The following provisions can be found on pages 230-41 of the proposed regs:  https://www.compassionatecarenc.org/non_members/legislative/ac/aco-proposed-regs-4-2011.pdf .  Installment 12b to follow.

 IIF—Shared Savings Determination:

• ACO will be given an option of two track: 
  • Track 1 (1-sided model for 1^st 2 yrs then 2 sided model in the 3^rd yr)
    • For the first 2 years of the 3-year agreement, shared savings would be reconciled annually using a one-sided shared savings approach, with ACOs not being responsible for any portion of the losses above the expenditure target.
    • For the third year of the 3-year agreement, an alternative 2-sided payment model will be established in which an ACO would be required to agree to share any losses that may be generated as well as savings.  ACOs that enter the SSP under Track 1 would be automatically transitioned to the 2-sided model in the third year of their agreement period. In that year, the ACO’s payments would be reconciled as if it was in the first year of the 2-sided model; however, quality scoring would still be based on the methods for the third year.  Thereafter, those ACOs that wish to continue participating in the SSP would only have the option of participating in Track 2, which is the 2-sided model.
  • Track 2 (2-sided model from the beginning)
    • More experienced ACOs that are ready to share in losses with greater opportunity for reward may elect to immediately enter the 2-sided model for the 3 year period. The ACO would be eligible for higher sharing rates than under the one-sided model (60% vs 50%)
  • CMS seeks comment on the 2 tracks.

• Overview of Shared Saving Determination (applies to both tracks)
  • Definitions:
    • Minimum Savings Rate (MSR) is the percentage that expenditures must be below the applicable benchmark to account for normal variation in expenditures based upon the number of Medicare FFS beneficiaries assigned to an ACO.  (Note: could the variations be greater than the benchmark and if so, how are they treated?)
    • Sharing Rate is the percent of the difference between such estimated average per capita Medicare expenditures in a year, adjusted for beneficiary characteristics, under the ACO and such benchmark for the ACO may be paid to the ACO as shared savings and the remainder of such difference shall be retained by the Medicare program.
    • Sharing Cap is the limit placed on the total amount of shared savings that may be paid to an ACO.
  • Step 1:  Methodology for establishing the benchmark (which is a surrogate measure of what the Medicare FFS Parts A and B expenditures would otherwise have been in the absence of the ACO) by determining:

1.  
   1. The patient population (based on assignment) for who the benchmark is calculated;
   2. Appropriate adjustments for beneficiary characteristics such as demographic factors and/or health status that should be taken into account in the benchmark;
   3. Whether any other adjustments to the 3-year benchmark are warranted, such as to avoid potentially disadvantaging various types of providers (e.g., hospitals that receive Medicare DSH payments or teaching hospitals that receive indirect graduate medical education (IME) payments) or ACOs located in high cost, or low cost, areas; and
   4. Appropriate methods for trending the 3-year benchmark forward to the start of the agreement period, and subsequently for updating the benchmark for each of the 3 performance years of the agreement period with the ACOs.

Under Step 1, # 1—establishing the CMS presented two different options for estimating the benchmark.  The first (option 1), which is the one being proposed, is based on beneficiaries who would have been assigned in each of the 3 years prior to the agreement period (benchmark would be determined prospectively); the second option (option 2) is based on beneficiaries who are actually assigned for each performance year (benchmark would be determined retrospectively).

• Option 1 (pp 238 -241)—ACO’s benchmark would be estimated for an agreement period starting with the TINs of ACO participants identified at the start of the agreement period.  The same rules that will be used to determine assignment of beneficiaries to ACOs during the agreement period would be applied to these data.  Claims records of these ACO participants would be used to determine a list of beneficiaries who received a plurality of their primary care services from primary care physicians participating in the ACO in each of the prior 3 most recent available years.  Using the per capita Parts A and B FFS expenditures for beneficiaries that would have been assigned to the ACO in each of these 3 prior years, a fixed benchmark would be estimated that is adjusted for overall growth and beneficiary characteristics, including health status using prospective Hierarchical Condition Category (HHC) adjustments.  This benchmark would then be updated annually during the agreement period based on the absolute amount of growth in national per capita expenditures for Parts A and B services under the original Medicare FFS.

1. Calculate annual Parts A and B FFS per capita expenditures for the beneficiaries who would have been assigned for each of the (3) benchmark years. To minimize variation from catastrophically large claims, an assigned beneficiary’s total annual Part A and B FFS per capita expenditures would be reduced to the 99^th percentile as determined for each benchmark year (e.g., approximately $100,000 in 2008).  An assigned beneficiary’s total annual Parts A and B FFS per capita expenditures also would be reduced to the 99^th percentile for each subsequent performance year.
2. Using the CMS Office of the Actuary national Medicare expenditure data for each of the years making up the benchmark, determine an appropriate growth index and trend them to benchmark year 3 (BY3) dollars (Note: trending to be discussed later).
3. Using health status measures for the beneficiary population in each of the years making up the benchmark, health status indices would be established for each year, which would then be adjusted so they are restated to reflect BY3 risk.
4. Compare a 3-year risk-and growth-trend adjusted per capita expenditure amount for the patient populations in each of the 3 benchmark years by combining the initial per capital expenditures for each year with the respective growth and health status indices. This yield risk adjusted per capital expenditures for beneficiaries historically assigned to the ACO in each of the 3 years used to establish the benchmark stated in BY3 risk and expenditure amounts.
5. Weight the most recent year of the benchmark (BY3) at 60%, BY2 at 30% and BY1 at 10% to ensure the benchmark reflects more accurately the latest expenditures and health status of the ACO’s assigned beneficiary population.  This weighting allows a lower MSR to be established since the weighting resulting in a more accurate benchmark.
6. For each performance year, the fixed benchmark would be updated by the projected absolute amount of growth in national per capita expenditures for Parts A and B services under the original Medicare FFS program using data from the CMS Office of the Actuary.

• Option 2 (pp 241-45) (Note:  while this option is not the one proposed, I am including it here since CMS has acknowledged problems with option 1, which are discussed on pp 240-41)—for each beneficiary assigned to the ACO during the agreement period, per capital Parts A and B FFS expenditures would be calculated during each of the 3 years immediately preceding the first year of the agreement period.  These amounts would be trended to the start of the agreement period (same as Option 1), and since Option 2 also requires risk adjustment, the benchmark would be adjusted for health status using the same prospective CMS –HCC risk adjuster , which would be applied to calculate the benchmark in the same way as in Option 1.  The annual per capital expenditures would be adjusted to account for changes in health status.  For beneficiaries without 3 full years of immediately-prior Medicare eligibility (such as beneficiaries who were not 68 in their first year assigned to the ACO) a further adjustment would be necessary under this option.

1. For beneficiaries with less than one full year of prior Medicare experience, either (a) a substitute for their own expenditures in the update amount within the benchmark would be used (that is, substitute the average per capita FFS expenditures for all Medicare beneficiaries during the year they are first assigned to the ACO, adjusted for health status); OR (b) exclude their experience from the shared savings computations.
2. For those assigned beneficiaries with more than 12 months prior Medicare experience but less than 36 months, either (a) compute a weighted-average (using number of months as the weight) that blend their prior expenditure experience and the average per capita Parts A and B FFS expenditures for all Medicare beneficiaries during the year before the first year they are assigned to the ACO, adjusted for health status; OR (b) use only their prior expenditure experience.
3. After the benchmark is adjusted for beneficiary health status, the benchmark also would be updated by the applicable projected amount of growth in national per capita expenditures for Parts A and B services under the original Medicare FFS program (as described in option 1).
4. For the second and third year of the agreement period, no further adjustments would be made for assigned beneficiaries who were also assigned in the first year of the ACO agreement period.  However, in the second and third year, there would also be newly-assigned beneficiaries as well as previously-assigned beneficiaries who are no longer assigned to the ACO. The benchmark would be adjusted to account for these changes by adding the experience of the newly-assigned beneficiaries for the 3 years prior to the agreement period, and by removing the prior experience of the no-longer assigned beneficiaries.  In the case of a beneficiary who was assigned during the first year, not assigned during the second year, and then assigned during the third year, the prior expenditure experience that would be used to adjust the benchmark in the third year would be the same amount initially used for their first year of assignment. These adjustments would yield a benchmark for each ACO that is estimated using beneficiary expenditures for the 3 years prior to the agreement period for only those beneficiaries that were actually assigned to the ACO during that year of the agreement period.  Does anyone else have a headache from trying to follow this?

Additionally, option 2 would require an adjustment for assigned beneficiaries who die during an agreement year.  About 5% of all Medicare beneficiaries die in a single year, and that their average monthly expenditures are often higher during this last year of life compared to the immediately preceding years. For these beneficiaries, the benchmark might therefore not be a fair basis for comparison with actual expenditures for purposes of determining shared savings. This would not be the case for option 1 as that benchmark approach would include the average per capita costs of beneficiaries who died during the benchmark period.  There are 2 methods that could be used to adjust for decedents:  (a) exclude the expenditures of deceased beneficiaries from actual expenditures during the agreement period (preferred method); OR (b) compare average expenditures for each deceased beneficiary during the agreement year to the average expenditure for beneficiaries included in the benchmark.   If the agreement year’s expenditures were 5% or less above the benchmark, no adjustment would be made; if the agreement year’s expenditures were greater than 5 % above the benchmark, an acceptable method for adjusting the accumulated expenditures for deceased beneficiaries would need to be determined.  CMS solicits comments on these or other methods to adjust for decedents.

CMS seeks comments about Option 1 and Option 2 approaches and is looking for other potential approaches.