How successes, failures of early CMMI models can inform a 'new direction'

While not all models developed and tested through the Center for Medicare and Medicaid Innovation (CMMI) were successful, each provided valuable lessons for future experiments in care delivery and payment from both government and private payers, according to an article published by Health Affairs.

Led by Rocco Perla, EdD, the former director of Improvement Dissemination and Model Expansion at CMMI and founding director of the CMS Regional Innovation Network, the article looked at the design, results and lessons learned from five early models launched by CMMI: Pioneer and Advance Payment Accountable Care Organization (ACO) initiatives, Bundled Payments for Care Improvement (BPCI), Comprehensive Primary Care (CPC), the Partnership for Patients initiative and Health Care Innovation Awards (HCIA).

Not all the models hit their targets, Perla and his coauthors wrote. Pioneer ACO, for example, “struggled to deliver the degree of savings they had anticipated.” But it was still a worthwhile experiment, they argued, for determining key elements of ACO design (like prospective attribution of patients) and finding what elements didn’t work (like the model’s “complex” methodology for setting spending targets). These lessons helped inform more advanced ACO models like the Medicare Shared Savings Program and got commercial payers to take notice.

“Approximately thirty-two million lives are covered under 1,366 active ACO contracts, more than half of which belong to commercial payers,” Perla and his coauthors wrote. “Investors and entrepreneurs recognize ACOs as a rapidly growing opportunity to develop products catering to providers in value-based payment arrangements. Venture capital firms have invested directly in ACOs, recognizing the organizations’ challenges in accessing capital to invest in care management infrastructure.”

In the case of CPC, the model’s flaws for the commercial market were exposed by testing it out. Perla and his coauthors wrote practices weren’t motivated enough by the shared savings the program offered and struggled to convince electronic health record (EHR) vendors to make direct reporting of quality data possible. These lessons resulted in changes in the model’s larger successor, CPC+.

Commercial payers did respond, however, to the results of BPCI. While episode-based payments have been a part of CMS models since the 1980s, promising outcomes (like savings from reducing the cost of postacute care) led to greater participation in other bundled payment models—and at least anecdotal evidence that “commercial payers have implemented bundled payments for discrete conditions or are considering doing so.”

The successes and failures were synthesized by Perla and his coauthors into three major lessons for future models:

• CMS testing of models speeds up innovation. Even when models didn’t achieve their targets, the takeaways came faster and on a greater scale than any commercial payer could hope to achieve.
• Early CMMI models may have been too ambitious. Perla and his coauthors wrote that it was “unrealistic to assume that any new model or combination of models would rapidly achieve the Triple Aim” due to complexity and competing interests in the current healthcare system.
• Innovation is not always about creating new models. The authors of the article expect the second phase of CMMI to look at ways to better integrate and coordinate existing models after the “wildfire of innovation” of its first phase.

Changes will be coming to CMMI, as the article referenced the “new direction” promised by CMS Administrator Seema Verma, MPH, which Perla and his coauthors guessed would mean a greater focus on “market forces and informed consumers.”

It may also mean changes come more gradually in response to feedback from providers. The article mentioned the cancellation of mandatory bundled payments in cardiac care and joint replacement, which had been criticized as too burdensome by groups like the Federation of American Hospitals. Though new HHS Secretary Alex Azar has voiced support for mandatory models, if CMS continues to pursue only voluntary programs, there won’t be as “compelling business case to drive rapid and widespread change.”

“Major change takes time, and voluntary programs will typically take longer,” Perla and his coauthors wrote.