Ethical Principles in Research

Concept

Ethics is the study of moral principles that distinguish right from wrong conduct, and research ethics applies them to the design, conduct and reporting of research. It protects the dignity, rights, safety and privacy of participants, protects animals and the environment, and safeguards the integrity of knowledge so that the public can trust science. David Resnik lists its purposes as promoting the aims of research (truth and avoidance of error), supporting values needed for collaboration (trust, fairness, accountability), keeping researchers answerable to the public, and upholding wider moral and social values such as human rights, animal welfare and legality. Research ethics applies to all fields, not just medicine, and it is wider than the law: an act may be legal and still be unethical. Responsible conduct of research rests on honesty, objectivity, carefulness, openness, confidentiality and respect for colleagues and intellectual property.

Concept

Modern research ethics grew from abuses. The Nuremberg Code (1947) came from the trial of Nazi doctors; its first principle is that the voluntary consent of the human subject is absolutely essential. The Declaration of Helsinki, adopted by the World Medical Association in 1964 and revised several times since, is the central statement for medical research and requires ethics committee review, informed consent and a favourable risk–benefit balance. The Tuskegee Syphilis Study (1932–1972), in which poor Black men in Alabama were left untreated to observe the disease, led to the US National Research Act (1974) and then to the Belmont Report (1979) with its three principles: respect for persons, beneficence and justice. The CIOMS guidelines (from 1982, revised in 2016) apply these principles internationally, and ICH-GCP governs clinical trials.

Concept

The Belmont Report’s principles are respect for persons (autonomy and protection of those with diminished autonomy, giving rise to informed consent), beneficence (maximise benefits and minimise possible harm) and justice (fair distribution of the burdens and benefits of research and fair selection of participants). Beauchamp and Childress, in Principles of Biomedical Ethics, name four principles: autonomy, beneficence, non-maleficence (do no harm, primum non nocere) and justice. A risk–benefit assessment must show that risks are justified by the expected benefit. The Tuskegee study violated all three Belmont principles, notably justice, because a vulnerable group bore the burden.

Concept

Before a study involving people starts, an independent committee reviews the protocol, the consent documents and the safeguards: an Institutional Ethics Committee (IEC) in India, an Institutional Review Board (IRB) in the United States and a Research Ethics Committee (REC) in the UK. Such committees are multidisciplinary and include scientists, a clinician, a legal expert, a social scientist and a lay member, and they monitor studies after approval. In India the ICMR National Ethical Guidelines for Biomedical and Health Research Involving Human Participants (2017) set the standards, with general principles such as essentiality, voluntariness, non-exploitation, social responsibility, privacy and confidentiality, risk minimisation, professional competence, maximisation of benefit, transparency and accountability, and totality of responsibility. Clinical trials are governed by the New Drugs and Clinical Trials Rules, 2019, and registered in the Clinical Trials Registry of India (CTRI).

Concept

Vulnerable groups include children, prisoners, pregnant women, people with mental illness or cognitive impairment, the elderly, the economically or educationally disadvantaged, refugees, and people in dependent relationships such as students or employees. They need extra safeguards, such as parental consent plus the child’s assent, or a legally authorised representative. Minimal risk means harm no greater than in ordinary daily life. Deception is acceptable only when the study has real value, no non-deceptive alternative exists, the risk is minimal and participants are debriefed afterwards. Classic cases are Milgram’s obedience experiments (deception and distress), Zimbardo’s Stanford prison experiment (psychological harm) and Humphreys’ covert observation study, Tearoom Trade.

Concept

A conflict of interest arises when a researcher’s personal, financial or professional interest could, or could appear to, compromise judgement or objectivity. Financial examples are sponsor funding, consultancy fees, shares and patents; non-financial examples include rivalry, close personal relationships and academic competition. Conflicts are not always wrong, but they must be managed: disclose them to the institution, ethics committee, funder and journal; step aside from decisions where they apply (recusal); and arrange independent oversight. Reviewers and editors with a conflict should decline to handle the paper.

Concept

Research misconduct is conventionally summed up as FFP. Fabrication is making up data or results; falsification is manipulating materials, equipment, processes or data, or changing or omitting results so that the record is not accurate; plagiarism is taking another’s ideas, words or results without credit. Honest error and genuine differences of opinion are not misconduct. Related questionable research practices are p-hacking, HARKing (hypothesising after results are known), selective reporting and cherry-picking. Terms sometimes used for data abuse are cooking (selecting data that fit), trimming (smoothing outliers) and forging. Consequences range from retraction to dismissal.

Concept

Authorship credit should follow contribution and responsibility. The ICMJE (Vancouver) criteria require all four: substantial contribution to conception, design, data or analysis; drafting or critically revising the work; final approval of the version published; and agreement to be accountable for the work. Contributors who do not qualify are thanked in the acknowledgements. Ghost authorship hides a real contributor; guest or gift (honorary) authorship credits someone who did not contribute. The corresponding author handles communication with the journal. Publication ethics also forbids duplicate and redundant publication, salami slicing, image manipulation and breach of confidentiality in peer review, and the CRediT taxonomy describes fourteen contributor roles. COPE guides editors on dealing with misconduct.

Concept

A whistleblower reports suspected misconduct in good faith through proper channels; institutions must protect them from retaliation and keep their identity confidential as far as possible. In India the Whistle Blowers Protection Act, 2014 covers disclosures about public servants. For animal research the guiding framework is the 3Rs of Russell and Burch (1959): Replacement (use non-animal methods where possible), Reduction (use the fewest animals) and Refinement (minimise suffering). The Five Freedoms come from the Brambell Report (1965). In India the CPCSEA (Committee for the Purpose of Control and Supervision of Experiments on Animals), set up under the Prevention of Cruelty to Animals Act, 1960, regulates animal experiments through Institutional Animal Ethics Committees, and the ARRIVE guidelines govern reporting of animal studies.

Concept

Research integrity statements set shared standards. The Singapore Statement (2010) names four principles: honesty, accountability, professional courtesy and fairness, and good stewardship; the Montreal Statement (2013) addresses collaboration across borders; the European Code of Conduct for Research Integrity lists reliability, honesty, respect and accountability. Open-science practices support integrity: sharing data and materials, pre-registration, registered reports and replication. Publication bias (the file drawer problem, a term from Robert Rosenthal) is the tendency to publish only significant results. The reproducibility crisis refers to the low rate at which many published findings can be replicated. Replicability means getting similar results with new data; reproducibility means getting the same results from the same data and code.