CMS-0057-F Prior Authorization API

The Prior Authorization API lets providers ask an impacted payer, over FHIR R4, whether prior authorization is required for an item or service, what documentation is needed, and then submit the request and track its decision, live by January 1, 2027. CMS names the Da Vinci implementation guides as the recommended path (recommended, not required): CRD 2.1 (is auth required?), DTR 2.0 (collect the documentation), and PAS 2.1 (submit and track). Decision timeframes (72 hours expedited, 7 calendar days standard) and specific denial reasons apply from January 1, 2026.

Must answer	Whether prior authorization is required, the documentation rules, and the request's decision status with a specific reason for denials
Recommended IGs	Da Vinci CRD 2.1, DTR 2.0, PAS 2.1 (PAS is the FHIR representation of the X12 278 prior-auth transaction)
Decision timeframes	From January 1, 2026: 72 hours expedited, 7 calendar days standard
API deadline	January 1, 2027
Also required	Public prior-auth metrics (approval, denial, and appeal volumes and timing) posted annually

Example

Readiness checklist
 1. Map your prior-auth intake to the three-IG chain: CRD answers "is auth
    required", DTR collects the questionnaire, PAS carries the $submit and its
    ClaimResponse.
 2. Install the three IG packages and validate against the full profiles, not
    the embedded subset stubs.
 3. Wire the decision pipeline to meet the 2026 timeframes: 72 hours
    expedited, 7 calendar days standard, denial reason populated.
 4. Publish a CapabilityStatement declaring the PAS Bundle and Claim profiles
    your endpoint serves.
 5. Gate each IG separately in CI so failures attribute to the right team:

pidgeon data install fhir-davinci-pas-2.1
pidgeon data install fhir-davinci-crd-2.1
pidgeon data install fhir-davinci-dtr-2.0

pidgeon conform --endpoint https://fhir.example.com \
  --ig http://hl7.org/fhir/us/davinci-pas/ --walk --ci

Repeat the conform step with --ig http://hl7.org/fhir/us/davinci-crd/ and --ig http://hl7.org/fhir/us/davinci-dtr/ as separate CI steps. Each run produces structural conformance evidence against the installed IG; add --probe-operations to also POST a synthetic PAS request Bundle to Claim/$submit and grade the response (no real patient data is sent).

Validation context

Strict and compatibility validation notes land here with the phase-1 data export.

This answer, in your terminal

pidgeon conform --endpoint https://fhir.example.com --ig http://hl7.org/fhir/us/davinci-pas/ --walk --ci

Wire it into your AI agent:

claude mcp add pidgeon -- npx -y @pidgeonhealth/mcp

Download Post free to browse this reference with generation attached.