For years, prior authorization operated on vibes. Payers called it a safety check; providers called it a wall; neither side had public numbers, and every policy argument ended in competing anecdotes that could not be checked. That changed when federal transparency rules forced insurers to publish their prior authorization metrics, and KFF's August 13 analysis offers the first systematic read of the new data. Across the 14 largest insurers, at least one in eight standard prior authorization requests was denied in 2025. Average denial rates ran 12 percent in Medicare Advantage, 14 percent in Medicaid managed care, and 18 percent in the ACA marketplaces - with insurer-level rates ranging from 2 percent to 25 percent.
The variation is the story. UnitedHealth posted the highest Medicare Advantage denial rate at 17 percent; Independence Health Group topped Medicaid at 23 percent; Centene led the marketplaces at 25 percent. Same services, same patients on paper, a twelve-fold spread in how often the front door stays shut. Denial rates are a policy choice, and for the first time, every payer's choice is on the record.
From public data to leverage
Provider organizations sit on the demand side of that disclosure without yet using it. Three practical uses:
- Benchmark your own rates against the public ones. If your denial rate with a payer runs far above its published average, you have evidence of outlier treatment - the starting point for a payer relations conversation with teeth.
- Sequence your automation and staffing. A payer at the top of the denial distribution justifies heavier pre-submission scrutiny and a standing appeal workflow. A payer at 3 percent does not need the same machinery. Effort should follow the data, and until this year the data did not exist at the payer level.
- Argue policy with facts. The 2-to-25 percent spread is now citable in comment letters, contract negotiations, and state advocacy. Prior to this year, that argument rested on anecdotes.
What the data still hides
KFF is careful about the gaps, and providers should be too. The published metrics are annual and payer-level: they do not break denials down by service line, by reason code, or by provider specialty - which is exactly the resolution billing teams need. A payer can post a flattering 5 percent average while denying a majority of post-acute requests, as earlier KFF analyses of MA prior auth showed. Treat the federal data as a floor for scrutiny, not a ceiling.
One more gap worth noting: the published data covers only the insurers that report it, and self-reported compliance metrics have a way of flattering the reporter. The 2-to-25 percent spread is likely wider than published, because the plans with the worst behavior have the most incentive to explain it in footnotes. Read the numbers as a lower bound on payer variance, and treat any payer that markets its transparency compliance while disputing your denial data with appropriate skepticism.
The same caveat applies to your internal data. If your organization tracks denials only in aggregate, you cannot see whether the payer's published rate matches your experience at the specialty level. Per-payer, per-service tracking is what turns this transparency exercise from journalism into operations.
Worth noting what this dataset changes permanently: prior auth denial rates are now a recurring publication, not a one-time reveal. Each year's release will show whether the voluntary reduction commitments moved the numbers, whether outlier insurers faced consequences, and whether aggregate rates drift up or down. Providers who build per-payer benchmarking into their operations this year will have a longitudinal record of payer behavior that no annual average can match - and that record compounds in value every year it is kept.
The takeaway
Prior authorization denial rates are now public, measurable, and wildly uneven across insurers - and that changes the posture providers can take. Aggregate anecdotes gave way to named, dated numbers that belong in every payer negotiation and automation decision from here forward. The remaining work is matching that resolution inside your own operation: per-payer, per-service denial tracking with an appeal workflow scaled to the payer's actual behavior. That is the measurement layer QHB runs daily, so the next set of published numbers finds you holding better data than the payers expected.