Six years ago Ezekiel Emanuel and Jeffrey Liebman made the foolish prediction that ACOs would eat the insurance industry’s lunch. “By 2020, the American health insurance industry will be extinct,” they wrote. “Insurance companies will be replaced by accountable care organizations….” This would happen, they argued, because ACOs are just so darned good at lowering costs compared with insurance companies.
The first Medicare ACO programs began in 2012. Today there are 800 to 1,000 ACOs in business.  But ACOs aren’t even close to displacing the insurance industry. The most obvious reason is they don’t want to be insurance companies – they don’t want to bear full insurance risk. And the reason for that is they can’t cut costs. The performance of the Medicare ACOs, which are the only ACOs for which we have reliable data, illustrates both problems: Very few want to accept “downside risk” (the risk of losing money if they can’t cut costs); and they are incapable of cutting costs.
ACO hype confronts reality: Reality wins
Anyone paying attention to the research knew even before 2012 that ACOs wouldn’t cut costs for a general population (as opposed to a small slice of the population that is very sick). The Physician Group Practice Demonstration, which was widely seen as the first test of the ACO concept, raised Medicare spending. According to the final evaluation of the demonstration, the ten participating ACOs raised Medicare’s costs by 1.2 percent over the five years the demonstration ran (2005-2010), and it might have been worse if the ACOs hadn’t upcoded.  This failure to cut costs occurred despite the fact that the ten participating “group practices”/ACOs were very experienced in managing risk. They had names anyone who studies health policy would recognize, including Dartmouth-Hitchcock Clinic, Geisinger Clinic, and Marshfield Clinic. According to the final report on the demo, “Seven of the ten participants had currently or previously owned a health maintenance organization ….” (p. 15)
The latest reports from CMS on its Medicare ACO programs, which report 2016 data, indicate the current crop of ACOs are equally incapable of cutting costs. The data contained in the reports indicates the ACO programs are merely breaking even, and that’s only if you don’t count the costs to the ACOs of doing whatever it is ACOs do. The staff of the Medicare Payment Advisory Commission (MedPAC) prepared a graph for the commission showing how each of CMS’s ACO programs did in 2016. This graph, presented at the commission’s January 2018 meeting, appears below. (The graph appears on slide 8 here )
The graph presents the results for the 458 ACOs that participated in CMS’s three ACO programs – the Medicare Shared Savings Program (MSSP, which began in 2012), the Pioneer program (which began in 2012 and ended in 2016), and the NextGen program (which started in 2016). You’ll see that 410 of the ACOs, or 90 percent, were in “track one” of the MSSP, which means they were willing to accept only upside risk (the chance to make money if they came in under a target level of spending). We see as well that all three programs roughly broke even (again that’s not counting the cost of the interventions ACOs deploy in their attempt to cut costs).
The graph’s most discouraging news, from the point of view of ACO advocates, is the underwhelming performance of the eight diehard ACOs that remained in the Pioneer program in 2016. The graph indicates those ACOs cut Medicare’s costs by a mere seven-tenths of one percent in 2016, the fifth and last year of the program. Like the ten ACOs that participated in the Physician Group Practice demo, those eight ACOs were the crème de la crème – the entities most likely to succeed at the ACO game. They were the crème de la crème of the cohort of 32 ACOs that CMS selected to enter the Pioneer program in 2012, and those 32 ACOs were in turn the crème de la crème of all ACO candidates in 2011 when CMS accepted applications for the Pioneer program.
Because ACOs in the Pioneer program would have to accept both up- and downside risk, CMS was very selective in choosing the original 32 Pioneer ACOs. As the Advisory Board put it in an article published just before the Pioneer program began, “The 32 Pioneers underwent a rigorous evaluation – including a comprehensive review of applications and in-person interviews….” And yet by 2016 only eight of these rigorously evaluated ACOs remained. The dropouts were those who lost money or expected to lose money. Thus, if we were to take an “intent to treat” approach in evaluating the Pioneer program and asked how well all 32 performed, the result would probably be worse than the 0.7 percent savings achieved by the eight diehard ACOs that participated for all five years.
Why did we need ACOs in addition to insurance companies?
What were ACO proponents thinking would happen? We know they thought ACOs would perform much better than they have. But how much better? Did they agree with Emanuel and Liebman? Did they think ACOs would gradually accept total insurance risk and go on to outcompete the insurance industry? We don’t know. We don’t know because ACO proponents never bothered to ask an elementary question: “Are we proposing to create an ACO industry because we think the HMOs and the managed care tools they popularized have failed, and if so, why do we think ACOs will succeed where HMOs and managed care plans have failed? What tools will ACOs use that the insurance industry hasn’t already experimented with?”
ACO proponents never indicated that they had given a moment’s thought to that question. They never came right out and said, “Managed care plans have had their day in the sun, they haven’t worked, and now we need to replace them with a new entity we’re calling the ‘accountable care organization,’ and here’s why.” They just demanded that Congress authorize CMS to set up an ACO program in Medicare alongside the Medicare Advantage program, and that private-sector payers create a private-sector ACO industry alongside the insurance industry.
Congress, with even less forethought, agreed. Congress inserted provisions into the Affordable Care Act authorizing CMS to set up Medicare ACO programs, President Obama signed the ACA in March 2010, and CMS implemented the Pioneer and MSSP programs early in 2012. In 2015, Congress did it again: They enacted MACRA, which encourages the formation of even more ACOs. At no time in this higgledy-piggledy ACO-creation process did any of the participants publicly address the question, “How will ACO magic differ from HMO magic, and if there is no difference, why are we doing this?”
A handful of ACO advocates did attempt to distinguish ACOs from HMOs after the ACA was enacted, but their distinctions were meaningless or misleading. Of these after-the-fact distinctions, the three most frequently heard were: (1) Doctors will be running ACOs, not “distant,” uncaring HMOs; (2) ACOs do not limit patients to a network of “preferred” providers as HMOs do; and (3) modern information technology makes it possible for ACOs to provide much better care than HMOs ever did. Emanuel and Liebman, for example, trotted out all these distinctions. 
The first claim – that doctors are running ACOs, not HMOs – is laughable. HMOs and other insurers are setting up their own ACOs, and even the ACOs that were not established by insurers have very little physician involvement in management.  The claim that ACOs do not limit choice of provider is, strictly speaking, true at this moment, but that is changing because ACOs are not happy with the enormous “leakage” they suffer because they can’t force their “attributees” to stay in network. Finally, the argument that information technology improves quality of care is an exaggeration of IT’s impact on quality, and in any event, the insurance industry has as much access to IT as ACOs do.
But even if these false distinctions between ACOs and HMOs were true, they wouldn’t tell us what it is ACOs do that HMOs don’t do. So what if doctors really are running ACOs? So what if ACOs have yet to force patients to stay within ACO networks? What does that tell us about what ACO doctors are doing that doctors micromanaged by the insurance industry are not doing?
Comparing one black box to another
The failure of ACO proponents to answer the question, Why do we need both ACOs and insurance companies, is creating problems for Medicare which now has to administer three separate programs – the traditional fee-for-service program, the new ACO program, and the Medicare Advantage (MA) program. Life for CMS, and for MedPAC (which has to advise CMS and Congress), was difficult enough when CMS had to administer just the FFS and MA programs. It has become infinitely more difficult with the insertion of the ACO program in 2012 and the enactment of MACRA in 2015. The complexity of today’s Medicare program has driven up CMS’s administrative costs, and has forced MedPAC into long discussions about how to ensure that all three programs are paid fairly.
The same problem now afflicts state Medicaid programs that were forced by their legislatures to balkanize into the same three sectors – a FFS sector, a privatized Medicare-Advantage-like sector, and an ACO sector.
In my next post, I will discuss the attempts MedPAC and Minnesota’s Medicaid program have made over the last few years to apply more uniform payment rules to their ACO and managed-care-plan programs. MedPAC and Minnesota’s Department of Human Services (DHS), which runs the Minnesota Medicaid program, apparently want to answer the question ACO proponents never bothered to ask: Are ACOs more efficient than insurance companies? MedPAC seems to want to answer that question by forcing all three Medicare sectors to compete on a level playing field to see which ones are driven to extinction by the others. Minnesota’s DHS seems to want to answer that question by forcing its ACO and MA-like sectors to compete on a level playing field. I use the word “seems” because MedPAC and DHS can’t bring themselves to come right out and say they expect ACOs to beat the insurance industry or vice versa. But they are willing to say they want to create competition between the two types of entities under uniform payment rules.
As we shall see, MedPAC’s attempts to create a level playing have failed, and DHS’s attempt appears to be well on its way to failing.
 Our knowledge about Medicare ACOs is very limited, but at least we know how many of them there are. We don’t even know the number of entities that call themselves ACOs that contract with private-sector payers (insurance companies and self-insured employers). The estimates are all over the map. A directory of ACOs claims to list 800 of them. Elliott Fisher, who invented the ACO label in 2006 with MedPAC, alleges there are 1,000 .
 According to the final report on the Physician Group Practice Demonstration, “[T]he demonstration saved Medicare .3 percent of the claims amounts, while performance payments were 1.5 percent of the claims amounts.” (p. 64)
 Here’s an excerpt from the Emanuel-Liebman op-ed in which they sought to persuade readers that ACOs are not your father’s HMOs. “ACOs are not simply a return to the health maintenance organizations of the 1990s. Although in both models patients are members of a provider network with a specific group of doctors and hospitals, and both are paid primarily per member rather than per procedure or test, there are big differences between them. HMOs were often large national corporations far removed from their members. In contrast, ACOs will consist of local health care providers working as a team to take care of patients who are likely to be members for years at a time. HMOs often cut costs not by keeping people healthy but by denying patients services and by forcing doctors and hospitals to take lower payments. In the 1990s, we lacked the information technology and proven models of integrated care delivery that we have now. These advances will allow ACOs to simultaneously improve health outcomes and reduce costs.
“A final bonus of A.C.O.’s is that they will lead to a better form of competition in health care markets. Today, consumers have to choose among insurance plans with a bewildering array of copayments, deductibles and annual out of pocket maximums — choices that few of us are any good at making. In the A.C.O. model, consumers will choose a primary care physician and the team of doctors and hospitals that are in the same group.”
 Claudia Schur and Janet Sutton report that only a tiny fraction of all doctors in Medicare ACOs play leadership roles, and up to half are unaware of who their ACO patients are or whether they are eligible for shared savings.
Kip Sullivan is a long-term health policy expert based in Minnesota