The Problem: Literature Reviews That Can't Be Evaluated or Reproduced
You read a paper that claims to review "all available evidence" on a topic. But you don't know: what databases were searched? What search terms were used? How were studies selected or excluded? How were conflicting studies handled? You can see the conclusions but you can't evaluate whether the process that produced them was rigorous or cherry-picked.
This is the problem with narrative literature reviews — the traditional form, where a researcher reads relevant papers and synthesizes them — and why they're systematically less trusted than systematic reviews. A narrative review's credibility depends entirely on trusting the author's judgment; its conclusions can't be independently reproduced or evaluated.
The PRISMA method for systematic reviews is a standardized framework for conducting and reporting literature reviews that makes every stage of the process explicit, transparent, and reproducible. It's the gold standard for evidence synthesis in medicine, public health, psychology, and increasingly in social science and education research.
Where the PRISMA Method Comes From
PRISMA stands for Preferred Reporting Items for Systematic Reviews and Meta-Analyses. It is a reporting guideline — a standardized checklist and flow diagram — developed by an international group of researchers, clinicians, and methodologists.
Historical context:
The Cochrane Collaboration (established 1993, named for British epidemiologist Archie Cochrane): founded specifically to produce, maintain, and disseminate systematic reviews of healthcare evidence. Cochrane reviews established the systematic review as the highest-quality evidence synthesis available, far superior to expert opinion or narrative reviews.
QUOROM (Quality of Reporting of Meta-Analyses, 1999): the first systematic reporting standard for meta-analyses in medicine, published in the Lancet. PRISMA superseded QUOROM.
PRISMA 2009 (Liberati et al., PLoS Medicine, 2009): the original PRISMA statement, with 27-item checklist and the PRISMA flow diagram. Became the standard reference for systematic review reporting across many disciplines.
PRISMA 2020 (Page et al., BMJ, 2020): updated version with 27 items across 5 sections (Title, Abstract, Introduction, Methods, Results, Discussion, Other Information), reflecting advances in methodology and changes in how research is conducted (automated search tools, living systematic reviews, preregistration).
PRISMA is now required or strongly recommended by hundreds of journals and is the standard framework in any field where evidence synthesis matters.
The PRISMA Method for Systematic Reviews, Step by Step
Step 1: Define the Research Question (PICO or Similar)
Before beginning a PRISMA-compliant systematic review, you need a specific, answerable research question. Medical systematic reviews typically use the PICO framework:
- Population: who are the participants?
- Intervention: what intervention is being studied?
- Comparison: what is it compared to?
- Outcome: what is measured?
PICO example: "In adults with type 2 diabetes (P), does a Mediterranean diet intervention (I) compared to a standard low-fat diet (C) improve HbA1c levels at 12 months (O)?"
Non-clinical reviews use variations (SPIDER, PEO, PICOC) adapted to qualitative or social science questions, but the principle is the same: the question is specific enough to determine exactly what evidence is relevant.
Preregistration (strongly recommended): Register the review protocol in PROSPERO (for health reviews) or the Open Science Framework before beginning the search. Preregistration prevents outcome reporting bias (deciding which analyses to report after seeing results).
Step 2: Design the Search Strategy
Define databases to search: For medical reviews: MEDLINE/PubMed, Embase, Cochrane CENTRAL, and field-specific databases. For psychology: PsycINFO. For education: ERIC. For social science: Web of Science.
Define search terms: Develop a comprehensive search string using:
- MeSH (Medical Subject Headings) terms or equivalent controlled vocabulary
- Free-text synonyms for each concept
- Boolean operators (AND, OR, NOT)
Example search string:
("type 2 diabetes" OR "T2DM" OR "type II diabetes") AND ("Mediterranean diet" OR "Med diet" OR "Mediterranean-style") AND ("glycated hemoglobin" OR "HbA1c" OR "blood glucose")
Document the search: Record the exact search string used in each database, the date of the search, and the number of results. PRISMA requires this documentation for the Methods section.
Additional sources: Search trial registries (ClinicalTrials.gov), reference lists of included studies, grey literature (government reports, conference proceedings), and contact experts if appropriate.
Step 3: Screen Results in Two Stages
Stage 1 — Title and abstract screening:
Apply your inclusion/exclusion criteria to each search result at the title and abstract level.
- Clearly articulate inclusion criteria (what makes a study eligible)
- Clearly articulate exclusion criteria (what disqualifies a study)
- Mark each result: Include, Exclude, or Uncertain
- Two independent reviewers screening each record is the PRISMA standard; resolve disagreements by consensus or third reviewer
Stage 2 — Full-text review:
For records not excluded at abstract screening, retrieve full texts and apply more detailed criteria:
- Does the study design meet requirements (e.g., must be RCT)?
- Does the sample meet the population definition?
- Is the outcome measured?
- Is the follow-up period appropriate?
Document all exclusions at full-text stage with reasons.
Step 4: Complete the PRISMA Flow Diagram
The PRISMA flow diagram is the most recognizable element of the method — a visual representation of the screening process:
Records identified through database searching (n = X)
+
Records identified through other sources (n = X)
↓
Records after duplicates removed (n = X)
↓
Records screened (title/abstract) (n = X) → Records excluded (n = X)
↓
Full-text articles assessed for eligibility (n = X) → Full-text excluded with reasons (n = X)
↓
Studies included in synthesis (n = X)
Every number in the flow diagram must be accurate and documented. The flow diagram makes the screening process fully transparent and reproducible.
Step 5: Extract Data from Included Studies
Design a data extraction form (usually a spreadsheet) and extract from each included study:
- Study identification (author, year, country)
- Study design (RCT, cohort, etc.)
- Population characteristics (sample size, age, clinical characteristics)
- Intervention details (type, dose, duration)
- Comparison details
- Outcome measures (what was measured, when, how)
- Key results
- Risk of bias assessment
Risk of bias assessment: For each included study, assess the risk of bias using a standardized tool:
- RCTs: Cochrane Risk of Bias 2 (RoB 2) tool
- Observational studies: Newcastle-Ottawa Scale or ROBINS-I
- Document the assessment in a risk-of-bias table
Step 6: Synthesize and Report Results
Narrative synthesis: Describe what the evidence shows, organizing by outcome, study design, or population. Compare and contrast studies.
Meta-analysis (if appropriate): When included studies are sufficiently homogeneous (similar populations, interventions, outcomes), pool results statistically using meta-analysis to produce a summary effect estimate (e.g., pooled mean difference in HbA1c).
GRADE the evidence: In health research, use the GRADE framework (Grading of Recommendations Assessment, Development and Evaluation) to rate the certainty of evidence for each outcome (High, Moderate, Low, Very Low).
Report completely: The PRISMA 2020 27-item checklist covers every aspect of the report. Key sections:
- Title (include "systematic review" and "meta-analysis" if applicable)
- Abstract (structured, with results)
- Introduction (rationale, objectives)
- Methods (eligibility, sources, search, selection, extraction, synthesis)
- Results (selection, characteristics of included studies, risk of bias, synthesis)
- Discussion (limitations, certainty of evidence, conclusions)
A Worked Example
A PhD candidate in public health, Sam, conducts a PRISMA systematic review on the effect of mindfulness-based interventions on workplace burnout.
PICO:
- P: Adult employees in workplace settings
- I: Mindfulness-based interventions (MBSR, MBCT, mindfulness apps)
- C: Control conditions (waitlist, treatment as usual, active control)
- O: Burnout (measured by Maslach Burnout Inventory or equivalent), 8+ weeks follow-up
Search: 4 databases (MEDLINE, PsycINFO, Embase, Cochrane CENTRAL); search terms combining mindfulness synonyms AND burnout synonyms AND workplace/occupational terms. Date: up to August 2026.
Screening:
- 3,847 records identified after deduplication
- 3,614 excluded at title/abstract (not RCT, not workplace, not burnout outcome, etc.)
- 233 full texts reviewed
- 187 excluded at full text (wrong follow-up, single-session intervention, not validated burnout scale)
- 46 studies included
PRISMA flow diagram: Sam completes the flow diagram with exact numbers at each stage.
Extraction and synthesis:
- 46 RCTs, n = 6,892 total participants
- Meta-analysis pooled effect: standardized mean difference (SMD) = −0.48 (95% CI: −0.62 to −0.34), indicating moderate reduction in burnout
- Risk of bias: 22 studies low risk, 18 moderate, 6 high
- GRADE certainty: Moderate (downgraded for some imprecision and risk of bias)
Report: Sam submits to a peer-reviewed journal with the 27-item PRISMA checklist as supplementary material. Reviewers can verify every methodological decision.
Where the PRISMA Method Shines
For health and clinical research: PRISMA is the required standard for systematic reviews submitted to major medical journals (JAMA, Lancet, BMJ, NEJM). Without it, systematic reviews of clinical evidence will not be accepted by major publications.
For policy-relevant research: Government health agencies (NICE, CDC, WHO) base clinical guidelines on systematic reviews; PRISMA's transparency makes it possible to evaluate the evidence behind guidelines.
For PhD dissertations in health sciences: A systematic review chapter following PRISMA is a standard and well-regarded dissertation component in medicine, nursing, public health, and psychology.
Where the PRISMA Method Breaks Down
For rapid reviews: PRISMA's rigor comes with significant time cost — a full systematic review typically takes 12-18 months for a team. Rapid reviews (conducted under time pressure for policy purposes) can follow simplified versions of PRISMA guidelines but involve methodological trade-offs.
For qualitative research: PRISMA was developed primarily for quantitative synthesis. Qualitative systematic reviews (meta-ethnography, thematic synthesis) have their own reporting guidelines (ENTREQ, COREQ) that parallel PRISMA's structure.
For non-health disciplines: PRISMA is being adopted in education, social science, and management research, but is less established. Some disciplines use adapted frameworks (PRISMA-P for protocols, PRISMA-EcoEvo for ecology).
PRISMA vs. Other Evidence Synthesis Approaches
| Approach | Transparency | Reproducibility | Time required | Typical use |
|---|
| Narrative review | Low | Low | Low | Editorials, expert opinion |
| Scoping review | Medium | Medium | Medium | Mapping literature, identifying gaps |
| PRISMA systematic review | High | High | High | Clinical guidelines, policy |
| Cochrane review | Highest | Highest | Highest | Gold-standard health evidence |
| Meta-analysis | High | High | High | Statistical pooling of multiple RCTs |
The PRISMA method sits between scoping reviews (less rigorous) and full Cochrane reviews (additional requirements: updated regularly, team review, Cochrane editorial oversight).
Tools That Support the PRISMA Method
| Tool | Role | Notes |
|---|
| PROSPERO | Protocol preregistration (health) | Required for most high-quality health systematic reviews |
| Covidence | Online screening platform | Two-reviewer screening; flow diagram auto-generated |
| Rayyan | Free abstract screening tool | Web and mobile; good for students |
| Zotero / EndNote | Reference management | Import search results; remove duplicates |
| Cochrane RevMan | Meta-analysis and risk of bias | Standard for Cochrane reviews; free |
| PRISMA 2020 checklist | Reporting compliance | Download at prisma-statement.org |
| WebSnips | Grey literature and web source capture | Clip government reports, conference abstracts, policy documents found in web searches |
WebSnips and PRISMA reviews: Not all relevant evidence lives in academic databases. Grey literature — government reports, NGO publications, unpublished studies, conference abstracts available online — is an important source in PRISMA reviews, and its omission is a recognized source of publication bias. WebSnips clips specific documents from web searches for grey literature (government agency sites, policy organization pages), organized into a collection that feeds the grey literature search step documented in the PRISMA Methods section.
Common PRISMA Method Mistakes
Mistake 1: Starting the search before preregistering the protocol.
Once you've seen the results, you can unconsciously bias your inclusion criteria toward studies that support a preferred conclusion. Preregistration in PROSPERO or OSF prevents this.
Mistake 2: Single-reviewer screening.
PRISMA 2020 recommends at least two independent reviewers for abstract and full-text screening. Single reviewer introduces undetected errors and selection bias. At minimum, have a second reviewer check a random 10-20% and calculate inter-rater agreement (kappa statistic).
Mistake 3: Inadequate documentation of exclusions.
"Did not meet inclusion criteria" is not sufficient. Each full-text exclusion needs a specific, documented reason. This is what makes your screening reproducible.
Mistake 4: Ignoring grey literature.
Studies with non-significant results are less likely to be published (publication bias). Grey literature — unpublished studies, conference proceedings, government reports — captures some of these. Neglecting grey literature inflates the apparent effect of interventions.
Frequently Asked Questions
Do I need to conduct a meta-analysis to follow PRISMA?
No. PRISMA applies to systematic reviews whether or not they include meta-analysis. "Systematic reviews and meta-analyses" refers to the breadth of evidence synthesis methods it covers, not a requirement to do both.
How many studies do I need to include?
There's no minimum. A systematic review of 3 highly rigorous studies is more valid than a narrative review of 50 studies with no methodology. However, meta-analysis typically requires 5+ studies for meaningful statistical pooling.
How long does a PRISMA systematic review take?
A solo reviewer on a focused question can complete one in 3-6 months. A full team review (2+ reviewers, complex question, large literature) typically takes 12-18 months. Cochrane reviews can take years.
Key Takeaways
- PRISMA method for systematic reviews is a standardized reporting and process framework (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) that makes literature reviews transparent, reproducible, and bias-resistant.
- Developed from the Cochrane tradition (1993); QUOROM (1999) → PRISMA 2009 → PRISMA 2020; now required by hundreds of journals.
- Six core steps: define PICO question and preregister → design search strategy → screen in two stages (title/abstract then full text) → complete PRISMA flow diagram → extract data and assess risk of bias → synthesize and report.
- The PRISMA flow diagram is the hallmark: a transparent count of records at every stage of screening.
- Two-reviewer independent screening is the standard; inter-rater agreement should be reported.
- Best for: clinical and health research, PhD systematic reviews, policy-relevant evidence synthesis; involves significant time investment (3-18+ months).
Conclusion
The PRISMA method for systematic reviews is the most rigorous framework available for synthesizing evidence across a body of literature. Its demands — preregistration, systematic search, two-reviewer screening, documented exclusions, risk-of-bias assessment — represent a significant time investment. But what PRISMA produces in return is unique: a literature review whose conclusions can actually be evaluated and reproduced, free from the hidden selection biases that make narrative reviews unreliable. For PhD candidates, researchers, and evidence-based practitioners, learning PRISMA isn't just about following a reporting checklist — it's about internalizing a systematic, transparent approach to evidence that distinguishes rigorous scholarship from expert opinion dressed up as research.
Try WebSnips free — capture grey literature and web-based evidence sources for your PRISMA review, organized by search stage, so your grey literature documentation is as complete as your database search.