The Case For ACOs: Why Payment Reform Remains Necessary

Medicare population-based payment models, broadly known as accountable care organization (ACO) models, of which the Global and Professional Direct Contracting Model (GPDC) is an example, were launched out of recognition of two largely inherent weaknesses of fee-for-service: FFS does not promote efficiency, nor does it promote equity. The Centers for Medicare and Medicaid Services (CMS) and the Center for Medicare and Medicaid Innovation (CMMI) at CMS are in the midst of redesigning ACO models and developing new ones to maintain a portfolio of models as other, similar models are phased out. The evolution of population-based payment models, particularly the launch of GPDC, has spawned recent controversy, with some calling for a halt to ACO programs. For this reason, it is worth stepping back and reviewing the original motivation for these models and their merits (and challenges). It is important to recognize that any policy option must be viewed relative to an alternative, in this case, the traditional Medicare fee for service (FFS) payment system.

ACOs Are More Likely Than FFS To Promote Efficiency

FFS Payment Does Not Encourage Efficient Care Delivery

It is widely acknowledged that there is a lot of waste and provision of low-value care in the FFS system, which can harm patients clinically and financially. This is apparent from the decades of research on geographic variation in utilization and in initiatives such as the American Board of Internal Medicine Foundation’s Choosing Wisely program, which identifies specific low-value medical practices. While FFS is not the cause of all low-value care and moving away from FFS would not eliminate all low-value care, at its core FFS does not create incentives for efficiency and in fact often creates incentives for overuse. For example, elimination of unnecessary pre-operative testing (a practice identified as low value by Choosing Wisely), is not in the financial interests of providers. Similarly, prescribing a lower-cost treatment or avoiding an unnecessary procedure is discouraged by FFS, which rewards the volume of care provided rather than its value. In a world in which we must become more efficient, having a payment system that discourages efficiency seems unwise.

This problem is likely to get worse as new modes of care delivery, like telemedicine, grow. The wide array of care that digital health can provide offers tremendous value to patients but is ill-suited to FFS. Defining the unit of service is complex and ensuring program integrity is difficult. We must find ways to ensure access to these services, but it would be difficult to add new codes (and associated rules for how they are used), and to set payment rates that are high enough to support use but not so high as to encourage overuse.

Because of challenges in managing utilization, cost containment in FFS often involves lowering prices. There is certainly room to lower some prices in the Medicare program (including Medicare Advantage (MA) benchmarks), and reforms in some areas such as prescription drugs can save money while still supporting innovation. Yet, relative to the challenges faced, savings opportunities from fee reductions are modest. In fact, Medicare fees are already set under current law to rise at very low rates. For example, fees for professional services (if clinicians are not in alternative payment models) are scheduled to rise at 0.25 percent in nominal terms.

Overall, aggregation of CMS projections for spending growth in Medicare Parts A and B suggests that Medicare fees will rise at a rate approximately 0.7 percent below the rate of inflation between 2021 and 2030, but Medicare utilization (i.e., volume and intensity, including new drugs) will grow at about 3.8 percent per year (adjusted for demographics). If the American health care system is to be fiscally sustainable, we must design systems to hold the rate of increase in utilization below current, largely FFS-based, projections.

ACOs Can Promote Efficiency Without Undercutting Quality

ACO models were born, in part, out of recognition in the ACA that FFS did not encourage efficient delivery of care. By providing bonuses for lower total cost of care (and sometimes penalties for higher total cost of care), ACOs promote reduction of waste. Moreover, a population-based payment system affords providers greater flexibility in care delivery by removing incentives that favor services which may not be best suited for patients.

The strength of ACO incentives to reduce unneeded care depends on the design of the models and varies by organizational structure. Incentives are stronger for physician groups than for hospital-based health systems. Current models, by reducing benchmarks if ACOs are successful, dampen these incentives. Nevertheless, considerable evidence indicates that ACOs reduce unnecessary utilization (and that physician groups are more successful than large systems). For example, there is evidence that ACOs have reduced the use of low-value care and generated savings in use of post-acute care, an area known for large practice pattern variation. And because ACOs are disincented to overuse care, they can provide digital health services with less of a need for regulatory limits to reduce the risk of overuse.

Concerns about ACOs encouraging stinting on care have not been borne out so far. ACO savings appear to have been generated without decrements to quality, access, or limits on freedom of beneficiaries to choose providers. Though quality measures are imperfect, performance on care processes have not systematically fallen, and early evidence suggests patient experiences have improved, particularly among high-risk patients.

That said, spending reductions achieved by ACOs have been modest and may not translate into savings for Medicare. For example, if benchmarks are set too high or not adjusted for coding effects, ACOs will prosper, not Medicare. Moreover, if programs are voluntary or if ACOs can adjust provider lists, non-random participation can result in higher Medicare spending even if ACOs succeed in reducing utilization. Similarly, if risk is one-sided, savings from lower use that arise by chance will lead to bonuses without offsetting penalties.

All of these issues can be mitigated with careful design of program parameters and in fact, despite all of these issues, population-based initiatives such as the Medicare Shared Savings Program (MSSP) have resulted in savings to Medicare. Recent incarnations have been less successful because of selective program participation. Clearly, more work must be done to improve model parameters to develop incentives and avoid gaming, and to streamline and harmonize available models. Nevertheless, the evidence of behavior change suggests promise.

ACOS Are More Likely Than FFS To Promote Equity

FFS Does Not Support Equity

Disparities in health care are well known. As with low-value care, FFS is not the sole cause of disparities and moving away from FFS will not eliminate them. That said, FFS is not well-suited to address disparities because it directs resources away from populations that have limited access to care. Certainly, we could do better under FFS, but it lends itself to a patchwork, as opposed to a population-based, approach that will diminish the likelihood of success.

ACOs Can Promote Equity

The population perspective of ACOs is better suited to reduce disparities. Specifically, in contrast to FFS, population-based payments can be more easily adjusted to ensure adequate resources are devoted to populations who are otherwise underserved. Whereas FFS allocates payment according to the provision of services, a population-based model can set payment according to population needs. Setting payments for underserved groups above current FFS spending levels also strengthens incentives for providers to attract those groups with enhanced care or additional services. In fact, evidence from the commercial sector demonstrates that ACOs can reduce disparities.

Moreover, health equity will require a more efficient health care system. To the extent payment reform improves affordability, it will help promote equity. Much work needs to be done to ensure that resources are passed through to patients in need in meaningful ways, but the ACO framework can greatly facilitate equity-oriented initiatives by directing more resources for populations with unmet needs and organizing the use of those resources under a less fragmented governance structure.

Misperceptions And Challenges Related To Population-Based Payment Models

Moving forward, policy must address misperception and challenges faced by the population-based models. Specifically, several points must be emphasized:

ACOs, Including The GPDC Model, Do Not Limit Patient Choice

ACOs do not alter the fundamental benefits or entitlements associated with the Medicare program. ACOs, including the GPDC model, do not have the ability to impose network requirements or limit benefits (they can offer added services). Beneficiaries can seek care from any Medicare provider, and ACOs, including direct contracting entities in the GPDC model, cannot deny claims or impose direct utilization management rules such as prior authorization. This contrasts with Medicare Advantage plans or the indefinitely delayed (and very different) Geographic Direct Contracting Model.

ACOs, Including The GPDC Model, Do Not Represent A Privatization Of Medicare And Are Not Analogous To Medicare Advantage

ACOs were conceptualized as being provider centric. This remains true. Because MA plans enroll the patient and control the network design, providers are compelled to come to agreement with MA plans or possibly lose patients. In contrast, ACOs, including direct contracting entities, cannot restrict beneficiary access to a provider network. If a provider severs a relationship with any non-provider-based ACO or with a direct contracting entity, the patient stays with the provider and the ACO/ direct contracting entity loses the beneficiary.

In some ACO models such as the direct contracting models, the money may flow from CMS to the direct contracting entity and then to providers. This may seem analogous to MA, but it is better understood as allowing cash flows conducive to transmitting incentives to partnering providers, not restricting patient choice. Compared to MA, the control resides more directly with the providers and, as a result, the direct contracting entity can be viewed as a support entity for providers. A preferred provider may assign its cash flow rights to the direct contracting entity, but only under terms agreed to by the provider. Unlike MA, the direct contracting entities’ primary leverage is if it provides value to the provider, as opposed to threats of excluding the provider from its patients’ network.

More broadly, some have criticized the GPDC model because it involves Medicare contracting with for-profit entities. As with other programs in Medicare, beneficiary protections and guardrails to protect the Medicare program from unintended spillover effects must be in place. Yet, it is worth noting that non-profit entities may engage in deleterious practices. For example, Sutter Health Care, subject of a well-known legal case, is a not-for profit entity. Similarly, private equity acquisitions and surprise billing have unfolded in the context of FFS. Moreover, for-profit entities can add value—as David Blumenthal notes, for example, they may play a role in saving primary care, particularly if given the right incentives.

Any payment system is susceptible to exploitation by bad actors. Focusing on behavior is more important than focusing on ownership type.

Coding Effects Can Be Managed In ACOs

Risk adjustment is a central component of population-based payment models. Coding effects in the MA program have attracted a lot of attention because more aggressive coding benefits MA plans. Conceptually the same could be true in ACOs.

However, this is not inevitable. Risk-score normalization and risk growth caps in the ACO program already mitigate much of the fiscal consequences of more aggressive coding and continued attention is important. In fact, the risk-score growth constraints applied in the GPDC Model, which include a model-wide coding intensity factor and participant-specific risk score growth cap, are intended to be more restrictive than those applied in either MSSP or MA and could inform next steps in addressing coding intensity in MA.

ACOs’ Impacts On Provider Consolidation, Should They Arise, Must Be Managed.

Consolidation of providers in the health care system has led to high prices in the commercial market. Because some scale is needed to bear total cost-of-care risk and to invest in population health programs, there is concern that ACOs may exacerbate consolidation via ownership or joint negotiation. Yet this incentive may be offset by the fact that that physician-group ACOs have greater opportunities for success than larger hospital-based ACOs, which may discourage some consolidation. Evidence on consolidation and the impact of value-based care is mixed, though there is some evidence that ACOs may facilitate joint price negotiations among separately owned entities.

Of course, there are many reasons for consolidation (e.g., to cover the fixed costs of information technology and quality measurement burdens and to counter the market power of insurers). Though the impact of ACOs on consolidation remains unclear, antitrust enforcement and perhaps other regulatory interventions will be needed to address consolidation that will occur regardless of the future path of ACOs. We should not hold payment reform in Medicare hostage to consolidation concerns that will need to be addressed anyway.

ACO Parameters, Including For The GPDC Model, Must Be Carefully Designed And Experimentation Will Be Needed

Despite the conceptual appeal of population-based payment models and early positive results, ACOs are far from achieving their potential. There are challenges including attribution, risk adjustment, symmetry and degree of risk sharing, and benchmark setting. Program designers must balance incentives to save, incentives to participate, incentives to promote equity, and incentives to create program savings. This includes large enough bonuses to overcome any ACO operating costs (though the geographic variation literature demonstrates that lower utilization need not be associated with higher administrative costs). Additionally, program designers must integrate population-based models with episode-based models.

There are undoubtably many opinions about how to accomplish these goals. This is why CMMI was established, to experiment with different design choices. A rigorous debate about program features is valuable. But that debate, or disagreement with specific design choices, should not motivate opposition to population-based models overall. Moreover, because CMMI models are temporary, stakeholders must expect experimentation with models they might not prefer.

Summing Up

The American health care system is expensive, characterized by meaningful waste and disparities. The dominant FFS system is an impediment to addressing these issues. Moreover, unless more money is infused into the system, it is likely that fees for professionals and facilities will grow at rates below inflation and input costs.

Population-based payment models offer providers the incentive to reduce low-value care and allow them to benefit as that low value care is removed from the system (or not generated in the first place). The models, if successful, will reduce patients’ out-of-pocket costs and promote equity and population health. Of course, the challenges are not small, but the rewards will be significant if we can design models that come closer to their conceptual potential. However, it will take considerable, continued effort.

Authors’ Note

Dr. Chernew has research grants from Blue Cross Blue Shield Association, Health Care Service Corporation, Ballad Health, Signify Health LLC; received personal fees from Blue Cross Blue Shield of Florida, Humana, America’s Health Insurance Plans; equity in Archway Health and Waymark Inc.; serves on advisory boards for National Institute for Health Care Management, Blue Cross Blue Shield Association; and serves as the current Chair of Medicare Payment Advisory Commission.

Dr. McWilliams reports serving as a senior advisor to CMMI and providing consulting services to RTI International and Blue Cross Blue Shield of North Carolina. He also reports serving as an unpaid member of the board of directors for the Institute for Accountable Care.

The content of this article solely reflects the views of the authors and not necessarily those of any of these organizations.