A federal investigation found Medicare Advantage insurers reversed 97% of their own rehab denials on appeal. The reversal rate is not evidence the system self corrects. It is the system working as designed.
A 97 percent reversal rate is not a sign the Medicare Advantage appeals process works. The reversal rate reveals that the denials were never meant to survive scrutiny, and the appeals process is the part of the system that gets billed as a safeguard while functioning as a filter for the patients stubborn enough to use it.
That is the structural reading of the new federal report from the HHS Office of Inspector General, reported exclusively by STAT's Casey Ross and Bob Herman, which found that dominant Medicare Advantage insurers denied post-acute rehabilitation care to older and disabled Americans at higher rates than their peers, then reversed those denials 97 percent of the time on internal appeal. The reversals were not self-correction. They were the predictable outcome of a denial apparatus built to be defeated only by the people with the stamina to fight it.
At the center of the pattern is NaviHealth, a UnitedHealth Group subsidiary that uses algorithmic tools to evaluate post-acute care requests, including stays in skilled nursing and inpatient rehabilitation facilities. STAT's prior investigation into insurers' reliance on NaviHealth, a Pulitzer Prize finalist, documented how that subsidiary scrutinized rehab requests and shaped denial decisions across the post-acute landscape. The OIG report extends that arc from individual reporting to federal confirmation. The high denial rates that drew attention in 2023 are now on the record as a pattern at the insurers that matter most.
The patient population is the part the policy debate has been slowest to absorb. These are older and disabled Americans on Medicare Advantage, most of them over 65, admitted to post-acute rehab after a hospital stay for a hip fracture, a stroke, or a joint replacement. The clinical question in front of the reviewer is whether the patient is safe to go home, whether they can walk to the bathroom, whether the wound is healed. The administrative question in front of the insurer is whether the plan's algorithm says the patient is ready to leave. The OIG report quantifies how often those two answers diverge, and how often the insurer's own second look produces a different answer than its first.
The 97 percent figure also reframes a piece of conventional wisdom inside the Medicare Advantage debate. Insurers and their trade groups have long argued that high reversal rates on appeal are proof the system works. Denials get reviewed, the right answer wins, the patient is protected. The OIG data does not support that framing. An internal appeal that reverses 97 percent of denials is not a quality-control mechanism. The appeal is a one-shot gate that almost never fires correctly on the first pass, and the cost of getting it wrong falls on the patient, who either pays out of pocket, accepts an unsafe discharge, or finds a lawyer.
That is the design implication, and the OIG report hands it directly to regulators. The Centers for Medicare and Medicaid Services now has a quantitative basis to revisit the prior authorization rules that govern post-acute care, including the auto-escalation thresholds at which a denial must be sent for independent external review, the audit-trail requirements for algorithmic decision support, and the reviewer-independence rules that govern who gets to overturn a denial issued by a plan's own clinical staff. The OIG evidence base is the kind of artifact that triggers each of those conversations at once.
Congress has a parallel set of levers, most of which have been floating in legislative drafts for two cycles. The Improving Seniors' Timely Access to Care Act, which would codify prior authorization timelines and electronic standards for Medicare Advantage, has stalled in part because the prior evidence base was anecdotal. A federal report that names the dominant insurers, names the subsidiary, names the post-acute setting, and quantifies the reversal rate changes the terms of the debate. The same logic applies to the reviewer-independence provisions that have surfaced in post-acute care hearings, where the central question is whether a denial issued by an algorithm owned by the same plan that pays the claim can be reviewed by anyone other than the plan itself.
The industry response so far has been the standard one. UnitedHealth has told STAT, in coverage of the prior investigation, that NaviHealth's tools support clinical judgment rather than replace them, and that the subsidiary's role is to help clinicians and families make informed decisions about post-acute care. AHIP, the insurer trade group, has argued that prior authorization is a tool for protecting patients from unnecessary or unsafe services and that the appeals process exists to correct the small number of cases where the initial decision is wrong. The OIG report does not refute those framings. The report makes them arithmetically uncomfortable. A 97 percent internal reversal rate is not a small number of wrong initial decisions.
The OIG report also gives federal weight to a critique patient advocates have made for years. The cost of fighting a denial is not distributed evenly. Patients with a working phone, a working family member, a working primary care physician, and a working copy of the plan's appeal form are likelier to get a denial reversed than patients without those resources. The same insurer that issues the denial processes the appeal. The same subsidiary that supplies the algorithm supplies the reviewer. The OIG data does not say the system is rigged. The data shows the system is structured in a way that requires the patient to win a bureaucratic fight to receive the doctor-recommended care, and that 97 percent of patients who do win the fight turn out to have been right.
What to watch next is narrow and concrete. CMS's response to the OIG report will land first in the audit and enforcement pipeline, where the agency can flag the named plans for closer review of their post-acute denial patterns. The more durable change is structural. A rule that requires auto-escalation of post-acute denials to independent external review above a defined threshold, and a rule that requires disclosure of algorithmic decision support in the denial record, would both close the loop the OIG report exposes. Both have been on the regulatory wish list for years. The OIG report is the first artifact to put a federal number behind them.
The 97 percent reversal rate is the headline. The design implication is the story. Federal investigators did not find a system that catches its own mistakes. They found a system whose mistakes are caught only when the patient pays the cost of catching them, and whose internal appeals process is part of the product, not part of the safeguard.