Only a third of CMMI models yield substantial net savings

Not many alternative payment and care delivery models from the Center for Medicare and Medicaid Innovation, or CMMI, have generated significant savings after a decade of CMMI model design and implementation. 

A recent analysis from Avalere Health studied 18 CMMI models and their impact on savings and quality of care using publicly available data and evaluation reports as of December 2024. The models studied in the analysis have at least two years of model performance, two published evaluation reports and that have affected over 25,000 beneficiaries or episodes of care. 

Avalere researchers found that only one-third of the CMMI models studied "yielded substantial net savings for the federal government."  

Another one-third of the CMMI models even generated substantial net losses and the remaining third had nominal financial effects per their last evaluation report. 

CMMI models generating the most net savings included the Maryland All-Payer Model, which reduced spending the most by $975 million, Maryland's Total Cost of Care Model, Accountable Health Communities Model and the Pioneer Accountable Care Organization (ACO) Model.  

Researchers pointed out that state and community-based models were more successful at reduced expenditures, with the models (Maryland All-Payer Model, Maryland Total Cost of Care and Accountable Health Communities Model) generating $2.1 billion in net savings after accounting for operational costs. 

In contrast, CMMIs two health plan-based models (Medicare Advantage Value-Based Insurance Design Model and the Part D Enhanced Medication Therapy Management Model) yielded the most net losses at $4.8 billion, including operational costs. 

Some ACO and disease-specific models, like the popular Bundled Payments for Care Improvement Advanced model and Comprehensive Care for Joint Replacement Model, also generated net savings to the federal government. However, the 12 ACO and disease-specific models studied by Avalere cost around $5 billion after accounting for operational costs. 

Researchers said spotty savings results suggested "careful design and population selection may help drive improved performance." Provider attrition, selection bias and unpredictable payments were also behind net losses across other ACO and disease-specific models. 

On the other hand, whether a model is voluntary or mandatory did not seem to have a significant impact on a model's financial performance. 

In addition to cost savings, Avalere also analyzed quality improvement and model transparency.  

The analysis also found mixed results with quality improvements, with just four CMMI models studied showing improvement, three showing nominal improvement, four showing no statistically significant impact on quality and seven with mixed results. CMMI models also faced challenges with improving patient experience, with only four models demonstrating nominal improvement. 

Avalere also identified opportunities to increase model transparency. The analysis found a general lack of opportunity for public input via public comments on rulemaking, formal requests for information and public listening sessions. 

CMMI could also enhance data transparency as Avalere found issues with participant-level data access compared to access to aggregate model results. 

CMMI has non-legislative authority to design and test models in Medicare, Medicaid and CHIP. The models, according to statute, must either improve quality of care without increasing spending, reduce spending without reducing quality of care or improve quality and reduce spending. Meeting one of these criteria makes a CMMI model eligible for nationwide expansion through rulemaking. 

However, only four of the 50 total CMMI models have been selected for nationwide expansion in the CMMI's decade of existence. Moreover, CMMI had not realized the $1.3 billion in net savings expected between 2010 and 2019, according to the analysis. 

Where CMMI goes in the next couple of years will depend largely on the Trump administration. It has already terminated four CMMI models -- Maryland Total Cost of Care, Primary Care First, End-Stage Renal Disease Treatment Choices and Making Care Primary. CMS expects to save about $750 million by stopping the models by Dec. 31, 2025. 

Jacqueline LaPointe is a graduate of Brandeis University and King's College London. She has been writing about healthcare finance and revenue cycle management since 2016.