Nursing Dissertation Guide

PRISMA Flow Diagram Mistakes in Nursing Systematic Reviews

Find and correct PRISMA 2020 flow-diagram errors involving search totals, duplicates, screening, full-text exclusions, reports and included nursing studies.

Written by Moses Updated August 25, 2026 8 min read
PRISMA Flow Diagram Mistakes in Nursing Systematic Reviews

PRISMA flow diagram mistakes usually begin before the diagram is drawn. Missing search logs, inconsistent duplicate counts, confused use of record, report and study, and vague full-text exclusion reasons can make a nursing systematic review difficult to audit.

The diagram should reflect the actual study-selection process from identification through screening and inclusion. It is not a decorative figure added after the review is finished.

What needs to reconcile

  • Final search totals must match saved database searches.
  • Pre-screening removals must be categorised correctly.
  • Records, reports and studies must not be treated as the same unit.
  • Title/abstract exclusions and full-text exclusions must be separated.
  • Reports not retrieved must not be given invented eligibility reasons.
  • Full-text exclusion reasons should be specific and reconcilable.
  • The final included-study count must match the characteristics table and synthesis.

Use the correct PRISMA 2020 template

PRISMA 2020 provides different flow diagrams for new and updated reviews and for reviews using databases/registers alone or additional identification methods. Choose the official template that matches what was actually done (Page et al., 2021a).

If a nursing review searched CINAHL, MEDLINE and PsycINFO and also used citation searching or an organisation’s website, the diagram should represent those additional sources rather than hiding them inside the database total.

Do not confuse records, reports and studies

These are different units:

  • Record: a database or register entry, such as a citation imported from CINAHL.
  • Report: a document that provides information about a study, such as an article, conference abstract or thesis.
  • Study: the underlying investigation.

One study can generate several reports. A trial might have a protocol, main outcomes paper and follow-up article. Those documents can appear as separate records early in screening but should not be counted as three independent studies in the final synthesis.

PRISMA 2020 distinguishes included reports from included studies for exactly this reason (Page et al., 2021b).

Record duplicate removal accurately

Database overlap means the same citation can be retrieved from several sources. Before removing duplicates, preserve the original result count for every database and export.

Audit field Example
Database/platform CINAHL Complete via EBSCOhost
Search date 14 July 2026
Records exported 286
Export file cinahl_2026-07-14.ris
Imported successfully 286
Deduplication method DOI/PMID match plus title-author-year review

Do not place every record removed before screening in the duplicate box. PRISMA 2020 separates duplicate records, records marked ineligible by automation and records removed for other reasons.

Check the arithmetic between every stage

The flow diagram should satisfy a simple accounting sequence:

  • Records screened = records identified − records removed before screening.
  • Reports sought = records screened − records excluded at title/abstract stage.
  • Reports assessed = reports sought − reports not retrieved.
  • Included reports = reports assessed − reports excluded after full-text review.

For example, if searches identify 940 records and 190 are removed before screening, 750 records remain. If 650 are excluded at title/abstract stage, 100 reports are sought. If four cannot be retrieved, 96 are assessed. If 84 are excluded after full-text review, 12 reports remain. Those 12 reports might describe only nine underlying studies.

If the arithmetic fails, return to the search and screening logs. Do not change a number simply to make the diagram look consistent.

Use only final search totals

Search strategies often change during development. A pilot PubMed search and the final translated strategy may produce different counts. The diagram should use the final saved search set that matches the method appendix and screening library.

Record the exact database, platform, strategy version, search date, limits and total. Updated searches should be logged separately rather than overwriting the original record.

The site’s systematic-review search-strategy guide explains how to build and document the retrieval stage.

Keep title/abstract and full-text exclusions separate

Initial screening excludes records based on the information available in titles and abstracts. Full-text screening evaluates retrieved reports against the complete eligibility criteria.

Detailed reasons usually belong at the full-text stage. Do not retrospectively invent detailed title-stage reasons if they were never recorded.

Use specific full-text exclusion reasons

“Not relevant” is not sufficiently informative. Exclusion categories should map to the eligibility criteria and be understandable to another reviewer.

Possible categories include:

  • wrong population;
  • wrong intervention or phenomenon;
  • ineligible comparator where required;
  • wrong outcome;
  • ineligible study design;
  • wrong setting;
  • not primary research; or
  • duplicate report of an already represented study.

Assign one primary reason to each excluded report so the category totals equal the number excluded. If several criteria fail, use a predefined hierarchy.

PRISMA 2020 also recommends identifying apparently eligible reports that were excluded and explaining why (Page et al., 2021b).

Do not count unavailable reports as full-text exclusions

If the full text was not obtained, the report could not be assessed fully against the eligibility criteria. Count it as not retrieved rather than inventing a reason such as wrong outcome.

Document reasonable retrieval attempts through institutional access, repositories, interlibrary loan, author contact or citation verification where appropriate.

Report automation transparently

Some review tools assist with duplicate detection, prioritisation or screening. If automation removed or classified records, explain what tool was used, the rule applied and how decisions were checked.

Do not call standard reference-manager deduplication “AI screening,” and do not put classifier removals in the duplicate box.

Students should also follow programme rules governing automation or AI-assisted review processes and retain an audit trail they can explain.

Build the diagram from the audit trail, not memory

Create the tracking record while searching and screening are underway. Retrospective reconstruction can create inconsistent counts, lost exclusion reasons and confusion about linked reports.

A useful file structure includes:

  1. search log;
  2. master reference library;
  3. deduplication log;
  4. screening export with decisions and full-text reasons; and
  5. study-report linkage table.

The Cochrane Handbook emphasises systematic reference management, documentation and study selection (Lefebvre et al., 2025).

Check the diagram against the rest of the dissertation

The final PRISMA numbers should agree with:

  • the methodology;
  • the evidence or characteristics table;
  • the narrative results;
  • the list of excluded full texts where required; and
  • the synthesis.

A flow diagram reporting nine included studies cannot coexist with a characteristics table containing eleven independent studies without an explanation.

Present the figure clearly

Use the official PRISMA template or approved flow-diagram tool. Keep labels legible in the final PDF and avoid replacing counts with percentages.

Introduce the figure in the results section with one concise sentence. For example:

The searches identified 940 records; after pre-screening removals and eligibility assessment, 12 reports describing nine studies were included.

The figure should provide the full route rather than the narrative repeating every number.

Common PRISMA flow-diagram mistakes

  • Using the wrong template.
  • Mixing pilot and final search totals.
  • Deleting duplicates before preserving source counts.
  • Treating reports as separate studies.
  • Assigning vague exclusion reasons.
  • Counting unavailable reports as full-text exclusions.
  • Using numbers that do not reconcile mathematically.
  • Hiding automation or manual pre-screening removals.
  • Building the diagram from memory.
  • Allowing the diagram, evidence table and synthesis to contain different included-study counts.

Final PRISMA checklist

  • The correct PRISMA 2020 template is used.
  • Search totals come from final saved strategies.
  • Other-source records use the correct identification pathway.
  • Duplicate, automation and other removals are separated.
  • Records screened reconcile with identification and removal totals.
  • Reports sought reconcile with title/abstract exclusions.
  • Reports assessed reconcile with reports not retrieved.
  • Full-text exclusion categories reconcile with total exclusions.
  • Multiple reports from one study are linked.
  • Included-study counts match the characteristics table and synthesis.
  • The method reports screening responsibility, disagreements and software accurately.
  • The figure remains legible in the final document.

Frequently asked questions

Does every nursing literature review need a PRISMA diagram?

No. PRISMA 2020 is intended for systematic-review reporting, with extensions for related evidence-synthesis designs. A traditional narrative review may not require one.

Should duplicates be reported by database?

Report source totals, then the total number of duplicate records removed after sources are combined. One duplicate can occur across several databases.

Can a conference abstract and journal article count as two studies?

Not when they report the same underlying study. Link the reports and count the study once.

What if one full-text report fails several criteria?

Use one predefined primary exclusion reason so the exclusion counts remain mutually reconcilable.

Can software generate the flow diagram automatically?

Yes, but the diagram is only as accurate as the underlying data. Verify every count against the saved search and screening records.

Project-specific support

If a PRISMA diagram does not reconcile with the screening log, use the nursing systematic review support route or the contact page for a focused review.

Related Nursing Guides

Conclusion

A reliable PRISMA flow diagram is produced from a reliable study-selection record. Use the correct template, preserve final search totals, separate removal types, distinguish records from reports and studies, reconcile every transition and use specific full-text exclusion reasons. When the diagram, screening log, evidence table and synthesis agree, the nursing review becomes much easier to audit and defend.

References

Related articles

Browse all