Whose patient is this?

CHAPTER 01

The denominator problem

Meet a clinic. Two practices, Northside Clinic and Riverside Practice, share six clinicians and belong to one accountable care organization, called Harbor ACO here. Around them sit the other places their patients also go: another primary care practice, a Florida clinic, an urgent care, a cardiology group. Forty-two patients, twenty-four months of synthetic visit history, every fact fabricated.

Now ask the only question that matters before any measurement can start: which of these patients belong to the clinic? Three contracts answer three different ways. A Shared Savings contract follows claims dollars through a federal rulebook. An election-style contract, common in Medicare Advantage arrangements, counts whoever picked a network clinician as their own. A visit-plurality contract, a common commercial shape, counts whoever shows up most often. The table below is the whole point of this page: read across any row and watch the same patient get three different answers.

CHAPTER 02

The rules

Only one of the three contracts runs on a published federal methodology: the Medicare Shared Savings Program, whose assignment rules CMS publishes in a versioned specification and in regulation. That is the rule this page implements in full. The other two columns are archetypes: simplified stand-ins for contract families whose exact rules vary payer by payer and are usually not public. The page is honest about that line, and everything below the line is the real thing.

First, who is even assignable

Before any claims are counted, a patient must clear four screens: at least one month of both Part A and Part B and no months of only one of them; zero months of Medicare Advantage or other private plan enrollment, so a single MA month removes a patient entirely; no alignment to another shared savings initiative; and US residence. That second screen is the bridge to this site's other pages: the same enrollment choice that builds the MA books subtracts, member by member, from every MSSP denominator.

Then, what counts as primary care

The methodology defines primary care as a specific list of billing codes: office and home visits, wellness visits, care management, behavioral health integration, even e-visits and virtual check-ins. It also defines who: primary care physician specialties, non-physician practitioner types, and other physician specialties that count in a fallback step. Everything else a patient does, a knee replacement, an ECG, a dermatology biopsy, is invisible to attribution.

The algorithm: a pre-step, then three steps

Assignment is computed from allowed charges, not visit counts: the methodology says plainly that dollars reduce ties. A pre-step asks whether the patient saw an ACO physician at all during the year. Then:

• Step 1. The patient goes to whichever organization billed the plurality of their primary care dollars through primary care physicians, nurse practitioners, physician assistants, and clinical nurse specialists.
• Step 2. Only for patients with no primary care from any of those clinician types anywhere: the same plurality test, run over the listed specialist physicians instead. It is a narrow door, not a general fallback.
• Step 3. New for 2025: patients the pre-step missed can still be assigned through a 24-month expanded window, if a nurse practitioner, physician assistant, or clinical nurse specialist saw them at the ACO this year and an ACO physician saw them within the two years.

Exact ties have their own cascade: the organization with the most recent primary care service from a primary care clinician wins, then the most recent from a specialist, and if the tie somehow survives all of that, the methodology assigns at random. On this panel the random rung is never reached, and the page treats reaching it as an error.

Two overrides sit on top of the claims math. A patient who designates an ACO clinician as their primary clinician on Medicare.gov is assigned to that ACO regardless of where their dollars went. And the mirror image, easy to miss: a patient who designates a clinician outside the ACO is blocked from assignment even when the claims say they belong. Patient 04 and Patient 05 in the roster demonstrate the pair.

CHAPTER 03

Flip a patient

Attribution feels abstract until you watch how little it takes to move someone. Open a patient's file, read how each contract decided, then change one fact and watch the verdicts recompute. The ten patients at the top of the roster were each built to carry one lesson; the rest are ordinary.

CHAPTER 04

The roster moves on its own

Here is the part that surprises people who live downstream of these lists: under the most common MSSP election, assignment is recomputed every quarter over a rolling twelve-month window, using claims with a week or less of run-out. The final list for the year is computed later, over the calendar year, with three months of run-out. So the roster changes every quarter even when not a single patient changes doctors: the window slides, late claims arrive, and patients enter and leave the denominator on paper.

The figure below runs this clinic's MSSP assignment at each 2026 quarterly point and at the final reconciliation run. The bars move. The decomposition underneath answers the question every analyst gets asked about a quarter like that: who left, who entered, and which part of the rule did it.