MedPAC report on alternative payment models

In 2021, CMS expects to operate 12 alternative payment models (APMs) offering 25 distinct tracks for providers to choose from that involve different payment options and risk arrangement… providers serving about 20 percent of Medicare beneficiaries participate in this APM

That is a quote from a June 2021 MedPAC report on CMS APM’s serving Medicare beneficiaries. As the quotation indicates, there are a large number of APMs that Medicare is using. The largest is the Medicare Shared Savings Program (MSSP); about one-third of traditional Medicare fee-for-service beneficiaries participate are attributed to providers in MSSP. CMS’s goal for implementing these programs was to either reduce spending without reducing the quality of care or improve quality without increasing spending. CMS used seven general category for defining different types of APMs:

• Accountable care models (e.g., MSSP, Next Generation ACO)
• Episode-based payment models (e.g., BPCI, CJR)
• Primary care transformation models (e.g., CPC+)
• Initiatives to speed the adoption of best practices (e.g., Partnership for Patients)
• Initiatives to accelerate the development and testing of new payment and service delivery model (e.g., Emergency Triage, Treat, and Transport Model)
• Initiatives focused on beneficiaries who are dually enrolled in Medicare and Medicaid (e.g., Financial Alignment Initiative for Medicare-Medicaid Enrollees)
• Initiatives focused on the Medicaid and CHIP populations (e.g., Strong Start for Mothers and Newborns Initiative)

Interest in these models increased when the Medicare Access and CHIP Reauthorization Act of 2015 (MACRA) increased by 5% if they participated in an Advanced-APM. Below are some of the APM programs:

• Next Generation ACO Model. ACO model that involves more financial risk than the MSSP, with participating providers subject to either 80 percent or 100 percent shared savings and losses.
• Global and Professional Direct Contracting (GPDC) Model. This APM is the successor to the Next Generation ACO Model. It offers primary care capitation payments combined with 50% shared savings or losses (in the “professional” option) or full capitation coupled with 100% shared savings or losses (in the “global” option)
• Geographic Direct Contracting (Geo) Model. Under the Geo Model, all Medicare FFS beneficiaries who live in certain CMS-selected geographic regions will be aligned to one of several participating organizations. These organizations’ annual spending targets for their attributed beneficiaries will be set based on bids they submit to CMS, rather than spending targets determined by CMS (as is the case for other APMs). Participating organizations are responsible for 100% of the shared savings and losses, but these organizations are able to make use of prior authorization and claim reviews not typically available under traditional Medicare FFS.
• Comprehensive ESRD Care Model. This shared savings model for dialysis clinics, nephrologists, and other providers treating beneficiaries with end-stage renal disease (ESRD) had a two-sided risk track which qualified as an A-APM.
• Comprehensive Care for Joint Replacement (CJR) Model. This APM is an episode-based (i.e., bundled) payment model for hip and knee replacements.
• Bundled Payments for Care Improvement (BPCI) Advanced Model. Whereas the CJR program focused on bundled payment for surgeries, BPCI was an episode-based (i.e., bundled) payment model for a variety of inpatient and outpatient procedures and conditions.
• Oncology Care Model (OCM). This model included hybrid payments for chemotherapy involving elements of episode-based payment and primary care transformation models. The two-sided track version was considered an A-APM. The Oncology Care First model was going to replace OCM, but was put on hold due to COVID-19.
• Comprehensive Primary Care Plus (CPC+) Model. Primary care transformation model that pays primary care providers partial capitation plus small performance bonuses.
• Primary Care First Model. This APM is the successor to CPC+, involving larger performance bonuses.
• Maryland’s Primary Care Program & Care Redesign Program. While modeled after CPC+, this program also included an option modeled after BPCI Advanced, as well as an option allowing hospitals to pay their care partners incentive payments for engaging in care redesign interventions (e.g., care coordination, discharge planning, improving clinical quality and patient experience).
• Vermont ACO Initiative. This APM took the Next Generation ACO Model approach, but applied it across payers. The goal is to make it easier for providers who accept reimbursement from multiple payers.

What happens to all these models? MedPAC notes that most of the programs typically last only 5 years. After 5-years, CMMI either abandons the program if it doesn’t show results or updates the program often under a new name.

Do these APM actually save money? Typically yes, but often not much and often not after accounting for bonus payments.

…almost all of CMS’s accountable care and episode-based payment models have generated relatively small gross savings for the Medicare program, before model payments (e.g., performance bonuses) are taken into account…. After bonuses are paid to model participants, gross savings are reduced and in some cases Medicare expenditures in APMs have exceeded what they would have been otherwise

The models that have generated net savings include a few of the early ACO models, CJR and some BPCI episodes (but not in aggregate). Why haven’t savings been larger? MedPAC poses the following potential explanations.

• Financial incentives aren’t strong enough. Despite financial incentives for shared savings and other programs, performance payments are often paid years into the future and these bonus payments may not be high enough to offset increased revenue from more fee-for-service utilization. Also, provider may care for a mix of patients both under APM and FFS and keeping straight who is paid under what program is difficult.
• Payment incentives are hard to understand. This stuff is complex. Clinicians are trained in medicine, not economics. Even if they were, the bureaucratic details and constantly shifting programs are difficult to understand.
• Payments to whom? While provider organizations may get bonus payments, its not clear if those payments trickle down to individual providers.
• Voluntary payment models lead to selection bias. Those who are most likely to produce savings are most likely to join the APM. Thus, the true causal effect is hard to estimate.
• Some clinicians may be unable to make the infrastructure investments needed to succeed in new payment models. What CMS is asking for (implicitly) under APM is population health management. Its not clear if all providers have the IT, staff, and expertise to to reach these goals.
• Beneficiaries are not aware of APM. Beneficiaries are typically assigned retrospectively to APM based on service use. Thus, they may not even be aware they are in an APM and if so, which one.

Despite those potential benefits, having so many models is confusing. MedPAC writes:

In many cases, providers participate in multiple CMS APMs simultaneously, and Medicare beneficiaries are attributed to multiple models at the same time. This overlapping participation can have unintended consequences. For instance, savings that are generated for a beneficiary served by different sets of providers participating in different APMs can be allocated to providers in only one of these models, thus diluting financial incentives in the other models. Overlapping participation can also make it difficult for evaluators to accurately assess the impact of a given payment model on program spending and quality

This is a sensible solution. MedPAC believes that this recommendation could produce modest savings. However, the Congressional Budget Office estimates that simplifying APM will have no net impact on Medicare spending. We’ll go to the data a few years from now to find out what the answer ends up being.