The Medicare Payment Advisory Commission (MedPAC) has done it again. At their October 4, 2017 meeting they agreed to repeal the Merit-based Incentive Payment System (MIPS), an insanely complex and evidence-free pay-for-performance scheme within the larger program known as MACRA. Instead of examining how they made such a serious mistake in the first place (MedPAC has long supported turning fee-for-service Medicare into a giant pay-for-performance scheme), they repeated their original mistake –- they adopted yet another vague, complex, evidence-free proposal to replace MIPS.
MedPAC’s history gives us every reason to believe that when they discuss their “repeal and replace MIPS” proposal at their December 2017 and January 2018 meetings, they will refuse to discuss their “replace” proposal in any detail; they will not ask for evidence indicating their proposal is safe and effective; and in their March 2018 report to Congress they will foist upon CMS the dirty work of figuring out how to make their lead balloon fly. CMS will dutifully write up a gazillion pages of gibberish describing how the new program is supposed to work, it won’t work, MedPAC will return to the scene of the crime years later and, pretending they had no part in creating it, propose yet another evidence-free tweak. And so on.
MedPAC is caught in a trap of their own making. They endorse health policy fads without any evidence and without thinking through the details; then when the fads don’t work, rather than review their defective thought process, they endorse other iterations of the fads, again without evidence and without thinking through the details. The tweaked version of the fad fails, and MedPAC starts the cycle all over again. Two analogies for this trap or vicious cycle occur to me. One is the tar pit where mastodons got stuck and died; struggle only caused the dimwitted creatures to sink faster. The other is the hedge fund that gradually becomes a Ponzi scheme. Investors like Bernie Madoff make bad investments, and when the investments go south, instead of admitting their mistakes, they induce their investors to throw good money after bad.
In this comment I will explain why MedPAC’s MIPS-repeal-and-replace scheme deserves ridicule. In my next comment I will describe the process by which MedPAC built the MACRA tar pit or, if you like, the intellectual Ponzi scheme, that now traps them.
The October 4 meeting opened with a statement by MedPAC staffer David Glass about why MIPS was bound to fail. The staff’s most fundamental objection to MIPS is that it’s not possible to measure accurately the “merit” or “total performance” (to use MACRA’s goofball language) of individual doctors. They cited several reasons for this, including: Small sample size; the freedom CMS gives to doctors to select their own “quality” measures from a list of “around 300” (page 5 of the transcript of the meeting ), which guarantees apples are not being compared to apples; and the poor correlation between most of these “quality” measures and, um, quality.
Oddly, Glass’s otherwise thorough critique of MIPS failed to mention the attribution problem. In order to measure physician “value” or “merit,” one must first decide which patients “belong” to which doctor. The MACRA statute instructs CMS to follow a bizarre attribution scheme based on whether the doctor was the “lead” doctor or a “supportive” doctor, and whether the condition was acute or long-term (see my discussion of this section of MACRA here). Understandably, CMS totally ignored these instructions and adopted a slightly less bizarre method of attribution: Medicare beneficiaries are attributed to doctors based on the plurality of primary-care services provided. This plurality method guarantees that doctors are rewarded and punished for patients they never see or, in the case of patients they do see, for treatment decisions made by other doctors. The attribution problem by itself guaranteed MIPS would be difficult to administer and that MIPS scores would be useless or at best difficult to interpret.
Despite overlooking the attribution problem, MedPAC’s staff reached the right conclusion: MIPS’ grading system is useless, both to physicians and to patients. As Glass put it, “[I]t is extremely unlikely that clinicians will understand their score, or what they need to do to improve it” (p. 7). The staff noted in addition how costly MIPS is to physicians. “CMS estimates that the clinician cost to comply with MIPS in the first year of the program is over $1 billion,” reported Glass (p. 6). Glass offered no information what it costs CMS to administer MIPS.
This would have been an ideal time for the staff and commissioners to stop, take a deep breath and ask, “How the hell did we fool ourselves into recommending a whacky pay-for-performance scheme to Congress, and why did it take us two-and-a-half years after MACRA was enacted to admit MIPS can’t work?” Or even: “How could we have done a better job of warning Congress not to pass such a useless bill?” That, of course, is not what happened. There would be no looking back to learn from history.
When Glass was done explaining why MIPS was a disaster in need of repeal, his colleague Kate Bloniarz presented two vague entities that might replace MIPS. Bloniarz and the other staff did not refer to these options as ACOs (they called them “groups”), but I will because they walk and quack like ACOs. The two options were: “Virtual” ACOs that doctors (with gobs of free time on their hands) will start up and “voluntarily” join; and geography-based ACOs that CMS will force upon all the retrograde physicians who refuse to move into “virtual” or existing Medicare ACOs (or other MACRA “alternative payment models”). The only information Bloniarz offered about these entities was that they had to be “sufficiently large to have statistically detectable performance on population-based measures” (p. 12) such as use of “low-value” services, “healthy days at home,” and the ever-popular “potentially preventable admissions” and a cost measure (“relative resource use”), all poorly adjusted for factors outside provider control. Doctors will, allegedly, be inclined to join some flavor of ACO –- virtual, geographic, or existing –- and be rewarded and punished for “performance” on “population measures” because, according to the staff’s proposal, if they don’t they will lose 2 percent of their Medicare FFS payments.
The one feature of Bloniarz’s vague proposal worth praising was the removal of the expensive MIPS reporting requirements. Bloniarz proposed that CMS calculate grades on the “population measures” using claims data only, that is, data doctors are already reporting.
If you were a commissioner and you had been exposed to Bloniarz’s annoyingly vague MIPS replacement proposal, what questions would have occurred to you? You’d want to know, among other things:
- What the staff meant by “sufficiently large;”
- how Medicare beneficiaries would be assigned to the new ACOs;
- whether it’s possible for CMS to risk-adjust accurately the quality and cost scores and, if not, whether ACOs would avoid inviting doctors with sicker and poorer patients to join them;
- whether ACOs of any stripe (existing, virtual, or geographically defined) will be available to all doctors;
- whether doctors in any ACOs, but especially the “virtual” and geographically-based types, would have any way of influencing each other or knowing why their group was being punished or rewarded; and
- whether ACOs anywhere are making enough money to at least offset the expenses of starting and running an ACO so that doctors would have some incentive to join.
With the exception of the attribution question, the commissioners as a group did a good job of posing every important question a rational person would think to ask. (Sloppy attribution is an obstacle to accurate measurement of physician “merit” at both the individual physician level and the ACO or group level.) But, sad to say, not one of those questions was answered by the staff or Chairman Crosson. The staff and Crosson either remained silent, or offered non-answers. To impress upon you how poorly prepared the staff was to defend their vague, Rube Goldberg proposal, I’ll use the rest of this comment to illustrate the non-productive, one-sided conversation between commissioners and staff.
Ask and you shall not receive
The commissioners who spoke first focused on whether it is reasonable to expect that ACOs in any of the three flavors will be available for all doctors to join. The importance of this question is obvious: If doctors are going to be robbed of 2 percent of their Medicare payments for not joining an ACO, they should at least have the opportunity to join one. These commissioners noted that the problem of ACO accessibility is two-fold: ACOs are not available everywhere, and many existing ACOs will want to avoid clinics and hospitals that serve poorer, sicker and more expensive people, such as clinics in rural areas and doctors who specialize in treatment of addiction. All the staff could say was that CMS might be able to create a “fall back option” for clinics and hospitals in rural or poor areas. (Note the staff did not say, “Excellent question, WE will figure out a good answer and not ask CMS to do the dirty work we refuse to do.”)
Commissioner Kathy Buto asked about “isolated providers” who might find it difficult to form groups large enough to create pools of patients large enough for statistical accuracy and, if in the event that large numbers of isolated providers could be herded into one ACO, whether they would feel they had any input into or control over the ACO’s decisions (pp. 18-19). All Glass could say is those problems afflict all ACOs. Commissioner Dana Gelb Safran asked, “[A]re we creating a kind of mini-version of the problem we had with the SGR where individuals really aren’t accountable to each other, even though they are grouped together…?” (p. 58) No one answered her.
Commissioner David Nerenz asked how big an ACO’s pool of attributed patients would have to be to make “population measures” accurate (by MedPAC’s low standards, I would have added). Bloniarz answered 1,000 to 10,000 depending on the “measure.” When Nerenz asked how it would be possible to apply a common set of measures to all ACOs if some didn’t have big enough patient pools for all measures, Bloniarz had no answer. When Nerenz rephrased his question to ask how many doctors an ACO would have to have, Glass replied, “It’s hard to say.”(pp. 20-22) When Nerenz asked how the staff proposed to weight the half-dozen “quality” measures and the cost measure prior to deriving an arbitrary “composite” score, Bloniarz excused the staff’s inability to answer on the ground that their proposal is “a bit exploratory.”
When Commissioner Pat Wang asked the only question any commissioner raised about attribution, staff dodged again. Wang asked if attribution would be prospective (meaning ACOs know in advance which patients are “theirs”) or retrospective (ACOs don’t find out till the end of the performance year who “their” patients were). Bloniarz replied, “I don’t think we have weighed in on [that issue].” “I don’t think we had a reaction to that,” added Director Mark Miller helpfully. (pp. 24-25)
When Commissioner David Grabowski asked what would happen to providers who take care of a disproportionate number of dual eligibles, Bloniarz replied, “I don’t think we have a great answer for that,” and then added wistfully, “[T]here would be risk adjustment,” as if CMS’s grossly inaccurate risk-adjustment method would solve the problem. (p. 29) Grabowski seemed dissatisfied. “If you don’t get the risk adjustment right …, you’re going to magnify disparities,” he replied. “You’re going to widen that gulf between the haves and the have nots.” (p. 31) Staff had no further comment.
Nerenz also warned that inaccurate risk adjustment put the poor and the sick at risk. He urged the commission to adopt criteria for the “quality” and cost measures that would guarantee they can measure quality and cost accurately. Nerenz quoted an article from BMJ that found that failure to adjust readmission rates for socio-economic factors punished hospitals in poorer regions. He warned that if the staff’s proposal was “rolled out” today it would “exacerbate socio-economic disparities.” (p. 72) Nerenz also cited a paper that found that quality of care is better in small clinics than in large networks, and yet the staff’s proposal will encourage further consolidation of the medical sector into large groups. The only reply Nerenz got was a “thank you for that” from Crosson and an “amen” from Commissioner Coombs.
Commissioners Safran and Craig Sammitt asked how the virtual and geographic ACOs would differ from existing ACOs. Neither Crosson nor staff answered.
Summarizing the unintelligible
At the end of the MIPS segment of the October 4 meeting, Chairman Crosson attempted to summarize the unproductive conversation that he and his staff had forced upon the commissioners. He began by saying “I’m going to try to summarize where I think we are. (p. 83) ….We have very close to consensus that MIPS should be repealed.” After that, his ability to construct complete, meaningful sentences deteriorated rapidly. I quote him at length so you can see for yourself what I mean: “I think we have consensus that it would be good to advance population health as the basis for accountability.  …. Where we had a difference was like how to do that –- well, I guess, whether we can do that at all with a replacement which would be the – – well, let’s just replace MIPS –- I mean let’s just eliminate MIPS and leave nothing with respect to measuring accountability for cost and quality in that practice environment. I think my notion was that we could take that direction, but we ought to try, before we do that, to think through whether or not there are some ideas that we could put forward which would — I guess I ‘m not sure I like the term ‘replacement of MIPS’ but to substitute something ….” (p. 85)
Will the other 16 commissioners insist on answers to their questions before they authorize the staff to write up some vacuous version of the staff’s MIPS replacement proposal and send it over to Congress? Don’t bet on it. The commission has caved into their staff and chairman repeatedly throughout its history. The commissioners are capable of asking obvious questions about the half-baked proposals their staff and chairman cook up, but they have never shown an ability to force the chairman and staff either to clarify a proposal and cite evidence for it, or to throw it away.
In my next comment I will explore the history of this habitual failure. I will focus on the commission’s endorsement of pay-for-performance in 2003 and how that endorsement led the commission into the MACRA tar pit.