ACADEMIC WRITING GUIDE
How to Create a PRISMA 2020 Flow Diagram for a Dissertation (With Worked Examples)
A systematic review may begin with hundreds of search results and finish with a few dozen included studies. Between those two points, readers need to understand what happened to every record and report. A PRISMA 2020 flow diagram documents that selection process without turning the Methods chapter into a confusing collection of unexplained totals.
This guide shows how to build a PRISMA flow diagram for a dissertation systematic review, reconcile screening counts, distinguish records, reports and studies, report exclusion reasons and write the accompanying Results paragraph. A complete fictional example follows two identification pathways and shows exactly how its counts reconcile.
Important: All numbers and screening decisions below are invented for teaching. They do not represent an actual search or a completed review. The appropriate reporting guideline depends on your review type and institutional requirements. A PRISMA diagram cannot turn an ordinary narrative review into a systematic review.
What is a PRISMA 2020 flow diagram?
PRISMA stands for Preferred Reporting Items for Systematic Reviews and Meta-Analyses. PRISMA 2020 offers reporting guidance, including an updated flow diagram that traces identification, screening and inclusion. The diagram helps readers understand how many potentially relevant items were discovered, removed, examined more closely and ultimately represented in the synthesis.
The official PRISMA 2020 resources provide checklists and flow-diagram templates. The official flow-diagram page offers versions suited to different search pathways and whether a review is new or updated.
A flow diagram is a report of work actually completed, not a substitute for an explicit eligibility protocol, reproducible search strategy or critical appraisal. Its totals must agree with your screening log and the included-study inventory.
Our guide to conducting a systematic literature search explains how to design and document searches. Here we focus on the separate, downstream task of accounting accurately for what those searches produced.
First, understand the three units: records, reports and studies
These terms are central to avoiding mismatched counts.
Record: A title-and-abstract entry or other reference identified by a database, register or information source. A single article can appear as several duplicate records across databases.
Report: A document conveying information about a study, such as a journal article, dissertation, preprint or trial report. A report may be sought, retrieved and assessed against eligibility criteria.
Study: The actual investigation. One study can have more than one report—for example, a conference abstract and a later full journal paper. Conversely, a report can occasionally describe more than one study; the link between reports and studies must be tracked explicitly.
The PRISMA 2020 explanation and elaboration distinguishes these units and explains how their numbers should be reported. Never assume that 34 included reports must mean 34 different included studies.
| Stage | Typical unit | What to count |
|---|---|---|
| Search results exported from databases | Records | Each retrieved citation before deduplication |
| Duplicate removal | Records | Redundant records removed before title/abstract screening |
| Title/abstract screening | Records | Unique records examined for possible eligibility |
| Full-text retrieval | Reports | Documents sought and successfully obtained |
| Eligibility assessment | Reports | Retrieved documents evaluated using the criteria |
| Final synthesis | Studies and reports | Unique eligible investigations, plus the documents describing them |
When a review includes websites or citation searching, the PRISMA 2020 template may use a separate other methods pathway. Choose the relevant official diagram rather than forcing everything into a single database count.
Step 1: Start with a documented eligibility question
Our fictional dissertation asks:
What is known about librarian-led digital-literacy workshops and university students' ability to evaluate online information?
The fictional review specifies eligible participants, the type of workshops or activities, relevant outcomes, eligible study designs, acceptable publication formats and any justified language or date limitations before final screening.
The example below assumes those criteria are documented in a review protocol. A transparent diagram cannot repair eligibility rules invented retrospectively to favour a desired answer.
If your research question is still too broad, consult our guide to developing a strong research question before building a screening form.
Step 2: Preserve the raw identification totals
Suppose the researcher searches two bibliographic databases and one study register:
Database A: 320 records.
Database B: 180 records.
Study register: 40 records.
These sources identify 540 records before deduplication.
The researcher also identifies 12 reports using other methods: 7 through targeted websites and 5 through backward citation searching. These are kept in a separate pathway in the worked example; they are not silently added to the 540 database/register records.
| Source | Items identified | Unit |
|---|---|---|
| Database A | 320 | Records |
| Database B | 180 | Records |
| Study register | 40 | Records |
| Databases and register combined | 540 | Records |
| Targeted websites | 7 | Reports |
| Citation searching | 5 | Reports |
| Other methods combined | 12 | Reports |
Maintain the search dates, platforms, full search strings and exported result files. The PRISMA-S reporting extension provides detailed guidance for documenting searches, beyond the selection diagram itself.
Step 3: Deduplicate before screening
Suppose the combined 540 database/register records include:
90 duplicate records removed.
10 records marked ineligible by a defined automation procedure before screening.
5 other records removed before screening for a documented reason.
That leaves 435 unique records to screen: 540 minus 90 minus 10 minus 5.
For a real review, describe what the automation did, the basis for removing records before screening and whether its limitations were assessed. Do not label manual exclusions as automated. Likewise, do not erase uncertain near-duplicates without checking whether they might be distinct reports.
Keep a deduplication log showing the retained record and identifiers associated with each removed duplicate. Your research data management plan should specify approved storage and versioning for these screening records.
Step 4: Screen titles and abstracts
The researcher screens 435 records against the prespecified criteria.
Records screened: 435.
Records excluded at title/abstract stage: 320.
Reports sought for retrieval: 115.
The bookkeeping identity is 435 = 320 + 115.
At title/abstract screening, it is acceptable for the PRISMA diagram to give a single total excluded rather than detailed reasons for every title. More specific reasons are generally most useful at full-text eligibility assessment.
Note whether screening was done by one or more reviewers, whether screening was independent, and how disagreements were handled. Do not claim that two independent reviewers participated when the dissertation researcher screened alone. Single-reviewer screening can carry a greater risk of missed eligible evidence and should be described transparently.
Step 5: Record reports not retrieved
The researcher seeks 115 full reports from the database/register pathway, but 9 cannot be retrieved despite documented attempts.
Reports sought: 115.
Reports not retrieved: 9.
Reports assessed for eligibility: 106.
The reconciliation is 115 = 9 + 106.
Keep a retrieval log with the report identifier, contact or library-access attempt, date and outcome. A report that was not obtained must not be described as assessed and excluded for an eligibility reason, because its full content was not actually assessed.
Step 6: Give one primary exclusion reason for each ineligible full report
Among the 106 reports assessed, the researcher excludes 75 using prespecified criteria:
Ineligible population: 30.
Ineligible intervention or context: 20.
Ineligible outcome: 15.
Ineligible study design: 10.
That leaves 31 included reports on the database/register pathway: 106 minus 75.
| Primary exclusion reason | Reports excluded |
|---|---|
| Population outside review criteria | 30 |
| Intervention or context ineligible | 20 |
| Outcome outside review criteria | 15 |
| Study design ineligible | 10 |
| Total reports excluded | 75 |
| Reports included via this pathway | 31 |
In this fictional example, one primary reason is assigned to each excluded report according to a predetermined hierarchy. This prevents double-counting a report that fails multiple criteria. The review should still retain sufficient detail to explain consequential decisions, including a list of apparently eligible full reports excluded for substantive reasons where required.
Our critical-appraisal guide addresses a different question: how to assess the strengths, weaknesses and potential biases of relevant evidence. An eligible study is not necessarily a high-quality study; inclusion and appraisal are separate methodological decisions.
Step 7: Reconcile the other-methods pathway separately
The 12 reports found through websites and citation searching were not already represented in the database/register pathway after cross-checking identifiers and study identity. In this teaching example:
Other-method reports identified: 12.
Reports sought: 12.
Reports not retrieved: 2.
Reports assessed: 10.
Reports excluded with recorded eligibility reasons: 7.
Reports included: 3.
The equations are 12 = 2 + 10 and 10 = 7 + 3.
For the seven other-method exclusions, imagine four concerned an ineligible population and three an ineligible outcome. These reason counts must appear in the eligibility log; the precise layout of the diagram may be adapted to the official template.
Important: These 12 reports were checked against all prior retrievals, so they are not additional copies of already counted reports. If an item appears in both pathways, document its identity and count it consistently rather than inflating the totals.
Step 8: Distinguish the final reports from the final studies
The database/register pathway yields 31 included reports, but they describe only 28 unique studies because three of those studies have two distinct included reports each.
The other-methods pathway contributes 3 reports describing 3 additional studies not present in the first pathway.
Therefore the final review includes:
34 reports: 31 plus 3.
31 studies: 28 plus 3.
This is not an error. Reports and studies are different units. The review should keep a study-level identifier linking all related reports so that multiple publications do not cause a single study to be treated as independent evidence twice.
Reconciled worked example at a glance
| Pathway and stage | Count | Reconciliation |
|---|---|---|
| Database/register records identified | 540 | 320 + 180 + 40 |
| Duplicate records removed | 90 | Before screening |
| Automation exclusions before screening | 10 | Before screening |
| Other removals before screening | 5 | Before screening |
| Records screened | 435 | 540 − 90 − 10 − 5 |
| Records excluded at screening | 320 | 435 − 115 |
| Reports sought, database/register | 115 | 435 − 320 |
| Reports not retrieved, database/register | 9 | 115 − 106 |
| Reports assessed, database/register | 106 | 115 − 9 |
| Reports excluded, database/register | 75 | 30 + 20 + 15 + 10 |
| Reports included, database/register | 31 | 106 − 75 |
| Other-method reports identified/sought | 12 | 7 websites + 5 citations |
| Other-method reports not retrieved | 2 | 12 − 10 |
| Other-method reports assessed | 10 | 12 − 2 |
| Other-method reports excluded | 7 | 4 + 3 |
| Other-method reports included | 3 | 10 − 7 |
| Total included reports | 34 | 31 + 3 |
| Total included studies | 31 | 28 from first pathway + 3 new studies |
The table serves as an audit-friendly companion to the figure. In a real dissertation, the exact counts should come from the documented screening log—not from adjusting boxes until the diagram looks tidy.
Step 9: Transfer reconciled numbers into an official diagram template
Use the PRISMA 2020 flow-diagram templates. Select the version for a new review with database/register searches and other methods, matching this example.
Enter the counts in their appropriate boxes. Retain the database/register and other-method pathways, report retrieval failures and include the final numbers of both reports and studies. Use your institution's requirements for captions, numbering and figure presentation.
Do not reproduce an old PRISMA 2009 figure merely because it appears in an earlier dissertation. The reporting standard, any relevant extension and the chosen template should match the review.
Accessibility: If you submit a visual diagram, include meaningful alternative text or a text equivalent where your format permits, and ensure the caption identifies what is counted. Keep the underlying selection ledger so the process remains understandable even if a figure cannot be viewed.
Do not invent records to populate blank boxes. Where a template contains a category not applicable to the actual review, follow the official template instructions and explain genuine zeros or omissions appropriately.
Step 10: Write the accompanying Results paragraph
A good Results paragraph describes the search and selection journey without duplicating every figure box.
Fictional dissertation-ready example
Database and register searches identified 540 records. Before screening, 90 duplicates, 10 records classified as ineligible by the prespecified automation procedure and five records removed for documented additional reasons were excluded, leaving 435 records for title-and-abstract screening. Of these, 320 were excluded and 115 reports were sought. Nine reports could not be retrieved; 106 were assessed for eligibility and 75 excluded, leaving 31 reports. An additional 12 reports were identified through websites and citation searching; two were not retrieved, seven of the ten assessed were excluded, and three were included. In total, 34 reports represented 31 distinct studies. The reasons for full-report exclusion are provided in the screening log, and the complete selection process is shown in Figure 1.
This paragraph models reporting structure. It does not describe an actual systematic review. Do not copy it as a factual account unless your own screening logs support its numbers.
What to include in the methodology chapter
The flow diagram typically appears in the Results section, but the methodological explanation should precede it. Explain:
The review question and eligibility criteria.
Databases, registers and other sources searched.
Date of each search and full search strategy or reproducible supplementary record.
Deduplication rules and software, if applicable.
Title/abstract and full-text screening arrangements.
Whether one or more reviewers independently screened items.
How disagreements or uncertainty were resolved.
How unavailable full reports were handled.
How multiple reports were linked to one study.
Where detailed eligibility and exclusion decisions can be audited.
Do not use PRISMA as a substitute for a genuine systematic method. A conventional structured literature search may borrow transparent recordkeeping practices without inaccurately describing itself as a full systematic review. Our dissertation methodology chapter guide explains how to align methodological claims with what was actually done.
Common PRISMA flow-diagram mistakes
Combining raw search hits with unique screened records
Duplicates are removed before screening in the usual flow. If you add raw database exports directly to the screened count, the diagram overstates work completed.
Confusing reports with studies
A trial registration, conference abstract and final article may describe the same study. Map related reports to one study identifier before calculating the final study total.
Treating an unretrieved document as a full-text exclusion
If the complete report was never accessed, it belongs under reports not retrieved, not an eligibility reason inferred from its abstract.
Giving overlapping exclusion reasons that do not sum correctly
For the flow, count each excluded full report once using the declared primary reason. Keep other relevant reasons in an audit log if useful.
Counting citation-search discoveries twice
Check citations and website items against the master library and included-study identities. Document the handling of duplicates across pathways.
Using software as a substitute for review judgment
Deduplication, machine classification and screening tools can help manage records, but their decisions and limitations require transparency. A generated figure cannot validate the underlying judgments.
Inventing reviewer independence
If one postgraduate student screened records alone, say so. Do not write that two independent reviewers resolved disagreements unless that occurred.
Failing to reconcile a final diagram after updates
If a search was updated, refresh the screening ledger, figures and included-study inventory together. Use the appropriate official version for updated reviews when applicable.
Frequently asked questions
Can I use PRISMA for a dissertation literature review?
Use PRISMA 2020 when your dissertation actually conducts a review to which its reporting guidance applies. For a conventional narrative review, you may document searching transparently without claiming full PRISMA compliance.
Is the PRISMA diagram the same as a PRISMA checklist?
No. The flow diagram describes selection counts. The checklist covers a much broader set of reporting items across the review.
Should I include records excluded by automation tools?
Yes, if automation removed records before screening, record that category transparently and explain the procedure. Do not invent automated removals if no such tool was used.
Do I need a reason for every abstract excluded?
A total number excluded at title/abstract screening is usually sufficient for the flow diagram, although maintaining decision records can improve auditability. At full-report eligibility assessment, report the principal exclusion reasons.
What if I cannot obtain a full report?
Record the retrieval attempts and count it as not retrieved if the full report is unavailable. Do not pretend its full text was assessed.
Why can included reports exceed included studies?
One study can have multiple publications or documents. Report both totals and link the reports to the underlying study.
Can I create a PRISMA diagram in Word or Excel?
Yes, using an appropriate editable template and your reconciled selection counts. Software is a drawing or data-entry aid; it does not establish that the underlying review was systematic or that your totals are accurate.
Must PRISMA 2020 be used for a scoping review?
Scoping reviews commonly use PRISMA-ScR, the scoping review extension, with appropriate selection reporting. Choose guidance that fits your actual review rather than assuming every evidence synthesis has identical reporting requirements.
Checklist: audit your diagram before submission
Have I used the appropriate PRISMA 2020 template or relevant extension?
Does the review question match the stated eligibility rules?
Can I reproduce the raw identification totals from database exports and source logs?
Are removed duplicates distinct from other pre-screening exclusions?
Are records screened equal to identified minus removed records?
Are title/abstract exclusions and reports sought reconciled?
Are reports not retrieved distinguished from full-report exclusions?
Does the eligibility log explain each excluded full report once?
Have cross-pathway duplicates been handled consistently?
Do I know which reports describe the same underlying study?
Do final included reports and included studies have separate accurate totals?
Does the Methods account match the actual review procedure and staffing?
Does the Results paragraph use exactly the same counts as the diagram?
Are all numbers and reasons grounded in a preserved screening log?
Is the final figure accessible, legible and properly captioned?
Final takeaway
A PRISMA 2020 flow diagram is a compact, auditable account of how evidence moved from search results to included studies. The most common source of trouble is not drawing the arrows. It is mixing units or losing track of decisions.
Preserve raw exports, deduplicate carefully, track retrieval and exclusions, link multiple reports to underlying studies, and reconcile every number before placing it in a diagram. Your systematic review will be more transparent when readers can follow both the selection process and the reasons supporting it.
References and further reading
Page, M. J., et al. (2021). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ, 372, n71. https://doi.org/10.1136/bmj.n71.
Page, M. J., et al. (2021). PRISMA 2020 explanation and elaboration. BMJ, 372, n160. https://doi.org/10.1136/bmj.n160.
PRISMA Statement. (n.d.). PRISMA 2020 official resources.
PRISMA Statement. (n.d.). PRISMA 2020 flow diagram templates.
Rethlefsen, M. L., et al. (2021). PRISMA-S: An extension to the PRISMA statement for reporting literature searches in systematic reviews. Journal of the Medical Library Association, 109(2), 174–200. https://doi.org/10.5195/jmla.2021.962.
Rethlefsen, M. L., & Page, M. J. (2022). PRISMA 2020 and PRISMA-S: Common questions on tracking records and the flow diagram. Journal of the Medical Library Association, 110(2).
https://doi.org/10.5195/jmla.2022.1449.
Editorial note: All search and screening counts in this guide are hypothetical and are provided only to explain the method. Students must use their own actual search records and approved review protocol.
Need help improving your dissertation systematic review? Explore Acadelyra's academic support services for help with search documentation, research-writing clarity, source organisation, editing and responsible presentation of methods and results. The actual screening decisions and research claims remain the researcher's responsibility.