Customize your JAMA Network experience by selecting one or more topics from the list below.
The Patient Protection and Affordable Care Act (ACA) is now more than 10 years old, and so is one of the law’s most important creations: the Center for Medicare and Medicaid Innovation (CMMI). Before the CMMI, the scientifically guided redesign of US health care financing and delivery had no principal governmental home or sponsor (ie, there was no analogue to the role the National Institutes of Health has for biomedical research). With CMMI, the health care system has that; or, more accurately, it could.
As stipulated in the ACA, the role of CMMI is “…to test innovative payment and service delivery models to reduce program expenditures…while preserving or enhancing the quality of care furnished to individuals….” The text of the ACA suggests, as examples, more than 20 such models that CMMI could assess.1
Congress was generous to CMMI. It appropriated $10 billion for 2011-2019 and another $10 billion for 2020-2029, thus relieving CMMI from annual appropriations debates and hearings. Furthermore, in a provision without precedent, the ACA gave the Secretary of Health and Human Services (HHS) the authority to issue regulations, independent of legislation, that could expand to national scale any model that CMMI demonstrates would either reduce spending without harming quality or improve quality without increasing spending.2
In its first decade, CMMI developed and tested more than 50 models. Fifteen of those models were in 3 categories that involved changing payment to support pursuit of the “triple aim” of better care for individuals, better health for populations, and lower costs: 6 tests were of primary care practice models, with payment support for advanced primary care practice care delivery such as the medical home; 5 tests were of accountable care organization models, which reward clinicians and hospitals for improving the quality and cost of care3; and 4 tests were of episode-of-care payment models, which use bundled payment to clinical organizations, such as for an entire 90-day period before, during, and after total joint replacement, rather than payment for individual services.
CMMI has devoted more attention to testing new payment models than to fostering specific care models. However, it has sponsored some direct tests of care delivery innovation, such as comprehensive perinatal care, diabetes prevention, and reduction of health risk for Medicare beneficiaries.
What has been learned in a decade of CMMI investment? And what changes in CMMI seem wise for the next decade?
According to a recent review, only 5 of the 54 CMMI models tested have “resulted in significant financial savings” according to independent evaluations.4 Evaluations on the CMMI website show a more favorable profile. Of 55 models identified on the CMMI website, 10 improved both costs and quality, 6 reduced costs without impairing quality, 9 improved quality without raising costs, 15 increased costs without changing quality, 5 had no effect, 4 had mixed effects, and 6 were not intended to affect quality and costs or were not evaluated.
The learning from the largest tests in terms of number of beneficiaries, number of organizations, or amount of Medicare expenditure affected, both positive and negative, is important. For example, primary care practice models clearly can improve quality but did not reduce costs. Accountable care organizations (ACOs) can produce significant savings and some quality improvement. Bundled payment models can reduce the costs for acute care and maintain quality but are subject to gaming if enrollment is voluntary.
Despite these lessons, to date the Centers for Medicare & Medicaid Services (CMS) actuary has certified only 4 models for national scaling. Only 2, the National Diabetes Prevention Program (in which public and private organizations offer lifestyle change supports to reduce the risk of type 2 diabetes) and a few elements of the Pioneer ACO model (testing downside risk for advanced ACOs), have actually been scaled.
Thus, in its first 10 years, CMMI’s investments have yielded considerable information about how clinicians, organizations, payers, and CMS can accelerate progress toward better care and lower costs. But CMMI has not yet become the engine that it could be for transforming the nation’s health care finance and delivery. To fulfill that potential, it needs to change, and HHS and CMS need to work more closely with CMMI to bring its lessons to much larger scale.
Recommendation 1: Connect the CMMI agenda more explicitly to a broad HHS and CMS and a strategic plan and aims for improving health and health care delivery. The models that CMMI tests should link tightly to a shared vision, fully embraced by all of HHS and CMS, of a transformed health care system, with better outcomes, more equity, and much higher value. Real change of that magnitude will require some mandatory models. CMMI should curate its portfolio of models and serve as the test bed and driver at the core of the CMS transformation strategy. The office of the HHS secretary should frequently review CMMI’s projects and investments to ensure alignment with the overall strategy.
Recommendation 2: Use CMMI authority to scale the ACO model nationally by making it mandatory for all Medicare participating clinicians and hospitals. Clinicians, hospitals, and payers find it difficult to operate in an ambiguous world straddling payment for volume and value. Although voluntary participation has made evaluation of ACOs difficult,5 the Medicare Payment Advisory Commission and others have concluded that different CMS ACO models during the last 15 years have consistently produced modest savings for CMS.6-8 CMS should gradually but steadily expand ACO adoption during the next 5 years until virtually all Medicare participating organizations and clinicians are operating within accountable organizations. Advanced primary care practice models will be a natural core feature. Part of the expansion should include, as much as feasible, progressing to capitation of ACOs for total cost of care.
Recommendation 3: Sponsor models directed at improving health equity. Health inequities in the US, driven by poverty, structural racism, gaps in rural care, and failures to improve the social determinants of health, far exceed inequities from differences in health care. CMMI should sponsor tests of specific new models of care that aim to improve health equity and include equity and social determinants of health in all evaluations. The White House could create an “all-of-government health equity campaign” that extends beyond HHS to bring together programs for social determinants of health from the US Departments of Agriculture, Education, Transportation, Veterans Affairs, Housing and Urban Development, Justice, and more.
Recommendation 4: Rebalance CMMI model tests toward delivery system redesigns, not just new payment models. Payment matters, but ultimately only changes in care at the patient and clinician level can produce better outcomes and lower costs. CMMI should actively solicit and support tests of radical new care delivery designs such as comprehensive alternative care models that minimize inpatient hospital use by fully integrating digital health, home health care, multifunctional teams, freestanding outpatient services, and hospital at home. According to CMMI results already available, the following delivery system models should be brought to national scale now: the Medicare Care Choices Model (which expands hospice benefits), Strong Start for Mothers and Newborns (in which one model, which could be scaled to Medicaid, was associated with lower cesarean delivery rates than the comparison group [17.5% vs 29%] and savings in total expenditures for mothers and infants from delivery until the infant’s first birthday of more than $2000 per birth), and the Partnership for Patients (which was shut down despite having achieved large improvements in patient safety and reduced costs).
Recommendation 5: Build much stronger cooperative innovation programs between CMMI and private-sector health care insurers and delivery, including academia. Federal health coverage programs, including the Defense Health Agency, the Veterans Health Administration, and the US Office of Personnel Management, should require participating insurers to align with and participate in CMMI programs. CMS and CMMI could develop teams that include both CMS career staff and, on 1- or 2-year details, clinicians and leaders from the health care industry to help design and manage tests of models. (The Defense Advanced Research Projects Agency uses an approach analogous to private-sector engagement, which CMMI could emulate.) CMMI also could recruit and certify a cadre of several thousand “Innovation Fellows” throughout the nation, volunteer clinicians and managers who work within health care and public health systems to further facilitate public-private synergy.
Recommendation 6: Change CMMI clearance processes, evaluation methods, and cycle times to better support the tempo and culture needed for effective innovation. An innovation engine for any organization should operate differently from its production engine. It should take more risks, move more quickly, and learn more intentionally. Instead, CMMI has been enveloped in the normal HHS approval processes used for regulations, with multiple clearance steps that create drag and slow cycle times. For instance, the Medicare Care Choices Model was first considered at CMMI in 2013 and was launched in 2016, but, despite positive cost and quality improvements, it has not been scaled to date. HHS should restructure CMMI processes to maximize speed to launch, test, learning, and spread, in cycle times of weeks or months, not years.
Evaluation methods for CMMI ought to be agile and rapid, in accord with the National Academy of Medicine’s image of a “learning healthcare system,” and fit for a constantly evolving health care environment demanding continual adjustments. CMMI should establish robust intramural research and evaluation capabilities and make all components of evaluations public.
Despite the often-inspiring efforts of its clinicians and staff, US health care fails far too often and costs far too much. Stressing current designs will not change that; only new designs at scale can. In CMMI, the ACA wisely created a badly needed center for the rapid study of bold changes in the financing and design of that care, and a route to convert learning into large-scale change. So far, that has not happened. It can, but only if CMMI is unleashed—protected from habitual, overcautious controls and doctrine-driven political pressures—to innovate fast and far-beyond-usual boundaries, and only if that energy is fully aligned with a shared vision of the new health care system that the nation needs.
Corresponding Author: Donald M. Berwick, MD, Institute for Healthcare Improvement (IHI), 131 Lake Ave, Newton, MA 02459 (firstname.lastname@example.org).
Conflict of Interest Disclosures: None reported.
Additional Information: Dr Berwick was the administrator of the CMS in 2010 and 2011, and in that role he oversaw the establishment of the CMMI. Dr Gilfillan was the director of the CMMI from 2010 to 2013.
Berwick DM, Gilfillan R. Reinventing the Center for Medicare and Medicaid Innovation. JAMA. 2021;325(13):1247–1248. doi:10.1001/jama.2021.3203
Create a personal account or sign in to: