Annual Performance Review — Medicare Advantage Prior Authorization System
Review Period: 2019–2024 | Classification: Administrative Competence Assessment
The industry's preferred characterization of the prior authorization process is that it protects patients from unnecessary care. The Bureau accepts this characterization on its face and notes, in the same filing, that the system reversed more than eight in ten of its own decisions when asked.
Both of these things are true. The Bureau presents them together as a matter of institutional completeness.
I. Volume Metrics
In 2024, Medicare Advantage insurers processed 50.2 million prior authorization requests. Of those, 4.1 million — 7.7 percent — were fully or partially denied.
Four million, one hundred thousand denied requests means 4.1 million individual determinations in which the system's first answer was: no.
The Bureau notes the volume without editorial embellishment. The number does not require it.
BUREAU NOTE: The Bureau's records show that prior authorization denials increased approximately 60 percent in absolute terms between 2019 and 2022 — from 2.1 million to 3.4 million — driven by both enrollment growth and a rising denial rate (5.7% to 7.4%). The Bureau's records also show that the denial rate then declined to 6.4% in 2023 before rising again to 7.7% in 2024. The Bureau regards an oscillating denial rate as evidence that the rate is being managed. The Bureau regards a managed rate as a policy choice. The Bureau does not adjudicate whose choice it is.
II. Quality Rating — Initial Determination
Score: 13%
In April 2022, the Office of Inspector General of the Department of Health and Human Services published Report OEI-09-18-00260. The OIG reviewed a stratified random sample of 250 prior authorization denials issued by 15 of the largest Medicare Advantage organizations during a one-week period in June 2019. Its finding: 13 percent of those denials met Medicare coverage rules. They should have been approved. They were not.
Thirteen percent is not a rounding error. In a system that issued 4.1 million denials in 2024, a 13 percent error rate produces a floor estimate of approximately 533,000 decisions wrong by the federal government's own coverage standard. The OIG identified the most common causes: insurers applied clinical criteria stricter than Medicare's rules, or rejected documentation the patient's own medical file contained.
The Bureau notes that "stricter than Medicare's rules" is a precise phrase. Medicare sets the coverage standard. The insurer sets a higher bar. The patient clears Medicare's bar. The insurer's bar does not move.
The record does not contain a discrepancy. The record is the discrepancy.
III. Quality Rating — Appeal Outcome
Score: 80.7%
In 2024, 12.2 percent of denied MA prior authorization requests were appealed. Of those, 80.7 percent were partially or fully overturned. In 2022, the reversal rate was 83.2 percent. Across all years examined by KFF's analysis of CMS data, more than eight in ten appealed denials were overturned.
The appeal reversal rate is the system's self-assessment. When a patient or provider contests a denial, the system reviews it and, in more than four out of five cases, concludes that the original determination was wrong.
The industry cites this rate as evidence of a robust patient-protection mechanism. The Bureau notes that a mechanism which corrects itself more than 80 percent of the time when challenged is also a mechanism that issued wrong decisions more than 80 percent of the time before being challenged. These are two descriptions of the same rate.
The wrong decision is still the first step. The appeal is not the safety net. The appeal is the second step of the process.
BUREAU NOTE: A factual note the Bureau considers worth filing separately: in 2024, approximately half of the physicians surveyed in a peer-reviewed NPJ Digital Medicine study reported that time constraints prevented them from appealing denials. Physicians reported spending an average of 13 hours per week on prior authorization work. The Bureau notes that the appeal mechanism's availability and its practical accessibility are two different metrics, and that the system reports only one of them.
IV. Product Assessment — The Algorithm
In October 2024, the Senate Permanent Subcommittee on Investigations released a majority staff report on Medicare Advantage denials. It found that UnitedHealth Group's skilled nursing facility denial rate increased ninefold between 2019 and 2022 — from 1.4 percent to 12.6 percent. Humana's long-term acute care denial rates grew 54 percent between 2020 and 2022. The three largest MA insurers were denying roughly a quarter of all post-acute care requests by 2022.
The increase coincided with the deployment of the NaviHealth nH Predict algorithm. UnitedHealth acquired NaviHealth in 2020, through its Optum division. The nH Predict tool was designed to assess prior authorization requests using predictive modeling. Between 2019 and 2022, UnitedHealth's skilled nursing denial rate increased by a factor of nine.
The Bureau notes that the algorithm did not produce a higher denial rate. It produced a measurable, documented, named output. The algorithm is not background to the story. It is the mechanism.
A ninefold increase has a purchase date.
V. Network Stability Assessment
At least 19 health systems — including Mayo Clinic, Johns Hopkins Medicine, Mass General Brigham, NewYork-Presbyterian, and UNC Health — exited or went out-of-network with major Medicare Advantage plans for 2026. Prior authorization denials and slow reimbursement are the documented primary drivers.
Mayo Clinic's departure from most UnitedHealthcare and Humana MA networks in Minnesota, Wisconsin, and Iowa took effect January 1, 2026, affecting approximately 1.6 million Medicare Advantage members in those states. Mayo cited reimbursement terms that do not align with its care model. Humana argued that Mayo's care costs are too high.
The Bureau notes that when the most prominent academic medical institutions in the country assess the prior authorization system and decline to participate, this constitutes a product review. The prior authorization is the product. The insurer retains the premium. The hospital leaves.
VI. Transparency Review
Effective in 2026, CMS required Medicare Advantage plans to publish aggregate prior authorization approval and denial metrics at the contract level. This represents a meaningful transparency enhancement.
On June 16, 2025, CMS suspended enforcement of two additional requirements from the same rulemaking: the requirement that utilization management committees include a member with health equity expertise, and the requirement that plans publish denial and approval rates broken down by dual-eligible, low-income, and disabled beneficiary status.
The system now reports that it denies care. It does not report which patients it denies care to.
The Bureau regards this as transparency as currently defined: the number exists. The names do not.
VII. Summary Assessment
The prior authorization system in Medicare Advantage processed 50.2 million requests in 2024. It denied 4.1 million. The federal government's own Inspector General confirmed that a meaningful share of those denials were wrong. The system reversed more than 80 percent of the denials that were contested. Fewer than 13 percent were contested.
The industry presents the appeal as the patient-protection layer. The Bureau accepts this framing and adds one observation: for a patient-protection mechanism to function, patients must use it. In 2024, 87.8 percent of patients whose denials were eligible for appeal did not appeal.
The protection was available. It was not practical.
The protection from the wrong decision is the ability to contest the wrong decision. The wrong decision is still the first step. For most patients, the first step is also the last step.
BUREAU NOTE: The Bureau of Prior Determination, Appeals and Outcome Reconciliation Division closes this review with the system's own most efficient summary: in 2024, 50.2 million prior authorization requests were processed. Four million were denied. Approximately 500,000 of those denials were appealed. Approximately 400,000 appeals succeeded — meaning the original denials were wrong. The remaining 3.6 million denials were not appealed. The Bureau does not know how many of those 3.6 million were also wrong. The system does not know either. The system was not required to find out. Annual review complete.
Bureau of Prior Determination, Appeals and Outcome Reconciliation Division — filing annual performance assessments on systems that work as described.