top of page

The Gate Moved. It Didn't Open.

Writer: Vitali Khvatkov
Vitali Khvatkov
6 hours ago
5 min read

Analysis · Pinnacle Services Corporation · September 2026

For the first time, the prior-authorization denial rate is a published number. KFF's analysis of the first year of disclosures mandated under the CMS interoperability and prior-authorization rule, released on August 13, found that insurers denied at least one in eight standard requests in 2025: 12% in Medicare Advantage, 14% in Medicaid managed care, 18% on the federal exchanges. Among the largest Medicare Advantage carriers the range ran from 5% at Elevance to 17% at UnitedHealthcare.

The number beside it matters more. Of the Medicare Advantage denials that were appealed, 67% were overturned; in Medicaid managed care, 47%; on the exchanges, 43%. At Centene the Medicare Advantage figure was 93%. Appeals stayed rare. A review process that loses two-thirds of its challenges but faces almost none is not clinical judgment. It is throughput that prices in silence.

Prior-authorization denial rates and overturn-on-appeal rates by program, 2025, per KFF analysis of CMS-mandated disclosures

The concession that wasn't

Three weeks later the largest carrier moved. On September 1, UnitedHealthcare sent providers the lists of roughly 1,700 procedure codes that lose their prior-authorization requirement on October 1 — about 30% of its preapproval rules, spanning oncology, cardiology, orthopedics, home health, and, by the company's own description, genetic and laboratory testing. The breakdown by plan: more than 800 codes on commercial plans, about 940 on the exchange product, some 1,400 on Oxford, and about 120 on Medicare Advantage and dual-eligible special-needs plans.

Read at the code level, the laboratory relief is narrower than the headline. Of the 793 codes freed on the commercial plan, thirty are laboratory tests, and twenty-eight of those are proprietary laboratory analysis codes — the PLA "U" codes that describe a single manufacturer's assay. There is not one conventional pathology or laboratory CPT code among them. The Medicare Advantage list contains no laboratory codes of any kind. The exchange list frees 39 laboratory-type codes, eleven of them the unlisted codes — 81099, 84999, 88199, 88299, 88399 and their kin — that payers rarely authorize in the first place.

The gate moved. It did not open.

Where the machinery is going

The direction of travel is toward faster gates, not fewer. On August 17, Epic announced that four health systems — Ochsner, Froedtert ThedaCare, Denver Health and Summit Health — had gone live checking prior-authorization requirements for UnitedHealthcare and Aetna inside the electronic record at order entry, five months ahead of the January 1, 2027 federal deadline that requires payers to expose real-time authorization APIs. Sixteen more payers were testing. The same week, R1 agreed to acquire Humata Health for its "touchless" authorization agents, folding them into a platform it describes as an autonomous revenue cycle; the target had shipped a product for independent practices only in June, a reminder of who this machinery is built for first.

Medicare has bought the model too. Records obtained by the Electronic Frontier Foundation under a Freedom of Information Act suit, reported on September 8 and 15, describe how WISeR, Medicare's AI prior-authorization pilot running in six states since January, pays its vendors: on averted expenditure, which is to say on what they deny, with a poor quality score trimming the fee by five to ten percent. Two vendors denied more than 20,000 requests in the pilot's first three months. One denied more than it approved. The 72-hour decision standard was routinely missed; one request waited 83 days. The pilot runs until 2031.

The heresy is spreading from unexpected quarters. On August 21 a health insurance chief executive, Sidecar Health's Patrick Quigley, told Becker's that prior authorization should be made illegal. The American Economic Liberties Project, in a brief published the same month, estimated that the practice consumes the equivalent of 99,000 clinicians and up to $32.7 billion a year.

Prior authorization was never where laboratory claims died

All of this is upstream of the specimen, and for a laboratory that is the point. A laboratory bills downstream of an encounter it did not conduct, for a patient who was never in its building. It sees the authorization decision, if at all, as a denial code weeks after the test was run.

The money shows it. On the twelve-month denial book of a national collection of more than 12,000 physicians, prior-authorization denials — CARC 197 — totaled $318,674 on 3,437 claims. Recovr's models classify 97.9% of that as avoidable, meaning the authorization should have been obtained, and $110 as recoverable after the fact. In August alone, missing precertification accounted for $27,000 of denied charges on 363 claims, under one percent of a $2.9 million denial month.

What grew in the same window was review after the fact. Denials for documentation the payer requested and then judged late or incomplete rose from $14,000 in July to $72,000 in August. Denials demanding records before adjudication, CARC 252, nearly doubled between June and August, to $50,600 on 906 claims, much of it surgical pathology — 88305, 88307, 88341 — across commercial and Medicaid plans. Our models score 53% of that pool as recoverable, provided the attachment is produced and the claim refiled before the filing clock expires.

Review did not disappear when the checkpoint was removed. It relocated past the specimen and past the bill, into the adjudication engine, where it arrives at machine volume with a filing deadline attached. Prior-authorization reform, whatever it eventually delivers for hospitals and physician practices, changes very little about that.

The asymmetry that remains

Payers have published their denial rates. The overturn rates published something else: that most contested denials are winnable, and that the industry's economics depend on most of them never being contested. Federal rules will put the request pipeline on APIs in January. Nothing in those rules automates the provider's side of the argument, and nothing in them touches the post-adjudication queue where laboratory denials actually accumulate.

The party that adjudicates at machine speed has already chosen its speed. For a laboratory, meeting it after the remit posts is now the decision that matters; the gate before the test was never the one that decided its revenue.

Sources: KFF, "Prior Authorization Metrics Provide New Insights into Insurer Practices, but Gaps Remain," August 13, 2026; Healthcare Dive and Healthcare Finance News on UnitedHealthcare's October 1 code lists, September 2, 2026, with the plan-level PDFs published on UHCprovider.com on September 1; xtelligent Rev Cycle Management on Epic's real-time authorization checks, August 17, 2026, and on R1's acquisition of Humata Health, August 18, 2026; Electronic Frontier Foundation, "New Records Reveal Problems with Medicare's AI Prior Authorization Experiment," September 8, 2026, and STAT, September 15, 2026; Becker's Payer Issues, "Health insurer CEO: 'Make prior authorization illegal,'" August 21, 2026; American Economic Liberties Project policy brief, August 2026. Physician-book figures are from Recovr's denial analytics for a national collection of more than 12,000 physicians in some sixty specialties and are model-identified, not realized recoveries.

 
 
bottom of page