Onyx Blog
Prior Authorization: The View from Backstage
By Immanuel Anthony, Tech Contributor, Onyx
Everyone in healthcare IT knows prior auth is broken. Fewer people have stood backstage and watched exactly how it breaks — and what it takes to fix it at the infrastructure level. That’s where this piece comes from
The premise of prior auth is reasonable: verify care is covered before it’s delivered. A doctor orders something, the payer says yes or no, and the patient gets care. But the moment you’re backstage, watching how it actually happens, you realize prior auth is not paperwork. It is a distributed workflow stitched together by humans, portals, PDFs, and yes, fax machines.
“None of these are exotic edge cases. They’re just… Tuesday.”
Here’s what happens more often than anyone wants to admit: delay, “need more info,” a PDF that’s anywhere from 2 to 45 pages, unclear receipt so docs get resent, and one critical detail buried in a note. Prior auth has survived on “best effort” — but that best effort is different in every system, and rarely standardized.
The stakeholder landscape spans EHRs like Epic and Cerner, clearinghouse networks, and payer stacks that include eligibility systems, utilization management workbenches, clinical policy tools, decision support, and nurse reviewers. A payer-facing ePA platform like ours becomes the connector between standards-based exchange and the systems that actually produce decisions.
A lot of people assume the main challenge is mapping fields into a standard format. It isn’t. The real challenge is that decisions depend on context, and context is scattered.
At Onyx, our Patient Access APIs, integration surfaces, and data pipelines each tell different parts of the story. When we connect those views — across payloads, responses, documents, and lifecycle changes — we can reduce the gaps that drive pends and resubmissions. And when patterns repeat, we can feed them back into the standards process through clarifications and ballots, so future FHIR IGs better match real workflows.
Everyone asks this. AI has huge potential here — summarizing documents, extracting key facts, guiding the next best question. But prior auth is loaded with PHI. That means security, privacy controls, auditability, and vendor governance are not optional; they are the whole game. I look at AI as something we enable responsibly over time, not something we bolt on recklessly.
ePA won’t make prior auth instantly easy. Policies will still vary, documents won’t become structured overnight, and human judgment will remain. But it does create something we’ve never really had: a trackable workflow. With standardized exchange, providers can trust status, payers can reduce avoidable pends, UM teams chase fewer artifacts, and engineering teams can measure bottlenecks instead of guessing.
We are not flipping a switch from fax to flawless. But for the first time, we can actually see the whole board — and that changes everything about how we play the game.
Immanuel Anthony is a Tech Contributor at Onyx, working on electronic prior authorization and standards-based interoperability. He focuses on connecting payer systems with FHIR-based exchange and is actively involved in HL7 standards development.
