Do You Need Ethics Approval for a Literature Review? (2026 Guide)
By Megan Carter 9 min read
Usually not, but "usually" is doing real work in that sentence. The answer is set by your institution rather than by your method, and two universities can read the same review two different ways.
A review that only reads published papers is secondary research, and many universities exempt it. The University of St Andrews states plainly that "literature reviews, including systematic reviews and meta-analyses, do not require ethical review" unless the review analyses raw data or raises specific ethical issues (St Andrews, ethical review). UNISA takes the opposite line: its Masters and Doctoral ethics guidance says all research must be considered for ethical clearance before it may commence, and that this "includes research involving human participants, animals, conceptual work, the use of secondary data, laboratory experiments and literature review studies" (UNISA, College of Economic and Management Sciences).
So the answer is not decided by your method. It is decided by your institution, and you have to go and read its policy rather than reason it out from first principles.
What counts as secondary research for ethics purposes?
Secondary research is research that analyses material which has already been published or made publicly available, without collecting any new data from people. A standard literature review is secondary research: you read published papers, compare what they found, and write about the pattern.
That definition is what most ethics policies are testing for when they ask whether your project involves human participants. You read papers, you do not recruit anyone, so there is no participant in your study to protect.
The University of Cambridge School of the Biological Sciences puts the boundary in one sentence. Studies that do not require ethical review include "those involving information freely available in the public domain (e.g. published biographies, newspaper accounts), and the analysis of datasets, either open source or obtained from other researchers, where the data are properly anonymised and informed consent was obtained at the time of original data collection" (Cambridge, School of the Biological Sciences). Notice the two conditions attached to the dataset half of that sentence. Anonymised, and consented at source. Reading the discussion section of a published paper clears both without you having to think about it. Downloading the supplementary file and re-analysing the participant rows does not.
Why does one university require clearance when another does not?
Because ethics committees are institutional, not national, and they set their own screening thresholds. As of August 2026, the published policies of six universities in South Africa and the UK split three ways on a review that reads nothing but published papers.
| Institution (country) | Literature-only review needs clearance? | What the policy actually says |
|---|---|---|
| UNISA (South Africa) | Yes, it must be submitted for consideration | "All research must ... be considered for ethical clearance before it may commence", and this includes "literature review studies" |
| UCT, Faculty of Commerce (South Africa) | No | Research "involving only the use of secondary data ... does not require the completion of either the pre-screening questionnaire (PSQ) or the completion of an application for ethical clearance" |
| Wits (South Africa) | Depends on the screening test | Clearance is for research that "may impact directly or indirectly on people, animals or the environment" |
| St Andrews (UK) | No, with two named exceptions | "Literature reviews, including systematic reviews and meta-analyses, do not require ethical review unless" the review analyses underpinning raw data, or raises issues such as social profiling or sensitive personal data |
| UCL (UK) | No | "Use of information available in the public domain" and "use of anonymised records and datasets in the public domain" are listed as minimal risk activity |
| Cambridge, Biological Sciences (UK) | No, if two conditions hold | Public-domain information and properly anonymised datasets with consent obtained at original collection |
UNISA is the outlier worth knowing about if you study in South Africa, and the wording is deliberate. The same page says "it is not for the candidate to decide whether his/her research requires ethical clearance." That sentence is the whole rule: at UNISA the screening decision belongs to the committee, not to you, even when you are confident the answer is obvious.
Does my review need approval if I analyse the data inside the papers?
Analysing participant-level data from inside a paper is the line most students cross without noticing, and yes, crossing it can pull a literature review into ethical review. The University of St Andrews names it as the first exception to its exemption for literature reviews: ethical review is required where the review "includes analysis of the underpinning raw data or of a dataset attached to a publication" (St Andrews, ethical review).
The practical test is what you open. If you extract the reported effect sizes, sample sizes and confidence intervals from the results tables of published papers and combine them, you are doing a meta-analysis of published summary statistics, which the policies of St Andrews, UCL and Cambridge all treat as secondary research. If you download the individual participant data that a paper deposited alongside itself and run your own analysis on those rows, you are working with a human dataset, and the two conditions set by the University of Cambridge School of the Biological Sciences apply: the data has to be properly anonymised and informed consent has to have been obtained at the time of original data collection (Cambridge, School of the Biological Sciences). An individual participant data meta-analysis is a different animal from a standard review, and it is worth reading the difference between a literature review and a systematic review before you commit to one.
What else turns a literature review into a study that needs review?
Three things, and none of them are about how many papers you read. The University of St Andrews lists the second exception to its literature review exemption as a review that "will raise ethical issues such as around social profiling, data management issues such as managing sensitive personal data, or access issues" (St Andrews, ethical review).
Scraping is the common trap. A study of what people post about a condition in public forums looks like desk research and reads like human subjects research to a committee, because the posts are attributable to identifiable people who did not consent to being studied. The University of Cambridge School of the Biological Sciences flags this directly, noting that even studies analysing public online resources may still need review to confirm participants gave informed consent and that data storage meets current protections (Cambridge). Sensitive personal data is the second trap: health, sexuality, religion, criminal records, immigration status. UCL adds a third that is not about risk at all, warning that approval may be required "by another external body, for example, to obtain research permissions, funding, or to publish results" (UCL, do I need ethical approval). A grant condition or a journal policy can create a requirement your own committee would have waived.
Do I need a NACOSTI research licence in Kenya for a desk review?
Assume yes until NACOSTI tells you otherwise. Kenya’s National Commission for Science, Technology and Innovation states that "all persons intending to undertake research in Kenya are required to apply for a Research Licence", and that "there are no exemptions from the NACOSTI Research Licence requirement for persons intending to undertake research in Kenya" (NACOSTI FAQs). Students are named in scope, as are Kenyan citizens, independent researchers and consultants.
What that page does not do is say whether a literature-only study counts as "undertaking research in Kenya", and I am not going to guess on your behalf. Two things are worth knowing before you ask. Licensing and institutional registration are separate processes at NACOSTI, so your university being registered does not cover you. And licence payments run through the eCitizen gateway, with no fee figure published on the FAQ page as of August 2026, so budget from the portal rather than from any number you read on a blog.
What do I write in my methods section if approval was not required?
Write one explicit sentence saying so, name the reason, and give a reference number if you were issued one. Something close to: "This study is a review of published literature and did not involve human participants or the collection of new data. In line with [institution]’s research ethics policy, ethical clearance was not required." Silence reads as an omission to an examiner, and an examiner who cannot tell whether you skipped approval or were exempted from it will ask.
Get the reference number if your institution offers one, even when the answer is no. The University of Cape Town’s Faculty of Commerce exempts secondary-data research from ethical clearance and still encourages researchers to apply anyway, in order "to receive certification and a reference number, in case funders, publishers, or any other legitimately interested party requires this" (UCT Commerce, ethics in research). That is five minutes of admin now against a journal query in eighteen months. If you are still drafting the surrounding chapter, how to write a methodology section covers where an ethics statement usually sits.
How do I find my own institution’s rule quickly?
Search your university domain for the phrase, not the topic. A query along the lines of "do I need ethical approval site:youruniversity.ac.za" reaches the screening page faster than the research office homepage does, because that page is almost always titled as a question. UCL’s page is literally called "Do I need ethical approval?" (UCL).
Then check three things on the page you land on. Whether it names literature reviews or secondary data explicitly. Whether a pre-screening form has to be filed even for exempt work, which the University of Cape Town’s Faculty of Commerce calls the PSQ (UCT Commerce). And which committee covers your faculty, since large universities run several rather than one central body, as the University of the Witwatersrand does (Wits, research ethics). If the page does not name your situation, email the committee and keep the reply. An emailed "not required" from the committee secretary is the record you want in your appendix.
Turn Your Reading Into a First Draft
Ethics clearance is the paperwork around the review. The review itself is still forty papers you have to read, hold in your head at once and turn into an argument.
Litrevu is an AI literature review assistant that turns the papers a researcher has already gathered into a cited first draft, with every citation traceable to the uploaded source. You upload the PDFs you chose, you get back a structured draft with the citations pointing at your own files, and then you do the part that is actually yours: read every line, check every citation against the paper it came from, and rewrite it in your voice. It is a starting point, not a submission.
Below is a format illustration, not a real Litrevu output and not a real study. It shows the shape of a drafted paragraph, with author and year left as tokens.
Ethical governance of secondary research
Institutional policy on secondary research diverges along the question of who makes the screening decision (Author, Year). One position holds that reviews of published material fall outside the remit of ethical review unless raw participant data is analysed (Author, Year). A second position requires that every project be submitted for consideration regardless of design, on the grounds that the exemption decision itself belongs to the committee (Author, Year). Both positions are defended on transparency grounds, which suggests the disagreement is procedural rather than substantive (Author, Year).
Sources: uploaded_paper_1.pdf p.4, uploaded_paper_2.pdf p.11, uploaded_paper_3.pdf p.2
If that is the stage you are stuck at, Litrevu gives you 800 words free with no credit card to see what it does with your own papers.
Start a draft from your own papersFrequently asked questions
Does a literature review count as research involving human participants?
No, not on its own. Reading and analysing published papers does not involve participants, which is why UCL lists use of information available in the public domain as minimal risk activity that does not require ethical review. The picture changes if you analyse participant-level data attached to those papers, or study identifiable posts by real people.
My supervisor says I do not need ethics approval. Is that enough?
It depends on where you study, and at some institutions it is explicitly not enough. UNISA’s guidance states that "it is not for the candidate to decide whether his/her research requires ethical clearance", which puts the decision with the committee. Ask your supervisor to point you at the policy page rather than relying on the verbal answer, and keep whatever the page says.
Do I need ethics approval for a systematic review?
At most universities, no. St Andrews names systematic reviews and meta-analyses alongside literature reviews as not requiring ethical review, subject to its two exceptions around raw data and ethical issues. A systematic review is more work than a narrative one but it is not a different category for ethics purposes, which is worth remembering while you are choosing between a scoping, systematic and narrative review.
What if I only realise halfway through that I need clearance?
Stop and apply, and say so honestly in the application. Most policies require clearance before research commences, which means retrospective applications are awkward but survivable, whereas an undeclared one discovered at examination is not. Your search and screening records help here, so keep your search strategy and PRISMA flow numbers dated as you go.
Does using an AI tool in my review need separate ethics approval?
Ethics approval and AI-use disclosure are different processes, and an exemption from one says nothing about the other. Ethics review covers risk to people and data; AI disclosure covers how you produced the work, and is usually governed by your assessment or academic integrity policy instead. Check both pages, and keep a record of what you wrote and when.