As summarized in our previous advisories, the Patient Protection and Affordable Care Act anticipates the creation of accountable care organizations comprised of physicians, hospitals, and other health care suppliers.

ACOs must be willing to enter into a three-year Shared Savings Program agreement with the Centers for Medicare and Medicaid Services and be accountable for the care of at least 5,000 Medicare beneficiaries. If quality performance standards are met, the ACO is eligible to receive shared savings bonus payments in addition to fee-for-service payments. The details are contained in the proposed ACO rule published by CMS on April 7, 2011.

This eighth advisory in our series on ACOs focuses on how the shared savings will be calculated. This article does not address shared risks, which will be examined in a future advisory. CMS refers to the shared savings methodology as the “one-sided model,” and the combination of shared savings and shared risks as the “two-sided model.”

CMS’ proposed methodology for calculating shared savings is neither simple nor particularly precise, making it challenging for providers to estimate their potential bonus payments before deciding whether to form an ACO.

Essentially, CMS proposes to share with the ACO a portion of the savings that are calculated by comparing the ACO’s future Medicare payments against a “benchmark” of what CMS would have paid for services to the same patient population absent the ACO’s efforts to achieve cost savings. To an unusual degree, CMS discusses the various options it considered and adopted or rejected, and invites public comment on all aspects of its methodology.


As a preliminary matter, no ACO will be eligible for any shared savings payments if it fails to meet CMS’ minimum quality standards. Please refer to our April 18, 2011 advisory, “Proposed Quality Measures for ACOs,” for an explanation of the standards.

Establishing an expenditure benchmark

The Medicare expenditure benchmark is designed to estimate the amounts CMS would otherwise have spent for Medicare Parts A and B for the population served by the ACO. If successful, the cost savings estimates that are based on this benchmark will reward an ACO that successfully coordinates patient care and achieves real savings.

The option that CMS selected to measure the initial benchmark analyzes Parts A and B fee-for-service payments on behalf of patients that would have been assigned to the ACO in the three years prior to the contract period (had ACOs existed).

Starting with the tax identification numbers of ACO provider participants, CMS will use their claim records to identify patients who received a plurality of their primary care services from physicians in the ACO in each of the three prior years. The per capita expenditures for the group of assumed ACO members will then be adjusted for demographic factors and health status, as well as inflation.

The option CMS considered but rejected would have based the benchmark on patients who were actually assigned (as opposed to “would have been assigned”) to the ACO, and analyzing the Parts A and B expenditures for those patients during the prior three years.

Adjusting the benchmark for the beneficiary characteristics

CMS also proposes additional adjustments intended to provide greater certainty that shared savings are actually the result of greater quality and efficiency in delivering health services. Specifically, CMS wishes to minimize the possibility that expenditures could go down based on random factors and to anticipate that ACO providers will provide diagnosis codes with more detail in the ACO context. In other words, the coding for the benchmark population might be apples, and the coding for the “real” ACO participants might be oranges.

To guard against these risks, CMS proposes to import a prospective risk adjustment model that has already been implemented in the Medicare Advantage program. The model is called the CMS Hierarchical Condition Category (CMS-HCC). The CMS-HCC is designed to use historical diagnoses to develop risk scores, which are then applied to their current year expenditures. CMS asserts that the model has been accepted by the Medicare Advantage HMOs.

CMS will calculate a single benchmark risk score for each ACO, based on the assigned ACO beneficiaries. That risk score will apply throughout the three-year contracting period. The benchmark risk scores for each of the three years prior to the contracting period will be calculated by applying the CMS-HCC model to the assigned beneficiaries, but changes in the risk scores between the benchmark period and the contract period will be ignored. According to CMS, this methodology will minimize the effects of “greater diagnosis coding intensity” that may occur going forward.

Trending the benchmark

CMS also had to decide how to account for inflation while calculating the expenditure trend in the benchmark period in order to project it into the future and compare it to the performance period. CMS proposes to use actual beneficiary expenditures (as opposed to national per capita expenditures), but to adjust them for inflation using growth factors based on national expenditures.

CMS reasons that the former will be the most neutral and comparable method across different geographic regions, while the latter will assist ACOs in both high- and low-spending markets and growth markets while moving the industry toward the establishment of national standards.

Minimum savings and sharing rates

CMS proposes to share savings with ACOs only when they achieve a minimum savings rate (MSR). To establish MSRs for different sized ACOs, CMS purports to rely on standard inferential statistical analysis to assign a sliding scale of statistical confidence intervals (CIs) which will be individually tailored to each ACO depending on its number of beneficiaries.

Because statistical confidence is lower with smaller numbers of beneficiaries, the smaller ACOs will have higher MSRs. For example, an ACO with 5,000 beneficiaries will have an MSR of 3.9 percent, while an ACO with 10,000 will have an MSR of 3 percent and an ACO with 50,000 will have an MSR of 2.2 percent.

With some exceptions, once an ACO has achieved its MSR, it will be entitled to receive 50 percent of the total savings in Medicare expenditures beyond the first 2 percent of savings, up to a limit of 7.5 percent of the benchmark. Exceptions to this 2 percent threshold are made for certain ACOs in smaller rural areas such as those served by critical access hospitals and federally qualified health centers (FQHCs).

Special bonuses for ACOs that Include FQHCs and/or RHCs

ACOs that incorporate the services of FQHCs and rural health centers (RHCs) will be given more generous shared savings, at least in the one-sided model. Once again, CMS proposes to pay such bonuses on a sliding scale that measures the numbers of ACO beneficiaries that use FQHC or RHC services.

Such ACOs can receive an increase in shared savings rate of up to 2.5 percent during the first two years of the performance period. The higher the numbers of such beneficiaries, the higher will be the potential additional bonus. To receive the maximum increase, an ACO would need at least 41 percent of its beneficiaries to have at least one encounter with an FQHC or RHC.

Shared savings, or shared risk?

As indicated above, CMS plans a quick transition of ACOs from sharing savings only to sharing both savings and risks. Shared risks mean that, if the ACO does save the program money, then the ACO will have to refund to CMS a portion of the payments it received.

The one-sided model is only available in the first two years of an ACO’s participation, and ACOs can elect the two-sided model from the beginning. After the initial two-year period, all ACOs will be subject to the two-sided model. Because every ACO is being asked to commit to three years of participation, that means every ACO must carefully consider the possibility that it may wind up paying CMS at the end of the third year, as opposed to receiving a bonus.

As noted above, the two-sided model will be the subject of another advisory in this series about the ACO proposed rule.


Under the right circumstances, an ACO could receive significant bonus payments in addition to Medicare’s fee-for-service payments. ACOs that elect to participate in this initiative under the rules CMS proposes, however, will initially have limited clinical and financial data. How well an ACO gathers, analyzes, and reacts to the data will determine how soon the goals of achieving greater efficiencies and coordination of care are realized.

Only time (and perhaps the comments to the proposed rule) will tell whether CMS has chosen the right path.