Release 5

This page is part of the FHIR Specification (v5.0.0: R5 - STU). This is the current published version in it's permanent home (it will always be available at this URL). For a full list of available versions, see the Directory of published versions . Page versions: R5 R4B R4 R3

FHIR Infrastructure icon Work GroupMaturity Level: 2Standards Status: Trial Use

TODO: Discussion (or reference to one) on Polling and Subscription

Diagram showing creation of Task on placer system workflow
  1. Placer creates the request in its own system via POST, an internal action, or POSTs it to a queue server system
  2. Placer creates a Task resource in its own system via POST, an internal action, or POSTs it to a queue server system, pointing to the request resource and seeking fulfillment.
    The Task may have a specified performer, in which case step 3 is expected to be done by that performer.
    If the Task does not have a specified "performer" (although may have performer type), then this is a case of an "open" task, where any number of fulfillers may attempt to "claim" the task. Who succeeds is determined by local policies and procedures.
  3. Fulfiller system uses either polling or pub/sub to become aware of the existence of the task
    1. A common case may be the conveyance of the Task id to a fulfiller by other means. For example, a lab test is ordered, and the patient takes a requisition to the lab of his choice. The requisition contains the Task id (as a bar code, or stored in the patient's healthcare smart card), and the lab system can execute a direct GET for the Task, thus eliminating the need for subscription or polling.
  4. Fulfiller system queries to retrieve the referenced request and updates the Task to indicate "acceptance" and agreement to fulfill
  5. Fulfiller may update the Task to indicate interim progress notes
  6. Placer is made aware of the acceptance of the Task and any changes to the Task either through its ownership of the resource or using polling or a subscription to a queue server system to determine the same
  7. Fulfiller creates an event resource in its own system via POST or internal action, or POSTs it to a queue server system
  8. Fulfiller PUTs an update to the Task resource to change its status to completed and to point to the event resource
  9. Placer is aware of the completion of the Task either through the ownership of the resource, or via polling or subscription to a queue server system, and retrieves the referenced event resource
  10. Placer updates the request resource to indicate completion via PUT or an internal action, or PUTs the update to a queue server system
  • Uses the FHIR REST API for managing the workflow
  • Both placer and fulfiller can track the state of the workflow
  • Can use this approach for request other than just fulfillment (e.g. to request status change or other updates)
  • There's an ability to negotiate fulfillment - i.e. the ability to say "no"
  • Explicit acknowledgement that filler has received and agreed to act on the request
  • Supports cancellation through ability of placer to change the status of the Task to cancelled.
  • Additional complexity of using Task
  • Additional complexity of setting up and maintaining a subscription or polling infrastructure
  • Placer and fulfiller must have a FHIR server, and placer's FHIR server must allow "write" updates to the Task resource
  • Placer and filler may need to be able to communicate directly (i.e. know each other's respective endpoints)

    • This could become unmanagable if there are a large (or dynamic) number of placers and fillers that need to communicate
    • Placer and fulfiller must know where to subscribe for content - this could be a large number of systems
    • Might not apply if there's a queue server
  • Does not give fulfiller control over whether the task can be cancelled
  • When following the state of the workflow is important
    • When there is a need to know that a fulfiller has accepted to perform the request
    • When there is a need to be able to change the performance of the request
    • When there are multiple steps in fulfilling the request
  • When requests are not (always) directed to a specific filler
  • When the placer needs to manage the Task resource

Lab order to known performer

A provider orders a bilirubin test for a patient, to be done at the provider's facility lab....

"Open" lab order

A patient is scheduled for their physical, and the provider sends her a secure message to have a fasting blood test performed before the office visit. The message has a link to a requisition, and a list of possible lab locations where the test can be done. The patient chooses a lab location near her place of employment and a test is scheduled. The lab gets the requisition from the patient and, using the information on how to retrieve the Task from the provider's system, claims the test to be performed...

Diagram showing POST of Task to filler system workflow
  1. Placer creates the request in its own system via POST or an internal action, or POSTs it to a queue server system
  2. Placer POSTs a Task resource to the filler system, pointing to the request resource and seeking fulfillment
  3. Fulfiller system GETs the referenced request
  4. Fulfiller updates the Task to indicate acceptance of the task
  5. Placer either polls the Task to note acceptance or uses a subscription to determine the same
  6. Fulfiller may further update the Task to reflect the progress made. Using the same method as in step 5, the placer becomes aware of these updates
  7. Fulfiller creates an event resource in its own system via POST or an internal action, or POSTs it to a queue server system
  8. Fulfiller Updates the Task resource to change its status to completed and to point to the event resource
  9. Placer either polls the Task to note completion and changes or uses a subscription to determine the same
  10. Placer system queries to retrieve the referenced event resource
  11. Placer updates the request resource to indicate completion via PUT or an internal action, or PUTs the update to a queue server system
  • Uses the FHIR REST API for managing the workflow
  • Subscription/polling is targeted to a specific resource instance - much simpler than general subscription/polling infrastructure
  • Both placer and fulfiller can track the state of the workflow
  • Can use this approach for request other than just when requesting fulfillment (e.g. to request status change or other updates)
  • There's an ability to negotiate fulfillment - i.e. the ability to say "no"
  • Explicit acknowledgement that filler has received and agreed to act on the request
  • Placer can attempt to cancel task in the event of order cancellation, but no guarantee
  • Filler has control over whether task can be cancelled
  • Additional complexity of using Task
  • Additional complexity of using subscription or polling
  • Placer and fulfiller must have a FHIR server and fulfiller's FHIR server must allow "write" creation of the Task resource by the placer
  • Placer and filler may need to be able to communicate directly (i.e. know each other's respective endpoints)
    • This could become unmanagable if there are a large (or dynamic) number of placers and fillers that need to communicate
    • Might not apply if there's a queue server
  • Placer might not know immediately when filler system has retrieved the request
  • When following the state of the workflow is important
    • When there is a need to know that a fulfiller has accepted to perform the request
    • When there is a need to be able to change the status of the workflow by either the placer or the fulfiller
    • When there are multiple steps in fulfilling the request
  • When the fulfiller needs to manage the Task resource

Outpatient prescription

A patient is suffering from a poison ivy rash and is prescribed a steroid medication to manage the reaction. The provider asks the patient for his preferred pharmacy and creates the Task at the specified pharmacy's system. The patient is running late and by the time he arrives at the pharmacy, it is closed. He calls the clinic where the provider practices and requests that the pharmacy where the order is to be fulfilled is changed to one that is open around the clock. The staff at the clinic cancels the Task at the original pharmacy and creates a new Task at the requested new one. By the time the patient arrives at the new pharmacy, the medication is ready for pickup. Once the request is fulfilled, the clinic receives the notification that the Task is complete, and the prescription is also marked as complete.

TODO: Still needs review and update

Diagram showing workflow broker workflow
  1. Placer POSTs the request to its own system or to a queue server system
  2. Broker detects that new un-assigned request (without a Task yet created and falling within the scope of the Broker to ensure fulfillment) via polling or subscription
  3. Broker POSTs a Task resource to its own system or a queue server system, pointing to the request resource and seeking fulfillment from a specific filler
    Task does not have a specified "performer" (but may have performer type)
  4. If the Task is rejected by one potential recipient, the broker may create a new task to seek fulfillment from others
  5. Continue as per Option G
  • Offloads responsibility for seeking fulfillment from the placer system, but more actively solicits fulfillment than a simple "post the task and see who takes it". Also, allows prioritized assignment of tasks (i.e. some fillers may be preferred over others)
  • Requires a broker to exist
  • Broker must know all available fillers and their capabilities to allow appropriate assignment
  • Broker must understand business rules for permitting changes to tasks such as cancelation

Appropriate in environments that have a workflow engine that takes on responsibility for ensuring fulfillment

Diagram showing mutual POSTs of a Task to filler system and Task to placer system workflow
  1. Placer creates the request in its own system via POST or an internal action
  2. Placer POSTs a Task resource to the filler system, pointing to the request resource and seeking fulfillment
  3. Fulfiller system GETs the referenced request
  4. Fulfiller updates the Task to indicate acceptance of the task
  5. Fulfiller POSTs a Task, which also points to the request resource and uses the Task.partOf attribute to point to the Task from step 2 (indicating it is a sub-Task)
  6. Fulfiller may further update both Tasks to reflect the progress made. Since the sub-Task is on the placer's system, the placer is aware of these updates
  7. Fulfiller creates an event resource in its own system via POST or an internal action
  8. Fulfiller Updates the Task resource to change its status to completed and to point to the event resource
  9. Fulfiller updates the sub-Task resource as completed and points to the event resource. Since the sub-Task is on the placer's system, the placer is aware of the completion and changes.
  10. Placer system queries to retrieve the referenced event resource
  11. Placer updates the request resource to indicate completion via PUT or an internal action
  • Uses the FHIR REST API for managing the workflow
  • No need to use subscription or polling
  • Both placer and fulfiller can track and change the state of the workflow
  • Can use this approach for request other than just when requesting fulfillment (e.g. to request status change or other updates)
  • There's an ability to negotiate fulfillment - i.e. the ability to say "no"
  • Explicit acknowledgement that filler has received and agreed to act on the request
  • Placer can initiate task cancellation, but filler decides whether it's possible to cancel the task or not
  • Additional complexity of using Task
  • Need to keep the two Task states in sync
  • Placer and fulfiller must each have a FHIR server and give the other side "write" access to create and update Task resources.
  • Placer and filler may need to be able to communicate directly (i.e. know each other's respective endpoints)
    • This could become unmanagable if there are a large (or dynamic) number of placers and fillers that need to communicate
    • Might not apply if there's a queue server
  • When following the state of the workflow is important
    • When there is a need to know that a fulfiller has accepted to perform the request
    • When there is a need to be able to change the status of the workflow by either the placer or the fulfiller
    • When there are multiple steps in fulfilling the request
  • When there is no need for polling or subscription infrastructure
  • When us of the FHIR REST API is preferred to any other method of communications
  • When the intended fulfiller is known in advance
  • When messaging-like functionality is needed, but a messaging infrastructure is not available

Outpatient referral

A patient is seen by her Primary Care Provider (PCP) for shortness of breath and the provider decides to refer her to a cardiologist. Because of the referral, the provider's system creates ServiceRequest resource and POSTs a Task resource to a cardiologist's practice system, requesting a referral. This places the referral request on a referral coordinator's work list to review. Upon accepting the referral, the Task is updated as accepted and a new sub-Task, referencing both the ServiceRequest and the Task created by the provider's system, is created on the provider's system. The patient gets an appointment with the cardiologist and the sub-Task is updated with that information, allowing the provider's system to inform the provider of the change.

The patient misplaces the information about the appointment and misses the visit with the cardiologist. The cardiologist's system updates the sub-Task to indicate the no-show and the PCP is alerted of the missed appointment. The provider's staff gets hold of the patient and schedules another appointment for her with the cardiologist. Because of the cardiologist's system scheduling the appointment, the sub-Task is updated with that information. The patient is seen and the sub-Task is updated accordingly. The cardiologist document the findings about the patient and updates the sub-Task as complete, adding references to the resources representing the findings. The PCP is alerted of the updates and marks the referral as complete, which updates the original Task and the ServiceRequest as complete as well.

TODO: needs more details

  1. Placer sends message to filler with a MessageHeader, where the focus element points to the Task resource, also contained in the message. The message might or might not contain any other relevant resources (e.g. the actual request resource), or an "event" code saying "please fulfill"
  2. Filler system sends a response containing the same Task resource, indicating receipt of the message and, optionally, an indication of their intention to fulfill the request
  3. Filler system may send incremental messages to the placer showing progress (e.g. specimen collected, preliminary results, final results) by including an updated Task resource
  4. Placer system may also send messages to the fulfiller containing the Task resource and updating the state of the workflow, for example cancelling the task
  • Reduced number of communications
  • All relevant data can be sent in one package
  • Responses can be asynchronous, and content may be routed
  • There's an ability to negotiate fulfillment - i.e. the ability to say "no"
  • Can request things other than just fulfillment (e.g. please suspend)
  • Explicit acknowledgement that filler has received and agreed to act on the request (though no need for the placer to check)
  • Messaging is "heavy"
  • Need to negotiate what allowed responses are and what data can be present in request and response messages
  • Additional complexity of using Task
  • Need message delivery infrastructure in place

Appropriate when existing messaging infrastructure can be used (e.g. HL7 over HTTP, v2 LTP, MLTP, WSI Web Services, Direct, VISA, REST, etc.), and a need to stay consistent with that architecture.

TODO: This scenario needs work - there's not a lot of experience using FHIR services to manage the fulfillment process

  1. Placer creates a request resource on their own system or a queue server
  2. Placer may create a Task resource on their own system or a queue server
  3. Placer invokes a service on the filler system saying "please fulfill this request", including the content or a reference to the request resource and any other relevant data
  4. Filler system responds (synchronously if using HTTP, but may be asynchronous if using SOAP or other transport mechanisms) with conformation of receipt and, optionally indication of intention to fulfill and/or results
  • No need to expose RESTful endpoints for either requests or tasks - acts like a remote procedure call
  • Requires support for a custom operation by both parties (typically at least a different operation for each type of request)
  • No way to get insight into progress without invoking another custom operation

TBD

This approach requires the presence of an active FHIR-aware intermediary/broker, that provides the appropriate interface to requester and fulfiller with different capabilities. The broker translates between multiple workflow fulfillment approaches. For example, messaging in, REST out, or REST in, operation out.

An example of a Prescription workflow is shown below:

Diagram showing a hybrid messaging/REST approach

These steps are for the example above, which includes the assumption that the filler is subscribed to any relevant Task resources on the broker.

  1. Placer sends message to broker with a MessageHeader, where the focus element points to the Task resource, also contained in the message. In this case the message also contains the MedicationRequest resource describing the prescription, as well as any other relevant resources.
    Note that in certain cases it may be possible to omit the Task resource, for example if MessageHeader.event can be directly mapped to Task.code. For support of complex workflow management, it is still recommended to include the Task resource in the message Bundle.
  2. Broker creates local copies of the resources contained in the message, and sends a subscription notification to Filler system that a new Task resource is available
  3. Filler searches broker for the known Task resource, using the relevant _include and _revinclude values to get the information about the prescription
  4. Filler system may PUT incremental updates to the Task resource on the broker showing progress
  5. Based on the updates of the Task resource, the broker may send FHIR messages to the placer to inform them of the progress. In order to include the relevant information about the updates, the broker may precede the sending of the message with a search on the filler to obtain the necessary details.
  • Allows interoperability among systems with different data exchange architectures
  • A sophisticated FHIR-aware broker can be used for multiple purposes, and provides the flexibility to include more participants at different levels of support for FHIR
  • Requires the presence of a broker that is capable of converting between the different approaches
  • Does not provide RESTful access to the clinical information on the non-RESTful side

This approach may enable a gradual introduction for FHIR RESTful exchange patterns

TODO: needs review and update. Possibly add options about using messaging and/or services instead of polling/subscription in above scenarios

This is a variation of Option H, where the Workflow broker is essentially merged with the fulfiller. It still allows the placer to only use a POST of the request and be made aware of the changes to the other resources via subscription or polling.

Diagram showing POST of "request" resource for filler system, response via Task workflow
  1. Placer system invokes a "create" action by POSTing a 'request' resource (e.g. ServiceRequest etc.) to the appropriate RESTful resource endpoint (e.g. [base]/MedicationRequest) on the filler, placer or queue server system and sets a "tag" on the resource that indicates the request is "actionable"
  2. Filler POSTs a Task resource to its own system or a queue server system, pointing to the request resource and indicating intent to fulfill or refusal to fulfill
  3. Placer system uses either polling or pub/sub to become aware of the existence of the task and fulfillment intent
  4. Fulfiller may update the Task to indicate interim progress notes
  5. Placer either polls the Task to note acceptance and changes or uses a subscription to determine the same
  6. Fulfiller POSTs an event resource to its own system or to a queue server system
  7. Fulfiller Updates the Task resource to change its status to completed and to point to the event resource
  8. Placer system becomes aware of the update via polling or subscription
  9. Placer system retrieves the event
  10. Placer system marks the request as "complete"