# Do Systematic Reviews Need Ethical Approval? A systematic review reanalyzes studies that other researchers already published, so it usually does not involve people or animals directly, and in most cases it does not require approval from an ethics committee or Institutional Review Board (IRB). The rule of thumb is simple. If your review touches only published literature and aggregated, de-identified data, you can normally proceed without an ethics application. If you plan to collect new data from anyone, handle records that identify individual patients, or use confidential datasets from third parties, you need to talk to your ethics office before you start. That rule covers the majority of cases, but the details matter, because the line between exempt and non-exempt research is easier to cross than most students realize. This article walks through the three realistic options for handling ethics in a systematic review, compares them side by side, works through scenarios you are likely to face, and covers what a review needs in place of ethics approval, such as protocol registration and a documented systematic literature search. ## Why Ethics Approval Exists in the First Place Ethics committees and IRBs exist to protect the people (and sometimes animals) who bear the risks of research. When a team runs a clinical trial, patients accept real burdens, possible side effects, blood draws, and lost time, so an independent body reviews whether the benefits and safeguards justify the risks. A systematic review does not create that kind of direct risk. The reviewers read papers, screen titles, extract numbers from tables, and synthesize results. Nobody donates a sample. Nobody fills out a questionnaire. Nobody gets randomized to anything. You can see this clearly in how published reviews describe their methods. One review of interleukin inhibitors in psoriasis searched five platforms, retrieved 7,444 records, and had two reviewers screen titles and abstracts independently before full-text review [1]. Every one of those 7,444 records was a document, not a person. A review of trauma surgery outcomes searched four databases and pooled data from twelve retrospective observational studies covering 2,854 patients, all of whom had been treated years before the review began [2]. The reviewers worked with numbers on pages and in spreadsheets. This is why most universities, hospitals, and funding bodies treat reviews of published literature as outside the definition of human subjects research. In the United States, the Federal Policy for the Protection of Human Subjects (known as the Common Rule) defines research as a systematic investigation designed to develop generalizable knowledge that involves obtaining information about living individuals through intervention or interaction, or obtaining identifiable private information. Reading and synthesizing already-published papers does neither. That said, the answer is not always a flat no, and the exceptions are where projects go wrong. ## Does a Systematic Review Need Ethical Approval? The Short Answer ![A three-question triage path leading to no application, a determination letter, or full ethics approval for a systematic review.](/article-images/illustrations/does-systematic-review-need-ethical-approval-ethics-triage.webp) The question "does systematic review need ethical approval" has three practical answers, depending on what your project actually involves. Most reviews fall into the first category. A smaller number need a formal determination letter. A minority need full approval. The comparison below is the core of the decision. | Option | What it involves | When it applies | Pros | Cons | |--|--|--|--|--| | No formal ethics application | Proceed directly, document your reasoning, register your protocol | Review uses only published, publicly available literature and aggregated data | Fastest start, no committee timelines, standard for most published reviews | You must recognize the boundary, because adding surveys or identifiable data changes the answer | | Exemption or waiver determination | Submit a short form to your IRB or ethics office, receive a written letter saying the project is not human subjects research or is exempt | Your institution requires written confirmation, your funder asks for it, your data sources sit in a gray zone | Cheap in time, gives you a document journals and examiners can ask for, protects you if the project scope shifts | Requires some lead time, forms vary by institution, a determination can still be overturned if the scope changes | | Full ethics approval | Complete a standard application describing participants, consent, data protection, and risks, attend review if required | Review includes a survey or Delphi panel, uses identifiable individual-level data, or uses unpublished confidential records | Fully compliant for any secondary analysis or mixed-methods extension, satisfies strict journals | Slowest path, committee cycles can take weeks or months, adds ongoing reporting duties | Notice that the pros and cons are mostly about time and paperwork rather than scientific quality. None of these options changes how good your synthesis is. They change whether an administrator, a journal editor, or a thesis examiner will accept that you handled the ethical dimension correctly. ## Option 1: Proceeding Without a Formal Application For the typical review, this is the correct path, and it is worth understanding why so you can defend it if anyone asks. A systematic review is an act of synthesis. The raw material is other people's completed research. Consider a recent review of early versus delayed drainage in necrotizing pancreatitis. The authors searched PubMed, Embase, and the Cochrane Library, screened literature according to the PRISMA guidelines, and included 11 studies covering 1,697 patients [3]. PRISMA, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses, is the international reporting standard for these projects. Every patient represented in that analysis was treated in a study completed long before the review team formed. The review team never interacted with a patient, never accessed a medical record, and never held identifiable data. Their dataset was a table of published summary statistics. The same pattern repeats across the literature. A review of immune checkpoint inhibition for melanoma brain metastases searched PubMed, Scopus, and Embase and pooled outcomes from 24 published studies of 3,518 patients [4]. A review of eating disorder prevention programs searched Scopus, CINAHL, Embase, PsycINFO, MEDLINE, and Google Scholar and analyzed 17 published randomized trials [5]. A radiomics review in gastric cancer screened four databases and extracted data from 13 studies of 2,447 patients [6]. In each case, the unit of analysis was a publication. Because the work involves no interventions, no interactions, and no identifiable private information, the standard position across research institutions is that such a project does not constitute human subjects research and therefore falls outside the scope of ethics committee review. Many institutions still encourage (or require) a quick check with the ethics office or a recorded self-assessment, which is a reasonable habit even where it is not mandatory. One caution applies even here. De-identification is doing a lot of quiet work in this argument. The moment your project starts handling records that identify individuals, you have left Option 1. ## Option 2: Getting a Formal Exemption or Waiver Determination Some situations call for a piece of paper even when no real ethical risk exists. A determination request is a short submission to your IRB or [research ethics](/blog/guides/research-ethics-fundamentals-a-practical-guide-for-early-career-scientists) office describing your project and asking the office to confirm in writing that it is not human subjects research, or that it qualifies for an exemption category. You should choose this option when any of the following applies: - Your institution's policy requires written confirmation for all research outputs, including theses and dissertations. - Your funder or sponsor asks for an ethics letter as a condition of the grant, regardless of study type. - Your data sources sit in a gray zone. For example, you plan to contact trial authors for unpublished data, or you will use a dataset released under access restrictions rather than fully open terms. - You are a student and your supervisor or graduate school wants the letter on file. The process is usually lightweight. You describe the sources of data (published articles, public registries, aggregate statistics), confirm that no individuals will be contacted or identified, and state that the output will be a synthesis of existing literature. The office issues a letter confirming the project is not human subjects research. That letter solves a recurring problem at manuscript submission, because some journal editors ask about ethics approval for review articles, and a determination letter answers the question in one line. A related but often confused requirement is protocol registration. Registration is not an ethics process. PROSPERO is the international prospective register of systematic reviews, and registering there documents your research question, eligibility criteria, and planned analysis before you begin, which protects the review against selective reporting. A recent review of traumatic axillosubclavian arterial injuries registered its protocol with PROSPERO under the registration number CRD420251018150 before searching PubMed, Embase, the Cochrane Library, and Scopus [2]. Registration and ethics review answer different questions. Registration proves you planned your synthesis honestly. Ethics review proves you protected human participants. Reviews frequently need the first and almost never need the second. Some teams go further and publish the protocol itself as a separate paper. A weight management review in adults with intellectual disabilities published its full protocol in BMJ Open, describing the review's aims, search plan, and synthesis approach before the results existed [7]. Publishing a protocol is optional for most reviews, but it strengthens transparency and is another item that is not an ethics approval. ## Option 3: When Full Ethical Approval Is Needed Three situations push a review project into genuine human subjects territory. **Surveys and expert panels embedded in the review.** Many reviews include a consensus step, such as a Delphi survey where experts score items, or a questionnaire sent to study authors. The people answering the survey are participants in research. They interact with you, you record their responses, and their contributions shape the findings. This requires full ethics approval with participant information and consent procedures. If your protocol starts as a pure literature synthesis but you later add an author survey, the project has changed category, and you need approval before sending the first email. **Individual patient (or participant) data.** A standard meta-analysis uses the published summary statistics of each study, such as pooled effect estimates and [confidence intervals](/blog/data-analysis/confidence-interval-formula-examples). A more powerful design, called an individual participant data meta-analysis, obtains the raw row-level data from the original studies. When those data are anonymized and transferred under a data-sharing agreement, treatment depends on the dataset and jurisdiction. When the data include identifiers, or when they were collected under consent terms that did not anticipate reuse, your ethics office may require review or a formal data-use determination. Treat any request for patient-level records as a trigger for a conversation with the ethics office, not as a routine download. **Confidential or restricted-access data.** Some datasets require signed access agreements, institutional attestation, or regulator approval. Whether the ethics committee is involved depends on the terms of access and local policy. The safe rule is to treat restricted access as a signal that a determination is needed. A useful boundary check is to ask two questions. First, will I interact with any living person as part of this project, including by email? Second, could any person be identified from the data I hold? If the answer to both is no, and all data are public and aggregated, you are almost certainly in Option 1 or Option 2. If the answer to either is yes, escalate. ## What a Systematic Review Needs Instead of Ethics Approval ![Side-by-side comparison of PROSPERO registration and ethical approval across purpose, what it covers, and when each is needed.](/article-images/illustrations/does-systematic-review-need-ethical-approval-registration-vs-approval.webp) A review without an ethics application is not a review without rigor requirements. The infrastructure that replaces ethics review has several parts, and skipping them causes far more manuscript rejections than missing ethics letters ever will. **A registered, prespecified protocol.** Registering the review question, eligibility criteria, outcomes, and analysis plan before you search is the reviews-equivalent of pre-registration. It takes an hour or two and it guards you against the most common criticism of post-hoc synthesis. **A documented systematic literature search.** The term systematic literature search refers to a reproducible, comprehensive search strategy across multiple databases with recorded dates, filters, and search strings. Published reviews demonstrate the expected scale. One economic evaluation review searched six databases from inception to June 2023 [8]. A cost-effectiveness review of digital health in primary care searched PubMed, Scopus, and Health Business Elite in January 2025 and then updated the search in September 2026, reporting the results in a PRISMA flow diagram [9]. A review of HSCT for Hurler syndrome applied a search strategy with no date limits across MEDLINE, EMBASE, and other databases, completed the searches on a fixed date, and used reference checking to ensure completeness [10]. A review of automated ICD coding searched six databases and screened 4,280 initial citations down to 54 included studies [11]. Each of these teams documented exactly where the evidence came from and when they looked for it. That documentation is what separates a systematic review from a narrative summary, and it is what readers of a paper on how to write a literature review most often fail to grasp. A narrative literature review selects representative studies. A systematic review commits, in writing and in advance, to finding everything that meets its criteria. **Independent dual screening and extraction.** Two reviewers should screen records independently and resolve disagreements by discussion or with a third reviewer. The psoriasis review mentioned earlier used exactly this structure across 7,444 records, with a third reviewer resolving disagreements [1]. A review of propensity score methods managed search results in Zotero to remove duplicates and had two reviewers screen titles and abstracts independently [12]. Software platforms help. One review of patient-reported outcome instruments in bladder cancer managed data extraction in Covidence, a dedicated screening platform, after searching six databases [13]. **Structured quality assessment.** Appraisal tools differ by study design, and naming the tool you used is now standard practice. Options include the Cochrane Risk of Bias tool (RoB-2) for randomized trials, as used in a review of school-based eating disorder interventions [5], ROBINS-I for non-randomized studies, as used in the arterial injury review [2], and the Joanna Briggs Institute checklists, used for cross-sectional prevalence studies in a prostate cancer osteoporosis review [14] and for economic evaluations in the digital health review [9]. Reviews of diagnostic accuracy studies, such as the radiomics work in gastric cancer [6], apply design-specific appraisal in the same spirit. For reviews of health measurement instruments, the COSMIN framework structures the appraisal of psychometric evidence [13]. **PRISMA reporting.** The PRISMA statement governs how you report the flow of records, your searches, your selections, and your synthesis. Almost every modern review in the sample declares PRISMA adherence, from the economic evaluation review of injury care [8] to the meta-analysis of drainage timing in pancreatitis [3] to the atrial fibrillation screening review, which followed PRISMA 2020 guidelines explicitly [15]. Put together, this package (registration, documented search, dual screening, structured appraisal, PRISMA reporting) is the quality framework that substitutes for ethics oversight in secondary research. Invest your effort here rather than in an unnecessary ethics form. ## Worked Scenarios **Scenario 1: A master's student reviews randomized trials of a rehabilitation technique.** The student searches four databases, screens with a second reviewer, and pools published effect sizes. No individuals are contacted, and no identifiable data are involved. This is Option 1. The student registers the protocol, follows PRISMA, and records a brief note in the thesis methods stating why ethics review was not required. If the university requires a determination letter, one short form covers it. **Scenario 2: A hospital team reviews cost-effectiveness studies and then surveys local clinicians about implementation barriers.** The literature synthesis portion needs no approval, following the pattern of published economic reviews that screened hundreds of studies from six databases [8]. The survey, however, involves living people answering questions. The team submits the survey component for full approval before recruitment, and keeps the literature component on the no-approval track. Splitting the project into components in the ethics application is normal and accepted. **Scenario 3: A doctoral researcher requests individual participant data from the investigators of seven included trials.** Even though the original trials had approval and consent, the reuse of row-level patient data is a new processing activity. The researcher contacts the ethics office for a determination before requesting any files, confirms the data-sharing terms of each trial, and documents what was agreed. Depending on the identifiability of the data and local policy, this may resolve as an exemption or may require a full application. Either way, the conversation happens before the data arrive, not after. ## Common Mistakes and Limitations The mistakes below come up repeatedly in real review projects, and every one of them is avoidable with a few minutes of planning. - **Treating PROSPERO registration as ethics approval.** They are unrelated. Registration documents your planned synthesis. It confers no ethical clearance, and ethics committees confer no registration status. You may need both, one, or neither, depending on your project and institution. - **Adding a survey late and skipping approval.** The most common route into non-compliance is scope creep. A review that begins as pure synthesis gains a Delphi round in month four, and nobody revisits the ethics question. Decide at the protocol stage whether any primary data collection is possible, and if it becomes possible later, stop and consult the ethics office. - **Assuming student projects are automatically exempt.** Exemption depends on the design and the data, not on the seniority of the researcher. Some universities exempt course assignments across the board, others do not. The only binding answer comes from your own institution's ethics office or the policy document that governs your degree. - **Confusing data access agreements with ethical clearance.** Signing a data-use agreement with a database provider satisfies the provider's terms. It does not substitute for an ethics determination if the data are identifiable or restricted. - **Failing to report the ethics decision in the manuscript.** Reviewers and editors increasingly ask. One sentence in your methods stating that the project used only published, aggregated data and required no ethics approval (or citing your determination letter) prevents a query at revision stage. - **Overstating what a search proved.** A documented systematic literature search is reproducible, but search dates matter. A review that searches only through a certain month misses everything published after, which is why good teams run updated searches, as the digital health review did with its September 2026 update [9]. Report your search dates precisely. One limitation applies to everything above. Policies differ between institutions, countries, and journals, and this article cannot tell you what your specific ethics office will decide. When the stakes are a thesis deadline or a funded grant, a two-line email to your ethics administrator settles the question authoritatively. ## Frequently Asked Questions ### Does a systematic review need ethical approval if it only uses published papers? Usually no. A review that reads, screens, and synthesizes published literature involves no interaction with people and no identifiable private data, so it generally falls outside the definition of human subjects research. Register your protocol and document your search, and record a sentence in your methods explaining why approval was not required. ### What is the difference between PROSPERO registration and ethical approval? They answer entirely different questions. PROSPERO registration documents your planned review question and analysis before you start, protecting against selective reporting. Ethical approval protects human participants. A review typically needs registration and does not need approval unless it involves surveys or identifiable data. ### Do reviews that use individual patient data need ethics approval? Often yes, or at minimum a formal determination. Row-level data can carry identifiers or consent limitations that the original trials never anticipated. Contact your ethics office before requesting any individual-level dataset rather than after. ### Do thesis or dissertation systematic reviews need ethics approval? It depends entirely on your university's policy. Many graduate schools require a written exemption or determination letter even for purely literature-based projects, while others require nothing for research that involves no human participants. Check your graduate school handbook or ask your supervisor in the first week of the project. ### Can a review project that started without approval require approval later? Yes, if the scope changes. Adding an author survey, an expert panel, or a request for identifiable records converts part of the project into human subjects research. Seek the determination or approval before you collect that new component. ### Do I need to state the ethics decision when I publish the review? Yes, briefly. State in your methods that the review used only published, aggregated data and required no ethical approval, or cite your exemption letter number. A single sentence answers the editor's standard query and avoids delay at revision. ## Sources 1. [Comparison of the Effects and Safety of Planned Limited-Time Withdrawal Versus Continuous Maintenance of Interleukin Inhibitors in Patients with Psoriasis: A Systematic Review and Network Meta-Analysis.](https://pubmed.ncbi.nlm.nih.gov/41851588/) 2. [Comparison of short-term outcomes in open versus endovascular management of traumatic axillosubclavian arterial injuries in the contemporary era: a systematic review and meta-analysis.](https://pubmed.ncbi.nlm.nih.gov/42231442/) 3. [Comparison of early and late drainage interventions in necrotizing pancreatitis: a systematic review and meta-analysis.](https://pubmed.ncbi.nlm.nih.gov/42174434/) 4. [Safety and efficacy of dual immune checkpoint inhibition for melanoma brain metastases and its comparison to ICI monotherapy: A systematic review and meta-analysis.](https://pubmed.ncbi.nlm.nih.gov/42934318/) 5. [Effectiveness of universal secondary school-based interventions addressing the psychological risk factors of eating disorders in adolescents: a systematic review.](https://pubmed.ncbi.nlm.nih.gov/42881424/) 6. [Radiomics for Prediction of Microsatellite Instability Status in Gastric Cancer: A Systematic Review and Meta-Analysis.](https://pubmed.ncbi.nlm.nih.gov/42599597/) 7. [Acceptability and impact of weight management interventions for adults with intellectual disability: protocol for a systematic review and meta-analysis.](https://pubmed.ncbi.nlm.nih.gov/42551997/) 8. [Economic Evaluation Methods in Post-Injury Care: A Global Systematic Review and Critical Appraisal of Their Suitability for Decision-Making.](https://pubmed.ncbi.nlm.nih.gov/42961599/) 9. [Cost-effectiveness of digital health in primary care: a systematic review.](https://pubmed.ncbi.nlm.nih.gov/43009961/) 10. [Efficacy and safety of haemopoietic stem cell transplant (HSCT) for mucopolysaccharidosis type I-Hurler syndrome (MPS-IH): a systematic review.](https://pubmed.ncbi.nlm.nih.gov/43004383/) 11. [Recent Advances in AI for Automated ICD Coding: A Systematic Literature Review.](https://pubmed.ncbi.nlm.nih.gov/42461368/) 12. [Performance of propensity score methods in observational studies: a systematic review.](https://pubmed.ncbi.nlm.nih.gov/41792825/) 13. [Patient-Reported Outcome Instruments in Non-Muscle-Invasive Bladder Cancer (NMIBC): A Systematic Review.](https://pubmed.ncbi.nlm.nih.gov/42642612/) 14. [Prevalence of osteoporosis and low bone mass in men with prostate cancer: a systematic review and meta-analysis.](https://pubmed.ncbi.nlm.nih.gov/42828578/) 15. [Opportunistic and systematic screening for atrial fibrillation in adults aged ≥ 65 years in primary care: a systematic review of randomized and cluster randomized trials.](https://pubmed.ncbi.nlm.nih.gov/42750002/)