Download as PDF

April 15, 2022 

Paul N. Casale, MD, MPH
Chair, Physician-Focused Payment Model Technical Advisory Committee
Office of the Assistant Secretary for Planning and Evaluation
U.S. Department of Health and Human Services
200 Independence Avenue, S.W., Room 415F
Washington, DC 20201 

Re: Population-Based Total Cost of Care Models Request for Input

Dear Chairman Casale:

The National Association of ACOs (NAACOS) appreciates the opportunity to submit comments in response to the request for input (RFI) on key issues and options related to the development and implementation of population-based total cost of care (TCOC) models as published on the Physician-Focused Payment Model Technical Advisory Committee (PTAC) webpage in conjunction with the March public meeting on March 7–8, 2022. This RFI provides an important opportunity for stakeholders to provide committee members with information about current perspectives on the role that population-based TCOC models can play in optimizing health care delivery and value-based transformation in the context of alternative payment models (APMs) and physician-focused payment models (PFPMs). 

NAACOS is the largest association of accountable care organizations (ACOs) and Direct Contracting Entities (DCEs) representing more than 12 million beneficiary lives through hundreds of Medicare Shared Savings Program (MSSP), Global and Professional Direct Contracting Model (GPDC), and commercial ACOs. NAACOS is a member-led and member-owned nonprofit that works on behalf of ACOs and DCEs across the nation to improve the quality of Medicare delivery, population health, patient outcomes, and healthcare cost efficiency. 

NAACOS is committed to advancing the value-based care movement, and our members want to see an effective, coordinated, patient-centric healthcare system that focuses on keeping all individuals healthy. Strengthening the ACO model and other TCOC models provides an important opportunity to improve clinical quality and reduce health inequities while controlling rising healthcare costs. We are pleased to provide feedback on the following topics, which respond to the indicated groups of questions noted below.

Future population-based TCOC model design and implementation
Questions: 1. The Center for Medicare and Medicaid Innovation (CMMI)’s Strategy Refresh includes a goal that all Medicare beneficiaries with Parts A and B will be in a care relationship with accountability for quality and TCOC by 2030. What should these future population-based TCOC models look like?

  1. What type(s) of entity/entities or provider(s) should be accountable for TCOC in population-based TCOC models? Could the accountable entities look like current ACOs or Medicare Advantage (MA) plans? Could the accountable entities be delivery systems taking on risk, a combination of delivery organizations and payers, or fully integrated systems?
    • Does the ability to manage TCOC vary by certain factors (e.g., type of provider, specialty, condition)?
  2. What are some options for addressing model overlap and incorporating episode-based payments within population-based TCOC models?
    • How might these options vary by differing factors (e.g., ACO ownership type, condition, specialty, type of episode)?
    • What are potential issues related to nesting, carve-outs, and other potential approaches?
  3. What types of services should be included in calculating TCOC in the context of APMs, PFPMs, and population-based TCOC models? To what extent do definitions of TCOC differ across specialties, models, payers, and other factors?
    • Should there be a single definition of TCOC in future population-based TCOC models? Are there considerations regarding why the definition of TCOC should potentially be allowed to differ by certain factors (e.g., payer type)?

Response:

NAACOS strongly supports the Centers for Medicare and Medicaid Services (CMS)’s stated goal of having all Medicare beneficiaries with Parts A and B in a care relationship with accountability for quality and TCOC by 2030, and we look forward to working with the agency to realize this goal by achieving the strategic objectives outlined by CMMI. The MSSP is the most established value-based care program in Medicare, currently serving 11 million Medicare beneficiaries, which is nearly one-third of traditional Medicare patients. Medicare data show that the success of TCOC, population-health models such as Medicare ACOs far outpaces the performance of narrowly focused APMs and therefore should be considered as the best path forward for implementing healthcare payment reform. Medicare ACOs such as the MSSP, the Pioneer ACO Model, the Next Generation ACO Model (NextGen), GPDC, and the recently-announced ACO Realizing Equity, Access, and Community Health (REACH) Model are designed to incentivize all providers to work together to provide coordinated, whole-person care, functioning as a cohesive system. This has been described as akin to operating a farm versus an individual part, like a silo, on that farm. Historically, our health system has been siloed, with primary care providers, specialists, hospitals, and post-acute care all working discretely and often without the full-patient picture. 

In a population-based model, the focus is not solely about caring for individual patient needs at a given time, but about managing and improving the overall health of that patient and of a broader population. Therefore, full accountability should not be placed on individual providers, but spread across the team. It is most appropriate to have the ACO act as the accountable entity and to provide the ACO with the flexibility to work collaboratively with the ACO’s providers to determine the amount of risk that individual provider types take on. This incentivizes a team-based approach without holding providers accountable for factors outside of their control and provides additional support for managing health-related social needs (HRSNs) and social determinants of health (SDOH). The ACO model has been successful because it allows local healthcare providers to determine who to collaborate with and come together voluntarily with those who share common goals. This provides the flexibility to form partnerships with shared accountability to support innovation and collaboration without forcing it, which allows ACOs to meet differing needs of differing communities. 

In recent years, specialty-focused and episode-based bundled payment models have proliferated in the healthcare industry which has had negative consequences on TCOC models. Overlapping models create confusion for patients served by multiple models as well as the clinicians participating in such models, and these issues have become increasingly complex due to the overlap of multiple models. NAACOS has encouraged CMS to work collaboratively with stakeholders to establish transparent and consistent overlap policies that protect and support TCOC models and, specifically, to exclude ACO patients from these models unless a collaborative agreement between the bundler and the ACO is in place. Fair and appropriate model overlap policies should give deference to TCOC models such as ACOs, which have outperformed other models to date and should, therefore, be prioritized when incorporating episode-based payments within population-based TCOC models. 

As discussed in the PTAC March 2022 public meeting, there are many considerations to account for when defining and calculating TCOC in the context of a population health model. For example, payer type can impact what data are available to facilitate management of TCOC. Given the barriers created by the separation of Part D in Medicare, Medicare TCOC models should remain inclusive of only Parts A and B. If CMS were exploring adding accountability for costs outside of Parts A and B, it would be important that the agency do so very thoughtfully, offering an optional ACO test for those ready to take accountability beyond Parts A and B. Further, CMS should not move forward with such an approach until the systems within Medicare are equipped to provide the data necessary to successfully manage those additional costs. 

Definitions of TCOC may also vary based on the needs of the given population or other factors, which is why an accountable entity, such as an ACO, should be provided flexibilities to collaborate with providers in ways that meet their needs. For example, allowing ACO participation at the Tax ID Number-National Provider Identifier (TIN-NPI) level would ensure that all providers in the ACO are voluntarily participating and share the goals of the ACO and a commitment to value-based care. 

Payment structure and financial incentives needed to support population-based TCOC models
Questions: 6. Based on your experience, what payment strategies have been particularly effective for supporting efforts to improve quality and reduce TCOC (e.g., shifting risk downstream to providers)? Why have these strategies been effective? What have been some challenges and opportunities related to these approaches?

  1. What are the pros and cons of using payment methodologies that rely on a fee-for-service (FFS) architecture with upside and downside risk versus payment methodologies that involve global budgets or capitated payments?
  2. Based on your experience, what are different methodologies for developing benchmarks used to determine payment under population-based TCOC models? What are the pros and cons of these approaches? How can approaches for developing benchmarks be improved?
  3. Based on your experience, what are different methodologies for risk adjusting measures used to determine payment under population-based TCOC models? What are the pros and cons of these approaches from a beneficiary, physician, or program perspective? Are there any unintended consequences of applying risk adjustment methodologies?

Response:

When considering payment strategies most effective in supporting quality improvement and cost control, there are a variety of factors at play, including practice type, resource allocation, and provider experience with value-based payment initiatives and population health models. Imposing downside risk on practices that aren’t prepared will limit program participation and growth, as was shown by the decline in MSSP participation following implementation of the Pathways to Success Rule, which included significant changes to mandatory downside risk and other program elements. An appropriate balance of risk and reward is necessary to ensure broad and successful participation in a population-based TCOC model. 

It is important to distinguish between risk and value, as there has been some misconception that value in healthcare requires downside risk. Shifting downside risk to providers is not the main answer to improving quality and reducing costs, and mandating downside risk or predicating the availability of payment waivers or other participation incentives on the level of risk that an ACO has assumed is detrimental to the value transformation and ignores the fact that shared savings only or one-sided risk arrangements produce significant savings for the Medicare program. ACOs in both shared savings-only and risk-based models have shown reductions in spending per beneficiary relative to their benchmarks, and risk-based ACOs are not necessarily successful because of the assumed risk but because of other aspects of value transformation that have taken root. It can take years for the clinical and cultural changes necessary to succeed in value-based care to develop, and allowing providers ample time in one-sided arrangements allows those changes to progress. In many cases, one-sided risk arrangements mature to risk-bearing arrangements over time, which conforms with evidence showing that ACO performance improves over time, with length of participation associated with increases in both quality improvement and savings generated. 

Appropriate incentives are necessary to ensure success when shifting risk downstream to providers, which is why Congress established a 5 percent bonus payment for providers participating in Advanced APMs through the Medicare and CHIP Reauthorization Act (MACRA). These bonuses have been instrumental in incentivizing providers to participate in high-risk models, but due to implementation challenges, participation in such models has not yet grown to the levels Congress envisioned when enacting MACRA in 2015. Currently, the opportunity to qualify for the Advanced APM bonus is set to expire on December 31, 2022, leaving minimal incentive for providers to remain in or join Advanced APMs. ACOs have reported using the bonuses to invest in ACO initiatives such as care coordination or data analytics and to support the ACO’s move to a risk-based model. If the bonus is not extended, it will be important to consider alternative incentives to promote increased participation in high-risk APMs. 

Offering optional capitation payments within a population-based TCOC model can be an effective strategy in the value transformation. Implementing optional capitation payments helps to break the FFS “wheel” that providers are accustomed to and allows the flexibility to transform care delivery. CMS has already been granted the statutory authority to implement partial capitation within the MSSP, and NAACOS has encouraged the agency to develop a new full risk option for ACOs in the MSSP, which could include 100 percent shared savings and loss rates, participation at the TIN-NPI level, options for capitated payments, and more advanced waivers such as those tested under CMMI’s NextGen Model. Due to diversity among providers and varying levels of experience with non-FFS payment structures, some providers may find it easier to manage capitation than others and, therefore, capitation payments should be an optional component of model design. Ultimately, payment strategies within a population-based TCOC model should include appropriate flexibilities for providers to select risk and capitation options that meet their needs and recognize their ability to manage risk and administer capitation at that time. 

NAACOS has long advocated for the creation of fair and accurate financial benchmarks and risk adjustment policies for ACOs. These benchmarks should create realistic opportunities for ACOs to generate and keep shared savings when they successfully lower patients’ total cost of care. TCOC models have been successful in Medicare and should be widely encouraged because of their ability to both encourage care coordination among providers that leads to higher quality care, and generate savings for Medicare that prolong the fiscal sustainability of the program. Both are contingent upon sound financial benchmarking and risk adjustment policies, which are at the heart of fairly holding providers accountable for cost and quality. 

While Medicare TCOC models differ in their financial methodologies, all follow the same basic pattern. They are based upon per capita expenditures for Parts A and B services under traditional Medicare. Benchmarks include beneficiaries who would have been assigned to the ACO in each of three previous calendar years used in the model. They are trended forward based on national growth rates in Medicare spending. There is also some sort of regional adjustment so that efficient providers are rewarded and inefficient providers aren’t unfairly rewarded. Spending is also risk adjusted to account for an aging population and sicker groups of patients ACOs may care for. These calculations are done across four beneficiary types – aged, non-disabled, disabled, dually eligible for Medicaid, and end-stage renal disease patients. 

In the past, NAACOS has generally supported ACO benchmarking policies that more closely resemble those used in Medicare Advantage (MA). These are based more on spending for a particular county or region and are adjusted based on quality and risk scores. They are also administratively adjusted on an annual basis. If Medicare were to move to such policies for TCOC models, that could create more predictable benchmarks for ACOs. We encourage PTAC and CMS to further explore this approach and to make robust data publicly available for researchers to model the effects of such a revised approach. 

In the meantime, NAACOS calls on CMS to fix the existing ACO benchmarking and risk adjustment policies by correcting longstanding flaws that unfairly penalize ACOs and stymie the growth of the Medicare ACO program.  For example, as we detailed in our comments in response to the proposed 2022 Medicare Physician Fee Schedule, CMS should refine the current benchmarking and risk adjustment policies employed in MSSP to create fairer, more equitable financial methodologies for ACOs. For starters, CMS should remove ACO-assigned beneficiaries from the regional reference population. This is often referred to as the “rural glitch” and systematically penalizes an ACO when it reduces costs. Because of the rural glitch, when an ACO lowers the TCOC for its assigned population, it also reduces the average regional costs and diminishes the positive effect of the regional adjustment. This defeats the purpose of a benchmark that is based in part on regional expenditure data, which CMS has acknowledged is fair and necessary for a viable ACO program long-term. Specifically, to correct this CMS should remove ACO beneficiaries from calculation of the regional risk-adjusted per member per year (PMPY). Research conducted by the Institute for Accountable Care has found that 90 percent of MSSP ACOs would benefit to some degree by this correction.  

To help create fair policies that account for the sickness of ACO patients, CMS employs a number of policies to limit so-called “upcoding.” Most recently in GPDC and ACO REACH, CMS uses a Coding Intensity Factor, which limits risk score growth across the entire model by effectively normalizing risk scores for all patients in the model. In the MSSP, CMS caps risk scores at growing no more than 3 percent over a five-year period. In contrast to traditional Medicare’s accountable care programs which have multiple controls in place used to limit risk score increases, there are fewer such controls in MA. As a result, the Medicare Payment Advisory Commission reported in March 2021 that higher coding intensity resulted in MA risk scores that were more than 9 percent higher than scores for similar FFS beneficiaries. 

These policies create an inherently uneven playing field for providers operating in APMs within traditional Medicare. NAACOS urges CMS to align risk adjustment policies across all of its programs, including traditional Medicare and MA to avoid arbitrage and profit seeking based solely on risk scores. NAACOS is very concerned about the growing imbalance in risk adjustment policies between MA and various ACO programs that operate within traditional Medicare. CMS estimates that risk scores in MA will increase by an average of 3.5 percent in 2023. Risk adjustment policies in MSSP, by comparison, can only increase by 3 percent over a five-year agreement period. Sound program fundamentals around benchmarking and risk adjustment methodologies, as well as a reasonable glide path to taking on downside risk and an appropriate balance of risk and reward, are necessary to attract participants and ensure long-term financial sustainability in a population-based TCOC model. 

Care delivery strategies for population-based TCOC models
Questions: 3.Based on your experience, what are some approaches and best practices for integrating and improving coordination between primary care and specialty care providers within population-based TCOC models?

  • Has provider participation in population-based TCOC models affected innovation with respect to the integration of primary care and specialty care?
  • What are some incentives that can help to improve care coordination and provider accountability for TCOC?
  • Based on your experience, what kinds of care delivery strategies (e.g., patient-centered medical homes, telehealth, and care coordination; addressing social determinants of health, addressing behavioral health needs, and focusing on seriously ill patients) have been particularly effective for improving quality and reducing TCOC? Why have these strategies been effective? What have been some challenges and opportunities related to these approaches?
    • What are options for incorporating these strategies when developing care delivery models for future population-based TCOC models?
    • What are some best practices for improving the affordability of care for beneficiaries (e.g., copayments, prescription drugs) within population-based models?

Response:

Care innovations are often developed and implemented within the context of a specific population’s needs and, therefore, it is important that population-based TCOC models such as ACOs have the flexibility to tailor care innovations to the communities they serve. In order to enable ACOs to identify and target the greatest opportunities to improve gaps in health outcomes and tackle unnecessary spending, they need access to timely and accurate data. 

Providing timely data allows ACOs to deliver the kind of patient-centered, well-coordinated care necessary to improve health outcomes and reduce inequities, with emphasis on providing the right care in the right setting and preventing avoidable and costly complications or hospital readmissions. CMS provides some data, but it is delayed by weeks or months and is therefore not always actionable. The data available in the HIPAA (Health Insurance Portability and Accountability Act) Eligibility Transaction System (HETS) is very meaningful and should be provided in real time to ACOs for their beneficiaries. This would allow ACO providers to communicate with treating providers at the hospital and to work with the beneficiary upon their release to ensure optimal treatment, medication adherence, and follow up care. NAACOS has repeatedly urged CMS to develop a mechanism to share more robust health data, including that from HETS, with ACOs in real time to enhance care coordination, improve outcomes and reduce costs. As PTAC evaluates strategies to improve value-based care modes, we request that robust, timely, and actionable data be a priority. 

Another data challenge is that ACOs lack access to substance use disorder (SUD) data despite the fact that Section 3221 of the Coronavirus Aid, Relief, and Economic Security (CARES) Act helped to align 42 CFR Part 2 (Part 2) with HIPAA. ACOs still lack access to vital SUD-related data on their patients due to the fact that under current regulations, care coordination is not considered by CMS to fall under treatment, payment, and health care operations. Design of any population-based TCOC model should ensure that accountable entities have access to whole-person data on their patient populations to facilitate care coordination and other strategies to improve patient care and outcomes. 

Primary care versus specialty participation and levels of integration in ACOs can vary by different factors such as organizational structure or geography. Many ACOs include specialists as a strategy to reduce TCOC, and in 2018 nearly two-thirds of physicians participating in the MSSP were specialists. While the Affordable Care Act (ACA) provided CMS and the U.S. Health & Human Services Department (HHS) Office of Inspector General (OIG) broad statutory physician waivers of the physician self-referral law (the “Stark” law) and anti-kickback statute (AKS) for ACOs participating in MSSP, ACO participants continue to have questions about the application and scope of the agency-issued waivers. In order to encourage and enable collaboration and integration among primary care providers and specialists in ACOs, the agencies should address remaining uncertainty concerning whether, as well as the extent to which, an incentive program offered to a physician with respect to assigned MSSP patients may, without creating potential Stark Law issues, also be offered to the same physician for non-MSSP patients. This can create confusion for providers and limit uptake of such incentive programs. NAACOS has provided recommendations on modifications to Stark Law and AKS waivers that could provide participants with certainty and stability, and we also called for the expansion of waivers for non-risk bearing ACOs. These entities’ accountability for TCOC promotes self-regulation, and key safeguards in waivers ensure appropriate use. Clarifying remaining uncertainty regarding current Stark and AKS waivers and flexibilities in place would allow ACOs and other TCOC model participants to more easily create relationships to deliver coordinated, integrated care. 

A variety of care delivery strategies have been implemented by ACOs across the country, but specific strategies with the greatest efficacy vary based on the needs of the population being served and the capabilities of the individual ACO. ACOs should have the flexibility to develop and implement care delivery strategies that work within the organizations’ resources and the patient populations’ needs. An ACO in its evolution to value may be able to deploy different strategies in the beginning and later advance them, and readiness for different strategies depends on where the providers are on the continuum, so it is important that TCOC model design not be overly prescriptive of care delivery methods. 

For example, ACOs that are immature or underfunded often approach innovations within value-based care in stages, initially tackling obvious opportunities that require minimal overhead. This can be as simple as implementing workflow changes for staff to include additional care management or pharmacy staff or creating a manual process to test the effectiveness of a new care model before purchasing more costly services such as data platforms or artificial intelligence (AI) to facilitate provider data, referrals, or consults. ACOs that are more experienced and have proven success in their model are often better-equipped to pursue additional innovation and the investments required to create the best outcomes, which often come at a financial or staffing disruption cost. Some examples of innovation for a mature ACO could include adding technology that has bidirectional communication between participant electronic medical records (EMRs), network consulting platforms, or sharing services for patient care such as paramedics or nurses for home care. Regardless of an ACO’s maturity, provider engagement in care delivery innovation takes time and proof of concept.  Primary and specialty care providers have worked in silos for decades and changing cultural patterns take time, persistence, and flexibility to meet providers where they are today, while allowing them to implement value-based care in a way that fits their practice to create better patient outcomes. 

There are other “wraparound” strategies that facilitate innovative care delivery and improved outcomes, such health information technology (IT) infrastructure and data analytics that are important to consider. Appropriate infrastructure for closed-loop referrals within and outside of the ACO can support transitions of care and help to manage HRSNs and other factors impacting health outcomes that may be outside of the primary care provider’s control. Waivers and payment support should be made available where appropriate to enable accountable entities to implement such wraparound strategies. Smaller and/or under-resourced entities may require additional support in order to implement and sustain these strategies. For this reason, NAACOS has recommended providing upfront funding and other financial supports to enable the development of ACOs in underserved communities. This could be accomplished through a program analogous to the ACO Investment Model (AIM), which encouraged ACO development in rural and underserved areas by offering pre-payment of shared savings in both upfront and ongoing per beneficiary per month payments and is considered to be one of the most successful CMMI models to date. 

Population-based TCOC models such as ACOs are incentivized to deliver the right care at the right time and in the right setting for a given patient. Copays and cost sharing should never be an impediment to this. A patient, who cannot afford the copay for a primary care visit at the onset of symptoms, may end up with a much higher bill from an emergency department visit later on. ACOs should have the flexibility to waive cost sharing for certain beneficiaries in order to encourage patients not to delay needed care. While the Beneficiary Incentive Program in the MSSP was intended to help eliminate financial barriers to accessing care, the current program requires an ACO to offer any beneficiary incentive payment to all beneficiaries equally, regardless of financial need or particular condition. This requirement makes the program cost-prohibitive and has significantly limited uptake. Under GPDC and ACO REACH, CMMI is testing a beneficiary engagement incentive that allows the accountable entity (DCE/ACO) to provide cost sharing support for an identified subset of beneficiaries, types of Part B services, or both. NAACOS supports this type of flexibility, which allows accountable care entities to allocate resources in the most appropriate and effective manner for the needs of their patient populations. 

NAACOS has also recommended that such waivers be extended to ACOs in shared savings-only arrangements to allow those activities to develop and grow. All ACOs, regardless of risk level, require substantial startup and ongoing operational costs. Therefore, the use of waivers and other tools to enable success should be seen as a necessary precursor for long-term program participation and the path to assuming downside risk.

Health equity and clinical quality considerations for population-based TCOC models

Questions: 8. What specific issues should be considered when applying population-based TCOC models to diverse patient populations and care settings?

  1. Are there potential unintended consequences associated with implementing population-based models (for patients, primary care providers, specialty providers, and others)?
  2. Are there potential issues related to health equity regarding the implementation of population-based TCOC models?
  3. What are the options for increasing the participation of underrepresented and underserved populations in value-based models, including population-based TCOC models?
  4. Based on your experience, what are the best performance metrics for evaluating population-based TCOC models, and their impact on the quality and cost of care?

Response:

There are vast differences in care needs and priorities across communities and even across populations within those communities. Therefore, careful consideration of potential unintended consequences is essential when designing and implementing any population-based TCOC model. When models include mandatory downside risk, this creates barriers that can prevent providers treating greater proportions of historically-disadvantaged patients from being able to participate and will result in ACOs only forming in areas with financially secure health care providers, which is a detriment to health equity. An equitable TCOC model should include sound program fundamentals around benchmarking and risk adjustment methodologies to meet providers where they are at and avoid unintentionally penalizing those serving historically under-resourced communities. As mentioned previously, the rural glitch is a prime example of how benchmarking methodologies, when not designed carefully, can have negative consequences on certain communities. 

Additionally, model design and implementation should consider the upfront costs and resources required to stand up an ACO or other population-based TCOC model. In order to be successful in these models, accountable entities need access to sufficient health IT infrastructure and data analytics, as well as appropriate staff such as community health workers (CHWs), care navigators, and peer support specialists. Providers working in under-resourced communities with significant unmet need likely will not have access to the necessary capital to participate in such models without upfront funding and a reasonable glide path to taking on risk. NAACOS has provided several recommendations of strategies to both enhance existing ACOs’ ability to identify and address health inequities and to enable and encourage the formation of ACOs in historically under-resourced communities with more vulnerable populations. These recommendations include upfront funding, adjusted financial benchmarks, and development of a “chronic social determinant management” service (akin to chronic care management codes) that would allow ACOs to deliver and bill for certain supplemental benefits that address social determinants, improve health equity, and meet social needs. 

Careful consideration needs to be taken when designing policies that address health equity as to avoid unintentionally penalizing those treating underserved populations or creating winners and losers as a result of a given policy. Those working to design and implement equitable population-based TCOC models must recognize that there is considerable unmet need and significant health disparities in many populations due to a variety of compounding systemic factors that are largely outside the control of a healthcare provider. Some policies that may seem to support health equity may actually be detrimental. For example, stratifying quality measures by race/ethnicity can help to identify health disparities so that they can be targeted for improvements. However, some have suggested adjusting quality benchmarks for race and ethnicity. NAACOS strongly opposes this, as doing so would endorse and accept that for an underserved population it is acceptable to have lower quality or poor outcomes. Instead, policies that enable ACOs to identify and eliminate gaps in health equity should be pursued. Health equity initiatives should include rewards for reducing gaps in health equity but should not penalize ACOs for existing gaps or suggest that said gaps are acceptable or immutable. 

CMMI’s recently-announced ACO REACH Model exemplifies how a population-based TCOC model can be intentionally designed to promote health equity through various provisions, including a health equity benchmark adjustment to provide additional financial support to ACOs serving a disproportionate number of underserved beneficiaries and demographic data collection requirements to help with model monitoring and evaluation. While this model seems promising, NAACOS was disappointed to learn that the health equity benchmark adjustment will be budget neutral, and any upwards adjustments to support ACOs serving higher proportions of underserved beneficiaries will come at the expense of other ACOs in the model. This limits the amount of funding available and creates other challenges given the fact that beneficiaries are defined as “underserved” in relation to other beneficiaries in the model, not based on a standard definition. Given the significant unmet need experienced by historically-underserved populations, models designed to support health equity should include additional funding for these populations, which is long overdue. 

CMMI has also acknowledged the need for additional collaboration with external stakeholders, such as beneficiaries, community based-organizations (CBOs), and patient advocacy groups, in designing models that advance health equity. In order to increase the participation of underrepresented and underserved populations in value-based models, diverse stakeholder input is needed at every stage of design, implementation, and evaluation. Feedback from those working with these populations, including local health care providers and CBOs, will be necessary to understand the barriers and enablers to participation. 

In evaluating population-based TCOC models, the historical focus has been on a model’s ability to generate savings. However, recent remarks from leadership at CMS indicate that there will be an increased focus on quality improvement and health equity in future model evaluations. This is critical given the significant unmet need being experienced by underserved populations. In the short term, it may not be feasible to achieve cost savings while improving quality and reducing health inequities but achieving high quality, equitable outcomes will be necessary for the long-term success and sustainability of our health care system. When evaluating cost savings, comprehensive difference-in-difference (DID) analysis should be used in addition to comparing benchmarks to performance, as DID can provide a more sophisticated estimate of the model’s impact. It is also important to note that actuarial evaluations may not capture care delivery transformation and the effect it has on patients. Therefore, it is important that model evaluation leverage a multifaceted approach with both quantitative and qualitative data to ensure a full picture of the model’s impact on quality, equity, and cost is being assessed. 

Impacts on quality can be difficult to measure. To date, CMS has used a variety of quality measures and measurement approaches across APMs. In some TCOC models, there are few if any clinical quality measures, and instead the focus is on administrative claims measures that assess hospitalization and readmission rates. In other TCOC models, CMS has aligned quality measurement approaches with the FFS program, the Merit-Based Incentive Payment System (MIPS). This misaligned approach creates uncertainty for participants and does not allow CMS to appropriately evaluate model impact on quality. 

Further, there is little room for stakeholder input on APM quality measures. Most CMS programs must involve stakeholder input through the Measure Applications Partnership (MAP), which makes annual recommendations to CMS on Measures Under Consideration for various CMS programs. However, to date, no CMMI models have engaged in the MAP process, leaving stakeholders no way to provide input and no transparency regarding the process for measure selection, as well as no way to question the validity of a certain measure or provide other critical input. CMS should be focused on advancing quality evaluations for APMs, and TCOC models in particular, in an aligned and transparent manner that involves stakeholder input. In particular, CMS’s recent actions to move MSSP ACO quality assessments to the FFS MIPS methodology is a step in the wrong direction and has the potential to erode participation in that model. 

Specifically in the Final 2021 Medicare Physician Fee Schedule Rule, CMS created a new APM Performance Pathway (APP) within the MIPS program to evaluate APMs who are, despite their participation in an APM, subject to MIPS. At the same time, CMS also removed the previous MSSP quality assessment methodology to replace it with the new APP methods. Aligning ACO quality assessments with MIPS assessments is a step backward for value-based care. Instead, CMS should look for ways to align the quality approach for value-based payment models, which are the future of healthcare delivery. 

While CMS has taken the approach of removing all clinical quality measures from certain population health models, such as the Direct Contracting Model, NAACOS believes there continues to be utility in maintaining some high value process measures focused on prevention, such as cancer screenings and immunizations. Further, while CMS often cites that measure performance is topped out on these process measures, when stratified by race/ethnicity and other factors, there is indeed still room for improvement. NAACOS believes TCOC models like ACO models provide an opportunity to look across a population served to identify inequities in quality so that they may be addressed appropriately. We have provided a detailed list of recommendations and considerations when addressing health equity in quality measurement for TCOC models, and we recommend PTAC and CMS consider these approaches when moving forward. When these issues are not addressed, there may be unintended consequences. For example, in moving the MSSP to the APP quality assessment structure, CMS has inadvertently established a policy which will punish ACOs serving under-resourced communities as their quality scores will appear lower, and, as a consequence, they may not be eligible for shared savings or even owe losses to CMS.  

Finally, while hospitalization and readmission rates can be a good indicator of reducing costs and maintaining health for a beneficiary by avoiding a hospital stay, to solely focus on these measures is missing the broader picture of quality improvement efforts ACOs engage in. Additionally, there are certain quality measures that lend themselves to a TCOC model or program, such as measures focused on screening for SDOH. CMS should begin to think more strategically about how to evaluate quality for APMs, and TCOC models in particular, gathering stakeholder input throughout the process, such as using the established MAP process.  

Issues related to provider readiness, participation incentives, and administrative burden associated with population-based TCOC models
Questions: 4. What are some options for evaluating and increasing provider readiness to participate in population-based TCOC models?

  1. Are there differences in provider readiness by specialty or other factors?
  2. To what extent can provider participation in models with some upside and downside risk help to increase provider readiness to participate in population-based TCOC models? If so, what are some options for improving provider readiness to take on risk?
  3. What are some of the provider-level barriers to participating in population-based TCOC models (including barriers for specialists)?
  4. Are there opportunities to improve multi-payer alignment and increase multi-payer participation in population-based TCOC models? What are the most important model design components related to increasing multi-payer alignment (e.g., clinical tools, outcome measures, payment)? 

Response:

There are a variety of factors that may impact a given provider’s readiness to participate in population-based TCOC models such as practice size and sophistication, staffing issues, culture, financial backing, or serving rural or underserved communities. Providers need to share the values and goals of the accountable care entity in order to achieve the necessary level of buy-in to be successful. Without this, providers will not be motivated to change clinical practice and workflows to deliver high quality, high value care. Past experience with APMs or value-based care strategies outside of the traditional FFS system may impact providers’ interest in and readiness to participate in a population-based TCOC model. As previously mentioned, a reasonable glide path to taking on downside risk is necessary to ensure that different types of practices and providers are able to participate and succeed in population-based TCOC models. NAACOS has previously advocated for no fewer than four years before an ACO be required to take on financial risk. It is important to recognize that not all providers are in the same position on the path to value and model design should be flexible enough to meet providers where they are at and provide sufficient time to demonstrate positive results. 

Model design should also avoid overly complex administrative burdens. Certain program requirements are associated with considerable costs, time, and resources. While meeting one specific program requirement, like securing a repayment mechanism or ensuring adherence with compliance requirements, may seem reasonable, the totality of administrative burdens on providers in APMs is rarely considered. Taken together, meeting these requirements takes away from staff time and resources that would otherwise go to direct patient care or quality improvement initiatives. We recognize that not all administrative burdens can be eliminated, but we encourage PTAC and CMS to consider and limit these growing burdens as models are developed and updated. 

NAACOS has provided detailed recommendations on strategies to reduce administrative burdens and allow more time and resources to be dedicated to care transformation efforts, such as reforming beneficiary notification requirements, addressing model overlap, providing ACOs with actionable and timely data, and other improvements to program elements. NAACOS has also provided recommendations to CMS on strategies to encourage MSSP participation and bolster ACO program growth including changes to quality assessments, benchmarking and risk adjustment methodologies, and shared savings rates. 

In addition, finding ways to align certain methodologies and program design elements across payers would go a long way in reducing burdens for providers wishing to participate in multiple models across multiple payers. This would not necessarily require a new model, but rather is a recommendation for efforts to align TCOC model building blocks and methodologies such as quality measurement approaches, risk adjustment methodologies, and payment rule waivers.  Alignment across models would allow ACOs to reduce the significant administrative burden associated with participating in multiple models, each relying on their own unique methodologies and program design elements. We recommend that HHS play an active role in collaborating with provider stakeholders and payers outside of Medicare to generate consensus and support for an ideal set of standard TCOC model elements. 

Other considerations and next steps
Question: 14. Are there any other important questions that should be considered related to the development of population-based TCOC models and PFPMs? 

Response:

In developing population-based TCOC models, it is important to leverage lessons learned from past models in order to structure models in a way that is attractive enough for providers to voluntarily participate. This includes recognizing the burdens of startup and ongoing operational costs, balancing risk and reward to encourage provider buy-in and prevent attrition, addressing issues of model overlap, and providing ample data and technical assistance for accountable entities to succeed in population-based TCOC models. NAACOS encourages diverse stakeholder collaboration in model development with ample provider input throughout each stage of model design, implementation, and evaluation. This will increase transparency, predictability, and fairness which will foster a higher level of trust between providers and payers entering into these value-based agreements, thus leading to a successful transition to value-based care. 

Conclusion:

Thank you for the opportunity to provide comments on the PTAC RFI on key issues and options related to the development and implementation of population-based TCOC models. Should you have any questions about our comments, please contact Allison Brennan, SVP, Government Affairs, at abrennan@naacos.com. 

Sincerely

Clif Gaus, Sc.D.
President and CEO
NAACOS