How to Write a Systematic Review
A complete guide to conducting and writing a systematic review, from protocol registration to PRISMA reporting, data extraction, and synthesis.
Guide
Systematic reviews are among the most impactful publications in evidence-based research. A well-conducted systematic review synthesizes all available evidence on a question using transparent, reproducible methods. Publishing one requires mastering a specific methodology: protocol registration, systematic search, formal screening, data extraction, quality assessment, and synthesis. This guide covers each stage of the systematic review process.
Registering your protocol
Before conducting a systematic review, register your protocol on PROSPERO (health research) or OSF (social and behavioral sciences). Registration documents your search strategy, inclusion criteria, and analysis plan before the search begins, preventing post-hoc modifications. The registration number must be cited in your published review. Many journals now require proof of prior registration.
Developing your search strategy
A systematic review search must be comprehensive: it covers all major databases relevant to your topic (PubMed, EMBASE, Web of Science, Scopus, CINAHL for health; PsycINFO for psychology), uses controlled vocabulary (MeSH terms) combined with free-text terms, and documents every decision. The search should be run by a trained librarian where possible. Include grey literature (theses, conference proceedings) to minimize publication bias.
Screening and study selection
Screening is done in two stages: title/abstract screening followed by full-text screening. Both stages should be conducted independently by two reviewers, with conflicts resolved by a third. Use pre-defined inclusion/exclusion criteria. Document the number of studies at each stage in a PRISMA flow diagram. The PRISMA flow diagram is required by most journals publishing systematic reviews.
Data extraction and quality assessment
Extract data using a standardized form, tested on a pilot sample. Extract study design, sample characteristics, intervention details, outcomes, and results. Assess risk of bias using validated tools (Cochrane Risk of Bias 2.0 for RCTs, ROBINS-I for non-randomized studies, GRADE for evidence quality). Two reviewers should independently assess each included study.
Synthesis and meta-analysis
If included studies are sufficiently homogeneous, conduct a meta-analysis. Report heterogeneity (I² statistic), pooled effect estimates with confidence intervals, and sensitivity analyses. If meta-analysis is not appropriate due to heterogeneity, use narrative synthesis with structured tables. Describe synthesis decisions transparently.
PRISMA reporting
All systematic reviews must follow PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines. PRISMA 2020 includes 27 checklist items and a flow diagram. A completed PRISMA checklist must be submitted alongside your manuscript. Non-compliance with PRISMA is a common reason for systematic review rejection.
Ready to check your manuscript?
Upload your paper and get a submission readiness report in under 2 minutes.
Check my manuscript, it's freeNo account · PDF, Word, LaTeX · Results in <2 min
Frequently asked questions
Yes: prospective registration on PROSPERO or OSF is now expected by most journals publishing systematic reviews. Registration must occur before the search begins. The registration number is cited in your published review to demonstrate pre-specified methods.
At minimum two independent reviewers are required for screening and data extraction, with a third to resolve disagreements. Single-reviewer systematic reviews are generally not acceptable in high-quality journals. Report inter-rater reliability (Cohen's kappa) for screening decisions.
For health research: PubMed/MEDLINE, EMBASE, Cochrane Library, Web of Science, CINAHL, and one or more grey literature sources. For social sciences: PsycINFO, Scopus, ERIC. The specific databases depend on your topic. Always consult a research librarian for database selection.
A full systematic review typically takes 6–18 months: 1–2 months for protocol development and registration, 1–2 months for searching and screening, 2–4 months for data extraction and quality assessment, 2–4 months for synthesis and writing. Rushed systematic reviews have higher risk of errors and bias.