The Black Box Cracks, but Only Partway

For years, the scale of impact of prior authorization operated with essentially zero transparency. Physicians submitted requests, payers approved or denied them, and the aggregate numbers stayed private. On March 31, 2026, that changed when insurers posted their first year of prior authorization metrics under the Centers for Medicare & Medicaid Services (CMS) Interoperability and Prior Authorization Final Rule (CMS-0057-F). KFF has now analyzed those disclosures, and its August 13, 2026 brief gives us the first real read on whether public reporting delivers accountability. The data covers 14 insurers representing roughly 71 million enrollees across Medicare Advantage, Medicaid managed care, and the Affordable Care Act (ACA) Marketplace. It confirms a pattern we have tracked for years. It also shows how much the rule was built to leave unseen.

What the New Public Data Shows

The first cross-market numbers put denials in plain view. Payers denied 12% of standard prior authorization requests in Medicare Advantage, 14% in Medicaid managed care, and 18% in the ACA Marketplace in 2025, according to KFF. When people appealed, payers reversed themselves often: 67% of appealed denials were overturned in Medicare Advantage, 47% in Medicaid managed care, and 43% in the ACA Marketplace. Across all three markets, denials were rarely appealed.

The plan-level spread is wide. Standard denial rates ranged from 5% (Elevance) to 17% (UnitedHealth Group) in Medicare Advantage, and from 3% (GuideWell) to 25% (Centene) in the ACA Marketplace. Median response times ran about 1 day for standard requests, well inside the regulatory maximums.

These figures sit on top of a longer trend. A companion KFF analysis released January 28, 2026 found that Medicare Advantage insurers made nearly 53 million prior authorization determinations in 2024, denying 4.1 million (7.7%), up from 6.4% in 2023. Only 11.5% of those denials were appealed, and 80.7% of appeals were overturned. The scale is not shrinking. It is growing with enrollment.

The Overturn Rate Is the Story

Read the two figures together: most appealed denials reverse, and almost no one appeals. In Medicare Advantage, payers overturned 67% of the small share of denials that patients contested in 2025, and the January companion analysis put the 2024 overturn rate at 80.7%. Either figure means most people who receive a denial never challenge a decision that would probably be reversed if they did. KFF describes overturned appeals as care that a provider ordered and that payers ultimately deemed necessary, delayed only by the added step of fighting the initial denial.

The clinical cost of that delay is documented. In the American Medical Association's physician survey, 93% of participating physicians reported that prior authorization delays care, 82% said it leads patients to abandon treatment, and 91% reported a negative clinical impact, as CANN has covered previously. When a payer reviewing paperwork overrides the judgment of the provider who examined the patient, that payer is making a medical decision. As CANN’s President & CEO Jen Laws stated plainly, “We do not pay insurers to practice medicine. We pay them to cover the cost of care.”

We do not pay insurers to practice medicine. We pay them to cover the cost of care.

The Gaps That Blunt Accountability

The rule was written with enough discretion that the resulting data documents the problem without letting patients act on it. KFF reports the metrics are aggregated across all medical items and services, so we cannot see which services get denied. Insurers report percentages rather than numeric counts, which means a low denial rate can still represent a large number of denied requests, and a reported zero cannot be distinguished from missing data. CMS published a template, but its use is not mandatory, so formats vary. Some ACA Marketplace insurers report national averages that tell a patient nothing about denial rates in the state where they actually live.

Gold-card programs distort the picture further. When a payer exempts its highest-approval providers from prior authorization, the providers left in the reported pool may have lower approval rates, which can make a plan's published denial rate look worse than its full book of business. We flagged the payer-adaptation problem in 2023, warning that gold-card programs could prompt insurers to expand prior authorization requirements, narrow formularies, and raise denial rates to disqualify providers. That instinct to watch for the workaround applies directly to reading these metrics. The design choices that make comparison difficult are not neutral. They protect the payer.

Usability compounds the problem. KFF notes that one-quarter of insured adults report difficulty understanding the terms their insurer uses, and that metrics are often hard to locate on insurer websites. A transparency rule that produces data patients cannot find or interpret is transparency in name.

The Suspended Equity Reporting Is the Deepest Hole

The metrics that would have shown who gets denied were removed before they produced anything. On June 16, 2025, the Trump Administration's CMS announced it would not enforce the CY2025 requirements that Medicare Advantage utilization management committees include health-equity expertise and publish plan-level analyses of approval and denial rates by dual eligibility, low-income status, and disability, as Georgetown's Center on Health Insurance Reforms documented. That reporting would have been plan-level and broken out by service category, not aggregated to the contract level like the metrics we now have.

The populations those reports would have covered face the highest stakes. CMS data showed that dually eligible Medicare Advantage enrollees, who are more likely to live in poverty and to have disabilities, faced denial rates up to twice those of non-dual enrollees despite submitting fewer requests. Even the suspended requirements would not have examined racial disparities. The rule we lost was already incomplete, and we lost it anyway.

What Comes Next, and What We Should Push For

CANN has long argued that utilization management tools like prior authorization and step therapy function as access barriers, and that the decision about whether care is necessary belongs to the patient and the provider who examined them, not the payer. We have made that case against payer limitations for over a decade, and most recently in calling to limit prior authorization and step therapy for chronic and life-threatening conditions. Measured against that standard, better reporting is a floor, not the objective. The objective is a system where payers stop practicing medicine, where that judgment sits with the people who examined the patient rather than a reviewer incentivized to deny care. Prior authorization, as the data now confirms, does not reliably sort appropriate care from inappropriate care. It taxes access until people give up. A process that denies 4.1 million Medicare Advantage requests, sees only 11.5% appealed, and reverses up to 80.7% of those appeals is not performing clinical review, as KFF's numbers show. It is performing attrition.

Some readers will note that prior authorization has a stated cost-control function, and that payers defend it as a check on low-value care. The evidence in these disclosures cuts against that defense. When the initial denial reverses two times out of three on appeal in Medicare Advantage, as the 2025 cross-market data show, the process is not screening out unnecessary services. It is delaying necessary ones and betting most patients will not fight back.

While we work toward eliminating these barriers, the pending rules and reporting mechanisms determine how much pressure we can bring in the meantime, and which of them produce usable data matters. The 2026 proposed rule (CMS-0062-P) would standardize denominators, require numeric counts, add a metric for requests that remain denied after appeal, and extend reporting to prescription drugs. Its comment period closed June 15, 2026, and a final rule is pending. It would not require breakout by service category. A July 2026 CMS template update clarified that burying metrics where ordinary navigation cannot reach them fails the "publicly accessible" standard, and that response times under one day must be reported in hours. CMS has also announced a Medicare Advantage pilot for plan-level and service-level data, which it intends to make mandatory starting with plan year 2027, though that data remains years away. States are moving faster: Massachusetts used a detailed data examination to prohibit prior authorization on high-volume services with high approval rates, and Iowa now requires public reporting. The operational floor CMS-0057-F already set matters too. Since 2026, payers must give a specific reason for each denial, and in Medicare Advantage and Medicaid managed care they must decide standard requests within 7 days and expedited requests within 72 hours (ACA Marketplace timeframes were unchanged), per CMS.

These are steps toward visibility, not toward removing the barrier itself. We should treat them as leverage and press on several fronts at once:

• Press CMS, as it drafts the final CMS-0062-P rule, to keep the numeric-count and standardized-denominator provisions and to add service-category breakouts so we can see which care gets denied. The comment period closed in June 2026, so this pressure now runs through direct engagement rather than the docket.

• Press CMS to hold the 2027 timeline for mandatory Medicare Advantage plan-level and service-level reporting rather than let it slip, and to reinstate the suspended population-specific equity reporting.

• Support statutory codification through the Improving Seniors' Timely Access to Care Act, including its requirement that plans report denials made through artificial intelligence.

Sunlight Is a Floor, Not a Fix

The opening of this data is a real gain. For the first time, patients, researchers, and policymakers can look at a given payer and ask how often it says no, how long it takes, and how often it is wrong when someone pushes back. Those are reasonable questions, and the answers should be uncomfortable for a lot of payers.

The reporting still cannot tell a person living with HIV or hepatitis C which plan will deny their care, for which service, or how often it denies care to people like them. It documents the pattern without dismantling it. The metrics we have prove that denials are common, appeals are rare, and reversals are the norm. That is the case for going further, not for declaring the problem addressed. The goal is not a better-labeled gate. The goal is unobstructed access to the care that patients and their providers, not their payers, decide they need.