Value-based care in the United States: attribution, total cost of care and the quality gate

How commvita runs the three numbers an accountable care organisation is judged on — who is attributed, what they cost against the CMS benchmark, and whether quality clears the gate — plus the billable care-management programmes that fund the work.

1 The numbers an ACO lives on

An accountable care organisation is judged on three things. How many lives are attributed to it. What those lives cost against the benchmark. And whether quality is good enough to unlock a share of anything saved. Get the third wrong and the first two stop mattering, because a shared-savings pool you don’t qualify for pays nothing. commvita puts all three on their own screens, computed from the same member record, and this document walks each one.

This is the first of three documents about the United States build. It assumes you already know what the platform is; if you don’t, the US health system overview covers the whole module pack, the jurisdiction setup and the payer and compliance rails, and this document doesn’t repeat it. Risk adjustment gets its own document next. Everything below is either Live, meaning an endpoint in the platform serves it, or Demonstrated, meaning the surface reads seeded content and says so.

2 Who counts as yours

Attribution is calculated quarterly on CMS methodology under 42 CFR 425. Members are assigned on the plurality of their primary-care visits. Prospective attribution runs off prior-year claims and tells you at the start of the year who you’re probably responsible for. Retrospective attribution is finalised when the performance year closes and tells you who you were responsible for in the end. Most organisations see one of those. commvita shows both on the same list.

Read the visit column. It’s three numbers per member, and it’s the column that decides attribution. A member with a thin visit history against your primary-care physicians is a member you’re about to lose, and losing them late in the year means you carried their cost without carrying them into the calculation. Filtering the list by attribution basis and by risk tier turns it into an outreach list. The Exclude action removes a member from the attributed cohort where the organisation has a reason to, and it’s restricted to an administrator.

Every figure on that screen is derived. The tiles count and average the member rows underneath them, so the total, the split between prospective and retrospective, the average risk-adjustment factor and the average year-to-date cost all reconcile with the list. The cohort itself is seeded demonstration data. It’s fifteen invented people, and it’s there to show the shape of the surface.

3 Total cost of care against the benchmark

Total cost of care is the number the whole model turns on. commvita holds actual spend per member per month against the risk-adjusted CMS benchmark for the performance year, and reports the difference as gross savings before any sharing rate is applied.

The methodology sits on the screen rather than in an appendix, which matters when a finance director asks where a number came from. Gross savings is the benchmark minus actual, multiplied by attributed members, multiplied by twelve. The benchmark is risk-adjusted, and the risk-adjustment factor used is shown on the same row as the money so the two are never separated. An executive can see immediately whether a favourable variance is real performance or a benchmark that moved.

The condition drivers underneath rank where the spend concentrates. Five conditions, ranked by share, and each one maps to a care-management programme further down this page. That’s the useful join: the same five conditions that drive cost are the ones the chronic care management registry is enrolling against.

4 The gate, and why it comes first

Shared savings are conditional. An ACO has to clear a quality threshold before it sees a dollar of what it saved, and the composite is built from measures drawn from three different sets. commvita tracks all of them on one board with the benchmark beside each rate and a gap count against each measure.

The gap count is the operational column. A measure that sits two points under benchmark with a hundred and twenty open gaps is a fixable problem with a name and a list behind it, and the platform serves that list: each measure drills to the patients who are missing the numerator event, and closing a gap writes back against the measure. The composite score at the top is the number that decides whether the organisation qualifies at all, and putting the qualifying threshold on the same banner stops that being something people rediscover in November.

Two measures in the demonstration set are flagged critical. That’s the point of the surface. A quality board where everything is green tells an executive nothing.

5 Getting the measures to CMS

Tracking a measure and reporting it are separate problems. Electronic clinical quality measures go to CMS as QRDA category I and category III documents, and a rejected submission close to a deadline is an expensive week. commvita holds each measure as a submission object with its numerator, denominator, reporting period and state.

Five states: draft, validated, submitted, accepted, rejected. Only a validated measure offers a Submit action, so a half-built measure can’t be pushed at a deadline. Rejections stay on the list with their rate and their period instead of disappearing, and a submission log records what went when. The counts across the top reconcile with the rows, which sounds obvious and is the thing that most often isn’t true of a reporting screen.

6 The programme that pays for the care manager

Chronic care management is how an organisation funds the people who do the population work. A patient with two or more chronic conditions consents to a care plan, a care manager spends time on them across a month, and once that time crosses twenty minutes the month becomes a billable claim. The whole thing turns on capturing minutes at the point they’re spent.

The three-state key is the design decision worth noticing. A session under the threshold is shown as not yet billable not hidden or rounded up, so a care manager can see that a patient is three minutes short and finish the work, not lose the month. Sessions carry the activity that generated them, which is what an audit asks for. The registry tab holds the enrolments with their conditions, risk tier, named care manager and last review date, and it records a patient who declined as declined instead of dropping them. The billing tab turns the qualifying months into claim records with the code, the minutes behind them and their state from ready through submitted to paid. Enrolments, sessions and billing are all served by the platform’s own endpoints.

7 The annual wellness visit

The annual wellness visit is the other billable anchor, and it does double duty. It’s a claim in its own right, and it’s the appointment where the health risk assessment gets done and the screening gaps that the quality board is chasing get closed.

The action button changes with the state, so the screen is a worklist instead of a report. Scheduled offers Start Visit. In progress offers Complete. Completed offers Bill. Billed offers nothing but the record. That sequence is enforced by the appointment record, so a visit can’t be billed before it happened. The dollar values shown are the CMS rates for the two codes and they’re illustrative figures on seeded appointments.

8 The thing this document isn’t about

You’ll have noticed a risk-adjustment factor on two of these screens. It sets the benchmark, so it sets the number every saving is measured against, and it deserves a document of its own not three lines at the end of this one.

The position commvita takes is worth stating plainly here anyway. Risk adjustment is about documenting conditions a patient already has. A person with stage three chronic kidney disease has it whether or not the coder found it, and if the record doesn’t carry it then the benchmark understates what caring for that person costs. The work is finding conditions that are present and evidenced in the record and are missing from the claim. It’s a completeness problem. The next document in this series covers how commvita surfaces those, what evidence it requires before it suggests one, and where a clinician has to decide.

Where it lives in commvita