Current Build
FHIR Infrastructure Work GroupMaturity Level: 5Ballot Status: Trial Use

The FHIR specification describes a set of resources, and several different frameworks for exchanging resources between different systems. Because of its general nature and wide applicability, the rules made in this specification are generally fairly loose. As a consequence, this specification allows that different applications may not be able to interoperate because of how they use optional features. As a consequence, applications claiming conformance to this specification make the claim in respect of a specific exchange framework, and in regard to specific details about their usage of those frameworks and resource contents.

Applications claim conformance to one (or more) of the following exchange frameworks:

To provide details about specific usage of the frameworks and resource contents, FHIR provides a conformance layer that implementers and national/regional programs can use to provide a computable statement about how the resources and their exchange paradigms are used to solve particular use cases. The conformance layer itself is implemented using the following key resources:

Value Set Defines a set of coded values (see "Using Codes" for more details)
StructureDefinition Makes rules about how a resource (or type) and its data elements are used in a particular context. A structure definition references value sets for the coded elements in a resource
CapabilityStatement A statement of the kinds of resources and operations provided and/or consumed by an application. The Capability Statement references profiles to describe specific use of resources by the application
Implementation Guide A single coherent collection of capability statements, profiles, value set, and documentation describing a set of interoperable applications

The specification also provides a number of tools that can assist with enforcing technical conformance to this base specification and profiles on it.

Conformance with this specification does not provide any guarantee of patient or data safety. However, choosing to not conform to this specification carries additional risk in two ways:

  • FHIR has been subject to a level of review and vetting unlikely to be received by any non-conformant variation; variations may result in introduction of undetected risks
  • FHIR-like solutions (based on FHIR, but not conformant) may set expectations by trading partners which are not met due to the non-conformance of the system and these un-met expectations may also result in risk

Any assertion of conformance to FHIR that does not reference a CapabilityStatement instance has limited utility. System capabilities that cannot be described by a CapabilityStatement cannot be considered FHIR conformant.

This specification uses the conformance verbs SHALL, SHOULD, and MAY as defined in RFC 2119 . Unlike RFC 2119, however, this specification allows that different applications may not be able to interoperate because of how they use optional features. In particular:

  1. SHALL: an absolute requirement for all implementations
  2. SHALL NOT: an absolute prohibition against inclusion for all implementations
  3. SHOULD/SHOULD NOT: A best practice or recommendation to be considered by implementers within the context of their particular implementation; there may be valid reasons to ignore an item, but the full implications must be understood and carefully weighed before choosing a different course
  4. MAY: This is truly optional language for an implementation; can be included or omitted as the implementer decides with no implications

The contents of a resource and the formats used to represent resources SHALL conform to the rules described in this specification, as defined in the narrative of the specification, and as controlled by the conformance properties defined below.

Data elements defined in resources and data types have 3 properties that are directly related to conformance: Cardinality, Is-Modifier, and Must-Support. These interact to place conformance requirements on implementations.

All attributes defined in FHIR have cardinality as part of their definition - a minimum number of required appearances and a maximum number. These numbers specify the number of times the attribute may appear in any instance of the resource type. This specification only defines the following cardinalities: 0..1, 0..*, 1..1, and 1..*. Profiles that describe specific use cases may use other values for cardinality within the limits of the cardinality defined by the base resource.

Note that when present, elements cannot be empty - they SHALL have a value attribute, child elements, or extensions. This means that setting an element to a minimum cardinality of 1 does not ensure that valid data will be present; specific FHIRPath constraints are required to ensure that the required data will be present.

In this specification, very few elements have a minimum cardinality of 1. Resources are used in many contexts, often quite removed from their primary use case, and sometimes even basic information is quite incomplete. For this reason, the only elements that have a minimum cardinality of 1 are those where they are necessary to any understanding of the resource or element that contains them. The minimum cardinalities should not be taken as a guide to what elements are expected to be present in any particular use of the resource, including their normal/primary usage purpose. In some cases, this specification publishes additional profiles that define which elements are required in particular situations. Similar profiles are published by jurisdictions, vendors, or projects.

For elements that have cardinality > 1, the order in which they appear may have meaning. Unless the element definition (either in this specification or the extension) defines a meaning to the order explicitly (using ElementDefinition.orderMeaning, the meaning of the order is not defined, and implementations are allowed to reorder the elements. Note that it is not possible to define a meaning for the order of the elements in a profile using a StructureDefinition. When there is no definition of the meaning of the order, implementations that need to choose a single element from a list of elements for some use SHALL do so based on the semantics of the content of the elements that repeats. Profiles and Implementation guides may often make rules about this selection process.

Clients should not depend on servers maintaining ordering of elements, unless the retrieved resource conforms to a profile which mandates maintenance of ordering. If a server cannot maintain ordering, it must strip off known profile tags which require maintenance of ordering, and strip off unknown profiles (since they might require maintenance of ordering).

Is-Modifier is a boolean property that is assigned when an element is defined, either as part of the base resource contents in this specification, or when extensions are defined. An element is labeled "Is-Modifier = true" if the value it contains may change the interpretation of the element that contains it (including if the element is the resource as a whole). Typical examples of elements that are labeled "Is-Modifier" are elements such as "status", "active", "refuted", or "certainty". Whether an element is a modifier cannot be changed when element usage is described in a profile (i.e. a constraining Structure Definition). When an element is labeled as Is-Modifier, the documentation must be clear about why it is a modifier.

A typical example of a modifier element is one that negates the element that contains it. For instance, in the following fragment of a resource definition:

NameFlagsCard.TypeDescription & Constraintsdoco
.. AllergyIntolerance DomainResourceAllergy or Intolerance (generally: Risk Of Adverse reaction to a substance)
... onset Σ0..1dateTimeDate(/time) when manifestations showed
... patient Σ1..1Reference(Patient)Who the sensitivity is for
... verificationStatus ?! Σ0..1codeunconfirmed | confirmed | refuted | entered-in-error
AllergyIntoleranceVerificationStatus (Required)
... criticality Σ0..1codeCRITL | CRITH | CRITU
AllergyIntoleranceCriticality (Required)

The value of the 'verificationStatus' element affects the entire meaning of the resource - if it is set to refuted, the entire resource must be understood differently, and so it is not safe for implementations to ignore it. As a consequence, it is labeled as 'is modifier = true'. In this tabular representation of the resource, this shows as the flag '?!'. The JSON and XML representations of a resource definition have their own representation of 'is modifier = true' status, and it is defined directly in a ElementDefinition.

Is-Modifier elements SHALL be represented in the narrative summary of the resource.

If the value of a modifier element is not explicit in the instance, or known by the context, the resource may not be able to be safely understood. Wherever possible, elements labeled "Is-Modifier = true" also have a minimum cardinality of 1, or a default value, in order to introduce certainty in their handling. However, sometimes this is not possible - much legacy data is not well described. Implementations producing resources SHOULD ensure that appropriate values for isModifier elements are provided at all times.

Implementations processing the data in resources SHALL understand the impact of the element when using the data. Implementations are not required to "support" the element in any meaningful way - they may achieve this understanding by rejecting instances that contain values outside those they support (for instance, an application may refuse to accept observations with a reliability other than "ok"). Alternatively, implementations may be able to be sure that, due to their implementation environment, such values will never occur. However applications SHOULD always check the value irrespective of this.

Note that processing the data of a resource typically means copying or filtering data out of a resource for use in another context (display to a human, decision support, exchange in another format where not all information is included, or storing it for this kind of use). Servers and background processes that simply move whole resources around unchanged are not "processing the data of the resource", and therefore these applications are not required to check Is-Modifier elements.

Every element in the base resource has a value of "true" or "false" for the Is-Modifier flag. The value of the flag cannot be changed by profiles on the resource, in either direction. When a profile defines an extension, it labels the extension with the Is-Modifier flag, and this cannot be changed in other profiles. Note that extensions that have is-Modifier = true are represented differently in resource instances ("modifierExtension" instead of "extension"), and there are additional rules about how they are handled.

Here is a list of all the modifier elements defined in the specification:

Labeling an element Must-Support means that implementations that produce or consume resources SHALL provide "support" for the element in some meaningful way. Because the base FHIR specification is intended to be independent of any particular implementation context, no elements are flagged as "must-support" as part of the base specification. This flag is intended for use in profiles that have a defined implementation context.

For this reason, the specification itself never labels any elements as must-support. This is done in Resource Profiles, where the profile labels an element as mustSupport=true. When a profile does this, it SHALL also make clear exactly what kind of "support" is required, as this can mean many things.

Note that an element that has the property IsModifier is not necessarily a "key" element (e.g. one of the important elements to make use of the resource), nor is it automatically mustSupport - however both of these things are more likely to be true for IsModifier elements than for other elements.

All elements may have constraints attached to them (also known as 'invariants'). Constraints defined on an element have the following properties:

KeyIdentifies the constraint uniquely amongst all the constraints in the context - typically, this is used to refer to the constraint in an error message
RequirementsAn explanation of why the constraint has been applied - what harmful conditions are being avoided
SeverityWhether the constraint is an error, or a warning. The exact difference in meaning of these depends on context, but an error is associated with "SHALL" and systems rejecting content, whereas a warning might not be
Human DescriptionA human description of the rule intended to be show as the explanation for a message when the constraint is not met
ExpressionAn FHIRPath expression that must evaluate to true when run on the element
XPathAn XPath expression that must evaluate to true when run on the element in the XML representation

Many constraints are defined in the base specification. In addition, additional constraints may be defined in profiles that apply to resources. Systems are not required to evaluate the constraints, just as they are not required to check for conformance, or schema validity. However, systems SHOULD always ensure that all resources are valid against all applicable constraints.

Elements can also be explicitly associated with constraints defined elsewhere. This is a notification to implementers that the element is affected by the constraint. It has no meaning when the constraints are evaluated.

Profiles may define additional constraints that apply to an element, but they cannot alter or remove constraints that are already applied.

In addition to the conformance metadata, each element has other metadata properties:

  • isSummary - How the element behaves in Summary Searches
  • defaultValue / meaningWhenMissing - What it means when an element is missing
  • maxLength - How long the representation of the primitive value is allowed to be (not including XML or JSON escaping)

This specification includes many examples. While every effort has been made to ensure that the examples are fully conformant to the specification, if the examples disagree with the specification, the specification is considered correct and normative, not the examples. This same rule applies to the reference implementations.

The examples reflected in this specification do *not* represent actual people. Any resemblance to real people - alive or dead - is entirely coincidental. In some cases, examples may be drawn from real clinical data. However, if this has occurred, the content has been scrubbed to remove any identifying information.