Six strategic choices made by CMS in the final ACO rules


1. The final ACO rules largely maintain the demanding economic parameters for mature ACOs (Track 2) found in the originally proposed version (relative, for example, to the original PGP demonstration project):

  • Potential for both downside and upside reward.
  • Maximum shareable savings of 60% (less than in the original PGP demo); and the rewards are limited to an upper bound of total costs. In this regard, the ACO contract payoff locks a lot like your classic “collar” financial option (see graphic below).
  • The starting cost benchmark based on the ACO’s actual historic costs (removing any “easy wins” for today’s low cost providers).
  • The cost benchmark is reset each three year period based on the ACO’ costs, thereby incorporating all the savings and applying a ratchet to push ACOs toward continuous improvement.
  • An anticipated start-up investment and first year operating cost of ~$2M for each ACO (clearly a non-trivial sum for most candidates).

2. Most the adjustments make the ACO model easier for inexperienced providers to test and adopt. These changes include:

  • Making the “beginner track” 1 more attractive by extending the “upside only” model for shared savings to the full 3 year contract term (vs. just the first two years of the term under the proposed rules). The final rules also make all the savings created available for sharing once a certain threshold is reached (the threshold can range from 2.0% to 3.9% lower costs depending on the number of patients) instead of just the savings above that threshold.
  • Removing overly prescriptive rules for managing an ACO (e.g. on organizational and governance structures) which would be expensive for newly forming ACOs to put in place. Also removing redundant requirements (e.g. 50% EHR adoption duplicating components in the quality metrics).
  • Quarterly interim feedback on which members are likely to be assigned to the ACO (versus the annual look-backs). (Note this is weaker than it might first appear – it is not prospective assignment but a prediction and provider still lack many tools to keep a patient’s care within the ACO).
  • More flexibility for providers to make the commitment by creating multiple start-dates in 2012 when ACOs be launched.
  • Creating an advance payment fund of $170M for rural providers and physician-owned hospitals to finance the ACO infrastructure costs

3. Overall this approach is consistent with a strategy of measured, predictable ACO roll-out which will allow the provider system time to structurally adjust, for the growth in the Medicare eligibles to fill capacity idled by ACO care patterns and thereby minimize provider resistance:
  • By retaining the relatively tough economics for mature ACOs, CMS will keep initial adoption to a small beachhead covering 1M and 5M Medicare eligibles (2-12% of those in FFS vs. Medicare Advantage and therefore eligible for the CMS ACO program – largely unchanged from the forecasts for the preliminary rules).
  • On the other hand, by easing the “on ramp” for providers to try out the model, CMS is also looking for a solid pipeline of maturing ACOs to expand the model beyond 2015. A credible pipeline is important to ensure that the broader provider system take the ACO model – and its associated care patterns – into account in their strategy making.

4. CMS has raised the upper bound of sharable savings from 10% to 15% of benchmark costs for risk taking ACOs (track 2) (and makes a relatively similar increase for upside only track 1). An upper bound mitigates against the risk that providers might avoid costly but appropriate care in order to create sharable savings. The trade-off is that providers might “game the system” by exploiting only the savings opportunities that reach the shareable level. By increasing the upper bound, CMS appears to have decided that the quality indicators offer sufficient protection against excessive care reduction and there are more savings opportunities out there which can be had without reducing care below appropriate levels (perhaps CMS was nudged in this direction by the successes of commercial ACOs seeing savings on the order of 13% of costs). This is particularly interesting given the simultaneous decision to reduce the number of quality indicators included in the ACO scoring from 65 to 33.

5. Broadly, the CMS strategy in pruning quality measures seems to have been:
  • Remove care quality measures for conditions where any gaps in care delivery are quite likely to lead to a high cost acute event within the 3 year measurement period of the ACO contract and before a patient may naturally rotate to another provider system (in other words, conditions where costs over a year interval is a pretty good indicator of care quality)
  • Retain care quality measures for conditions like diabetes where it may take time for low quality care to lead to a high cost acute event (e.g. a hospitalization). This will discourage any “betting” that a patient may move on to another provider before the high cost event occurs.

For example, in the “at risk population” category of measures where adherence to specific evidence-based medical standards are monitored. Notable areas of trimming were: heart failure (where 6 measures were eliminated and only beta blocker therapy was retained); CAD (where 4 measured were eliminated and only anti-cholesterol therapy and ACE, ARB therapy were retained); COPD (where all 3 measures in the proposed rules were eliminated). At the same time, diabetes saw much less reduction in metrics: the proposed rules had 10 metrics in the “at risk” category – one of which was a composite of the others — and retained 6 of them.

6. Finally, CMS has shifted its strategy regarding ACOs and community health centers to allow them to form their own ACOs (vs. the original proposed rule offering an additional reward for ACOs to include community health centers). Given the critical role that Medicaid plays for many community health centers, CMS appears to be supporting the idea emerging in multiple corners (the professional literature, in some of the legislative proposals for leading laboratory states such as Massachusetts and implicit — we would argue– in some of the emerging operating models of major providers) that an ACO model could make Medicaid a more viable financial proposition for providers.