H.R. 2, Medicare Access and CHIP Reauthorization Act of 2015

Title I

“(A) IN GENERAL.—[T]he Secretary shall establish an eligible professional Merit-based Incentive Payment System (… ‘MIPS’) under which the Secretary shall—

“(i) develop a methodology for assessing the total performance of each MIPS eligible professional … for a performance period … [of] a year;

“(ii) using such methodology, provide for a composite performance score … for each such professional for each performance period; and

“(iii) use such composite performance score of the MIPS eligible professional for a performance period for a year to determine and apply a MIPS adjustment factor ….


H.R. 2, the bill passed by the House of Representatives late in March to repeal the Sustainable Growth Rate formula, instructs the Secretary of Health and Human Services to measure the “total performance” of hundreds of thousands of doctors every year. “Total performance” is to be measured by a “composite performance score.” This score will be some number between zero and 100.

According to the authors of H.R. 2, “total performance” refers to both the cost and quality of care. It is extremely difficult and costly to measure accurately either “total cost” or “total quality” alone, especially at the level of the individual doctor (as opposed to large groups of doctors). Combining an inaccurate score for quality with an inaccurate score for cost to derive a “composite performance” is not a good idea. But even if each score were accurate, it would still not be a good idea because the decision about how much weight to give to each score is arbitrary. In the Infinite Wisdom of Representatives John Boehner and Nancy Pelosi, who negotiated the final version of H.R. 2, the cost score will account for 30 percent of the composite score.

If Boehner and Pelosi had proposed that CMS share the grossly inaccurate and arbitrary composite score with physicians privately, the worst we could say about it is that it will be a waste of money. No human being, including doctors, can make use of feedback that is inaccurate. But Boehner and Pelosi are proposing to use the score to publish report cards listing “good” and “bad” doctors, and to punish “bad” doctors by withholding 9 percent of their reimbursement and using the savings to reward “good” doctors with a 9 percent increase. This pay-for-performance scheme is the heart of H.R. 2’s so-called Merit-based Incentive Payment System (MIPS).

The negative consequences will vastly outweigh any positive consequences. Costs will rise, physician morale will be further damaged, sicker and poorer patients of all ages will be harmed, and concentration within the health care system will increase as rising administrative costs force small clinics to close and join large hospital-clinic fiefdoms.

To construct a “composite score” on each doctor for cost or quality, the HHS Secretary will have to solve several difficult issues. Of these, the most important are:

  • Determining which patients “belong” to which doctors (the “attribution” problem); and
  • adjusting grades for factors outside physician control (the “risk adjustment” problem).

I’ll focus the rest of this comment on the attribution problem. A brief explanation of its mind-boggling complexity should be enough to cause reasonable people to oppose MIPS.

Here are H.R. 2’s instructions to the Secretary on how to attribute patients to doctors who bill Medicare:

In order to facilitate the attribution of patients … to … physicians or applicable practitioners…. [t]he Secretary shall develop patient relationship categories and codes that define and distinguish the relationship and responsibility of a physician or applicable practitioner with a patient…. Such patient relationship categories shall include different relationships of the physician or applicable practitioner to the patient (and the codes may reflect combinations of such categories), such as a physician or applicable practitioner who—

“(i) considers themself to have the primary responsibility for the general and ongoing care for the patient over extended periods of time;

“(ii) considers themself to be the lead physician or practitioner and who furnishes items and services and coordinates care furnished by other physicians or practitioners for the patient during an acute episode;

“(iii) furnishes items and services to the patient on a continuing basis during an acute episode of care, but in a supportive rather than a lead role;

“(iv) furnishes items and services to the patient on an occasional basis, usually at the request of another physician or practitioner; or

“(v) furnishes items and services only as ordered by another physician or practitioner. [emphasis added]

Note first of all how different H.R. 2’s attribution method is from the methods used to attribute patients to “accountable care organizations” and “medical homes.” The standard method used by Medicare and Medicaid to attribute patients to ACOs and “homes” is to attribute patients to the primary care doctor who provides the plurality of primary care services, measured either by visits or expenditures, to the patient during a one-year period. (In the case of ACOs, the patient is further attributed to an ACO if the doctor belongs to one.)

But H.R. 2 uses the phrase “patient relationship.” The Secretary is to develop “patient relationship” codes based on at least two criteria: whether the doctor’s role is a “lead” role or a “supportive” role; and whether the patient’s condition is acute or chronic. “Supportive” doctors are further divided into those who order services on their own versus those who order or provide services on the orders of another physician.

As if this weren’t vague enough, the Secretary is authorized to create codes that combine these categories. Thus, a “lead doctor,” say a primary care doctor caring for a patient with coronary artery disease over a long period of time, might bill as a hybrid “lead-supportive” doctor during a heart attack (an “acute episode”), at which time much of the “primary responsibility” for the patient shifts to a cardiologist. How will the division between “lead” and “supportive” be determined? The mind bridles and balks. But let us push on.

Despite the odd language in H.R. 2 about what a doctor “considers themself,” it’s a safe bet that doctors won’t be allowed to “consider” any relationship they like and enter the code for that relationship on the claim form (doctors will have to enter a relationship code on every claim). Because doctors will have every reason to think the Secretary’s risk-adjustment scheme will not protect their composite score from being dragged down by sicker patients, they will have an incentive to “consider” that they were not the “lead” doctor for difficult, sicker patients and, conversely, that they were the “lead” doctor for easier, healthier patients.

So, if doctors are not going to be allowed to select any relationship that appeals to them, the Secretary will have to develop “percent of services” attribution algorithms that resemble those in use now in the Medicare and Medicaid ACO and “medical home” pilots. These algorithms are already causing problems for ACOs, which consist of dozens and even hundreds of doctors, and which stand to lose only a percent or two of their incomes. The problems these algorithms will cause doctors under MIPS are much more severe. Under MIPS, individual doctors will eat all losses, and these losses could amount to 9 percent of their Medicare income.

ACOs are complaining about attribution algorithms because the use of a plurality threshold means many patients are assigned to doctors who really are not the patient’s primary doctor (see the request for an “attestation” requirement in this letter from the National Association of ACOs, p. 5). Consequently, many patients assigned to an ACO visit doctors outside of the ACO. Analysts and business consultants refer to this problem as “leakage.”

The small body of research on this “leakage” problem indicates it is serious. One study that simulated leakage under Medicare’s ACO algorithm estimates it amounts to 30 percent. In other words, of the visits CMS assigned to ACOs, only 70 percent actually occurred to providers within the ACO. The other 30 percent “leaked” – they saw providers outside the ACO.  For specialists, the leakage rate is 67 percent according to a study published last year.

A close cousin of the ACO “leakage” rate is the ACO “churn” rate – the rate at which patients are assigned to a different ACO each year. The estimated annual churn rate for the 10 large hospital-clinic chains that participated in the Physician Group Practice Demonstration (regarded widely as a test of the ACO concept) was 25 percent (see Section II.E of CMS’s final rule for the Medicare Shared Savings Program, p. 67861.)

It is possible to reduce “leakage” and “churn” by using a formula that attributes patients to doctors who provide a high percent of all services to a patient rather than the under-50-percent threshold used now. (Minnesota uses a 20-percent threshold for its Medicaid “home” program.) Such a method would attribute only the most “loyal” patients to doctors. But this would create another problem: Relatively few patients could be attributed. Which would mean Boehner and Pelosi’s pay-for-performance scheme would apply to only a small minority of patients and would, therefore, affect only a small portion of the average doctor’s Medicare income.

MIPS, ACO, and “home” advocates must choose their poison: They can choose an attribution formula that cannot determine accurately which patients belong to which doctor but which will maximize the financial pressure on doctors; or they can choose a formula that will attribute far fewer “phantom” patients to doctors but which will greatly reduce the number of patients assigned to doctors and, consequently, the financial pressure doctors will be under to “perform.”

Let me close with a brief description of the two remaining calculations that will determine the “composite score” required by H.R. 2: Adjusting physician scores for factors outside their control, and merging the quality and cost scores into a “composite score.” The risk-adjustment calculation will be crude; the weighting of the composite score by cost and quality will be arbitrary.

As I mentioned at the outset, the MIPS pay-for-performance scheme depends not only on an accurate attribution method, but on an accurate method of risk-adjustment – adjustment of physician cost and quality scores for factors outside physician control such as patient health, income, and breadth of insurance coverage. Even the best risk-adjustment schemes are deplorably inaccurate. Medicare’s risk-adjustment scheme for the Medicare Advantage program, the most studied scheme in America and probably the world, can predict no more than 11 percent of the variation in expenditures among Medicare enrollees (see page 8 here).

But as is the case with the attribution problem, there is no feasible solution to the risk-adjustment problem. Improved risk-adjustment will require the collection of much more medical and demographic data on all patients, which will be very expensive.

Finally, let us ask by what logic or moral principle the House of Representatives decided to give the cost score 30 percent of the weight of the composite score. The question is rhetorical. There is no rational explanation for that choice.

In sum, the MIPS composite score will be a meaningless number for three reasons: The attribution method will be grossly inaccurate, the risk-adjustment method will be grossly inaccurate, and the useless cost and quality scores these methods will produce will be mashed together by an arbitrary 70-30 weighting ratio.

We will pay a heavy price for this latest experiment in the never ending experiment with managed care. MIPS will make the Sustainable Growth Rate formula look like a stroke of genius.

Kip Sullivan, J.D., is a member of the board of Minnesota Physicians for a National Health Program. His articles have appeared in The New York Times, The Nation, The New England Journal of Medicine, Health Affairs, the Journal of Health Politics, Policy and Law, and the Los Angeles Times.