Monday, April 27, 2009

The Trouble With “Shared Savings”

There is a lot of discussion today about using “shared savings” as an approach to healthcare payment reform. Medicare has used it as the key element of its Physician Group Practice Demonstration, and it has been proposed as the key mechanism for encouraging the creation of “accountable care organizations.”

The basic concept is fairly simple: if a healthcare system or provider reduces total healthcare spending for its patients below the level that the payer (Medicare or a private health insurance plan) would have otherwise expected, the provider is rewarded with a portion of the savings, i.e., the payer still spends less than it would have otherwise, and the provider gets more revenue.

Unfortunately, there are some fundamental weaknesses in the shared savings approach that make it far less desirable as a payment reform than it might first appear:

1. It’s P4P, Not Payment Reform. Shared Savings is just another form of pay-for-performance (P4P). It doesn’t actually change the current payment system at all – key primary care services that aren’t paid for today (like nurse care managers for chronic disease patients, phone and email consultations with physicians, etc.) still wouldn’t be paid for, services where fees are too low to cover costs would still lose money, etc. (Although shared savings proponents would argue that it does pay providers for things that are under- or non-compensated today, the problem is that the payment, if it comes, arrives long after the service is delivered, and may or may not be adequate to cover the cost of the services delivered). It’s not even clear if the reward through shared savings would be sufficient to offset the profit that the provider is losing by reducing the number of services it provides, particularly if only one payer is sharing savings but the provider changes its approach with all of its patients. Creating a P4P incentive based on total spending is a good idea, but only if it is coupled with changes in the underlying payment system.

2. It Gives Providers Risk Without Resources.At first glance, shared savings looks like the perfect deal for the healthcare provider – if the provider is successful in reducing total costs, it gets a bonus; if it’s not successful, it suffers no penalty. The flaw in the logic is assuming that what the provider needs to do to achieve success is costless. For example, although there are programs that have been demonstrated to reduce preventable hospitalizations, those programs require an increase in upfront spending by the primary care practice or the health system that implements them. Even if the shared savings payment would ultimately cover the provider’s costs, it has no way of knowing whether this will be the case, and there is a non-zero risk that even if it reduces costs, it will not meet the threshold established by the payer to declare “savings,” or it will end up spending more money on a net basis even after any shared savings are paid. Moreover, when multiple providers are involved, shared savings creates a variant of the “prisoner’s dilemma” – if one provider makes the investment to improve care but others don’t, the savings may not be sufficient to cover that provider’s costs; conversely, if most providers make the investment, any individual provider can increase their profit by sharing in the savings without making any upfront investment themselves.

3.  It Rewards Improvement Rather Than Performance. The communities and providers that have the most to gain from shared savings are the ones that are “wasting” the most resources today, through high rates of hospital admissions, use of unnecessary procedures, etc. In contrast, the communities with relatively low costs and high quality of care are already “saving” Medicare and other payers significant amounts of money but with no reward. The first group can improve relatively easily, since they have so much “low-hanging fruit” to pursue. The latter group, even if it can still improve further, may need to invest significantly more resources to do so, yet it will likely receive far less reward relative to the costs it incurs. In effect, shared savings exacerbates the current inequities in the payment system. The problem is even worse if the payer chooses to take a disproportionate share of the savings first, as Medicare has done in the Physician Group Performance Demonstration. Although this is intended to avoid rewarding providers for reductions in spending simply due to random variation, the practical effect is to increase the amount the provider has to spend or lose before receiving any reward through shared savings.

4.  It’s Inherently Arbitrary. How much of the savings should be shared? Merely picking a percentage between 0 and 100% provides no assurance to the provider that the costs they incur or the losses they would sustain will be offset by the amount shared. Although it would theoretically be possible to select a percentage that matches the amount of the savings to the cost of the specific investments that the providers plan to make, this would have to be done on a case-by-case basis, rather than on a uniform basis across the country.

5.  It’s Not a Sustainable Approach. Even in a place where there is the potential for significant savings, what happens after most of the savings is achieved?

• If shared savings is viewed as a one-time bonus payable in a year that savings is achieved, then the motivation it creates for providers to change disappears when there is little no additional bonus to be achieved in the future. Moreover, if there is no downside risk to the provider for increased costs, one-time shared savings creates the perverse incentive for the provider to again increase utilization of services in order to create a new pool of potential “savings” to go after in the future. (Similarly, it creates an incentive for low-spending regions to increase spending temporarily in order to lower it again and get a bonus payment they would not have otherwise received.)
• If the shared savings payment were not a one-time payment, but were to be paid repeatedly each year once achieved, there would need to be a mechanism for determining what kind of future growth is acceptable in a region that has achieved savings initially. For example, if a region reduces costs by 5% during the first savings calculation period, but then it experiences cost increases slightly higher than other regions in the second year, would its shared savings payment be reduced, and if so, by how much? What if the payer impose an even more stringent set of standards for spending in the future, which would reduce or eliminate repeated payments? From the provider’s perspective, it would need to understand in advance the requirements to achieve shared savings payments over an extended period of time in order to make investments that require a multi-year payback period.

These problems do not meant that shared savings is totally without merit, but it cannot effectively serve as the primary mechanism for healthcare payment reform.

 Comments (0)