Introduction to Systematic Reviews

A systematic review is a rigorous, transparent, and reproducible method for answering a specific research question. It involves identifying, selecting, critically appraising, and synthesizing all relevant studies on a topic. Although the approach originated in medicine, it can be applied to social sciences, public policy, education, and many other fields. The hallmark of a systematic review is its transparency: anyone who follows the documented steps should arrive at the same set of results. This distinguishes it from narrative or traditional literature reviews, which are often selective and less reproducible. Meta‑analysis, a statistical technique for pooling numerical data, is frequently attached to systematic reviews but is not mandatory.

Formulating a Research Question

The first and most crucial step is to craft a clear, focused research question. Frameworks such as PICO (Population, Intervention, Comparator, Outcome) and PEO (Population, Exposure, Outcome) provide structured ways to break down the question. For qualitative work, the SPIDER framework (Sample, Phenomenon of Interest, Design, Evaluation, Research type) is useful. In a workshop example, the topic “social media use and depression among adolescents” fits the PEO model:
- Population – adolescents
- Exposure – social media use
- Outcome – depression

A well‑defined question guides every later decision, from keyword selection to inclusion criteria.

Protocol Registration

Before any searching begins, the review protocol should be registered on an international repository such as PROSPERO. Registration locks in the eligibility criteria, search strategy, outcomes of interest, and risk‑of‑bias tools, ensuring methodological transparency and preventing unnecessary duplication of effort. It also provides a public record that reviewers can refer to if they need to justify methodological changes.

Building a Search Strategy

A strong search strategy is the engine that drives study identification. The process starts with listing primary keywords and their synonyms, then combining them with Boolean operators AND (to require both concepts) and OR (to capture alternatives). Truncation symbols (e.g., *) broaden terms to include multiple word endings.

Key databases to search include PubMed, Scopus, Web of Science, and PsycINFO. In the practical demonstration, PubMed was used with both MeSH terms and title/abstract keywords. For example, a PubMed query for the workshop topic might look like:

(adolescents[MeSH Terms] OR teen*) AND (social media[Title/Abstract] OR "online networking"[Title/Abstract]) AND (depression[MeSH Terms] OR depressive[Title/Abstract])

Defining Inclusion and Exclusion Criteria

Clear criteria prevent bias and keep the review focused. Inclusion criteria typically specify the target population, study design (e.g., randomized trials, cohort studies), outcomes, language, and publication date range. Exclusion criteria might rule out studies lacking a control group, those not peer‑reviewed, or papers published in languages the team cannot translate. Explicit criteria make the screening process objective and reproducible.

Study Selection Process

Screening proceeds in three stages:

1. Title screening – discard obviously irrelevant records.
2. Abstract screening – evaluate whether the abstract meets the criteria.
3. Full‑text review – confirm eligibility and extract detailed information.

At least two independent reviewers should conduct each stage, resolving disagreements through discussion or a third reviewer. The PRISMA flowchart visualizes the number of records at each step, providing a transparent audit trail.

Data Extraction

Once studies are selected, a standardized extraction form captures essential details: author, country, sample size, study design, exposure or intervention specifics, outcome measures, effect sizes, and key findings. This structured data collection directly supports answering the original research question.

Risk of Bias Assessment

Assessing methodological quality is vital for interpreting results. The choice of tool depends on study design:

• Cochrane RoB for randomized trials
• Newcastle‑Ottawa for observational studies
• CASP for qualitative research
• AXIS for cross‑sectional studies
• PEDro for physiotherapy trials

A systematic review should report the bias assessment for each study; high‑risk studies may be excluded or discussed as limitations.

Data Synthesis

Two main synthesis approaches exist:

• Meta‑analysis pools quantitative results when studies are sufficiently homogeneous, providing a combined effect estimate with confidence intervals.
• Narrative synthesis summarizes findings textually when data are heterogeneous or qualitative.

Often, reviewers employ both: a meta‑analysis for comparable outcomes and a narrative overview for the broader context.

Common Beginner Mistakes

New reviewers frequently stumble on:

• Vague or overly broad research questions
• Weak search strategies that miss key studies
• Relying on a single database
• Forgetting to register the protocol
• Mixing incompatible study designs in the same analysis
• Ignoring risk‑of‑bias assessment

These errors can compromise the credibility and usefulness of the final review.

Practical Roadmap for Beginners

1. Pick a focus topic that interests you and has feasible literature.
2. Frame the question using PICO, PEO, or SPIDER.
3. Register the protocol on PROSPERO or a similar platform.
4. Develop a comprehensive search strategy across multiple databases.
5. Screen studies using title, abstract, and full‑text stages with at least two reviewers.
6. Extract data with a standardized form.
7. Assess risk of bias using the appropriate tool for each study design.
8. Synthesize the evidence via meta‑analysis, narrative synthesis, or both.
9. Report the findings following PRISMA guidelines.

Importance and Hierarchy of Evidence

Systematic reviews occupy the apex of the evidence pyramid. Because they aggregate all available data on a question, they carry substantial weight in shaping clinical guidelines, policy decisions, and future research agendas. A well‑conducted review can therefore influence practice on a large scale.

Support Resources from Centini

Centini offers training, journal‑submission assistance, and collaborative software (Centini Gate) designed for systematic review teams. Their Full Systematic Review and Meta‑Analysis Course and the Systematic Review Network (SRN) provide additional tutorials, checklists, and community support for both novices and experienced reviewers.

Q&A Highlights

• Mixed‑design checklists: Reviewers should select appraisal tools that match each study’s design rather than using a single checklist for all.
• Risk of bias vs. critical appraisal: The former focuses on methodological flaws that could skew results; the latter evaluates overall study quality.
• Manual searching: Reference lists of included papers are valuable for locating additional studies not captured by database searches.
• Adjusting the question: If preliminary searches reveal insufficient literature, the question can be broadened or reframed while documenting the change in the protocol.
• PubMed demo: Demonstrated how to combine MeSH terms, synonyms, and Boolean operators to capture the full scope of “social media use and depression among adolescents.”