CMS has now conducted three demonstrations of the “accountable care organization,” and all of them have failed. The Physician Group Practice (PGP) Demonstration, which ran from 2005 to 2010, raised Medicare costs by 1.2 percent.  The Pioneer ACO program, which ran from 2012 through 2016, cut Medicare spending by three- or four-tenths of a percent on average over its first four years. And the Medicare Shared Savings Program (MSSP), which began in 2012 and may lumber on indefinitely, has raised Medicare costs by two-tenths of a percent on average over its first four years.
It is way past time for CMS and health policy researchers to determine why all three ACO demos failed. In the first two installments in this three-part series I laid out one of the reasons: CMS’s method of assigning patients to ACOs guarantees ACOs must apply their magic to a rapidly changing pool of patients and doctors. In the first essay , I demonstrated that this method, which assigns patients first to doctors based on where they get the plurality of their primary care visits and then to ACOs if their doctors are in ACOs, guarantees high churn rates among doctors and patients, shunts sicker patients away from the ACOs, and assigns few ACO patients to each ACO doctor. In the second essay I reviewed the series of evidence-free decisions that led to CMS’s plurality-of-visits method. I noted that the first of these decisions was one Congress made: They instructed CMS to figure out how to assign patients to ACOs without making patients enroll in ACOs.
In this last installment I ask what if anything can be done to reduce the patient churn rate, and whether reducing the churn rate would do anything to improve the performance of ACOs. I’ll first review the argument some ACO advocates make that the churn rate could be greatly reduced if “attributees” were induced to stay within “their” ACO. We will see that there is little reason to believe that. Then I’ll go on to ask whether ACOs would work even if “attributees” were forced to enroll with an ACO, a requirement that would effectively define ACOs as HMOs. I’ll answer that question in the negative on the ground that the cost-control tactics available to ACOs, namely those that have long been available to HMOs and other insurance companies, have not worked.
Finally, I’ll discuss the only feasible option I see for salvaging something from the entire ACO/HMO experiment: Requiring that CMS and Congress abandon the fantasy that ACOs/HMOs can do something positive for entire “populations” and instead require them to provide defined services to patients with specific diagnoses.
I would like to apologize in advance for the frequent use of quote marks around words like “attributee” and “attestation.” But I insist on using them to warn readers that when we attempt to understand the conversation about ACOs we enter a strange world where words don’t mean what they seem to mean.
When it became clear that the ACO would be enshrined in the Affordable Care Act, many observers asked whether ACOs weren’t merely warmed over HMOs. Proponents replied that HMOs were different from ACOs in several respects, the most important of which was that Americans would not have to enroll with ACOs and use only ACO doctors and hospitals. 
But it wasn’t long before ACO advocates began to complain that this distinction was making it impossible for ACOs to lower costs. Pioneer ACO managers voiced this complaint to L&M Policy, the author of the final evaluation of the Pioneer ACO program. “Representatives from many of the Pioneer ACOs noted that it was more difficult than initially anticipated to manage beneficiary utilization and patient visits outside of the ACO because beneficiaries did not face financial incentives to use ACO providers,” L&M reported. “Some ACOs reported frustration with translating existing care management programs to the ACO population without the benefit of traditional managed care tools (e.g., enrolled population, utilization management, prior authorization)….” (p. 70)
ACO proponents have responded to this lament with two related proposals: (1) CMS should ask “attributees” to “attest” that an ACO doctor is “their” doctor and, if they say yes, CMS should count them as ACO patients; and (2) ACO “attributees” should be given financial incentives to sign the letter and stay within the ACO network. As is always the case with managed care proposals, advocates have proposed these “reforms” without a hint of evidence that they’ll work or what they might cost.
You can readily see these two proposals together recommend enrollment without using the “enrollment” word. So why don’t proponents of these “reforms” just admit they want Medicare recipients to enroll in ACOs and suffer financially if they go outside the ACO network? I submit they know there was a hellacious HMO backlash in the latter half of the 1990s, triggered primarily by American hostility to limitations on choice of doctor, and that many Medicare recipients in FFS Medicare are there because they don’t want to enroll in a Medicare Advantage plan, and they’re hoping Medicare beneficiaries will be too dumb to notice that they’re being offered a bribe to enroll in an HMO dressed up as an ACO.
Avid ACO proponent and former CMS administrator Mark McClellan, for example, offered the enrollment-by-another-name proposal in a list of “reforms” of the MSSP program he sent to CMS Administrator Marilyn Tavenner in February 2015. He proposed CMS should mail an “attestation” form to “attributees” that “attributees” could sign and thereby “positively identify their primary care provider … and thus declare their active participation in an ACO.” (p. 6) Got that? “Active participation” would be “declared.” What does “active participation” in something as vague as an ACO mean? Would it mean signers agree to stay within the ACO’s network? McClellan didn’t say. That was for Tavenner to figure out.
But McClellan hinted that that’s exactly what he meant when he described an additional proposal to expose “attributees” to financial incentives: “We also recommend that CMS consider additional incentives for patient participation in the ACOs, including waiving or reducing copays and deductibles for patients when they receive care from their primary care physician or other providers in the ACO.”
McClellan offered no evidence that his back-door enrollment proposal – one which would eliminate one of the few distinctions between ACOs and HMOs – would work. And he offered no estimate of what it would cost.  Obviously, contacting patients and urging them to sign “attestation” documents would cost money, and so would paying patients to stay within ACOs/HMOs.
Explaining Groupthink to Doctors
Let’s focus first on the “attestation” nostrum. Let’s set aside the cost issue for now, and ask the question that McClellan and other ACO advocates never address or address only with happy talk, namely: Even if CMS spends whatever it takes to contact every ACO “attributee” and urge them to sign a piece of paper that “declares their active participation” in an ACO, what good will that do? Remember, the message from CMS about the “attestation” form would be in addition to the notice CMS already sends out to Medicare beneficiaries indicating they have been assigned to an ACO.
Let me begin by reviewing evidence from the final Pioneer ACO evaluation on how poorly ACO doctors understand the vague “ACO” concept. Once you comprehend how confused the doctors are, you’ll shake your head and wonder how anyone could think patients will ever understand the flabby ACO concept. Remember, we’re discussing the crème-de-la-crème here – CMS selected the 32 hospital-clinic and clinic-clinic chains to participate in the Pioneer program because of their size and previous experience “managing care.” If doctors anywhere understand ACOs, you’d think it would be the doctors in the Pioneer ACOs.
Although ACO proponents claimed ACOs would be “run by doctors” and would therefore be kinder and gentler than HMOs , it turns out doctors not only didn’t run the Pioneer ACOs they had a poor understanding of what their ACOs were doing to and for them. According to L&M’s final evaluation of the Pioneer program, “The vast majority of providers participating in Pioneer ACOs were not directly part of the decision to participate, but rather were employed by or part of a medical group that joined the ACO.” (p. 23) The evaluation went on to say, “Survey results indicated that only 2 percent of Pioneer physicians had served on the ACO board of directors and only 9 percent had served on an ACO committee. Even for those reporting such involvement, only half of Pioneer physicians said they were satisfied with their participation….” (p. 43)
Because the vast majority of doctors weren’t involved in the decision to join an ACO, and because the ACO is so poorly defined , doctors had little comprehension of what their ACOs were doing. “In several respects, physicians were not particularly knowledgeable about the ACO,” reported L&M. “When asked if they knew which of their patients were aligned with the Medicare ACO, just over a third of Pioneer physicians reported knowing which beneficiaries were aligned and a similar proportion reported not knowing their aligned beneficiaries at all. When asked about the elements of their compensation, almost half of physicians participating in the Pioneer model reported not knowing whether they were eligible to receive shared savings from the ACO….” (p. 43)
Given their incomprehension of the ACO thing, Pioneer physicians understandably were unable to generate the evidence-free hype about ACOs that ACO advocates can generate in their sleep. For L&M and ACO management, physician confusion about ACOs reflected a defect in physicians, not in the wonderful ACO project. “Nearly all Pioneer ACOs reported interest in improving physician engagement,” L&M reported, “with many reporting frustrations in the perceived lack of engagement by physicians with the ACO.” (p. 41)
But maybe, just maybe, the problem wasn’t with the knuckle-dragging doctors. Maybe it was that ACOs were either doing so little that most doctors couldn’t detect anything that might be called an ACO intervention, or that doctors did perceive some ACO interventions but viewed them as not helpful. L&M did report that 70 percent of ACO physicians thought they were already practicing in a manner consistent with the goals trumpeted by ACO advocates, and for that reason, “they may have believed that they did not need the structure or strategies provided by the ACO to adapt to new approaches to care delivery.” (p. 42)
The one feature of ACOs the great majority of doctors understood clearly was that ACOs impose more paperwork on doctors. “Approximately three-quarters of Pioneer physicians indicated that participation had required them to increase time spent on administrative, documentation, and reporting tasks…, ” said L&M. (p. 45)
Explaining Groupthink to the Unwashed Masses
If doctors who have been roped into ACOs don’t understand ACOs, is there any hope that their patients will sign “attestations” with anything resembling informed consent? Informed consent, it should go without saying, will be necessary if “attestation” is to have any chance of influencing patient behavior.
In fact, the Pioneer program did make use of “attestations” (but not financial incentives). According to a December 2011 CMS press release, “beneficiaries may affirmatively attest that their primary provider is in a Pioneer ACO, and can then be aligned with the ACO and benefit from the enhanced care coordination that it offers.” (p. 2) Obviously “attestation” didn’t work. Why not? Because the ACO is so poorly defined ordinary mortals can’t understand it.
L&M discovered that “attributees” in Pioneer ACOs didn’t have the faintest idea what an ACO is. L&M concluded, “Despite the annual notification letter and Pioneer ACOs’ efforts to engage beneficiaries, in small group discussions with beneficiaries … we learned that beneficiaries were generally unaware of the ACO organization and the term ‘ACO.’ In the few cases where the beneficiaries reported hearing the term ACO, they were not able to describe what an ACO is….” (p. 51) Although L&M did a great job measuring and reporting high churn rates among Pioneer doctors and “attributees,” they made no attempt to ask why “attestation” didn’t work. 
If you were in charge of contacting “attributees” and urging them to sign ACO “attestations,” what would you say? Don’t bother asking ACO advocates like McClellan for help. Here’s all McClellan had to say on this topic in his letter to Tavenner: “CMS should also support better patient education around the goals and features of Medicare ACOs, and how patients can work with their providers to improve care.” (p. 6)
Would ACOs Work if “Attributees” Had to Enroll in Them?
For the sake of argument, let’s pretend that CMS could explain ACOs to Medicare recipients, that all recipients assigned to ACOs could be persuaded to “attest,” and that this back-door enrollment requirement reduced “attributee” churn. Would CMS’s ACO programs finally start cutting Medicare spending? (Note that I’m not asking whether the costs ACOs incur to do whatever ACOs do would exceed any savings for Medicare. I’m only asking whether Medicare costs might drop.)
To answer that question, we would need to know what interventions ACOs would apply to “their” patients. Sad to say, a decade after the phrase “accountable care organization” was invented, we don’t know what ACOs do. Again, that’s because the definition of the ACO is so vacuous. As L&M put it in their first evaluation of the Pioneer program, “The ACO ‘treatment’ under investigation is not a prescribed set of activities or interventions. Rather, it is a financial arrangement….” Is it possible to be any more abstract? Have you ever seen a more useless definition of anything?
Even the managers of the Pioneer ACOs were in the dark. “The vast majority of Pioneer ACOs entered the model with care management experience,” L&M stated in the final evaluation. “However, there remained a decided lack of consensus on what makes care management effective.… Even through the third year of the [demonstration], Pioneer ACOs continued to report using trial and error to make incremental changes and improvements to their care management programs.” (p. 48)
So how does one rationally inquire why ACOs are failing when no one knows what they’re supposed to be doing? Lawton Burns and Mark Pauly, economists at Wharton, addressed this question in a 2012 paper in Health Affairs. They politely characterized the flabbiness of the ACO concept as a “lack of consensus … over what the new entities should do, or stop doing, to reduce spending and how they should control out-of-network utilization.” Then they attempted to do that which ACO advocates should have done before promulgating the ACO “arrangement”: They listed the tools or interventions available to ACOs and reviewed the evidence on each one.
Burns and Pauly made the reasonable assumption that ACOs had no choice but to use the same managed-care tools that HMOs and kindred insurers were using, such as utilization review, pay-for-performance, health information technology, and the ever-popular “care coordination.” After reviewing the literature on these tools they concluded, “The evidence … suggests that components of accountable care organizations have limited and uncertain impact, especially on cost savings, and thus provide little support for the [claim that] better care coordination will improve quality at any given cost, and … the organizations will lower Medicare’s rate of spending growth.” (p. 2412) 
If the tools available to ACOs can’t lower costs, then reducing churn among ACO “attributees” by forcing or bribing them to enroll in ACOs will do little to improve the ACOs’ ability to cut costs. To put it another way, the high rate of turnover among ACO “attributees” is not the most significant reason ACOs are failing. They’re failing because the cost-containment tools available to them are so ineffective. These same tools didn’t work for HMOs and other insurance companies, and of course enrollment is required for insurance.
Salvaging Something From the ACO Experiment
The taxpayer has now financed three ACO demonstrations – the PGP, Pioneer, and MSSP demos.  All of them have failed to cut Medicare spending; all of them are raising costs if ACO intervention costs are taken into account. The purpose of demonstrations is to learn something. Sad to say, we have learned nothing from the ACO demos. No one – not CMS, not MedPAC, not ACO buffs – can explain the chronic failure of Medicare ACOs (see my discussion of MedPAC’s bafflement here ).
The ACO demonstrations will not have been a total waste of money if CMS can determine whether any ACOs did anything that improved the quality of care of specific patients. It is possible that some of the managed care tools ACOs are expected to use would work (that is, they would at least improve quality and might lower costs as well) if they were applied to subsets of the chronically ill. It’s clear those tools are never going to work if they are applied to entire “populations.” Managed care proponents must stop thinking in terms of structures (ACOs, HMOs, “integrated systems,” “medical homes”) that apply managed care magic to entire populations, and start thinking in terms of specific services delivered to subsets of the chronically ill.
 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” over the five years the demo ran, for a net loss of 1.2 percent. (p. 64)
 Austin Frakt and other ACO proponents claimed that another difference between ACOs and HMOs is that ACOs bear less risk than HMOs. They argued that ACOs would not bear total insurance risk but would instead be subjected to a lower level of risk-sharing called “shared savings,” and this would give doctors and their ACO managers less incentive to deny necessary services to patients. But other ACO advocates claimed just the opposite – that ACOs were expected to move gradually from the limited risk of shared savings contracts to full blown insurance risk delivered via capitation aka premium payments. For, example, the final evaluation of the Pioneer ACO program stated, “CMS intended the model to allow these provider groups to move more rapidly from a shared savings payment model to a population-based payment model.” (p. xii)
Some ACO advocates also sought to distinguish ACOs from HMOs with the claim that ACOs would be more “accountable” than HMOs ever were because the spread of electronic medical records and alleged advancements in quality measurement since the “HMO backlash” of the 1990s would make it easier for unidentified third parties to monitor the quality of ACO medical care (see, for example, CMS’s brave promise to “routinely analyze data surrounding utilization of services” at p. 2 of this 2011 press release ). Thus, even if the shifting of some but not all insurance risk to ACOs induced ACOs to short-change their patients, all-seeing monitors would detect this behavior and correct it. Here are other examples of experts who claimed ACOs were not HMOs in drag: Ezekiel Emanuel in a 2012 comment for the New York Times blog; two experts quoted in a 2015 article by Kaiser Health News.
 Here is the totality of the “evidence” McClellan offered CMS administrator Tavenner in his February 2015 letter to her in support of his proposal that CMS allow ACOs to try to induce “attributees” to “attest” to their loyalty to an ACO doctor: “Results regarding the extent to which patients choose to join an ACO and the potential long-term impact are not yet available, but private-sector programs have had promising results, and increased beneficiary engagement is a fundamental objective for Medicare.” (p. 6)
 Obama adviser and ACO advocate David Cutler claimed ACOs would not “dictate to doctors and patients what they are allowed to do and what they cannot” as HMOs did (see Austin Frakt’s New York Times comment in footnote 2).
 For a discussion of the flabby, aspirational definition of “ACO,” see my comment here
 The problem ACOs face in inducing “attributees” to give a hoot about “engaging” with “their” ACO may go beyond the inability of mortals to comprehend what “ACO” means. According to L&M, many Pioneer ACO managers believe the problem is resistance to HMO-like interference in the doctor-patient relationship. I quote from the final Pioneer evaluation: “Pioneer ACOs highlighted difficulties managing FFS beneficiaries; these ACOs often contrasted ACO-aligned beneficiaries to beneficiaries in MA [Medicare Advantage] plans. … Some Pioneer ACOs asserted that beneficiaries often choose traditional Medicare FFS specifically because they do not want provider network limitations, and these patients may have construed efforts to engage them in the ACO as an effort to limit provider choice.” (p. 51)
 The conclusion by Burns and Pauly that none of the managed care tools work is consistent with other analyses using different methodologies. I reported in a literature review published in Health Affairs in 2000 that insurance companies that use managed care tactics are not cutting costs. In a 2015 paper financed by the leading lights of the American health care establishment, Lawton Burns and Jeff Greenfield examined the evidence for “Kaiser-like entities,” aka “integrated delivery systems.” In a blog comment about that paper, the authors stated, “We reviewed more than 30 years of academic literature on vertical integration and diversification in healthcare, and found virtually no measurable benefits – either to society or to the sponsoring healthcare enterprises themselves – of putting health insurance, hospitals and physician services under the same structure.”
 Strictly speaking, only the PGP and Pioneer experiments were “demonstrations.” The MSSP program was authorized by the Affordable Care Act and is a permanent program as long as the ACA remains law.