Patient confidentiality is the duty of doctors and other health professionals to keep information about their patients private. It is one of the oldest rules in medicine, reaching back to the Hippocratic Oath, and today it rests on professional ethics, civil law and, in some countries, criminal law. It exists so patients can speak openly: without trust that what they say stays private, they hold back, and care suffers.
An everyday example
A GP, Dr Clarke, sees a patient, Mr Patel, about anxiety and a recent prescription change. Patient confidentiality means:
- Dr Clarke may share what is needed with the team directly involved in Mr Patel's care, such as the practice nurse or a specialist he is referred to.
- Dr Clarke may not tell Mr Patel's employer, his family or an insurer about the consultation without his consent or another legal basis.
- The practice must keep the notes secure, including on any software it uses.
So far, so familiar. The less familiar part comes later, in what happens to the record once his name is taken off it.
The details, simply
What it covers
Confidentiality covers information a patient shares, or a professional learns, in the course of care: diagnoses, test results, medication, what was said in the consultation, and often the fact that the person is a patient at all. It applies to doctors, nurses, therapists, pharmacists, dentists and the staff who work with them.
Where it comes from
The principle is shared across Europe, the UK and the US, but the legal roots differ:
- United Kingdom. There is a common-law duty of confidentiality: information given in confidence must not be disclosed without consent or another lawful justification. The General Medical Council's guidance on confidentiality sets out what doctors must do, and breaches can lead to fitness-to-practise proceedings. The Caldicott Principles guide how health and care organisations handle identifiable information.
- Germany. Breaking medical confidentiality is a criminal offence under § 203 StGB. Doctors, dentists, pharmacists, psychotherapists and others who disclose a patient's secret without authorisation face up to one year in prison or a fine, and up to two years if they do it for payment or to cause harm.
- United States. The HIPAA Privacy Rule protects "protected health information" held by covered entities (health plans, healthcare clearinghouses and most healthcare providers) and their business associates, who must sign a business associate agreement. Crucially, data that has been de-identified under HIPAA, either by removing 18 listed identifiers (the Safe Harbor method) or by an expert determination, is no longer protected health information, and the Privacy Rule no longer restricts its use.
- France. Medical secrecy (secret médical) is protected by the Public Health Code and is a criminal offence under Art. 226-13 of the Penal Code: up to one year in prison and a €15,000 fine.
- Italy. Disclosing a professional secret without just cause, where harm may result, is a crime under Art. 622 of the Penal Code (rivelazione di segreto professionale), punishable by up to one year in prison or a fine, and prosecuted on the victim's complaint.
- Poland. Doctors are bound by medical secrecy (tajemnica lekarska) under Art. 40 of the Act on the Professions of Physician and Dentist, and disclosing information learned through one's profession is a crime under Art. 266 of the Penal Code, punishable by up to two years in prison.
- Across the EU. GDPR adds a data protection layer on top. Health data is special category data under Art. 9, and processing it for care is allowed only by or under the responsibility of a professional bound by secrecy. See special category data.
The usual exceptions
Confidentiality is strong but not absolute. Typical exceptions are:
- Consent from the patient, explicit or, for direct care, often implied.
- A legal requirement, such as certain notifiable diseases or a court order.
- Public interest, for example to prevent serious harm to others. This is a high bar.
Each country draws these lines a little differently, and professional bodies publish detailed guidance.
This is general information, not legal advice.
What patient confidentiality does not protect
Confidentiality protects the identifiable patient. Remove the identity and, in most legal systems, the duty no longer bites. The US writes this into the law itself: under HIPAA, properly de-identified health data is no longer protected health information at all.
- In England, the Court of Appeal decided in 1999 (R v Department of Health, ex parte Source Informatics) that pharmacists passing anonymised prescription information to a data company did not breach confidence, because the patients' identities were not disclosed.
- Under GDPR, Recital 26 says data protection does not apply to anonymous information.
- In Germany, § 203 StGB protects secrets about an identifiable person. Data that can no longer be linked to anyone is, legally, no longer that person's secret.
The result is a large and lawful market. De-identified prescriptions, diagnoses, lab values and treatment pathways are pooled, licensed and sold to analytics firms, insurers, pharmaceutical companies and AI developers. What leaves is what has value: the clinical detail. Only the name stays behind.
Two more risks sit on top:
- Weak anonymisation. Health records re-identify easily, because a rare condition, an age and a postcode can be enough. See pseudonymisation vs anonymisation.
- Licences in terms of service. Software that handles patient data may have terms allowing the provider to use content for "aggregated" or "de-identified" purposes. That is often how records enter the market in the first place.
We show this path, record by record, on the data trail, and the legal mechanism with sources on legal to sell.
What this means when you use AI tools
Doctors increasingly use AI to summarise letters, draft referrals or write up consultations. Before you use any tool with patient information:
- Check the legal basis for disclosure. Patient consent or another legal permission is what makes sharing lawful, and that responsibility stays with you, not the vendor.
- Check the provider's role. Is it your processor under a data processing agreement, acting only on your instructions?
- Read the licence. Do the terms allow the provider to anonymise, aggregate or otherwise reuse what you upload?
- Check training and retention, for the tool and every AI model provider behind it.
Wysor is built for this work. We act only as your processor under Art. 28 GDPR and take no licence to reuse your content. We never train on it and never sell it, anonymised or not. Those commitments are in the contract and bind anyone who could ever own Wysor. Data is encrypted in transit and at rest, hosted in the EU, and every connected AI model provider runs with zero data retention. For clinical use cases, see our guides to medical AI and the AI scribe.
FAQ
What is doctor-patient confidentiality? It is the duty of a doctor to keep what they learn about a patient private, sharing it only with consent, with the care team where needed, or where the law requires or permits it. It applies to other health professionals and their staff too.
When can a doctor break patient confidentiality? Typically with the patient's consent, when the law requires disclosure, or when there is a serious risk of harm to others that outweighs the duty. The exact rules depend on the country and the professional guidance.
Is breaking patient confidentiality a crime? It depends on the country. In Germany it is a criminal offence under § 203 StGB. In the UK it is mainly a matter of civil law and professional regulation, and serious breaches can end a doctor's registration.
Does patient confidentiality apply to anonymised data? Generally not. Confidentiality and data protection law protect identifiable information. Properly anonymised data falls outside both, which is why de-identified health data can be legally sold.
Can I use AI with patient data? It can be done, but you need a lawful basis for disclosure, such as consent, and a provider that acts only as your processor, does not reuse or train on the data, and keeps it secure. Check the terms, not only the marketing page.
Further reading
- What is special category data? GDPR Article 9 explained
- Pseudonymisation vs anonymisation
- Private AI for medicine
- The data trail
Questions about using AI with patient data? Reach us at [email protected] or through the in-app contact form.
