In my last post , I made fun of Ezekiel Emanuel and Joseph Liebman for predicting that ACOs would replace insurance companies by 2020. I noted that the 800 to 1,000 ACOs reportedly in existence today are nowhere near ready to accept full insurance risk because they have shown no ability to cut costs. 
Although Emanuel and Liebman were foolish to predict most ACOs would quickly evolve into successful insurance companies, they shared with all other ACO advocates the understanding that that’s what ACOs are supposed to do – over time they’re supposed to bear more and more financial risk. It’s just never been clear how much financial risk ACOs were supposed to take on. The Affordable Care Act, for example, which authorized CMS to establish an ACO program within Medicare, did not address that question.
A reasonable interpretation of statements and papers by ACO proponents is that they expected a substantial portion of ACOs to morph into insurance companies. According to an early paper by several prominent ACO proponents, including Elliott Fisher, ACOs were expected to accept more risk in three stages: In stage 1 they would accept only upside risk (the opportunity to share in savings if they stayed under a target spending level); in stage 2 they would accept both up- and downside risk; and in stage 3 they would accept full insurance risk (they would no longer be paid fee-for-service, but would instead be paid “capitation” payments, aka premium payments), and would, therefore, have to set aside reserves. Presumably stage 3 ACOs would also have to be licensed by the insurance regulators in the states where they operate.
But the ACO industry is still in Fisher et al.’s stage 1, and it appears to be stuck there. According to the latest CMS reports on Medicare’s three ACO programs, 90 percent of the ACOs were in stage 1 at the end of 2016 and only 10 percent had entered stage 2 (see the bar graph in my last post). (We have almost no information on the amount of risk currently accepted by ACOs set up by private-sector payers.)
The failure of the Medicare ACOs to cut costs and to evolve past stage 1 presents CMS, Congress and the Medicare Payment Advisory Commission (MedPAC) with difficult questions: Why are the ACOs failing to cut costs, and what should be done about the Medicare ACO program if it continues to spin its wheels?
A level playing field for ACOs and MA plans?
The rational approach would be to conduct studies of ACOs to determine what interventions they use and whether those interventions save money after taking into account the cost of the interventions. (In a sane world, this would have been done before ACO proponents lobbied Congress to endorse ACOs, but we’re talking managed care here, and due diligence is never done prior to the launch of managed care fads.) But MedPAC has never done that, and no one else has either. The extremely rare studies of the activities of particular ACOs merely demonstrate the axiom, “If you’ve seen one ACO you’ve seen one ACO.” The cause of this information vacuum is the flabby definition of “ACO” and the inexplicable failure of ACO proponents to address the question, What is it ACOs will do that the insurance industry hasn’t already tried?
Since CMS implemented the first two ACO programs early in 2012, MedPAC members have asked numerous variations of that question. For example, commissioners have asked how ACOs that cut costs a percent or two differ from those that raise costs a percent or two, and whether ACOs are better at managing the chronically ill than Medicare Advantage (MA) plans. The staff have had to say they don’t know. 
MedPAC has attempted to cope with the ACO information vacuum by developing a scheme that would “level the playing field” for Medicare’s three programs – the ACO, MA, and traditional fee-for-service (FFS) programs – by making the payment rules uniform. The rationale appears to have been, “If we can’t judge efficiency by peering inside ACOs and MA plans, the least we can do is set up a fair fight between the ACO black box, the MA black box, and the FFS program and see which one wins.”
The commission began discussing this approach at their September 12, 2013 meeting. MedPAC staffer David Glass began the discussion with this question: “[D]o we want to move to a level playing field across traditional fee-for-service, ACOs, and MA?” (See page 73 of the transcript .) The ensuing discussion indicated the answer was yes. The commission continued their discussion at their November 7, 2013 meeting by which time they were using the word “synchronization” to describe their goal of harmonizing the payment rules for all three programs.
The commissioners agreed early on to let a long-held principle they call “financial neutrality” be their guide. “Financial neutrality” means Medicare should pay the same amount for a given enrollee regardless of which of the three Medicare programs the enrollee winds up in. Over the next four years they discussed whether a proposal called “premium support” could achieve financial neutrality.
“Premium support,” first proposed in the 1990s, would give Medicare beneficiaries a voucher useful only for the purchase of health insurance. Under the version MedPAC studied, Medicare’s FFS program would be treated as a “plan” that would compete with ACOs and MA insurers. Beneficiaries could use their voucher to buy insurance from one of the Medicare ACOs, one of the MA plans, or from the FFS program. The financial neutrality requirement would be met because the value of the voucher would be the same regardless of whether the beneficiary used the voucher to buy insurance from an ACO, an MA plan, or the FFS program.
But the implementation issues turned out to be unsolvable. Despite devoting hours of discussion time to the “premium support” concept, as well as multiple, mind-numbing chapters in their June 2013, 2014, 2015, 2016, and 2017 reports, the commission failed to produce a workable proposal. Here are the closing words of the commission’s latest, and for the time being, final chapter on premium support, which appears in its June 2017 report : “The Commission makes no recommendation on whether premium support should be used. Rather, we discuss an array of complex issues that the Congress may want to address if it decided to develop a premium support system.” (p. 127)
The first sign of how impossible it would be to build a level playing field for the ACO, MA and FFS programs was the commission’s unannounced and unexplained decision to pull the ACO program off the table, leaving only the FFS and MA programs in the financial neutrality, premium support discussion. This decision was implied rather than stated in MedPAC’s June 2015 report with this cryptic, parenthetical remark: “Because beneficiaries in ACOs are part of FFS Medicare, only two of Medicare’s payment models – FFS and MA – were relevant to the analysis of premiums.” But even with the complex ACO program off the table, the staff and commissioners still could not solve the multiple issues that had to be resolved to put the FFS and MA programs on a level playing field. By the time the commission issued its June 2017 report, it was clear they were giving up even on the simpler FFS/MA premium support idea. The issues they would have had to solve to synchronize the payment rules for just those two programs were too numerous and too complex.
If I had to rank, by degree of intractability, the multiple issues that thwarted premium support, I would rank risk adjustment number one. Throughout the 2013-2017 period during which MedPAC discussed premium support, staff and members asserted over and over that accurate risk adjustment was essential to its success. And yet even after publishing a chapter in their June 2014 report stating accurate risk adjustment is not possible, they kept discussing premium support. That chapter reported that CMS’s hierarchical condition categories (HCC) risk adjuster is so crude it vastly overpredicts the cost of (overpays for) healthy seniors and vastly underpays for sicker seniors (see the table in my comment here ). That chapter also reported MedPAC does not know how to improve the HCC adjuster’s accuracy.
Inaccurate risk adjustment wasn’t the only mind-bending problem premium support presented to the commission. Drawing arbitrary boundaries around “markets,” “attributing” patients to ACOs, and, starting in 2015, figuring out how MACRA meshed with premium support are other examples. (I suspect the enactment of the absurdly complex MACRA was the reason MedPAC dropped ACOs from the financial neutrality discussion in 2015. The imposition of MACRA on the FFS program changed what was already an impossibly complex discussion into an insanely complex discussion.) But no other problem was so intractable and so fundamental. Without an accurate risk adjuster, effective competition between entities paid a set fee per enrollee is not possible. Any system that underpays for the sick and overpays for the healthy will induce many players to “compete” by avoiding the sick, and those who refuse to “compete” in that fashion will suffer loss of resources, which will in turn harm their sicker and poorer patients.
Don’t ask, don’t tell
The good news about MedPAC’s four-year struggle to think through premium support is that it clearly showed that real competition among and between ACOs and insurance companies is not possible. The bad news is that policy-makers are not paying attention. At the federal level, the Trump administration made statements last week suggesting they are eager to impose some version of premium support on Medicare and Medicaid. At the state level, the experiment with Medicaid ACOs continues at full throttle.
Minnesota is one of the states that inserted ACOs into its Medicaid programs shortly after the Affordable Care Act elevated the ACO from obscure idea to stardom. Under the authority of legislation enacted by the Minnesota legislature in 2010, the state’s Department of Human Services (DHS) started a Medicaid ACO program in 2013, just one year after CMS implemented its Medicare ACO program. Although Minnesota’s Medicaid program had long ago been turned over to HMOs, it didn’t occur to state legislators or then-governor Tim Pawlenty to ask why Medicaid needed ACOs in addition to HMOs.
DHS is finally getting around to asking a version of that question. In November of 2017, DHS posted a proposal that seeks to harmonize the HMO and ACO payment rules. (The proposal ignored the small FFS sector that remains within Medicaid.) In my next post, I will discuss DHS’s proposal, the comments DHS got back, and DHS’s reaction to the comments. The dialog between DHS and the public is similar to the four-year conversation MedPAC had about premium support. The same mind-bending issues arose, and at this date it appears that none of the major, must-solve issues are going to be solved.
 My last article generated some commentary here and elsewhere to the effect that I ignored research indicating Medicare’s ACOs are doing slightly better than breaking even, which is what CMS’s reports indicate. I assume those comments referred to articles by J. Michael McWilliams et al. and David Nyweide et al. I am well aware of those articles. They are simulations of Medicare’s ACO programs, not studies of the actual programs. I discussed them in an essay on THCB entitled “Simulated ACOs versus real-world ACOs.”
I concluded the authors’ simulations changed every important parameter in CMS’s ACO programs, including “the control group, the experimental group, the method for calculating how the ACOs affected costs and, with that, the size of and distribution of rewards and penalties.”
 The MedPAC staff’s most definitive statement about the impossibility of ever knowing what ACOs do came in the course of discussing why Medicare Advantage plans raise rather than lower Medicare’s costs. At MedPAC’s November 3, 2016 meeting, the staff told the commission MA plans were raising Medicare’s costs. This provoked a discussion about why. Several commission members asked why “integration,” which is supposed to be the hallmark of MA plans and ACOs, wasn’t working. Commission member Warner Thomas asked, “[C]an we look at actual integration and clinical integration of services?” (p. 37 of the transcript of that meeting ) Staff member Jeff Stensland replied that it’s just not possible to distinguish mere financial integration from clinical integration (clinical integration is what MA plans and ACOs are supposed to achieve). Here is how he put it: “It’s very difficult to distinguish [between] a truly integrated entity, where people are really talking to each other and cooperating and improving care, and an entity that just looks, on the surface, like they taught to the test to make it look like they’re doing this, but they really aren’t doing it underneath.… [T]hat difficulty of distinguishing between truly good integrated entities that are coordinating care and reducing costs and improving quality, and integrated entities that say they’re doing that but really aren’t… — almost impossible to distinguish between those two.” (p.38) (See my discussion of this issue here. )