Guide
How to do a literature review: a step-by-step guide
A literature review is not a summary of everything ever written on a topic. It is a structured answer to a specific question, built from papers you found deliberately, screened honestly, and organized well enough to write from. Here is the process that works — whether this is your first thesis chapter or your fiftieth review.
Step 1: Define your question before you search
The most common literature review mistake is searching first and deciding what you're looking for later. Write one specific question down. For clinical topics, the PICO framework helps: Population, Intervention, Comparator, Outcome. "Does yoga reduce inflammatory markers in adults with chronic stress?" is searchable. "Yoga and inflammation" is a rabbit hole.
Step 2: Choose your databases — plural
PubMed is where most reviews start, and where too many of them stop. PubMed is enormous, but it is not everything: regionally published clinical work — much of it in Portuguese, Spanish, Japanese, French, Thai, or Korean — often never appears there. A defensible search for a health topic usually includes:
- • PubMed and Europe PMC — the core biomedical indexes, overlapping but not identical.
- • Cochrane CENTRAL — controlled trials, the backbone of intervention reviews.
- • ClinicalTrials.gov — registered trials, including completed-but-unpublished results.
- • Regional indexes — LILACS (Latin America), J-STAGE (Japan), ThaiJO (Thailand), KoreaMed, HAL (France).
- • Open indexes — DOAJ, OpenAlex, and CrossRef catch open-access and cross-disciplinary work.
Searching these one at a time means a dozen tabs and a dozen query syntaxes. Symmathy runs one query across all twelve in parallel and translates non-English titles and abstracts inline — the original record is always one click away.
Step 3: Run the search — and write down what you did
Reproducibility separates a review from a ramble. For every search, record the exact query, the databases searched, and the date. If you later write up methods — or a supervisor asks where a paper came from — you need this trail. Symmathy logs sources queried, query terms, and timestamps automatically for every search.
Step 4: Screen and organize as you go
Read titles and abstracts first; most records will be obvious excludes. Save everything that might matter to one library — not a downloads folder — and attach a note to each paper the moment you read it: why it's relevant, what it measured, what worried you. Future you, writing the discussion section at midnight, will be grateful. When you need records elsewhere, export to RIS, BibTeX, or CSV for Zotero, Mendeley, EndNote, Covidence, or Rayyan.
Step 5: Summarize, then compare
For each included paper, extract the same fields: population, methods, endpoints, effect sizes, limitations, funding. A structured summary makes this fast — and for papers you only have as PDFs, a PDF scanner can pull those fields out with page references. Then lay the papers side by side: where they agree, where they diverge, and what the group collectively reports. That comparison is the raw material of your review.
Step 6: Write it
Organize by theme or chronology, not by paper. Every claim gets a citation. Say plainly what the evidence shows, what it doesn't, and where the gaps are. A literature review that only reports what supports its thesis isn't a review — it's an argument with references.
Frequently asked questions
How many databases should I search?
At minimum two, but more is better for health topics. PubMed alone misses regionally published and non-English studies. Adding Europe PMC, Cochrane, ClinicalTrials.gov, and regional indexes like LILACS, J-STAGE, and ThaiJO meaningfully changes what you find.
What's the difference between a literature review and a systematic review?
A narrative literature review surveys a topic with flexible methods. A systematic review follows a pre-registered protocol with documented, reproducible searches, explicit inclusion criteria, and formal risk-of-bias assessment. Both start with the same skill: searching well.
How should I organize the papers I find?
Save every relevant paper to a single library as you screen, attach notes to each record, and group by theme. Export to RIS, BibTeX, or CSV when you need the records in a reference manager or screening tool.
Should I include non-English studies?
Yes, when you can. The Cochrane Handbook warns that language restrictions can introduce bias. Inline translation lets you screen Japanese, Portuguese, Spanish, French, Thai, and Korean records in English while keeping the original one click away.