Response to CMS-1461-P Medicare Program, MSSP, and ACOs

Re: CMS-1461-P Medicare Program; Medicare Shared Savings Program;
Accountable Care Organizations

The Health Care Transformation Task Force (Task Force) is pleased to provide input on the Centers for Medicare & Medicaid Services (CMS) in response to proposed policy and payment changes set forth in CMS-1461-P Medicare Program: Medicare Shared Savings Program: Accountable Care Organizations (ACO) notice of proposed rulemaking (NPRM).

As we described in previous communications, the Task Force is an emerging group of private sector stakeholders that is coming together to accelerate the pace of delivery system transformation. Representing a diverse set of organizations from various segments of the industry – currently including providers, health plans, employers, consumers and academic institutions – we share a common commitment to transform our respective business and clinical models to deliver the triple aim of better health, better care and reduced costs. Our organizations aspire to put 75 percent of their business into value-based arrangements that focus on the Triple Aim of better health, better care and lower costs by 2020. We hope to provide a critical mass of policy, operational and technical support from the private sector that, when combined with the work being done by CMS and other public and private stakeholders, can increase the momentum of delivery system transformation.

The Task Force’s shared principles reflect our commitment to a specific timeline for the migration from fee-for-service toward payment models that promote patient-centered care, improved population health, and lower total cost of care. Our outputs reflect agreement on common private and public approaches that will best facilitate transformation.

Current ACO Environment

In considering revisions to the MSSP regulations, it is essential for CMS to consider the current reality of the ACO marketplace. The initial program approach has generated great interest and participation. CMS has made significant efforts to be an effective payer partner and has made remarkable progress in providing information, data and claims to the participants. Participants have tried many new approaches to improving and coordinating care. Beneficiaries have experienced better care and better outcomes and we have all learned a great deal as a result.

However, as ACO participants, consultants and industry experts gain greater experience with ACO Shared Savings Models in general, it has become increasingly clear that to be sustainable, the financial opportunity must provide sufficient reward to support the investments needed to improve care and yield a meaningful return. Unfortunately, many are coming to the conclusion that the MSSP does not meet that test.

Many aspects of the current regulations make it more difficult for the ACOs to obtain a positive financial return. The minimum savings rate, the beneficiary data-sharing opt-out process, delay in receiving claims, inability to communicate with beneficiaries, uncertainty of the benchmark, expected rebasing that would decrease savings opportunities in future years, and the inability to use more advanced approaches with skilled nursing facilities, home healthcare, and telemedicine, all limit ACOs’ ability to generate sufficient savings. These factors, as well as the newness of attempting to manage a population’s experience in the “fee-for-service open network, no referral” world creates great uncertainty about ACOs’ ability to deliver sufficient savings to obtain a return on their investment. Furthermore, the program’s retroactive beneficiary assignment causes instability in the ACOs’ benchmarks1 and creates a moving target in terms of identifying the patients for whom the ACO would be held accountable.

The right strategy for CMS to get maximum impact on improving quality and decreasing costs is to better help ACOs be successful in every Track and continue in the program. We recommend structuring the program to afford more opportunity to explore alternative care approaches, with greater potential to share in more of the savings to encourage potential providers to invest more in care coordination and therefore produce meaningful improvement. Most of our comments, therefore, take positions that increase the likelihood of wider provider participation in the program and further investment in infrastructure that support improved, patient-centered care delivery and enable transition to full risk. While there is much in the NPRM that suggests that CMS understands the need for this approach, there are several areas in the NPRM that we believe should be revised. These are:

  • Program Evolution – Encourage two-sided risk, but acknowledge the diversity of ACOs; allow a prospective attribution option; support further exploration of bundled payments, pre-payments or capitation in MSSP.
  • Benchmarking – Explore regionally-based benchmark updating and resetting in addition to refinements to historically-based benchmark option; evaluate improvement in risk adjustment methodology.
  • Regulatory Flexibility – Support waiver of fee-for-service payment policy, beneficiary communication restrictions and benefit design parameters.
  • Attribution – Respect beneficiary choice in primary care provider; allow ACOs to identify providers on whom attribution should be based.
  • Data – Support CMS run data-opt out process; Provide consist and current data including for beneficiaries who have received primary care, make public not just descriptions of MSSP formulas, but the code and formula itself; evaluate inclusion of substance-abuse claims in at least aggregate form.

Program Evolution

Encouraging 2-Sided Risk and Beyond

With so few ACOs selecting two-sided risk, CMS makes several proposals to encourage two-sided risk:

  • Reducing the shared savings rate to 40/60 in Track 1 for the second contract
  • Introducing a variable MSR/MLR in Track 2
  • Creating a new Track 3 with a set 2 percent MSR/MLR and a shared savings rate of 75/25

Yet, the impact statement assumes little success of these proposals, indicating that 90% of ACOs will still most likely choose Track one in 2016. Half of all MSSP ACOs started in 2012 and 2013, and if 90% of those choose Track one, participation in two-sided risk would be little better than it is today.2 The core issue is whether the benefits (increase in savings rate, changes in MSR/MLR, caps on savings and losses, waivers that make savings more likely) outweigh the risk of taking on losses. We applaud CMS for seeking to mitigate some of the downside elements, thereby modifying the current balance to incentivize two-sided risk. ACOs participating in two-sided risk will be more successful at reaching the programs goals than other ACOs. However, the market reality is that for many providers, accepting downside risk exposure at this stage in the evolution of the MSSP appears ill-advised. There are just too many uncertainties regarding benchmark methodologies, ability to coordinate care in the FFS market, data availability, as well as CMS and provider operational challenges.

Changes to Track 1

While we agree that a two-sided risk arrangement provides more incentive for an ACO to reduce costs and coordinate care, we suggest CMS keep in mind that even absent downside risk, ACO participation in the upside only model has been a good investment for CMS. The incentive to create savings is still present and has led to systematic efforts to increase value for both ACOs and CMS. As discussed in changes to Track 2 and 3 below, while movement to two-sided risk is important, one-sided risk still holds short-term opportunity for CMS.

ACOs come in all sizes and forms. Some are set up by organizations with a history of taking on risk, others are created by small physician practices banding together for the first time. CMS should acknowledge those differences when offering options to increase financial risk. The organizational skills to manage financial risk are not necessarily the same ones that result in excellent care coordination or other activities that reduce costs. Moreover, for small ACOs, the variable MSR under Track 1 makes it harder to obtain shared savings compared to larger ACOs, which means they are less likely to avail themselves of the savings necessary to recover their investment and generate the reserves necessary for two-sided risk. To encourage MSSP participation and drive CMS closer to the recently announced HHS goals for value-driven care, CMS should maintain the 50/50 split into the second contract—supporting those ACOs that are skilled in generating savings, but not yet prepared for financial risk.

Furthermore, CMS should explore whether, for an ACO that generates sub-MSR savings in two or more contract years, it might pool the spending across the years and recalculate the MSR to see if the ACO meets a valid MSR over the longer time period.

Finally, we strongly urge CMS to reconsider its policy on how to account for quality in the shared savings calculation. Under current policies, the maximum shared savings rates for Tracks 1 and 2 are only achievable by a small subset of ACOs in years two and three when rates are based on quality performance rather than just quality reporting. The reduction in the shared savings rate that the majority of ACOs can expect when factoring in quality is a major deterrent to continuing participation. CMS should reward ACOs by permitting the maximum shared savings rates when they are able to achieve a quality achievement or improvement score at the median. So in Track 1, an ACO whose quality performance is above the median would get more than a 50% share, while those below median would get less than a 50% share.

Changes to Track 2

We are concerned that the proposed changes in the NPRM actually have the effect of making Track 2 less attractive. The primary driver of low participation in Track 2 is that the 10% increase in the shared savings rate is insufficient to encourage ACOs to take on risk. Changing the MSR/MLR does not address this fundamental problem and could make the Track less attractive by making savings harder to achieve for smaller groups otherwise willing to take on risk. CMS should explore additional opportunities to make Track 2 more attractive to providers.

CMS should continue to explore opportunities to make Track 2 (and two-sided risk in general) more attractive to providers to encourage faster movement in that direction. For example, a strategy missing from the proposed rule is making the MSSP models more viable for ACOs by increasing the number of their patients in ACO models through contracts with commercial payers. CMS should play a leadership role in creating efficiencies between Medicare and private sector ACO requirements to facilitate alignment.

Changes to Track 3

ACOs are divided over whether prospective or retrospective attribution is ultimately more beneficial to an organization’s overall efforts. Some ACOs crave the certainty that comes with an upfront identified population of patients under prospective attribution. Other ACOs emphasize the importance of accuracy in being held accountable only for the beneficiaries for whom they have provided care. Both camps seek certainty with regard to their assigned patient population and during times of transition, few things are more valuable. In a retrospective model, CMS is certain that the beneficiaries for whom they pay out savings received services from the ACO and it is on the basis of those services that the savings were generated and not due to beneficiary selection by the ACO. Similarly, the ACO is certain that the beneficiaries on whom savings are based received services from the ACO.

However, this presents a false dilemma. The closer the prospectively attributed population matches the retrospectively attributed population; issues of uncertainty and inaccuracy are diminished.

Therefore, we suggest that CMS not only institute a prospective option, but also find ways to improve prospective attribution such that the need for retrospective reconciliation and the retrospective attribution model as a separate option is less necessary. Currently, CMS reports a 24 percent difference between prospectively assigned populations and those assigned retrospectively. However, that discrepancy decreases to 17 percent after accounting for eligibility. It is in that 17 percent where the uncertainty for most ACOs lies.

While not clear in the data, one source of difference is a beneficiary moving to another geographic area. These beneficiaries should be excluded from the prospective list (just like eligibility exclusion) if they move in the first half of the year. There is no way for an ACO to game this situation, and there is no way for an ACO to maintain care coordination of these patients.

Another situation that should trigger a potential change in eligibility is when a patient starts a long-term care arrangement. CMS could exclude those patients who do not have a visit to an ACO provider after the admit date.

CMS should continue to conduct detailed data analysis (such as that reported in the proposed rule) and continue to solicit stakeholder feedback on how to reduce the difference between prospective and retrospective lists. The less variation between prospective and retrospective lists the more certainty there is that beneficiaries are linked with their chosen primary care provider.

CMS should continue its monitoring for gaming through quality measures and consideration of overall population risk.

While prospective attribution is being refined, CMS should allow ACOs to elect which attribution method they prefer. We are not aware of any circumstance where the preliminary attribution list for a retrospective ACO cannot serve the same purpose as a prospective attribution list for a prospective ACO in determining waiver applications and other such decisions that benefit from knowing who the beneficiaries are at the start of the performance year.

Benchmark Adjustments for Prospective Attribution

We agree with CMS that timeframe adjustments are needed to operationalize prospective attribution and concur with the proposed adjustments.

Shared Savings/Losses

CMS should improve the savings opportunities in both Tracks. Decreasing Track one to 40% simply makes an uncertain opportunity almost definitely unattractive. Lacking adequate return, ACOs are likely to underinvest and therefore not be successful which creates a self-fulfilling prophecy rather than an effective pathway for risk. We agree that an increase in the sharing rate is necessary to incentivize increased participation in two-sided risk. The Track 3 approach is sufficiently different from our proposed Track 1 savings rates. The variation in the shared savings cap is also an incentive to move ACOs to two-sided risk. Along with CMS, we place great value on the move to two-sided risk as an accelerator to reaching the program’s goals; something that more than compensates for the reduced savings to CMS.


Similar to the attribution challenge, it is not really possible to determine the “best” MSR/MLR for all ACOs. As CMS points out, setting the variability corridor at 2% was somewhat arbitrary. The scaling by number of beneficiaries in the ACO only accounts for one consideration in a multifactorial situation. The MSR/MLR that is appropriate for a given ACO is best determined by a combination of factors which include an ACO’s ability to handle risk, its number of beneficiaries and its local market. While the number of beneficiaries in the ACO is a known quantity to both CMS and the ACO, the local market and an ACO’s individual ability to handle risk are more difficult to discern. In other comments, we have attempted to address the geographic aspect by supporting (see section on Benchmarking) an option of moving to a regional benchmark, which makes the local market more of a known quantity.

The remaining aspect unaccounted for is an ACO’s willingness to absorb financial risk. CMS should not attempt to substitute its judgment for that of the ACO on their ability and willingness to bear risk. An MSR/MLR of zero may be seen as heightened risk to some ACOs, but others will view it as more likely to result in shared savings. As discussed under Track 2, there is no evidence that moving the MSR/MLR slightly in one direction or the other will make a dramatic difference. This means the choice of the individual ACO remains one of the most critical factors in the appropriateness of the MSR/MLR. Rather than attempt a guess, CMS should allow an ACO to select from a few MSR/MLR options in the range of 0 to the size based MSR/MLR available if they were under Track 1. However, the preservation of the symmetry in the MSR/MLR creates protection for CMS.

Beyond Two-Sided Risk

Shared savings is an excellent transition to a value based payment model, but its very construction means it cannot be the end point of value based payment. Changes to the underlying payment system such as full capitation, risk-adjusted capitation, bundled payments and prepayment are the logical evolution of the shared savings model. We encourage CMS to use the final rule to begin a public discussion about where its vision for underlying payment system reform is going and to begin to put out details about what is next and when. Just as with ACOs, some providers are more ready than others.

Benchmarking and Rebasing

While a transition to regional benchmark is the best option presented by CMS in the rule and we discuss it at length, the Task Force believes there is an even better option that takes into account the variability between ACOs. Because CMS cannot predict the best pace of the transition for each ACO, it should provide a set of options for moving to a regionally-based benchmark. These options would be more influenced at the beginning of the transformation process by historical spending, but blend in regional spending over time. Providing such a menu of options (including maintaining a historically-based benchmark methodology for another contract period, one or two transition glide paths, and a complete conversion to a regional benchmark) would allow more advanced ACOs to move more quickly to beating a regional target, while at the same time providing a pathway to success for less experienced ACOs.

Although CMS lays out several options, which we discuss in great detail below, we are concerned about a lack of data and detail regarding those options. We urge CMS to consider issuing a proposed rule to propose one or a combination of options. At a minimum, CMS should issue a final rule with comment to allow stakeholders to weigh in on the details of the finalized option.

Taskforce Support for Option 5 MSSP Transition to Regional Benchmarks

We support option 5 as most aligned with the principles of the Task Force. There are core elements to making this option successful: defining the region, managing changes in patient acuity and adjusting the pace of the transition. CMS should flesh out these proposals and model them to receive the best input from the stakeholder community. We offer these preliminary thoughts on the three domains.

  1. Defining the Region: Every county where at least 10% of the self-attested patients or the preliminarily or prospectively assigned beneficiaries to the ACO have their primary residence is in the region.
  2. Defining the Comparison Group: All Medicare FFS beneficiaries (including those assigned to other ACOs) not preliminarily or prospectively assigned to the ACO in the region’s counties that make up the comparison group.
  3. Dealing with small numbers: For a few ACOs, the region might not contain enough beneficiaries in the comparator group for a valid comparison. While the inclusion of beneficiaries assigned to other ACOs should go a long way to preventing small number problems, CMS should expand the included counties to any county with a preliminarily assigned beneficiary. If a valid minimum of beneficiaries in the comparison group is still not attainable, then the group can be expanded to contiguous counties.
  4. Risk adjustment: Use HCC coding comparison between the ACO assigned beneficiaries [DL1] and the comparison group to create a risk adjustment factor. This factor should be adjusted in both directions when health status changes. CMS should study whether the MA adjustment is valid or whether an ACO specific adjustment is needed. CMS should also explore the feasibility of concurrent risk adjustment which could be superior to the current prospective model.
  5. Transitioning from ACO historical costs to regional benchmark: Offer at least three paths for those organizations that choose to transition-
    1. Below regional benchmark at the end of the first contract
      1. 2nd contract: 50% (Historical Benchmark) / 50% (Regional Benchmark)
      2. 3rd contract: 20% (Historical Benchmark) / 80% (Regional Benchmark)
      3. 4th contract: 100% (Region)
    2. Above the regional benchmark at the end of the first contract
      1. 2nd contract: 80% (Historical Benchmark) / 20% (Regional Benchmark)
      2. 3rd contract: 50% (Historical Benchmark) / 50% (Regional Benchmark)
      3. 4th contract: 20% (Historical Benchmark) / 80% (Regional Benchmark)
      4. 5th contract: 100% (Regional Benchmark)
    3. Move straight to regional benchmark.
  6. Achieving Budget Neutrality: We believe this benchmark is a sustainable model that will attract more ACOs to the program, generating a greater denominator from which to generate savings to offset the shift to regional benchmarks for those in transition path (a). Furthermore, we believe that those in transition path (b) will almost certainly continue with the program in the 2nd contract and likely will continue in the 3rd contract. The reduced paid out savings (but unchanged program savings) due to the inclusion of the region in the 2nd and 3rd contract will also offset the transition costs for those in path (a). We urge CMS to consider this stakeholder feedback when examining the budget neutrality of this option.

Comments on Option 4: A 6-Year ACO Specific Benchmark

We agree with CMS that this change will encourage more ACOs to remain in the program than resetting the benchmark for each contract based on ACO specific historical costs. However, we believe that the transition to a regional benchmark is an excellent option due to its greater equity, greater incentives for continual improvement by high and low cost ACOs alike and established stakeholder support. We are concerned that the existing ACOs at the end of their first agreement period will not have sufficient time to understand the implications of the final program regulations prior to having to commit for the 2016 performance year. In addition, with only one year of fully reconciled data, it will be challenging for the ACOs to make a truly informed decision as to whether they are ready to move to a two-sided risk Track.

By extending the contract period, CMS would provide the ACOs more time to determine which Track would be the best next move. CMS would then have more data to determine which ACOs should not be allowed to continue in the program. These benefits are true even outside of rebasing. Therefore, CMS should extend the agreement period from three years to five years under all models for this and all subsequent agreement periods.

Taskforce Support for Option 3: Regional Updates to Benchmarks in All Contract Years

While unnecessary in a transition to regional benchmarking, if that approach is not chosen, CMS should consider replacing the national update factor with a regional update factor. Using regional updates increases the likelihood that savings are the result of ACO-specific improvements furthering the goals of the MSSP. If CMS moves toward regional updates, it should revisit its statistical analysis of the minimum shared savings rate for Track 1 to see if the power would be increased if it were to move to regional benchmarks and therefore a lower MSR is possible.

To define the region for the regional update factor described in the previous paragraph, we refer CMS to the definition of region described previously in our discussion under Option 5.

Comments on Option 1 and 2: Even Weight of Prior Years and Returning Earned Shared Savings to the Benchmark

We support the transition to regional benchmarks as an excellent solution to increasing the equity of and sustaining the MSSP. However, we recognize that some ACOs would prefer to reset the benchmark at the end of each contract on ACO specific costs (either as part of a blend of historical benchmark and regional benchmark or solely on a historical benchmark). Both the proposal to evenly weight the historic baseline years and add back the savings in resetting the benchmarks would increase the sustainability of the MSSP; however, we stress our commitment to more wholesale reform in the near term.

Should Option 2 be adopted, we recommend one refinement. We encourage CMS to return to the benchmark all program savings. If CMS does not do so, then with each new contract the shared savings rate for the ACOs will effectively be cut in half. This will not provide sufficient incentive for providers to remain in the program.

Risk Adjustment

We take this opportunity to highlight the absolute importance of risk adjustment for ACOs. There are two types of risk in health care. The first risk is that a person has an unavoidable accident, or that an ACO attracts a population with unusually high burden of disease. This is insurance risk, which is addressed through risk adjustment. The second risk is that a person will receive sub- optimal healthcare; this is risk the ACO can influence. Better risk adjustment of insurance risk

leads to better ACO programs. An ACO program that transfers too much insurance risk to health care providers is not sustainable. We recommend that CMS allow for risk scores to go up as well as down for continuously enrolled beneficiaries. CMS should transparently evaluate risk adjustment methodologies including concurrent models and models more aligned with Medicare Advantage.

Regulatory Flexibility

As recognized in the NPRM, there are many regulatory policies in place that make sense in a FFS marketplace but limit the ability of ACO to coordinate and improve care in the MSSP context. The waivers discussed in the proposed rule are all potential tools for an ACO to improve care coordination and reduce costs, and thus generate savings. Consistent with our earlier comments, CMS should give every ACO the maximum opportunity to be successful in the Program. We recognize, however, that CMS may have concerns about potential abuse of waivers by ACO providers still operating in a FFS environment. We suggest that CMS consider whether waiver review protocols, consumer protections, and quality criteria could be built into the waiver application process in such a way that could allow for extension of these waivers to ACOs in both one-sided and two-sided risk Tracks.

CMS asked whether the waivers should be limited to only ACOs in their second contract and whether the waivers should be limited to those beneficiaries who are preliminarily or prospectively assigned to the ACO. Subject to the additional suggestions noted in this section, we see no reason to limit the waivers to ACOs only in their second contract. Furthermore, we support making all waivers available, on a consistent basis, to all beneficiaries for whom an ACO can request data. In addition, we recommend that ACOs be able to educate beneficiaries about the waivers.

SNF 3-Day Rule

Avoiding unnecessary hospitalizations is one of the primary goals of coordinated care and therefore of an ACO. As Medicare beneficiaries are financially barred from using the SNF without a hospitalization, ACOs are unable to prevent an unnecessary hospitalization in the event that the necessary care could be provided in a SNF. We see limited likelihood for abuse of this waiver by most ACOs, with the exception of SNF-based ACOs or a health system with SNFs. CMS should simply monitor at-risk entities for evidence of excessive utilization and waive the 3-day rule for all ACOs. Thus, we support the waiver of the SNF 3-day stay rule.


When setting a national policy, it is necessary to use national definitions of an originating site. However, granting the waiver and allowing the ACO to use their much more extensive knowledge of local resources aligns the service to the needs of the area. Furthermore, tele-health with originating site requirements can also generate savings because it allows for greater access to physicians. To date there has been limited adoption of tele-health services because of limited reimbursement opportunities. The MSSP program presents an opportunity for CMS to learn more about the potential value of tele-health, such as substituting these services for more difficult to obtain specialty visits. To protect against any abuses, CMS should monitor for ACOs that are outliers for these services and do not achieve savings. Thus, we support the waiver of the originating site policy under the tele-health benefit.

Homebound Requirement for Home Health Services

Home health services can be critical to chronic care management and support the waiver of the current homebound requirement. We recommend, however, a change to which beneficiaries are eligible. Those who would most benefit from the services are those beneficiaries who need the full range of services, but do not quite meet the definition of homebound.

Referrals to Post-Acute Care Providers

The ability to develop a care coordination relationship with post-acute care providers is very valuable. However, that relationship only has real value when the patients use those providers. ACOs – regardless of Track — should be able to not only provide information on the quality of care provided by post-acute care providers, but also recommend facilities with which the ACO has an established relationship. This should not stand in the way of beneficiaries choosing another facility if they prefer.

Reducing Barriers to Wellness and Care Coordination While not proposed by CMS, we ask CMS to consider giving ACOs the ability to offer certain financial incentives that reduce barriers to care and facilitate care coordination. (The private sector offers examples of successfully implementing such incentives.) As well documented in benefit design, people respond to even small increases and decreases in cost-sharing under their health coverage. CMS should consider how ACOs should be able to leverage this effect to provide better care coordination. An example of such services is the new Chronic Care Management code which requires ~$8 a month in co-insurance. This co- insurance may serve as a barrier to accessing care and prevent beneficiaries from using a service that would improve health outcomes and generate savings opportunities. Thus it would make financial sense for ACO participants to waive the co-insurance. Similarly with respect to encouraging beneficiaries to stay within the ACO when seeking care, an ACO may find it beneficial to waive co-insurance for primary care providers. We encourage CMS to carefully consider the possibility of this type of flexibility. We also ask the CMS articulate its reasoning for not granting such waivers, particularly as to whether the barrier to the waiver is a policy position of CMS or a legal barrier.

Securing OIG Feedback on Waivers

Nothing discourages the taking of risk more than uncertainty in policy. In terms of the willingness to take risk, the clarity of the language of the waivers only matters if it successfully reduces the anxiety of providers to utilize the waivers. The power of the waivers is proportional to the providers’ willingness to use them, and that willingness is based on the perception of the clarity of the language not the clarity as determined by long legal review. Providers will be more likely, ultimately, to accept downside risk if they understand clearly what is permitted under the waivers. To improve that clarity, we encourage CMS to work with OIG on a feedback process for ACOs that is simpler and timelier than the current OIG opinion process.


Honoring Patient Choice

One of the central tenets of MSSP is that beneficiary choice of providers is maintained. Beneficiary choice in provider relationships should also be honored if they choose to identify a primary care provider. As CMS discussed in the preamble of the proposed rule, there are many situations where, for a particular year, the plurality of primary care services may shift away from a beneficiary’s primary care provider. Simple and common examples include dealing with an acute illness or condition requiring specialized evaluation and management services, extended time away from primary residence, low health care utilizers where a single service plays a big role in determining plurality, and many other circumstances. Beneficiaries should be able to declare that despite the data from a single, peculiar year, “this physician, this nurse practitioner, this physician assistant is whom I have a special relationship with, this is who I want to coordinate my care.”

Just as all people using Medicare Part A and B are eligible to be in an ACO, they should all be eligible to make the choice of their primary care provider. Certainly honoring patient choice of their provider is a prerequisite for a truly patient-centered program. However, we acknowledge the sheer size of that base creates operational and communication challenges. We propose three congruent approaches for supporting beneficiary choice.

Step 1: Applicable to All Medicare Part A and Part B Beneficiaries

CMS should leverage the proposed data opt-out process. CMS should expand the description in Medicare and You handbook to provide this full range of options for beneficiaries:

  1. Choose your Primary Care Provider: If that provider is in an ACO, the beneficiary will then be assigned to that ACO, regardless of any potentially contradictory service methodology.
  2. Do nothing: Beneficiary is then assigned by the service methodology and data is made available to the ACO to which they are assigned (if they are assigned).

Further, we urge CMS to consider how to best address assignment of beneficiaries who opt- out of data sharing. Having a beneficiary opt-out of data sharing takes away many of the tools ACOs use to improve quality and coordinate patient care, thus reducing their likelihood for success. At the same time, including beneficiaries who opt-out of data will provide more comparable benefits across patients such as access to additional services under the payment waivers. Recognizing the need to balance these dynamics, we recommend CMS work with stakeholders to better understand beneficiaries’ concerns and providers’ uses of the data to further explore innovative solutions to improving their care.

Step 2: Applicable to Beneficiaries Who Are Preliminarily Assigned to an ACO

For beneficiaries who will be assigned to an ACO, CMS should provide an additional opportunity for the beneficiary to make an informed choice. CMS should mail a targeted letter to each beneficiary who will be preliminarily or prospectively assigned to an ACO no later than November 1 the year before the performance year. Unlike the handbook, which covers all ACO options, this letter is solely focused on making sure the beneficiary is matched to the right primary care provider to coordinate their care.

The letter should identify the primary care provider and/or ACO to whom they will be assigned and list other primary care physicians from whom they have received a primary care service. The letter should let them know that if they agree with the identified provider/ACO, they can simply do nothing and would be assigned using the service methodology. If they disagree and want either one of the other identified physicians/providers or to name another physician/provider to be their primary care provider, they should be able to call 1-800-Medicare and make the change on a form and return it to the ACO or CMS. Please note that this differs significantly from the Pioneer experiment in two respects. The beneficiary is presented with multiple options in an attempt to have them make an affirmative choice.

Step 3: Applicable to Beneficiaries Who See An ACO Professional During the Performance Year

At face-to-face office visits, ACO primary care providers should be allowed to inform beneficiaries about their ability to designate their provider as the ACO primary care provider responsible for coordinating their care. CMS should provide guidance on appropriate ways to engage in such activity. Beneficiaries would do so either in writing at the office visit or by calling 1-800-Medicare after the visit. Such an informed patient choice should be honored in all circumstances, including if the beneficiary had been prospectively assigned to an ACO.

Additional Considerations

The framework above does not constitute an additional administrative burden to ACOs. The ACO is only involved in one of the three pathways and it can choose whether or not to engage beneficiaries in this way.

In addition, we urge CMS to reconsider limiting beneficiary choice to ACOs participating in two- sided risk… We recommend that CMS allow beneficiary choice for all beneficiaries.

This approach should reduce churn in the performance year, as well as correct churn in the historical year. One cause of churn is that a beneficiary was incorrectly assigned to an ACO provider during the historical period, due to peculiarities of the historical year. Moving them out of the historically assigned ACO to another provider could be moving them back to their long-term primary care provider (i.e. good churn). CMS data presented in the proposed rule indicates this type of correction could be as high as 13 percent.

This framework provides excellent beneficiary protection. The only opportunity for ACOs to “recruit” beneficiaries is at face-to-face office visits. The visit confirms the relationship between the ACO provider and the beneficiary, which also limits the possibility of systematic targeting that is possible in a broader marketing effort. Further protection is provided by annual risk adjustment and vigilance regarding complaints and patterns suggestive of selection.

Finding the Right Providers

We support CMS proposals to narrow the provider types included in attribution. As we discussed in our section on two-sided risk, decreasing the difference between prospective and retrospective attribution should be the goal rather than choosing between the two. To that end, CMS should allow ACOs to designate which providers should be used for purposes of attribution. Such a designation would increase the likelihood that a beneficiary is attributed to a provider who will coordinate and be responsible for their care, making both prospective and attribution more accurate. Using the CMS enrollment data, it would not be difficult for CMS to monitor these designations for any signs of gaming.

The Role of NPs, PAs and CNS in Primary Care

We applaud CMS for its elegant solution to acknowledge the critical role nurse practitioners, physician assistants and certified nurse specialists play in primary care. We encourage CMS to finalize it changes to step 1 of the attribution process. The combination of these changes and ACO provider designation should go a long way to increasing the accuracy of attribution while preserving patient choice in providers.

Continued Improvement in Attribution

As discussed in the Track 3 section of this letter, continued improvements in attribution provide greater certainty to everyone in the process. CMS should explore these following options to see if any would have reduced churn and/or created better alignment with primary care providers:

  • Use plurality of events rather than costs
  • Increasing the time period of attribution to a 24 month period
  • Increasing the weight of annual wellness visits, transition of care management services and chronic care management services as indication of care coordination
  • Using the most recent data point as a tie-breaker

With the exception of chronic care management services, all of these could be subject to analysis of past ACO attribution and their effects be considered for the final rule.


Substance Abuse and Alcohol Treatment Claims

We encourage CMS to work with SAMHSA and other stakeholders, including Congress, to review the regulations governing the disclosure of substance abuse and alcohol treatment data in light of new technology and new payment models.

While those efforts are underway, we make two recommendations to the current processes employed by CMS. CMS should consider whether beneficiary privacy rights would be adequately protected if claims for other services that contain, as ancillary information, a reference to a diagnosis of alcohol or substance abuse, were scrubbed to remove such references and then provided to the ACO. Provided that such an approach adequately removed any implication of 42 CFR Part 2, it would improve an ACO’s ability to provide optimal care to that patient. (As CMS already deploys technology to identify the claim, removing an already identified data element is in the realm of operational feasibility.) Second, CMS should consider whether it would be beneficial to give beneficiaries the option to consent to allow claims that have been scrubbed of all reference to a diagnosis of alcohol or substance abuse to be shared with the ACO to which they have been assigned, as is currently done under the Pioneer program. We also recommend that CMS consider providing ACOs with aggregate reports that include complete aggregate-level data.

CMS Data

CMS should use open source methods or make open source their methods and codes for all data and calculations in the MSSP. This creates greater clarity for the ACOs and fosters research and policy efforts as well. Narrative description of the calculations is not sufficient, CMS should make available the code and artifacts.

CMS should experiment with carving-in additional care components (such as Part D costs) to support efforts to move to full accountability/global budget for organizations ready to do so.

CMS should improve the comparability of existing data sets and seek to provide additional data sets (MDS, CCW, Oasis). CMS should continue to improve its revision process for these data sets possibly considering outside certification of its process.

Global Concerns

Finally, we encourage CMS to work across payers to promote comparability of data sets and program design. Such efforts would greatly aid groups with smaller Medicare populations to increase their patient pool through contracts with commercial payers, thereby helping justify both the infrastructure investments and facilitate a transition to downside risk under Tracks two and three.

Please contact Tonya Wells at with any questions.


Lee Sacks
EVP Chief Medical Officer

Advocate Health Care

Francis Soistman
Executive Vice President & Head of Government Services

Farzad Mostashari
Founder & CEO
Aledade, Inc.

Peter Leibold
Chief Advocacy Officer

Emily Brower
Executive Director, Accountable Care Programs
Atrius Health

Dana Gelb Safran
SVP Performance Measurement & Improvement
Blue Cross Blue Shield Massachusetts

Kristen Miranda
VP Strategic Partnerships & Innovation
Blue Shield of California

Mark McClellan
Director, Health Care Innovation and Value Initiative
Brookings Institute

Tony Clapsis
VP and EA to Chairman, President, and CEO
Caesars Caesar’s Entertainment Corporation

Lynn Guillette
Director of Revenue
Dartmouth – Hitchcock

Elliot Fisher
The Dartmouth Institute for Health Policy and Clinical

Lloyd Dean
President & CEO
Dignity Health

Chris Dawe
Evolent Health

Steve Ondra
SVP and Enterprise Chief Medical Officer
Health Care Service Corporation – Illinois Blues

Richard Merkin
President and CEO
Heritage Provider Network, Inc.

Debra Ness
National Partnership for Women & Families

Jay Cohen
Senior Vice President/Executive Chairman

Kevin Schoeplein
President & CEO
OSF HealthCare System

David Lansky
President & CEO
Pacific Business Group on Health

Timothy Ferris
SVP, Population Health Management
Partners HealthCare

Jay Desai
Founder and CEO

Blair Childs
Senior Vice President
Premier, Inc.

Joel Gilbertson
Senior Vice President
Providence Health & Services

Steve Wiggins
Remedy Partners

Michael Slubowski
President & CEO
SCL Health

Gaurov Dayal
President, Health Care Delivery, Finance and Integration
SSM Health Care

Judy Rich
President & CEO
Tucson Medical Center Healthcare

Paul G. Neumann
EVP & General Counsel
Trinity Health 

1 We understand CMS’ stated policy position to not identify a fixed population for practices so that ACOs would make an investment in improving care for their total population. We agree that some investments in practice improvements can be made that will affect all patients without regard to scale, such as an open appointment system. But some direct investments in care management, such as hiring registered nurse care managers, are driven by the volume of patients to be managed. Given that there is an estimated 20 to 40% turnover in ACO-attributed populations, CMS was, in effect, asking ACOs to invest in care for all Medicare patients touched by their practices, even though the potential shared savings would only be available for 60-80% of the managed population. This policy created
additional uncertainty about the value of making significant investments in improving care.

2 Rebasing in their second contract will play a significant part in ACOs decision and is addressed in the Benchmarking
Section of this letter.