Evaluation Of Medicare Alternative Payment Models: What The Data Show

The COVID-19 pandemic has impacted every aspect of life in the US and challenged the capacity of the health care system like no other event in modern times. While policy makers struggle with how to deploy and rebuild the public health infrastructure and the biomedical and pharma industries race to a vaccine and therapeutic cures, it is important to evaluate the best model for delivering care in the coming decades. Physicians, hospitals, and other providers work in a myriad of organizational structures ranging from single-person medical practices to massive multistate integrated delivery systems. Providers are paid through a range of models, from piecework fee-for-service to full capitation for packages of benefits. In between, there are myriad financial models incentivizing providers to deliver high-quality care at lower costs. 

As policymakers take stock of the new line-up in Washington, they will consider how best to implement health care payment reform. The evidence presented below suggests that accountable care organizations (ACOs) are the best path forward. We think the data in Medicare are clear: The success of total cost of care, population-health models such as Medicare ACOs far outpaces the performance of narrowly focused alternative payment models (APMs). As leaders of the National Association of ACOs, we believe policy makers need to recognize this disparity in Medicare and stop wasting time and energy trying to develop and fine-tune other medical home and episodic-based payment models.

After a decade, we have enough knowledge to know we should focus on ACOs in our delivery system reform efforts, although we recognize that there may be other models worth exploring if they don’t interfere with total cost of care models. Furthermore, we have learned during this pandemic that total reliance on a fee-for-service payment model is dangerous. As providers learn to appreciate the value of capitated payments, ACOs provide a natural bridge from fee-for-service to capitated payments.

While Medicare Advantage has grown to serve almost 36 percent of Medicare beneficiaries, not everyone sees the program as the best national model. The Medicare Payment Advisory Commission (MedPAC) estimates the health system spends about $1,000 per patient per year more on Medicare Advantage enrollees than on traditional Medicare enrollees. Conversely, they estimate Medicare ACOs are reducing costs by 1 percent to 2 percent compared to traditional Medicare. Many desire arrangements that would transform the remaining 64 percent of Medicare to a direct contracting arrangement, rewarding providers that lowered costs, improved access, provided coordinated patient-focused care, and improved quality and outcomes.

We now have eight years of experience with ACO and Center for Medicare and Medicaid Innovation (CMMI) models. In this piece, we explore the results of these past eight years of Medicare experimentation and the real-time experiences of health care delivery amid the pandemic.

Evidence For Population Health Model

Population-focused, total cost of care models, such as ACOs, incentivize all providers to work together to care for the whole patient and provide quality care throughout the continuum to address patients’ social needs, manage comorbidities, and coordinate medications. Their strength is akin to operating a farm versus an individual part, like a silo, on that farm. Health care for too long has been siloed, with specialists not working with primary care or hospitals not working with postacute care. While other authors have recently pointed out the need for condition-based APMs to offer better financial incentives to specialists, almost two-thirds of the nearly 500,000 clinicians practicing in Medicare ACOs are specialists.

The evidence indicates that ACOs continue to realize value-based care’s goal of providing better care at lower costs. Data from MedPAC, researchers at Harvard University, and the analytic firm Dobson DaVanzo and Associates show that ACOs are lowering Medicare spending annually by 1 percent to 2 percent. Knowing Medicare Parts A and B cost $636 billion in 2018, a 2 percent reduction in spending would save nearly $200 billion when compounded over a decade, assuming Medicare spending would grow at 4.5 percent per year without ACOs.

Importantly, based on the authors’ analysis of Centers for Medicare and Medicaid Services Public Use Files, ACOs also hit an average quality score of more than 94 percent in 2019, the latest year for which Medicare data are available. Also, the Office of Inspector General found that ACOs outperformed fee-for-service providers on most quality measures and improved quality over time in the program. Finally, stories from the experiences of ACO care during the pandemic illustrate the advantages of being responsible for the whole patient and using state-of-the-art data and care coordination best practices.

Medicare ACO performance has improved each year since it started in 2012 (exhibit 1). The Shared Savings Program lowered Medicare spending by $2.6 billion in 2019 alone—$1.3 billion after accounting for shared savings bonuses and shared loss payments. Total gross savings have totaled more than $7 billion.

Exhibit 1: Gross federal savings from the Medicare Shared Savings Program

Source: Authors’ analysis of Centers for Medicare and Medicaid Services data, 2020.

The Centers for Medicare and Medicaid Services (CMS) has acknowledged the work of the Medicare Shared Savings Program (MSSP) in the impact analysis of the proposed “Pathways to Success” rule in August 2018. Using claims data, the Medicare Actuary looked at spending in ACO markets versus non-ACO markets. When ACOs lower spending across the fee-for-service system, this also lowers payments to Medicare Advantage plans since those payments are based, in part, on fee-for-service spending. The agency estimates the overall impact of ACOs, including “spillover effects” on Medicare spending outside of the ACO program, lowered spending by $1.8–$4.2 billion in 2016 alone.

There are other benefits as well. ACOs offer a way for independent physician practices to receive the benefits of operating in a larger organization without having to merge with a hospital or health system, as Seema Verma, administrator of CMS, noted in a January Health Affairs Blog post. Finally, for the patient, there is one organization and one set of providers that care for all their health problems and integrate that care into a seamless delivery system as opposed to myriad of disconnected silos for each clinical condition.

Other Narrow-Focused APMs Have Proven Lackluster

As part of an effort to test new payment models, CMMI continues to roll out new APMs for various conditions or provider types. While the goals of testing new payment models and concepts have value, they compete with and confuse the care provided by ACOs. CMS released more than a dozen new payment models in 2019. Evaluation of the early models is limited, but data show other APMs have not proven as successful as ACOs (exhibit 2).

Exhibit 2: ACO performance versus other APMs




Size of Model by Annual Spending

Demonstrated Savings

Shared Savings Program



11.2 million

$94 billion

Significant Savings- (1) In 2019, the Shared Savings Program generated $1.2 billion in total net savings across 541 ACOs. (2) Spending 1-2% lower than comparison practices

Next Generation ACO Model




$5.1 billion


Significant Savings– “Spending was reduced by approximately $100 million (a 1.7% decline), or $62 million after adjusting for shared savings/loss payments (a 1.1% net savings).”

Comprehensive Primary Care Plus (CPC+)



2 million Medicare

3.3 million non-Medicare


No number found


No Proven Savings –“CPC+ slightly increased CMS’ expenditures for these beneficiaries by 2 to 3 percent, or by over $1.1 billion across the two tracks, when including the enhanced payments CMS made to practices.”

Comprehensive Primary Care





No number found


No Proven Savings– “In other words, although CPC did reduce Medicare Part A and B expenditures slightly relative to expenditures in comparison practices, it is highly unlikely that these Medicare savings generated by CPC were enough to cover the CPC care management fees Medicare paid.”

Bundled Payment for Care Improvement


797,000 episodes

No number found


No Proven Savings– “Despite these encouraging results, Medicare experienced net losses under BPCI after taking into account reconciliation payments to participants.”

Comprehensive Care for Joint Replacement




101,000 episodes

$92 million


No Proven Savings– “After accounting for reconciliation payments earned by participants, the CJR model likely resulted in net savings to the Medicare program of $17.4M, although we cannot conclude this with statistical certainty.”

Comprehensive ESRD Care Model


72,000 beneficiaries

$537 million

No Proven Savings– “Medicare experienced aggregate net losses of $46 million after taking into account shared savings payments made to ESCOs.”

Oncology Care Model


380,000 episodes

No number found

No Proven Savings– “Episode payments for high-risk cancers declined, but increases in episode payments for low-risk cancers offset these impacts, leading to a non-significant overall estimate for PP1-3. When model payments from the first two PPs were included (MEOS and PBP), OCM resulted in net losses for Medicare.”

Independence at Home Demonstration


10,000 beneficiaries

$399 million

No Proven Savings– “Over its first five years, there is little evidence that the IAH payment incentive reduced Medicare spending or improved quality of care for beneficiaries with chronic illness and functional limitations.”

Source: Authors’ analysis of the Center for Medicare and Medicaid Innovation formal evaluation of models.

If CMS’s goals are to lower cost and improve quality, data do not support wider use of bundled payments or medical homes. Initial results from CMS’s Bundled Payments for Care Improvement (BPCI) found Medicare spending increased after payments to participants were reconciled. Those findings remained true through the third year of the program, as confirmed by the program’s formal evaluation. More recently, Comprehensive Primary Care Plus was found to have fee-for-service spending that was 2 percent to 3 percent higher than comparison practices after accounting for incentive payments. This is consistent with other research that have found medical home models do not save Medicare money.

After assessing the evaluations of more than a dozen CMS initiatives over the past eight years, the chart above shows no other model comes close to the success of ACOs. This could be a result of clinicians in ACOs examining patients’ total cost of care, enabling tradeoffs between sites of care such as nursing homes versus patients’ homes and being incentivized to choose care patterns that better manage long-term outcomes, especially for high-need, high-cost patients. Making providers accountable for the whole patient and total costs incentivizes a delivery system to identify problems early and intervene to prevent long-term consequences. These incentives and the close clinical relationship to the patient through data may be the reason ACOs produce better results and why they are the best model for health care in the decades to come.

MSSP Grew Fast And Then Faltered Due To Policy Changes

By early 2018, the MSSP had grown to be by far Medicare’s largest APM. However, in a Health Affairs Blog post last fall, it was reported that a surprisingly small number of ACOs joining MSSP in 2019. This pattern continued in January, when CMS released 2020 participation data that showed a decline in the number of new ACOs joining the program.

The turning point in ACO growth can be traced back to CMS’s “Pathways to Success” rule, which was finalized in late 2018 and redesigned the MSSP, the most popular ACO program. Since the rule took effect, an average of 38 new ACOs have joined the program each of the past two years, a startling decline after the program averaged 107 new ACOs annually between 2012 and 2018. We believe the combination of more aggressive requirements to take on risk and a reduction in the portion of savings ACOs are able to keep, among other “Pathways to Success” changes, will continue to result in slower growth of the MSSP.

Furthermore, CMS has announced that due to the pandemic of 2020 there will be no opportunity for new providers to apply for the MSSP in 2021, guaranteeing a continued shrinking of the program. If CMS’s goal is to grow value-based payment models and APM participation, then it needs to reevaluate policies that have stunted the growth of its largest and most successful model to date (exhibit 3).

Exhibit 3: New Medicare Shared Savings Program ACOs

Source: Authors’ analysis of Centers for Medicare and Medicaid Services data, 2020.

COVID-19 Pandemic And Population Health Care

While it is impossible to assess the full impact of the pandemic, we know that timely identification of persons at risk who have complex and confounding health conditions should be prioritized for risk mitigation. Based on interviews with selected organizations, population-focused organizations such as ACOs are uniquely positioned to both early identify patients at risk and manage the comorbidities of the pandemic survivors. Capabilities include keeping current patient registries of clinical histories and problems, a staff of clinical coordinators to manage high-risk patients with chronic conditions, established contractual relationships with postacute facilities to deliver effective care transitions, and established telehealth services.

After years of testing a myriad of APMs, data show that population-focused, total cost of care models are consistently producing savings that episodic-based or disease-specific models are not. While CMS continues to experiment by rolling out new models and fine-tune existing ones, we have enough evidence now to support what is working better. Based on these results, it may be time to slow the proliferation of narrowly focused APMs and place priorities on how to expand and improve the population-focused accountable care model.

Laisser un commentaire