The numbers have an agenda: Why physicians must question ACO shared-savings calculations

Shared-savings statements arrive looking authoritative: clean tables, precise dollar figures and risk scores carried to three decimal places. But precision is not accuracy. Every shared-savings number is the product of a chain of assumptions about attribution, risk, benchmark construction, trend, quality and distribution. Each step can materially change who appears to have created savings and who gets paid.

Physicians should not assume that an accountable care organization’s (ACO) distribution is correct simply because it comes from an actuary, a finance committee or a payer report. Over the past five years, I have identified material distribution errors that redirected hundreds of thousands of dollars to clinicians who had actually generated the savings.

The central question is not, “Did the ACO report savings?”

It is: “Can the ACO show how those savings were calculated, validate the result and explain why your practice was paid what it was paid?”

Why should physicians be skeptical of an ACO’s shared-savings report?

Because it is not a fact; it is an estimate produced inside a contractual and actuarial framework.

A shared-savings calculation requires judgments about:

• Which patients were attributed to which practices
• How long those patients were actually attributed
• How illness burden and risk were measured
• What baseline spending was used
• How spending was expected to trend
• Which claims and services were included or excluded
• Whether quality thresholds changed the payment
• How the ACO decided to distribute earned dollars among practices

A technically correct formula can still produce a wrong result if it uses the wrong population, denominator, baseline or quality definition. The spreadsheet may be mathematically elegant while the conclusion is economically false.

Are these errors common enough to matter?

Yes. Assume there may be errors until the ACO can demonstrate otherwise.

Most problems are not fraud or dramatic misconduct. They are quieter: a static member count substituted for member-months, an outdated quality threshold, a benchmark built from a population that no longer resembles the current one or a distribution formula that has never been revisited.

The immediate consequence is financial. A practice may be underpaid despite reducing avoidable utilization, improving access, managing chronic disease and carrying a more complex patient panel. The longer-term consequence is worse: Physicians lose faith in the contract and stop investing in the care-management work that the ACO claims to reward.

Why should I care about member-months?

Because a member count is a snapshot; member-months measure actual exposure.

Suppose a practice starts the year with 1,000 attributed patients. In July, an employer group representing 300 patients leaves the plan. The practice ends the year with 700 patients.

Actual exposure is:

If allowed cost is $5.1 million, the correct per member per month (PMPM) cost is:

If the ACO instead uses the opening population of 1,000 members for a full year, it creates 12,000 member-months and reports a PMPM cost of $425. If it uses the year-end count of 700, it produces an implied PMPM cost of $607.

All three numbers may look plausible. Only one reflects the period during which the practice was actually accountable.

At a $520 PMPM benchmark:

• Correct calculation: $520 × 10,200 member-months = $5.304 million expected cost, compared with $5.1 million actual cost: $204,000 in savings

If the ACO instead uses the year-end census of 700 members as though those were the only members attributed for the entire year, it counts only 8,400 member-months. Expected spending falls to $4.368 million, making the same $5.1 million in actual spending appear to be a $732,000 loss. The practice’s care did not change. A static year-end count turns $204,000 in genuine savings into a $732,000 reported loss — a $936,000 swing created entirely by the denominator.

The reverse happens when membership grows during the year. A practice can improve care and lower avoidable spending, yet appear to have performed poorly because the denominator understated its actual exposure.

The physician question is simple: “Did the ACO calculate PMPM using actual member-month eligibility, and does the denominator match the same period reflected in the cost numerator?”

How can a benchmark make good performance disappear?

A benchmark represents what spending supposedly would have been without the practice’s intervention. That is not an observable fact. It is a modeled counterfactual.

Benchmarks can be distorted by:

• Incorrect or inconsistent baseline years
• Changes in attributed membership
• Employer-group turnover
• Benefit-design changes
• Network changes or leakage patterns
• Regional or payer-wide trend factors that do not reflect the local market
• Comparison populations with different geography, demographics or illness burden
• Inclusion or exclusion of high-cost categories without a clear contractual basis

A flawed benchmark can create savings that never occurred. It can also erase genuine savings.

Physicians should ask whether the benchmark reflects the practice’s actual population and market — not merely a convenient actuarial average. If the ACO cannot explain how the benchmark was built, it cannot credibly explain why a practice did or did not earn savings.

How can risk adjustment affect my payment?

Risk adjustment sets expected cost. Expected cost determines whether actual spending looks favorable. That determination influences whether shared savings exist at all.

Commercial payers increasingly use proprietary risk models that may differ materially from CMS-HCC or HHS-HCC methods. The same patient can receive different risk scores across different contracts because models may vary in:

• Diagnosis hierarchies
• Chronic-condition weights
• Diagnosis-frequency requirements
• Treatment of behavioral health conditions
• Treatment of social risk
• Coding-intensity adjustments
• Recalibration methods

If the model understates the burden of a practice’s patient population, the benchmark is too low, and clinically appropriate spending can be labeled poor performance. If the model overstates risk, it can manufacture savings that reflect model design rather than better care.

A physician should be skeptical when panel complexity clearly rises while the ACO reports flat or declining risk, or when the same practice appears successful with one payer and unsuccessful with another despite similar clinical performance.

Can quality reporting determine whether I am paid?

Absolutely. In many contracts, quality performance is a payment gate, a savings adjustment or a withhold calculation. A modest error can turn a payment into no payment.

Common failures include:

• Applying prior-year thresholds
• Using an outdated measure specification
• Miscalculating the eligible denominator
• Mishandling exclusions
• Treating a payment gate as merely informational
• Failing to account for changes in patient attribution or data capture

A physician should not accept a quality result merely because it appears on an ACO dashboard. If the result is inconsistent with practice records, historical performance or known operational conditions, it deserves reconciliation.

The question is not whether the ACO’s quality report is necessarily wrong. The question is whether anyone independently verified it before using it to reduce physician compensation.

Why should physicians care about the ACO’s distribution formula?

Because the ACO may receive a valid payer payment and still distribute it unfairly.

Often, an ACO does not have sufficiently timely or reliable clinic-level data to calculate each practice’s precise contribution to shared savings. Payers may withhold attribution detail, claims detail, benchmark logic or proprietary risk-adjustment methodology. Faced with incomplete information, the ACO may use a fixed share, pay only a subset of practices or rely on internal estimates of who “likely” generated savings.

That can be reasonable only if the limitations are disclosed and the methodology is transparent, consistent and tied as closely as possible to measurable performance.

An arbitrary distribution formula based on a percentage of clinics paid can overpay a practice that did not generate savings and underpay a practice that did. “The data were incomplete” is an explanation for uncertainty; it is not a justification for secretive or permanent inequity.

Where exact clinic-level savings cannot be calculated reliably, a tiered distribution model is often more defensible. Practices can be grouped into several performance tiers using the best available cost and quality measures, with stronger combined performance receiving a greater share and lower-performing practices receiving less rather than nothing.

That approach does not create false precision. It acknowledges uncertainty without pretending that every practice contributed equally — or that only a favored few did.

What should an ACO be required to show physicians?

At minimum, the ACO should provide a written shared-savings methodology that identifies:

• Data sources and data limitations
• Attribution rules and attributed membership
• Member-month eligibility data
• Risk-adjustment method and material assumptions
• Benchmark baseline periods and construction method
• Trend assumptions and source population
• Included and excluded spending categories
• Quality measures, thresholds, denominators, exclusions and payment effects
• Shared-savings distribution formula
• Appeals, discrepancy and correction process

The ACO does not need to disclose a payer’s protected intellectual property. But it must provide enough information for a physician, analyst or independent actuary to determine whether the payment is financially credible.

A physician should insist on reproducible work papers. Another qualified analyst should be able to follow the logic, rerun the calculations using the available data and understand why the result differs from the payer’s report, if it does.

What is a shadow calculation?

A shadow calculation is an independent internal estimate of contract performance using available practice, ACO, claims, enrollment, attribution, clinical and financial data.

It will not perfectly match the payer’s result. Nor should anyone expect it to. The payer may use data feeds, exclusions, completion lags and proprietary models unavailable to the practice.

Its purpose is to identify unexplained variance and test whether the reported result is plausible. A useful shadow calculation asks:

• Does the attributed population match what the practice sees?
• Are the member-months plausible?
• Are allowed-cost totals consistent with available claims data?
• Does the risk trend make clinical sense?
• Does the benchmark reflect the agreed baseline?
• Are high-cost services handled consistently?
• Is the quality result reproducible?
• Does the distribution methodology correspond to actual measurable contribution?

The goal is not to rebuild the payer’s black box. The goal is to determine whether the ACO has enough evidence to distribute physician money fairly.

What should physicians ask their ACO leadership?

Physicians should ask direct questions before accepting a distribution as final:

• How many attributed member-months were assigned to my practice?
• What claims and costs were included or excluded?
• What baseline years created the benchmark?
• What trend factor was used, and why?
• What risk model was applied?
• How did my practice’s risk score change, and is that clinically plausible?
• Which quality measures affected payment?
• What were the eligible denominators, exclusions, thresholds and final results?
• Can I review the distribution methodology before payments are made?
• What data support the conclusion that my practice did — or did not — contribute to savings?
• Was the methodology reviewed independently?
• What is the formal process for disputing an error?

An ACO that cannot answer these questions is asking physicians to trust a result it cannot defend.

What governance should physicians demand?

Shared savings should be treated as a governed financial process, not an actuarial deliverable handed down at year-end.

Physicians should expect:

• A written, board-approved methodology
• Advance disclosure of distribution rules
• Formal reconciliation between payer results and internal calculations
• Sensitivity analysis for high-impact assumptions
• Contract-specific validation of quality measures
• Independent review when distributions are material
• A documented appeals and correction process
• Finance committee or board oversight of methodology changes
• Actuarial documentation of assumptions and data limitations
• Clear identification of assumptions that materially affect payments

At minimum, the ACO’s actuaries should attest that the calculation followed the contract, used the available data appropriately, disclosed material limitations and applied the documented methodology consistently.

What should sensitivity testing reveal?

A valid methodology should not collapse when one reasonable assumption changes.

The ACO should test:

• Member-months versus static member counts
• Alternative valid baseline periods
• Membership churn
• Plausible ranges of risk-score trend
• Different attribution cutoffs
• Inclusion or exclusion of unclear high-cost claim categories
• Alternative interpretations of quality denominators
• Alternative distribution formulas

If a modest change in one reasonable assumption materially changes physician payment, the result is uncertain. It should not be used casually to determine compensation, recruit or terminate practices, expand a program or conclude that care management has failed.

What is at stake?

If the calculations are wrong:

• Practices that genuinely reduce cost and improve care may be underpaid
• Practices whose results are methodological artifacts may be overpaid
• Physicians lose confidence in value-based contracts
• Care-management investment declines
• Boards make strategic decisions using distorted data
• The ACO reinforces the wrong clinical and operational behavior
• A false story takes hold about who is creating value

A spreadsheet error can be corrected. Physician disengagement is much harder to reverse.

The bottom line

Physicians should not ask only whether the ACO reported shared savings. They should ask: “Can the ACO trace, reproduce, validate and credibly explain why those savings occurred and why my practice received the payment it did?”

When payers use proprietary models and limit transparency, every shared-savings result contains judgment. That does not make value-based contracting illegitimate. It makes governance essential.

An ACO that expects physicians to change referral patterns, invest in access, improve coding, hire care managers, manage high-risk patients and accept downside exposure has an obligation to show its work. If it cannot explain the calculation, test the assumptions, reconcile discrepancies and distribute savings transparently, physicians have every reason to be skeptical.

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.