Literature Reviews
How to conduct a literature review
A good literature review answers a defined question with a search that someone else could repeat. This guide takes you from the question to the written synthesis.
- Author
- By ToProt editorial team
- Published
- Updated
- Reading time
- 6 min read
Reviewed by ToProt editorial team on
What a literature review is, and the kinds there are
A literature review collects, assesses and summarises the existing research on a question. It shows what is known, where the evidence conflicts and what remains unanswered, and it is the basis for the background section of a protocol or thesis.
Reviews differ in how formal they are [1]:
| Type | Purpose | Method |
|---|---|---|
| Narrative review | Give an overview or context | No fixed method; selection of studies is the author's judgement, so it is open to bias |
| Scoping review | Map what research exists on a broad topic | Systematic search and selection, but usually no assessment of study quality [3] |
| Systematic review | Answer a focused question with the least bias | A written protocol, a comprehensive reproducible search, selection and appraisal by two reviewers, and a structured synthesis [2] |
| Meta-analysis | Combine numerical results across studies | Usually part of a systematic review, with statistical pooling |
Choose the type from your purpose and resources. A thesis background often needs a structured narrative or scoping review. A systematic review is a project in itself and needs a team, usually including a librarian and a statistician.
Step 1: Define the question
Write the question before you search. For questions about interventions or exposures, structure it as population, intervention or exposure, comparison and outcome (PICO). For broader scoping questions use population, concept and context. A question that is too broad returns thousands of irrelevant records; one that is too narrow finds nothing.
Write down your eligibility criteria: the study designs, populations, interventions, outcomes, languages and dates that you will include and exclude, and why. Writing them first stops you from choosing studies because of what they found.
Step 2: Plan the search
Choose several sources, because no single database covers everything.
- Bibliographic databases: PubMed or MEDLINE for biomedicine, plus Embase, Scopus, Web of Science or Cochrane CENTRAL where you have access. Subject databases add coverage, for example CINAHL for nursing.
- Other sources: the reference lists of relevant studies (backward citation searching), the studies that cite them (forward searching), trial registries and, for systematic reviews, grey literature such as reports, theses and conference abstracts.
Build the search from your question's concepts. Within a concept, join synonyms with OR; between concepts, join with AND. Combine controlled vocabulary (indexing terms such as MeSH in PubMed) with free-text words in the title and abstract, and use a wildcard () to catch word endings, such as adheren for adherence and adherent.
Test the search against a few studies you already know are relevant. If it misses them, add terms. Save the exact search string for every database and the date it was run. Reporting the full search lets others repeat it [4].
Step 3: Collect, de-duplicate and screen
Export all records to a reference manager and remove duplicates, because the same paper appears in several databases. Automatic de-duplication misses some, so check the remainder [5]. Then screen in two stages:
- Title and abstract: remove records that clearly do not meet the criteria.
- Full text: read the remaining papers and record the reason for each exclusion.
For a systematic review two reviewers screen independently and resolve disagreements by discussion or a third reviewer. Keep count at every stage and show the flow in a diagram [2].
Step 4: Extract data and appraise quality
Design a data extraction form before you start and test it on a few studies. Typical columns are: study (author, year, country), design, population and sample size, intervention and comparison, outcome measures, main results, and funding. Extract in a table so studies can be compared.
Then appraise how trustworthy each study is. Use a tool that matches the design: for example the RoB 2 tool for randomised trials [6], and checklists such as the Newcastle-Ottawa Scale, the CASP checklists or the Joanna Briggs Institute checklists for other designs. A study at high risk of bias should carry less weight in your conclusions, and a scoping review may choose not to appraise.
Step 5: Synthesise the evidence
Synthesis is more than listing studies. Group them by theme, population, intervention or outcome, and compare their findings and their quality. Say where they agree and where they conflict, and look for reasons, such as different populations, doses or measurement methods.
- A narrative synthesis explains patterns in words and tables.
- A meta-analysis pools numerical results from similar studies into one estimate, and shows how much the studies vary (heterogeneity). It is appropriate only when the studies are similar enough to combine, and it needs a statistician's help.
State how confident you are in each conclusion and what would change it.
Step 6: Write and report
Write the review with a clear structure: introduction and question, methods (search, criteria, screening, appraisal, synthesis), results (the flow of studies, the study characteristics and the findings) and discussion (what it means, limitations, gaps for future research). Systematic reviews are reported with the PRISMA 2020 guideline and its search extension [2, 4], and scoping reviews with PRISMA-ScR [3]. Cite your sources consistently with a reference manager so the bibliography is built from the same records you screened.
Registering a systematic review protocol in advance, for example in PROSPERO, shows that the question and methods were set before the results were known.
Limitations to state
Every review has limits, and you should say what yours are.
- Publication bias: studies with positive or striking results are more likely to be published, so the published literature can overstate effects.
- Language and database bias: restricting to one language or one database misses relevant work.
- Search limits: a search is a snapshot taken on a date, and new studies appear after it.
- Reviewer judgement: screening and appraisal involve judgement, which is why two independent reviewers are used for systematic reviews.
- Quality of the evidence: a review of weak studies produces weak conclusions, however carefully the review itself is done.
Common mistakes
- Starting to search without a defined question and eligibility criteria.
- Searching only one database, or only for words in English.
- Not recording the search string and date, so that it cannot be repeated.
- Choosing studies that support a view and ignoring those that do not.
- Listing studies one by one instead of synthesising them.
- Treating all studies as equally reliable.
- Not de-duplicating records, so that one study is counted twice.
Use it in ToProt
The Literature Explorer searches OpenAlex, PubMed and Crossref in one place and lets you save records to a study. The reference manager in ToProt holds the saved records, finds duplicates and exports to RIS, BibTeX, EndNote XML and CSV for a review tool. The citation formatter formats references in common styles, the research question builder helps you structure the question, and the writing editor builds a bibliography from your study's references.
References
- 1.Grant MJ, Booth A. A typology of reviews: an analysis of 14 review types and associated methodologies. Health Info Libr J. 2009;26(2):91-108. doi:10.1111/j.1471-1842.2009.00848.x
- 2.Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. doi:10.1136/bmj.n71
- 3.Tricco AC, Lillie E, Zarin W, et al. PRISMA extension for scoping reviews (PRISMA-ScR): checklist and explanation. Ann Intern Med. 2018;169(7):467-473. doi:10.7326/M18-0850
- 4.Rethlefsen ML, Kirtley S, Waffenschmidt S, et al. PRISMA-S: an extension to the PRISMA statement for reporting literature searches in systematic reviews. Syst Rev. 2021;10:39. doi:10.1186/s13643-020-01542-z
- 5.Bramer WM, Giustini D, de Jonge GB, Holland L, Bekhuis T. De-duplication of database search results for systematic reviews in EndNote. J Med Libr Assoc. 2016;104(3):240-243. doi:10.5195/jmla.2016.24
- 6.Sterne JAC, Savovic J, Page MJ, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366:l4898. doi:10.1136/bmj.l4898
This guide is general educational information. It is not a substitute for advice from your supervisor, a statistician or your ethics committee.