Netherlands - Generic Functions for data exchange Implementation Guide
0.2.0 - ci-build
Netherlands - Generic Functions for data exchange Implementation Guide
0.2.0 - ci-build
Netherlands - Generic Functions for data exchange Implementation Guide, published by Stichting Nuts. This guide is not an authorized publication; it is the continuous build for version 0.2.0 built by the FHIR (HL7® FHIR® Standard) CI Build. This version is based on the current content of https://github.com/nuts-foundation/nl-generic-functions-ig/ and changes regularly. See the Directory of published versions
This FHIR Implementation Guide defines the specifications for Generic Function Addressing (GFA), a national initiative led by the Dutch Ministry of Health, Welfare and Sport (VWS). GFA aims to establish a standardized, interoperable system for discovering and sharing current (digital) addresses of healthcare providers, enabling reliable and efficient exchange of health data across systems.
In today’s healthcare landscape, organizations rely on both physical and digital addresses for communication and data exchange. However, access to accurate and up-to-date digital addressing information is often inconsistent or unavailable, hindering interoperability and timely access to health information. GFA addresses this challenge by providing a unified framework that ensures all current, valid addresses for data exchange are easily accessible and traceable to authoritative sources.
This guide outlines the technical requirements and architectural principles underlying GFA, with a focus on trust, authenticity, and data integrity. Key design principles include:
By adhering to these principles, this Implementation Guide supports consistent and secure address discovery, fostering improved interoperability within the healthcare ecosystem.
GFA follows the IHE mCSD profile. The mCSD profile provides multiple options for deployment. This guide specifies the choices made for The Netherlands. Most impactful/striking choice are:
Here is a brief overview of the processes that are involved:
Each component, data model, and transaction will be discussed in more detail.
The Administration Client is responsible for managing the registration and maintenance of addressable entities within a healthcare organization. It should be able to create, update, and delete records for healthcare services, organizations, departments, locations, endpoints, or practitioners in the Administration Directory. Addressable entities MUST conform to the Data models.
The Administration Client of the LRZa provides an user interface for healthcare providers to administer their Administration Directory endpoint (url).
The Administration Directory persist all addressable entities of one or more healthcare organizations. The Administration Directory MAY implement these capabilities for an Administration Client to create, update and delete resources. If you've implemented both an Administration Client & Directory, you can also choose to use proprietary formats/APIs/transactions between these components.
The Administration Directory MUST implement these capabilities to publish changes of addressable entities. These changes are consumed by an Update Client.
The performance/availability requirements for an Administration Directory:
The Update Client is responsible for aggregating and synchronizing addressable entity data from multiple Administration Directories. It periodically retrieves updates, including new, modified, or deleted records, and consolidates this information into a Query Directory.
The Update Client uses a FHIR 'history-type' transactions and (optionally) parameter '_since' to get updates from Administration Directories, for example:
GET https://somecareprovider.nl/fhirR4/Organization/_history?_since=2025-02-07T13:28:17.239+02:00&_format=application/fhir+json
Besides using the 'history-type' transaction, the Update Client should be able to do query everything in the Administration Directory using a search transaction. Either for the initial load or periodically for a full reload to fix edge-case scenario's (e.g. Administration Directory backup restores).
During consolidation, multiple Administration Directories may have overlapping or conflicting entities. An Update Client MUST obey these guidelines:
identifier of system http://fhir.nl/fhir/NamingSystem/ura (URA), it's name and status. When the Healthcare provider's Administration Directory also provides a name or status value (for an Organization-instance with a URA-identifier), these values should be ignored. Other elements from the Healthcare provider's Administration Directory should be added. This way, a Healthcare Provider can add an alias or endpoint using it's own Administration Directory.After consolidation, the Update Client writes the updates to a Query Directory. For each instance, the meta.source element is populated with the url of that instance at the (source) Administration Directory. The Update Client MAY use the same interactions a Administration Client uses to register entities in an Administration Directory.
The Query Directory persist all addressable entities it receives from the Update Client. The Query Directory MAY implement these capabilities for an Update Client to create, update and delete resources. If you've implemented both an Update Client & Query Directory, you can also choose to use proprietary formats/APIs/transactions between these components. Due to the consolidation process of the Update Client, not all (intermediate) changes are replicated between Administration Directories and Query Directory
The Query Directory serves/exposes all addressable entities to one or more Query Clients. The Query Directory MAY implement these capabilities for a Query Client to search and read resources. If you've implemented both an Query Client & Query Directory, you can also choose to use proprietary formats/APIs/transactions between these components.
The Query Client is used to search and retrieve information from the Query Directory, which contains consolidated data from all Administration Directories. It enables practitioners, EHR systems, and other healthcare applications to discover healthcare services, organizations, departments, locations, endpoints, or practitioners across the entire ecosystem. By querying the Query Directory, users can efficiently find up-to-date and authoritative addressable entities for care coordination, referrals, and electronic data exchange.
Within GF Addressing, profiles are used to validate data. They are based on both mCSD-profiles and nl-core-profiles (TODO: use Nictiz nl-core package as soon as dependency-bug is fixed). Ideally, these profiles are merged in the nl-core-profiles in the future.
A brief description of the data models and their profile for this guide:
Organizations are “umbrella” entities; these may be considered the administrative bodies under whose auspices care services are provided. An (top-level)Organization-instance has a URA identifier, type, and name and may have additional attributes like the Endpoints it uses. Departments of an institution, or other administrative units, may be represented as child Organizations of a parent Organization. The NL-GF-Organization profile contains no extra constraints on top of mCSD & nl-core profiles.
An Organization may be reachable for electronic data exchange through electronic Endpoint(s). An Endpoint may be a FHIR server, an DICOM web services, or some other mechanism. If an Organization does not have an Endpoint, it may still be reachable via an Endpoint at its parent Organization or an affiliated Organization. The NL-GF-Endpoints profile has an extra value set constraint on .payloadType (source: ADR-8) and an additional extension 'replacedBy'. This can be used to point to a new endpoint if the old one has changed.
Changing endpoints and the continued integrity of references
Healthcare records (e.g. conditions, observations, or procedures) will contain links (references) to addressable entities. Some entities may be referenced by an identifier (e.g. a URA or DEZI-number), but most references will use either a local ID (e.g. Patient/880e50a3) or URL (https://somecareprovider.nl/fhirR4/Patient/880e50a3). The (local) IDs and (remote) URLs are definitive and widely supported in the FHIR ecosystem. Identifiers may be harder to resolve, may expose sensitive data (like a Citizen number; Dutch BSN) and can be ambiguous (multiple instances carrying the same identifier). IDs and URLs are easy in use and (should be) resolvable, but they may break over time which leads to broken references and unresolvable medical records. Try to use endpoints that can remain stable for long periods (5-10 years) and use universally unique IDs (UUIDs) in stead of e.g. incrementing numeric values. If the Endpoint.address (the base part of URLs) must be changed, use the
statusandreplacedByextension to properly redirect and fix broken links, ensuring continued integrity of references. Don't delete Endpoint-instances.For example, care provider 'CP1' uses an EHR from software vendor 'SV1'. This vendor SV1 uses endpoint-url 'https://sv1/fhirR4' for every customer (care provider). Now if CP1 wants to switch to a new vendor, this endpoint-url may risk lots of broken references in many EHR-systems. A better option would be url 'https://sv1/cp1/fhirR4' where a change can be applied to all references in a system OR url 'https://cp1/fhirR4' that may prevent a change altogether (until FHIR R4 is outdated…).
Healthcare services are used to publish which (medical) services are provided by an (child) Organization. Examples include surgical services, antenatal care services, or primary care services. These services in HealthcareService.type can be extended by references to specific ActivityDefinitions and PlanDefinitions that are supported. The combination of a Healthcare Service offered at a Location may have specific attributes including contact person, hours of operation, etc.
The NL-GF-HealthcareService profile contains a value set constraint on .type and .specialty and an extension on .type to refer to Activity/PlanDefinitions.
Locations are physical places where care can be delivered such as buildings (NL: Vestiging), wards, rooms, or vehicles. A Location may have geographic attributes (address, geocode), attributes regarding its hours of operation, etc. Each Location is related to one (child) Organization. A location may have a hierarchical relationship with other locations (e.g. building > floor > room). The NL-GF-Location profile has no extra constraints on top of mCSD & nl-core profiles.
PractitionerRole resources are used to define the specific roles, specialties, and responsibilities that a Practitioner holds within an Organization. PractitionerRole enables precise modeling of relationships between practitioners and organizations, supporting scenarios like assigning practitioners to departments, specifying their roles (e.g., surgeon, nurse), and linking them to particular healthcare services or locations. A PractitionerRole may have contact details for phone, mail, or direct messaging. The information in a PractitionerRole-instance is equivalent to the info on a Dutch 'UZI-pas'. The NL-GF-PractitionerRole profile: no extra constraints on top of mCSD & nl-core profiles.
This resourcetype is out-of-scope for this IG-version Practitioner is a health worker such as physician, nurse, pharmacist, community health worker, district health manager, etc. Practitioners have a name and may have qualifications (like in the Dutch BIG-register). The registration (Administration Directory) of Practitioners may be assigned to the Dutch BIG-register or similar organizations. The NL-GF-Practitioner profile: no extra constraints on top of mCSD & nl-core profiles.
This resourcetype is out-of-scope for this IG-version OrganizationAffiliation resources are used to represent relationships between organizations, such as a software vendor managing the Endpoint that is used by a care provider. It could also be used the represent multiple care providers working together under some agreement (e.g. in a region). The NL-GF-OrganizationAffiliation profile has no extra constraints on top of mCSD & nl-core profiles.
An overview of the most common elements and relations between data models:
The service provider of an Administration Directory may choose whether to require mTLS certificates. Certificate Authority PKI-O should be trusted. For cross-border data exchange using mTLS, support for additional CA's is required. If mTLS is not used, Administration Directory endpoints must be unconditionally available to everyone.
The patient, Vera Brooks, consults with her physician who recommends surgery. The physician can assist the patient in finding a suitable care provider, taking into consideration the location and specialty for orthopedic surgeons.
Dr. West just created a referral (for patient Vera Brooks from use case #1). The EHR has to notify Hospital East and the Orthopedic department of this referral. This may include some recurring requests: