Systematic Reviews Explained: Steps, Standards and Student Options

A systematic review is a research project in its own right. Instead of surveying a topic selectively, it follows a defined recipe to find all the relevant evidence on a precise question, judge its quality and combine it. Systematic reviews are the backbone of evidence-based practice in health and are increasingly used in education, social sciences and business.

They are also demanding, and many students are told to do one before they understand what is involved. This guide explains what a systematic review is, how it differs from other reviews, the main steps and whether it is realistic for a student project.

What makes a review systematic

A systematic review has these features:

• A focused question, usually in a structured form.
• A protocol, written in advance, stating the methods.
• Explicit criteria for including and excluding studies.
• A comprehensive search of several sources, with the search strategy reported.
• Screening and selection done carefully, ideally by two people independently.
• Quality or risk-of-bias assessment of each study.
• Structured data extraction.
• Synthesis, narrative or statistical.
• Transparent reporting, following a guideline.

The aim is to reduce bias and make the process reproducible.

Systematic review and other reviews

Our guide to writing a literature review covers the narrative approach used in most dissertations.

The main steps

1. Frame the question

Systematic reviews need a precise question. In health, a common framework is PICO: Population, Intervention, Comparison and Outcome. For example: In adults with chronic low back pain (P), does exercise therapy (I) compared with usual care (C) reduce pain (O)? Other frameworks exist for qualitative and other kinds of questions, such as SPIDER. A narrow, well-defined question makes the rest manageable.

2. Write a protocol

A protocol sets out your question, inclusion and exclusion criteria, search strategy, screening process, appraisal tools and plan for synthesis, before you start. It protects against changing methods after seeing the results. In health, protocols are often registered in a public register, such as PROSPERO, for reviews with health outcomes. Check what your program and field expect.

3. Search

Search several databases suited to your topic, plus other sources such as reference lists, trial registers and grey literature, which means material not published in journals. Build a detailed search string with controlled vocabulary and free-text terms, combined with Boolean operators. Record the exact strategy, the date and the number of results for each source. Many librarians help with search design. Our guide to finding peer-reviewed articles covers the basics.

4. Screen and select

Remove duplicates, then screen titles and abstracts against your criteria, then read full texts of the remaining papers. Best practice is for two reviewers to screen independently and resolve disagreements by discussion. Record reasons for excluding full texts. Software tools can help manage this.

5. Extract data

Design a form to extract the same information from each study: author, year, country, design, sample, intervention, outcomes and results. Pilot it on a few studies. Two people extracting independently, or one extracting and another checking, reduces errors.

6. Appraise quality

Assess how trustworthy each study is, using a recognized tool suited to the design, such as the Cochrane risk-of-bias tool for randomized trials or checklists from the Joanna Briggs Institute or CASP. Use the appraisal in the synthesis and the discussion. Do not simply list scores.

7. Synthesize

Combine the findings.

• Narrative synthesis: describe and compare results in a structured way, grouped by theme, population or outcome.
• Meta-analysis: pool numerical results statistically where studies are similar enough, using software and ideally statistical guidance.
• Qualitative synthesis: approaches such as thematic synthesis combine findings from qualitative studies.

Explain how heterogeneity, meaning differences between studies, affects what you can conclude.

8. Report

Follow a reporting guideline. For most systematic reviews, this is PRISMA, whose 2020 version includes a checklist and a flow diagram that shows how many records were found, screened, excluded and included. Reporting in a standard way lets readers judge the review and compare it with others.

Meta-analysis in brief

A meta-analysis pools results from multiple studies to produce an overall estimate and a confidence interval. Results are often displayed in a forest plot, which shows each study's estimate and the combined effect. Meta-analysis needs studies that are similar in question and measures, and it requires statistical care, such as checking for heterogeneity and publication bias. If studies differ too much, a narrative synthesis is more honest than a forced pooled number.

Is it realistic for a student?

A full systematic review can take a team many months. For a student project:

• Time: screening thousands of records, extracting data and appraising studies take weeks to months.
• People: the method calls for at least two reviewers, which a solo student lacks.
• Skills: you need training in search design, appraisal and synthesis.
• Support: you need a supervisor and ideally a librarian experienced in reviews.
• Rules: your program must allow it as a dissertation.

Options that are often more realistic include:

• A narrow systematic review, with a tightly focused question and limited scope.
• A rapid or scoping review, with clearly stated limits.
• A systematic approach to a narrative review, documenting the search and selection clearly.
• A review with a second reviewer for a sample of records, such as 20 percent, to check agreement.

Discuss feasibility with your supervisor early, and be honest in your methodology about any shortcuts. Our guide to choosing a research topic offers a feasibility check that applies here as well. Graduate nursing students who want feedback on a review chapter can find it through coaching on a scholarly project draft.

Structure of a systematic review dissertation

The methods chapter replaces the usual research methodology, and the results chapter reports the synthesis.

Grey literature and publication bias

Studies that find an effect are more likely to be published than studies that find none, so a review that uses only journal articles can overestimate benefits. This is called publication bias. Systematic reviewers try to reduce it by searching trial registers, conference abstracts, theses and reports, and by contacting researchers. They may also use plots and statistical tests to check for signs that results are missing. For a student project, the minimum is to search at least one source of unpublished work and to discuss publication bias honestly as a limitation.

Common mistakes

• A question that is too broad.
• No protocol, or changing the methods after seeing results.
• Searching only one database.
• Not recording the search.
• A single reviewer with no checks.
• Ignoring study quality.
• Pooling studies that are too different.
• Not following PRISMA.
• Underestimating the time.

A quick checklist

• Is my question focused and structured?
• Do I have a protocol?
• Did I search multiple sources and record the strategy?
• Are my criteria explicit and applied consistently?
• Did a second person check screening or extraction?
• Did I appraise study quality?
• Does my synthesis fit the data?
• Have I reported according to PRISMA?
• Is the scope realistic for my project?

Final thoughts

A systematic review is valued for its transparency: another researcher should be able to follow your steps and reach the same studies. That makes it rigorous and labor intensive. Be honest about what you can do alone, choose a narrow question, plan carefully and report each step. A well-scoped, honestly reported review is far better than an over-ambitious one with hidden shortcuts.