4. Inclusion: identifying the final studies and reports included in the review.

The flow diagram does not decide which studies are eligible. Those decisions come from the review protocol and predefined inclusion and exclusion criteria. PRISMA documents what happened when the criteria were applied.

Records, Reports, and Studies Are Different

PRISMA distinguishes between records, reports, and studies.

During title and abstract screening, researchers generally count records. During full-text assessment, they count reports. At the end, they may report both included studies and the reports describing them.

One study can produce a protocol, conference abstract, primary article, follow-up analysis, and secondary publication. A review may therefore include 40 reports representing only 36 studies.

Keeping these units separate prevents duplicate evidence and inconsistent totals.

What to Record at Each PRISMA Stage

Identification

Record the number of results retrieved from every database, register, and additional source before removing duplicates.

The search log should include:

• Database or source name
• Search date
• Search strategy
• Filters applied
• Number of results
• Results from updated searches

Capture these figures when each search is completed. Database result counts may change as new records are added or metadata is corrected.

When a citation appears in multiple databases, count it in each original export and then record the additional copies as duplicates during removal.

Removal Before Screening

Records may be removed before screening because they are duplicates, have been marked ineligible by an approved automation process, or meet another documented removal rule.

Deduplication should consider titles, authors, publication years, journals, and identifiers such as DOI or PMID. Exact title matching alone may miss records with punctuation, spelling, or metadata differences.

Preserve the original search exports and document the deduplication method. Researchers should be able to explain how the initial search total became the number screened.

Title and Abstract Screening

The remaining records are screened against the review’s eligibility criteria.

The diagram reports:

• Records screened
• Records excluded
• Reports sought for retrieval

Individual reasons are not normally required for every title and abstract exclusion. However, the underlying decisions should remain available in the review-management record.

The calculation should be straightforward:

Records screened − records excluded = reports sought for retrieval

Retrieval and Full-Text Assessment

Potentially eligible reports move to full-text retrieval and assessment.

Track three separate figures:

• Reports sought for retrieval
• Reports not retrieved
• Reports assessed for eligibility

A report that cannot be obtained is not a full-text exclusion because the reviewers could not evaluate it completely.

For every retrieved report that is excluded, record one primary reason. Common reasons include:

• Wrong population
• Wrong intervention or exposure
• Wrong outcome
• Ineligible study design
• Ineligible publication type
• Insufficient data
• Outside the specified date, language, or setting criteria

Using one primary reason per excluded report keeps the exclusion totals consistent. Review teams can define a hierarchy for situations in which several criteria apply.

Final Inclusion

The final stage records the number of included studies and the number of reports describing them.

If several reports describe the same participants, trial, dataset, or investigation, group them under one study identifier. Retain relevant reports, but do not treat each publication as an independent study.

A separate figure may show studies included in a meta-analysis when statistical synthesis is performed. A systematic review does not always require a meta-analysis.

How to Track Study Selection Step by Step

Step 1: Create a Search Log

Record every source, search date, search strategy, filter, and result count. Keep database, register, website, and citation-search totals separate.

Step 2: Assign Stable Identifiers

Give every imported record an internal identifier. Retain DOI, PMID, registry number, or another persistent identifier when available.

Step 3: Deduplicate Before Screening

Use automated matching where appropriate, then review uncertain matches manually. Record the exact number removed as duplicates.

Step 4: Save Screening Decisions

Store an include, exclude, or uncertain decision for each record. When two reviewers screen independently, preserve both decisions and the final resolution.

Step 5: Track Retrieval Status

Record whether each potentially eligible report was retrieved successfully. Keep unavailable reports separate from reports excluded after full-text assessment.

Step 6: Apply Consistent Exclusion Reasons

Create a controlled list of full-text exclusion reasons before assessment begins. Assign one primary reason to every excluded report.

Step 7: Link Reports to Studies

Compare authors, participants, locations, sample sizes, dates, interventions, and registry numbers to identify multiple publications from one study.

Step 8: Reconcile the Numbers

Check every transition before creating the final diagram:

• Identified records minus removed records equals screened records.
• Screened records minus excluded records equals reports sought.
• Reports sought minus reports not retrieved equals reports assessed.
• Reports assessed minus reports excluded equals included reports.

The number of included reports may be larger than the number of included studies.

Completed PRISMA Tracking Example

The following example shows how the counts move through a systematic review.

The 225 full-text exclusions could be documented as follows:

The example contains 40 included reports but 36 included studies. This difference is valid because several studies produced more than one report.

Every figure is connected to the previous stage, and all excluded reports are assigned one primary reason.

Common PRISMA Flow Diagram Mistakes

Reconstructing Counts at the End

Search results, duplicates, and screening decisions are easily lost. Maintain the selection log from the first search.

Using Unclear Exclusion Categories

Labels such as “irrelevant” or “not suitable” do not explain why a report failed the eligibility criteria. Use categories connected to the protocol.

Counting One Report Under Several Reasons

Several criteria may apply, but the diagram needs consistent totals. Assign one primary exclusion reason.

Ignoring Updated Searches

If the search is rerun before publication, add the new records and document how they moved through screening.

Hiding the Use of Automation

Report whether automation removed, prioritized, or classified records. Retain human verification for important eligibility decisions.

Using the Diagram Instead of a Methods Section

The flow diagram reports numbers. The methods section must still explain databases, search strategies, screening procedures, reviewer roles, eligibility criteria, and dispute resolution.

How JournalLabs Supports Literature Review Organization

JournalLabs helps researchers organize scientific literature, compare studies, and understand evidence across multiple papers. Its AI literature review feature can support the discovery and review of relevant research.

Researchers should maintain a separate decision log for search totals, duplicates, screening decisions, retrieval status, and exclusions. After verifying the counts, they can transfer the results into the appropriate PRISMA 2020 flow diagram.

AI can reduce repetitive literature work, but the review team remains responsible for eligibility decisions and final reporting.

Frequently Asked Questions

Is a PRISMA flow diagram required for every literature review?

It is mainly used for systematic reviews and related evidence-synthesis methods. A traditional narrative review may not require one unless it follows a documented systematic selection process.

Should title and abstract exclusions include individual reasons?

The flow diagram generally reports only the total excluded during title and abstract screening. Specific reasons are normally reported for exclusions made after full-text assessment.

Does every systematic review include a meta-analysis?

No. A systematic review may use narrative or another form of synthesis when statistical pooling is inappropriate.

Which PRISMA flow diagram template should researchers use?

Use the PRISMA 2020 template that matches the review type: new or updated, and databases and registers only or databases plus other identification methods.

Can AI create a PRISMA flow diagram automatically?

AI can help organize records and detect possible duplicates, but all counts and selection decisions require verification by the review team.

Conclusion

A PRISMA flow diagram makes study selection transparent and reproducible. Accurate reporting depends on tracking evidence from the original search through screening, retrieval, eligibility assessment, and final inclusion.

Researchers should distinguish records, reports, and studies; apply consistent exclusion reasons; and reconcile every number before publication. The completed diagram then provides a clear numerical account of how the review’s evidence base was created.

Build a More Transparent Systematic Review

Use JournalLabs to organize scientific evidence and understand relationships across studies while maintaining verified selection records for accurate PRISMA reporting.