Most prior authorization denials are never appealed. Here is what that costs a practice.

Most prior authorization denials are never appealed, and that is where a practice loses the most money. When denials are appealed, most of them get reversed. An analysis of 2025 data by Dr. Archelle Georgiou, a former UnitedHealthcare chief medical officer, found that 58% of appealed UnitedHealthcare prior authorization denials and 65% of Humana's were overturned. A denial that sits unworked is, more often than not, a payment the practice would have received.

Key numbers

• 58% of appealed UnitedHealthcare PA denials and 65% of Humana's were overturned in 2025 (AELP report, citing Georgiou).
• Nearly 60% of physicians do not believe an appeal will succeed, based on past experience (AMA 2025 survey, via AELP).
• Private insurers initially denied 70% of new brand-name prescriptions in 2025, up from 57% in 2021. After appeals, the denial rate fell to 24% (IQVIA, via AELP).
• Applying the overturn rates to denials nobody appealed, at $100 per claim, implies roughly $100 million a year in avoided costs across just UnitedHealthcare and Humana.
• 79% of physicians say prior authorization leads patients to abandon treatment (AMA 2025).
• In California workers' comp, Independent Medical Review overturned only 10.2% of denials in 2025 (daisyBill, citing the DWC IMR annual report).

Why do so many denials go unappealed?

Because the people who would file the appeal have learned not to bother. The AMA's 2025 survey found nearly 60% of physicians doubt an appeal will work. Only 16% say the "peer" on a peer-to-peer call is often or always qualified to judge the case. Physicians and staff already spend about 13 hours a week on roughly 40 prior authorization requests per provider. CMS puts the cost at about $34,000 and 700 hours per provider per year. When the next denial arrives, the rational individual choice is to let it go.

That choice is wrong at the portfolio level. Any single appeal might lose. Across a hundred of them, the published rates say most will win. Physicians are right about the experience and wrong about the odds.

The payer's incentive runs the other way. A denial that is never appealed costs the plan nothing. Georgiou ran the arithmetic. Take the overturn rates, apply them to the denials nobody challenged, assume $100 per claim, and two plans alone keep around $100 million a year. The denial is an opening bid, and most practices accept it.

What does the prescription data show?

The clearest picture comes from pharmacy. IQVIA found private insurers initially denied 70% of new brand-name prescriptions in 2025, up from 57% in 2021. After appeals, the rate dropped to 24%. About two-thirds of the initial denials did not survive a challenge.

Read that as a billing manager rather than a policy analyst. The payer's first answer is a filter, and it only holds if nobody pushes back.

What does unappealed leakage look like for a five-provider practice?

Here is a worked example. Two of the inputs are published. The rest are placeholders you should replace with your own numbers.

• Volume (published): about 40 PA requests per provider per week. Five providers, 50 working weeks: roughly 10,000 requests a year.
• Initial denial rate (your number): assume 8%. That is 800 denials.
• Average allowed amount per denied service (your number): assume $600.
• Share of denials you appeal today (your number): assume 30%. That is 240 appealed and 560 left alone.
• Overturn rate (published, as a starting prior): 60%, the midpoint of 58% and 65%.

The appeals you file recover about 240 × 0.60 × $600, or $86,400. The denials you did not file leave about 560 × 0.60 × $600, or $201,600, on the table. The leakage is more than twice what you are recovering, and it comes from work you are already set up to do.

Unappealed leakage equals denials not worked, times overturn rate, times average allowed amount.

One caution. Appealed denials may be the stronger cases, so the true overturn rate on the unworked pile could be lower than 60%. Use the published rate as a starting prior. Once you have 30 or 40 outcomes with a given payer, swap in your own number.

The patient side of the ledger does not show up in that math. The Commonwealth Fund found that among insured working-age adults with a PA denial in 2026, 41% said their care was delayed and 28% said a health problem got worse. Every unappealed denial is also a treatment that may not have happened.

What should a billing manager measure?

Most practices can tell you how many appeals they won. Far fewer can tell you how many denials they never touched. Start with these:

1. Denials received versus denials worked, by payer. The gap is the leakage denominator. If your system cannot produce this count, that is the first fix.
2. Overturn rate by payer and by denial reason, on the appeals you file. Compare it to the published 58 to 65%. A payer well below that range deserves a closer look at your packets. A payer well above it is telling you to appeal everything.
3. Days from denial to appeal filed. Filing windows close. A denial that ages past the window is not leakage you can recover.
4. Peer-to-peer outcomes and the reviewer's specialty. See what the AELP report means for billing teams for why this matters.
5. Decision deadline misses. A payer that misses its clock has handed you a procedural win. See prior authorization decision deadlines in 2026.

Report the first metric monthly to whoever owns the budget. "We won 71% of appeals" sounds like success. "We appealed 30% of denials" is the number the budget owner should be asking about.

How UtilReview tracks unappealed leakage

UtilReview turns every denial into a case whether or not anyone appeals it. Cases nobody has touched show up as the leakage report, with dollars at risk attached, and each one carries an expected recovery figure.

Until a practice has its own outcome history for a payer and service, published overturn rates like the 58% and 65% figures above serve as the prior. As outcomes land, the prior gives way to the practice's own rate for that payer, that CPT code and that denial reason. The queue is then ranked by expected recovery, so the highest-value unworked denials come first.

The AI drafts the appeal and checks it against the records. A person reviews it and sends it. Nothing goes to a payer without a human signature.

Count the denials you did not work last month. Multiply by 60% and by your average allowed amount. That is the budget line nobody is reporting, and it is usually bigger than the one they are.