CMMI and the slow march toward value
Congress created the Center for Medicare & Medicaid Innovation under the Affordable Care Act to design, test and expand new payment and care delivery models that cut spending without lowering the quality of care or increased quality without raising expenditures.
It was also supposed to hit on the third leg of the so-called Triple Aim by improving the overall health of Medicare and Medicaid beneficiaries.
Despite the lofty agenda, the CMMI has a modest track record. Its muted success in bringing down healthcare costs is due in part to poorly targeted and designed pilots, and a tendency to pursue incremental progress over comprehensive reform.
Still, the Innovation Center is expected to save Medicare $18 billion between 2017 and 2026, according to a report from consulting firm Avalere Health. “I think (CMMI is) one of the bigger impacts of the ACA that’s rarely talked about,” said Dr. Patrick Conway, former CMMI director and former CEO of Blue Cross and Blue Shield of North Carolina. That $18 billion is only slightly more than half of the $34 billion projected by the Congressional Budget Office in 2016.
Those figures might seem trivial relative to the more than $1 trillion the federal government spent on Medicare and Medicaid in 2018 alone. But many experts say the numbers don’t reflect the CMMI’s most significant impact: changing how providers think about and manage the healthcare delivery system through its emphasis on value-based care.
“I think it’s changed the cultural fabric of how providers approach healthcare,” said Dr. Amol Navathe, a Medicare Payment Advisory Commission member and associate director of the Center for Health Incentives and Behavioral Economics at the University of Pennsylvania.
In studies that analyze the effects of the CMMI’s alternative payment models, the control groups have provided more efficient care over time, according to forthcoming research by Navathe and his colleagues. That means providers are increasingly delivering better-value care, even if they aren’t involved in a specific alternative payment model. That might be because they’re participating in another CMMI model or commercial value-based payments, or because cost-effective practices are increasingly common thanks to peer influence.
Researchers “are comparing a particular model against a baseline that is already improving,” Navathe said. “The leading hypothesis for why that baseline trend is improving is that there’s been this psychological shift in the delivery system … because of the innovation center and Medicare’s …
deliberate efforts to move toward higher-value care.”
The CMS for decades tried new models of paying for and delivering care before Congress created the CMMI, but the process was complicated and burdensome. Many of the demonstration projects were underfunded, overly restrictive or failed to hold the agency or providers accountable. The CMMI’s approach has noticeably improved each of those areas because the CMS finally has a dedicated agency with the power to experiment with new models.
“Prior to CMMI, it was a bit of a mess as to how we fostered innovation in the Medicare program,” said MedPAC Commissioner David Grabowski, a professor in Harvard Medical School’s department of healthcare policy.
Andy Slavitt, former acting administrator of the CMS and founder of United States of Care, said the CMMI was imbued with a sense of urgency. “When I was at CMS, my guidance to (CMMI) was ‘Go figure out how fast people can change and try to go about 10% to 15% faster.’ ”
But that doesn’t mean the CMMI has been an unqualified success. Industry experts contend that the agency’s approach to designing, testing, implementing and evaluating new models leaves a lot to be desired. The endgame is to transform the healthcare delivery system, not to test models.
The CMMI has tested more than 40 alternative payment
When I was at CMS, my guidance to (The Center for Medicare and Medicaid Innovation) was Go figure out fast peoplecan change and try to go about 10% to 15% faster." Andy former administrator United Slavitt CMS States acting and of Care founder of
approaches since its creation, but just two of them—the Pioneer Accountable Care Organization model and the Medicare Diabetes Prevention Program—have been certified for expansion by the CMS actuary.
“There have been far too many misses relative to successes,” Grabowski said.
But questions even plague models deemed ready for prime time. The Diabetes Prevention Program, for example, was supposed to help as many as 110,000 people avoid Type 2 diabetes each year, but only about 200 people enrolled in 2018 because of regulatory obstacles that blocked providers from participating. The CMS estimated the program would save Medicare $182 million from 2018 to 2027, but it’s unlikely to deliver those savings.
That’s left some observers wondering whether the agency has focused enough on addressing the right problems and creating appropriate solutions.
Some of the agency’s demonstration models haven’t been well-designed to achieve the Triple Aim. For example, the comprehensive primary-care initiative seems intended to realize other policy goals like increasing payments to primary-care providers rather than reducing spending or improving quality.
“My big concerns with CMMI haven’t been with the goals of the agency,” Grabowski said. “But that we haven’t designed a lot of programs to actually target those goals.” Plus, the Innovation Center’s demonstrations don’t include many different types of providers because the agency has mainly focused on developing new iterations of its existing models. Without including the full spectrum of providers in the transition to value-based payment, the dream of healthcare transformation could prove elusive.
“CMMI has focused more on (developing new versions of existing models) rather than expanding its scope of models, and, as a result, there are
I don't think that people have an appreciation for what it takes to stand up one of these models. It's a significant investment. We have to be very selective about the models that we stand up." Seema CMS administrator Verma
significant specialties that have been left relatively unaffected with no particular plan or pathway to include them,” said Dr. Randy Pilgrim, enterprise chief medical officer for SCP Health, a practice-management company operating in 30 states.
Instead of continuously developing new iterations of existing demonstrations, the center should focus on a small number of models that are most likely to reduce spending or increase the quality of care, some experts said. Those models should include all types of providers since an out-and-out redesign of the U.S. healthcare delivery system is the ultimate goal of the CMMI.
But many industry insiders say that expansion shouldn’t be the standard used to judge CMMI models because it’s too high of a bar—much higher than for private payers—and it obscures the progress of both the agency and providers on value-based care. Changing healthcare is a learning process that requires constant testing and development, so the industry shouldn’t expect one model or program change to be a silver bullet.
“Even if a model isn’t certified, they take what they’re learning and apply it to a new Innovation Center model
or apply it to Medicare Shared Savings Program,” said Emily Brower, senior vice president of clinical integration and physician services for Trinity Health. Many of the lessons from ACOs have been applied to the Medicare Shared Savings Program, a mandatory program created by Congress rather than the CMMI.
Critics may also underestimate how difficult it is for the center to design, test, evaluate and expand a model. The regulatory process is lengthy and complicated because the agency needs to make sure that its demonstrations won’t have substantial adverse effects on providers or quality.
“I don’t think that people have an appreciation for what it takes to stand up one of these models,” CMS Administrator Seema Verma said. “It’s a significant investment. We have to be very selective about the models that we stand up.”
Yet this cautious approach to policymaking could be a barrier to meaningful progress. The CMMI’s models have come under fire because many of them have offered financial rewards that are too small to encourage significant changes in healthcare delivery. They also tend to expose doctors and hospitals to little, if any, risk.
“Most of the payment reforms that have been implemented so far have been relatively modest,” said Dr. Mark McClellan, a former CMS administrator and founding director of the Duke-Margolis Center for Health Policy at Duke University. “Healthcare reform is hard, especially when it’s not that hard to keep making money the old-fashioned way through fee-for-service.”
A greater focus on nonfinancial interventions could make payment reforms more effective too. The agency hasn’t experimented with demonstrations that incorporate the lessons of behavioral economics, such as how information is delivered to providers.
The CMMI’s demonstrations typically have a reconciliation period of three to six months, so there’s a lag between the time that providers deliver care and when they get paid. That weakens the ability of financial rewards to affect provider behavior. It also makes it hard for providers to understand how the agency is evaluating them. But the claims system prevents the CMMI from significantly reducing the time between care delivery and payment. “If you want to maximize the results of a particular behavior change, you want to link the reward to the behavior change,” Navathe said. “CMMI’s models are doing a terrible job of this.”
The CMMI’s short demonstration periods may also stymie progress, as providers don’t have the time to adopt necessary changes and respond to program incentives. A recent study of the Bundled Payment for Care Improvement Advanced model found that most of the demonstration’s savings came from providers who joined the program early.
Many experts think that too many of the CMMI’s models are voluntary rather than mandatory. When the agency first started, the models needed to be optional because neither the center nor hospitals knew how providers or the demonstrations would perform because so few people had experience beyond fee-for-service reimbursement.
Now that the CMMI and providers have a decade’s worth of experience under their belts, it’s time for the agency to test and expand
more mandatory models. Volunteer models have proven that doctors and hospitals can respond to alternative payment models’ incentives. But they’re limited. The only providers that participate are the ones that think they’ll be successful.
“You may be inevitably paying for some of the success that would have happened … anyway,” said MedPAC Vice Chairman Paul Ginsburg, chair in health policy studies at the Brookings Institution and professor of health policy and management at the University of Southern California. “That makes it hard to reduce spending overall or save money for the Medicare program.”
Mandatory demonstrations have a greater potential for impact because they compel providers to undergo some of the management changes necessary to succeed under value-based payment, even if they’re reluctant. Providers with the most room for improvement usually haven’t adopted the reforms with the best return on investment, so getting them on board is critical for reducing costs or improving quality.
But mandatory models need to be designed differently because the CMMI runs the risk of negatively affecting the providers least prepared to adapt to payment and delivery reform efforts. That could cause access or quality problems for Medicare and Medicaid beneficiaries.
“Payment models that change too drastically can’t be implemented,” Slavitt said. “The American healthcare system needs to be able to adapt … do you move as fast as the fastest markets, or do you move as fast as the slowest markets can go?”
In place of mandatory models, “taking voluntary participants and randomizing them is probably the best way to go,” Navathe said. The CMMI would be able to test different intensities of interventions without running into some of the selection problems that are typically associated with voluntary models.
Uncertainty is another critical issue for CMMI models and providers. Since the CMS won’t certify most demonstrations for expansion, doctors and hospitals might not be willing to make all the changes necessary to maximize cost savings because it’s unlikely that a model will become permanent. That could limit the upside of CMMI demonstrations and slow the transition to value-based payment.
“I think the lack of long-term commitment on the part of CMMI does impact how people in businesses or practices are willing to invest in the infrastructure that’s necessary to succeed,” Brower said.
Trust is an issue too because some providers feel like CMMI has burned them in the past. In 2018, seven ACOs announced that they would exit the Next Generation ACO model after the agency decided to retroactively lower the average risk score for 2017 by 4.82%, which wiped out their ability to earn a bonus or avoid paying penalties. These providers believe the CMMI changed the rules in the middle of the game.
“I would not participate in a CMMI model,” said Alison Fleury, senior vice president of business development for Sharp HealthCare in San Diego. “I would not participate in any model that provides (CMMI with) that much latitude. … You can’t be in a business transaction where terms are dictated one way.”
But the CMMI is a federal agency, and it’s responsible for spending federal funds appropriately. It can’t focus exclusively on the wants and needs of providers and must balance them with the requirements of beneficiaries and the public, CMS officials say.
“I think we have an obligation to taxpayers,” Verma said. “I think that if we can clearly see that a model is not going well … we have an obligation to take action on that.”
New flexibilities that allow the CMMI to pay for innovative delivery and treatment changes like genomic medicine could lead to more rapid adoption. Some of these treatments might require modest increases in spending, but they have the potential to deliver outsized benefits in the quality of care. The center doesn’t have the power to roll out models that would increase spending, so Congress would have to grant it new authority.
“If we’re evaluating a model, and it says it’s going to increase quality without a savings, then that doesn’t actually qualify to become a model,” Verma said.
But despite the complexity and difficulty of transforming the healthcare delivery system, many industry insiders remain optimistic about the CMMI’s continued progress.
“Sometimes we get into the details of these models, (but) we take for granted that the providers that participate in these models are better able to understand the conditions and patterns of care for the people they serve and, therefore, change them,” Brower said.
Still, that change might not be happening fast enough. Healthcare spending grew 4% on a per capita basis in 2018, and the CMS projects overall spending will reach nearly 20% of GDP by 2027.
“One of the most disappointing things you learn when you’re at CMS is that everyone’s filled with good ideas, and then you quickly learn that if you throw all these things at the healthcare system, they scream because it feels like a lot of change,” Slavitt said.●