How to Write a Systematic Literature Review: PRISMA Guide

Home / Blog / How to Write a Systematic Literature Review: PRISMA Guide

What Is a Systematic Literature Review—and Why Does It Matter?

If you’ve written a literature review before, you might wonder: what makes this “systematic” approach different?

Here’s the honest truth: most literature reviews you’ve written are narrative reviews. They summarize what you’ve read, organized thematically. That’s perfectly fine for a dissertation chapter or a class assignment.

But a systematic literature review is something else entirely. It’s a research methodology designed to answer a precise research question by identifying, evaluating, and synthesizing all relevant studies—minimizing bias at every step.

The key difference? Reproducibility. If another researcher follows your protocol, they should arrive at the same studies you included. A narrative review doesn’t guarantee that.

A systematic literature review is a rigorous research methodology, not just a summary of existing papers. It follows a strict protocol to ensure transparency, reproducibility, and minimal bias.
PRISMA (Preferred Reporting Items for Systematic Reviews) provides a 27-item checklist for reporting your review—standard practice in health sciences, education, and social sciences.
The process has seven core steps: define your question, register a protocol, conduct a comprehensive search, screen studies, assess quality, extract data, and write using PRISMA guidelines.
Most students skip protocol registration and underreport search strategies—two errors that reviewers flag immediately.

Systematic Review vs. Narrative Review: The Real Differences

Feature	Systematic Review	Narrative (Traditional) Review
Search methodology	Defined protocol, multiple databases, fully documented	Flexible, based on author’s selection
Inclusion criteria	Explicit and pre-defined	Author’s judgment
Bias assessment	Formal quality appraisal tools required	Optional
Reproducibility	Another researcher can replicate your search	Not reproducible
PRISMA reporting	Required	Not required
When to use	Policy questions, meta-analysis, field expects methodology	Background context, thesis literature chapters

Think of it this way: a narrative review tells you what you read and thought about a topic. A systematic review tells you what the entire body of evidence says—and how you proved it.

This guide walks you through the entire process, step by step, using the PRISMA 2020 guidelines—the current standard. Whether you’re writing a thesis, a dissertation chapter, or preparing a manuscript for publication, this is the methodology your committee expects.

Step 1: Formulate a Focused Research Question

Your research question determines everything—the databases you search, the criteria you use, the studies you include. A vague question produces a messy review. A focused one produces clean, defensible results.

Use the PICO framework (or SPIDER for qualitative research):

Population: Who or what are you studying? (e.g., graduate students, clinical patients, organizational settings)
Intervention or Exposure: What treatment, phenomenon, or variable? (e.g., mindfulness intervention, AI tool usage, remote learning)
Comparison: What’s the alternative or control? (e.g., traditional instruction, no intervention, standard practice)
Outcome: What are you measuring? (e.g., stress reduction, detection accuracy, learning outcomes)

Example PICO question:

“In master’s level graduate students (P), does mindfulness-based stress reduction (I) compared to a waitlist control (C) reduce anxiety symptoms (O) as measured by the GAD-7 scale?”

Alternative framework: SPIDER (for qualitative or mixed-methods research)

Sample: Who is the population?
Phenomenon of Interest: What are you exploring?
Design: What study types?
Evaluation: What outcomes?
Rearch type: Qualitative, quantitative, or mixed?

💡 What I’d choose: If you’re unsure which framework to use, pick PICO. It’s the most widely recognized. Use SPIDER when your research is exploring experiences or perceptions rather than measuring effects.

Step 2: Develop and Register a Protocol

This is where most students make their biggest mistake—and it’s the one examiners hit hardest.

A protocol is your research blueprint. It documents your objectives, inclusion criteria, search strategy, data collection methods, and analysis plans before you begin. It prevents you from changing your criteria after seeing what the search turns up.

Register your protocol on:

PROSPERO (for health-related reviews)
Open Science Framework (OSF) Registries for other disciplines
Institute of Evidence-Based Healthcare if working in clinical fields

Even if registration isn’t required in your field, write a protocol anyway. It forces methodological clarity and protects you against accusations of bias.

Your Protocol Should Include:

Research question (PICO/SPIDER)
Inclusion and exclusion criteria (study designs, populations, dates, languages)
Database list and planned search strategy
Data extraction approach
Quality assessment tools
Synthesis plan (narrative or meta-analysis)

Step 3: Conduct a Comprehensive Search Strategy

A systematic search is not a casual Google Scholar query. It’s a documented, reproducible process designed to capture every relevant study.

3.1 Choose Your Databases

Run your search in at least three databases. For most disciplines:

Scopus + Web of Science (broad, interdisciplinary coverage)
Field-specific database: PubMed (medicine), ERIC (education), PsycINFO (psychology), CINAHL (health sciences)

3.2 Build Your Search String

Use Boolean operators, MeSH terms, keywords, and truncation:

Example PubMed search:

(("mindfulness"[MeSH Terms]) OR "mindfulness-based stress reduction"[Title/Abstract])
AND
("anxiety disorders"[MeSH Terms])
AND
("randomized controlled trial"[Publication Type])

3.3 Search Grey Literature

Don’t ignore:

Conference proceedings
Thesis repositories (ProQuest, DSpace)
Government reports
Citation chaining (following reference lists of key papers)

Record the number of results from each database before and after deduplication—this feeds directly into your PRISMA flow diagram.

3.4 Remove Duplicates

Export all results to a citation manager (EndNote, Zotero, or Mendeley) and remove duplicates. Document how many duplicates were removed.

Practical tip: Some citation managers auto-deduplicate as you import files. Capture the raw numbers before deduplication so your flow diagram stays accurate.

Step 4: Screen Studies for Eligibility

Screening happens in two distinct phases. Don’t skip either.

Phase 1: Title and Abstract Screening

Apply your inclusion/exclusion criteria to each result. When in doubt, keep the paper for full-text review. It’s easier to exclude later than to re-run your entire search.

Tools that help:

Covidence (automatically generates PRISMA flow diagram numbers)
Rayyan (AI-assisted screening for student reviewers)

Phase 2: Full-Text Screening

Retrieve and read the full-text articles that passed title/abstract screening. Confirm they meet every inclusion criterion.

Document every exclusion reason. You’ll report these in the PRISMA diagram:

Wrong study design
Wrong population
Wrong outcome
No comparison group
Incomplete data

Best Practice: Two Independent Reviewers

For rigorous reviews, two reviewers assess each paper independently. Disagreements are resolved by consensus or a third reviewer. Report inter-rater reliability (Cohen’s kappa) if you use this approach.

Step 5: Assess Study Quality Using Formal Appraisal Tools

This is where many student reviews fall short—and it’s the one examiners are most likely to probe.

Systematic reviews require formal quality assessment. You can’t just summarize papers and call it systematic. You need to evaluate whether the included studies are methodologically sound.

Quality Assessment Tools by Study Type:

Study Type	Recommended Tool	What It Assesses
Randomized controlled trials	Cochrane Risk of Bias 2 (RoB 2)	Randomization, blinding, outcome reporting
Observational studies	Newcastle-Ottawa Scale (NOS)	Comparability, exposure assessment
Qualitative studies	CASP (Critical Appraisal Skills Programme)	Rigor, credibility, transferability
Mixed methods	MMAT (Mixed Methods Appraisal Tool)	Qualitative, quantitative, and mixed quality

Present your findings transparently:

Traffic-light plots (green = low risk, yellow = some concerns, red = high risk)
Summary tables showing risk of bias per study domain

What I’d recommend: Even if your program doesn’t require formal quality assessment, include it. It strengthens your review significantly and shows you understand research methodology.

Step 6: Extract Data and Synthesize

Create a data extraction form before you begin. Apply it consistently to every included study.

What to Extract:

Author, year, country
Study design
Sample size and participant characteristics
Intervention/exposure details
Outcome measures and findings
Quality appraisal score

Two Synthesis Approaches:

1. Meta-analysis (quantitative synthesis)

Combine results statistically when studies are sufficiently homogeneous
Calculate pooled effect sizes with confidence intervals
Use I² statistic to assess heterogeneity
Requires ≥10 studies for publication bias assessment (funnel plots)

2. Narrative synthesis (qualitative approach)

Describe findings thematically when studies vary in methodology, populations, or outcomes
Note where study differences explain the variation
Compare findings across studies

Step 7: Write Using PRISMA 2020 Guidelines

The PRISMA 2020 checklist is a 27-item reporting guideline that ensures transparency and completeness. It covers every section of your manuscript:

PRISMA Checklist Sections:

1. Title & Abstract (Items 1–2)

Title must include “systematic review” (or “systematic literature review”)
Structured abstract with background, objectives, methods, results, conclusions

2. Introduction (Items 3–4)

Rationale: Why is this review needed?
Objectives: Explicit PICO/PECO research question

3. Methods (Items 5–16)

Protocol registration (Item 5)
Eligibility criteria (Item 6)
Information sources: Full database list with dates (Item 7)
Search strategy: Full search strings for every database (Item 8)
Selection process: Number of reviewers, disagreement resolution (Item 9)
Data extraction: Methods, forms, reviewer process (Items 10–11)
Risk of bias assessment: Tool and process (Item 12)
Effect measures and synthesis methods (Items 13–16)

4. Results (Items 17–22)

PRISMA flow diagram (Item 17): Mandatory
Study characteristics table (Item 18)
Risk of bias within studies (Item 19)
Individual study results (Item 20)
Synthesis results (Item 21)
Reporting bias assessment (Item 22)

5. Discussion (Items 23–25)

Summary of main findings (Item 23)
Limitations at review and study level (Item 24)
Conclusions with implications (Item 25)

6. Funding & Other (Items 26–27)

Funding sources (Item 26)
Conflicts of interest, data availability (Item 27)

The PRISMA Flow Diagram

This is your most important visual element. It maps the study selection process across four stages:

Identification: Records from databases and additional sources
Screening: Duplicates removed → titles/abstracts screened → excluded
Eligibility: Full-text assessed → excluded (with reasons)
Included: Studies in qualitative and quantitative synthesis

You can download the official PRISMA 2020 flow diagram template at prisma-statement.org.

Common Mistakes Students Make (and How to Avoid Them)

❌ Mistake 1: Treating PRISMA as the “Method”

Many students think a PRISMA flowchart replaces their methodology. It doesn’t. PRISMA is a reporting guideline—it tells you what to write, not how to conduct your research. You still need a protocol, search strategy, screening criteria, and quality assessment.

❌ Mistake 2: Vague Search Strategy Reporting

Writing “We searched PubMed, PsycINFO, and Google Scholar” is insufficient. PRISMA requires the full search strings for every database, ideally in an appendix.

❌ Mistake 3: Missing or Incorrect Flow Diagram Numbers

If your exclusion numbers don’t add up, reviewers will flag it immediately. Use Covidence or Rayyan to track decisions automatically. Double-check:

Identified → Duplicates removed → Screened → Excluded at screening → Full-text assessed → Excluded at full text → Included

❌ Mistake 4: Skipping Quality Assessment

Simply stating “two reviewers assessed quality” without naming the tool or presenting results is a major red flag.

❌ Mistake 5: Unclear Synthesis Method

Failing to report the effect measure, statistical model, heterogeneity metrics (I²), or software used makes your synthesis impossible to evaluate.

❌ Mistake 6: Omitting Protocol Registration

While some student reviews publish without a registered protocol, the expectation is rising. Registering demonstrates methodological rigor.

What We Recommend

If you’re writing a systematic literature review for a thesis or dissertation, here’s my practical recommendation:

Start with a protocol. Don’t skip it. Even an unregistered protocol forces clarity.
Use PRISMA 2020. It’s the current standard—journals and examiners expect it.
Document every number. Your flow diagram is a primary evidence piece. If the numbers don’t add up, your review loses credibility.
Include quality assessment. Even if your field doesn’t require formal appraisal tools, include a narrative quality discussion.
Be transparent about limitations. Don’t hide gaps in your search. Acknowledge them. That’s what rigor looks like.

The process takes more effort than a traditional literature review. But the payoff is significant: your review becomes a defensible piece of research rather than a summary of what you’ve read. That distinction matters to examiners, to reviewers, and to your future citations.

Related Guides

Frequently Asked Questions

Can I use a systematic literature review for my dissertation?

Yes—if your research question requires comprehensive evidence synthesis and your field expects it (medicine, public health, education, social policy are the most common). Check with your committee whether a systematic approach is expected or optional.

How long does a systematic review take?

Typically 6–12 months, depending on the scope. The search and screening phases are the most time-consuming. Protocol registration can take 2–4 weeks.

What if I can’t find enough studies?

That’s actually a valid finding. Systematic reviews report what they find—not what they wish they found. If only a few studies match your criteria, discuss why (e.g., emerging field, narrow research question) and proceed with a narrative synthesis.

Can I use ChatGPT or AI tools to write my systematic review?

You can use AI for organizing notes, drafting search strings, or summarizing individual studies—but you must conduct the systematic process yourself. AI cannot ensure proper screening, quality assessment, or bias evaluation. Using AI for the full review is academically dishonest and methodologically invalid.

What if my field doesn’t require PRISMA?

Even without formal requirements, following PRISMA strengthens your review significantly. It demonstrates methodological sophistication and makes your work citable.

Need Help With Your Systematic Literature Review?

Writing a systematic review is one of the most methodologically demanding tasks in graduate research. If you need support—whether it’s developing your search strategy, structuring your PRISMA compliance, or synthesizing complex findings—our team of qualified researchers can help.

Get a custom quote for your dissertation chapter or review