AI and Insurance Denials: 80.7% of Appeals Succeed

Overview

Medicare Advantage insurers made nearly 53 million prior authorization determinations in 2024 — decisions about whether to cover care a doctor has recommended.

Two figures describe what happens when those decisions go against a patient. Of denials that were appealed, 80.7% were partially or fully overturned. Of denials overall, only 11.7% were appealed.

Put together, they describe a system in which the great majority of challenged denials turn out to have been wrong, and the great majority of denials are never challenged.

Against that backdrop, 84% of large health insurers surveyed use AI for operations including utilization review — and a Senate investigation found that post-acute care denial rates at major insurers rose between 54% and 108% after predictive AI tools were adopted.

53 Million Decisions a Year

Prior authorization requires a doctor to obtain insurer approval before providing certain treatments. Medicare Advantage insurers made nearly 53 million such determinations in 2024, up from around 50 million the previous year.

That volume is what makes automation attractive. Fifty-three million decisions cannot be individually reviewed by physicians without enormous cost, so insurers have strong incentives to process them at scale.

The overall denial rate sits in the region of 7 to 8% of requests. Applied to 53 million determinations, that means several million denials annually — each one a case where a treating doctor recommended care and the insurer declined to cover it.

80.7% of Appeals Are Overturned

The most important statistic in this area is what happens when a denial is challenged. In 2024, 80.7% of appealed prior authorization denials were partially or fully overturned.

An overturn rate above 80% is extraordinary for any decision-making process. It means that when a denial is examined more carefully, the original decision is found to be wrong four times out of five.

Federal oversight has found the pattern even starker in specific categories. A 2026 HHS Office of Inspector General report found Medicare Advantage organisations overturned nearly all appealed prior authorization denials for skilled nursing facility admission — a rate the inspector general explicitly flagged as raising concerns about the quality of the initial decisions.

But Only 11.7% Are Appealed

The overturn rate would be reassuring if appeals were routine. They are not. Only 11.7% of denials were appealed at all.

Roughly seven in eight denials are simply accepted. The reasons are readily understood by anyone who has faced one: appeals require paperwork, persistence and time, and they arrive when someone is ill and least equipped to fight a bureaucratic process.

The people least likely to appeal are also predictably those with least capacity to — the seriously ill, the elderly, people without an advocate, people unfamiliar with the system or the language it operates in. The barrier sorts by resources rather than by whether the denial was justified.

What That Combination Means

The two figures together produce an uncomfortable inference, and it is worth stating explicitly.

If 80.7% of appealed denials are overturned, and denials are not appealed on the basis of whether they were correct — but on whether the patient had the capacity to challenge them — then a similar proportion of unappealed denials would likely also have been overturned.

The unappealed denials are the majority. So the arithmetic points toward a large volume of care that was recommended by a doctor, denied by an insurer, would likely have been approved on review, and simply never happened. That is the cost of the appeal gap, and it does not appear in any denial statistic.

84% of Insurers Use AI

A 2024 survey of 93 large health insurers found 84% using AI for operational purposes including fraud detection and utilization review.

Not all of that is objectionable. Automating administrative processing, flagging genuine fraud and routing straightforward approvals faster all benefit patients — a system that approves clearly covered care in seconds rather than days is an improvement.

The concern concentrates on the specific application of algorithmic tools to denial decisions, where speed and volume interact badly with the appeal gap. A system that produces denials faster than patients can challenge them shifts outcomes even if its per-decision accuracy is unchanged.

Denials Rose 54% to 108% After AI

A 2024 Senate investigation examined what happened at UnitedHealthcare, Humana and CVS/Aetna after they began using predictive AI tools to manage post-acute care.

Denial rates for that care rose by between 54% and 108%. Post-acute denials ended up three to sixteen times higher than the same insurers' overall denial rates.

Post-acute care means skilled nursing, rehabilitation and recovery services after a hospital stay — typically for elderly patients recovering from surgery, stroke or serious illness. It is care where the consequence of a wrong denial is a patient discharged without adequate support, and where the population affected is least able to appeal.

One Insurer Went From 8.7% to 22.7%

The Senate findings included specific figures for individual insurers. One major insurer's post-acute services denial rate rose from 8.7% in 2019 to 22.7% in 2022.

Its skilled nursing home denial rate increased ninefold over a comparable period. These are not marginal adjustments to a threshold; they represent a fundamental change in how frequently recommended care was refused.

Correlation is not proof of causation, and denial rates can move for reasons including changes in utilization patterns or policy. But an increase of this magnitude, concentrated in the specific care category where predictive tools were deployed, and occurring across multiple insurers adopting similar tools, is a pattern that warrants the scrutiny it received.

The 90% Error Rate Allegation

One widely cited figure requires careful handling. A class-action lawsuit alleges that a major insurer's predictive tool, nH Predict, has roughly a 90% error rate — defined as the share of its denials later overturned on appeal.

That figure is an allegation made in litigation. It has not been established in court, the insurer disputes the characterisation, and the case is a legal claim rather than a finding of fact.

It is included here because it circulates widely and is frequently reported as established. The verified figures — 80.7% of appeals overturned across Medicare Advantage generally, and near-total overturn rates for skilled nursing admissions per the HHS inspector general — are serious enough without relying on a contested number from an unresolved lawsuit.

How to Appeal

Given that 80.7% of appeals succeed and only 11.7% are filed, the single most useful thing in this article is that appealing is worth doing.

Request the denial in writing with the specific reason and the clinical criteria applied. Insurers are generally required to provide this, and a denial that cannot articulate which criterion was not met is weaker on appeal.

Involve the treating doctor. A physician's statement explaining medical necessity addresses the substance of the decision, and many insurers offer a peer-to-peer review where the treating doctor speaks directly with a reviewing clinician.

Observe the deadlines, request an expedited appeal where delay would harm health, and know that Medicare Advantage denials can be escalated to an independent external review beyond the insurer itself. The process is designed to be tiring, and the overturn statistics indicate that persistence is usually rewarded.

Conclusion

Medicare Advantage insurers made nearly 53 million prior authorization decisions in 2024, and the two numbers that matter most sit either side of a large gap. Of denials that were appealed, 80.7% were partially or fully overturned. Of denials overall, only 11.7% were appealed.

That gap is where the harm accumulates. Denials are not appealed on the basis of whether they were correct, but on whether a patient had the capacity to fight — which means a substantial volume of care that would have been approved on review was never provided, and never counted.

AI has intensified the pattern rather than created it. Eighty-four percent of large insurers use AI operationally, and a Senate investigation found post-acute denial rates rising 54% to 108% after predictive tools were adopted, with one insurer's rate going from 8.7% to 22.7%.

This article summarises published research and official findings for general information. It is not medical, legal or insurance advice. Anyone facing a coverage denial should consult their treating physician and review the appeal rights specific to their plan.

Data Source and Attribution

KFFHHS Office of Inspector GeneralHealthcare Dive (Senate report)

Prior authorization volumes, denial rates, appeal rates and overturn rates come from KFF analysis of Medicare Advantage data for 2023 and 2024. Findings on skilled nursing facility admission appeals come from a 2026 report by the HHS Office of Inspector General. Figures on denial rate increases following adoption of predictive AI tools come from the 2024 US Senate investigation into Medicare Advantage insurers. Insurer AI adoption figures come from a 2024 survey of 93 large health insurers. The nH Predict error rate is an allegation in ongoing litigation and is identified as such.

FactsFigs reviews, cleans, and cross-checks every source dataset before shaping it into a data story. Each visualization is created and designed in FactsFigs Design Studio — an internal tool developed and owned by FactsFigs — and is the original work of a FactsFigs author, not an AI-generated copy of any existing graphic. Individual assets within a visual may or may not be produced with AI tools, but the design of the visual itself is solely FactsFigs' own.

This content is for information only and is not medical, legal or insurance advice. Coverage rules and appeal rights vary by plan and jurisdiction.

2026-07-20