Extended definition
A research ethics committee is an independent institutional body that ethically evaluates research projects involving human participants before data collection begins, with authority to approve, require modifications, or refuse. National structure varies: in Brazil, the CEP/CONEP system regulated by CNS Resolution 466/2012, with mandatory approval via Plataforma Brasil for human research; in the US, IRB (Institutional Review Board) under 45 CFR 46 (Common Rule); in the UK, REC (Research Ethics Committee) coordinated by the HRA (Health Research Authority); in the European Union, variable system with convergence via directives; variable systems in Asia, Africa, and Latin America. International ethical foundations include the Nuremberg Code (1947, post-Holocaust), the Declaration of Helsinki (WMA, 1964, periodically updated), the Belmont Report (1979, US) which articulated three principles — respect for persons, beneficence, justice — and Beauchamp and Childress’s work (Principles of Biomedical Ethics, 2019, 8th ed.) which expanded to four principles including non-maleficence. Emanuel, Wendler, and Grady (2000, JAMA) proposed an operational framework of seven requirements for ethically valid clinical research: value, scientific validity, fair participant selection, favorable risk-benefit ratio, independent review, informed consent, respect for enrolled participants.
When it applies
Ethics approval is a regulatory requirement in research with human participants: clinical trials, surveys with identifiable data, interviews, focus groups, observational studies with medical records, research with human biological material, studies with Indigenous peoples (with collective informed consent and specialized ethics committees in some countries), research with vulnerable populations (children, elderly, prisoners, persons with cognitive impairment). It also applies to animal experiments via specific committees (CEUA in Brazil, IACUC in the US). Approval is a precondition for publication in top-tier journals — ethics approval declaration with protocol number and committee is part of the methods section. It applies in international research with inter-institutional cooperation: approval in all involved jurisdictions is standard practice.
When it does not apply
Ethics approval does not apply to research not directly involving human subjects: analysis of fully anonymized public data (with care — re-identification is a growing risk in big data), systematic reviews of published literature, purely theoretical or mathematical research. It does not apply to some educational activities classified as program evaluation (with formal criteria), although the boundary is contested. In minimal-risk research with public data, some systems offer expedited review or exemption — formally requested, not self-decided by the researcher. It does not replace informed consent: ethics approval and consent are complementary instruments, not alternatives.
Applications by field
— Health and biomedical sciences: CEP/IRB approval mandatory; ICMJE requires declaration at submission. — Empirical social sciences: interviews, surveys, ethnography require ethics approval; similar processes in REC/IRB. — Education: research with students and teachers frequently requires ethics approval, especially in vulnerable populations. — Applied ML research: clinical data use requires approval; social media data use is a gray zone in transformation.
Common pitfalls
The first pitfall is submitting an incomplete or generic project to the ethics committee expecting fast approval: substantive documentation (informed consent forms, instruments, analysis plan, data management plan) is required, and absence delays the process. The second is confusing initial approval with permanence: substantive project changes require amendment submission; collecting outside the approved scope is an ethical violation. The third is failing to respect informed consent in fact: a signed but uncomprehended consent form (overly technical language, pressure to participate, vulnerability) is ethically problematic, even if formally compliant. The fourth is reporting approval in the manuscript without essential information (protocol number, date, institution) — top-tier journals reject for absence. The fifth is assuming anonymized data are outside ethical scope: re-identification is a growing risk especially in big data — case-by-case ethical analysis is prudent.