If you have ever been told to “do a systematic review” and felt your stomach drop, you are not alone. A systematic literature review sounds like one of the biggest, scariest assignments in research — and honestly, it is a big job. But it is not a mysterious one. It is simply a literature review done in a careful, transparent, and repeatable way, so that another researcher could follow your steps and reach the same set of studies.
In this guide, you will learn how to do a systematic literature review step by step, following the PRISMA 2020 guidelines. PRISMA stands for Preferred Reporting Items for Systematic Reviews and Meta-Analyses. It is a widely used reporting standard that tells you what to report and how to show your search and screening process clearly. By the end, you will have a realistic plan you can actually follow, plus a worked example and the mistakes to avoid.
What Is a Systematic Literature Review?
A traditional narrative literature review summarises what the author found interesting or important. A systematic review is different in three ways. First, it starts with a clear, focused research question. Second, it searches for evidence using a planned, documented method rather than whatever turned up first. Third, it screens, appraises, and synthesises every study that meets pre-set criteria, and it reports the whole process transparently — including the studies it excluded and why.
That transparency is the whole point. A reader should be able to see exactly how you searched, what you included, what you left out, and how you judged study quality. That is also why systematic reviews are trusted: they reduce the risk that the author cherry-picked only the studies that agreed with them.
A systematic review does not always include a meta-analysis. A meta-analysis is a statistical pooling of results across studies. You can do a systematic review with a narrative synthesis instead, where you describe and compare findings in words and tables. Both are valid — the choice depends on your question and your data.
Before You Start: Three Decisions That Save You Weeks
Most students rush straight to searching. Resist that urge. These three decisions will save you far more time than they cost.
First, narrow your question. “What do we know about anxiety?” is not a review question; it is a lifetime of reading. A good systematic review question names a population, an intervention or exposure, a comparison, and an outcome — the PICO framework. For example: “In university students (population), does mindfulness-based intervention (intervention), compared with a waitlist control (comparison), reduce symptoms of anxiety (outcome)?” If your topic is qualitative, you can use a framework like SPIDER instead, which covers sample, phenomenon of interest, design, evaluation, and research type. Either way, a tight question makes every later step easier.
Second, write a protocol before you search. A protocol is a short plan that states your question, your inclusion and exclusion criteria, the databases you will search, and how you will screen and extract data. Many reviewers register their protocol on PROSPERO, a free international register of systematic review protocols, where registration is appropriate. Even if you do not register, writing the protocol down matters: it stops you from quietly shifting your criteria halfway through because the results are not what you hoped for.
Third, decide your inclusion and exclusion criteria in advance. Criteria usually cover study design (for example, randomised controlled trials only), population, language, publication date range, and publication type. Be realistic: if you include every study design ever published, your screening pile will be enormous. If you restrict too tightly, you may end up with almost nothing. A quick scoping search — a few test searches in one database — helps you judge whether your criteria are workable before you commit.
Step 1: Write Your Research Question and Protocol
Turn your topic into a structured question using PICO or SPIDER, as described above. Then draft your protocol: question, criteria, databases, search terms, screening process, data extraction plan, and how you will assess study quality. Share it with your supervisor and get feedback before you move on. Fixing a vague question now takes an afternoon; fixing it after screening two thousand records takes weeks.
Step 2: Build a Search Strategy
List the key concepts in your question — usually the population, the intervention, and the outcome — and brainstorm synonyms and related terms for each. For our mindfulness example, the intervention concept might include mindfulness, meditation, mindfulness-based stress reduction, and MBSR. Combine terms within a concept with OR, and combine concepts with AND.
Search at least two or three major databases relevant to your field, such as PubMed, PsycINFO, Scopus, or Web of Science, plus a check of reference lists of included studies. Use database subject headings where available, such as MeSH terms in PubMed, alongside free-text keywords. Record your exact search strings, the date of each search, and the number of records retrieved. You will need all of this for your PRISMA flow diagram and your methods section, and “I searched Google Scholar” is not a reproducible strategy on its own.
Step 3: Screen Studies in Two Stages
Export all records into a reference manager and remove duplicates. Then screen in two stages. First, screen titles and abstracts against your criteria, keeping anything that might be relevant — when in doubt, keep it in. Second, read the full texts of the remaining studies and apply your criteria strictly.
Ideally, two reviewers screen independently and resolve disagreements by discussion. If you are working alone, as many students do, be honest about that in your methods section, and consider asking a colleague or supervisor to check a sample of your decisions. Keep a record of every full text you excluded and the reason — this is required for the PRISMA flow diagram.
Step 4: Extract the Data
Design a simple extraction table before you start: study authors and year, country, sample size and population, intervention details, outcome measures, and key results. Extract the same information from every study, in the same format. This discipline is what makes synthesis possible later. If a study reports results unclearly, note that rather than guessing — missing information is itself a finding about reporting quality.
Step 5: Assess Risk of Bias and Study Quality
A systematic review does not treat all studies as equal. Use an established appraisal tool suited to your study designs. For randomised trials, the Cochrane Risk of Bias tool is the standard choice; for observational studies, tools such as the Newcastle-Ottawa Scale are commonly used. Record your judgements study by study. This step is not about criticising other researchers — it is about understanding how much confidence your conclusions deserve.
Step 6: Synthesise the Findings
If your included studies are similar enough, you may pool their results statistically in a meta-analysis. More often in student reviews, you will synthesise narratively: group studies by population, intervention type, or outcome, describe patterns, and note where findings agree and where they conflict. Tables help enormously here — a summary table of included studies lets readers see the whole evidence base at a glance. Be careful not to overclaim: if the studies are small, varied, or at high risk of bias, say so plainly.
Step 7: Report Using PRISMA 2020
PRISMA 2020 provides a 27-item checklist for reporting a systematic review and a flow diagram showing how many records you identified, screened, excluded, and included at each stage. Work through the checklist as you write, and build your flow diagram from the numbers you recorded during screening. Your methods section should let a reader repeat your search; your results section should report what you found, including the risk-of-bias judgements; and your discussion should state the limitations of both the evidence and your review itself.
Step 8: Write It Up Clearly
Structure your review like a research paper: introduction and rationale, methods, results, discussion, and conclusion. In the introduction, explain why this review was needed and what gap it fills. In the discussion, answer your question directly, compare your findings with earlier reviews, and state what future research should do differently. End with a clear, honest conclusion — not a grand claim your evidence cannot support.
A Worked Example
Let us walk through a small example. Suppose your question is: “In university students, does a mindfulness-based intervention reduce anxiety compared with a waitlist control?” Your PICO is clear. Your protocol says you will include randomised controlled trials of mindfulness interventions in university students, published in English, measuring anxiety with a validated scale.
Your search concepts are university students, mindfulness interventions, and anxiety. You search three databases and retrieve a large set of records, remove duplicates, and screen titles and abstracts. Perhaps a few hundred records become a few dozen full texts, and after applying your criteria, a handful of trials remain. You extract sample sizes, intervention lengths, anxiety measures, and results into one table. You appraise each trial’s risk of bias. In synthesis, you notice that longer interventions tend to show larger effects, but two small trials show no effect — and you report both facts, not just the pleasing one. Your PRISMA flow diagram shows every number along the way, so your reader can see exactly how you got from thousands of records to your final set. That is a systematic review.
Common Mistakes to Avoid
The first mistake is starting without a question or protocol, then drifting as you read. The second is searching only one database or only free web sources, which misses relevant studies and biases your conclusions. The third is screening alone without any checking and without recording exclusion reasons — you will regret this when you try to draw your flow diagram. The fourth is treating all included studies as equally trustworthy and skipping quality appraisal. And the fifth is overclaiming in the conclusion: a review of five small, varied studies cannot prove that an intervention works for everyone.
How Long Does a Systematic Review Take?
Be honest with yourself about time. A full systematic review is a major project that typically takes months, not days, especially the screening stage. If you have a tight deadline, discuss scope with your supervisor early: a narrower question, a smaller date range, or a review of reviews may be more realistic. A smaller review done rigorously is worth far more than a huge one done carelessly.
Frequently Asked Questions
How is a systematic review different from a normal literature review?
A normal literature review summarises selected literature, often without reporting how studies were chosen. A systematic review follows a pre-planned, documented method, screens all studies against set criteria, appraises their quality, and reports the full process so it can be checked and repeated.
Do I need to do a meta-analysis?
No. A meta-analysis is a statistical technique for pooling results, and it is only appropriate when studies are similar enough to combine. Many excellent systematic reviews use narrative synthesis instead, describing and comparing findings in words and tables.
What is the PRISMA flow diagram?
It is a diagram that shows the flow of studies through your review: records identified from each database, duplicates removed, records screened, full texts assessed, studies excluded with reasons, and studies finally included. PRISMA 2020 provides the standard template, and you fill it with your own recorded numbers.
Can I do a systematic review on my own?
Yes, many students do, but be transparent about it. Ideally two reviewers screen independently. If you work alone, say so in your methods, consider having someone check a sample of your screening decisions, and be extra careful with your records.
Do I have to register my protocol?
Registration on PROSPERO is good practice and expected for many health-related reviews, but it is not possible or required for every topic. What matters most is that you write a protocol before you search and follow it, so your decisions are planned rather than made up as you go.
Final Thoughts
A systematic literature review is demanding, but it is a skill that will serve you throughout your research career. Start with a tight question, write your protocol, search carefully, screen honestly, and report transparently with PRISMA 2020. Do that, and your review will stand up to scrutiny — and you will actually understand your topic far better than when you started. Good luck, and take it one step at a time.