Prior Authorization in Florida: Denials, Resubmissions, and the Half of the Problem Vendors Skip

The 2024 AMA Prior Authorization Physician Survey put the operational damage at 24 hours of staff time per physician per week. The CMS Interoperability and Prior Authorization Final Rule, CMS-0057-F, addresses the submission side — by 2027 payers will be required to expose FHIR-compliant APIs that let practices submit electronically. The rule does not address the part that actually costs Florida practices money.

That same AMA survey found 94% of physicians reporting care delays attributable to prior authorization, and 24% reporting a serious adverse patient event as a direct result. Those numbers are not anecdotes from one specialty. They represent a structural friction point that Florida practices, already operating under Florida Medicaid managed-care PA requirements that sit on top of commercial payer rules, absorb differently than a large health system with a centralized revenue cycle team.

A four-physician orthopedic group in Coral Gables does not have a denial management department. A two-physician cardiology practice in Brickell is not staffed to run payer-specific resubmission workflows by hand.

I have come to believe the vendor market has misdiagnosed the problem entirely. The pitch you hear at every health IT conference is faster submission. That is the wrong surface.

Submission volume is not where revenue disappears

Every platform selling prior authorization automation today, from Availity to Waystar to the PA modules baked into AthenaHealth and Cerner, emphasizes submission speed and portal consolidation. Those are real improvements. A practice that previously toggled between twelve payer portals now moves through a single queue. Turnaround on initial submission drops.

But the AMA survey data makes the actual bottleneck visible: the cost is not in sending the request. It is in what happens after the denial comes back.

In my work with Florida practices, initial denial rates across commercial payers typically run between 15% and 22% depending on specialty, with orthopedics and oncology sitting at the high end. Each denial triggers a decision tree: a clinical documentation deficiency, a wrong CPT or ICD-10 pairing, a payer-specific formulary rule that changed in Q1, or a peer-to-peer request that needs physician time on a call that may or may not result in an overturn.

Submission automation does not touch any of that. The PA coordinator still opens the denial letter, reads the reason code, looks up the payer's appeal requirements, pulls the chart, identifies what documentation is missing, and starts over. That loop is where the 24 hours per week goes.

Denial reason classification at the point of receipt

The first place a well-designed workflow changes the economics is denial intake. When a denial arrives from Florida Blue, Humana, or Sunshine Health (one of Florida's contracted Statewide Medicaid Managed Care plans under the 2024-2030 SMMC awards), it carries a reason code. PA denials typically ride on an X12 278 response transaction or on payer-proprietary portal output, and the reason categories are not standardized across payers in any operationally useful way. A medical-necessity denial from UnitedHealthcare and a free-text "medical necessity not established" string from Sunshine Health can mean structurally different things requiring different documentation to overturn.

An automated classification layer that maps incoming denial reason strings to a taxonomy of resolution pathways changes what happens next. Instead of a coordinator reading free-text and making a judgment call, the system routes the denial to the correct resubmission template for that payer, that code, and that specialty. The coordinator's job becomes review and approval rather than diagnosis and construction from scratch. That is a meaningful difference in time per denial.

What most Florida practices encounter today is the generic denial queue inside AthenaHealth or Cerner with no payer-specific routing intelligence layered on top. The classification layer has to be built or bought separately, and most practices have done neither.

Payer-specific resubmission workflows

This is the piece the vendors almost universally skip in their demos, and it is the part that actually determines overturn rate.

Florida Statewide Medicaid Managed Care is split across Sunshine Health, Molina Healthcare of Florida, Simply Healthcare, and a handful of others under the current AHCA contract awards. Each has a different PA appeal window (typically 30 to 60 days from denial), different peer-to-peer scheduling mechanisms, different documentation templates for clinical necessity letters, and different historical patterns of what overturns versus what does not. A practice that sees 200 PA denials per month across a mixed payer panel is managing 200 different mini-processes that share almost no procedural DNA.

Automated workflows that encode payer-specific logic, what documentation Molina accepts for a lumbar MRI medical necessity appeal versus what Simply Healthcare requires, reduce the cognitive load on the coordinator and reduce the error rate on resubmission. The variance in documentation requirements is not large in any single case, but across 200 monthly denials it compounds into hours of rework and a meaningful number of missed appeal windows. Missing the appeal window on a procedure authorization in the $4,000 range, for example, converts a recoverable denial into a write-off.

CMS-0057-F has a layered timeline that practices should understand precisely. The denial-reason disclosure provisions take effect for impacted payers for plan years beginning January 1, 2026, and the FHIR Prior Authorization API compliance date is January 1, 2027. The 2026 disclosure piece is what makes denial reasons more transparent at the policy level; the 2027 API piece is what makes them machine-readable at the point of receipt. That second milestone is the necessary precondition for the classification layer described above to work at scale. Practices building or buying workflow infrastructure now should verify that any platform they deploy has a roadmap for FHIR API ingestion of denial reasons, not just submission outputs.

A reasonable counterargument: once the FHIR PA API and the gold-carding and auto-approval provisions in CMS-0057-F and state-level proposals take effect, some share of low-risk PA volume will be auto-approved at submission, which compresses the denial pool the resubmission thesis depends on. I take that seriously. But auto-approval will land first on the cleanest, most rule-codifiable procedure categories. The denials that remain will be concentrated in exactly the high-friction specialties (orthopedics, oncology, advanced imaging, specialty pharmacy) where resolution-side workflow already drives the economics. Auto-approval shrinks the denominator. It does not shrink the workflow problem on what remains.

Clinical documentation gap-filling before the denial happens

The most expensive denial is the one that did not have to happen. A significant share of prior authorization denials in Florida, particularly for imaging and specialty procedures, trace back to incomplete clinical documentation at the point of initial submission: a diagnosis not carried forward from the referring note, a conservative treatment history not explicitly documented, a functional limitation scale that the payer's criteria require but the physician did not record using the payer's preferred terminology.

A workflow that runs a pre-submission gap check against the payer's medical necessity criteria for that specific procedure code can surface those deficiencies before the request leaves the practice. This is not a spell-checker. It is a structured comparison of what the payer's criteria require and what the chart actually contains, flagging the gaps for clinical staff to address.

Some EHR-native PA tools are beginning to incorporate criteria libraries from MCG Health (formerly Milliman Care Guidelines) and InterQual (currently held within Optum following the Change Healthcare acquisition, with ongoing divestiture discussion). The question for a Florida practice evaluating these tools is whether the criteria library is current (payers update medical necessity policies quarterly in some cases), and whether the gap-check output produces an actionable documentation request rather than a generic alert.

The win-win here is alignment: the practice submits a cleaner request, the payer processes it faster, and the patient receives authorization without a 14-day delay that sometimes becomes a care abandonment.

Peer-to-peer scheduling as an operations problem

The peer-to-peer call is one of the highest-value levers in PA denial management. Specialty-society and MGMA reporting put overturn rates for peer-to-peer reviews in select procedure categories well above 50%, and in some imaging and specialty pharmacy contexts higher still. It is also structurally broken at most Florida practices because scheduling a peer-to-peer requires a physician to be available during a window that payers control, which typically runs 9:30am to 4:00pm EST on business days.

For a physician already seeing patients, coordinating a peer-to-peer call is friction on top of friction. Practices that systematically track which denials should go to peer-to-peer (versus direct appeal), automate the scheduling request to the payer, and surface the pending call to the physician in the EHR workflow will recover more of the denials that are recoverable. That tracking and surfacing function is an operations workflow, not a clinical one, and it belongs in the PA coordinator's dashboard, not in the physician's inbox.

No major EHR vendor has solved this natively as of mid-2025. The operational pattern that does work, and that practices can implement Monday without buying anything: a shared coordinator dashboard that tags each denial as P2P-eligible based on payer-specific historical overturn data, queues the eligible calls into a single 15- or 30-minute physician window each afternoon, and gives the coordinator authority to schedule the payer call into that window directly. The physician receives one batched calendar block per day, not five ad hoc interruptions. The dashboard can live in Salesforce Health Cloud, in a structured Airtable, or in the practice management system, but the principle is the same: aggregate the calls, protect the physician's clinical time, and stop losing recoverable denials to scheduling friction.

Four questions to put to any vendor selling PA automation in Florida

Before you commit to a platform or an agency building PA workflow automation, I would put these directly to the vendor:

One: What is the taxonomy of denial reason categories your system classifies, and how do you handle payer-proprietary reason strings that fall outside X12 278 standard responses? Ask them to show you how the system routes a medical-necessity denial from Florida Blue versus a free-text denial reason from Sunshine Health.

Two: Which Florida SMMC managed care plans have payer-specific resubmission templates built into your workflow, and when were those templates last updated against current payer policy? If they cannot name Sunshine Health, Molina Florida, and Simply Healthcare specifically, the platform was not built for this market.

Three: What is your roadmap for ingesting CMS-0057-F FHIR Prior Authorization API denial reason data, and which payers in your network have committed to early API availability before the January 2027 mandate? This tells you how future-proof the architecture is.

Four: Does your pre-submission gap check run against a current medical necessity criteria library, and what is the update cycle for that library when payers revise their policies? If the answer is "we use a static criteria set," the gap-check produces false confidence.

Submission speed was never the bottleneck. Resolution speed is. Any platform that cannot answer those four questions with specificity is selling you the wrong half of the problem.