- Deliver the protocol, search log, reconciled flow and extraction table.
Question and protocol
Where appropriate, define population, intervention or exposure, comparison and outcome. Specify eligibility, sources, screening and risk-of-bias methods before examining results. Protocol registration depends on the topic and registry requirements.
Search and screening
Record reproducible search strings, execution dates and databases. Deduplicate records, screen titles and abstracts, then assess full texts against predefined criteria. Document reasons for full-text exclusion.
Extraction and appraisal
Record design, sample, instruments, effect estimates and limitations. Reporting quality and risk of bias are distinct. Select appraisal tools suitable for the study design. Quantitative synthesis requires a defensible comparison of questions and methods.
Reporting
PRISMA is reporting guidance, not a quality certificate or a replacement for review methods. Report the flow diagram, full search strategies, study characteristics and limitations. Keep conclusions within the evidence.
Choose the review type
Systematic, scoping and narrative reviews have different purposes and methods. Match the question to evidence volume and team resources before starting. Mapping concepts differs from estimating an intervention effect; do not select the label afterward merely for appeal.
Check extraction and synthesis
Pilot the extraction form and define its fields. Explain conflict resolution and team roles in screening and extraction. Check units, direction and design before combining estimates. The availability of meta-analysis software does not justify pooling every study.
Preserve the review record
Retain the protocol, full searches, flow diagram, study table, bias assessment and synthesis rationale. Record update dates and added or removed studies. Explain the consequences of language or full-text restrictions.
Worked case and implementation decisions
The following is a fictional teaching case. Do not use its numbers or wording as actual study findings.
For a fictional active-learning review, define the question and eligibility before a reproducible search. Illustrative accounting: 120 records, 20 duplicates, 100 screened, 15 full texts assessed, five studies included and ten full-text exclusions with reasons. Replace these synthetic counts with actual records and distinguish record, report and study: one study can have multiple reports. PRISMA guides reporting; using its flow diagram alone does not validate searching, eligibility or risk-of-bias assessment.
| Stage | Teaching example | Verification question |
|---|---|---|
| Protocol | Question and eligibility before search | Are changes dated? |
| Search | Full query, interface and date | Can a colleague repeat it? |
| Screening | 100 illustrative records | Are disagreements and exclusions logged? |
| Extraction | Study ID and linked reports | Is double counting avoided? |
| Synthesis | Method suited to heterogeneity | Are bias and limits reflected? |
Exercise output: Deliver the protocol, search log, reconciled flow and extraction table.
Sources and further reading
Official sources for verification and further reading