Fundamental Ethical Principles and Swiss Law

Why This Module Is the Load-Bearing Wall

The hardest decisions in clinical research are not the ones where the rules are unclear. They are the ones where two legitimate values pull in opposite directions. The consent form is signed, but you suspect the participant did not really grasp what they agreed to — autonomy against beneficence. The protocol needs a placebo arm to give a clean answer, but the condition is serious — scientific rigour against non-maleficence. A disadvantaged population can be enrolled quickly, but the resulting treatment will be priced beyond their reach — recruitment against justice.

A framework that only tells you "follow the rules" is no help here, because both options follow some rule. What you need is a way to reason — and behind the reasoning, a body of law that turns the reasoning into concrete, enforceable answers about who may run which study and who has to approve it first.

Switzerland has built exactly that. The four ethical principles sit underneath a three-part legal framework — the Human Research Act and its two main ordinances — that decides how risky a study is, who authorises it, and how its data must be protected. This module is where those principles and that law are taught in full, because almost every other obligation in the course is downstream of them.

After this module, you can:

• Apply the four ethical principles to resolve real conflicts between competing legitimate interests
• Judge whether a study's science, equipoise and risk–benefit balance are sound — and what extra protection a vulnerable population needs
• Map the Swiss research-law framework — HRA, KlinV, HRO — and locate where ICH-GCP and recognised standards fit
• Categorise a study (A/B/C) and route it to the ethics committee, Swissmedic, or both
• Apply the nDSG and HRA data-protection rules to research data and samples

Here is the shape of what follows. The module runs in two halves. The first is the ethics: the four principles, and then the working concepts that turn them into protections you can actually apply. The second — the larger half — is the law that makes those protections enforceable: the framework of Swiss research law, how a study is sorted by risk, who has to authorise it, and how its data is protected.

The Four Principles

The four principles — autonomy, beneficence, non-maleficence, and justice — are not abstract philosophy. They are the framework for working through the tensions the intro opened with.

The four come from Beauchamp and Childress's Principles of Biomedical Ethics (1979). They built on the Belmont Report of the same year — the response to the Tuskegee study (Module 2 tells that story in full), which named three principles: respect for persons, beneficence, and justice. Beauchamp and Childress drew autonomy out of respect for persons and split non-maleficence off from beneficence, giving the four that the SwissEthics curriculum names and that Swiss law encodes.

Swiss law states the master principle that orders the other four. When the interests of science and society collide with the interests of the individual participant, the individual wins. HRA

We take them one at a time, in the order they tend to collide on a real study: autonomy first, then beneficence and its mirror-image non-maleficence as a pair, and justice last.

Autonomy and Respect for Persons

Autonomy is the capacity of a person to make an informed, voluntary decision about their own life. In research, respecting autonomy means treating participants as agents who can understand information, weigh it, and decide — not as objects of study.

The Declaration of Helsinki grounds its entire consent framework in this principle: participation by people capable of giving informed consent must be voluntary, and no such person may be enrolled in research unless they freely agree. Declaration of Helsinki

Respect for persons has two limbs, and the second is the one most often forgotten:

• Respect autonomous individuals — give them the information they need, confirm they understand it, and honour their decision.
• Protect individuals with diminished autonomy — children, adults with impaired decision-making capacity, and others with reduced ability to decide get additional safeguards. The answer is not to exclude them, but to enrol them with protections.

The mechanics of how you obtain and document consent are Module 7's subject. What matters here is the principle: autonomy is why consent exists at all, and the Human Research Act turns it into a hard requirement and a set of special provisions for people who cannot consent for themselves. HRA

Beneficence and Non-Maleficence

These two are usually said in the same breath, but they are distinct duties.

Beneficence is the obligation to actively promote participants' well-being and to maximise the potential benefit of the research. In design terms, it means running studies that are likely to produce genuinely useful knowledge, and giving participants the best feasible monitoring and care while they are in your study.

Non-maleficence is the obligation to avoid causing harm. It does not mean avoiding all risk — research by definition carries uncertainty. It means minimising risk relative to the expected benefit, and never exposing participants to risk that the importance of the question cannot justify. The risk–benefit assessment, which is where this principle becomes operational, has its own section below.

Justice

Last of the four, justice concerns the fair distribution of the burdens and benefits of research. It operates at two levels.

• Individual justice — one participant should not carry burdens for the benefit of others without proper justification and consent.
• Social justice — the selection of participants should not load research risks onto disadvantaged or vulnerable groups while the benefits flow mainly to the better-off.

Tuskegee (Module 2) is the textbook justice violation: a disadvantaged population bore the entire burden of the research — including the deliberate withholding of treatment — while the knowledge was meant to benefit everyone else. Swiss law answers this directly with a non-discrimination rule: no group may be disproportionately included in or excluded from research without good reason. HRA

That cuts both ways, and the second direction surprises people. Systematically excluding a population — say, everyone over 75 — to make a trial simpler is itself a justice problem when that population carries the disease burden, because the results will not generalise to the patients who actually have the condition.

From Principle to Protection

Principles are easy to nod along to and hard to act on. What turns "do no harm" into something you actually do on a Tuesday morning is a handful of working concepts — five of them, in the order they come here: scientific accuracy, equipoise, the risk–benefit assessment, the handling of vulnerable populations, and confidentiality in its wider context. Swiss law makes each one a legal requirement, not merely a virtue.

Scientific Accuracy and Equipoise

A scientifically weak study is an ethical failure, not just a methodological one. If a study cannot answer its question — too small, wrong endpoint, biased design — then any risk or burden it imposes on participants is unjustified, because nothing useful comes out the other end. This is beneficence and non-maleficence working together.

Swiss law makes scientific quality a precondition, not an afterthought. Research may only be carried out if it addresses a scientifically relevant topic, and the sponsor and investigator must define a question grounded in the current state of knowledge, use appropriate methodology, and secure the necessary resources and infrastructure. HRA KlinV

Equipoise is the ethical condition that justifies a comparison in the first place. It means there is genuine, honest uncertainty in the expert community about which arm of the study is better. If you already know one treatment is superior, randomising patients to the inferior arm cannot be justified — you would be knowingly disadvantaging half your participants.

Equipoise is also dynamic. If evidence accumulates during the trial — from your study or from outside it — that one arm is clearly better, equipoise has collapsed, and continuing as before is no longer defensible. That is the same logic that drives the risk–benefit reassessment below.

The Risk–Benefit Assessment

This is where non-maleficence becomes a calculation you actually perform. Swiss law states the test plainly: in every project the risks and burdens to participants must be minimised as far as possible, and the likely risks and burdens must not be disproportionate to the expected benefit. HRA

Risks in clinical research include:

• Direct physical risk from the intervention (adverse drug reactions, device failures)
• Risk from study procedures (blood draws, biopsies, extra imaging)
• Psychological risk (distress, uncertainty about one's prognosis)
• Social risk (a confidentiality breach affecting insurance or employment)

Benefits include:

• Direct benefit to the participant (a potentially better treatment, closer monitoring)
• Indirect benefit (knowledge that may help future patients)

The assessment must be honest about both columns. Overstating benefit ("this drug may cure you") or understating risk ("it's just an observational study") is both a consent failure and a distortion of the ethical calculation.

A recurring trap is the procedure that serves the science but not the participant. A biopsy added purely for mechanistic data, or a blood-draw schedule heavier than the endpoints require, has to be justified by the specific scientific value of that data — "minimise as far as possible" is a legal instruction, not a slogan. HRA

Whose job is the reassessment? Everyone's, at their own level. Any site team member who sees a safety signal must escalate it; the investigator must act on it and report findings that could affect participant safety or the risk–benefit balance to the sponsor; ultimate responsibility for the site's response rests with the principal investigator. ICH E6(R3)

Vulnerable Populations

The fourth working concept is about people, not procedures. Some people are at greater risk of being wronged or harmed as research participants — because of a fixed factor (age, impaired capacity) or a contextual one (dependence, deprivation, an emergency). The ethical response is not exclusion, which would deny them research that addresses their health needs, but enrolment with proportionate extra safeguards.

Swiss law builds this in two layers. First, subsidiarity: a project involving particularly vulnerable persons may only run if equivalent findings cannot be obtained any other way. HRA

Second, a tiered set of conditions for people who cannot fully consent — children, adolescents, and adults lacking capacity. The pattern is consistent: the lower the participant's capacity and the smaller the direct benefit to them, the higher the bar. Research with no direct benefit to a person who lacks capacity is only permissible if it carries no more than minimal risk and burden and promises substantial findings for others in the same situation. HRA

The detailed consent process for these groups — who signs, the role of a legal representative or trusted person, re-consent on regaining capacity — is Module 7's territory. Here the point is the principle and the legal architecture: vulnerability triggers stronger protection, written into specific articles, not investigator discretion.

Confidentiality, Privacy, and Local Context

Confidentiality is the last of the five working concepts, and it flows from the respect-for-persons principle; the detailed legal regime that enforces it is the data-protection section at the end of this module. A participant who entrusts you with sensitive health information is owed protection of that information, and a breach is a real harm — it can affect someone's insurance, employment, or standing in their community.

Two further factors belong here because the Declaration of Helsinki and SwissEthics expect investigators to weigh them. Societal, religious and cultural factors can change what a given risk or procedure means to a participant, and what counts as respectful information and consent. Local conditions matter too: a study that is appropriate in one setting may be exploitative in another if the population cannot access the eventual benefit. The classic example is a trial run in a low-income population that carries all the risk while the resulting drug is priced out of their reach afterwards — the justice principle applied to where and among whom you run the study.

The Architecture of Swiss Research Law

Principles tell you what is right. They do not tell you which form to file, or which authority signs it off. That is the law's job — and Swiss research law is built in layers, so the first move with any study is working out which layer it lands in, because that decides its rules, its categorisation, and who approves it.

At the top sits the Human Research Act (HRA) — in German the Humanforschungsgesetz, SR 810.30. It is the umbrella statute for all human research concerning disease and the structure and function of the human body, covering research on persons, on deceased persons, on embryos and foetuses, on biological material, and on health-related personal data. HRA

So the first fork — is this even under the HRA? — turns on two questions. Is it research: a method-driven search for generalisable knowledge? Routine clinical care and pure quality-assurance or audit are not, so they fall outside. And does it touch one of those five things — a living or deceased person, an embryo or foetus, biological material, or health-related personal data? The decisive edge is identifiability: fully anonymised data or material, which cannot be traced back to a person, is outside the Act (the anonymisation line the data-protection section returns to), while coded data, still linked to someone by a key, stays inside. HRA

The HRA sets the principles and the high-level rules, then delegates the operational detail to ordinances. Two of those ordinances do most of the day-to-day work, and the split between them is the first thing to get right:

• The Clinical Trials Ordinance (KlinV / ClinO), SR 810.305, governs clinical trials — research where participants are prospectively assigned to a health-related intervention to study its effects.
• The Human Research Ordinance (HRO), SR 810.301, governs all other research projects — observational studies, and research re-using already-collected data or biological material.

A separate ordinance, the Clinical Trials with Medical Devices Ordinance (ClinO-MD, SR 810.306), carves device trials out of the KlinV; device specifics are covered in Module 10. ClinO-MD

So the routing for any new study runs in two steps: first the scope test above — is it under the HRA at all? — then, if yes, is it a clinical trial (KlinV) or another research project (HRO)? That second fork determines almost everything that follows.

That fork is worth naming, because its logic recurs. A study is interventional — and falls under the KlinV — when the researcher assigns the exposure: who receives which treatment, at what dose, on what schedule. That deliberate assignment is what lets a trial support a claim about cause, where an observational study — the HRO's domain — only watches and records what happens in ordinary care, and so can show associations without, on its own, proving that one thing caused another.

The framework also adds protections that are pure participant-welfare law. Anyone running a project involving persons is liable for damage suffered in connection with it, and that liability must be covered — in practice, by trial-specific insurance that has to be in place before the trial begins and is checked by the ethics committee. HRA

Risk-Based Categorisation

It would be strange to make a trial of a drug licensed for thirty years, used exactly as its label says, clear the same bar as a molecule no human has ever swallowed. Swiss law agrees. It sorts every study by risk into a category — set in the ordinances, not the HRA — and that category decides how heavy the gate is. This is the proportionality idea, the GCP "risk-based approach" from Module 8, made concrete.

The categories differ by study type, so read them off the right list.

For clinical trials of medicinal products under the KlinV, there are three categories. If a study would fit more than one, it takes the highest. KlinV

The other study types use simpler schemes, but on the same logic — lower category for standard-of-care or minimal-risk work, higher for novel or riskier work:

• Transplantation trials (organs, tissues, cells): Category A if the transplantation under investigation is recognised as standard, Category C if it is not; trials of embryonic or foetal tissue are always C. KlinV
• Other clinical trials (interventions not involving a therapeutic product): Category A if the intervention carries only minimal risks and burdens or is recognised as standard, Category B otherwise. KlinV
• Research projects under the HRO (observational studies, further use of data or material): Category A if the planned sampling or data collection carries only minimal risks and burdens, Category B if more than minimal. HRO

Notice the thread running through every scheme: "recognised as standard in guidelines prepared in accordance with internationally accepted quality criteria" is what pulls a study down into the lowest category. That phrase is the categorisation face of the recognised-standards rule from the previous section — standard-of-care, done to recognised guidelines, is treated as low risk.

Who Authorises What: the Ethics Committee and Swissmedic

The categorisation feeds straight into the question that trips up the most people: before a trial starts, who has to say yes? In Switzerland the answer is split between two kinds of body, and the split depends on the category. (The ethics committee here is the independent cantonal body whose membership, independence and review process are Module 5's subject; this section is about its place in the authorisation map, not how it reviews.)

The ethics committee authorises every clinical trial — all categories, no exceptions. No committee sign-off, no trial: that bright line holds whatever the category. The investigator, or the sponsor on their behalf, submits the dossier, and the committee normally decides within 30 days.

What does it actually check? Quite a lot at once: the requested categorisation, the science and the risk–benefit ratio, the consent materials, the safeguards for vulnerable participants, the insurance, and whether the investigator and site are up to the job. HRA KlinV

Swissmedic — the Swiss agency for therapeutic products — only gets involved for the riskier drug trials: Categories B and C. Category A drug trials are exempt from Swissmedic approval by law. And when Swissmedic does review, it looks at something the committee does not: not the ethics, but the product. Is the medicinal product safe and well made — its pharmacology, toxicology, manufacturing, GMP (Good Manufacturing Practice, the manufacturing-quality standard)? Swissmedic decides within 30 days, the same clock the committee runs on. KlinV HMG / TPA

So the two bodies are not duplicates. The ethics committee is the guardian of the participants and of the ethical, legal and scientific integrity of the trial; Swissmedic is the product regulator that vets the medicinal product's safety and quality. They run on parallel tracks — applications may be filed simultaneously, and the ethics committee and Swissmedic coordinate their assessments and keep each other informed. KlinV

Two edges complete the map. For transplantation trials, the second authority is not Swissmedic but the Federal Office of Public Health (FOPH), which authorises Category C transplantation trials. KlinV

For medical-device trials, authorisation runs under ClinO-MD with Swissmedic as competent authority (Module 10).

For a multicentre trial spread across cantons, one lead ethics committee — the committee where the coordinating investigator is based — gives the authorisation, after the local committees report on local conditions; it decides within 45 days. The mechanics of the lead committee are introduced in Modules 3 and 5; what matters here is that it is still the ethics committee arm of the map, not a different authority. HRA KlinV

Data Protection in Research

A leaked diagnosis can cost someone a job, an insurance policy, or their standing in their community — which is why health data sits in the most protected tier of the law. Confidentiality — the working concept from earlier, which flows from respect for persons — is enforced through a specific legal regime, and two instruments interlock to do it: the general data-protection law, and the HRA's own research-data provisions. This section is the legal framing; the operational mechanics of coding, securing, and archiving data live in Module 12.

The general law is the Federal Act on Data Protection (nDSG / FADP), SR 235.1, revised and in force since September 2023. It protects the personal data of natural persons and applies to processing by both private persons and federal bodies. Health data is explicitly classed as sensitive personal data, alongside genetic and biometric data — the category that gets the strongest protection. nDSG

The nDSG's core principles are the ones to internalise, because every data decision in a trial has to satisfy them:

• Lawfulness, good faith, proportionality — process data only as far as the purpose genuinely needs.
• Purpose limitation — collect data for a specific, recognisable purpose and use it only in ways compatible with that purpose.
• Data minimisation and deletion — destroy or anonymise data once it is no longer needed for that purpose.
• Explicit consent for sensitive data — processing sensitive personal data requires the data subject's explicit consent. nDSG

The nDSG does cut research some slack — but with strings attached. You may process data for purposes not tied to specific people, research included. In return, three conditions follow. Anonymise the data as soon as the purpose allows. Publish results so that no individual can be picked out. And handing sensitive data to a third party is not a free pass: only fully anonymised data — where no one can re-identify the person — sits outside the regime. Coded data does not qualify, because the key still exists and the person is therefore still identifiable, so disclosing sensitive data in identifiable (including coded) form needs a proper legal basis — consent or the further-use rules below — plus the data-transfer agreements covered in Module 12. nDSG

On top of the general law, the HRA sets the rules specific to research re-using biological material and health data. Further use of identifiable material or data generally requires consent (or at least a non-objection for coded data); the threshold rises with how identifiable the data is. HRA

The single most reliable way to step outside the whole regime is anonymisation: the HRA does not apply to anonymously collected or anonymised health data, and the nDSG's principles fall away once data can no longer be traced to a person. That is why "coded versus anonymised" is such a consequential distinction — and why getting it precisely right is treated in depth in Module 12. HRA

Module Summary

The four principles are not a checklist — they are a set of lenses, and the hardest cases ask you to look through all of them at once. Swiss law then takes the reasoning and makes it enforceable.

You can now:

• Apply the four principles — autonomy, beneficence, non-maleficence, and justice — to work through real conflicts, with the primacy of the individual as your tie-breaker.
• Judge a study's science, equipoise and risk–benefit balance, and recognise when a vulnerable population needs subsidiarity and tiered safeguards rather than exclusion.
• Map the Swiss framework — the HRA umbrella over the KlinV (clinical trials) and the HRO (other research) — and locate where ICH-GCP and recognised standards enter the law.
• Categorise a study A/B/C and route it correctly: the ethics committee for every trial, Swissmedic only for Category B and C medicinal-product trials.
• Apply the data-protection rules of the nDSG and the HRA to research data and samples, and know why anonymisation changes everything.