Medicare Advantage margins are shrinking. Advance care planning could be the fix you haven’t tried
November 3, 2026
Medicare Advantage margins are shrinking. Advance care planning could be the fix you haven’t tried
The easy levers have already been pulled. The one left is care nobody chose.
Most Medicare Advantage plans file advance care planning under quality or member experience, somewhere next to satisfaction surveys and annual wellness visit outreach. It almost never shows up in a conversation about the budget. That is a missed connection, and it is an expensive one.
Here is the math behind why it should not be missed. MA plans’ aggregate medical loss ratio hit 90 percent in 2024, tighter than every line of business except Medicaid managed care, according to KFF data reported by Becker’s Payer Issues. Then in 2025, health benefit costs grew another 9.6 percent, while per-member administrative costs barely moved because plans had already cut staff and overhead about as far as they could go, per the Sherlock Company’s 2026 benchmark. Put plainly: margin is compressed, and most of the easy levers have already been pulled.
That leaves one place left to work, and it is a place most plans have not touched yet: unwanted, low-value utilization in the single most expensive phase of care: the last year of life. That phase accounts for roughly a quarter of all Medicare spending, according to KFF. Much of this care is care that is unnecessary or unwanted by patients.
Advance care planning, done as ongoing infrastructure rather than a one-time conversation or a signed form, is one of the few levers that reduces that kind of utilization by giving patients the care they asked for, and educating them on their options so that care aligns with their goals and values. Lower cost and higher member satisfaction rarely move in the same direction. This is one of the rare cases where they do.
Why the connection gets missed
The reason this does not show up in most MLR conversations is simple: a meaningful share of the claims running through that 90 percent ratio is care patients did not choose, and none of it shows up as waste on a claims report.
An unwanted intubation. A fourth hospitalization in six months for a condition the patient had already decided to manage differently. An ICU stay that happens because nobody in the building knew a POLST existed, not because anyone judged it to be the right care. All of that looks identical, in the data, to care a patient actively wanted. There is no line item for utilization nobody asked for.
There is no line item for utilization nobody asked for.
The only thing that can separate goal-concordant care from care nobody asked for is a mechanism that knows what the patient wanted in the first place, and most ACP programs today are not built to do that. They are built to document a single conversation, not to catch the moment a patient’s situation changes.
The data that moves the needle
Not all ACP shows savings, and that distinction matters more than any topline number.
A 2019 JAMA Network Open study of more than 18,000 seriously ill Medicare Advantage beneficiaries found that patients with a billed ACP encounter actually had $1,635 higher adjusted monthly costs and more hospitalizations than those without one. The likely explanation is confounding: sicker patients are more likely to have an ACP conversation billed in the first place. A single conversation, counted as a line item, does not isolate the effect, and on its own, it can look like ACP makes things worse.
Comprehensive, ongoing ACP programs tell a different story. An ACO-wide program studied by AJMC and Iris-Aledade found $292 lower cost per patient per month over 14 months, roughly $4,000 per patient. A three-year case-control study from Iris Healthcare found $9,500 lower cost per participant, close to a two-to-one return after program cost. And in Koda Health’s own data in a IRB study of Houston Methodist Coordinated Care patients showed 19% lower total cost of care in the last year of life and 79% fewer terminal hospitalizations. At the population level, Koda deployments have shown:
NPS
Fewer terminal hospitalizations
Fewer ICU admissions
Higher hospice utilization
Lower total cost of care
The pattern holds across every program-level study available: a single documented conversation does not always move the needle. Infrastructure that keeps watching after that first conversation, catching the moment a patient’s situation or preferences change, is what reduces avoidable utilization. The complete MLR benchmarks, objection handling and implementation detail are in our guide for MA plans.
This is not a denial of care strategy
That distinction matters even more for an MA plan, because Star Ratings and member trust are also on the line, and this kind of initiative looks very different from traditional utilization management.
ACP built as infrastructure does not deny or delay care. It identifies what a patient wants, documents it somewhere a clinician can find it at 2 a.m., and makes sure someone revisits it as the patient’s situation changes. That is why the Koda results above pair an 87 NPS with the utilization reductions rather than trading one for the other. Patients and families consistently report feeling more in control and less blindsided, not less cared for.
Most cost containment levers ask a plan to trade member satisfaction for savings. This is one of the few that does not ask for that trade.
There is a revenue side too
The MLR case is not the only financial mechanism at work, and it is worth understanding the second one alongside it, because both are moving in the same direction.
CMS is shifting Star Ratings weight toward member-reported outcomes through the Health Outcomes Survey, and survey-based measures are on track to make up nearly 40 percent of total Stars weight by 2029. Total Quality Bonus Payment spending is projected to exceed $12.7 billion in 2025. Small shifts in Stars performance now translate directly into bonus payment revenue, not just a quality score.
Retention carries its own number. Annual plan switching runs around 23 percent, and nearly half of new enrollees leave within five years. On a 100,000-member book, each point of retention is worth roughly $2 to $3 million in annual revenue.
Here is what connects this back to advance care planning: the Health Outcomes Survey does not ask members to rate a program. It asks whether they feel supported, whether their health improved, whether they can do what they want to do. Members remember a small number of moments, not every touchpoint, and a well-run advance care planning conversation, someone asking what matters most, clarifying uncertainty, being there when a decision got hard, is one of the moments members actually remember.
It is not two initiatives competing for budget. It is one capability showing up on both sides of the P&L.
The same infrastructure that reduces unwanted utilization and protects MLR is also the kind of experience that moves HOS scores and retention.
Where to go from here
This post covers the headline numbers: what MLR measures, why a single ACP conversation does not move cost while ongoing infrastructure does, and why the revenue side reinforces the same case.
The full picture goes further: how MA plans’ MLR compares to every other line of business, the objections a Star Ratings committee or a CFO will raise, and exactly how Koda operationalizes population-level advance care planning for a Medicare Advantage plan. If you are building this case internally, that is the version worth bringing with you: MLR, Stars, and Retention: The Financial Case for Advance Care Planning in Medicare Advantage.
CEO and Co-founder, Koda Health
