By KIP SULLIVAN, JD

The managed care movement thrives on misleading words and phrases. Perhaps the worst example is the incessant use of the word “quality” to characterize a problem that has multiple causes, only one of which might be inferior physician or hospital quality. [1] To illustrate with a non-medical analogy, no one would blame auto repair mechanics if 50 percent of their customers failed to bring their cars in for regular oil changes. We would attribute the underuse of mechanics’ services to forces far beyond the mechanic’s control and would not, therefore, refer to the problem as a “quality” problem.

But over the last three decades it has become acceptable among American health policy experts and policy-makers to characterize any measurement of under- or over-use of medical care, or any measurement of a medical outcome, no matter how poorly adjusted to reflect factors outside provider control, as an indication of “quality.” The widespread, inappropriate use of “quality” long ago set off a vicious cycle. It helped spread the folklore that the quality of America’s doctors and hospitals is awful, and that in turn was used to justify taking even more crude measurements of quality, and so on. [2]

In the early 1990s when the measurement craze was limited to a few “quality” report cards, and grades on those report cards were not used to punish or reward doctors and hospitals, the craze posed less risk to providers who treated a disproportionate share of the poor and the sick. But over the last 15 years, inaccurate “quality” grades on report cards have been linked with financial penalties and rewards. Payment schemes that link rewards and punishments to quality scores were dubbed “pay-for-performance” (P4P) circa 2003. Payment schemes that link rewards and punishments to both quality and cost scores, which began to materialize in the late 2000s, were dubbed “value-based purchasing” (VBP) schemes. (“Value” implies both quality and price have been taken into account). These VBP schemes go by various names, the most important being “accountable care organization” and “medical home.” MACRA is one big VBP scheme.

A growing body of research indicates P4P and VBP schemes (I will refer to both from now on as VBP schemes) pose a much greater threat to the health of poorer and sicker Americans than the old report cards did. They do so because they pose a much greater threat to the financial health of the providers who treat poorer and sicker Americans. Because minorities represent a disproportionate share of our nation’s sickest and poorest, and because the responsibility for treating minorities falls disproportionately on a few clinics and hospitals, many of them staffed by minority health care professionals, VBP schemes punish minority providers for no reason other than that they see more minority patients.

In this article I will document what I have just said by focusing on black doctors as well as hospitals and ACOs that serve a large number of black patients. I will argue that those who promote VBP are guilty of racial profiling – punishing black doctors and other providers who treat a high proportion of black patients and, ultimately, black patients. We will see that three intractable conditions guarantee that VBP penalizes black providers and their patients:

(1) Black physicians, and a small number of hospitals, treat a disproportionate share of black patients;

(2) black patients tend to be sicker and poorer than white patients, which means the clinics and hospitals that treat a disproportionate number of them have to spend more on them and yet have fewer resources to do that, and are therefore less likely to score well on cost and quality measures; and

(3) administrators of VBP schemes are incapable of adjusting quality and cost scores accurately to take into account factors outside provider control, including the lower health status, lower incomes, worse insurance, worse access to transportation, and greater exposure to stressors such as crime and food insecurity suffered by black Americans.

Note that I am not criticizing measurement per se. Measurement is essential to quality improvement in every field of human endeavor. I’m criticizing inaccurate measurement and the reckless use of inaccurate measurement.

Intractable problem 1: High concentrations of black patients

Let us begin with problem 1. It is well established that a minority of the nation’s clinics and hospitals treat a disproportionate share of black patients. Peter Bach et al. found that “22 percent of primary care physicians … account for 80 percent of all visits by black [Medicare] patients….” Moreover, Bach et al. found that those primary care physicians were likely to be black. As Bach et al. put it, “visits by black patients were markedly more likely than visits by white patients to be to black physicians (22.4 percent vs. 0.7 percent).” (p. 579)

The concentration of black patients at the hospital level is slightly higher than it is at the clinic level. Ashish Jha et al. reported : “The 5 percent of hospitals with the highest volume of black patients cared for nearly half of all elderly black patients, and the hospitals in the top quartile by volume of patients cared for nearly 90 percent of elderly black patients. Hospitals with a high volume of black patients were … more often teaching hospitals located in the southern United States …. Hospital care for black patients in the United States is remarkably concentrated in a small percentage of hospitals”

If concentration of black patients is high at the clinic and hospital level, it follows it should be high at the ACO level as well (because ACOs consist of hospitals and clinics). A recent paper by Valerie Lewis et al. found that among 306 Medicare Shared Savings Program ACOs, “a small number … had very high proportions of minority patients,” and that blacks constituted 60 percent of those minority patients. “Each ACO in the top 5 percent of ACOs in terms of proportion of minority patients had a patient population with more than 50 percent minority patients,” according to Lewis et al. (p. 60)

Intractable problem 2: Sicker and poorer patients, fewer resources

I’m sure every reader of this blog is well aware of the ocean of evidence indicating black Americans are in poorer health than the average American. Their morbidity and mortality rates are far above the American average. [3] Other conditions outside physician and hospital control that affect health, such as income and insurance status, are also worse among blacks.

Perhaps the worst of these other non-health factors is the low level of resources available to providers who treat a disproportionate share of black patients compared with other providers. Black physicians and other high-black-proportion providers must not only treat more difficult and expensive patients, but they must do so with less income and less support. The study I mentioned above by Bach and colleagues determined that black physicians are significantly less likely to say they have enough resources to provide high-quality care and yet they provide more charity care than other primary care doctors. [4] Similarly, Lewis et al. reported gigantic differences in resources available to “ACOs with a high proportion of minority patients and … other ACOs.” The former “were significantly less likely than other ACOs to offer … routine specialty care (57 percent versus 75 percent), outpatient rehabilitation (29 percent versus 50 percent), pediatric health care (40 percent versus 59 percent), and palliative and hospice care (30 percent versus 53 percent).” (p. 62)

The reason why clinics, hospitals and ACOs that treat a high proportion of black patients have fewer resources should be obvious. Black people are more likely to be uninsured, to be on Medicaid if they have insurance (Medicaid reimburses at very low rates compared even to Medicare), and to be unable to afford out-of-pocket payments. But as we shall see in my next comment in this two-part series, managed care advocates routinely ignore or soft-pedal the resource problem. Acknowledging the resource problem would rebut, or at least undermine, the argument that VBP will save money. Managed care advocates would much rather place the blame on doctors and hospitals because demonizing providers makes it easier to justify the managed care solution – shifting insurance risk onto providers and micromanaging them with various VBP schemes. God forbid we should admit the problem is a resource problem, not a “quality” problem. What reason would managed care advocates have to get out of bed in the morning if America didn’t suffer from rampant “quality” problems requiring the expertise of managed care advocates to solve?

Intractable problem 3: Lousy risk adjustment

If the science of risk-adjustment were advanced enough to adjust cost and quality scores accurately to take account of patient health status and other factors outside provider control, the presence of the first two problems I have just discussed wouldn’t matter. Risk-adjustment algorithms would adjust away both problem 1 (a small number of providers treat a large share of the black population) and problem 2 (black patients tend to be sicker and poorer).

But no one has produced a remotely accurate risk adjuster, and no one ever will. The best risk-adjustment system in America, and possibly the world, is the one CMS began using a decade ago to adjust capitation (aka premium) payments to Medicare Advantage plans. This method, known as the hierarchical condition category (HCC) method, can only explain 12 percent of the variation among Medicare beneficiaries (see Table 2-1 p. 6 here.) This means that CMS overpays Humana et al. for healthier enrollees and underpays for sicker enrollees. Here’s how the Medicare Payment Advisory Commission (MedPAC) put it in its June 2014 report to Congress: “[T]he CMS–HCC model predicts costs that are higher than actual costs (over-predicts) for beneficiaries who have very low costs and lower than actual costs (under-predicts) for beneficiaries who have very high costs.”

MedPAC illustrated this statement with the data in the table below. The table indicates CMS’s risk-adjustment method results in overpayments equal to 62 percent for the healthiest quintile of enrollees and underpays by 21 percent for the sickest 1 percent. Mind you, MedPAC discovered this much inaccuracy when they applied CMS’s HCC method to a pool of 12 million Medicare beneficiaries (see the discussion of MedPAC’s methodology in the sidebar on pages 28 and 29 of their June 2014 report). How bad does it get when CMS and other VBP advocates apply their crude risk adjusters to pools of 20, 25 and 50 patients?

Gross inaccuracy plagues quality measurement as well. For evidence of the crudeness of one of the best risk-adjustment methods in the world for a quality report card, see my discussion in an earlier comment on THCB  about the New York Department of Health’s annual report cards on heart surgery. That report card, considered to be the gold standard for quality measurement, cannot distinguish the vast majority of the 185 New York coronary artery bypass surgeons from one another, nor can it distinguish the vast majority of the 40 hospitals where this procedure is performed from one another. (This is true even though the NY Department of Health relies on rolling three-year periods of data, which of course makes the reports somewhat dated by the time they are released.) Nevertheless, New York’s report card has been shown to harm sicker patients in general and black patients in particular (see Rachel Werner et al., “Racial profiling: The unintended consequences of coronary artery bypass graft report cards,” Circulation, 2005;111:1257-63).

The gross inaccuracy of quality measurement plagues quality measurement in all of CMS’s VBP programs, including their ACO, medical home, and “excess readmission” programs. Unlike the old days when report cards were simply published in newspapers or online where patients would ignore them, these days, thanks to VBP advocates, CMS punishes ACOs, “homes” and hospitals that get inferior grades on CMS’s crude report cards.

Medicare ACOs, for example, lose a portion of the shared savings (assuming they achieve any) if they don’t score well on CMS’s “quality” measurements, the great majority of which are not risk adjusted at all [5]. But risk adjustment is so inaccurate that even when it is applied to CMS’s ACO “quality” measurements it makes little difference. In the paper by Valerie Lewis et al. I cited above, the authors reported that the patients in the high-minority-proportion ACOs were much sicker than patients in other ACOs and that CMS’s crude HCC risk-adjustment method made almost no difference in the “quality” scores CMS attributed to the ACOs. The authors reported that patients in the high-minority ACOs “were more likely to be … dually eligible for Medicare and Medicaid, disabled, and female and to have end-stage renal disease…. In addition, the HCC scores for patients ages 65 and older who were not dual eligibles and for disabled patients were higher in ACOs with a high proportion of minority patients….” (p. 60).

These differences cry out for accurate risk adjustment. But none is available. Lewis et al. reported that high-minority ACOs suffered lower “quality” scores almost across the board, and that risk adjustment via CMS’s HCC method (plus adjustments for the percent of ACOs doctors in primary care) made virtually no difference. “The proportion of minority patients was associated with score performance … on 29 of 36 measures unadjusted for patient and ACO characteristics,” they report, “and on 25 of 36 measures… adjusted for those characteristics.” Lewis et al. concluded, “[A] large array of other unmeasured patient characteristics – including education, income and wealth, social support, neighborhood resources, and health literacy – may explain some or all of the associations reported here.” (p. 59)

Similarly, the VBP craze punishes hospitals with sicker, poorer, and blacker patients. One of the VBP fads authorized by the Affordable Care Act, and now inflicted on hospitals that treat Medicare patients, is the Hospital Readmission Reduction Program (HRRP). This program punishes hospitals for “excess” readmissions (an alleged quality measure) of patients with specific diagnoses such as congestive heart failure. But CMS’s attempt at risk adjustment of readmission rates is pathetic, in part because CMS thinks it is useful and fair to use pools of patients as small as 25 to determine what constitutes “excessive” readmission rates (see MedPAC’s June 2013 report to Congress ) “The result,” according to MedPAC, “is a great deal of random variation in these condition-specific rates.” (p. 102)

According to a paper published by NEJM Catalyst hsponsored by the Missouri Hospital Association, the risk adjuster CMS uses to determine “excess” readmissions for congestive heart failure (CHF) explains only 18 percent of the variation in readmission rates. That leaves room for “a great deal of random variation,” as MedPAC put it, which means hospitals that treat sicker and poorer patients suffer a disproportionate share of the penalties CMS dishes out, and that in turn means hospitals that treat a disproportionate number of black people suffer a disproportionate share of the penalties.

Manipulative language facilitated this mess

Lives are at stake here! Research  at the University of Michigan suggests the HRRP program is killing up to 5,000 CHF Medicare patients annually. A disproportionate percent of those deaths are probably suffered by black people.

The managed care movement’s penchant for manipulative language played a substantial role in creating this mess. The problems managed care advocates glibly call “quality” problems are often resource problems or other problems clinics and hospitals cannot fix – that only society can fix. I will discuss the role that manipulative language has played in downplaying the resource problem and vastly exaggerating the “quality” problem in my next comment.

[1] For example, in a 2003 paper entitled, “The quality of health care delivered to adults in the United States,” Elizabeth McGlynn and colleagues at the RAND Corporation found that patients received recommended care only during 55 percent of their encounters with doctors. For example, the authors found that doctors referred alcoholics for “further treatment” during just 5 percent of the visits by their alcoholic patients.

Although McGlynn et al. made no effort to determine the causes of so much underuse (nor of the far fewer examples of overuse), they made it clear they had already made up their minds the problem was primarily physician incompetence. They conveyed this message implicitly by using the word “quality” over and over to describe the problem they were studying, and explicitly by calling for supervision of “clinical decision making” via information technology. (p. 2644)

[2] An example of how measurement advocates rely on the widespread abuse of “quality” to justify more inaccurate measurement appeared in a comment by four representatives of employers posted here on THCB on October 6: “Given the widely acknowledged … quality-of-care issues produced by the fee-for-service system, the fact that there are rough spots on the road to value-based payment is hardly a justification for slowing down reform.”

[3] Here is an excerpt from a recent literature review on disparities in cardiovascular health:

Disproportionate rates of cardiovascular disease (CVD) are seen in the U.S. African American population. African American subgroups have a greater burden of myocardial infarction, heart failure, stroke, and other cardiovascular events. On a population level there is a higher overall prevalence of risk factors that are unrecognized and therefore not treated, which places these individuals at a greater likelihood of experiencing adverse outcomes and therefore potentially higher morbidity and mortality. Among the various minority population groups, African American men have the highest overall death rate from cardiovascular disease. This extends to African American females, whose death rates from CVD are higher than those of white females. Equally as concerning is the fact that deaths also occur much earlier in African Americans compared to whites.

African American adults have among the highest rates of hypertension in the world. They and Mexican Americans tend to have higher blood pressure and exhibit lower levels of blood pressure control than whites in the U.S., even after consideration of modifiable health behaviors, suggesting that other racial/ethnic differences underlie these disparities.

[4] Peter Bach et al. found that physicians treating black patients had fewer resources available to them and yet provided more charity care. “Physicians treating black patients provided more charity care, derived a higher percentage of their practice revenue from Medicaid, [and] more often practiced in low-income neighborhoods … than physicians treating white patients,” they reported. “When physicians were asked if they were able to provide access to high-quality care for all of their patients, 27.8 percent of physicians treating black patients responded that they could not do so …, as compared with 19.3 percent of physicians treating white patients. They were also more likely than physicians treating white patients to report that they could ‘not always’ provide access for their patients to subspecialists of high quality (24.0 percent vs. 17.9 percent), high-quality diagnostic imaging (24.4 percent vs. 16.6 percent), nonemergency hospital admissions (48.5 percent vs. 37.0 percent), and high-quality ancillary services (36.6 percent vs. 27.7 percent).”

[5] Valerie Lewis et al. (see link to this paper in the text) stated that MSSP ACOs are graded on “33 Medicare ACO quality metrics and two disease composite measures,” and these “quality measures are not risk adjusted – except for unplanned readmissions and the two measures of admissions, which are adjusted for case mix.” (p. 58)