What makes it systematic
A systematic review is defined by explicit, reproducible methods—not simply by reading many papers. The eligibility criteria, search strategy, screening process, data extraction, appraisal and synthesis should be planned before the review results are known.
Approval and governance
Usually needs formal review / authorisation
- Institutional research-office approval if required by local policy for research conducted under the institution.
- Permission/terms compliance for any non-public datasets or licensed full-text/data sources used.
May follow a lighter or different route
- Reviews restricted to published or publicly available aggregate data often do not involve new human participants or identifiable patient data and may not require human-subject ethics review in many systems. The institution should provide the applicable determination.
- Protocol registration is not ethics approval. PROSPERO or another registry may be appropriate depending on review type and eligibility.
Do not do this
- Do not label a narrative search as “systematic” if the search, screening and selection methods are not reproducible.
- Do not alter inclusion criteria after seeing which studies support the preferred conclusion without documenting and justifying an amendment.
Mediclinic Middle East publicly states that research projects carried out at MCME are to receive approval from its internal Research and Ethics Committee and applicable local regulatory authorities before initiation. The exact route varies by project, facility and emirate. Dubai projects may involve DSREC depending on applicability. This hub must therefore route users to the Research Office and current local forms rather than declaring a project “ethics exempt.” Institution-specific forms and contacts will be inserted after verification.
Step-by-step workflow
Form a review team
Use at least two people for critical selection/extraction decisions where feasible, include clinical/methods expertise, and involve an information specialist for complex searches. Resolve authorship and conflict-of-interest expectations early.
Define the review question
For intervention questions, PICO is common: Population, Intervention, Comparator, Outcomes. Other review types may use different frameworks. Make the question narrow enough that eligibility can be applied consistently.
Check existing and ongoing reviews
Search recent systematic reviews and protocol registries. Decide whether the new review is needed because evidence is outdated, methods were inadequate, populations/outcomes differ, or a new question is being asked.
Write the protocol before screening
Prespecify objectives, eligibility, information sources, search approach, screening, extraction, risk-of-bias tool, effect measures, synthesis, heterogeneity, subgroup/sensitivity analyses and certainty assessment. Record protocol version/date.
Register the protocol when appropriate
Use an eligible registry such as PROSPERO when the review type qualifies, or another public repository. Registration should occur before outcome-driven decisions are made. Keep registry and protocol consistent and document amendments.
Design the search strategy
Translate the question into concepts, controlled vocabulary and keywords. Avoid adding outcome terms when they unnecessarily reduce sensitivity. Adapt syntax for each database and record exact search dates.
Choose information sources
Select databases appropriate to the field, plus trial registries, reference lists, citation searching and grey literature when relevant. A single database is rarely enough for a high-quality clinical systematic review.
Run searches and preserve the raw exports
Save the exact strategies, result counts and original export files. Deduplicate using a reproducible method and record the number removed.
Pilot eligibility criteria
Two reviewers independently test a small sample of titles/abstracts and full texts. Clarify ambiguous criteria before full screening without changing the scientific question.
Screen titles/abstracts and full texts
Apply prespecified criteria consistently. At full text, record one defensible primary reason for exclusion so the PRISMA flow and excluded-study record can be reproduced.
Create and pilot the extraction form
Capture study characteristics, population, intervention/exposure, comparator, outcomes, time points, numerical data, funding/conflicts and information needed for risk-of-bias assessment.
Extract data with verification
Use duplicate independent extraction or one extractor plus independent verification for critical data. Resolve discrepancies by consensus or a third reviewer.
Assess risk of bias with a design-appropriate tool
Risk of bias is not the same as a generic “quality score.” Use a tool suited to the included study designs and report domain-level judgments with supporting reasons.
Decide whether synthesis is appropriate
Group studies according to clinical and methodological similarity. A systematic review does not require a meta-analysis. If pooling would hide incompatible populations, interventions, outcomes or designs, use a structured narrative synthesis instead.
Conduct quantitative synthesis if justified
Prespecify effect measures and model, examine heterogeneity, investigate planned subgroups, perform sensitivity analyses and avoid data-driven subgroup fishing. See the Meta-analysis guide for details.
Assess certainty of evidence when appropriate
For questions where recommendations or strength of evidence matter, use an explicit framework such as GRADE or the approach appropriate to the review type. Separate certainty from statistical significance.
Write with PRISMA 2020
Complete the 27-item PRISMA checklist and the correct flow diagram. Report registration/protocol, search strategies, selection, extraction, risk of bias, synthesis methods, results, limitations, funding and conflicts.
Update and archive
Store protocol versions, searches, screening decisions, extraction tables, risk-of-bias assessments, analysis code and the final manuscript. Record the final search date so future updates are possible.
Common errors
- Searching only PubMed and calling the review comprehensive.
- One-person screening with no verification for a high-stakes review.
- Changing outcomes or eligibility after seeing results.
- Using a pooled estimate despite major clinical incompatibility.
- Treating “no statistically significant difference” as proof of equivalence.
- Not providing full reproducible search strategies.
- Failing to report excluded full texts and reasons.
Final checklist
- Question and eligibility criteria prespecified.
- Protocol written and versioned.
- Registration considered/completed where appropriate.
- Search reproducible across appropriate sources.
- Duplicate/verified screening and extraction process defined.
- Risk-of-bias tool appropriate to included designs.
- Meta-analysis performed only if justified.
- PRISMA 2020 checklist and flow diagram completed.