Research Workflow
Master the systematic literature review process. Complete guide covering protocol development, search strategy, screening, data extraction, and synthesis.
A systematic literature review is the most rigorous form of research synthesis.
It's also the most misunderstood.
Many people confuse it with:
A true systematic literature review is different.
It's:
This comprehensive guide walks through every step of conducting a systematic literature review.
Use a systematic review when you need to:
Examples:
Don't use systematic review if you:
| Type | Purpose | Time | Rigor | When to Use |
|---|---|---|---|---|
| Narrative | Overview, background | Fast | Low | Exploratory |
| Scoping | Scope of a topic | Medium | Medium | Exploratory |
| Systematic | Answer specific question | Slow | High | Evidence synthesis |
| Meta-analysis | Quantitative synthesis | Very slow | Very high | When data pooling appropriate |
Develop protocol, define research question, create search strategy
Search all relevant databases, document search terms
Screen titles/abstracts, then full texts
Extract data from included studies in standardized format
Analyze and synthesize data (qualitative or quantitative)
Write up findings, create PRISMA diagram
Use the PICO framework:
P = Population: Who/what is the study about?
I = Intervention: What treatment/factor are you examining?
C = Comparison: What's the comparison?
O = Outcome: What results matter?
Full question: "In adults with generalized anxiety disorder, how effective is cognitive-behavioral therapy compared to waitlist control for reducing anxiety symptoms?"
Inclusion criteria (must meet ALL):
Exclusion criteria (any exclude):
Search every relevant database:
Create search terms using Boolean operators:
(anxiety) AND (cognitive-behavioral OR CBT)
AND (therapy OR treatment)
AND (RCT OR randomized OR trial)
Search each database with identical search terms.
Document every search (database, date, terms, number of results).
Create a standardized form for extracting data from each included study:
Study: [Author, Year]
Design: [RCT/Quasi-experimental]
Population: [N, demographics]
Intervention: [Description]
Comparison: [Description]
Outcome: [Measurement, result, effect size]
Bias Risk: [High/Medium/Low]
Use same form for every study (standardization = reproducibility).
Register your protocol (publicly):
This prevents changing your question after you see results.
For each database:
Example searching log:
Date: 2025-01-15
Database: PubMed
Search: (anxiety) AND (CBT OR cognitive-behavioral) AND (trial)
Results: 342
Exported: 342 to reference manager
Search for papers manually in key journals:
Look at last 5 years of published articles.
This catches papers databases might miss.
Look at references in included studies.
Check citations of included studies ("cited by" in Google Scholar).
This finds papers that databases might miss.
If you can't find the full paper:
Email the author. Ask for a copy.
Most authors respond.
After all searching:
Example:
Total records from databases: 1,500
After deduplication: 1,200
Ready for screening: 1,200
Read title and abstract of all 1,200 papers.
Apply inclusion criteria strictly (if unclear, include).
Remove papers clearly outside scope.
Before: 1,200 papers After: 200 papers (approximate)
Have TWO reviewers independently screen each paper.
If reviewers disagree, they discuss and reach consensus.
This reduces individual bias.
Documentation:
Reviewer 1: Include
Reviewer 2: Exclude
Discussion: Both reviewers agree: Exclude (not about anxiety)
Retrieve full text of remaining papers (~200).
Read full text carefully.
Apply inclusion/exclusion criteria strictly.
Document reason for exclusion if applicable.
Before: 200 papers After: 45 papers (your included studies)
Use your pre-designed extraction form for each study.
Extract:
Two reviewers independently extract data from each study.
Compare extractions. Resolve discrepancies.
Document the process.
For each study, assess risk of bias:
Use a tool like Cochrane Risk of Bias assessment.
Rate each as: High risk / Low risk / Unclear
If studies are too heterogeneous to combine:
Summarize findings narratively:
By outcome:
By study type:
By population:
If studies measure similar outcomes with sufficient detail:
Pool results statistically:
Example forest plot interpretation:
Follow PRISMA (Preferred Reporting Items for Systematic Reviews) checklist.
27-item checklist ensuring you report:
Visual representation of review process:
Records identified (n=1,500)
↓
Records screened (n=1,200)
↓
Excluded (n=1,000)
↓
Full texts assessed (n=200)
↓
Excluded (n=155)
↓
Studies included (n=45)
Create appendices documenting:
This allows reproducibility.
Zotero (free) or Mendeley (free tier available)
Covidence (paid) or DistillerSR (paid)
Excel or Google Sheets
Revman (free) or Comprehensive Meta-Analysis (paid)
| Phase | Duration | Output |
|---|---|---|
| Planning | 2–4 weeks | Protocol, registered |
| Searching | 1–2 weeks | Comprehensive list of papers |
| Screening | 2–4 weeks | 40–50 included studies |
| Extraction | 3–6 weeks | Standardized data from all studies |
| Synthesis | 2–4 weeks | Results, forest plots, conclusions |
| Reporting | 1–2 weeks | Final manuscript |
| Total | 11–22 weeks | Published review |
(With 1–2 reviewers, part-time)
You see papers and think "Oh, I should also look at anxiety + depression."
You change your criteria mid-review.
This introduces bias.
Fix: Lock your criteria before searching. Changes allowed only if documented.
One person screens, extracts, and assesses bias.
Increases subjectivity and error.
Fix: Always dual review (two independent reviewers).
You only search one database.
You miss papers.
Your review is incomplete.
Fix: Search ≥3 databases. Do hand searching and reference tracking.
You include all studies equally.
Low-quality studies skew your results.
Fix: Assess risk of bias for every study. Report it separately.
Later, nobody remembers why papers were excluded or how decisions were made.
Fix: Document EVERYTHING. Every decision. Every discussion.
✅ Rigorously synthesize evidence from multiple studies
✅ Provide reproducible, transparent methodology
✅ Reduce bias through standardized processes
✅ Guide clinical/policy decisions with high confidence
✅ Identify research gaps
❌ Guarantee "perfect" truth (they're limited by included studies)
❌ Happen quickly (takes 3–6 months minimum)
❌ Replace critical thinking (you still interpret results)
❌ Prevent all bias (they reduce it, not eliminate)
A systematic literature review is the gold standard for research synthesis.
Process:
Why it matters:
Timeline: 3–6 months for complete review
Start this week if you're planning a systematic review:
For more on research, see Build a Research Workflow. For citation management, check Citation Best Practices.
Be systematic. Be transparent. Be reproducible.
Conduct reviews that advance knowledge.
More WebSnips articles that pair well with this topic.
Understand when to use primary vs secondary research in digital workflows. Practical examples, source evaluation, and tools for each research type.
The exact Chrome extension stack for serious research workflows. From citation managers to web clippers to academic search tools.
Use web clipping to capture, organize, and cite research sources. Ideal for students, academics, and researchers managing large reference libraries.