The PRISMA 2020 Workflow: From the Research Question to a Verifiable Flow Diagram

PRISMA 2020 supports transparent systematic-review reporting, but checklists and flow diagrams cannot replace protocols, searches, screening decisions or synthesis methods.

SYSTEMATIC REVIEWS AND PRISMA 2020Prepared by Lê Minh12 minUpdated 29/10/2026

KEY ARGUMENT

PRISMA 2020 guides reporting, not review conduct. Every flow count should be traceable to search files, deduplication logs, screening decisions, full-text exclusions and report–study links.

Start reading

1. What is PRISMA 2020 for?

PRISMA means Preferred Reporting Items for Systematic Reviews and Meta-Analyses. The 2020 version specifies minimum reporting of rationale, methods and findings. Core materials include a 27-item checklist, expanded checklist, abstract checklist and flow templates (Page, McKenzie, et al., 2021).

Interpret “reporting” correctly. PRISMA helps assess completeness and transparency; it does not select databases, risk-of-bias tools or synthesis models. Explain those review-specific methodological choices through a protocol and specialized guidance (Page, Moher, et al., 2021).

Seven-step PRISMA workflow, from protocol to reporting and archiving. — Figure 1. PRISMA reports a documented workflow. Every step should produce assessable files or logs.

Table 1
Component	Scope	Source
PRISMA 2020 checklist	27 systematic-review reporting items	Page, McKenzie, et al. (2021)
Flow diagram	4 templates by review type and identification sources	PRISMA (n.d.-b)
PRISMA-S	16 search-reporting items	Rethlefsen et al. (2021)

Do not call PRISMA a research method

State that the review followed a prespecified protocol and was reported according to PRISMA 2020. Methods must still describe questions, eligibility, searches, screening, extraction, risk of bias and synthesis.

2. Choose appropriate checklists and extensions

PRISMA 2020 mainly addresses intervention-effect reviews. Other systematic reviews can use it while checking relevant extensions. PRISMA recommends the main guideline together with design- and evidence-specific extensions (PRISMA, n.d.-c; Page, Moher, et al., 2021).

Meta-analysis is not required for a systematic review. Without pooling, describe study grouping, synthesis methods and each synthesis in sufficient detail to assess conclusions (Page, Moher, et al., 2021).

Table 2
Product	Guidance	Avoid
Systematic review	PRISMA 2020 and expanded checklist	Submitting only a flow diagram
Protocol	PRISMA-P	Writing a protocol after learning results
Search strategy	PRISMA-S alongside PRISMA 2020	Listing isolated keywords only
Scoping review	PRISMA-ScR	Calling every descriptive review systematic
Diagnostic, IPD, harms or network meta-analysis	PRISMA 2020 plus the relevant extension	Ignoring evidence-specific requirements

3. Fix the question, eligibility and protocol

A protocol records rationale, objectives and planned methods before the team fully knows eligible-study results. PRISMA-P provides a 17-item protocol checklist (Moher et al., 2015); its explanation details each item’s rationale and requirements (Shamseer et al., 2015).

For intervention reviews, PICO defines population, intervention, comparator and outcome. The first three often underpin eligibility; prespecify outcomes, but they are not always exclusion criteria. Excluding a study for not reporting an outcome differs from excluding one that never measured it (McKenzie et al., 2024; Page, Moher, et al., 2021).

Other fields may need different frameworks. Match the framework to the question and operational inclusion/exclusion rules that two reviewers can apply consistently to one report.

Table 3
Component	Advance decision	Record to preserve
Question	Population, phenomenon/intervention, outcome and context	Question version and approval date
Criteria	Design, dates, languages and publication status	Criteria dictionary and borderline examples
Sources	Databases, registers, websites and citation searching	Platform, coverage and search date
Screening	Reviewer count, independence and disagreement resolution	Forms and conflict log
Synthesis	Extracted variables, risk of bias and analysis methods	Extraction form and codebook

Record every amendment

Protocol changes are not inherently unreliable. Undisclosed or result-informed changes create risk. Log each amendment’s content, date, reason and effect, and distinguish planned from performed methods.

4. Design reproducible searches

Record database and platform, execution date, date coverage, verbatim queries, limits, filters and supplementary methods. MEDLINE via PubMed and Ovid are not equivalent descriptions because syntax, fields and term mapping differ. PRISMA-S has 16 items supporting assessment and reproduction (Rethlefsen et al., 2021).

Build a concept matrix

List synonyms, spelling variants and controlled terms for each concept.

Develop the master query

Use OR within concepts and AND between concepts. Test against known relevant papers.

Translate syntax

Adapt fields, truncation, quotes and proximity operators for each platform.

Run and timestamp

Save date, count, export format, filename and verbatim query. Preserve each update rather than overwriting.

Supplement searches

Specify backward/forward citation searching, websites or expert contact.

Update before submission

Describe updates and integration of new batches.

Reporting example in English

We searched [database] through [platform] from [start, if any] to [last date]. Controlled and free-text terms addressed [concepts]. Verbatim queries, filters and source counts appear in Appendix [X]. Supplementary [citation/web/expert] searches occurred on [date]. Language, date and document limits were justified by [reasons].

5. Manage records and deduplicate

Preserve each source export as a separate batch before merging, with source ID, date, count and identifiers. This supports identification-branch counts.

Do not treat deduplication as a black box. Report tools, versions, matching fields and manual checks. Missing DOIs do not prove records differ, and similar titles do not prove duplicates. Compare titles, authors, years, publication sources, pages and identifiers together (Rethlefsen et al., 2021).

Table 4
Field	Example	Purpose
batch_id	SCOPUS_2026-10-20	Trace source and export batch
record_id	Unique internal identifier	Link processing stages
duplicate_of	Retained record ID	Audit deduplication decisions
decision_timestamp	Date/time and reviewer	Version decision records

First reconciliation

Records screened equal identified records minus duplicates and other prescreening removals. Trace every term to management tables rather than reconstructing counts from memory.

6. Screen and record exclusion reasons

PRISMA 2020 requests reviewer counts per record/report, independence, disagreement handling and automation roles (Page, McKenzie, et al., 2021). It does not assume one staffing arrangement for all projects; describe actual practice so readers can assess missed or misclassified evidence (Page, Moher, et al., 2021).

Pilot eligibility criteria

Reviewers independently apply criteria to a small common set, compare decisions and resolve ambiguities before full screening.

Screen records

Save include, exclude or uncertain decisions and their history.

Retrieve full text

Separate unavailable reports from retrieved but ineligible reports.

Assess full text

Assign each excluded report one main reason from a defined list.

Resolve disagreements

Record the final decision, conflict and any clarified rule.

Avoid “irrelevant” as an exclusion reason

Use criterion-specific labels such as wrong population, intervention/exposure, design or context. Distinguish unmeasured outcomes from measured but unreported results.

7. Distinguish records, reports and studies

A record is an indexed title, abstract or entry for a report. A report supplies study information, such as an article, preprint, thesis or registration. A study is the actual data-generating investigation. One study may have several reports and one report multiple duplicate records (Page, Moher, et al., 2021).

Multiple records may identify one report; multiple reports may describe one study. — Figure 2. Records, reports and studies are different counting units. Report–study mapping prevents duplicate evidence.

Table 5
Unit	Example	Common pitfall
Record	One Scopus entry	The same report appears in several sources
Report	Journal article, preprint or thesis	Several reports describe one study
Study	One original investigation or dataset	Counting reports as studies inflates evidence

Create report–study links

Use study_id and report_id. Link shared samples, trial registrations or collection rounds before extraction and study counting.

8. Extraction, risk of bias and synthesis

PRISMA extends beyond inclusion. Report extractor counts, independence, author confirmation, sought variables, risk-of-bias tools and synthesis methods (Page, Moher, et al., 2021).

Pilot extraction forms. Define field types, units, measurement times and source reports. Prespecify selection among multiple eligible results. For derived values, save formulas, inputs and the person responsible.

For each synthesis, state eligible studies and grouping reasons. Meta-analyses need pooled estimates, precision and appropriate heterogeneity measures. Without pooling, describe synthesis methods and avoid vote counting by statistical significance (Page, Moher, et al., 2021).

Table 6
Block	Content	Control
Study characteristics	Design, context, sample and timing	Codebook and permitted values
Outcome/theme	Definition, instrument and results	Source-page links
Risk of bias	Domains, judgments and reasons	Design-appropriate tool
Synthesis	Study groups and data used	Trace results to source reports

A diagram does not replace quality assessment

It documents the literature pathway, not low bias, correct extraction or defensible synthesis.

9. Build the diagram and reconcile arithmetic

Count identified records, prescreening removals, screened records, reports sought/unavailable/excluded with reasons, and included studies/reports. PRISMA provides four templates for new/updated reviews and identification with/without additional sources beyond databases/registers (PRISMA, n.d.-b).

The following counts are hypothetical illustrations, not an actual review.

Hypothetical reconciliation from 1280 identified records to 39 included reports and 34 studies. — Figure 3. Hypothetical arithmetic reconciliation illustrates checking and does not replace the official PRISMA 2020 template.

Table 7
Step	Hypothetical count	Reconciliation
Database records	1.248	Sum of exports
Other-source records	32	Supplementary search log
Duplicate records	286	1.248 + 32 - 286 = 994
Title/abstract exclusions	812	994 - 812 = 182 reports sought
Reports not retrieved	9	182 - 9 = 173 reports assessed
Full-text report exclusions	134	Reasons must sum to 134
Included reports	39	173 - 134 = 39
Included studies	34	39 reports linked to 34 study_id values

Use the correct official template

Choose the PRISMA 2020 template by two decisions: new or updated review, and identification using sources beyond databases/registers. Do not simplify away reports not retrieved or reports excluded.

10. Complete reporting and the handover package

Check reporting while writing, not only at submission. Give a page/section for every item or explain inapplicability. Official materials provide the 27-item and expanded checklists (PRISMA, n.d.-a). Read Explanation and Elaboration for rationale, essentials and examples (Page, Moher, et al., 2021).

Table 8
File	Content	Role
protocol	Plan and amendments	Distinguish planned from post hoc methods
search-log	Sources, queries, dates and counts	Reproduce searches
records-master	Records, sources and deduplication	Reconcile identification
screening-log	Reviewer decisions	Reconcile screening
full-text-exclusions	Excluded reports and reasons	Populate the exclusion branch
study-report-map	report_id and study_id links	Count studies accurately
PRISMA-checklist	Reporting location for every item	Assess completeness

Reporting example in English

This review was reported according to PRISMA 2020 (Page, McKenzie, et al., 2021). Its protocol was [registered/archived] at [platform, ID/URL] on [date]. Searches covered [databases/platforms] through [date]; full queries appear in Appendix [X]. Removing [n] duplicates and [n] prescreening records left [n] title/abstract records. [Number] reviewers screened [independently/nonindependently], resolving disagreements through [procedure]. Full-text assessment covered [n] reports, excluded [n] for defined reasons, and included [n] studies in [n] reports.

11. Common errors

Table 9
Weak practice	Problem	Correction
The study used the PRISMA method	PRISMA does not specify all conduct	Describe protocol, search, screening, extraction and synthesis
Only drawing a diagram	Does not cover 27 items	Complete the checklist with page locations
Not saving queries	Searches cannot be reproduced	Preserve queries, dates, platforms, filters and counts
Counting papers as studies	One study can have several reports	Create study_id and report_id
Combining unavailable and ineligible reports	Different stages and statuses	Keep retrieval_status and full_text_decision separate
Excluding as irrelevant	Does not identify the failed criterion	Use specific reason categories
Inventing counts at project end	Arithmetic errors and lost provenance	Generate counts from record and screening tables
Not reporting AI/automation	Unclear which records were affected	State tools, versions, roles and human verification

12. Submission checklist

Report the protocol, registration and all amendments.
Make eligibility reproducibly operational.
Name each source’s platform and last search date.
Include verbatim queries in an appendix.
Describe deduplication tools, versions, matching fields and checks.
State reviewer counts, independence and conflict resolution.
Separate unavailable reports from full-text exclusions.
Assign one main reason per full-text exclusion.
Count records, reports and studies correctly.
Reconcile diagram counts, decision tables and subtraction.
Make extraction, bias assessment and synthesis assessable.
Describe AI/automation and human verification.
Give reporting locations for all checklist items.
Match the diagram template to review type and sources.

Conclusion

A trustworthy PRISMA workflow starts with a protocol, operational criteria and preserved queries rather than filling diagram boxes. Record, report and study tables generate counts and document decisions.

Checklists and diagrams report documented work. Traceable boxes and explained protocol changes let readers assess searches, study selection and reporting completeness.

References

Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., Shamseer, L., Tetzlaff, J. M., Akl, E. A., Brennan, S. E., Chou, R., Glanville, J., Grimshaw, J. M., Hróbjartsson, A., Lalu, M. M., Li, T., Loder, E. W., Mayo-Wilson, E., McDonald, S., ... Moher, D. (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., Moher, D., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., Shamseer, L., Tetzlaff, J. M., Akl, E. A., Brennan, S. E., Chou, R., Glanville, J., Grimshaw, J. M., Hróbjartsson, A., Lalu, M. M., Li, T., Loder, E. W., Mayo-Wilson, E., McDonald, S., ... McKenzie, J. E. (2021). PRISMA 2020 explanation and elaboration: Updated guidance and exemplars for reporting systematic reviews. BMJ, 372, n160.https://doi.org/10.1136/bmj.n160
PRISMA. (n.d.-a). PRISMA 2020 checklist.https://www.prisma-statement.org/prisma-2020-checklist
PRISMA. (n.d.-b). PRISMA 2020 flow diagram.https://www.prisma-statement.org/prisma-2020-flow-diagram
PRISMA. (n.d.-c). PRISMA extensions.https://www.prisma-statement.org/extensions
McKenzie, J. E., Brennan, S. E., Ryan, R. E., Thomson, H. J., Johnston, R. V., & Thomas, J. (2024). Chapter 3: Defining the criteria for including studies and how they will be grouped for the synthesis. In J. P. T. Higgins, J. Thomas, J. Chandler, M. Cumpston, T. Li, M. J. Page, & V. A. Welch (Eds.), Cochrane handbook for systematic reviews of interventions (Version 6.5). Cochrane.https://training.cochrane.org/handbook/current/chapter-03
Moher, D., Shamseer, L., Clarke, M., Ghersi, D., Liberati, A., Petticrew, M., Shekelle, P., Stewart, L. A., & PRISMA-P Group. (2015). Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Systematic Reviews, 4, 1.https://doi.org/10.1186/2046-4053-4-1
Rethlefsen, M. L., Kirtley, S., Waffenschmidt, S., Ayala, A. P., Moher, D., Page, M. J., Koffel, J. B., & PRISMA-S Group. (2021). PRISMA-S: An extension to the PRISMA Statement for reporting literature searches in systematic reviews. Systematic Reviews, 10, 39.https://doi.org/10.1186/s13643-020-01542-z
Shamseer, L., Moher, D., Clarke, M., Ghersi, D., Liberati, A., Petticrew, M., Shekelle, P., Stewart, L. A., & PRISMA-P Group. (2015). Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015: Elaboration and explanation. BMJ, 350, g7647.https://doi.org/10.1136/bmj.g7647

Methodological note: This is general reporting and workflow guidance. Fields, questions and evidence types may require specialist guidance. All example counts are hypothetical.