ACO Payments May Hinge More on Geography than Quality

Location! Location! Location! When it comes to Accountable Care Organizations, the federal government is borrowing the adage from realtors that says it’s not so much the house as the neighborhood.

A study in the New England Journal of Medicine suggests that the shared-savings payments that participating providers receive in the Medicare Pioneer and Shared Savings ACO programs  might be more dependent upon geography than performance.

"There is a lot of variability in how favorable the payment methodology would be to organizations across the country because spending growth varies so much across different areas," J. Michael McWilliams, MD, a co-author of the study, tells HealthLeaders Media.

"Also we found it is likely these losses or gains would be unrelated to any organizational effort to improve quality of care and lower costs," says McWilliams, a general internist and assistant professor of healthcare policy and medicine at Harvard Medical School.

"It will probably even out over a few years for any given organizations but for half of them it will persist over time and it presents a pretty large gamble to perspective ACOs."

"If you happen to be in an area with high spending growth, your spending target will be well below that," he says. "And even if you do a good job, you might not quite get there. You might lose money or at the very least you might not get the shared savings that reflect your true performance relative to what spending would have been."

The federal government is structuring the payments within these 306 hospital referral regions to address geographic variations now seen in Medicare spending growth.

"There was some discussion in those regulations about using a national growth factor to help compress the geographic variation in Medicare spending," McWilliams says. "So the idea would be in a high-spending growth area using a mean national growth rate to set the spending target would necessarily bring that spending down in the high-spending growth area and vice-versa for a low-spending area. It would bring it up."

Unfortunately, McWilliams says, HRRs can’t draw a correlation between spending growth levels—based on a single year of data or an average across years—and spending growth rates—the percentage change across more than one year.

"There are such things as low-spending level high-spending growth areas, and high-spending level low-spending growth areas. In fact they make up about half of hospital referral regions," he says. "So using a national growth factor would not seem to accomplish the stated rationale in the regulations of compressing geographic variations in spending."

Such a structure makes it difficult for healthcare organizations within specific HRRs to estimate whether the use of a national growth factor to set a spending target would benefit them or not.

"That is because spending growth in one time period does not seem to be correlated at all with spending growth in another time period for a given area," McWilliams says. "For an organization deciding whether or not to participate in these ACOs, it would be unclear to them whether or not it would pay off."

McWilliams and study coauthor Zirui Song suggest instead that the federal government rely more on local growth factors to gauge spending targets. "There would be less of this problem of gains or losses for the organizations that are unrelated to their efforts to lower costs and improve quality," McWilliams says.

McWilliams says the study did not directly examine the affect of the HRRs on rural ACOs, but he identified some vulnerability.

"The variation in spending growth had two components. One was a random component that’s due to changes in healthcare needs of Medicare beneficiaries from year to year," he says. "The other is a more systematic component and the result of market forces that drive spending growth in particular areas. For example, if provider organizations are competing on the basis of attracting the latest and greatest technology to attract patients, spending growth ought to be uniform in the area."

"I would not expect the systematic component to be any more variable in rural versus urban areas," McWilliams says. "But clearly the random component would be because rural areas are more sparsely populated. A prospective ACO in a rural area would probably face a bigger gamble in terms of a one-year loss or gain that is due to the payment methodology than would an ACO in a more densely populated area."

"Again this is all the consequence of using national growth factors to set the spending targets for the ACOs, which is essentially mean spending growth in the nation," he says. "There is going to be a distribution across areas in local spending growth and rural areas will tend to be further out to either end of that distribution than the densely populated areas."

In other words, rural ACOs could be big winners or losers, or break even based largely upon where they are. But they won’t really be able to predict it either way.

"That is one of the concerns of our findings," McWilliams says. "If the current payment methodology is presenting a sizeable gamble to these organizations then some may decide not to participate, which, if the program turns out to be successful, would be a real shame."

Any payment scheme that does not necessarily reflect the work—good, bad, or mediocre—of a particular healthcare organization could put the Medicare ACO experiment in jeopardy.

"If it is determined that some of the losses and gains have happened not because of something the organization did, that would raise a red flag that might limit expansion of the programs," McWilliams says. "Some of the organizations that participate and experience big losses because of this could leave the program and that may cause other organizations to be more reticent to join."