"Refusal of Recovery" — Majority Staff Report on Medicare Advantage

An investigation into the three largest Medicare Advantage insurers found they denied post-acute care (skilled nursing, inpatient rehab, and similar) at far higher rates than their overall denial rate — UnitedHealthcare and CVS around 3 times their overall rate, Humana around 16 times.ruleSenate PSI Majority Staff Report (Oct 17, 2024): "Refusal of Recovery" — how MA insurers denied post-acute care

The investigation reviewed the insurers' own internal documents and found evidence that predictive algorithms and AI-assisted tools played a role in flagging post-acute care requests for denial, raising questions about whether individualized clinical review was actually happening for every case.

Post-acute care denials (skilled nursing, inpatient rehab, home health) aren't just common — the investigation found they're specifically where the largest insurers denied care at the most disproportionate rates, which is exactly the category most of CaseWhy Appeals' contested-item pages cover.

If an algorithm or automated tool played any role in a denial, that's worth asking about directly — the underlying Medicare coverage rule still applies regardless of what tool the plan used to reach its decision, and an appeal that names the real rule doesn't depend on knowing how the denial was generated.

This was a bipartisan Senate investigation, not a partisan document, and its findings were based on internal company documents the subcommittee obtained directly from the insurers themselves — not just public statistics — which is part of why it drew significant attention when it was published.

The report's title, "Refusal of Recovery," refers specifically to its central finding: the pattern of denying post-acute care — the kind of care that helps a patient recover after a hospital stay — at rates far higher than the insurers' own general denial rate for other kinds of care.

The subcommittee's own investigative authority let it obtain documents that wouldn't otherwise be public, which is part of why this report surfaced details (like the specific role of predictive tools) that earlier public-data-only analyses hadn't been able to show as clearly.