What Is a Systematic Review? Definition, Steps and Protocol Template

What is a systematic review? ¶

If another reader cannot repeat your search and reach the same set of studies, the review is not yet systematic.

A systematic review is a research study that collects and critically appraises all available studies on a clearly defined question, using explicit, pre-specified methods to minimize bias. Reviewers search several databases, select studies against fixed criteria, assess their quality and synthesize the findings, sometimes with a meta-analysis.

The Cochrane Handbook describes the defining features as:

• a clearly stated set of objectives with pre-defined eligibility criteria;
• an explicit, reproducible methodology;
• a systematic search that attempts to identify all studies that would meet the criteria;
• an assessment of the validity of the included studies, for example through risk of bias assessment;
• a systematic presentation and synthesis of the characteristics and findings of the included studies.

Because of this rigor, well-conducted systematic reviews of randomized trials sit at or near the top of most evidence hierarchies. The phrase systematic literature review means the same thing and is common in education, management and engineering.

Systematic review vs literature review ¶

The difference between a systematic review and a literature review lies in method. A traditional (narrative) literature review summarizes a field selectively, guided by the author's expertise. A systematic review follows a published protocol so that every decision can be checked.

Two neighbors are often confused with it. A scoping review maps how much evidence exists and what it looks like, usually without appraising quality. A meta-analysis is the statistical pooling of results; we compare the two in our guide to meta-analysis vs systematic review.

Pros and cons of systematic reviews ¶

Strengths

• Transparent and reproducible methods that readers and reviewers can audit.
• Reduced selection bias, because inclusion does not depend on which studies the author already knows.
• A defensible answer for clinicians, policymakers and guideline panels.
• A clear map of research gaps that can justify a new primary study or a PhD chapter.

Limitations

• Time and labor: dual screening and extraction need at least two people.
• The review can only be as good as the studies it includes.
• Narrow questions may exclude useful context.
• Results can date quickly in active fields, so reviews need updating.

How to do a systematic review: the 8 steps ¶

Most guidance, including the Cochrane Handbook and the JBI Manual for Evidence Synthesis, follows the same sequence. Each step produces a record you will report later.

1. Formulate the question with a framework such as PICO.
2. Write and register a protocol, for example on PROSPERO.
3. Search several databases and other sources with a documented strategy.
4. Screen titles and abstracts, then full texts, against your eligibility criteria.
5. Extract data with a piloted form.
6. Assess risk of bias in each included study.
7. Synthesize the evidence narratively, statistically or both, and rate its certainty.
8. Report using the PRISMA 2020 checklist and flow diagram.

Steps 1 and 2: the research question, PICO and protocol ¶

A good review question is narrow enough to answer and important enough to matter. PICO splits it into Population, Intervention, Comparator and Outcome. Other frameworks suit other questions: PECO for exposures, SPIDER for qualitative evidence, and PCC (Population, Concept, Context) for scoping reviews.

Writing the protocol

The protocol fixes your methods before you see the results. Report it using PRISMA-P, the 17-item checklist for systematic review protocols. It should state the rationale, eligibility criteria, information sources, draft search strategy, selection and extraction process, risk of bias tools, synthesis plan and how you will assess certainty.

Registering on PROSPERO

PROSPERO, run by the Centre for Reviews and Dissemination at the University of York, registers systematic reviews with a health-related outcome. Register as early as possible, ideally before screening begins; PROSPERO will not accept reviews that have already completed data extraction. Scoping reviews and reviews outside health are usually registered on OSF Registries instead.

SYSTEMATIC REVIEW PROTOCOL

1. Title: [Intervention/exposure] for [outcome] in [population]: a systematic review [and meta-analysis]
2. Registration: [PROSPERO / OSF ID, date]
3. Authors and contributions: [names, affiliations, roles; guarantor]
4. Amendments: [how changes will be dated and reported]
5. Funding and competing interests: [sources; declarations]

6. Rationale: [what is known, what is uncertain, why this review is needed]
7. Objective and review question (PICO):
   Population: 
   Intervention/Exposure: 
   Comparator: 
   Outcomes (primary; secondary): 

8. Eligibility criteria:
   Study designs: 
   Settings and dates: 
   Languages: 
   Exclusions: 

9. Information sources: [databases with platforms; trial registers; grey literature; citation searching; contact with authors]
10. Search strategy: [draft full strategy for one database, in an appendix]
11. Study records:
   Data management: [software]
   Selection process: [two independent reviewers; title/abstract then full text; conflict resolution]
   Data collection process: [piloted form; independent double extraction]
12. Data items: [list of variables to extract]
13. Outcomes and prioritization: [primary and secondary outcomes; time points]
14. Risk of bias in individual studies: [tool per design; how used in synthesis]
15. Data synthesis: [criteria for meta-analysis; effect measure; model; heterogeneity (I2, tau2); subgroup and sensitivity analyses; narrative synthesis method if pooling is not appropriate]
16. Meta-bias: [funnel plot / small-study effects if 10 or more studies; selective reporting]
17. Certainty of evidence: [GRADE; summary of findings table]

Timeline: [search, screening, extraction, analysis, write-up dates]

Steps 3 and 4: searching and screening ¶

Search at least two relevant bibliographic databases, such as MEDLINE (via PubMed or Ovid), Embase, CINAHL, PsycINFO, Scopus or Web of Science, plus trial registers and grey literature. Combine free-text words and subject headings (MeSH or Emtree) for each concept, join synonyms with OR and concepts with AND.

Record the date, database, platform and full string for every search; PRISMA 2020 item 7 asks you to present the full strategy for all databases. Export results to a reference manager or screening tool and remove duplicates.

Screening happens in two stages: titles and abstracts first, then full texts. Two reviewers should work independently and resolve disagreements by discussion or a third reviewer. Tools such as Covidence, Rayyan and EPPI-Reviewer manage blinding, conflicts and counts. Keep a reason for every full-text exclusion, because you will need those counts for the PRISMA flow diagram.

Steps 5 and 6: data extraction and risk of bias ¶

Design a data extraction form around your question and pilot it on two or three studies. Typical fields include study design, country, setting, sample size, participant characteristics, intervention details, comparator, outcomes, time points, results with measures of variance, funding and conflicts of interest. Ideally two reviewers extract independently.

Then assess each study's risk of bias with a tool matched to its design:

Report judgments per domain, not only an overall score, and explain how risk of bias influenced your synthesis.

Steps 7 and 8: synthesis and reporting ¶

If the included studies are similar in population, intervention, comparator, outcome and design, you can pool them in a meta-analysis and present a forest plot. When they are too different, use a structured narrative synthesis and report it with the SWiM (Synthesis Without Meta-analysis) guideline. Either way, rate the certainty of evidence for each key outcome, commonly with GRADE.

Report the review using PRISMA 2020, which has 27 items (several with sub-items) and a separate 12-item checklist for abstracts. A clear structured abstract should state the objective, data sources with search dates, eligibility criteria, risk of bias method, synthesis method, number of included studies and participants, main results, limitations, funding and registration number.

If you would like a second pair of eyes on your protocol, search strategy or PRISMA reporting, our systematic review support works alongside you while you keep full authorship of the review.

Systematic review protocol template ¶

The outline below follows the PRISMA-P structure and the fields PROSPERO asks for. Copy it, replace each prompt with your own decisions, and keep the finished protocol with your registration record. Any later change should be dated and explained in the final paper.

Doctoral students often reuse this protocol as the methods chapter of a thesis. Our guide to PhD thesis structure shows where a review chapter usually fits.