US Core Implementation Guide
8.0.0-ballot - STU8 Ballot United States of America flag

This page is part of the US Core (v8.0.0-ballot: STU8 Ballot 1) based on FHIR (HL7® FHIR® Standard) R4. The current version which supersedes this version is 7.0.0. For a full list of available versions, see the Directory of published versions

General Guidance

Page standards status: Trial-use

Referencing US Core Profiles

The search expectations and US Core Profiles have been developed and tested using logical FHIR ids. Therefore, a reference to a US Core resource SHOULD include a logical id (Reference.reference), not an identifier (Reference.identifier).

FHIR resource elements of the Reference datatype reference other FHIR resources. For all references, US Core Responders SHOULD return resources that conform to a US Core profile if a US Core profile exists for the resource type. In some cases, US Core profiles explicitly constrain these references to point to the relevant US Core profile because the element is typically a Must Support or Additional USCDI Requirements element. For example, CareTeam.subject in US Core CareTeam Profile references the US Core Patient Profile. In other cases, referencing elements point to the base FHIR resource because they are not a Must Support or Additional USCDI Requirements element and, therefore, are unconstrained by US Core. (This is an editorial decision to simplify and improve the readability of the Profile rendering in this guide.) For example, although US Core does not constrain Condition.asserter, its reference to Patient or Practitioner SHOULD be a valid US Core Patient Profile or US Core Practitioner Profile.

Contained Resources

When responding to a query, Servers SHOULD not use inline contained resources to represent the returned data. A contained resource can only be used when the source data exists within the context of the FHIR transaction. For example, the guidance in the Medication List page describes how a contained Medication in MedicationRequest is used to represent the medication. In addition, if referencing a contained resource in a US Core Profile, the contained resource SHOULD be a US Core Profile if a US Core Profile exists for the resource type. Further guidance about the general use case for contained can be found in the base FHIR specification.

Suppressed Data

In situations where the specific piece of data is hidden for security or privacy reasons, using a code from the DataAbsentReason Code System such as masked may exceed the data receiver's access rights to know. However, masking data SHOULD be handled based on implemented policies. For elements with a minimum cardinality = 0 (including elements labeled Must Support), the element SHOULD be omitted from the resource. For Mandatory elements (in other words, where the minimum cardinality is > 0), use the code "unknown" following the guidance on Missing Data in the Conformance Sections.

SNOMED CT United States Edition

SNOMED CT (Systematized Nomenclature of Medicine – Clinical Terms) is a comprehensive clinical terminology widely used in healthcare to support the electronic exchange of clinical health information. US Core uses the US Edition of SNOMED CT, a standalone release that combines the content of the US Extension and the International releases of SNOMED CT. It is used extensively in US CORE for various clinical concepts, including problems, procedures, allergies, and laboratory results. When using SNOMED codes in US Core Profiles, implementers MAY use the default system URI, which refers to an unspecified edition/version, as shown in option one below. However, for terminology Servers to be able to validate US Edition-only codes, implementers SHOULD provide the accompanying system URI to describe the edition as shown in option two.

SNOMED CT Edition and Version Options

  1. Using only the system http://snomed.info/sct refers to an unspecified edition/version. For example:

     "code": {
         "coding": [
             {
                 "system": "http://snomed.info/sct,
                 "code": "763875007",
                 "display": "Product containing sulfonamide (product)"
             }
         ],
         "text": "sulfonamide antibacterial"
     },

  2. Using the system plus the unspecified version US Edition URI http://snomed.info/sct/731000124108. For example:

     "code": {
         "coding": [
             {
                 "system": "http://snomed.info/sct",
                 "version": "http://snomed.info/sct/731000124108",
                 "code": "763875007",
                 "display": "Product containing sulfonamide (product)"
             }
         ],
         "text": "sulfonamide antibacterial"
     },

For more details, see Using SNOMED CT with FHIR.

Using UCUM codes in the Quantity datatype

Some US Core Profiles bind the Quantity.code element in the Quantity datatype to the UCUM code system. For example, in the US Core Laboratory Result Observation Profile, Observation.valueQuantity, Observation.referenceRange.low, and Observation.referenceRange.high are bound to UCUM. Systems should also use UCUM for the optional valueRange and valueRatio datatypes (which are complex datatypes with Quantity elements). A FHIR UCUM Codes ValueSet that defines all UCUM codes is in the FHIR specification. This guidance specifies how to represent the Quantity datatype when the correct UCUM units are missing or when the units are missing, which will likely occur in the real world.

UCUM code provided

 "valueQuantity": {
    "value": 26.0,
    "unit": "g/mL",
   "system": "http://unitsofmeasure.org",
   "code": "g/mL"
  }

free text units only:

  • If UCUM units are unavailable, represent units in the unit element.
 "valueQuantity": {
    "value": 26.0,
    "unit": "RR",
     }

no units

 "valueQuantity": {
    "value": 26.0
 }

Representing Deleted Information

A FHIR Server SHOULD not delete records. A FHIR Server SHOULD update the appropriate resource status to entered-in-error or inactive (refer to the next section). If a system supports the deletion of records, they SHOULD refer to the Deletion Safety Checks in the FHIR specification.

Representing Entered in Error Information

Clinical information entered in error in the patient's record needs to be represented by the FHIR Server so that Clients can expose the corrected information to their end users.

Server Recommendations:

  • A FHIR Server SHOULD not delete resources.
  • A FHIR Server SHOULD update the appropriate resource status to entered-in-error or inactive.
  • A FHIR Server SHOULD allow these resources to be searchable by Client applications.
  • If the FHIR Server has updated the resource status to entered-in-error:
    • For patient facing applications, A FHIR Server SHOULD remove the resource's contents, leaving only an id and status. Note that this typically will not conform to the US Core or FHIR StructureDefinitions.
    • For provider-facing applications, the content may be supplied with content and additional detail (such as the reason for the status change) that the patient viewing system would typically not have access to.

Narrative

The US Core CarePlan Profile requires a narrative summary of the patient assessment and plan of treatment. However, any US Core Profile can include a human-readable narrative containing a summary of the resource and may be used to represent the resource's content to a human. For further guidance, refer to the Narrative documentation in the FHIR Specification.

Language Support

There is a basic need to access records in your language, and the data provider SHOULD do their best to translate (safely) to the requested language. Understand that this will be variably complete depending on the nature of the record. For example, translating the following elements is relatively straightforward:

  • Coding.display
  • Generated narrative
  • Attachment.title

The following guidelines outline requesting and returning a resource in the requested language.

  • Clients MAY request language/locale using the http Accept-Language header.
  • Servers SHOULD make reasonable efforts to translate what can be safely translated.
  • Servers SHOULD populate the Resource's language element with a code based on the underlying language of record, not the requested language.

    • Using the Human Language Extension when the language of a display, etc, is known to differ from the stated (or inferred) language.

      Example

          <?xml version="1.0" encoding="UTF-8"?>
    <Patient xmlns="http://hl7.org/fhir">
        <id value="language-example-1"/>
        <meta>
        [...snip...]
        </meta>
        <language value=" es"/>
        <!--0..1 Language of the resource content, in this case, Spanish-->
        <text>
        [...snip...]
        </text>
        <extension url="http://hl7.org/fhir/us/core/StructureDefinition/us-core-race">
            <extension url="ombCategory">
                <valueCoding>
                    <system value="urn:oid:2.16.840.1.113883.6.238"/>
                    <code value=" 2106-3"/>
                    <display value="White">
                        <!--Human Language extension-->
                        <extension url="http://hl7.org/fhir/StructureDefinition/language">
                            <valueCode value=" en"/>
                            <!--English is different from stated language-->
                        </extension>
                    </display>
                </valueCoding>
            </extension>
        [...snip...]

  • Using the Translation Extension when the Server provides additional translations by its own choice or in response to a different Accept-Language than what the resource is stored in.

    Example

          <?xml version="1.0" encoding="UTF-8"?>
      <Patient xmlns="http://hl7.org/fhir">
          <id value="language-example-2"/>
          <meta>
          [...snip...]
          </meta>
          <language value=" en"/>
          <!--0..1 Language of the resource content: English-->
          <text>
          [...snip...]
          </text>
          <extension url="http://hl7.org/fhir/us/core/StructureDefinition/us-core-race">
              <extension url="ombCategory">
                  <valueCoding>
                      <system value="urn:oid:2.16.840.1.113883.6.238"/>
                      <code value=" 2106-3"/>
                      <display value="White">
                      <!-- Translation -->
                      <extension
                           url="http://hl7.org/fhir/StructureDefinition/translation" >
                       <extension url="lang"> <!-- 1..1 Code for Language -->
                        <valueCode value="es"/><!-- 0..1 Value of extension -->
                       </extension>
                       <extension url="content"> <!-- 1..1 Content in other Language -->
                        <valueString value="Blanca">
                       </extension>
                      </extension>
                      </display>
                  </valueCoding>
              </extension>
              [...snip...]

  • Servers SHOULD make it known what languages are supported in their CapabilityStatement(s) using this extension1:

    http://hl7.org/fhir/5.0/StructureDefinition/extension-CapablilityStatement.acceptLanguage

For further guidance on language and locale for the generation of the resource narrative, see Multi-Language support in FHIR (Localization/Internationalization) in the core specification.

Read (Fetch) Syntax

Fetching resource interactions are defined by the FHIR RESTful API. Guidance is included on the US core Profile pages for conformance and informative purposes:

GET [base]/[Resource-type]/[id] {parameters}

  • GET is the HTTP verb used for fetching a resource
  • Content surrounded by [] is mandatory for the Client to supply and will be replaced by the string literal identified.
  • Content surrounded by {} is optional for the Client to supply and will be replaced by the string literal identified.
    • parameters: URL parameters as defined for the particular interaction (e.g., "?_format=xml"}

For more information, see the FHIR RESTful API

Search Syntax

Searching resources is defined by the FHIR RESTful API and included here for informative purposes. The US Core FHIR RESTful Search API Section documents the Server and Client rules for the RESTful interactions described in this guide.

All the search interactions in this guide use the GET command with the following syntax:

GET [base]/[Resource-type]?[parameter1]{:m1|m2|...}={c1|c2|...}[value1{,value2,...}]{&[parameter2]{:m1|m2|...}={c1|c2|...}[value1{,value2,...}]&...}

  • GET is the HTTP verb used for fetching a resource
  • Variables surrounded by "[]" are mandatory for the Client to supply and will be replaced by the string literal identified.
  • Variables surrounded by "{}" are optional for the Client to supply and will be replaced by the string literal identified.
    • base: The Service Root URL (e.g., "https://fhir-open-api-dstu2.smarthealthit.org”)
    • Resource-type: The name of a resource type (e.g., "Patient")
    • parameter: the search parameters as defined for the particular interaction (e.g.,"?patient=Patient/123")
    • value: the search parameter value for a particular search
      • When searching using the token type searchparameter (how to search by token), the syntax {system|}[code] means that the system value is optional for the Client to supply.:
      • When searching using the reference type SearchParameter (how to search by reference), the syntax {Type/}[id] means that the Type value is optional for the Client to supply:
      • When searching using the date type SearchParameter (how to search by date), the syntax data={gt|lt|ge|le}[date] means the date comparators "gt", "lt", "ge", and "le" are optional. Date type searches without a comparator prefix are equivalent to searches with the "eq" comparator even if a Server does not support the comparator.
    • {:m1 m2 …}: The list of supported search parameter modifiers
    • {c1 c2 …}: The list of supported search parameter comparators
    • {,value2,…}: Optional multiple 'OR' Values
    • {&parameter2={:m1 m2 …}={c1 c2 …}[value1{,value2,…}&…}: Optional multiple 'AND' search parameters

In the simplest case, a search is executed by performing a GET operation in the RESTful framework:

GET [base]/[Resource-type]?name=value&...

For this RESTful search, the parameters are a series of name=[value] pairs encoded in the URL. The search parameter names are defined for each resource. For example, the Observation resource has the name "code" for searching on the LOINC code. For more information, see the FHIR RESTful Search API.

Note that when the patient is implicit in the context of some implementations (e.g., using SMART), the patient parameter can be omitted:

GET [base]/[Resource-type]{?other-parameters}

Searching Multiple Patients

Currently, most EHRs permit queries that provide a single patient resource id but do not support the comma-separated query or a query where the patient parameter is omitted, as described in the standard FHIR REST API. Instead, a user-facing app can perform multiple "parallel" queries on a list of patient ids. Alternatively, the FHIR Bulk Data Access (Flat FHIR) specification can be used to perform a "back end" system-level query to access large volumes of information on a group of individuals. It can also be used to identify and query against an unknown population, such as when looking for population-based research data or population-level queries for public health surveillance.

However, neither specification defines how a user-facing provider app can seek real-time "operational" data on multiple patients (such as all patients with recent lab results). Opportunities to add this capability to this guide are discussed in Future of US Core.

Searching Using lastUpdated

Servers SHOULD support the _lastUpdated search parameter for US Core Profiles and SHOULD populate Meta.lastUpdated for US Core Profiles as accurately as possible. Servers SHALL document in CapabilityStatement.rest.resource.searchParam.documentation the types of changes that can be detected using the _lastUpdated search parameter (see example snippet below).

Example CapabilityStatement snippet for a Server supporting the _lastUpdated search parameter for US Core Laboratory Result Observation Profile

{
"resourceType": "CapabilityStatement",
    ...
    "rest": {
      {
        "mode": "Server",
        ...
        "resource": [
            {
              "type": "Observation",
              "supportedProfile": [
                  "http://hl7.org/fhir/us/core/StructureDefinition/us-core-observation-lab",
                  ...
                ],
                 ...
              "searchParam": [
                  {
                  "name": "_lastUpdated",
                  "definition": "http://hl7.org/fhir/SearchParameter/Resource-lastUpdated",
                  "type": "date",
                  "documentation": "This parameter is supported for Observations with the category
                  \"laboratory\" and allows searching for Observations that have been created or the 
                  status updated since the specified date."
                  },
                  ...
                ]
            }
        ]
    }
}

Many Servers are unable to accurately populate the Meta.lastUpdated element.

Note to Clients:

  • Updates to Meta.lastUpdated may not reflect a change in the resource and resource updates may not result in updates to Meta.lastUpdated.
  • _lastUpdated search results may not reflect changes Clients can access and all updates to a resource may not be returned using the _lastUpdated search. (in other words, false negatives and false positives _lastUpdated search results are possible).
  • Updates to Meta.lastUpdated may reflect changes a Client can not access. (for example, the Client may not be authorized to see the changed data).
  • Supporting Meta.lastUpdated in a resource does not imply support for searches using the _lastUpdated search parameter.
  • Support for searches using the _lastUpdated search parameter does not require Servers to support Meta.lastUpdated; Servers can use an alternative method to track changes to an instance.

Note to Servers:

  • Updating the timestamp too frequently is better than missing updates.

Work is in progress to enable FHIR Topic-Based Subscription for notifications on relevant events as an alternative to search polling.

US Core Servers are not required to support patient compartment based searches.

Across Platform Searches

US Core Servers are not required to resolve absolute URLs external to their environment.

Limiting The Number Of Search Results

Servers can choose to return the results in a series of pages to manage the number of search results returned. The search result set contains the URLs used to request additional pages from the search set. For a simple RESTful search, the page links are included in the returned bundle as links. See the managing returned resources in the FHIR specification for more information.

To reduce API volume and ensure reliable use when implementing FHIR RESTful searches:

  1. When the Server supports it, Clients SHOULD use the _include search parameter to retrieve referenced content instead of searching for a resource and then performing a separate search for other references (for example, Patient, Encounter, and Location) in the result set.

  2. If the Server does not support the _include search parameter, Clients SHOULD consolidate duplicate searches before searching for referenced resources in the result set

  3. The HTTP Cache-Control response header stores a response associated with a request and reuses the stored response for subsequent requests. If Cache-Control is present in the Server response headers, the Clients SHOULD NOT search the same data within the time stated in Cache-Control headers.

Footnotes:

  1. This extension is converted from a new element in a future version of CapabilityStatement. ↩