This page is part of the Da Vinci Prior Authorization Support (PAS) FHIR IG (v0.1.0: STU 1 Ballot 1) based on FHIR R4. The current version which supercedes this version is 1.1.0. For a full list of available versions, see the Directory of published versions
This section of the implementation guide defines the specific conformance requirements for systems wishing to conform to this Prior Authorization Support implementation guide. The bulk of it focuses on the Claim $submit operation, though it also provides guidance on privacy, security and other implementation requirements.
Before reading this formal specification, implementers should first familiarize themselves with two other key portions of the specification:
The Use Cases & Overview page provides context for what this formal specification is trying to accomplish and will give a sense of both the business context and general process flow enabled by the formal specification below.
The Technical Background page provides information about the underlying specifications and indicates what portions of them should be read and understood to have necessary foundation to understand the constraints and usage guidance described here.
This implementation guide uses specific terminology to flag statements that have relevance for the evaluation of conformance with the guide:
SHALL indicates requirements that must be met to be conformant with the specification.
SHOULD indicates behaviors that are strongly recommended (and which may result in interoperability issues or sub-optimal behavior if not adhered to), but which do not, for this version of the specification, affect the determination of specification conformance.
MAY describes optional behaviors that are free to consider but where the is no recommendation for or against adoption.
This implementation guide sets expectations for two types of systems:
This specification makes significant use of FHIR profiles and terminology artifacts to describe the content to be shared as part of prior authorization requests and responses.
The full set of profiles defined in this implementation guide can be found by following the links on the Artifacts page.
The primary interaction supported by this implementation guide is submitting a prior authorization request and receiving back a response. To perform this, a FHIR Bundle resource is constructed by the client (EHR) system. That Bundle will contain a Claim resource (which FHIR uses to submit prior authorization requests), together with various referenced resources needed to support the population of the 5010 version of the ASC X12N 278 prior authorization request and the 6020 version of the ASC X12N 275 additional documentation transactions.
This Bundle will then be sent as the sole payload of a Claim/$submit operation. The server on which the operation is invoked will convert the Bundle into an ASC X12N 278 and 0..* additional unsolicited 275 transactions and execute them all against the target payer system. It will then take the resulting 278 response and convert it into a response FHIR Bundle containing a ClaimResponse and associated resources. All of this will happen synchronously with a maximum of 15 seconds between the user initiating the prior authorization request and seeing the resulting response. (Less time is better.)
In the event that the prior authorization cannot be evaluated and a final response returned within the required timeframe, a response in which one or more of the requested authorizations are ‘pended’ will be returned. The client (or other interested systems - e.g. patient, caregiver or performing provider systems) can then query the server endpoint for the final results using either a polling or subscription-based mechanism. During this period of time, the same $submit operation can be used to request cancellation or modification of the prior authorization.
NOTE FHIR uses a pair of resources called Claim and ClaimResponse for multiple purposes - they are used for actual claim submission, but they are also used for managing prior authorizations and pre-determinations. These are distinguished by the Claim.use code. All references to Claim and ClaimResponse in this implementation guide are using it for the prior authorization purpose.
The Claim/$submit operation is executed by POSTing a FHIR Bundle to the [base url]/Claim/$submit endpoint. The Bundle can be encoded in either JSON or XML. (Servers SHALL support both syntaxes.) The first entry in the Bundle SHALL be a Claim resource complying with the profile defined in this IG to ensure the content is sufficient to appropriately populate an X12N 278 message. Additional Bundle entries SHALL be populated with any resources referenced by the Claim resource (and any resources referenced by those resources, fully traversing all references and complying with all identified profiles). Note that even if a given resource instance is referenced multiple times, it SHALL only appear in the Bundle once. Bundle.entry.fullUrl values SHALL be the URL at which the resource is available from the EHR if exposed via the client’s REST interface and SHALL be of the form “urn:uuid:[some guid]” otherwise. All GUIDs used SHALL be unique, including across independent prior authorization submissions - with the exception that the same resource instance being referenced in distinct prior authorization request Bundles can have the same GUID.
In addition to these core elements needed to populate the 278 message, any “supporting information” resources needed to process the prior authorization request (whether determined by DTR processes or by other means) must also be included in the Bundle. Relevant resources referenced by those “supporting information” resources SHALL also be included (e.g. prescriber Practitioner and Medication for a MedicationRequest). All such resources SHALL comply with relevant US Core profiles. All “supporting information” resources included in the Bundle SHALL be pointed to by the Claim resource using the Claim.supportingInfo.valueReference element. The Claim.supportingInfo.category should be populated appropriately if possible, using ‘other’ if no appropriate category is known. The Claim.supportingInfo.sequence for each entry SHALL be unique within the Claim.
All resources SHALL comply with their respective profiles. FHIR elements not marked as ‘must support’ MAY be included in resources within the Bundle, but client systems should have no expectation of such elements being processed by the payer unless prior arrangements have been made. Servers that do not process such elements SHALL ignore unsupported elements unless they are ‘modifier’ elements, in which case the server MAY treat the presence of the element as an error.
Details on how to map the FHIR Bundle to the relevant X12N 278 and 275 messages are expected to be published by ASC X12N. The server is responsible for performing full conversion of all mapped elements, including execution of terminology translations when necessary. In addition, the server SHALL produce an additional 275 message whose binary segment contains a base64-encoded copy of the entire FHIR Bundle resource. This serves two purposes - it provides full audit traceability for the payer and it also allows the payer to directly process the FHIR content, potentially extracting elements not present in the X12 messages if needed. (Note: there is no requirement that payers take any such action.)
This IG treats everything that happens beyond the defined operations endpoint receiving the FHIR bundle as a black box. This black box includes any business associate(s), clearinghouse(s), payers, contracted review entities, and other intermediaries that may be involved in the PA request and response. It is up to that black box to ensure that any regulatory requirements are met and to perform all processing within the allowed timeframe.
The response to the prior authorization is processed in the reverse order as the request. The server is responsible for converting the ASC X12N 278 response into a FHIR Bundle. The Bundle SHALL start with a ClaimResponse entry that contains information mapped from the 278 response. As well, just like for the prior authorization request, additional Bundle entries must be present for all resources referenced by the ClaimResponse or descendent references. When converting additional Bundle entries, the conversion process SHOULD ensure that only one resource is created for a given combination of content. E.g. if the same Practitioner information is referenced in multiple places, only one Practitioner instance should be created - referenced from multiple places as appropriate. When echoing back resources that are the same as were present in the prior authorization request, the server SHALL ensure that the same fullUrl and resource identifiers are used in the response as appeared in the request.
The resulting Bundle is returned as the HTTP body of the POST response.
This is an example of a standard Referral Request / Response sequence between a Primary Care Provider and a Utilization Management Organization. The request example will show how a PCP can request a referral to a specialist for a patient from a UMO. The response example will also show the response.
Joe Smith is a subscriber to Maryland Capital Insurance Company. During a regular physical, Dr. James Gardener, Joe’s primary care physician, diagnoses a potential heart problem. Dr. Gardener determines that it would be best to refer Joe to Dr. Susan Watson, a cardiologist, for a consultation.
Dr. Gardener is required by Maryland Capital Insurance to submit a request for review seeking approval to refer Joe to Dr. Watson.
After review, Maryland Capital approves the referral and responds.
When the ClaimResponse.outcome is ‘queued’ or ‘partial’, it means that the payer requires additional time to make a final determination on all items within the prior authorization request. In this situation, the client system will need to retrieve the prior authorization response at a later point once a final decision has been made. There are two possible options - polling and subscription. Implementers SHOULD support subscription. Servers SHALL support polling in situations where either party is unable to use the subscription approach.
Note: There are use-cases for multiple systems potentially needing to check on the status of status of a pended prior authorization. In addition to the provider who submitted the prior authorization request, the status might also be of interest to:
As a result, queries seeking the status of the prior authorization response may come from multiple systems. Servers SHALL permit access to the prior authorization response to systems other than the original submitter. They SHALL require a match on both patient coverage id (identifier on the Claim.patient) and prior authorization id (Claim.identifier) to ensure access is only granted to individuals who know both - and thus have demonstrated a need to know.
In this approach, the Client regularly queries the Server to see if the status of the prior authorization has changed. This is done by performing a query on the server’s “ClaimResponse” endpoint, searching based on prior authorization identifier and patient coverage information. The query should look like this:
[base]/ClaimResponse?identifier=[authorizationresponseid]&patient.identifier=[patientid]&status=active
The authorizationresponseid is the ClaimResponse.identifier returned in the original synchronous response to the authorization request and represents the payer’s identifier for that transaction. The patient.identifier is the member identifier that was submitted as the patient.identifier in the original claim.
As per FHIR’s token search parameter, the identifiers can contain either only the Identifier.value or the Identifier.system. Clients SHOULD send the Identifier.system if it is known, but MAY search by the Identifier.value strings only. Systems wishing to reduce bandwidth impact can also filter using _lastudated
to only retrieve the record if it has changed since the previous query. Servers SHALL support this parameter.
Clients SHALL perform this query no more than 4 times within the first hour and no more frequently than once every hour after that. They SHOULD perform this query at least once every 12 hours.
The project is seeking feedback on whether these maximum frequency requirements are acceptable.
The intermediary SHOULD execute a 278i to return the status. However, if the payer does not support that function, the intermediary SHALL return the most recent copy of the prior authorization response as received from the payer. (Note that in this latter case, the intermediary would need to persist prior authorization information and would therefore be a ‘covered entity’ for HIPPA purposes.)
Notes:
Subscriptions require more sophistication than polling, but reduce communication overhead by ensuring that queries only occur when data has changed. When using the subscription retrieval mechanism, the client will POST a new Subscription instance to the server’s [base]/Subscription endpoint. The Subscription.criteria SHALL be of the form: “identifier=[authorizationresponseid]&patient.identifier=[patientid]&status=active”. (Order of parameters with the search does not matter.)
The project is seeking feedback on whether the implementation guide should mandate one of the subscription channel approaches (either rest-hook or websocket). Feedback for or against standardization as well as an expression of preference is welcome.
Once the subscription has been created, the server SHALL send a notification over the requested channel indicating that the prior authorization response has changed. This may happen when the response is complete, but may also occur when information on one or more of the items has been adjusted but the overall response remains as ‘pended’.
Upon receiving a notification, the client SHALL - when convenient - execute the same query as shown above in the polling section.
If the retrieved ClaimResponse has an outcome of ‘complete’ or ‘error’, the client SHALL perform a DELETE on the Subscription.
Systems other than the requesting system may choose not to poll or subscribe to the prior authorization response but instead to check the status at the request of a user. This query is performed in the same manner as the polling query.
In some cases, the needs associated with a prior authorization may change after the prior authorization request was submitted. This might be a change to one of the services needed, the timeframe over which the service is provided, the quantity of the service or product, or even the elimination of the need for a given service.
There are four types of changes possible:
In the first case, the Bundle is resubmitted using the $submit operation with the status code changed to ‘cancelled’. In all other cases, the change is requested by creating a new Bundle containing a new Claim resource with its own unique Claim.identifier and posting it using the $submit operation. That resource will point to the previous Claim using the Claim.related element. The relationship type will be ‘replaces’. The new Claim instance will comply with the Revised Prior Authorization profile.
From an X12 perspective, only those items/attachments that are being added/cancelled/revised need to be present. From an HL7 perspective, resources are generally represented as a cohesive whole, not a set of deltas from a previous resource. This standard therefore provides two different mechanisms for representing the revised prior authorization request:
In this case, the prior authorization request is handled in a typical FHIR manner and all items and supporting information is included in the Bundle - including items that have not changed at all. Changed information is flagged as follows:
The intermediary will create 278 and/or 275 submissions that instantiate the changes (by looking for those items and supportingInfo elements) and will ignore the unchanged items.
The benefit of this approach is that it is consistent with the way the prior authorization would need to be passed around if ever shared in a RESTful manner. However, it can be bandwidth intensive if the prior authorization contains a large number of items, but only a small number of those have changed. Unfortunately, the typical FHIR patch mechanism cannot be used in this implementation guide because the intermediaries do not have (and cannot have for regulatory reasons) a locally stored copy of the original prior authorization, nor any mechanism to retrieve one.
An example of a changed full request can be found at Updated Homecare Request along with the original Homecare Request.
In this case, only the Claim, related resources needed to support the Claim (e.g. submitting organization) and those items and supportingInfo elements that have actually been changed/added are included. I.e. The Claim instance doesn’t represent the full prior authorization request, but only the overall prior authorization metadata and the subset of elements that are different. The submitted Bundle will be identical to that above, however it will omit all items that do not have a changed extension. It will also exclude any resources in the Bundle that are now no longer needed because the references to them have been removed with the of the removal of the non-changed Claim.item and/or Claim.supportingInfo elements.
Note that if the change is to cancel the entire request, in the differential approach, there is no need to send any items or supporting Info.
Because this Claim instance doesn’t represent the ‘full’ authorization request, but only a subset, the Claim instance SHALL also declare a security tag with a value of SUBSETTED to make clear that the resource is incomplete.
An example of a differential request for the same Homecare Request scenario can be found at Differential Homecare Request.
Just as the submission of a changed prior authorization request can be submitted in two different modes, a payer can choose to respond in two different modes. Some payers may include responses for all items in the authorization. Others might only include responses for those items that were specifically changed. (In theory, some payers could also return the items that were changed as well as those that are still pended and thus considered ‘open’.) As for the request, if a ClaimResponse does not contain items corresponding to all that are part of the revised prior authorization (including those cancelled or unmodified), it SHALL declare a security tag with a value of SUBSETTED to make clear that the resource is incomplete.
The intermediary would populate the ClaimResponse Bundle based on the approach the payer had chosen in their 278 response.
NOTE: When querying for the current status of a prior authorization, the prior authorization response SHALL include all items, even if the identifier queried against corresponds to a prior authorization response whose synchronous response was a differential. For example, if a prior authorization revision was submitted changing one item out of four, the synchronous prior authorization response might only contain one item (and a subsetted flag). However, a subsequent query for the status of that prior authorization would always return a prior authorization resource that contained all four items.
PAS servers SHALL ensure that prior authorizations that were initially pended remain available for query for at least 6 months after the anticipated completion of the services whose authorization was requested.
If the prior authorization response is a refusal, it is not permitted to send an ‘update’ to the request in the hopes of receiving a different answer. Instead, a new request must be initiated.
Note that data submitted by client systems will comply with US Core profiles, meaning that codes for medications, conditions, etc. will be those used for clinical purposes, not billing. The intermediary will be responsible for performing any necessary mappings (e.g. SNOMED diagnosis codes to ICD10)
It is the intent of this implementation guide to provide specifications for the exchange of prior authorization in a way that is conducive to developing test scripts and a reference implementation (RI) that can be used to validate/exercise the IG at connectathons and during piloting and production deployment. It is also the intent of this guide that any test scripts will include testing of: