The case is In Home Health, LLC v. Kennedy, decided by the Sixth Circuit Court of Appeals on July 27, 2026. The case involved a Medicare-certified hospice that underwent post-payment review. A Medicare contractor initially determined that 252 of 374 reviewed claims did not meet Medicare coverage requirements and sought repayment of almost $1 million. The hospice appealed, and some of those determinations were overturned along the way, but a significant number remained denied. Eventually, the case made its way through the administrative appeal process and into federal court.
Here’s what makes this decision so interesting for hospice.
The Sixth Circuit did not say the patients were eligible. In fact, the court found that substantial evidence supported the ALJ’s decision denying coverage for the claims that remained in dispute. But the court also said something incredibly important: A determination that Medicare should not have covered a hospice claim does not necessarily mean the hospice is financially responsible for paying that money back.
Medicare has what is essentially a limitation-on-liability safe harbor under 42 U.S.C. § 1395pp. Congress specifically extended that protection to hospice. Under certain circumstances, even when Medicare ultimately determines that a patient was not terminally ill for purposes of coverage, the hospice may still be protected from financial liability if it did not know, and could not reasonably have been expected to know, that Medicare would not cover the care.
That distinction matters.
In this case, the ALJ essentially concluded that the hospice should have known the claims would not be covered because Medicare providers are expected to know the applicable CMS guidance, LCDs, manuals, contractor notices, and accepted standards of practice. The Sixth Circuit rejected that approach. The court explained that simply knowing an LCD exists is not enough to answer the safe-harbor question. The real question is whether the hospice, acting in good faith, reasonably could have interpreted the applicable Medicare guidance as supporting eligibility for that particular patient based on the clinical information available at that time.
That is a much more realistic way to look at hospice prognostication.
Anyone who has worked in hospice for any length of time knows that determining a six-month prognosis is not an exact science. In fact, the court acknowledged that very point. Hospice eligibility requires clinical judgment, and the LCD itself contains multiple factors that have to be considered together. Two clinicians can sometimes look at the same profoundly debilitated patient and reasonably reach different conclusions about prognosis.
The court also discussed something I have been teaching hospice nurses for years: stability does not automatically equal ineligibility.
Once hospice interventions are working, we should expect some symptoms to stabilize. That is the point of hospice. Pain may improve. Dyspnea may become better controlled. Edema may decrease. Anxiety may improve. A patient may even have an unusually good day. None of those things, standing alone, necessarily means the patient’s terminal prognosis has disappeared.
This is also why I have such a problem with teaching nurses that every note must “paint the picture of decline.” Most hospice patients do not demonstrate measurable decline from visit to visit. What we absolutely can do at every visit is document the clinical findings that continue to support eligibility. Then, at meaningful intervals such as recertification, admission, or a significant change in condition, we can appropriately compare and document decline when it is actually present.
That distinction becomes incredibly important when you look at this court decision.
Imagine a patient with advanced dementia who remains completely dependent for care, incontinent of bowel and bladder, unable to meaningfully communicate needs, unable to independently obtain nutrition, sleeping most of the day, experiencing recurrent infections, and requiring extensive caregiver support. Maybe the patient’s weight didn’t change dramatically during one certification period. Maybe one measurement remained stable. Maybe the patient had several good weeks because hospice interventions were working.
Years later, an auditor may look at that record and conclude there wasn’t enough measurable decline to support the six-month prognosis.
The question after this decision may not end there.
The first question remains: Was the patient actually eligible for Medicare hospice coverage? But if Medicare ultimately says no, there may now be another critically important question: At the time the care was provided, could the hospice reasonably and in good faith have believed that this patient met Medicare’s terminal-illness requirements based on the guidance and clinical evidence available?
Those are two very different questions.
That doesn’t give hospices permission to admit inappropriate patients or maintain patients who clearly no longer meet eligibility requirements. The Sixth Circuit repeatedly focused on reasonableness and good faith. This is not a loophole for poor clinical decision-making, and it is certainly not a substitute for good documentation.
It is also important to understand the geographic limitation. This is a decision from the Sixth Circuit Court of Appeals, which covers Kentucky, Michigan, Ohio, and Tennessee. That means its strongest precedential effect is within those states. Hospices elsewhere may certainly look to the court’s reasoning, but this should not be interpreted as though one federal appeals court just created a new nationwide Medicare rule.
Still, I think the reasoning matters far beyond four states because it recognizes something hospice providers have struggled with for years. Prognosis is inherently clinical and inherently uncertain. Medicare itself recognizes that some patients will stabilize or improve while receiving hospice services and may nevertheless continue to have a prognosis of six months or less.
For organizations facing ADRs, TPEs, SMRC reviews, or significant overpayment demands, I think appeal reviews need to become even more thoughtful. We should certainly continue challenging an eligibility denial when the clinical evidence supports doing so. But when a denial is upheld, organizations and their advisors should also be evaluating whether the limitation-on-liability provisions of §1395pp may apply based on the specific circumstances of the claim.
And this brings me right back to documentation.
A legal argument years later is only as strong as the record that existed when the decision was made.
If your physician reasonably believed Patient had a prognosis of six months or less, why? What clinical findings supported that judgment? What functional limitations existed? What disease-specific findings were present? What nutritional compromise, cognitive impairment, recurrent infections, symptom burden, ADL dependence, or other factors contributed to that prognosis? What was the entire clinical picture at that moment?
Those things need to be in the record.
And Now We Have Another Documentation Challenge: AI
I love technology. I built a technology company around making hospice documentation easier. I think AI has enormous potential in healthcare, and I believe we’re only beginning to understand what it will eventually allow us to do.
But I’m also going to say something that hospice leaders need to hear: Right now, we are seeing problems with AI-generated hospice documentation.
In the records we’re reviewing, we’re beginning to see AI-generated notes that introduce information that isn’t accurate, change diagnoses, restate normal findings in ways that can contradict eligibility, or create narrative that simply doesn’t match the clinical picture. In other words, AI can hallucinate, and when it hallucinates inside a hospice medical record, that is not just an annoying technology problem. It can become a reimbursement problem.
The danger is that an AI-generated note often sounds really good. It is polished. It is grammatical. It sounds clinical. That can make it very easy to skim instead of carefully reviewing every statement for accuracy.
Every clinician using AI-generated documentation needs to read every line of every note before signing it. The clinician’s signature makes that documentation part of the medical record. If AI changed the diagnosis, added an assessment finding that never occurred, contradicted another discipline, or documented something that undermines eligibility, “the AI wrote it” is not going to be much of a defense several years later when Medicare pulls the chart.
Hospice organizations also need independent oversight of their records. That doesn’t mean correcting every comma or telling nurses that every note has to sound exactly alike. It means having someone who understands hospice eligibility, documentation, Medicare review, and the patterns auditors are actually looking for regularly evaluate the record from the perspective of the person who may eventually be deciding whether you get to keep the money Medicare already paid you.
That is exactly what we do through The Amity Group’s Audit Protection Division. Our audit nurses spend their days reviewing hospice records. We work with organizations during routine ongoing audits, and we work with hospices when records are already under Medicare review or payment is at risk. Because we see these problems over and over across organizations, we know where contradictions tend to occur, what documentation raises questions, and what is likely to become important when someone outside the hospice eventually reads that record.
We also built DAHN, the Documentation App for Hospice Nurses, very intentionally not to write the nurse’s note using generative AI. The nurse remains in control of the documentation. DAHN provides hospice-specific wording, examples, and structure, including the HEARTS model, but the clinician chooses what applies to Patient and builds the note based on what was actually assessed. The technology helps the nurse find the words. It doesn’t replace the nurse’s clinical judgment.
I think that distinction is going to become increasingly important.
This new Sixth Circuit decision may give hospices another meaningful argument when Medicare later disagrees with a good-faith eligibility determination. But if we’re going to argue that the hospice reasonably believed Patient was eligible at the time, we need a medical record that shows why that belief was reasonable.
That is where this entire conversation comes full circle. Strong clinical judgment matters. Strong documentation matters. Independent oversight matters. Technology can help tremendously, but it cannot replace any of those things.
Because when Medicare comes back years later and asks you to defend the decision you made today, your defense is only as good as the records you have to back it up.
Protect the Record Before You Have to Defend It
The best time to find a documentation problem is before Medicare requests the record. The Amity Group’s experienced hospice audit nurses review records every day, including those already under Medicare review. We know where documentation breaks down, what creates eligibility concerns, and what can make an otherwise defensible patient difficult to defend. Our Audit Protection Division provides ongoing third-party review to identify those vulnerabilities early and help strengthen your records before reimbursement is at risk. Learn more at AmityStaffing.com.
Strong records also begin with the nurse writing the note. That’s why we created DAHN, the Documentation App for Hospice Nurses. DAHN is not generative AI and doesn’t write the note for the nurse. The nurse stays in complete control of every word in the medical record while DAHN provides hospice-specific wording, phrases, examples, and structure to make documentation easier and help support eligibility. Your nurses know their patients. DAHN simply helps them find the words to document what they see. Learn more at GetDAHN.com.
Because in the end, your defense is only as strong as the record you have to support it.
