The CMS Prior Authorization API is not one endpoint. It is a set of capabilities — determining whether authorization is required, discovering documentation requirements, submitting requests, and returning decisions — that payers generally deliver through the HL7 Da Vinci Coverage Requirements Discovery (CRD), Documentation Templates and Rules (DTR), and Prior Authorization Support (PAS) implementation guides. Impacted payers must implement the API by January 1, 2027, while the operational decision timeframes of 72 hours for expedited and seven calendar days for standard requests already applied from January 1, 2026.
Read the requirement as capabilities, not an endpoint
CMS describes the Prior Authorization API in the singular, which understates the work. Its published workflow requires that providers be able to determine whether a prior authorization is required, identify what documentation the payer needs, submit the request electronically, and receive the payer's decision. Each of those is a distinct interaction with distinct data dependencies inside your utilization management stack.
Treating the requirement as four capabilities changes the delivery plan. Requirement checks depend on published coverage rules. Documentation discovery depends on machine-readable rule authoring. Submission and decisions depend on integrating with the adjudication path that already exists for X12 278 transactions.
How the Da Vinci guides line up with the four capabilities
Most payer programs implement this through the HL7 Da Vinci Burden Reduction guides rather than inventing a proprietary interface, because provider EHRs are building against those same guides.
- CRD (Coverage Requirements Discovery): CDS Hooks calls from the provider workflow that answer whether prior authorization is required.
- DTR (Documentation Templates and Rules): delivers payer rules to the provider so documentation is gathered in-workflow rather than by fax or portal.
- PAS (Prior Authorization Support): carries the request and the decision, mapping FHIR content to and from the X12 278 transaction where that remains the transactional standard.
- Patient Access API: prior authorization information must also be surfaced to members, which is a separate obligation on an existing API.
A build order that reduces rework
The dependency runs backwards from how teams usually plan it. PAS submission is meaningless without a rules source, and CRD answers are only trustworthy if the coverage rules behind them are governed and versioned. Start with the rules asset, then the transaction path, then the in-workflow hooks.
- Inventory the services subject to prior authorization and the current rule source of truth for each.
- Establish rule authoring, versioning, and approval before exposing any rule through an API.
- Prove the PAS path end to end against a single service category and a single provider partner under production-like conditions.
- Add CRD and DTR once rules are stable, then expand service categories on the proven pattern.
Do not let the API plan hide the operational obligations
The operational provisions of CMS-0057-F carry earlier dates than the API provisions. Impacted payers, excluding QHP issuers on the federally facilitated exchanges, must send decisions within 72 hours for expedited requests and seven calendar days for standard requests, provide specific denial reasons, and publicly report prior authorization metrics annually. Those obligations are process and reporting work owned by utilization management, and they do not become easier because an API project is underway.
How BlockInterop supports this work
- FHIR and API Implementation consulting and implementation — Design and implement FHIR-based APIs, SMART on FHIR applications, US Core-aligned workflows, and scalable integration services.
- CMS Interoperability consulting and implementation — Prepare for and implement CMS interoperability, prior authorization, payer-to-payer, provider access, and patient access requirements.
Sources
Every figure cited in this article links to its public source. BlockInterop publishes no client names, outcomes, or internal performance statistics.
