The departure of seven accountable care organizations from CMS' Next Generation ACO Model program will likely not threaten the viability of the program, but the reasons behind the early exit are costing providers millions.
The Accountable Care Coalition of Chesapeake, Allina Health, Fairview Health Services, KentuckyOne Health, Lifeprint ACO, MemorialCare and Sharp HealthCare have left the program for 2018 although they will likely still be on the hook for any losses from the 2017 performance year. There are now 51 ACOs that participate in the Next Generation model.
Four of the seven ACOs indicated that they decided to withdraw because changes made to the model's design, including the chosen risk adjustment, would directly hurt their ability to make money in the program.
"Even as one of the top performers in utilization nationwide, we do not believe we can be successful achieving the program's goals with the current design," Dr. Mark Schafer, CEO of MemorialCare Medical Foundation, Fountain Valley, Calif., said in a statement.
The biggest change to the model's design was the CMS' decision to lower the average risk score for 2017 by 4.82%, effectively making it more difficult to save money and earn a bonus or avoid a penalty.
In a letter to Next Generation ACOs, the CMS explained that the change was made to "account for a significant increase in coding intensity that otherwise threatens the financial sustainability of the Next Generation ACO Model." The CMS was likely referring to the impact that implementing electronic health records and ICD-10 in 2015 had on coding practices.
At least one provider is considering legal action. Officials for Sharp HealthCare, San Diego, believe that their risk scores have intensified because their Medicare fee-for-service population has actually grown sicker, not because their coding has improved, said Alison Fleury, senior vice president of business development.
Based on preliminary data, Sharp has calculated that because of the risk score change, it will experience financial losses under the program for 2017 even though Sharp previously expected to have a financial gain. Fleury said she expects Sharp to know in August. what it officially owes the CMS for 2017.
She said Sharp is looking into "what our legal options are," arguing the CMS didn't give the customary 30-day notice before they made the change. "This was not the program we agreed to; they unilaterally adjusted the program through that amendment." Sharp says that its 2017 Medicare population is on average older than its 2014 Medicare population, "which tells me these are actually sicker beneficiaries," Fleury said.
Sharp's Medicare fee-for-service population is likely riskier because in San Diego, where the integrated health system is based, there is a strong Medicare Advantage market so younger—and presumably healthier—beneficiaries are signing up for those plans instead, Fleury said.
She added that Sharp has been using EHRs for "many years" and hasn't changed its coding system. The CMS told Sharp if it didn't lower the risk scores, the program would no longer be financially sustainable and would have to be terminated.
The risk scores are used in ACO arrangements to determine the expected clinical costs of beneficiaries. Under the Next Generation Model, risk scores for a performance year are determined using beneficiary costs in an ACO's population from the prior year. A baseline risk score from 2014 is also used to control how high or low the risk score can fluctuate each year. Ideally this setup should allow ACOs to predict what their shared savings or losses will be under the program before performance years officially begin.
Experts say the program itself shouldn't be affected by the departures. Travis Broome, vice president for policy at Aledade, a consulting firm, said he doesn't believe that the long term viability of the program will be threatened by the loss of those ACOs. There are still more than 50 Next Generation ACOs participating, he said, and, furthermore, the program is due for a natural revamp in 2020 as that will be its sixth year.
David Muhlestein, chief research officer at consulting firm Leavitt Partners who first noted the departure of the seven Next Generation ACOs last week, said that the Next Gen model is already a demonstration program that's designed to end. So while it's a challenge to have ACOs leaving because they don't believe it works for them, CMS officials will learn from that, he said.
The CMS announced the risk score change on Dec. 7 and ACOs had until the first week of March to decide if they wanted to stick with the model, although they would still be on the hook financially for 2017.
Hospitals and physicians have gotten better at coding patients for risk factors like various chronic conditions, which has likely made the Medicare population appear sicker now than they were before the rollout of ICD-10 and EHRs.
"We saw a big jump in 2016 in Medicare beneficiary risk scores overall that coincided with the implementation of ICD-10," Broome said. To offset that, what it described as a "coding intensity," the CMS lowered the risk score for the Next Generation ACO Model.
Broome said he thinks the CMS should allow risk scores to be unique to each ACO and their population so it's a more predictable model. He said a common reason for ACOs to leave the Next Generation program, which often happens in enrollment periods, is because of the unpredictability of risk adjustment under the model.
Indeed, Fleury said if Sharp "had known that this was going to happen, we wouldn't have ever elected to be in the Next Generation program."
Others argue that it's no surprise that experimental models from the CMS' Center for Medicare and Medicaid Innovation are unpredictable. "CMS' goal isn't to make everyone happy, it is to make the program better," Muhlestein said.
The CMS' ultimate goal is for the Next Generation program to mature out of the Innovation Center and into an established program under Medicare. In fact, Track 3 of the Medicare Shared Savings Program is modeled closely after the Pioneer ACO Model, an Innovation Center graduate.
"They (the CMS) take the model, experiment, figure out things that work and don't, and then they are able to put those into the standard Medicare Shared Savings Program," Muhlestein said. "When you make a change, not everyone is going to love it … that is something the CMS has to grapple with."
An edited version of this story can also be found in Modern Healthcare's March 26 print edition.