Hospice is not a uniform good
Most ACO leaders treat hospice as a fixed input: a service line patients enter near end-of-life that exists outside the ACO's strategic decisions. The number that gets watched is hospice utilization rate. The assumption is that more hospice is better, because more hospice means less hospital. The assumption is wrong on both counts.
Hospice quality varies enormously between providers, and the variation has economic consequences that dwarf most other levers in an end-of-life strategy. When we did the underlying analysis for Pearl Health's hospice network, the addressable spend in a representative 25,000-life ACO came in north of $20 million annually, larger than most ACOs' entire shared-savings opportunity. Almost none of it is being captured today.
There's a second reason this matters that doesn't show up in the dashboards. The kind of hospice a patient ends up with shapes whether their last weeks of life look anything like what they actually wanted. High-quality hospice supports patients to die at home, with their families, free of pain, on their own terms. Low-quality hospice, particularly the for-profit operators running benefit-abuse cycles, produces burdensome transitions, hospital deaths, and care experiences that bear no resemblance to what any patient or family would have chosen if they'd been asked.
This post is about both: why the financial stakes are this large, and why the patient-experience stakes are at least as serious.
What the data actually shows
The national hospice live-discharge rate has risen from 16.0 percent in FY 2020 to 19.0 percent in FY 2024. That national average obscures a sharp divide: for-profit hospices run at 22.4 percent live discharge, non-profits at 14.6 percent. In 2017, 472 hospices (462 of them for-profit) had live-discharge rates above 50 percent.
The Office of Inspector General has explicitly identified "abnormally high rates of still-living patients discharged from hospice care" as a fraud indicator. CMS doesn't set a numerical threshold, but the regulatory orientation is unambiguous: high live-discharge rates clustered around regulatory milestones (60-day, 180-day marks), correlated with approaching aggregate payment caps, and accompanied by burdensome transitions, are pattern signals that something is wrong.
Industry expert consensus places appropriate live-discharge rates at 20–30 percent depending on patient mix, treats 30–50 percent as warranting investigation, and treats rates above 50 percent as a strong indicator of inappropriate admission practices or active benefit abuse.
How the abuse cycle works
The mechanics are not complicated, which is why the problem persists.
A low-quality hospice, typically for-profit, often part of a roll-up, aggressively enrolls a patient who isn't truly terminal. The patient might have moderate Alzheimer's with a two-to-three-year prognosis, or a chronic condition that's stable enough to disqualify them from real hospice eligibility. Hospice gets enrolled anyway. Per-diem revenue starts flowing.
When the patient develops an acute issue that requires therapeutic care (a urinary tract infection requiring IV antibiotics, a pneumonia, a fall) the hospice revokes the benefit. The patient is admitted to the hospital. The acute episode is resolved. The patient is then re-enrolled in hospice. The cycle repeats.
Each revocation drives roughly $30,000 in post-discharge spend. The hospice keeps billing per-diem fees for the periods the patient is on service. The acute care costs land on the ACO. The patient experiences burdensome transitions in the worst period of their life, the kind of transitions that the patient and their family did not consent to in any meaningful way, because they were never given the chance to understand what they were signing up for.
A single Alzheimer's case run through this pattern over six months can generate $82,000 in combined hospice fees and acute care costs. Properly managed (a high-quality hospice declining inappropriate enrollment, a PCP managing the chronic conditions in place, an honest conversation with the family about what hospice is and isn't) the same patient's care comes in around $9,700, with a much better human experience. The differential per case is roughly $72,000.
This is not an unusual pattern. It's a business model.
What network curation looks like
The response is a curated hospice network: a defined set of high-quality hospice partners that ACO patients are preferentially routed to, with clear quality criteria and an explicit decision to exclude poor performers.
The criteria we used at Pearl:
A live-discharge rate below 15 percent, indicating appropriate admission practices and sustained engagement to end-of-life rather than revocation cycling. CAHPS hospice survey scores above the 80th percentile across the eight domains the survey covers: communication, timeliness, respect, emotional support, symptom management, family training, overall rating, and willingness to recommend. CMS quality ratings of four stars or higher. A balanced length-of-stay distribution that avoids both crisis-mode short stays (under seven days) and gaming-mode long stays (above 180 days, which often correlate with approaching aggregate payment cap manipulation).
These metrics, taken together, identify hospices whose financial model is built around sustained, high-touch care for genuinely terminal patients, not enrollment funnel optimization. They also identify hospices that consistently produce the kind of end-of-life experience patients and families would actually choose. The two things turn out to correlate strongly, which shouldn't be surprising: hospices that take care of patients well financially perform well too.
The negotiation that makes the network work
Network curation only generates ROI if it changes patient flow. Curating a list and not steering patients toward it is an academic exercise. The mechanics that make it work in practice:
The ACO communicates the curated network to its participating practices and care managers, with clear documentation of why each network hospice was selected. Pearl's risk signals and care management workflows preferentially generate referrals to network partners. Practices commit to network-first referrals as a default, with clinically appropriate exceptions.
In return, network partners typically agree to negotiated fee discounts (5 percent off the Aggregate Payment Cap is a standard ask) and to participate in shared performance reporting on the quality metrics. The volume-for-quality trade is straightforward: high-quality hospices want more appropriate referrals, and they'll trade modest fee concessions and reporting transparency to get them.
The discount itself is not the point. The patient flow is. A hospice network that captures the majority of an ACO's hospice referrals removes the abuse cycle from a meaningful percentage of cases. That avoided-cost figure is what justifies the operational work of building the network; and the patient-experience improvement is what justifies the moral weight of doing it well.
The integration with the rest of the EOL program
This piece doesn't work as a standalone. A curated hospice network without a palliative care wraparound and without POLST infrastructure earlier in the patient journey is a tail-end intervention that catches a smaller subset of cases. The earlier interventions (POLST completions, palliative care engagement on identified seriously ill patients) are what generate the population of patients who arrive at the hospice decision in the first place, having had the conversations that let them choose what they want.
What the hospice network does is make sure that when those patients do arrive, they don't get caught in a benefit abuse cycle that erases the savings produced by everything upstream, and, more importantly, that doesn't put them through experiences they never agreed to.
This is one of three posts on Pearl Health's end-of-life program. The companion pieces examine why POLST is the right wedge for primary care, and why palliative care is not a single strategy but a multifactorial design decision that varies by practice. Read together, they describe a three-layer end-of-life strategy: POLST as the entry point, properly matched palliative capacity as the wraparound, and a curated hospice network as the back-end protection against the largest avoided-cost lever in the entire system.
Why almost nobody does this
The honest answer is that it's hard. Building a curated hospice network requires going down to the level of individual hospice operator data, evaluating it against criteria most ACO leadership teams haven't internalized, negotiating with network partners, communicating the rationale to participating practices, and changing the default referral patterns of care managers and discharge planners.
It's also the part of the strategy that's hardest to copy from the outside. The work is in the analytical detail and the partner negotiations, not in a slide that says "curate a hospice network." That's why ACOs that invest the operational effort capture savings their peers don't, and produce outcomes for patients that look meaningfully different from the rest of the field.