Payment and Delivery System Reform in Medicare: A Primer on Medical Homes, Accountable Care Organizations, and Bundled Payments

The Affordable Care Act (ACA) established several initiatives to identify new payment approaches for health care that could lead to slower spending growth and improvements in the quality of care. Many of these new delivery system reforms are currently being implemented and tested in traditional Medicare. This Primer describes the framework and concepts of three broad alternative payment models—medical homes, ACOs, and bundled payments—and reviews their goals, financial incentives, size (number of participating providers and beneficiaries affected), and potential beneficiary implications. It also summarizes early results with respect to Medicare savings and quality.

Context for Delivery System Reform in Medicare

Delivery system reform in Medicare focuses on shifting a portion of traditional Medicare payments from fee-for-service (FFS) (which reimburses based on the number of services provided) to payment systems that incorporate some link to the “value” of care as determined by selected metrics, such as patient outcomes and Medicare spending. The Department of Health and Human Services has announced a goal to have 90 percent of traditional Medicare payments linked to quality or value by 2018. 1 The agency also reports that it has met its goal of having 30 percent of Medicare payments tied specifically to alternative payment models (such as Accountable Care Organizations (ACOs), bundled payments, and medical homes) by the end of 2016, and expects to reach 50 percent by the end of 2018. 2 Additionally, recent legislation to reform Medicare payments for physician services, the Medicare Access and CHIP Reauthorization Act (MACRA), includes bonus payments for physicians and other health professionals who participate in qualifying alternative payment models. 3

To establish a central place for designing, launching, and testing new payment models, the ACA created the Center for Medicare and Medicaid Innovation (CMMI), also referred to as the “Innovation Center,” housed within CMS. The ACA granted CMMI wide authority to design and test new models that aim to either lower spending without reducing the quality of care, or improve the quality of care without increasing spending. 4 The intent of designing and launching multiple new models is that the cream of all of these approaches will rise to the top, providing direction as to what works and what does not—so best practices can be quickly disseminated across the country. In fact, the ACA gives CMMI unprecedented authority to expand models across the U.S. when they are found to be successful.

Medical Homes, ACOs, and Bundled Payments

Medical homes, ACOs, and bundled payment models, combined, account for care provided to over 10 million Medicare beneficiaries 5 and are frequently cited by media, researchers, and policymakers as major examples of delivery system reforms implemented widely across the U.S. 6 These models are described briefly below:

• Medical homesare team-based models of patient care that rely heavily on the primary care practice (provider and care team) as the main and central source for delivery and coordination of the majority of health, illness, and wellness care for Medicare beneficiaries. Health insurers that support the medical home model typically provide monthly care management fees or other payments in addition to fee-for-service reimbursement for activities related to patient care and coordination.
• ACOsare groups of doctors, hospitals, and other health care providers who agree to share collective accountability for the quality and cost of care delivered to the patients attributed to their ACO. Payments to ACOs incorporate varying levels of financial incentives in the form of shared savings or losses (sometimes referred to as bonuses or penalties) for performance on identified spending and quality metrics.
• Bundled payments focus on discrete episodes of care by establishing an overall budget for services provided to a patient receiving a course of treatment for a given clinical condition over a defined period of time. In contrast to paying for each service individually, bundled payments provide incentives for providers to come in “under budget” for episodes of care.

Within each of these three delivery system reforms, CMS is managing the implementation and assessment of multiple payment models (Table 1).

• Multi-Payer Advanced Primary Care Practice (MAPCP)
• Comprehensive Primary Care (CPC)
• Federally-Qualified Health Center (FQHC) Advanced Primary Care Practice*
• Independence at Home (IAH)
• Comprehensive Primary Care Plus (CPC+)

• Medicare Shared Savings Program (MSSP) (Tracks 1-3)
  • Advance Payment ACO
  • ACO Investment Model
  • Bundled Payment for Care Improvement (BPCI) Models 1–4
    –
    Focus of bundles:
    • Model 1: Inpatient hospital services
    • Model 2: Inpatient hospital, physician, and post-acute services
    • Model 3: Post-acute services
    • Model 4: Inpatient hospital and physician services
    • Oncology Care Model (OCM)
    • Comprehensive Care for Joint Replacement (CJR)

    Ongoing Evaluation of these Models and Programs

    The ACA requires that each model managed by CMMI be evaluated for changes in both spending and quality of care provided. 7 In fact, the ACA incorporates assessments of both quality and cost in the criteria for granting the Secretary of Health and Human Services the authority to expand models nationally if they either reduce spending without reducing the quality of care or improve the quality of care without increasing spending. 8

    By working with many types of providers who are treating a variety of patients in various kinds of facilities, CMMI is evaluating cost and quality not just for the models overall, but also how these results may differ based on other variables, such as patient and provider characteristics. CMS is also examining the attractiveness of the models to providers based on model participation rates, feedback from participants regarding the process (challenges, barriers, success strategies, etc.), and other operational, quantitative, and qualitative measures.

    Preliminary results from these models are modest and mixed at this point, with some models showing more promise than others (Table 2). This might be expected given their early stages, the diverse number of approaches being implemented, and methodological challenges associated with calculating savings. Many of the models are meeting quality targets and showing improvements in quality of care, but to date, overall net savings to Medicare are relatively modest, with large variation between models as well as among the individual programs running within them. Given the difference in beneficiary counts, care models, and evaluation time periods for each model, it is difficult to compare the relative magnitude of savings among them.

    • Savings: In the MAPCP model, one state out of eight achieved savings over the first two years, net of care management fees. Although the CPC model almost broke even in the second year, neither it nor the FQHC/APCP model achieved savings net of care management fees in the first two years. In contrast, the IAH model, which has no care management fees and focuses on providing care to chronically ill patients in their own homes, achieved savings—over $25 million in year 1 and over $10 million in year 2.
    • Quality: Small differences in quality were found between CPC practices and comparison primary care practices, as well as between FQHCs participating in the APCP model compared with FQHCs not participating. In year 1 of the IAH model, all participating practices met quality goals on at least three of six quality measures; in year 2, all participating practices showed further improvement from the previous year on at least two of six quality measures.
    • Provider participation: Generally stable across models.
    • Savings: From the most recently available results, both MSSP and Pioneer ACOs had lower beneficiary expenditures than their benchmarks in 2015. For MSSPs, bonus payments exceeded savings, resulting in a total net cost of $216 million. For Pioneer ACOs, net savings after bonuses totaled under $1million. Nearly one third of the MSSP ACOs and half of the Pioneer ACOs received shared savings. The CMS Office of the Actuary certified the Pioneer ACO model as cost-saving.
    • Quality: In addition to quality improvements over time, CMS states that ACO performance on quality measures is comparably as good as or better than the traditional Medicare program overall. ACOs that participated for multiple years performed better on quality measures and were more likely to share in savings than those that withdrew.
    • Provider Participation: The number of MSSP ACOs has about doubled, growing from about 200 MSSP ACOs in the initial year to over 400 in 2016. Participation in the Pioneer ACO model declined to 9 from its initial count of 32; several of the withdrawing Pioneer ACOs became MSSP ACOs.
    • Savings: In the most prevalent model (Model 2), first year results show spending differences between BPCI and control group episodes in 2 out of 6 clinical categories. Namely, for orthopedic surgery (mostly hip and knee replacements), Medicare payments declined more for BPCI episodes than for their control group’s episodes; in contrast, for spinal surgery, spending increased more for the BPCI episodes relative to the control group. For the other clinical categories in Model 2, as well as most other categories in the other 3 BPCI models, results did not show significant differences in spending between BPCI and control group episodes.
    • Quality: Most clinical episode groups across the 4 BPCI models did not differ on quality outcomes relative to comparison episodes. When differences were detected, they varied both between BPCI models and within the models by several factors, including: clinical episode groups (e.g., orthopedic, cardiovascular, etc.), whether or not beneficiaries had a surgical procedure, and whether or not beneficiaries received post-acute care.
    • Provider Participation: 2 of the 4 models grew steadily in provider participation during the first year, but the other two 2 models experienced multiple withdrawals among their participants.

    Implications for Medicare Beneficiaries

    Questions about how beneficiaries are involved in these new payment approaches and the implications these models have on beneficiary care—particularly for those with multiple chronic conditions—are key issues, yet at this stage, the answers are not well understood. For the most part, Medicare beneficiaries in these models do not experience any changes in their Medicare benefits, as they are not required to see certain providers and do not experience differences in their levels of cost-sharing. Accordingly, even though beneficiaries are notified of their providers’ participation in these models, it is not clear that beneficiaries are aware of their “attribution” to them, nor the inherent implications of the model’s payment incentives for less costly care. Beneficiaries whose primary care provider is transforming their practice into a medical home may notice differences in the delivery of their care—for example, greater reliance on clinical teams and electronic health records—but for the ACO and bundled payment models, beneficiaries would not necessarily notice changes in care practices.

    On the one hand, having Medicare attribute beneficiaries to these new delivery models is least disruptive to beneficiaries and places greater responsibility on the providers to engage with their patients and deliver better quality care. It is also less susceptible to selection issues, whereby beneficiaries with higher health costs may be somehow discouraged from enrolling in an ACO or medical home. On the other hand, some analysts propose that beneficiaries could play a greater role in lowering health care costs and improving quality if they took a more active role in selecting a medical home or ACO and encountered incentives to see providers affiliated with or recommended by their ACO or medical home, or if the providers did more to actively engage them. (CMS is exploring several of these approaches via new ACO models that started in 2016.)

    Issues about directing patients to certain providers also play a role in bundled payment approaches. For example, when beneficiary and provider incentives are aligned, patients leaving the hospital may seek or receive recommendations for post-acute providers and settings that deliver high-quality care at relatively lower overall costs. However, other considerations, such as patient convenience and the absence of family caregiver support are factors that may play a far greater role in beneficiary preferences than provider interests. These issues highlight the need to monitor care provided to the vulnerable patients in these models of care and to institute patient protections if quality and access problems are found.

    Looking Ahead

    In traditional Medicare, delivery system reforms and alternative payment models are being implemented and tested to determine how they might encourage better care coordination for patients across settings—particularly for those patients with chronic conditions and high health care needs—while simultaneously reducing provider inefficiencies and potentially lowering costs (or cost growth) for Medicare. By design, these goals are intrinsic to these models, and stakeholders (including CMS, other payers, providers, patients/ consumers, and the Congress) naturally want to see real-world results from their investments as soon as possible. This pressure presents a clear tension between the desire for immediate results versus the time it takes to test fully-implemented interventions.

    CMS’s commitment to “rapid cycle” improvements in the demonstrations means that CMS is able to refine the payment models as they are being implemented in light of early results and unforeseen circumstances. 9 While potentially advantageous, tracking and monitoring the changes will be difficult, and designing evaluations that can hold constant all other factors, including site-specific variables, to determine definitively what works and what does not, presents methodological challenges. 10 Additionally, the evaluation of medical homes, ACOs, and bundled payments are conducted separately, which may preclude assessing how they might work together in a coherent way.

    CMMI has also launched a number of new models in 2016, several of which are designed to address issues raised by stakeholders with respect to the initial model offerings. For example, new ACO and bundled payment models, described later in this Primer, will provide advance payments to providers to address their concerns about start-up resources needed for transforming their patient care and business models. CMMI is also offering models that aim to inform and engage beneficiaries about the goals of the new payment approaches. Consumer involvement has not, as yet, been a major component of delivery system reform, particularly outside of medical home models. It will be helpful in future models to see how beneficiaries’ understanding of the models affects provider approaches, changes in service utilization, and patient outcomes. With increased beneficiary involvement also comes the need to incorporate appropriate measures for beneficiaries who have cognitive impairments.

    Introducing delivery system reform in Medicare is an extremely ambitious effort and one that comes with a number of challenges. Health care is changing rapidly, and while it is difficult to guarantee a controlled experiment in which only the changes of interest (typically targeted payment incentives) are allowed to vary, understanding the ability of providers to lower costs while maintaining or improving quality may help illuminate the ingredients of success (or failure). Ultimately, a key consideration for CMS and Congress to track is how Medicare patients fare in these delivery system reform models, especially those with the greatest health care needs. This issue, along with overall spending and quality, will help policymakers identify which models to pursue, and what it will take to disseminate the most successful payment and delivery models more broadly.