Certainly no Medicare provision of the Affordable Care Act (ACA) has generated more interest among health care providers, policy analysts, and consultants than the Medicare Shared Savings Program (MSSP) for accountable care organizations (ACOs). Because there are so many design elements for which the Secretary of Health and Human Services must “determine”, “establish”, and “specify” the details, all potential participants have been anxiously awaiting the proposed rule before deciding how, and if, to proceed. These details will have important implications for the future of traditional fee-for-service (FFS) Medicare and the program’s acceptance by physicians and beneficiaries.
The MSSP is based on the Physician Group Practice (PGP) Demonstration and incorporates a “shared savings” model (SSM) of potential bonuses for eligible groups of providers that meet annual performance standards of beneficiary quality and per capita expenditures. Payment for covered services will remain fee-for-service, and all of the fundamental and enduring FFS freedoms of choice, equality, and privacy are to be preserved. There will be no financial penalties or down-side risks for these groups other than their unreimbursed services and other investments. The SSM is not yet another version of managed care by at-risk entities for enrolled beneficiaries.
In spite of the clear Congressional intent to include this model, some are now questioning its adequacy to achieve delivery system “transformation” and its possibility for “unearned” bonuses, i.e., groups participate – do nothing – hope for a favorable random outcome. Therefore, now being considered for the MSSP is the “option to use other payment models”, as authorized but only cryptically referenced, such as “partial capitation”, or “two-sided risk” recently suggested by MedPAC.
The “sentinel” issue should not be which model to promote, but which models and their differentiated design elements. This would optimize initial participation, ongoing “upward” progression (from one model to the next), and ultimate program impact.
The SSM will allow interested groups that are not willing or able to engage in risk the opportunity to form, develop, and mature as an ACO. Any concerns for “free-loading” can be addressed by requiring “proof-of-capability” rules for participation and “proof-of-compliance” rules for payout. The alternative risk models will allow integrated delivery systems, managed care organizations, and health plans with experience and resources a different and higher set of performance requirements and rewards. “One size does not fit all.”
However explained, risk models move FFS toward “managed care” for both beneficiaries and physicians. Physicians should focus on the plans of care for patients, not their records of cost; this is the difference between accountable care and managed care. Seeking improved quality should come first, and reduced cost will follow; better health costs less.
Attributing Medicare Beneficiaries To ACOs
One of the major design elements for ACOs identified by CMS in their November “Request for Information” is “the process of attributing beneficiaries to an ACO … important to ensure that expenditures, as well as any savings achieved by the ACO, are appropriately calculated and that quality performance is accurately measured”. The following table lists suggested distinguishing features for beneficiary attribution among the SSM and two other likely and distinct ACO payment models permitted under the MSSP: “2-sided risk” (FFS with variable, symmetric or asymmetric, shared gains and losses) and “partial capitation” (fixed payment with “limits”)..
|Medicare Accountable Care Organizations: Anticipated Program Models
Key Distinguishing Features – Beneficiary Attribution
|Services:||Primary Care||Primary Care||N/A|
|Practitioners:||Any Physician||PCP + Med Spec||N/A|
|Identified to ACO:||No||Yes||Yes|
|Beneficiary Notice:||Yes (By CMS)||Yes (By ACO)||Yes (By ACO)|
The implications of when beneficiary attribution (or “assignment” or “alignment”) is made, either at the start (“prospective”) or the end (“retrospective”) of the performance period, are profound. With retrospective attribution, which was used in the PGP demo, any care coordination strategy would have to be applied to all applicable FFS beneficiaries without discrimination as they would all be potential attributees. The absence of specific identification, especially at both the start and end, i.e., “invisible” attribution, would assure the ultimate non-discrimination and privacy protections, and would avoid the need for any beneficiary action. This option seems appropriate for the SSM.
There would also be no need to give SSM ACOs beneficiary-level information as to the benchmark or actual claims experience. They should instead receive an aggregate accounting and comparative data summaries of their categories of beneficiaries. Of course, the ACO would know from its own records who it is treating, for what condition(s), and who may benefit from pro-active interventions. In the interest of “transparency”, beneficiaries should still receive information from CMS as to the general nature of the ACO program and the identities of eligible ACOs and associated professionals within their communities, and ACO professionals should also be obligated to identify themselves to beneficiaries and should want to promote their advantages, thereby encouraging patient engagement and compliance.
For risk-based models, however, it would seem necessary to have “prospective” attribution (and financial benchmark and utilization information) so they “can target care coordination strategies to those beneficiaries whose cost and quality information will be used to assess the ACO’s performance.” Consequently, the freedoms of the targeted beneficiaries would clearly be different from the non-targeted. This will be so even if these beneficiaries are given choices to (a) acquiesce or consent (“opt-in”), or (b) object, and have their experience excluded (“opt-out”); having to leave their existing doctor(s) should not be considered a “choice”. Unless there is also a beneficiary “lock-in”, which might be hard to justify under the statute, ACOs with prospective attribution would end up having accountability for beneficiaries who choose to receive some, or even most, of their care from non-ACO members. Adding the sub-option of a retrospective “reconciliation” to reduce this would introduce considerable, unwanted possibilities for beneficiary risk-avoidance (i.e., “de-selection”).
ACO Assignment Should Not Hinge On Physician Specialty
Other important elements of attribution are which services, for what professionals, and how assessed. The MSSP specifies that “The Secretary shall determine an appropriate method to assign Medicare fee-for-service beneficiaries to an ACO based on their utilization of primary care services provided under this title by an ACO professional described …” This language clearly suggests the legislative intent to consider the primary care services of only physicians and without regard to specialty, as physicians are the only professionals described and there is no reference to specialty, such as “primary care ACO professional” used elsewhere. Also, the PGP demo based attribution on services of physicians without regard to specialty.
Despite this language and experience, some are recommending that the basis should be limited to primary care practitioners (PCPs), advancing the usual arguments. Unfortunately, the nation has an under-supply of PCPs, particularly in urban areas, which will not change soon enough for a national ACO program. More importantly, Medicare beneficiaries usually have multiple chronic conditions and are treated by multiple specialists. For many such beneficiaries, even if they retain a relationship with a PCP, the PCP would not be responsible for most of their “primary care”. Therefore, the decision by government in this regard will have significant implications to the qualification and the scope of accountability of many organizations, particularly multi-specialty practices and academic medical centers. For the SSM, to comply with the legislative intent and to allow the greatest flexibility, the basis should be without regard to specialty; however, to remove any doubt as to the linkage of accountability with “responsibility”, a “majority of charges” method, more rigorous than the “plurality” in the PGP demo, would seem appropriate. For the risk models, assuming prospective attribution, the alternatives listed might be more appropriate.
As for these attribution issues, I would encourage CMS to use its discretionary authority to select distinguishing and escalating features on all other critical design elements, such as: (1) the financials, e.g., the percentage shares and maximum amounts of bonus and risk (the controversial “minimum saving percent” could be waived for the risk models); (2) quality measures and standards, e.g., “core” vs. “expanded” sets, levels of attainment (and improvement); and (3) eligibility requirements, e.g., patient-centeredness criteria, information technology capabilities.
Most importantly, I encourage all stakeholders to support having multiple MSSP models, including a distinct shared savings model with distinguishing features that preserve FFS and full beneficiary freedoms, and giving these the chance to operate and succeed along with the other ACA payment models in the spectrum of delivery system reform.