Medicare Payment and Reimbursement

Accounting for Non-Performance-Related Variation in Shared Savings Contracts

January 22, 2015 1:38 pm

Understanding—and adjusting for—the variety of reasons why PMPM costs rise and fall is a critical challenge that must be confronted if the industry is going to successfully adopt value-based contracts.

In theory, the fundamental purpose of any value-based payment model is rewarding the provision of high-quality, cost-effective care. In practice, however, factors unrelated to performance, such as patient attribution and product mix, often come into play, says Phil Kane, director in the network and competitive analytics group for Florida Blue, the Blue Cross Blue Shield brand in Florida.

As a result, accountable care organizations (ACOs) are sometimes rewarded—or fail to be rewarded—with shared savings incentives due to factors unrelated to the cost or quality of care provided to ACO members, Kane explained during a September 2014 HFMA webinar.

“One of the key takeaways for us is that small changes in medical expense performance can have a large impact on the financial incentive arrangement,” Kane says

Obtaining a reliable estimate of shared savings incentives is important to both providers and payers. Hospitals and health systems often use the incentives to fund improvement initiatives and recruit physicians. Payers, meanwhile, rely on projections to forecast cash flow and determine the amount of funds that should be set aside for the incentive pool. Faulty estimates that fall well below the projected mark can cause frustration and confusion, Kane says.

“The key is looking at the difference between the target and the actual results, and making adjustments to those particular calculations that reflect the true efficiencies of the care system itself,” Kane says.

How Variation Affects Incentives

The upside for providers participating in ACOs is the potential to share in cost savings that result from the cost-effective delivery of care. The “reward” comes in the form of incentives, which are often based on the savings derived from the difference between actual and targeted per-member-per-month (PMPM) medical expenses, Kane says.

For example, let’s say an ACO contract sets a baseline PMPM cost of $362, based on the previous 12 months of patient claims (see the exhibit below). The contract might also estimate that PMPM costs will increase by an estimated 5 percent over the following year, resulting in a target PMPM payment of $380.

If the actual PMPM expense turns out to be $375, the savings is $5 PMPM. For an ACO with 20,000 members, the incentive pool would amount to $1.2 million (20,000 x $5 x 12).

However, just slight variations in the PMPM cost over the course of a year can produce wide swings in the incentive pool—an activity Kane refers to as leveraging. A PMPM cost that swings from $375 to $376, then to $371, and again to $377 from one quarter to the next can mean the gain and loss of thousands of dollars in year-end incentives.

For instance, consider an ACO arrangement that calls for a 50/50 sharing of savings. If the total yearly savings averages $3 PMPM versus $5, projected incentives can slide from $600,000 ($1.2 million ÷ 2) to nearly half that amount ($375,000) by the end of the contractual year.

Reasons for Variation

That variation between targeted and actual PMPM costs can be due to several factors, including legitimate reasons that truly reflect the performance of the ACO, Kane says. For example, an ACO may achieve efficiency by shifting care to more appropriate settings, improving care coordination, and increasing the use of generic medications. The resulting metrics would show a reduction in costs resulting from a reduction in the number of admissions, outpatient visits, and brand-name medications, Kane says.

But other reasons for variation, while not reflective of provider performance, still affect incentives. Some of these non-performance-related reasons include the following:

Provider expansion/contraction. Adding or losing a physician practice during the contract period can affect the ACO’s risk level. A new pediatrics practice, for instance, would likely lower the overall PMPM cost, Kane says. Pediatric populations typically have a lower risk of illness and use fewer resources than an adult population.

If, for example, adding a pediatrics practice results in a 10 percent decline in morbidity, PMPM costs would reflect that decline. The cost savings, however, is not reflective of the efficiency of the ACO, but the demographics of the patient population.

Member expansion/contraction. The addition of a large group of individual members can also affect PMPM costs. For example, early claims experience from members who obtain insurance through the new health insurance marketplaces has shown a greater use of resources, Kane says. The greater need may not be due to higher acuity levels, but to a pent up demand for healthcare services, Kane says. These members may have previously avoided obtaining care due to a lack of insurance.

For example, if the average annual cost of an ACO with 20,000 members was $4,558, but 200 health exchange members are added with 54 percent higher costs ($7,000 per member), the total annual medical expense cost would rise from $91.2 million to $92.6 million, resulting in a 0.5 percent increase in the PMPM cost (from the targeted $380 to $382).

Again, the variation is not due to an ACO’s care management strategies, Kane points out, but to a change in patient population.

Product mix. The mix of insurance products can also produce variation. For example, if the ACO includes a large, self-funded employer group with 500 members, and the employer carves out (or does not cover) certain benefits, such as pharmacy, the PMPM costs of that patient population will be lower than the rest of the ACO—again, irrespective of the efficiency of the ACO, Kane says.

“This has occurred in our ACOs,” says Kane. “We’ve added a substantial number of members who obtained coverage under the Affordable Care Act. In these cases, the addition of these members increased the actual PMPM and is not a reflection of the ACO’s care management. If this comes up during discussions with providers, we indicate that risk-adjustment methods should account for the addition of these members.”

High-cost members. Just one member with a serious condition can cause significant variation. A candidate for a heart transplant who is attached to a device that helps his heart pump may incur costs as high as $500,000, significantly skewing the PMPM costs (and incentives) by considerable amounts, Kane says.

The accuracy of estimated trend increases (e.g., medical trend estimation) for PMPM costs can also cause variation between targeted and actual PMPM costs. If the trend used to project a target was inaccurately estimated, the ACO may end up performing worse (or better).

“For example, suppose an ACO uses a projected medical trend of 3 percent (based on past experience) to establish a target PMPM,” says Kane. “Then suppose there’s a bad flu season and medical trends actually increase at a 5 percent rate. The ACO would be penalized due to factors that are somewhat outside of its control.

Attribution. Additionally, PMPM cost levels can be affected by how members are assigned or attributed to the ACO. Ideally, there should be an established relationship between the patient and the primary care physician (PCP). For example, the patient chose the PCP. This scenario improves the likelihood that the PCP can effectively manage the patient’s care as the physician is familiar with the patient and his or her health history, etc.

However, attribution does not always work that way. “Some health plans have not yet developed product and benefit designs around ACO payment models, which require members to choose a PCP,” says HFMA’s Susan Horras, director, healthcare finance policy, health plan and population heath initiatives. “So consumers/members are not aware they have been attributed to an ACO or what an ACO even is. Therefore, accountability of the patient/member hasn’t been established, and leakage occurs that impacts the ACO care model from the provider perspective.”

In some methods of attribution, a member is automatically assigned to a PCP, perhaps by a computer-generated algorithm. In such a method, there may not be an established relationship between the physician and member, Kane says. Members may instead be assigned to physicians based on claims utilization.

“For an ACO to work and provide care and coordinate care, you really need to have a relationship between the patient and the primary care provider,” Kane says. “That’s the kind of the basis on which you want to provide quality care and really do the assignment in theory. The reality is that in some of the attribution models that I’ve seen, it’s an arbitrary decision.” Kane says.

Not Ready for Risk

Managing the inherent variability within an ACO is not yet a perfect science, although there are some steps that providers and payers can take.

See related article: Six Ways to Address Non-Performance-Related Variation in ACO Contracts

“Ultimately, this requires the payer and provider to work very closely together,” says Kane. “We’re trying to get a little more understanding around the ups and downs and how to jointly work through it.

Karen Wager is a freelance writer who regularly contributes to HFMA publications and Forums.

Quoted in this article:
Phil Kane is director in the network and competitive analytics group for Florida Blue, an independent licensee of the Blue Cross and Blue Shield Association, Jacksonville, Fla.

Susan Horras, director, healthcare finance policy, health plan and population heath initiatives, HFMA.

Discussion Starters

Forum members: What do you think? Please share your thoughts in the comments section below.

  • How can providers and payers work together to solve these variability issues around shared savings incentives? Is your organization involved in talks around these issues? What lessons learned have you identified to date?


googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text1' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text2' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text3' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text4' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text5' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text6' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text7' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-leaderboard' ); } );