In June 2024, Medicare Advantage plans denied 12% of requests to admit a patient to a skilled nursing facility (SNF). Enrollees and their providers appealed 18% of those denials. Of the appeals filed, 95% were overturned in favor of the enrollee (Source: HHS Office of Inspector General, OEI-09-24-00331, June 2026). Those three numbers describe two separate decisions: the plan’s initial authorization decision, and the decision by an enrollee or provider to pursue an appeal. This article is about the second one. Why so few post-acute prior authorization appeals get filed — and what it would take to tell the contestable denials from the rest, while it still matters.
On July 14, 2026, Senators Richard Blumenthal and Josh Hawley sent records requests to UnitedHealthcare, Humana, and CVS Health. They asked how many post-acute denials each company issued, how many were appealed, and how many were overturned. By facility type, by year, back to January 2023 (Source: Senate Permanent Subcommittee on Investigations, July 14, 2026).
Most coverage focused on a different question in the same letter: whether artificial intelligence is producing the denials. That is the right question for Congress. It is not the question that changes anything in your building next quarter.
What Does a 95% Overturn Rate Mean?
A note on terms: throughout this article, “appeal” follows OIG’s terminology for a challenge brought by a beneficiary or provider after an initial prior-authorization denial. Individual plans may label earlier review steps differently — reconsideration, peer-to-peer review, or resubmission with additional documentation.
It means the initial decision frequently did not survive review once an appeal brought the case back in front of the plan. HHS-OIG examined 19 Medicare Advantage organizations (MAOs). SNF denial rates ranged from 0.4% to 23% — a wide spread on the same benefit, under the same federal rules. OIG’s own reading is deliberately careful: the extremely high overturn rate indicates that some enrollees were initially denied medically necessary care, and it raises concerns about the much larger group of denials that never reached appeal (Source: HHS-OIG, OEI-09-24-00331, June 2026).
That distinction matters, and this article holds to it. A 95% overturn rate among appealed cases does not prove that every unappealed denial would have been reversed. Providers may appeal their strongest cases first, which would produce exactly this pattern. What the number does establish is that the cases that reached appeal were far more contestable than an 18% appeal rate suggests.
What the Contractor Data Shows
One detail is harder to explain away. The contractor naviHealth processed roughly half of all SNF admission requests in OIG’s review, denied 14% of them, and saw 97% of those denials overturned when appealed. OIG said the pattern raises concerns about whether MAOs provide appropriate oversight of the contractors they use (Source: HHS-OIG, OEI-09-24-00331, June 2026).
How the Other Post-Acute Settings Compare
The other post-acute settings run cooler. Long-term care hospital (LTCH) denials were overturned 36% of the time, inpatient rehabilitation facility (IRF) denials 43%. But IRF overturn rates ranged from 14% to 86%, depending on which plan issued the denial (Source: HHS-OIG, OEI-09-24-00330, June 2026).
| Post-acute setting | Appealed denials overturned | What varies by plan |
|---|---|---|
| SNF — skilled nursing facility | 95% | Denial rates ranged from 0.4% to 23% |
| IRF — inpatient rehabilitation facility | 43% | Overturn rates ranged from 14% to 86% |
| LTCH — long-term care hospital | 36% | Plan-level range not reported |
That roughly sixfold variation across MAOs warrants scrutiny of how different plans make and review these determinations. The two reports cover different reviewed populations and settings, so these figures are not a direct ranking of denial rates across care settings.
UnitedHealthcare has said it does not use AI for clinical denials and that the reports take “a narrow and incomplete view” of its review process. That question belongs to the Subcommittee. The overturn rate belongs to the provider either way, because whatever produced the denial, an appeal is one formal mechanism for testing whether that decision holds on review.
Why Do So Few SNF Denials Reach Appeal?
Not every denial should be appealed, and the OIG data cannot tell us how many of the unappealed ones would have been reversed. Post-acute prior authorization appeals go unfiled for reasons that have nothing to do with the merits:
- the patient discharges by another route;
- the family declines to pursue it;
- a notice deadline passes;
- nobody on the provider side holds the role or delegated authority to file;
- the clinical picture changes, or placement availability does;
- documentation genuinely doesn’t support the request.
But an 18% appeal rate sitting next to a 95% overturn rate is a strong reason to look hard at how appeals get triaged.
Triage happens under time pressure, against a queue. Someone has to locate the functional assessments, therapy notes, medical-complexity evidence, physician-documented discharge criteria, and the plan’s stated rationale. Quickly enough for the decision to still be worth making.
When that evidence is scattered across the electronic health record (EHR), discharge-planning notes, and payer correspondence, the choice defaults to what the worklist actually displays: payer, status, deadline, dollar exposure.
This is partly a staffing and process problem, and it would be dishonest to pretend otherwise. It becomes unmanageable when the evidence that makes an appeal defensible is fragmented across three systems that don’t meet. The gap isn’t that teams lack a reason to appeal. It’s that the denial record and the clinical record supporting it are rarely in front of the same person at the moment the decision has to be made.
What Does the Delay Cost While an Appeal Runs?
It is paid in beds and time, and the accounting is indirect. HealthLeaders’ analysis of the June 2026 OIG reports reported a median wait of roughly five to six days for an appeal decision on post-acute cases, with nearly 17% waiting 10 days or longer — secondary reporting, not a figure confirmed in the OIG report text. Those delays can hold a patient in an acute-care bed after the treating team has already sought post-acute placement, consuming capacity while the authorization dispute is resolved.
For a critical access hospital running a handful of med-surg beds, that isn’t a line item. It’s a bed a new admission can’t have. The denial surfaces in the revenue cycle; the capacity cost surfaces in throughput; and the two are usually owned by people who don’t sit in the same meeting.
What Would Have to Change Architecturally?
The denial record has to arrive joined to the clinical evidence, not merely near it. Three concrete moves:
Join at receipt. When an adverse determination lands, pull the encounter’s functional assessments, therapy notes, and documented discharge criteria into the same record automatically. Not searchable — attached. This can be implemented through EHR integration using FHIR (Fast Healthcare Interoperability Resources) where the required structured data and documents are available, with supplementary document and workflow integration where they are not.
Rank by evidence readiness, not only by amount. Order the queue by how completely the record already supports medical necessity. A $4,000 denial with complete therapy documentation may be a better use of an hour than a $22,000 denial missing the physician’s discharge rationale. Most queues can’t express that difference today, so they don’t act on it.
Keep the trail. Every appeal your team assembles is also a record of what a given plan accepted and what it rejected. Structured, that becomes an input to denial prediction and, separately, part of the audit trail you already owe. One record, three uses.
None of this needs a new platform bought off a shelf. It needs the utilization-management record and the clinical record to be joined at the point where someone decides whether to fight. That is the same architectural move behind an automated appeals pipeline — applied one step earlier, at triage rather than at assembly.
Build It or Subscribe to It?
That depends on how closely your workflow matches the one the product assumes. There are packaged denial-management and prior-authorization products on the market, several sold as monthly subscriptions inside your EHR vendor’s own marketplace. They work. Where a hospital’s process matches the model they were built around, buying is the right call and building would be waste.
Where a Packaged Connector Stops Reaching
The mismatch shows up in the seams. A packaged product ingests the determination formats it was designed for, and reads the evidence its connectors know how to read.
Then it meets your building:
- therapy documentation;
- the functional assessment instrument you actually use;
- case-management records;
- your payer mix;
- the two systems you kept through the last merger.
Those specifics decide whether an appeal is assemblable — and they are exactly what a general product has to average over.
So the question isn’t build versus buy in the abstract. It’s narrower, and it’s answerable. How much of the evidence that decides your appeals sits somewhere a packaged connector doesn’t reach? And what does it cost per month to keep closing that gap by hand?
Where the answer is “most of it,” this is a build. Interoperability between the systems you already run, specified around your configuration rather than an average one, and owned by you when it’s done. Not a template with your logo on it, and not a capability you rent back every month.
This is the part of the work Itirra is built around. Better to be specific about what that means than to claim it.
The configurations a packaged product has to average over are the ones we specify against:
- a rehab module that never exposed a shareable document;
- an assessment instrument living in a proprietary flowsheet;
- the second system retained through a merger;
- payer determinations that still arrive as PDFs.
Working those out, and building what doesn’t exist on the market for them, is the engagement — not a customization tier on top of a template.
What Changes in Year Two
There's a second-order difference that shows up later. A packaged product changes when its roadmap says so. A system specified around your configuration changes when you need it to. Payer alters its determination format, you adjust. Retiring a source system, adding a service line — same answer. Customizable after delivery is not a feature list; it's who controls the schedule.
Why This Isn’t a Payer-Blame Argument
Because the plans’ conduct is already being examined by someone else, and your appeal rate isn’t. The Subcommittee’s requests are outstanding. CMS has three open OIG recommendations on SNF denial patterns with updates expected December 7, 2026 (Source: HHS-OIG recommendation tracker, June 2026). None of that resolves inside this fiscal year. The 82% is the part of the process a hospital can measure locally: which denials reached its teams, which ones reached appeal, why the others did not, and whether the clinical evidence was available when triage happened.
Frequently Asked Questions
What percentage of Medicare Advantage SNF denials are overturned on appeal?
95% of the ones that reach appeal. In June 2024, the 19 Medicare Advantage organizations reviewed by HHS-OIG overturned 95% of appealed skilled nursing facility admission denials in favor of the enrollee. Enrollees and providers appealed 18% of denials (HHS-OIG, OEI-09-24-00331, June 2026).
How do post-acute denial and overturn rates compare across SNF, IRF, and LTCH?
In the OIG’s June 2024 reviews, MAOs denied 12% of SNF admission requests and overturned 95% of appealed SNF denials. For LTCHs, 36% of appealed denials were overturned; for IRFs, 43%, with IRF overturn rates ranging from 14% to 86% by plan. The two reports use different reviewed populations and settings, so these figures should not be treated as a direct ranking of denial rates across care settings (HHS-OIG, OEI-09-24-00330 and 00331, June 2026).
Did the Senate subpoena Medicare Advantage insurers over AI denials?
No. On July 14, 2026, Senators Blumenthal and Hawley sent letters requesting records from UnitedHealthcare, Humana, and CVS Health, with a July 28 response deadline. Letters requesting documents are not subpoenas, and no subpoena has been reported as of this writing.
Why aren’t more post-acute denials appealed?
The OIG data does not establish why the other 82% of SNF denials never reached appeal, or whether they would have been overturned. In practice, post-acute prior authorization appeals are constrained by filing deadlines, documentation completeness, discharge planning, enrollee preference, whether the provider holds the role to file, staffing, and the availability of an alternative care path. The actionable step is for a hospital to measure its own appeal rate and examine how evidence is assembled at triage.
Should a hospital buy a denial-management product or build its own integration?
It depends on how much of the evidence that decides an appeal sits somewhere a packaged connector can reach. Packaged products, including those sold as monthly subscriptions inside EHR vendor marketplaces, are built around an assumed workflow and work well where a hospital matches it. Where therapy documentation, functional assessments, case-management records, or systems retained through a merger fall outside that assumption, the integration has to be specified around the organization’s actual configuration — which is a build, and one the organization owns.
What does an appealed post-acute denial cost operationally?
An appeal consumes utilization-management, case-management, clinical-documentation, and revenue-cycle time, and it can delay the transition to post-acute care. A HealthLeaders analysis of the June 2026 OIG reports cited a median wait of five to six days for an appeal decision, with nearly 17% of cases exceeding 10 days — secondary reporting, and cited as such.
So Here’s the Question
Do you track your own appeal rate on post-acute prior authorization denials — not your win rate, your appeal rate?
Custom integration is what Itirra does — not as a fallback when a product doesn’t fit, but as the practice. We build the interoperability layer between the EHR, the utilization-management record, and the clinical documentation that decides an appeal, specified against your systems and your configuration. It stays customizable after delivery, because payer rules and source systems keep moving, and you own it rather than licensing it back monthly. Let’s talk about your project — contact Itirra.
Sources
- HHS Office of Inspector General — “Medicare Advantage Organizations Overturned Nearly All Appealed Prior Authorization Denials for Skilled Nursing Facility Admission”, OEI-09-24-00331, issued June 8, 2026 (June 2024 data)
- HHS Office of Inspector General — “The Three Largest Medicare Advantage Organizations Denied Requests for Long-Term Acute Care and Inpatient Rehabilitation at Some of the Highest Rates”, OEI-09-24-00330, issued June 8, 2026
- U.S. Senate Permanent Subcommittee on Investigations — Letter from Ranking Member Blumenthal and Sen. Hawley to Humana, July 14, 2026
- Office of U.S. Senator Richard Blumenthal — “Blumenthal & Hawley Press Medicare Advantage Insurers About Refusal of Care for Vulnerable Seniors”, July 15, 2026
- HealthLeaders Media — “OIG Reports Expose Flaws in Prior Authorization Processes for Post-Acute Care” — secondary source, cited only for the appeal-decision wait times
- Becker’s Payer Issues — “Senators press UnitedHealth, Humana, CVS on AI-driven Medicare Advantage coverage decisions” — secondary source for UnitedHealthcare’s stated position
- 86% of Denials Are Avoidable. Most Hospitals Are Still Appealing Them. — Itirra
- Building an Automated Appeals Pipeline: From RCM Cost Center to Revenue Engine — Itirra
- The AI Rules You Already Signed — Itirra
- The 20-Minute Referral Gamble — Itirra