LN: Page et al. (2021) — The PRISMA 2020 Statement
Bibliographic Reference
Citation: 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
Pass 1 — Bird’s Eye View (5 Cs)
| C | Assessment |
|---|---|
| Category | Reporting guideline |
| Context | Update to the original PRISMA 2009 statement. Endorsed by the International Committee of Medical Journal Editors and adopted across disciplines including SE |
| Correctness | Consensus document from 49 international experts; tested for clarity with systematic review authors |
| Contributions | (1) 27-item PRISMA checklist; (2) PRISMA flow diagram (updated to show records from databases vs. registers); (3) Abstract checklist; (4) Guidance on AI-assisted screening transparency |
| Clarity | Excellent — prescriptive with worked examples for each checklist item |
Relevance: ⭐⭐⭐⭐⭐
PUMA’s PRISMA-trAIce protocol is built on PRISMA 2020 extended with AI transparency requirements (Marco Veritas). Every AI-assisted screening decision in PUMA’s SLR is documented per the PRISMA 2020 flow.
Pass 2 — Key Concepts
The PRISMA 2020 Flow Diagram
Identification
├── Records from databases (n=__)
├── Records from registers (n=__)
└── Records from other sources (n=__)
↓ Duplicates removed (n=__)
Screening
├── Records screened (n=__)
├── Records excluded (n=__)
└── Full-text assessed (n=__)
↓ Full-text excluded (n=__, reasons)
Included
└── Studies included in synthesis (n=__)
The 2020 update separates database and register records (previously merged), and adds a path for “other sources” to improve transparency.
The 27-Item Checklist (Key Items for PUMA)
| Item | Description | PUMA Section |
|---|---|---|
| 1 | PRISMA statement identified | Methods |
| 2 | Structured abstract | Abstract |
| 5 | Research questions (PICO) | Methods 2.1 |
| 6 | Eligibility criteria (inclusion/exclusion) | Methods 2.2 |
| 7 | Information sources (databases, dates) | Methods 2.3 |
| 8 | Search strategy (full string) | Appendix |
| 9 | Selection process | Methods 2.4 |
| 10 | Data extraction | Methods 2.5 |
| 12 | Quality assessment | Methods 2.6 |
| 16 | Study selection flow diagram | Figure 1 |
| 17 | Study characteristics | Results 3.1 |
| 25 | Funding sources | Acknowledgements |
| 27 | Registration | Methods 2.1 |
AI-Assisted Screening and PRISMA 2020
PRISMA 2020 explicitly addresses AI-assisted screening: reviewers must document which steps were AI-assisted, which model was used, and confirm that human validation was performed. This aligns with PUMA’s PRISMA-trAIce protocol.
PUMA Integration
- Ch.2 Literature Review: PRISMA 2020 flow diagram as Figure 1 (search → screened → included)
- PRISMA-Log: Documents all 27 checklist items
- PRISMA-trAIce: PUMA’s extension of PRISMA 2020 for AI-assisted screening (adds AI tool, model version, validation date)
Related Notes
- PN-Wilcoxon-FINER-Cornell-PRISMA — PRISMA integration with PUMA research protocol
- LN-Kitchenham-2007-SLR — SLR protocol that PRISMA documents
- PRISMA-Log — PUMA’s active PRISMA flow log
- PN-Veritas-Framework — Marco Veritas: AI transparency layer on PRISMA