Why this system bought back 4.6 hours of physician time per week.
Authora deployed inside Epic via SMART-on-FHIR and CDS Hooks. The PA workflow now lives at the point of order — not in the in-basket, not in the after-clinic backlog, not on the appeals desk three weeks later.
What was broken, in dollar terms.
The system's CMO ran the math at the 2025 medical-group retreat: 460 physicians × 13 hours per week on PA × $280/hour internal loaded cost = $87M of physician time spent on PA per year. The number was so large the CFO initially refused to believe it.
- Average physician time on PA: 13 hours/week — consistent with the 2024 AMA national survey (12 hours)
- PA-related in-basket messages: 2,300/day across the medical group
- Peer-to-peer call backlog: 6.2 days, with three full-time hospitalists rotated off floor duty to handle the queue
- Patient drop-off after PA delay >5 business days: 22% never returned for the ordered service
The integration, the workflow change, the 90-day arc.
The deployment was about workflow placement, not just adjudication. Putting Authora behind the EHR was the difference between a faster denial and a structural change.
- SMART-on-FHIR launch from inside Epic at order signing — Authora pre-flights the PA before the physician closes the encounter
- CDS Hooks card surfaces missing evidence at order time, with one-click DTR questionnaire completion against the payer's actual criteria
- PAS submission happens in the background; the physician sees the decision (or the gap) inside the EHR within seconds
- Peer-to-peer requests now arrive with the complete clinical context attached — the conversation moves from 'what is the patient's history' to 'do you agree with the criteria gap we identified'
The seven seeded categories, by the numbers.
Pilot ran across the seven CPT categories that, in combination, accounted for roughly 62% of the customer’s PA volume. Auto-approval rates reflect the share of cases that cleared without human review under the customer’s own medical-policy library.
| CPT | Category | Auto-approval | P50 latency | Weekly volume |
|---|---|---|---|---|
| 72148 | MRI lumbar spine | 74% | 29s | 186/wk |
| 27447 | Knee arthroplasty | 63% | 38s | 72/wk |
| 78815 | PET/CT skull-to-thigh | 71% | 32s | 54/wk |
| 96413 | IV chemotherapy admin | 81% | 22s | 194/wk |
| 90867 | TMS therapy | 59% | 44s | 38/wk |
| J9271 | Pembrolizumab inj. | 79% | 24s | 104/wk |
| E2102 | DME — CGM sensor | 86% | 19s | 228/wk |
“Our physicians did not want a faster denial. They wanted the conversation about evidence to happen while the patient was still in the room. That is the actual product. The time savings followed.”
Year-one expansion plans.
- Roll Authora to the system's three affiliated FQHC partners under a shared-services agreement
- Integrate with the system's home-health and post-acute network for DME and HHA prior auth
- Build a real-time gold-card eligibility view inside the EHR so physicians see, per-payer, which orders are pre-authorized
- Publish year-one outcome data jointly with the system's quality affairs office to NEJM Catalyst
The trust posture is the same as the pilot.
Every pilot runs under BAA, in us-east-2, with the same SOC 2 Type II + HITRUST r2 posture documented in /trust. The pilot economics are documented in /pilot.