Much has been written about the potential for Medicare payment reform to improve both efficiency and quality of care, including a recent post in the JAMA Forum by David Cutler, PhD. Given that Medicare is the largest single entity paying for US health care services, accounting for about a quarter of all hospital and physician payments, there’s no question that the program can be a powerful force for payment reform for good or ill. But I have concerns about Medicare’s ability to advance payment reform without legislative changes.
Successful Medicare payment reform could transform health care delivery, but missteps could set back meaningful change for the foreseeable future. Clinicians, hospitals, and other providers need Medicare, Medicaid, and private insurers all pulling in the same direction so that they avoid a situation (beyond a transition period) in which a large proportion of their patients is under a fee-for-service (FFS) payment system and a large share under reformed payment approaches—leaving them, figuratively, as many have pointed out, standing with each leg in a separate boat. So payment reform will not succeed without success by Medicare.
Through authority and funds in the Affordable Care Act (ACA), Medicare has pursued a variety of new payment models vigorously, especially accountable care organizations (ACOs). The result has been hundreds of contracts with organizations of physicians or physicians and hospitals using these approaches and a rising percentage of payment made through these models. Indeed, the US Department of Health and Human Services (HHS) Secretary Sylvia Mathews Burwell recently indicated her goal of having half of Medicare payments based on non-FFS models by the end of 2018.
But all is not well. Some ACOs have identified serious problems in algorithms that attribute the use of services by a beneficiary to a specific ACO and pointed out barriers to engaging beneficiaries (for example, lack of incentives to choose clinicians and facilities linked to the ACO). There are also difficulties in quality measurement, especially in how the spending benchmarks are set. Unless ACOs and other organizations continue volunteering to contract with Medicare using these models, expanding the reach of reformed payment will slow or stop.
Although these organizations are happy to share in savings, they have been less willing to risk sharing losses—what’s known as 2-sided risk. I suspect this reluctance reflects lack of confidence in the payment models being offered more than limited capacity to bear risk. If they can sign on to share only gains, many are willing to put aside concerns and enter contracts to gain experience with approaches that might become the norm in the future. Indeed, I and others have argued that the ACO model could be greatly improved by attention to the issues mentioned above and by adding a network dimension, so that beneficiaries pay less when they use clinicians, hospitals, or other facilities that are part of the ACO. This has the potential to further engage beneficiaries as well as clinicians and hospitals, but likely would require legislation.
But the most serious and daunting problems in the transition to reformed payment concern setting targets for spending per beneficiary or per episode, known as benchmarks. Ideally, benchmarks should reflect experience with FFS spending regionally or nationally rather than historical spending of clinicians and providers. Each entity’s spending would then be compared with the benchmark and savings or losses shared. But such an approach is impossible when participation is voluntary, because only those whose costs are already low would take part, and large sums of money would be spent on rewarding them for continuing this performance.
Currently, benchmarks reflect provider-specific spending, with organizations being rewarded for improved performance but not for good performance—an approach that is suitable for a period of experimentation but not as a long-term strategy. Benchmarks will need to be updated, and the options for doing this are not attractive. Updated benchmarks can be based on more recent spending data, but this would undermine the business case for investing in improved delivery because gains from those investments would not continue to accrue to the ACO or other organization. Or the original benchmarks can be updated by a market-wide or national trend. Although this would maintain the business case for investing in improved delivery, it would perpetuate the inequities of basing each organization’s payment on their historical spending at a moment in time.
The only long-term resolution of the ACO benchmark issue is to make clinician and provider participation mandatory or make opting out and remaining under FFS much more onerous. Previous Medicare payment reforms have been mandatory. The hospital inpatient prospective payment system, which replaced cost-based reimbursement in 1983, applied to all hospitals from the beginning. Over 4 years, each hospital transitioned from payment based on its historical costs to national per-case rates adjusted for local wage rates and teaching activity. Imagine what would have happened if each hospital could have decided whether or not to participate, and instead of a national benchmark, each hospital had its own benchmark that was periodically updated on basis of its actual spending.
Many believe that Medicare could proceed with bundled payment on a mandatory basis by developing bundles for different types of episodes (for example, an oncology chemotherapy bundle would be quite different from a hip replacement bundle). This would allow benchmarks based on national or community experience in FFS. If clinicians and providers, such as nursing homes, succeeded in reducing spending per episode, this could lead to higher margins for them and savings for Medicare.
A softer approach could be used, along the lines of the legislation proposed to fix the sustainable growth rate mechanism. Under this legislation, physicians with a substantial portion of revenues through “alternative payment mechanisms” would receive higher Medicare payment rates. This would allow benchmarks to be based partially on national or community experience and lead physicians to press for more viable alternative payment models. Although not in the legislation, payment differentials could also apply to hospitals and other health care providers, further accelerating the movement to alternative payment and pressure for better payment models.
Medicare’s recent announcement of a “next generation” ACO model moves in the right direction on some of the issues I’ve mentioned here and at least signals the HHS’s awareness of problems. However, the model appears to be limited for now to a small number of the most advanced delivery systems.
The most critically needed changes—especially related to beneficiary engagement and clinician and hospital participation—will require legislation. Only when Congress allows the type of steps outlined above to engage beneficiaries directly—rather than hide ACOs and other alternative payment models from them—will Medicare have a real chance to meet Secretary Burwell’s ambitious payment reform goals.