Editor’s Note:This post is part of a series of Health Affairs Blog posts examining the proposed rule implementing the Medicare Shared Savings Program, issued March 31 by the Centers for Medicare and Medicaid Services. You can read earlier posts in the series by Mark McClellan and Elliott Fisher, Douglas Hastings, and Steve Lieberman.

No proposed regulation in decades has been more eagerly awaited by the US health establishment than the one recently released for the Medicare Shared Savings (or Accountable Care Organization (ACO)) Program mandated by section 3022 of last year’s Affordable Care Act (ACA).  Although largely unnoticed by the lay press and the general population, anticipating and discussing these statutory provisions and regulatory prospects has overwhelmed all other planning issues among virtually all health plans, hospitals, medical professionals and their societies, and health information vendors, providing an unintended but significant economic stimulus to health consultants and lawyers.

This obsession is not illogical: Everyone knows that the status quo is not sustainable, that some forms of payment and delivery system reform are inevitable, and that changes to Medicare have a way of influencing all providers and payers.  In addition, the concepts and objectives of ACOs seem eminently logical and achievable, and the expectations of supplemental revenues from performance-based shared savings and retention of the beneficiary and provider freedoms inherent in traditional fee-for-service are especially welcome.

Finally, we and our patients all seek such a solution for providers to directly collaborate to coordinate care to enhance treatment quality and efficiency without the restrictions and limitations of HMOs or capitation payment models.  The keyword is “accountability” for those patients who they take care of; significantly, this was not envisioned to be a new “insurance” program for enrolled members.

Unfortunately, the proposed rule creates a program that is likely to have few participants.  This is because it is overloaded with provisions to mitigate the likelihood that any conceivable negative possibility will occur, rather than trusting the design and encouraging and supporting participation.  Existing ACO-like integrated systems will find the financial deal inadequate (they want capitation), while provider groups contemplating becoming an ACO will find the organizational and operational requirements and costs excessive.  Unless major revisions are made to this proposed rule, we will have certainly missed a real opportunity for both wide-spread, voluntary enhancements of Medicare’s predominant fee-for-service program as well as general delivery system reform.

Shared Savings as an Overlay to Fee-for-Service

Most health professional and institutional providers are currently paid on the basis of the lesser of their charges or a predetermined schedule of maximum allowances, for each and every service code (Current Procedural Terminology or Diagnosis Related Group code) they render.  These allowances are uniform for all providers of the same type, as opposed to varying based on performance.  Such uniform “fee-for-service” compensation offers providers no incentive to focus on the value or relative cost of each service they render or order, the cumulative cost of their own services or the services their patients receive from others, or the “outcomes” of either their own services or the episodes of their patients’ care.

Efforts to control increasing total costs by either reducing fees or constraining fee increases, e.g., the Sustainable Growth Rate (SGR), have been largely ineffective, as have prior private payer utilization review initiatives.

The vision of “shared savings” is to strengthen fee-for-service by providing the “business case” for providers to devote some of their resources and attention beyond the services they are routinely paid for to services and initiatives that enhance the efficiency of the care and improvement of the health of their patients, but for which providers are not typically paid.  Initiatives such as disease surveillance, case management, treatment protocols, patient monitoring, and care planning should lead to reductions in hospitalizations and rehospitalizations, complications and adverse events, use of higher cost settings and providers, and unnecessary and duplicative services. This should help to achieve the oft-referenced “three-part aim” of better care, improved health, and lower cost.

Eliminating the “One-Sided” Shared Savings Model

Subsection (d) of section 1899 of the Social Security Act (where ACA section 3022 placed the Shared Savings Program) clearly, and by specific legislative intent, establishes a payment methodology for maintaining payment “under the original Medicare fee-for-service program” plus eligibility for shared savings subject to specified conditions and limitations.  Nowhere is there any mention of “shared losses”.  This is not surprising as it was widely acknowledged that this program was modeled on CMS’ Physician Group Practice (PGP) Demonstration, which also included only shared savings.  Furthermore, contained in the only written CMS document prior to this rule, available on their website for ten months since June 7 (“Preliminary Questions & Answers”), is the statement that “An ACO will share in savings if program criteria are met but will not incur a payment penalty if savings targets are not achieved.”

As discussed in this rule, some have questioned the adequacy of the intended shared savings model to provide new ACOs a sufficient incentive to (or disincentive not to) transform their care processes.  In particular, the proposed rule specifically cites the recommendation of the Medicare Payment Advisory Committee (MedPAC) to add the option of a “two-sided” risk model (hence, the derivation of the reference to the alternative as the “one-sided” model).  Unfortunately, the proposed rule does not simply to add the two-sided risk model (“Track 2”) as MedPAC recommended, but it would also eliminate the one-sided risk model after the second year of the performance period (“Track 1”).

Although the proposed rule notes that this option would give a group choosing Track 1 an opportunity to evolve and learn during their first two years, before the risk of shared losses begins, from a practical perspective the group would have no idea of their performance before they assume the risk.  This is because CMS and its contractors could not even begin to analyze the first year’s performance until the third quarter of the second year (in order to wait the proposed six months to allow for 99 percent-plus of claims to be submitted and processed), and probably would not be able to communicate the results of the first year’s performance to an ACO until after the start of the third performance year.

One view is that the real reason for eliminating the intended shared savings model was to reduce the possibility of an increase in program outlays for “unearned” savings.  Unearned savings would be those that are calculated for a population of assigned beneficiaries for whom no changes in care were provided.  Yet the proposed rule contains several safeguards against such a possibility, including a hurdle of a size-related minimum savings rate (MSR) to deal with year-to-year random variations in expenditures; a “net” sharing rate determined by each ACO’s overall quality score, and a required written and enforceable certification of compliance in order to receive payment.

Moreover, an ACO would need to treat its participation in the Shared Savings Program as a serious organizational and financial undertaking.  The impact analysis in the proposed rule cites figures, based on the 2008 GAO analysis of the PGP Demo, that the average start-up and first year operating costs would exceed $1.7 million.  Groups simply cannot make the business case to make that level of financial commitment without being serious about undertaking the significant efforts necessary to receive a return on such an investment.

So why should we care?  First, we should care because we want to encourage the greatest possible participation in this program, especially of entities not currently organized as ACOs As several CMS officials and private experts have suggested –and I agree –, the greatest potential for ACO success is fostering new alliances and organizations that do not yet exist to form.  These are the groups who will find a one-sided shared savings model as more acceptable.

Second, we should care because this intended shared savings model is so fundamentally different from one that includes the risk of shared losses  that incorporating them both, with more clearly distinguishing features, would advantage the Shared Savings Program’s acceptance by both providers and beneficiaries.  These include the details of assignment, the timing and content of information to be shared by CMS, the choices offered to beneficiaries, and the quality measures.  Please see my Health Affairs blog of January 25, 2011 for a more thorough discussion of these distinctions.

Beneficiary Issues

The proposed rule asserts that “this program does not affect the beneficiary’s freedom of choice regarding providers or care”.  Such freedoms are highly coveted and expected for traditional fee-for-service Medicare (but not necessarily for Medicare Advantage).  Unfortunately, the proposed rule includes provisions that suggest that beneficiaries’ freedom of choice will indeed be affected.

Before the start of the agreement period each ACO would be provided with a list identifying (by name, date of birth, sex, and Health Insurance Claim Number (HIC)) each “historically assigned beneficiary” as a potentially assigned beneficiary for the ACO’s agreement period.  This is an attempt to balance two positions.  One is to adopt “retrospective” (as opposed to “prospective”) assignment, in order to  ensure that accountability is for beneficiaries fitting the criteria during each performance year, and to  promote uniformity in care for all beneficiaries.  The second is to facilitate efforts by ACOs to target their care management activities and focus their resources.

The proposed rule does not include a provision that the potentially assigned beneficiaries so identified to an ACO are to be told that their information has been provided to the ACO.  Nor does it provide an option to “opt-out” of such private information transfer.  This is notable as it is proposed that beneficiaries would be extended such an option regarding their monthly claims history (to be requested by an ACO primary care physician).  I believe that beneficiaries should have, and are entitled to, such an option in both instances.

The beneficiary’s freedom of choice comes into play as follows.  Once they are identified to the ACO as being “potentially” assigned, they will be categorized and treated differently by the ACO than those not so identified.  As the purpose for such identification is to then care for these beneficiaries differently, their choices and their freedoms are affected.  If beneficiaries continue with the providers within the ACO, their freedom of choice regarding care, e.g., kinds, amounts, settings, referrals, will be affected.  If beneficiaries do not want their care to be affected by the ACO’s participation in the Shared Savings Program, they would have to choose alternative providers (if these are identifiable and available to them); the proposed rule clearly continues this option to choose alternative providers.

Unless beneficiaries are also given the option, not currently proposed, to “opt-out” of having their claims experience counted in the ACO’s experience but still continue with their current provider(s) with the ACO, their freedoms of choice of providers or care will not be preserved.  This is only an issue if the identities of potentially or actually assigned beneficiaries are provided to an ACO.  For the one-sided shared savings model, assignment could be “invisible” to both providers and beneficiaries.  Retrospective assignment without prospective identification by CMS would preserve all current beneficiary freedoms; of course, ACOs would already know the identities of the patients they treat, and the conditions they treat them for, without needing to be told by CMS.

On the matter of data release, the proposed rule may be incorrect in the interpretation of the “routine uses” exception in the Privacy Act of 1974.  The exception applies when the disclosure “is compatible” with the purpose of the original collection.  A good case could be made that the data was obtained solely for the purpose of adjudicating a unique claim for payment.  Therefore, sharing even four elements of such beneficiary data, for the purpose of identifying beneficiaries who were assigned to an ACO (either historically or ongoing), information that is not essential to the administration of either the Shared Savings Program or the Medicare program, can hardly be considered “compatible”.  Reinforcing this position, in the context of “claims-based reporting” of measures, the proposed rule states “the claims processing system was designed for billing purposes and not for the submission [and, therefore, obviously not use] of quality data”.

A similar lack of compatibility of purpose raises concerns about the legality of including in the monthly claims transfer identification of the provider of service, whose data is also covered by the Privacy Act.  If this interpretation of incompatibility is correct, then consent in some form (including an opt-out) would be necessary from both beneficiaries and the non-ACO providers.

Provider Issues

On numerous recent occasions CMS officials have identified their vision for the agency as a “trustworthy partner” with providers for the continual improvement of health and healthcare for all Americans.  Unfortunately, the proposed rule does not evidence a presumption of trustworthiness of providers.  For example:

  • Although ACOs must submit to, and receive approval from, CMS a “self-executing method” for repayment of losses, 25 percent of any savings would also be withheld by CMS to offset any losses.  This applies not just for Track 2 but also for those choosing Track 1, where no shared losses are possible until year three is tallied (which would probably be  in year five).
  • In the event of early termination of an ACO’s Agreement by CMS, or discontinuation by the ACO, all of the withheld amounts would be forfeited.  There are sixteen specified grounds for termination, all by CMS “in its sole discretion”.  This forfeiture applies even if the reason for an ACO’s discontinuation was in response to changes in the rules made by CMS during the three-year agreement period (changes which CMS would be allowed to make for any but three specified features).
  • As a condition of receiving shared savings payments an ACO must submit a separate written request that certifies to both compliance with program requirements and “the accuracy, completeness, and truthfulness of any information submitted”.  This certification would be  in addition to assurances already contained in the application/agreement process, and could only have been included to impose a higher bar on receiving shared savings payments in a way that could place some of these payments in jeopardy.
  • CMS has historically relied on a personal “attestation” to assure the integrity of a provider’s submitted program information, e.g., for the Physician Quality Reporting System, or for  “Meaningful Use”, accompanied by the possibility of a random audit.  Instead, for all ACOs a detailed audit process is proposed and elaborated for every GPRO quality measure, including one required and two conditional phases, possibly leading to a conclusion of failing to report accurately, which would be subject to termination or sanction.
  • As opposed to setting forth guidelines for the content of “materials, communications, and activities”, the proposed rule would require CMS to approve in advance of initial use, and of subsequent revisions, that which is “used to educate, solicit, notify, or contact Medicare beneficiaries or providers and suppliers regarding the ACO and its participation”.  In addition, such materials must not “suggest that we endorse the ACO, its ACO participants, or its ACO providers/suppliers”.
  • The proposed rule would require that ACO providers/suppliers (as well as the ACO, ACO “participants”, and their contracted entities) must maintain and submit to “the Federal Government or their designees, the right to inspect” every conceivable source of evidence regarding their activities in the Shared Savings Program, and that all such evidence be maintained “for a period of 10 years from the end of the agreement period or from the date of completion of any audit, evaluation, or inspection, whichever is later, unless we determine there is a special need to retain a particular record or group of records for a longer period”.
  • A lot has been said and written about finding the right point on the regulatory continuum between flexible and permissive at the “low” end, to maximize provider participation and innovation, and rigid and prescriptive at the “high” end, to minimize program risks or advance specific agendas.  However, when it comes to “patient-centeredness”, probably the least developed aspect of patient care, the proposed rule is toward the high end, requiring that an ACO must have in place all of eight very specific categories of detailed, although largely untested (or even possible), requirements.  Examples include  “patient involvement in ACO governance”, “a process for communicating clinical knowledge/evidence-based medicine to beneficiaries in a way that is understandable to them”, and “internal processes for measuring clinical or service performance by physicians across the practices, and using the results to improve care and service over time”.

Statutory Guidance Issues

There are several provisions in the proposed rule that appear to be contrary to both statutory guidance and to good public policy.  One is the proposal to consider only complete reporting of the specified quality measures as sufficient for the first performance year to meet the “quality performance standards”.  Subsection (b)(3) contains two distinct provisions: “(B) Reporting Requirements” and “(C) Quality Performance Standards”.

In order to assure beneficiaries and others that quality is a priority, some standards should be imposed at the outset.  Of course, subparagraph (C) included the requirement for the Secretary to “seek to improve the quality of care furnished by ACOs over time by specifying higher standards, new measures, or both”.  Although the proposed rule suggests that the reporting-only decision “would provide CMS with the opportunity to learn … using data reporting for the first reporting year”, the timing (of collection, analysis, etc.) would not allow for any impact until after the third performance year already commenced.

Another questionable provision is the proposal for the one-sided model, under subsection (d), to compute savings, once the minimum savings rate (MSR) is met, as only the difference between actual expenditures and two percent below benchmark expenditures, rather than the full difference. (The full difference is allowed only for smaller ACOs under specified circumstances).  Although  this approach is certainly allowable for “other payment models” pursuant to subsection (i), subsection (d)(2) includes no such authority.  Obviously, this proposed provision would save the program some money, while denying some ACOs some added revenue.

But probably the most questionable provision, and the one most likely to adversely impact the participation of multi-specialty faculty practices and the more complex beneficiaries treated by them, is the one to base beneficiary assignment on their utilization of primary care services by only primary care physicians, not by any physician.

The statutory language directing the Secretary (subsection (c)) is short and clear: “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 in subsection (h)(1)(A)”. [underscore added]  This subsection cites services of “a physician (as defined in section 1861(r)(1))”, i.e., MDs and DOs.  There is no mention of the specialty of a physician in the context of beneficiary assignment.  This was not an oversight, and was the legislative intent.  The words “appropriate method” refer to issues of computation, such as plurality vs. majority, not to specialty; the words “primary care services” refer to issues of procedure codes, not to specialty.

The only mention of physician specialty was included in another subsection, (b)(2)(D), dealing with eligibility requirements, specifying that “the ACO shall include primary care ACO professionals that are sufficient for the number of Medicare fee-for-service beneficiaries assigned to the ACO under subsection (c)”.  That is, once the number of assigned beneficiaries is determined without regard to physician specialty, a separate determination of sufficiency of primary care “ACO professionals” must be met, which includes “physicians” and “practitioners” described in subsection (h)(1)(B).

Besides this concern for respecting statutory guidance and legislative intent, it is essential to recognize the existing care patterns of Medicare beneficiaries and assign them accordingly.  Non primary care physicians render sixty percent of all “primary care services” to Medicare beneficiaries, with the majority of these by medical specialists.  For the most vulnerable of Medicare beneficiaries, those with multiple chronic conditions, the vast majority of their primary care is rendered by their specialists.  We have a national under-supply of primary care physicians, especially in urban areas, and we should not ignore the reality that Medicare beneficiaries face.  Finally, the assignment model during the full five years of the PGP Demo included the primary care services of any physician, not just primary care physicians.


I believe the opportunity for accountable care organizations is exceptional; the effort of the involved federal agencies is extraordinary; the prospects for participation in the shared savings program, at least in its current proposed rule form, is poor.  However, with a major makeover, the prospects for success, measured by provider participation, beneficiary acceptance, and program savings, can be excellent.