A systematic review is a literature review that answers one focused question from all the relevant studies. You find, screen, appraise and combine those studies using methods you plan in advance and report so others could repeat them. The point is to minimise bias in the answer.
A full systematic review normally starts with a written protocol and involves more than one reviewer, so that no single person decides which studies count. It often takes more than a year. If your undergraduate or master’s dissertation asks for a systematic review, you’ll usually do a systematised review instead: the same steps on a smaller scale, usually carried out by one person.
What makes a review ‘systematic’?
A review is systematic when every decision about which studies go in, and what you conclude from them, follows a method you set out before you started. If you choose papers because you know them or because they support your argument, it’s a narrative review, however carefully you wrote it.
In practice, a systematic review:
- asks one specific, answerable question;
- follows a protocol, sometimes registered in advance;
- searches in a way that is thorough and could be repeated;
- applies explicit inclusion and exclusion criteria;
- appraises, extracts and combines the data in a systematic way;
- reports every step, so a reader can check it or repeat it.
The aim is to be comprehensive within practical limits, to minimise bias and to be transparent. That’s why this approach is the one to use when reliable evidence is needed to inform policy or practice. Bias can still creep in at any stage if your methods aren’t planned, applied consistently and reported. It can come from how you frame the question, which databases you search or how you read the results.
What this means for you: if your supervisor asks for a ‘systematic’ review, your method section will be read closely. A reader should be able to repeat your search and screening from it.
Systematic review vs literature review, meta-analysis and scoping review
These terms describe different jobs. A systematic review answers a focused question from all the relevant evidence. A narrative review (often just called a ‘literature review’) gives an overview chosen by the author’s judgement. A meta-analysis isn’t really a separate kind of review: it’s a statistical step that pools the results of studies inside a systematic review.
| Review type or method | What it does | How it’s done | Choose it when |
|---|---|---|---|
| Narrative (literature) review | Summarises and interprets the literature on a topic | Author chooses what to include; no protocol needed | You need background and context, such as the literature review chapter of a dissertation built on your own data |
| Systematic review | Answers one focused question from all the relevant studies | Protocol, comprehensive search, screening against criteria, appraisal, synthesis, full reporting | You need a reliable answer to a specific question and have the time and a second reviewer |
| Meta-analysis | Combines the numerical results of several studies into one estimate | Statistical pooling of effect sizes, carried out within a systematic review | The included studies measure the same outcome in a similar enough way to combine |
| Systematised review | Uses the steps of a systematic review without being fully comprehensive | Systematic-style search and appraisal, usually by a single reviewer | You’re working alone on a dissertation with a fixed deadline |
| Scoping review | Maps what research exists on a broad topic | Broad search; no formal quality appraisal; results often charted and grouped by theme | You want to find gaps, clarify concepts or see what kinds of evidence exist |
| Rapid review | A shortened systematic review | Limited search and simplified extraction and appraisal | A decision needs evidence quickly |
| Umbrella review | Reviews existing systematic reviews | Finds and summarises published reviews | Several published reviews already cover your question |
Source: UCL library guide to review types, with Southampton and Reading library guides and Munn et al. (2018).
Two differences catch people out. The first is the team. A full review usually involves at least two people, and professional reviews often more, while one person usually writes a narrative review.
The second is the question. A scoping review suits a broad question (‘what research has been done on X?’), while a systematic review suits a narrow one (‘does X work for Y?’). A scoping review can also come first, to check there’s enough evidence for a full review and to firm up its question and criteria.
Weighing up a narrative review instead? Our guide to the difference between an annotated bibliography and a literature review explains how a literature review is put together.
Can you do a systematic review for a dissertation?
Usually you’ll do a systematised review rather than a full systematic review. It follows the same principles, often for a research methods module or dissertation, but is less rigorous and less in depth because you’re working alone and to a deadline.
That doesn’t make it an easy option. You still need a focused question, a search strategy you can report, clear inclusion criteria and a record of every decision. What changes is the scale: you may search fewer databases and skip unpublished (grey) literature, and you’ll screen and appraise studies on your own. Note each of these as a limitation.
Terminology is used loosely, so your lecturer may still call your project a systematic review. Check your module or assignment brief for what’s expected. Then describe what you actually did in your method section: for example, say that one reviewer screened all the records, and list the databases you searched. Being accurate about the limits is better than claiming a standard you couldn’t meet.
Which route fits you?
- Dissertation built on your own data (interviews, experiments, a survey): your literature review chapter is normally a narrative review. You can still borrow systematic habits, such as recording your search terms.
- Dissertation that is itself a review (for example in health or business): plan a systematised review, unless your brief says otherwise.
- PhD: a full systematic review is realistic if someone else can screen with you; PROSPERO strongly recommends more than one reviewer. If it’s going to be published, our guide to publishing your first paper covers the next step. Our PhD literature review guide explains what examiners expect from the literature review in a thesis.
How to do a systematic review: the steps
The steps are the same whether you’re doing a full review or a systematised one. What changes is mainly the scale, the depth and the number of people.
- Set a focused question. For questions about interventions or treatments, a common framework is PICO: population, intervention, comparison (or comparator) and outcome. A broad question can leave you with more studies than you can screen.
- Write a protocol. Set out your question, search plan, inclusion and exclusion criteria, and how you’ll appraise and combine the studies, before you start. This stops you changing the rules once you see the results.
- Build and run your search. Search more than one database, using the full search features of each. Fuller reviews also search grey literature, such as theses, reports and trial registers. Save your exact search strings and the date you ran each one.
- Screen the results. Check titles and abstracts against your criteria, then read the full text of everything that passes. A full review is usually run by at least two people, to avoid bias.
- Appraise the studies. Judge each included study’s quality or risk of bias with a recognised critical appraisal tool. Our guide to reading a journal article is a good warm-up for this.
- Extract the data. Record the same details from every study: its methods, participants and results.
- Combine the findings. Combine numerical results with a meta-analysis if the studies are similar enough. If they aren’t, write a narrative synthesis. Qualitative studies can be combined with methods similar to thematic analysis.
- Report it. Write it up using the PRISMA 2020 checklist, with a flow diagram showing how many records you found and how many studies you included.
Worked example: from question to search string
Say you’re a nursing student asking whether mindfulness programmes reduce anxiety in university students, compared with no programme. This is an illustration, not a real search.
| PICO element | Your concept | Search terms (synonyms joined with OR) |
|---|---|---|
| Population | University students | student* OR undergraduate* OR "higher education" |
| Intervention | Mindfulness programmes | mindfulness OR meditation |
| Comparison | No programme | (left out of this search; checked at screening) |
| Outcome | Anxiety | anxiety OR anxious |
You join the concepts with the Boolean operator AND:
(student* OR undergraduate* OR "higher education") AND (mindfulness OR meditation) AND (anxiety OR anxious)
The asterisk finds word endings (student, students), and double quotation marks keep a phrase’s words together. Truncation can also bring in words you don’t want (‘student*’ finds ‘studentship’ too), so check a sample of your results. Each database has its own syntax and subject headings, so adapt the string for each one and record every version. Before you run it, ask your university’s subject librarian to check your strategy.
What is PRISMA?
PRISMA 2020 is the current version of PRISMA, a reporting guideline for systematic reviews. It tells you what to include when you write the review up, so readers can see exactly what you did. It replaced the 2009 version, which shouldn’t be used any more.
The PRISMA 2020 statement has four parts:
- a 27-item checklist;
- an expanded checklist, with the reporting recommendations for each item;
- an abstract checklist;
- flow diagrams for new reviews and for updated ones.
The flow diagram is the part you’ll draw yourself. It shows the number of records your search found, how many you kept and excluded at each stage, and how many studies you finally included. There’s a version for reviews that searched only databases and registers, and it’s often the right one for an undergraduate or master’s project.
Common mistake: treating PRISMA as a quality checklist. The statement itself says it is for reporting, not for judging how well a review was done; separate tools exist for that. Protocols have their own guideline, PRISMA-P. The PRISMA 2020 statement explains each item and points to fillable checklist and flow diagram templates on the PRISMA website.
Do you need to register your review on PROSPERO?
Usually not for a dissertation review, because PROSPERO only accepts some reviews. It’s an international register of systematic reviews with a health-related outcome, including reviews in areas such as education, social care, crime and justice where there’s a health outcome. Registration keeps a dated, permanent record of your plan, which makes it transparent and helps others avoid duplicating your review.
PROSPERO’s eligibility questions are there to stop ineligible projects before you fill in the form. They rule out:
- scoping, literature or mapping reviews;
- reviews without at least one outcome directly related to human health, unless it’s a methodology review with a clear link to human health;
- mini or partial reviews done for a training course or classwork, including a review a whole class does together, or any other project that is less than a full systematic review, so a systematised review won’t qualify;
- reviews whose results won’t be made available to others, such as master’s dissertations that it’s known won’t be shared.
If yours is a training or classwork review, you can still fill in the registration form and save it as a PDF to show your supervisor. Don’t submit it: the saved copy isn’t published and doesn’t get a registration number.
PROSPERO also strongly recommends having more than one person on the review team. For now it still accepts single authors, but it says it may require more than one in future. Check the current rules in PROSPERO’s eligibility questions before you start.
If your review has no outcome related to human health, PROSPERO doesn’t apply. Your supervisor can tell you whether your field expects a protocol to be published somewhere else.
How long does a systematic review take?
A full systematic review takes many months. A 2017 study looked at 195 completed reviews of medical treatments registered on PROSPERO. They took a mean of 67.3 weeks, more than a year, from their registered start date to publication, and had a mean of five authors. Reviews that reported funding took longer and had larger teams.
The data in that study run to 2014, but we found no newer estimate of this size. For comparison, the average Cochrane review takes about 18 months.
The same study found that, on average, only about 3% of the records a search turned up made it into the review, and in half the reviews it was under 1%. So expect to reject most of the records you screen.
What does that mean for a dissertation? A systematised review has to fit into weeks, not years. Keep your question narrow, limit your search to the databases that matter most for your topic, and pilot your search early. If it returns far more records than you can screen, narrow the question before you start rather than cutting corners halfway through.
Next steps
Before you start a systematic review, agree with your supervisor which kind of review you’re doing and what your brief expects. Then write your question in PICO form (or a framework that suits your field) and draft your protocol before you search.
If your review will form a dissertation, our guide to dissertation structure shows the standard order of chapters, including where the method, results and discussion go. For a wider view of research designs, including the secondary research where reviews belong, see our overview of types of research.
Sources
- UCL Library: What are systematic reviews?
- UCL Library: Types of review
- University of Southampton Library: About systematic reviews
- University of Reading Library: Doing a systematic review
- York St John University Library: Systematic and systematised reviews
- University of Lincoln Library: Writing a systematised review in business
- Cochrane: What are systematic reviews?
- Cochrane Handbook for Systematic Reviews of Interventions: Chapter 3, Defining the criteria for including studies
- Borah R et al. (2017) Analysis of the time and workers needed to conduct systematic reviews of medical interventions using data from the PROSPERO registry. BMJ Open 7:e012545
- Munn Z et al. (2018) Systematic review or scoping review? Guidance for authors when choosing between a systematic or scoping review approach. BMC Medical Research Methodology 18:143
