Research Ethics and Integrity

The social licence. Medical research rests on a social licence: participants expose themselves to risk so that future patients may benefit. Research ethics is the body of principle and law that keeps that licence trustworthy, and it protects three things at once:

• The participant, from harm and exploitation
• The integrity of the knowledge produced
• The researcher and the institution, from the legal, professional and reputational consequences of a breach

The governing idea. Every code says the same thing: the well-being of the individual participant takes precedence over all other interests, including the interests of science and society. Every principle, approval and consent process below is a mechanism for enforcing that one idea.

Why examiners return to it. The historical disasters that created the field (wartime experimentation, the Tuskegee study, retention of organs without consent) were failures of judgement by otherwise competent doctors. Examiners therefore test whether you can recognise the principles, and not merely recite the names.

The duties are also concrete and operational. Every surgeon who runs a registry, audits an implant or co-authors a paper is already inside this framework, whether they realise it or not.

The penalties are severe and public: retraction, deregistration, loss of funding, referral to the regulator and, in some jurisdictions, criminal liability. A wrong answer signals an unsafe practitioner.

The frame. Examiners look for a structured answer, and the cleanest frame is the chronological ladder of codes. Each was a direct response to a scandal, and together they harden the same core idea into progressively enforceable rules. Asked "what governs research ethics?", climb the ladder, then name the gatekeeper: the independent committee that applies Helsinki to the actual protocol.

1. Nuremberg Code (1947). Drawn from the judgement at the Doctors' Trial. Its first article makes the voluntary, informed consent of the participant "absolutely essential", and it requires that the experiment be for the good of society, avoid unnecessary suffering, and be stoppable. It is the cornerstone of modern consent.
2. Declaration of Helsinki (1964). The World Medical Association's ethical principles for medical research involving human participants, and the document research ethics committees actually apply. It requires independent ethics review, a favourable risk-benefit balance, special protection for vulnerable groups, and the participant's right to withdraw. It is revised periodically, most recently in 2024.
3. Belmont Report (1979). The US National Commission distilled research ethics into three principles: respect for persons, beneficence and justice. It underpins the US "Common Rule" and is the cleanest three-principle summary an examiner will accept.
4. CIOMS International Ethical Guidelines (2002, revised 2016). These extend the principles to research in low-resource settings and to research involving identifiable data and biological materials.
5. National law and good clinical practice. ICH-GCP for trials, and each jurisdiction's own law and committees, compared side by side in the final section.

The four principles. The dominant framework in day-to-day clinical and research ethics is the four-principles approach of Beauchamp and Childress, set out in their Principles of Biomedical Ethics. It is a shared vocabulary for naming the values in conflict, which is what an examiner wants to see you do; it is not a decision algorithm. Name the four whenever an ethical conflict arises:

• Autonomy — respect the participant's capacity to decide; consent must be informed and voluntary
• Beneficence — act in the participant's best interests and maximise possible benefits
• Non-maleficence — avoid harm (primum non nocere); weigh risk against benefit
• Justice — fair distribution of burden and benefit, including fair selection of participants

Equipoise. The concept examiners reach for specifically in trial ethics is equipoise: honest, genuine uncertainty within the expert community about which of two treatments is better. Charles Fried framed the problem in 1974, and Benjamin Freedman refined it in 1987 into "clinical equipoise": randomisation is ethically permissible only while the community of competent practitioners is collectively undecided.

Why it answers the objection. The common objection is "how can a doctor randomly assign a patient?" The doctor is entering the patient at a moment when no one yet knows which arm is superior, so the choice is not arbitrary, and the trial is the means of resolving that uncertainty.

When it ends. Equipoise is disturbed the moment a treatment proves superior. That is why trials have stopping rules and data-monitoring committees, and why continuing to enrol once the question is answered is itself unethical.

The committee. Before any participant is approached, the protocol must be reviewed and approved by an independent committee: an Institutional Review Board in the United States or a Research Ethics Committee elsewhere. The committee exists precisely because individual researchers cannot be expected to judge their own risk-benefit balance objectively. It tests three things:

• Whether the risk-benefit balance is favourable
• Whether the selection of participants is fair
• Whether the consent process is adequate

What approval covers. Approval is a precondition, and it covers the protocol, the participant information sheet, the consent form, and any later amendment.

Research, audit or service evaluation? Where this boundary lies is a frequent practical and exam question, because it determines whether ethics approval is needed at all.

Where the border blurs. A quality project that prospectively assigns patients to different pathways, or that uses data in a way patients would not have expected, has crossed into research and needs approval. When there is genuine doubt, ask the committee in writing before starting, rather than proceeding and arguing afterwards. The duty is to the participant and the integrity of the record, not to the timeline.

A higher standard. Consent in research is a process, and a signature alone does not make it. It is held to a higher standard than clinical consent because the participant may derive no personal benefit. Five elements must all be present, and the right to withdraw runs through all of them:

• Capacity — the participant can understand, retain and weigh the information
• Information — adequate disclosure, including that this is research
• Comprehension — the information is understood and checked, not just handed over
• Voluntariness — the decision is free of coercion and undue influence, including therapeutic misconception
• Specificity — consent to this protocol, re-confirmed if material new risks emerge

What must be disclosed. The purpose of the study, the procedures and their alternatives, the foreseeable risks and discomforts, the absence of guaranteed benefit, the right to withdraw without any penalty to ongoing care, and the handling of data.

Therapeutic misconception. A recurring trap: the participant wrongly believes the trial is individualised treatment chosen for their benefit. The researcher must correct this explicitly, because a consent built on a misunderstanding is not valid consent.

The case-law ladder. The standard for what counts as adequate disclosure in consent law has itself moved, and examiners expect the sequence.

Montgomery in research. The practical effect of Montgomery is that risk disclosure is now judged from the patient's perspective, not the doctor's. In research this reinforces the duty: a trial participant must be told everything a reasonable person in their position would consider significant, including the uncertainty that justifies the trial in the first place.

Vulnerable participants. Children, adults lacking capacity, pregnant women, prisoners, and patients in emergency or intensive-care settings require additional safeguards. Consent may need to be obtained from a legally authorised representative, with assent from the participant where possible. The study must be one that cannot equally be done in a non-vulnerable group.

Integrity and misconduct. Research integrity is the positive duty to conduct, record and report work honestly, and research misconduct is the deliberate breach of that duty. The widely used definition, from the US Office of Research Integrity, centres on three acts, known as FFP.

Questionable research practices. A large and damaging grey zone surrounds these: practices that fall short of formal misconduct but corrode the record.

• Salami-slicing one study into several papers
• Selective reporting of outcomes
• p-hacking
• Image manipulation
• Ghost and gift authorship
• Failing to publish negative results

The Singapore Statement on Research Integrity (2010) addressed exactly this zone by setting four principles for the whole conduct of research: honesty, accountability, professional courtesy and fairness, and good stewardship.

When you suspect misconduct. This is the part examiners most often probe, and the answer is a process, not a confrontation:

1. Do not confront the individual informally
2. Do not alter any records; preserve the evidence
3. Report in confidence to your institution's research integrity officer or your defence union
4. Let the formal process determine due process

The four ICMJE criteria. Authorship is a statement of responsibility, both credit and accountability, and it is governed by the International Committee of Medical Journal Editors (ICMJE) Recommendations. To be listed as an author a contributor must meet all four criteria; there is no half-measure.

1. Substantial contribution to the conception or design, or to data acquisition, analysis or interpretation
2. Drafting or critically revising the work for important intellectual content
3. Final approval of the version to be published
4. Agreement to be accountable for all aspects of the work

Acknowledgements. Contributors who fall short of all four (for example providing funding, collecting data without intellectual input, or general supervision) belong in the Acknowledgements, with their consent, not on the author list.

Gift and ghost authorship. Two recurring abuses are explicitly misconduct in publication ethics, although they sit outside the narrower FFP definition and among the questionable practices above. Gift or honorary authorship adds a senior name who did not meet the criteria to lend credibility. Ghost authorship hides a real contributor, often a writer funded by industry, from the byline. Both distort the record of who is accountable, and both are condemned by the ICMJE and by the Committee on Publication Ethics.

Beyond the byline. Four further duties apply:

• Trial registration — clinical trials must be publicly registered before the first participant is enrolled, in a recognised registry such as ClinicalTrials.gov, with reporting of results
• Reporting guidelines — must be followed: CONSORT for randomised trials, PRISMA for systematic reviews, STROBE for observational studies, and CARE for case reports
• Conflicts of interest — financial and non-financial, must be disclosed at submission, because undisclosed conflict is a common cause of retraction
• Correction — when published work is wrong, the duty is to correct it: by a correction or erratum for an honest error, by a retraction for serious flaws or misconduct, never by silent deletion

Artificial intelligence. One point has hardened in the last few years: artificial-intelligence and large-language-model tools cannot be authors, because they cannot take responsibility, and their use in writing or analysis must be disclosed. Using them to generate text without declaration is a breach of the same integrity duty.

The Swiss cheese model. Research ethics and patient safety share an underlying idea that examiners reward: most failures are produced by systems, not by a single bad person, and integrity is sustained by design rather than willpower. James Reason's model of accident causation pictures the defences against failure as slices of Swiss cheese, each with holes; a harm reaches the patient only when the holes line up.

Layers for integrity. The robust response is to build layers rather than rely on individual honesty alone:

• Independent review
• Dual data entry
• Trial registration
• Open data
• Conflict declaration

The checklist as a case study. The WHO Surgical Safety Checklist is included here because it shows the whole loop the topic sits inside: a research question, an ethics-approved study, honest reporting, and a published standard that institutions then adopt. The global study by Haynes and colleagues reported a fall in death from 1.5 to 0.8 per cent and in complications from 11 to 7 per cent after checklist implementation, roughly a ~47%lower death rate relative reduction.

What happened next. Haynes was a before-and-after study at eight sites with no control hospitals. When Ontario mandated checklist adoption across 101 hospitals, adjusted operative mortality did not change significantly (0.71 per cent before, 0.65 per cent after). Read together, as the evidence cards below set out, the two studies show that the 47 per cent figure was never a reliable estimate of what mandating the checklist would achieve.

Same principles, different machinery. Research ethics is governed worldwide by the same Helsinki principles, and everywhere the committee must be independent of the researcher. The mechanisms differ by jurisdiction; examiners accept either system if the principles are right.

Professional codes. Professional bodies translate these principles into surgeon-specific duties. The AAOS Code of Medical Ethics and Professionalism and its Standards of Professionalism in the United States, and the GMC Good Medical Practice in the United Kingdom, both make research honesty, valid consent and accurate authorship explicit professional obligations. None of these frameworks includes country-specific billing or reimbursement item numbers, which are irrelevant both to the ethics and to the exam.

Everything below condenses the topic for revision and viva practice — the high-yield points, the memory hooks, worked vivas, and a one-screen cheat sheet.

Related pages: Core Bioethics Principles for the four-principles framework this page applies to research, and for the clinical setting in which the same reasoning is used; Informed Consent and Shared Decision-Making for consent to treatment, which shares Montgomery's materiality test but differs in one decisive way - a research participant may receive no benefit at all, which is why the consent bar is higher; Professional Boundaries, Confidentiality and Social Media for the duties owed to participant data once it leaves the clinical record; Medicolegal Principles and Negligence for Bolam, Bolitho and Montgomery in their original setting; Conflicts of Interest and Industry Relationships for the disclosure obligations the ICMJE recommendations impose and the reason they exist; Evidence-Based Orthopaedics and Critical Appraisal and Levels of Evidence for the appraisal skills the checklist case study above demands; and Study Design Types for why a before-and-after design could not have supported the effect size it reported.