The movement to value-based care is alive and well, with the Trump administration eager to put their own stamp on the program. In one of his first major speeches as Secretary of Health and Human Services, Alex Azar said value-based care “needs to accelerate dramatically.” This reprises the administration’s draft 2018–2022 strategic plan for HHS and statements by Seema Verma, the Administrator of the Centers for Medicare and Medicaid Services (CMS).
Given the sustained support for models that encourage paying for value, it is worth asking just how value-based payment models in Medicare might evolve to meet the administration’s objectives and approaches. The centerpiece of CMS’ alternative payment models is accountable care organizations (ACOs), which now cover 10.5 million seniors in the Medicare Shared Savings Program (MSSP) alone, and many millions more commercially insured patients.
We believe that greater technical and policy alignment of the ACO program with Medicare Advantage could encourage enrollment and benefit providers and patients alike, by adding predictability and reducing complexity. Those would be welcome developments in our nation’s ongoing move toward value-based care.
Three Recommendations for ACO Policy
Limit one-sided risk. Today, the vast majority of Medicare ACOs (460 of 561, or 82%) are still in one-sided risk models. The Track 1 MSSP model undeniably serves as a critical on-ramp for providers to gain experience with total cost of care models, particularly for the physician-only group of ACOs that have demonstrated the greatest ability to generate savings for Medicare. However, upside-only models do not force organizations to make a commitment to a new business model centered on value and outcomes, rather than on volume and market power.
Administrator Verma recently expressed her intention to take a close look at the current ACO model “to make sure that they are not driving out smaller practices.” A major driver of provider consolidation has been differential payments for “facility fees” when private practices are purchased and rebadged as hospital outpatient departments (which the 2018 President’s Budget has proposed to eliminate). But short of practice acquisition, hospitals can use one-sided ACO models as a safe haven for a softer form of consolidation: increasing “in-network utilization” for purposes other than lowering costs. Allowing providers to stay in one-sided risk also decreases savings for Medicare.
To address these concerns, we predict that CMS will almost certainly not extend one-sided risk for the third MSSP contract and may even propose regulations that would limit ACOs to one 3-year contract in Track 1, beginning in 2019, perhaps focused on hospital-based ACOs. This would certainly result in some ACOs dropping out of the program; among the most recent cohort ending their initial 3-year contract, only 8 out of 65 (12%) voluntarily moved to risk.
However, we believe that if two-sided risk is made less risky, and more predictable, then most successful ACOs will be willing to move up the risk continuum. The success of the value-based movement can be measured not only by the number of ACOs, but also by their ability to generate results.
Make downside risk less risky. The Medicare ACO Track 1+ model, which was unveiled by the CMS Innovation Center in late 2016, took a big step toward creating a two-sided model that is feasible for organizations of differing finances by introducing the concept of revenue-based downside risk. This model qualifies as an Advanced Alternative Payment Model under the “more than nominal risk” test of MACRA, as it provides for penalties of up to 8% of practice revenue for poor performance. For organizations with profit margins of 2–3%, that is certainly sufficient to assuage concerns that ACO waivers could lead to higher costs.
Track 1+ had a strong debut in 2018, with 55 ACOs entering this Innovation Center model. In contrast, only a few ACOs entered the MSSP’s current two-sided models — two in Track 2 and eight in Track 3. To generate better evidence on the level of risk that yields the best results for the ACO program, CMS should extend revenue-based risk to the MSSP program, and offer additional incentives for organizations that take on more risk.
Make the benchmark more predictable. The original ACO financial benchmarking methodology was a political compromise in Congress intended to move money from regions with high per-capita Medicare spending to regions with lower spending, while still rewarding efficiency. It has proven unsuccessful at both. It’s time to create a better measure of whether an ACO actually generates savings to the Medicare program compared to the alternatives.
One of the major hesitations that ACOs have about entering into two-sided risk is the complexity and unpredictability of the program’s current benchmarking methodology. Sophisticated statistical analysis by Harvard Medical School Department of Health Care Policy researchers has shown that the current benchmarks do not accurately share savings based on a given ACO’s activities because they do not account for local variations in cost trends. As a consequence, some ACOs generate “savings” against a benchmark that was not attributable to their actions, while other ACOs are told that they did not generate any savings, even as they have worked hard to improve patient outcomes and reduce hospital and emergency department utilization. Both scenarios sap provider confidence to take on two-sided risk, and reduce the program’s ability to reduce costs.
CMS introduced a regional benchmarking approach last year to account for regional trends, but the complex benchmark calculations conducted between the close of the performance year and the “final reconciliation” are not possible for ACOs to replicate. The complex regional benchmarking methods also inadvertently introduced a new problem that systematically disadvantages rural ACOs by including their population in the regional comparison group. This policy decision also exacerbated the confounding effects of the approach toward risk adjustment in the ACO program, wherein risk scores for continuously enrolled patients can be reduced, but not increased.
As a result, ACOs are faced with a no-win situation, penalized whether their risk scores decrease or increase. The Next Generation ACO program has also had shortcomings with respect to the benchmarking methodology, and many of its most promising features are rarely used.
As one recent study of independent ACOs observed, the lag between performance and evaluation, the “black box” of risk adjustment, and benchmarks that are perceived as constantly moving targets, all contribute to a reluctance to move ahead with two-sided risk. Pushing ACOs to take more risk, while creating a more predictable and equitable benchmark, can lead to greater savings to the taxpayer without encouraging further provider consolidation.
Improve ACOs by Linking to Medicare Advantage
The path forward for ACOs could involve continued regulatory tweaking of the benchmarks and experimentation across the five different ACO models at CMS. But further complexity itself can carry unintended consequences. As Administrator Verma recently commented, “We know that the complexity [emphasis added] of certain models might have encouraged consolidation within the health care system, leading to fewer choices for patients. But strengthening the health care system will require health care providers to compete for patients in a free and dynamic market.”
A radically simpler solution would be for CMS to move all ACO benchmarking toward a methodology based on Medicare Advantage (MA) rates. This approach could also be used to develop an improved version of the Next Generation ACO program that provides an on-ramp for smaller practices. To preserve the existing historical-to-regional transition of MSSP, the ACO’s benchmark could initially be set at their historical percentage of the MA rates in their area (120% or 80%, etc.) during one-sided risk, and then begin the transition toward the actual MA rates as soon as the ACO takes on two-sided risk.
The processes for establishing these rates are well understood, and the rates themselves are much more predictable. The rates are set prospectively, and do not require extensive analysis of cost trends months after the conclusion of the performance years. Improving the timeliness and predictability of benchmarks would greatly benefit ACOs at no loss to CMS; in fact, it would greatly reduce the cost and complexity of maintaining the ACO program for Medicare, since the MA program has already invested in the policy and analytic tools for solving many of the technical problems that ACO benchmarking faces.
Tying ACO benchmarking to MA rates would also have the advantage of giving risk-taking providers greater competence — and confidence — in taking on risk for Medicare Advantage patients, and partnering with plans to create more MA options for seniors.
Reward (and simplify) quality. Currently, ACO quality scores appear to be uncorrelated with savings against benchmarks. It is reassuring that the savings are not coming at the expense of patient care, and there is no evidence ACOs are stinting on needed health care. However, there is an opportunity to incentivize improved patient experience and quality outcomes in addition to savings, similar to the Medicare Advantage program. A simplifying approach aligned with the Patients Over Paperwork initiative would be for CMS to use identical clinical and utilization measures for the ACO programs and the MA STAR rating program, evolving both toward more meaningful outcomes and patient-reported measures over time. Such an approach would reward quality through increases in the benchmark, reduce administrative burden for CMS and providers, allow consumers to make informed choices between ACOs and Medicare Advantage, and provide an opportunity for making improvements in both.
Engage consumers. ACOs face limitations in using benefit design to align financial incentives with beneficiaries, as is done in MA plans. Individuals who have elected to remain in traditional Medicare cannot suffer any reduction in benefits for being attributed to an ACO, but they could receive positive incentives, such as the modest payments to beneficiaries with respect to qualifying primary care services provided for in the Bipartisan Budget Act of 2018. Providing greater flexibility for ACOs in engaging consumers would be welcome as long as it does not come at the cost of greater administrative burden, such as requiring each patient to fill out additional paperwork.
In conclusion, we predict that the current administration will put their own stamp on value-based payment models at Medicare. It is also a safe bet that any such changes will be criticized by some as a reversal of CMS’ leadership of the national movement toward value. But far from a retreat, these changes can move the nation toward improved patient outcomes and savings while emphasizing simplicity, greater accountability, support for independent practices and market competition, reduced administrative burden, and patient choice.
The authors would like to acknowledge Mark McClellan, MD, and Sean Cavanaugh, MPH, for their thoughtful review and feedback.