Some of the most promising strategies for controlling spending and improving the quality of care delivered in the United States are payment reforms that aim to give health care providers an incentive to improve value. Health care providers are often in the best position to identify ways to reduce waste and help their patients chose the most efficient sites and types of care. Giving health care providers a financial stake in driving value can be much more effective and palatable than runaway health care spending, pushing the risk onto patients, or subjecting them to one-size-fits-all insurer rules.
There are several types of payment reforms. Some approaches target total population spending, such as Accountable Care Organizations. These models typically provide incentives for physician groups or delivery systems to reduce per-capita spending and improve quality. The savings are generally shared with the organization that employs the primary care physician. Other payment models focus on episodes (bundles) of care, creating incentives for providers to limit spending during the episode while achieving quality benchmarks. The savings typically accrue to the organization that controls the hospital or specialist responsible for the episode. Medicare is currently experimenting with both approaches.
In this issue of JAMA Internal Medicine, Navathe et al1 study the effect of episode payment on lower extremity joint replacement in a single hospital system. Their findings are striking: After approximately 5 years under 2 different bundled payment programs for these procedures, spending at the Baptist Health System was about 20% lower. Much of that stems from savings on postacute care, suggesting the importance of whether postacute care is included in the bundle. The changes they document are much larger than most of those seen in other studies of similar bundles. For example, an earlier study2 examining all participants in 1 of the 2 bundled payment demonstrations studied by Navathe et al at Baptist Health System found average savings of about 4%.2 This could reflect differences in the duration of the episode (shorter in the study by Navathe and colleagues), experience with episode payment (greater in the study by Navathe and colleagues), research methodology (Navathe and colleagues do not formally incorporate an external control group), or variation across program participants (Navathe and colleagues examine 1 system). While the results of the study by Navathe et al are promising, further research will be needed to assess how well this comparison of spending before vs after the reform captures the causal effect of the payment reform and how broadly these results would generalize to other hospital systems. The headline results may not capture all of the other dimensions along which providers may respond. For example, there is some evidence from the study by Dummit et al2 that health care providers paid through episode models select healthier patients: the number of patients staying at a skilled nursing facility or using home health care before joint replacement decreased after an episode payment model was introduced. Moreover, Dummit et al reported that the number of lower extremity joint replacement episodes per hospital increased enough to offset savings per episode.2,3 Navathe et al also report large increases in volume. Some of this may reflect broad trends for greater use or shifts in care toward the participating facilities. Much more work is needed to assess how health status changes affect calculated savings and the extent to which volume increases offset per episode savings.
More broadly, the effectiveness of these alternative payment models in improving quality and lowering spending hinges on design and implementation choices. How the benchmarks against which spending is evaluated are set and updated is crucial to generating the right targets and attracting the right participants. The share of savings providers get to keep (including upside vs downside risk) is a key determinant of the strength of the incentives to reduce resource use. The scope of services covered (such as whether postacute care is included) and the range of conditions covered affect not only the incentives to save, but the magnitude of the potential system-level savings.
So which is more promising, episode-based or population-based payment reforms? Either could be better than the fee-for-service system that dominates Medicare now—particularly with broad scope and real financial stakes—but both seem likely to generate only modest savings in their current incarnations. The greater share of spending potentially covered by population-based payments suggests that, absent broader reach of episode-based models, population-based approaches might eventually have a bigger impact system-wide, although savings to date have been modest; estimates suggest that ACOs cover about 25% of Medicare Parts A and B spending and generate 1% to 2% savings (potentially rising over time).4 Even if episode-based models result in somewhat higher savings for covered spending, as currently constituted they are likely to cover a smaller fraction of spending than population-based payments could. Moreover, savings to date have been driven by a small subset of episode types, suggesting expansions of the program may yield even lower returns, further eroded by any increase in the number of episodes.5,6
It is also important to note that the Medicare program does not capture all of the savings in either model—that is the “shared” part of shared savings. The population-based saving programs share a large portion of savings with health care providers. Over time, savings to Medicare would grow if benchmarks rose more slowly than they otherwise would. In the episode-based models, benchmarks are set a few percent below estimated spending, guaranteeing that Medicare will reap some savings, but any greater savings go entirely to health care providers.
The good news is that neither type of reform seems to lower the quality of care thus far and both have the potential to be dialed up to increase savings to the Medicare program and overall.5,7 Ideally, both could be deployed in concert, designed so that the strengths of each complement the weaknesses of the other. But there are concerns that they compete to capture savings, and there are currently so many different options and demonstrations in play that the effectiveness of each may be dulled by the discordant incentives and systems with which health care providers must grapple. The existence of so many competing options—even leaving aside other Medicare programs like the Merit-based Incentive Payment System, Medicare Advantage, and Medicaid—may well undermine the potential for payment reform to drive health care providers toward real delivery-system improvements.
Corresponding Author: Katherine Baicker, PhD, Harvard Chan School of Public Health, Harvard University, 677 Huntington Ave, Boston, MA 02115 (email@example.com).
Published Online: January 3, 2017. doi:10.1001/jamainternmed.2016.8280
Conflict of Interest Disclosures: Dr Baicker serves on the Board of Directors of Eli Lilly and has speaking and consulting arrangements with health care provider groups and insurers. Dr Chernew is an advisor to Archway Health, is a partner in VBIDHealth, and has speaking and consulting arrangements with insurers, health care providers, pharmaceutical, and related companies.
Additional Information: The authors gratefully acknowledge expert research assistance from Christopher Barbey, MA. He was not compensated beyond his salary as a research assistant.
Baicker K, Chernew ME. Alternative Alternative Payment Models. JAMA Intern Med. Published online January 03, 2017. doi:10.1001/jamainternmed.2016.8280