Published Online: December 12, 2013
Jon Christianson, PhD
There is no question that the number of accountable care organizations (ACOs) in Medicare and total cost of care contracts (TCOCs) in the private sector is growing, along with the amount of care provided under these contracts.1 Common characteristicsof both types of contracts are that a provider group agrees to care for an attributed population of patients under a fixed budget, with the potential to benefit financially through “shared savings” and by meeting quality goals.2 The hope is that paying providers in this way will improve quality of care and population health, promote efficiency in care delivery, and ultimately “bend the cost curve.”3 While all these are commendable goals, the greatest emphasis appears to be on the potential for ACOs and TCOCs to reduce the rate of growth in per capita healthcare costs, without negatively affecting quality.
In some ways, we have been down this path before. In the 1980s, when capitated contracts first came into vogue, it was argued that health plan contracts that transferred financial risk to providers would curtail provision of unnecessary services, ultimately reducing costs to payers or at least the rate of growth in their costs. There were concerns that changes in provider behavior under capitated contracts could harm quality of care, but many health plans attempted to guard against this possibility by placing quality-related bonuses in contracts, specifying upper bounds on provider gains and losses, and analyzing claims data to detect trends that might indicate inappropriate care.4 What is different now? One important difference is that consumers do not enroll in ACOs or with provider groups being paid under TCOC contracts. Instead they are attributed to these entities based on their past history of service use and can continue to seek care from the providers of their choice. Presumably this will alleviate consumer concerns about health plan influence on provider decisions. At the same time, better data and quality measures combined with greater provider performance transparency will make it easier for health plans and consumers to identify providers who deliver poor quality care and will shame providers into improving their performance. It is clearly too early to tell if ACOs and TCOCs will improve provider efficiency without also reducing quality, although early findings from analyses of a TCOC contract between the Blue Cross and Blue Shield of Massachusetts and providers in that state certainly are promising.5
To date, most of the analytic attention being given to ACO and TCOC contract designs has focused on how best to encourage providers to become more efficient. This attention is certainly warranted, as there are many “moving parts” that must be in sync if contract goals are to be achieved. For instance, if attribution algorithms inappropriately assign patients to providers, incentives for efficiency are attenuated. The same is true if methods used to “risk adjust” attributed populations of patients are not adequate to avoid penalizing providers who care for sicker patients. And, there has been great debate in the design of Medicare ACOs regarding risk-sharing specifications. These decisions are complicated by the fact that provider groups vary in their experience in managing panels of patients and assuming financial risk. For instance, under private sector TCOC contracts some larger provider groups with experience managing financial risk do not see the need for “shared gain/shared loss” provisions. Instead, they prefer to adjust their degree of risk exposure through the purchase of reinsurance.
How these issues relating to contract design are resolved will have important impacts on the ability of ACO and TCOC contracting approaches to reach their goals. However, an issue that could prove more important, but has received less attention, is the setting of global budgets initially and their subsequent adjustment over time. Here it is important to distinguish between costs to providers and costs to payers. The incentives that these contracts create for providers could lead to reductions in their costs of providing care. Under typical ACO or TCOC contracts, payers can capture at least a portion of the provider cost savings (should they occur) through a combination of shared savings as specified in yearly contracts, but also—and likely more important and controversial—through adjustment of budget targets over time. There are technical aspects to the setting of budgets, of course, but for ACO contracts political clout also will play a role, and for TCOC contracts market leverage will come into play. In both cases, experience suggests that, even if efficiencies are obtained without harming quality, a favorable impact on longrun cost trends for payers and, ultimately, consumers is far from certain.
There are 3 general approaches that can be taken in setting budgets and, perhaps more importantly, year-to-year increases in budgets. Using the language of health service researchers, the problem is finding an appropriate “comparison group.” Some argue that the ideal comparison group would yield an exact picture of what the performance of the provider group under contract would have been if it had not signed the contract. Of course, there is no such ideal comparison group, so payers are forced to settle for proxies. In practice, there are 2 options. First, payers can set initial budgets (payments) based on a similar group of providers, tracking the performance of this group over time and adjusting budgets of contracting groups based on the rate of change in the comparison group. Because comparison groups are never perfect, some risk adjustment is required. This is essentially the approach that Medicare took in the early days of the current Medicare Advantage program, using performance in fee-for-service Medicare to set rates and make rate adjustments for contracting health plans, and in its “physician group practice demonstration.”6 There are all sorts of drawbacks to this approach, including continual disputes over the choice of the comparison group (eg, with Medicare health plans, should the fee-for-service comparison group be drawn from the same county, the same region, or nationally?), and the fact that the comparison group, and the environment in which it operates, can change in ways that make its use less defensible over time. In some cases,the comparison group can simply disappear. For example, when some Medicaid programs began contracting with health plans on an experimental basis to serve beneficiaries, they used the experience of Medicaid beneficiaries not in these plans to guide changes in contract rates. However, as they enrolled more beneficiaries in plans, the number in comparison groups shrunk. Beneficiaries in these groups became less representative of the beneficiary population in Medicaid health plans and, in some cases, too small for reliable estimates. Some health plans have used external comparison groups to inform the setting of budgets in their private TCOC contracts, but their exact methods are not clear.7
The second approach, a version of which was adopted in CMS’ ACO program, is to use patients served by the provider group under contract to set the budget.8 In a sense, the group is its own comparison. Historical provider expenditures for the group of patients are calculated; 2 or more years generally are included, with more weight typically given to more recent years. This historical yearly expense then is trended forward to the contract year. The choice of the trend rate is critical,9 and it is typical to choose a trend rate that is tied to an externally calculated number, such as regional healthcare expenditures, or the trend in expenditures for health plan or public program participants not covered by the contract. Usually, in the private sector, the trend rate is negotiated as part of the contract, as it can incorporate the mutually agreed upon goals of payers and providers. As such, it has an aspirational element, while also reflecting the relative negotiating leverage of the parties involved.
A third approach relies on payers to create incentives that encourage providers to “reveal” the dollar amount that they think is adequate to provide necessary care to an attributed population. his typically has involved some sort of competitive bidding process. Medicare currently uses this approach in setting rates for medical equipment and supplies and, in the past, used its demonstration authority to test competitive bidding to set payment rates for contracting health plans.10 Medicaid programs also have used competitive bidding approaches to set rates when contracting with health plans.
PDF is available on the last page.