Medicare is still swiping left on your patients

Between one quarter and the next, my Medicare Shared Savings Program (MSSP) roster lost a few dozen patients and gained about the same number. On paper, the net change was close to zero. In practice, more than 50 patients changed hands in a single quarter. Many who left had been with me for more than a decade. Many of the “new” ones had been with me nearly as long. It’s hard to call that accountable care.

The Centers for Medicare & Medicaid Services (CMS) seems to agree that part of the problem is real. Beginning in 2027, the proposed Physician Fee Schedule would fix quality reporting by limiting it to patients actually assigned to the accountable care organization (ACO). Beginning in 2028, it would change who gets assigned. Those are real steps. But none of them touches the plurality-by-dollars rule that causes most of the churn.

CMS sees the light

Right now, Medicare uses two rulebooks to decide who a patient’s doctor is: one for quality and one for attribution. One visit can make a physician responsible for a patient’s quality scores. Decades of care can still leave that physician unattributed for savings.

Starting in 2027, the proposal would count only assigned patients as eligible for Medicare clinical quality measures (CQMs). The quality population would finally match the attributed population. One visit would no longer make you responsible for a stranger’s diabetes, and a patient who isn’t yours would no longer become yours when something is overdue.

Starting in 2028, CMS proposes two changes to assignment itself, with the stated goal of expanding the number of beneficiaries in accountable care relationships:

• Closing a side door. Some ACO clinicians bill for the same patient both inside the ACO and through a practice outside it. When the outside practice wins the plurality, the patient drops out of assignment. CMS would stop counting those outside charges, citing concern that the current rule allows ACOs to avoid accountability for high-cost, at-risk patients. It estimates almost 98,000 added assigned person-years, or about 1%.
• Opening the front door wider. Under the current rule, the patient also needs a clean record for the whole assignment window: no months of Part A-only, Part B-only or Medicare Advantage enrollment. Under the proposal, one month of Part A and Part B without Medicare Advantage would be enough, no matter what coverage the patient had in the other months. CMS estimates about 248,000 more assigned person-years, roughly 2.5%.

Measuring physicians on the patients actually assigned to them is a simple idea that took more than one full performance year to rectify. If CMS finalizes it, call it progress. Everything depends on assignment, and the 2028 changes make the list bigger, not steadier.

Why attribution is the whole game

In an ACO, attribution isn’t paperwork. The ACO’s historical benchmark comes from the spending of patients who would have been assigned to it during the benchmark years. Performance-year results are measured based on the patients assigned now. Savings, losses, risk scores, per-capita costs and, under the proposal, quality denominators all depend on who is on the list.

When the list moves, the math moves with it. If a well-managed patient with years of controlled chronic disease drops off, the ACO loses the patient whose care shows what it can do. If a high-cost stranger joins, the ACO takes on spending it never had a chance to manage. An ACO can't bend a cost curve for patients it can’t identify.

CMS wants every person in traditional Medicare to be in an accountable care relationship by 2030, with ACOs as the main vehicle. Yet the assignment method treats a decades-long primary care relationship as one data point in a 12-month billing comparison. CMS is trying to fill the house while the front door keeps swinging. In 2028, the door gets wider.

A typical quarter

The patients who leave aren’t drive-bys. Some have been with the same physician for 10 to 20 years or more. Apparently, Medicare decided their physician had cared for them long enough — “Dear Doctor: Thank you for your service.” The “new” patients get the opposite treatment — “Welcome to the ACO.” Never mind that you have been their physician since the Clinton administration.

Currently, the quality side is more creative still. A physician sees a new patient with diabetes once, late in the quarter of a performance year. Months later, that patient’s care gaps belong to the physician: hemoglobin A1c control, eye exam, kidney health evaluation. Nothing builds a therapeutic relationship like owning a year of diabetes control for someone you met once.

Then there are long-standing patients missing from the MSSP roster but present on the quality gap list. They aren’t the physician’s patients when it comes to savings. They are when something is overdue.

None of this is a glitch. It is the design working exactly as written.

Rulebook one: Instant commitment (proposed for retirement in 2027)

Since performance year 2025, ACOs have reported Medicare CQMs. A patient becomes eligible after a single qualifying claim during the measurement period from an ACO primary care physician or other ACO provider. No plurality test, no tiebreaker. One visit puts the patient in the denominator.

CMS sends the ACO a quarterly list of these patients, built from all Medicare fee-for-service claims. But it’s only a list of names. The ACO will have no information on whether their screenings, tests or other quality-measure services were already done elsewhere. Medicare hands over the patient and the homework. You were her doctor for 15 minutes, after all.

CMS has acknowledged that this list caused confusion. The 2027 proposal would retire it.

Rulebook two: Prove it, in dollars (still standing)

Assignment for shared savings works on a different principle, and under the proposal it would drive quality too. CMS checks whether the ACO’s clinicians billed more primary care allowed charges than any other ACO or outside practice. It’s a plurality, measured in dollars. The number of visits don’t count. Years don’t count. Nor do they capture the times you talked a patient out of an MRI they didn't need.

What counts is the specialty code on the claim, not on who actually provides primary care. MedPAC, the Medicare Payment Advisory Commission, states that nurse practitioners (NPs), physician assistants (PAs) and clinical nurse specialists “are designated as primary care clinicians regardless of their true practicing specialty.” Plenty of outside clinicians bill the same primary care codes as the internist –– a Florida physician the patient sees every winter, an urgent care center, a weight-loss or men’s health clinic that bills Medicare for office visits. So does an NP or PA in a specialty practice, such as a cardiology family NP who sees the patient every three months. If any one of them bills more primary care dollars than the internist’s ACO, the patient goes with them because CMS counts charges, not relationships.

The most common setup for losing a patient is not a big competitor but a thin relationship on paper. When the internist bills only a visit or two a year, a few urgent care visits, a specialty-practice NP's follow-ups, or a wellness clinic’s visits can out-bill the physician managing the patient’s diabetes, blood pressure, medications and everything in between. If there is a tie, CMS goes with the most recent visit. After more than 10 years, the relationship can come down to a coin toss.

Then there’s timing. Prospective assignment uses a 12-month window that closed before the performance year began. Preliminary prospective assignment with retrospective reconciliation reruns quarterly lists on a rolling window, with the final answer coming months after the year ends. A practice can care for a patient all year and learn the next year that the patient wasn’t theirs, or be graded on visits from a year and a half ago. Either way, the ACO manages a population it can only see in the rearview mirror.

Here is what the proposal does to the attribution rulebook:

• 2027: A few more services count as primary care for assignment, including advance care planning.
• 2028: The outside-practice charges stop counting, and patients with partial-year enrollment become eligible.
• No change: The plurality-by-dollars test, the tiebreaker, the specialty-code definition of a primary care physician and the 12-month window all stay. The side door gets closed. The coin toss stays.

The patient’s opinion counts, technically

A patient can formally name their own clinician, and that choice counts. It is the one place where the patient’s voice beats the claims file, yet almost nobody knows it exists. The proposal doesn’t change it. If CMS wants assignment stable enough to carry both quality and savings, letting patients say who their doctor is would be a good place to start.

The patient may not know they are being counted, either. Today, an ACO must provide a standardized written notice at or before a patient’s first primary care visit and then follow up within 180 days. Starting in 2027, CMS proposes issuing one notice per agreement period, generally due by May 30, with no follow-up, saying that follow-up didn’t improve beneficiaries’ understanding. Fair enough, but one letter every few years is a casual courtship. Meanwhile, the backup plan is still the sign in the waiting room, somewhere between the flu shot poster and the hand sanitizer.

The translation layer

By the time CMS files reach a practice, physicians can’t see the plurality calculation, but they can see the care gap list. When the method is this opaque, even good-faith intermediaries struggle to explain why a 10-year patient disappeared. The physician is left to guess whether the cause was an urgent care visit, a specialty-practice NP or a billing error that no one will ever confirm, and the patient has usually passed through the ACO and its analytics vendor. Physicians can’t see which practice won the patient.

Attribution disorder, unspecified

If the proposal is finalized, physicians will no longer own quality scores for patients they met once or care gaps for patients Medicare won’t assign to them. By 2028, an ACO clinician’s outside billing would no longer pull patients out of the ACO. That’s real relief.

What stays is the larger problem. A patient you have cared for for over 10 years can still become a stranger because someone else billed more this year. More part-year patients join the roster. The physician’s notice is still a list that shows up months after the visit.

ACOs can’t be accountable for patients the program can’t consistently assign, and benchmarks can’t be fair if the population under them keeps changing. If a patient couldn’t keep their doctor straight, we would question their reliability. When Medicare can’t, the physician pays for it.

Physicians are graded on precision. Attribution is graded on a curve. The proposed fee schedule may cure one symptom and treat another, but the underlying condition remains: a program that demands accountability and can’t keep track of its own patients. In the exam room, that’s a care gap. At CMS, it’s attribution.

Robert Resnik, M.D., MBA, is a board-certified internal medicine physician practicing in Cary, North Carolina. He earned his medical degree from Eastern Virginia Medical School and completed his residency at East Carolina University. He also holds an MBA from Duke University.