Medicare's Accountable Care Organization (ACO) program is the Centers for Medicare and Medicaid Services' (CMS) flagship population-based payment model. In the ACO program, groups of providers form ACOs and take accountability for the spending and quality of care for the Medicare beneficiaries they serve. The ACO is given a spending target (benchmark) and receives a bonus (i.e. gets to share savings) if actual spending is below the target. In some ACO programs, the ACO must return money to Medicare if spending exceeds the target.
The release of the Office of the Inspector General's (OIG) report on the savings resulting from the Medicare Shared Savings Program's (MSSP) ACOs has rekindled the debate about the success of ACOs and their viability as a means to curb health care spending growth. This is an important debate, but clarity requires addressing crucial conceptual and semantic questions about the meaning of "savings."
In a previous post, we discussed the analytic issues about savings (specifically issues related to comparison groups). Our main point was that savings must be judged against a reasonable estimate of what the spending on ACO-attributed beneficiaries would have been in the absence of the ACO program. In other words, actual spending must be compared against an appropriate counterfactual level of spending. While CMS's benchmarks are useful for setting spending targets, as elaborated below they do not always represent the appropriate counterfactual against which to judge ACO savings (nor should they). Savings computed based on comparisons of spending versus benchmarks may thus give misleading estimates of the success of the MSSP, or of ACOs in general.
However, even if there is agreement about the counterfactual, there is continued confusion about the meaning of the term "savings" in the context of the ACO program. The complexity of the payment structure in the ACO program means that there are a few ways one could choose to measure savings. The semantic questions here are surprisingly challenging and variation in how terms are used can lead to considerable confusion. Below we attempt to clarify the meanings of "savings" and draw out the implications for the program.
Consider three types of "savings."
- Utilization Savings: The difference between Medicare Fee-for-service (FFS) spending on ACO beneficiaries (without counting the bonus) and the Medicare FFS spending on those beneficiaries had there been no ACO. This is driven by changes in utilization associated with the ACO.
- Payer Savings: The difference between Medicare program spending with, versus without, the ACO (including bonuses paid and other program fiscal effects, including administrative costs).
- Societal Savings: The savings achieved because fewer services are delivered and offset by payer and provider administrative costs associated with the ACO program, but ignoring changes in prices or distribution of savings and losses.
Utilization Savings
Utilization savings is the most common measure of savings in the literature. It reflects changes in spending due to changes in the ACO providers' pattern of health service provision. Commonly, measures of utilization savings are limited to ACO-attributed patients (i.e. those patients who are part of the population accounted for in the initial budget-setting process). Utilization savings is an important concept because it captures providers' response to the program's incentives and speaks to the ability of new payment models to change underlying utilization, which is likely needed if we are to create a fiscally sustainable health care system. However, utilization savings do not measure savings from a payer perspective. Conceptually, it is these utilization "savings" that get "shared" in a "shared savings" program.
Payer Savings
Payer savings refers to the budget impact on payers. Payer savings differ from utilization savings for a number of reasons. Most importantly, by definition the payer will not capture all of the utilization savings because they are shared between the payer and ACO. As a result, the payer savings will tend to be lower than utilization savings. Semantically, the idea is that providers create savings and then the payers share them.
However, other aspects of the program impact payers' savings apart from the utilization savings less bonuses. For example, if the benchmarks reflect any discounts off standard fee-for-service payments then payers may 'save' even if the providers do not create any efficiencies in practice patterns. (Payers also can lose if they set the budget targets too high.) For example, in the NextGen ACO model, the benchmark is deliberately set about 3 percent below what CMS expects would truly be spent caring for ACO-attributed patients if there were no ACO in place. As a result, if CMS's implied counterfactual is correct (which may not be the case), Medicare would save 3 percent even if there are no behavior changes.
This aspect of the program is equivalent to a price cut. These would be real savings from the payer perspective (again assuming CMS's counterfactual is correct), but the implications of program savings driven by a price cut are different from those related to behavior changes. Similar issues arise in the MSSP program, where benchmarks are now set to converge to regional averages.
A truly thorough computation of payer savings should include a few other considerations in addition to the calculation mentioned above. In ACO programs there are spillovers that result from changes in provider behavior in response to the incentives: When ACOs change their practice patterns for ACO patients, they likely change it for other patients as well. For example, some Medicare patients may not be attributed to an ACO but seek care from ACO providers. If those providers deliver more conservative care because of the ACO, they may deliver more conservative care to non-ACO patients as well.
In the case of Medicare, there is an additional type of spillover: Utilization savings in the ACO program reduce benchmarks in the Medicare Advantage program. Those savings should also be incorporated in estimates of payer savings. Finally, payer savings should also reflect the administrative costs borne by the payers to operate the program.
Societal Savings
Societal savings reflect lower costs (not the price) achieved by reducing utilization (including via spillovers), minus the costs of payers and providers operating the program — for example, providers may hire case managers and payers may hire staff to manage the data requirements associated with the ACO program. Societal savings do not reflect price changes or the distribution of savings or losses because those are transfers between payers and providers.
Interpreting This Framework
None of the preceding concepts of savings is inherently better than the others. Rather, each has certain uses. Utilization savings is the narrowest concept and captures the underlying behavior change. It corresponds most closely to the narrative about how alternative payment models are supposed to reduce utilization by incenting providers to be more efficient. Payer savings most closely corresponds to the concept budget analysts must use, but it may not reflect the long-term potential of the program because it does not directly measure whether the payment reform caused providers to change behavior. Societal savings captures the broadest measure of savings and would most closely correspond to the cost concept recommended for economic evaluations of clinical services, cost effectiveness. However, this may not be the relevant savings concept for many stakeholders and is the hardest to measure.
The evidence base for any of these savings concepts is still emerging. In general, the utilization savings from ACOs are initially small, (though the rise a bit over time). Early evidence shows that the longer beneficiaries are attributed to ACOs, the more savings they generate. In 2014, the per-beneficiary savings in the 2012 MSSP cohort grew to 2.6 percent, while the 2013 cohort had only managed to save 0.7 percent, and the 2014 cohort had yet to achieve any savings. Savings were greater for ACOs composed of independent physician groups and in ACOs with higher initial spending for their region; savings were disproportionally from post-acute care and shifts away from high-priced hospital outpatient departments. In Pioneer ACOs, some savings was also derived from reductions in use of low-value care, but this was not observed in the MSSP program.
All of this is encouraging in direction, but the magnitudes are disappointing. The trends in the commercial context are similar, though the levels are higher. For example, the savings from Blue Cross Blue Shield of Massachusetts ACO program—the Alternative Quality Contract (AQC)—for the first cohort rose from 2.4 percent in 2009 to 13.2 percent by 2012. The bigger impact partially reflects the ability of commercial ACOs to save by shifting care to lower-priced providers.
In Medicaid, Oregon has experimented with a variation of ACOs, Coordinated Care Organizations (CCO). According to one assessment, utilization savings from CCOs was about 7 percent per Medicaid enrollee. However, not all types of utilization were measured and certain sub-populations were excluded.
Analysis of payer savings is more challenging because of data issues and difficulty in measuring concepts like spillovers. Early savings under the AQC were completely offset by bonus payments, but the savings eventually grew beyond payments. In Medicare's Pioneer program, savings were $42 million in the first year after bonus payments and recoupments. In the MSSP, bonus payments initially exceeded the direct savings because CMS could not recoup losses from poor-performing ACOs, but savings grew to exceed bonuses in their second year. One estimate, accounting for spillovers and impact on Medicare Advantage suggests the Medicare savings from the MSSP were about 1.6 percent per ACO beneficiary. This does not take into account Medicare program costs, which would offset savings to some extent but could likely be streamlined.
The literature does not contain rigorous estimates of societal savings. There are reasons to believe that if any exist, they are small because the savings from reduced use measured by prices are higher than they would be if measured by the avoided cost of delivering the forgone services. But adding in spillovers would improve the financial profile. This would need to be weighed against payer and provider administrative costs, which have been estimated anecdotally to be relatively high. For example, MedPAC evidence suggests that provider program costs run at 1-2 percent of spending.
All of this would suggest little, if any, societal savings and maybe losses. However, as Muhlestein, Saunders, and McClellan document, there has been steady growth in ACO participation, suggesting at least some organizations are doing better than it seems or expect they can do even better. The jury is still out.
What Are The Implications For The ACO Program?
Given these modest results to date, some may ask why we should continue the ACO experiment. There are three compelling reasons. First, ACOs allow providers to benefit financially because ACOs generate efficiencies. As money gets tight—and it will get tight—providers will seek greater control over the dollars. With Medicare Access and CHIP Reauthorization Act calling for greater than 10 percent reductions in inflation-adjusted physician fees over the next decade and facilities fees under pressure from the Affordable Care Act productivity adjustments, ACOs give providers a mechanism to withstand fee cuts while providing better quality care. Participation is voluntary; those that want to participate in alternative payment should have a mechanism to do so. Second, from the Medicare perspective, ACOs seem to generate some savings, with a potential for the savings to grow over time. To the extent that payers can develop successful payment models, they can satisfy the demands of the ultimate payers (taxpayers, employers, and beneficiaries) for a better system. Finally, beneficiaries benefit from lower premiums and improved quality. If the alternative strategy for cost containment is higher cost sharing at the point of service and the shifting of risk on to patients (which it likely is[1]), then the ACO model looks even better from the beneficiary perspective. ACO programs, if they can succeed, may offer a way to slow the march towards greater patient risk.
However, a move towards population-based payment will not be easy or without challenges. Not all providers will succeed and some providers will do worse financially than they otherwise would. But this is true for any strategy to reduce spending. At least ACO savings are likely partially derived from waste reductions, and some of the savings flows back to the delivery system.
Payers must confront challenges with designing ACOs and managing technical details such as benchmark setting, risk adjustment and quality measurement. However, ACOs must be compared to the alternative, which is likely very low fee increases and higher patient cost sharing. Other countries have managed quite well with a FFS system and lower fees, but it unclear if the U.S. would want such a system (and the associated challenges). Even if we would, it is unclear how to get there.
A voluntary ACO program seems to offer a conceptually appealing way to give providers control over spending. Given that early results show both utilization savings and payer savings—albeit modest, but growing—working to improve the model seems wise.
[1] For example, the average annual deductible for a single individual rose from $1,097 in 2012 to $1,505 in 2017.
from Health Affairs BlogHealth Affairs Blog http://ift.tt/2yRxtq8
No comments:
Post a Comment