Health sector: master the RGPD in your EDS

Setting up a health data warehouse requires strict compliance with the RGPD and the CNIL standard, based on clear governance, comprehensive documentation and high security measures. The DPO is the guarantor, ensuring transparency and compliance at each stage of the project.

By
Laurent Chollat-Namy
1
Min
Share this article
Doctor with a tablet and data

GDPR and Data Warehouse

Health data warehouses have become strategic devices for healthcare institutions, but their implementation requires strict compliance with the GDPR.

Between legal basis, governance, security and reuse, this type of treatment raises complex technical and legal questions.

In this article, we explore the requirements for creating a compliant data warehouse, the mistakes to avoid, and the levers to activate to ensure security and transparency.

An operational guide for DPOs, CISOs, lawyers and data controllers.

DPO notice

The health data warehouse should be perceived as a massive storage of pseudonymized data whose reuse must then be justified by research, study and management projects that often themselves rely on foundations of public interest.

In the context of the public interest, the EDS must respond to the CNIL standard on EDS.

  • A health data warehouse is a structured process subject to the requirements of the GDPR.
  • The CNIL standard allows a declaration of conformity for projects of public interest
  • AIPD, ethical governance and documentation are indispensable
  • Pseudonymization is not enough to rule out the GDPR: the data remains identifiable
  • The DPO must be integrated from the design stage to secure the entire project

Adequacy has a Health module, where the CNIL Health Data Warehouse repository is registered and usable.

Understanding health data warehouses in light of the GDPR

Health data warehouses (EDS) have become strategic tools for health institutions wishing to value their data for the purposes of management, research or optimization of care. However, their implementation requires a rigorous legal framework, in particular with regard to the General Data Protection Regulation (GDPR).

Definition of a health data warehouse

A health data warehouse is a structured processing of personal data, implemented by a health institution or a hospital group, in order to allow the reuse of data collected during care¹. This data may relate to:

  • Strategic management
  • Improving the quality of care
  • Medical decision support
  • Feasibility studies
  • Research, under conditions

This treatment is distinguished from a traditional database by its reuse objective, its volume, its structuring for analytical purposes and its strengthened governance framework.

Health data: a category of “sensitive” data

The GDPR (article 9) considers health data to be sensitive data, subject to a prohibition of processing in principle, except for clearly defined exceptions². Because of the sensitivity of the information they aggregate (pathologies, treatments, hospital stays, medical procedures, etc.), data warehouses are therefore subject to a high level of vigilance.

Specific GDPR requirements include:

  • A solid legal basis
  • Appropriate technical and organizational security measures
  • A strict limitation of purposes
  • Transparent governance

Legal basis: the three applicable regimes

Data processing under an EDS may be based on several legal bases:

Base légale (RGPD) Finalité du traitement Exemple d'usage Régime CNIL
Article 6-1.e + 9-2.g Mission d'intérêt public Entrepôt hospitalier pour pilotage Déclaration de conformité au référentiel
Article 6-1.a + 9-2.a Consentement explicite Projet de recherche hors mission publique Pas de déclaration, mais respect du RGPD requis
Article 6-1.c + 9-2.i Obligation légale / santé publique Surveillance épidémiologique Nécessite souvent autorisation spécifique
Base légale (RGPD) Finalité du traitement Exemple d'usage Régime CNIL
Article 6-1.e + 9-2.g Mission d'intérêt public Entrepôt hospitalier pour pilotage Déclaration de conformité au référentiel
Base légale (RGPD) Finalité du traitement Exemple d'usage Régime CNIL
Article 6-1.a + 9-2.a Consentement explicite Projet de recherche hors mission publique Pas de déclaration, mais respect du RGPD requis
Base légale (RGPD) Finalité du traitement Exemple d'usage Régime CNIL
Article 6-1.c + 9-2.i Obligation légale / santé publique Surveillance épidémiologique Nécessite souvent autorisation spécifique

Important note: The most frequently used basis in EDS is the public interest mission, which allows data to be processed without individual consent, as long as the processing is supervised by a legal or regulatory system (CNIL reference system, authorization).

Purpose and transparency: elements to be documented

Each purpose of an EDS must be:

  • Legitimate
  • Determined
  • Explicit
  • Compatible with the public interest

It is imperative to clearly inform the persons concerned about this via an accessible and understandable privacy policy.

Formal prohibitions: Treatment may under no circumstances be used for commercial purposes, for the promotion of drugs or for the adjustment of insurance guarantees (in accordance with article L.5311-1 of the CSP and the restrictions of the CNIL standard).

To remember

The RGPD requires a detailed reading of the treatments implemented via an EDS. The choice of the legal basis, the definition of the purpose, the nature of the data, and governance are all determining factors in ensuring compliance.

Implement a GDPR-compliant data warehouse

The creation and operation of a health data warehouse (EDS) require a structured approach, in accordance with the requirements of the RGPD and the recommendations of the CNIL³. From governance to documentation, each step must ensure the legality, security and traceability of the treatments implemented.

Legal framework: compliance or authorization?

The CNIL proposes a specific framework to frame the creation of an EDS as part of a mission in the public interest4. If the project is in strict compliance with this standard, a declaration of conformity is sufficient. Otherwise, a request for authorization must be sent to the CNIL (article 66 III of the “Informatique et Libertés” law).

Summary table of procedures

Situation du projet Procédure CNIL Délai estimé
Conforme au référentiel CNIL Déclaration de conformité Mise en œuvre immédiate
Hors référentiel (ex. : autre finalité, gouvernance partielle) Demande d'autorisation Délai d'instruction par la CNIL (plusieurs mois)
Situation du projet Procédure CNIL Délai estimé
Conforme au référentiel CNIL Déclaration de conformité Mise en œuvre immédiate
Situation du projet Procédure CNIL Délai estimé
Hors référentiel (ex. : autre finalité, gouvernance partielle) Demande d'autorisation Délai d'instruction par la CNIL (plusieurs mois)

Mandatory documentation

Regardless of the regime, GDPR documentation is a central pillar of compliance. The data controller must keep up to date:

  • A register of processing activities (article 30 GDPR)
  • A privacy impact assessment (AIPD) (article 35 GDPR)
  • A governance file specifying the bodies in charge of steering and validation
  • Security and authorization protocols

Important: These documents must be available at all times in the event of an inspection, and updated in the event of changes in the perimeter of the warehouse.

Recommended governance

Strong governance is essential. The CNIL standard requires the implementation of:

Mandatory governance structure

Instance Rôle Responsabilités
Comité de pilotage Orientations stratégiques Définit les orientations stratégiques et tient un inventaire exhaustif des données
Comité scientifique et éthique Évaluation des demandes Évalue chaque demande de réutilisation des données et valide les projets de recherche
DPO Conformité RGPD Garant de la conformité et conseil méthodologique
DIM Information médicale Expertise technique et validation des aspects médicaux
Instance Rôle Responsabilités
Comité de pilotage Orientations stratégiques Définit les orientations stratégiques et tient un inventaire exhaustif des données
Instance Rôle Responsabilités
Comité scientifique et éthique Évaluation des demandes Évalue chaque demande de réutilisation des données et valide les projets de recherche
Instance Rôle Responsabilités
DPO Conformité RGPD Garant de la conformité et conseil méthodologique
Instance Rôle Responsabilités
DIM Information médicale Expertise technique et validation des aspects médicaux

Requirement: These bodies need to be formalized, and their roles clearly documented.

Lifecycle of the compliance stages

With regard to information to the persons concerned and the need for a transparency portal, read the article from the Aumans Avocats law firm.

Checklist — Launching a warehouse

Critère de conformité Statut
Les finalités sont clairement définies
La base légale est justifiée (intérêt public ou consentement)
Une AIPD a été réalisée
Une gouvernance a été formalisée
Les outils de sécurité sont opérationnels
Une information transparente est mise à disposition
Le registre de traitement est à jour
Une politique de revue régulière est en place
Critère de conformité Statut
Les finalités sont clairement définies
Critère de conformité Statut
La base légale est justifiée (intérêt public ou consentement)
Critère de conformité Statut
Une AIPD a été réalisée
Critère de conformité Statut
Une gouvernance a été formalisée
Critère de conformité Statut
Les outils de sécurité sont opérationnels
Critère de conformité Statut
Une information transparente est mise à disposition
Critère de conformité Statut
Le registre de traitement est à jour
Critère de conformité Statut
Une politique de revue régulière est en place

The key role of the DPO

The DPO is both a guarantor of compliance and methodological support, here are the missions of the DPO within the framework of an EDS:

  • Validates the purposes and the legal basis
  • Coordinates the implementation of the AIPD
  • Ensures that people's rights are respected
  • Advises on minimization and pseudonymization measures

Advice: It is imperative to integrate the DPO from the warehouse design phase.

Data Security, Anonymization and Reuse: Requirements and Limits

The implementation of a health data warehouse involves a high level of requirements in terms of security, risk management and respect for fundamental rights. The GDPR imposes a strict framework on the management of access, the reuse of data and the guarantee of confidentiality, especially when it comes to health data.

Data security: specific obligations

Article 32 of the GDPR requires data controllers to put in place appropriate technical and organizational measures to ensure a level of security adapted to the risks.

Mandatory safety measures in an EDS

Catégorie Mesures techniques Mesures organisationnelles
Comité de pilotage Orientations stratégiques Définit les orientations stratégiques et tient un inventaire exhaustif des données
Comité scientifique et éthique Évaluation des demandes Évalue chaque demande de réutilisation des données et valide les projets de recherche
DIM Information médicale Expertise technique et validation des aspects médicaux
Catégorie Mesures techniques Mesures organisationnelles
Contrôle d'accès Authentification forte Journalisation Principe du moindre privilège et gestion des habilitations
Catégorie Mesures techniques Mesures organisationnelles
Protection des données Chiffrement en transit et au repos et cloisonnement des environnements Séparation production/recherche et traçabilité exhaustive
Catégorie Mesures techniques Mesures organisationnelles
Surveillance Audits réguliers et tests d'intrusion Plan de réponse aux incidents et procédures de notification CNIL

Legal obligation: Establishments must also have an incident response plan, including a notification procedure to the CNIL and, where applicable, to the persons concerned (articles 33 and 34 RGPD).

Examples of recommended safety measures:

  • Double authentication (2FA) for access to warehouse environments
  • Isolated servers in a private network, auditable by logs
  • Logging of extractions with historization
  • Implementing VPN and TLS/SSL encryption
  • Regular penetration tests and update policy

Pseudonymization vs anonymization: two distinct logics

A frequent confusion persists between pseudonymization and anonymization, which are however quite distinct legally:

Technical and legal comparison

Technique Données à caractère personnel ? Objectif Réversibilité Cadre juridique
Pseudonymisation Oui Réduire le risque en masquant l'identité Réversible avec clé de correspondance Soumis au RGPD
Anonymisation Non Supprimer tout lien identifiable Irréversible par définition Hors champ RGPD
Pseudonymisation
Données à caractère personnel ?Oui
ObjectifRéduire le risque en masquant l'identité
RéversibilitéRéversible avec clé de correspondance
Cadre juridiqueSoumis au RGPD
Anonymisation
Données à caractère personnel ?Non
ObjectifSupprimer tout lien identifiable
RéversibilitéIrréversible par définition
Cadre juridiqueHors champ RGPD

Critical point: Pseudonymized data remains personal data within the meaning of the GDPR. They therefore require all legal guarantees (legal basis, security, information, etc.).

On the other hand, truly anonymized data is outside the scope of the GDPR, provided that no re-identification is possible, even by crossing.

For more information, watch our video by Patrick Tiev on this subject.

Secondary reuse: what are the conditions?

The reuse of EDS data, for example for research purposes or medical-economic studies, is possible under certain conditions:

Reuse conditions in accordance with the GDPR

Critères Exigence Validation requise
Compatibilité des finalités Compatible avec celle initialement déclarée (art. 5.1.b RGPD) Analyse de compatibilité
Validation scientifique Procédure de validation scientifique et éthique Passage en comité compétent
Nouvelle finalité Si finalité change ou dépasse le cadre initial Nouvelle AIPD + autorisation CNIL
Méthodologie Reposer sur des méthodologies de référence CNIL Respect des référentiels existants

Special procedure: In the event of a project not covered by the existing framework or methodologies, specific authorization must be requested.

Mistakes to avoid:

  • Use warehouse data for unintended purposes
  • Lack of justification on the scope of the data collected
  • Lack of impact assessment or insufficient documentation
  • Reidentification possible by crossing bases
  • Sharing data with third parties without a clear contractual framework

To go further:

Rigorous documentation of the processing chain, from feeding the warehouse to reuse, is essential. It is a guarantee in the event of an audit by the CNIL, but also a valuable internal management tool for data controllers and DPOs.

Use cases, common mistakes and recommendations for DPOs

DPOs, CISOs and lawyers play a central role in bringing health data warehouses into compliance. Their intervention from the design phase makes it possible to avoid many pitfalls. This section presents concrete examples of best practices, frequent errors and optimization levers for compliant and controlled EDS management.

Case study 1 — A compliant project: the structured hospital

Background: A hospital institution wants to create an EDS to analyze care pathways and assess the relevance of care.

Success Factors:

  • The treatment is based on a mission in the public interest
  • The project complies with the CNIL standard
  • Clear governance is in place (steering + ethics)
  • A comprehensive AIPD impact assessment is carried out
  • Data is pseudonymized, with restricted access
  • The reuse policy is defined and validated by the committee

Result: The project is operational, without a CNIL alert, with total transparency towards patients.

Case study 2 — A non-compliant project: the uncontrolled crossing

Background: An institution retrieves data from several HIS (hospital information systems), without clear documentation of the purposes. Personal data is cross-referenced with an external database for internal research purposes.

Observed failures:

  • Unclear or multiple purposes
  • Absence of an ethics committee
  • No AIPD or formalized legal basis
  • No effective pseudonymization
  • The data is used by several unauthorized services

Result: After an inspection, the CNIL requests the immediate suspension of processing, due to lack of compliance.

Table — Common Mistakes and Corrective Responses

Erreur constatée Risque encouru Réponse recommandée
Absence d'AIPD Sanction CNIL, arrêt du projet Intégrer le DPO en amont, formaliser l'analyse
Données identifiantes non protégées Réidentification, violation de données Appliquer pseudonymisation, accès restreints
Finalité non définie Illégalité du traitement Clarifier et documenter chaque finalité
Gouvernance absente Opacité du traitement Créer comités de pilotage et d'éthique
Réutilisation non cadrée Détournement de données Documenter et valider chaque projet de réutilisation

The DPO, compliance conductor

The Data Protection Officer must be integrated from the warehouse design phase. Its role is transversal and includes:

DPO missions within the framework of an EDS

Phase du projet Actions du DPO Livrables
Conception Accompagnement à la rédaction des documents et validation des bases légales et finalités Documents de conformité
Mise en œuvre Suivi de l'AIPD et sensibilisation des équipes projet AIPD validée
Exploitation Dialogue avec les comités de gouvernance et alerte en cas de dérive Rapports de conformité

Recommendation: It is strongly recommended to include the DPO in the validation circuits for any reuse of data from the EDS.

Practical recommendations

For optimal compliance management:

  • Conduct an annual compliance review with an update of the AIPD
  • Formalize a reuse procedure (request, validation, archiving)
  • Integrate a register specific to the EDS project, separate from the main register
  • Provide continuous training for stakeholders on RGPD issues
  • Follow regulatory changes (e.g.: CNIL directives, AI Act)

By capitalizing on these recommendations, DPOs can truly act as project facilitators, while securing the organization from legal and reputational risks.

FAQ - Frequently asked questions

  • Does a data warehouse always have to be subject to a DPIA?

Yes. Any health data warehouse requires an impact assessment (AIPD), as it involves large-scale processing of sensitive data.

  • Can an EDS be used for commercial projects?

No Reuse for commercial purposes is strictly prohibited. Only public interest or research purposes are authorized.

  • Who is responsible in the event of a data breach in a warehouse?

The data controller is responsible, even if the subcontractor is involved. He must demonstrate that he has put in place appropriate measures.

  • Do patients need their consent to feed an EDS?

No, if the processing is based on a mission in the public interest in accordance with the RGPD and the CNIL standard.

Yes, if the treatment is based on article 9.2.a (explicit consent).

  • What are the differences between an EDS and a traditional database?

An EDS is designed for the structured and secure reuse of health data, with governance, multiple purposes and comprehensive GDPR documentation.

List of EDS registered in France by the CNIL

Sources and references

1. CNIL, “The CNIL adopts a framework for health data warehouses”, cnil.fr

2. Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 (RGPD), Article 9

3. CNIL, “Health data warehouses: the CNIL publishes a 'checklist' for compliance with its framework”, cnil.fr

4. CNIL, “Framework relating to the processing of personal data implemented for the purpose of creating a health data warehouse”, cnil.fr

5. Public Health Code, Article L.5311-1

6. Law No. 78-17 of 6 January 1978 relating to information technology, files and freedoms, Article 66 III

Document updated in September 2025 - Complies with the latest CNIL regulatory developments

The latest news

They have trusted us for years

Discover Adequacy

One of our experts introduces Adequacy to you in a real situation.