Prior Authorization: I Killed It Once. They Brought It Back.

Prior Authorization: I Killed It Once. They Brought It Back. Now the Data Shows Why.

A former health plan CMO followed the money — and the numbers are worse than the industry wants you to know.

Archelle Georgiou, M.D.'s Substack

Archelle Georgiou

Apr 11, 2026

In 1999, I led an effort inside UnitedHealthcare to eliminate prior authorization (PA).

I was Chief Medical Officer. We believed the process was inefficient, burdensome, and — most importantly — a barrier between physicians and their patients. We dismantled much of it. And it worked: we redirected the savings from administrative review into care coordination, reducing hospital readmissions and improving outcomes. We weren't alone — while no other insurer took as dramatic a step, many loosened their rules and made it easier for patients to get care without the hassle.

In the 2000’s, medical costs rose — industry wide — and prior authorization crept back in at UnitedHealthcare and other insurers. Moreover, advances in automation and technology made it increasingly economical for insurers to expand the list of services requiring prior authorization, and today it has become one of the most pervasive and frustrating features of American healthcare.

Now, more than 25 years later, we have real data for the first time about the true scope of prior authorization. And it is more troubling than the industry wants to acknowledge.

The Industry’s Argument: “It’s About Quality”

Health insurers will tell you that prior authorization is a quality tool. AHIP, the industry’s trade group, frames it as a patient safety opportunity — a check against unnecessary care, dangerous drug interactions, and treatments that deviate from clinical evidence. There is a kernel of truth in that.

But it is not the full story.

For the First Time, We Have Real Data — And I Analyzed It

Until recently, the prior authorization debate has been fought largely with anecdotes, surveys, and industry-selected statistics. That changed on March 31, 2026.

For the first time, health plans were required by CMS to publicly submit PA data across Medicare Advantage, Medicaid, and exchange (ACA) plans.

I downloaded the Medicare Advantage (MA) data for UnitedHealthcare and Humana — two of the largest MA insurers in the country — and analyzed it myself. What I found should concern every patient, every physician, and every policymaker paying attention to this issue.

The Reality: 1.75 Million Patients Denied in Just Two Plans

Combined, UnitedHealthcare and Humana denied care to 1.75 million people – net AFTER appeals. The denial rate for Humana was 7% and for UnitedHealthcare 12%. Conversely, the approval rates were 93% and 88% respectively.

UnitedHealthcare made their data more easily accessible for download and analysis. (Thank you!) They also added an “easy to read” infographic on their website reporting an approval rate of 95.4%.

I kept getting 88%.

I recalculated. I rechecked. I went back to the raw data. I found the reason for the discrepancy in a footnote — in small print beneath the infographic — explaining that UnitedHealthcare had excluded certain categories of prior authorization requests that CMS explicitly required to be included in the public reporting.

The difference between a 95% approval rate and an 88% approval rate is not a rounding error. It is the difference between a 5% denial rate and a 12% denial rate — a fundamentally different picture of how often patients are told no.

Regulators specifically designed these reporting requirements to capture the full picture, including all prior authorization requests across all arrangements. Presenting a selective subset to consumers, in my view, is misleading.+

The Mechanism Nobody Talks About: Denials as Strategy

At UnitedHealthcare, only 10% of denials are appealed. At Humana, only 3% are. But when denials are appealed, at UnitedHealthcare, 58% of appealed denials are reversed. At Humana, 65% are reversed.

When patients or physicians push back, they win more often than they lose. That is not a quality system working as intended. That is a system betting on inertia.

Patients win when they appeal. But, most never do.

If all the unappealed denials had been challenged — and the same overturn rates applied — I calculated that nearly 991,000 additional patients would have received the care their care providers originally recommended.

Instead, the denials stand and the care was not received. Not because they are medically correct. But because the appeals process is too complex, too slow, and too opaque for most people to navigate. The friction is not a flaw in the system. It is the strategy.

Follow the Money — All the Way to Earnings Per Share

Here is the most important statistic in this entire discussion.

For those 991,000 patients, the denials are a clinical problem for their doctors — and a financial strategy for their insurer.

Stick with me here.

Using a conservative estimate of $100 per claim — applied to each denied case at the same overturn rate seen in actual appeals (58% for UnitedHealthcare; 65% for Humana) — the numbers are striking. Combined, across just two Medicare Advantage plans, wrongly denied and unappealed claims generated an estimated $100 million per year in avoided costs.

·        UnitedHealthcare’s unappealed but overturnable denials represent approximately $58 million in avoided costs annually, contributing roughly $0.05 per share to earnings.

·        Humana’s represent approximately $42 million, contributing roughly $0.27 per share — a more consequential figure given the intense financial pressure Humana is currently under.

This is not a byproduct of an imperfect system. This is a structural feature of the business model — and for the first time, it is measurable.

The Voluntary “Reforms” That Weren’t

Last year, health plans announced voluntary commitments to reduce prior authorization burdens. One of those commitment was to decrease the number of services requiring prior authorization. On the surface, it looked like progress. Maybe. But it was also a great strategy to avoid regulation.

Soon after the March 31 deadline to submit their data, AHIP issued a press release announcing that participating plans had reduced prior authorizations by 11%. Reduced relative to what? The denominator matters enormously.

Apparently, this reduction was calculated based on a survey of health plans. As of the writing of this article, I can’t locate the survey methodology or results beyond the headline.

A Better Model Already Exists

Earlier this week, at the University of Minnesota’s Convene Conference, I moderated a conversation with Dr. Sachin Jain, CEO of SCAN Health Plan. An analysis of SCAN’s data shows a prior authorization approval rate of approximately 98%. Dr. Jain offered a perspective that every policymaker should hear: health plans should stop doing prior authorization entirely. Instead, providers and health systems should build the internal accountability structures to monitor and regulate themselves — without inserting an insurer between the physician and the patient.

That model is already working, at scale, in a real health plan. The industry argument that prior authorization is indispensable to quality is not supported by SCAN’s results.

What Needs to Happen

Prior authorization is not going away overnight. But for the first time, we have the data to hold the system accountable.

·        Approval rates depend entirely on how you define the denominator

·        Denials are overturned more than half the time — when challenged

·        Most denials are never challenged

·        The financial benefit of that inaction is now calculable

Congress is actively examining this issue. CMS now has a reporting framework. The data exists. What is needed now is the will to act — and the refusal to accept industry-defined metrics as a substitute for real transparency.

I tried to eliminate prior authorization once, from inside the largest health insurer in the country. We showed it could be done, and that outcomes were better without it.

Maybe the mistake was not that it came back. Maybe the mistake was that we never made the case clearly enough — backed by data — about what prior authorization really is: a cost-control mechanism that has learned to dress itself as clinical quality.

Now we can make that case. And the data makes it for us.