FEATURES:
Are Doctors Biased?

By Sally Satel and Jonathan Klick

The factors underlying disparities in treatment

Two 50-year-old men arrive at an emergency room with acute chest pain. One is white and the other black. Will they get the same quality of treatment and have the same chance of recovery? We would hope so, but many experts today insist that their race will profoundly affect how the medical-care system deals with them and that the black patient will get much inferior care. Is this really true? And if so, why? Are differences in treatment due to deliberate discrimination or other (less invidious) factors?

Interest in the determinants of minority health has grown considerably since the publication of the Report of the Secretary’s Task Force on Black and Minority Health by the U.S. Department of Health, Education, and Welfare in 1985. The academic literature falls into two categories. One line of inquiry emphasizes overt or subtle racial discrimination by physicians. Research reports in this category assert that many physicians treat their white patients better than their minority patients on the basis of race alone. We call this the “biased-doctor model” of treatment disparities.

The other line of research focuses on the influence of so-called third factors that are correlated with race. These factors can influence care at the level of the health system, the physician, or both. They include, for example, variations in insurance coverage (insured versus uninsured versus underinsured; public versus private health plans; profit versus not-for-profit health plans), quality of physicians, regional variations in medical practices, and patient characteristics (such as clinical features of disease or health literacy).

Of course, it is possible that both of these mechanisms — “biased doctors” and “third factors” — could operate simultaneously. Yet it is the biased-doctor model that has acquired considerable weight in both academic literature and the popular press. It enjoyed a great boost in visibility from a 2002 report from the Institute of Medicine (iom), part of the National Academy of Sciences. The iom provides lawmakers with advice on matters of biomedical science, medicine, and health and issues high-profile reports written by panels of outside experts. Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care was widely hailed as the authoritative study on health disparities. It concluded that the dynamics of the doctor-patient relationship — “bias,” “prejudice,” and “discrimination” — were a significant cause of the treatment differential and, by extension, of the poorer health of minorities. Much media fanfare greeted the iom report, and virtually every story ran the triumphant remark of Dr. Lucille Perez, then president of the National Medical Association, which represents black physicians: “It validates what many of us have been saying for so long — that racism is a major culprit in the mix of health disparities and has had a devastating impact on African-Americans.”

In this essay, we evaluate some of the studies routinely put forth as evidence of harmful discrimination. We report evidence not considered by the iom panel. These additional findings indicate that race-related variables, such as geography and socioeconomic status, shine important explanatory light into the recesses of the treatment gap. Without adequate controls for just such variables, it is simply not possible to distinguish care patterns that correlate with race from those that are due directly to race. Indeed, as we will see, when researchers employ designs that control for more third factors, the magnitude of any race effect shrinks considerably, if it does not disappear altogether.

Public health as civil rights

First, a brief sketch of how we got here. Just before Christmas 2003, the Agency for Health Care Research and Quality of the U.S. Department of Health and Human Services (hhs) released the National Healthcare Disparities Report. It documented an all-too-familiar problem: the poorer health status of individuals on the lower rungs of the socioeconomic ladder and the fact that they often receive inadequate treatment compared to people with more resources and education.

The report sparked a heated controversy over whether hhs had downplayed the charge of racial bias in the health-care system. At issue were revisions made in a prepublication draft shortly before its release. Those consisted, among other things, of using the more neutral word difference instead of disparity to describe discrepancies between the health of whites and minorities. This might seem like an innocuous substitution, but it was not. In public health circles, the word “disparity” has come to connote unfair difference due to a patient’s race or ethnicity. It “has begun to take on the implication of injustice,” observed epidemiologist Olivia Carter-Pokras at the University of Maryland.¹ Architects of the agency report, however, argued that the more neutral term difference more accurately describes the findings of the report.

The switching of difference and disparity prompted Henry Waxman, ranking minority member of the House Government Reform Committee, to send a harsh letter to then-hhs secretary Tommy Thompson. The word substitution, Waxman wrote, “alter[ed] the report’s meaning . . . and fit a pattern of the manipulation of science by the Bush Administration.”² The revision also set alarm bells ringing among a range of constituencies. “By tampering with the conclusions of its own scientists, hhs is placing politics before social justice,” wrote members of the Congressional Black Caucus, Congressional Asian Pacific American Caucus, and Congressional Hispanic Caucus in a joint press release.³ The National Medical Association pronounced itself “appalled.”⁴ Physicians for Human Rights bemoaned “remov[al] from the text [of] any inference of prejudice on the part of providers, and [its] focus on individual responsibility for disparities.”⁵

The critics who scolded hhs for its revised executive summary cited the 2002 iom report as proof that bias was common among physicians. To be fair, the iom report acknowledged the roles of other factors in minority health, but it put heavy emphasis on the failure of the medical profession to purge its ranks of prejudice — a shortcoming that was, as the report put it, “rooted in historic and contemporary inequities.”

Although the iom report is now the most widely cited source for this claim, it was hardly the first to make the argument. A decade earlier in the Journal of the American Medical Association, Secretary of Health and Human Services Louis Sullivan cast minority health as a civil rights issue, writing, “There is clear, demonstrable, undeniable evidence of discrimination and racism in our health care system.”

The Reverend Al Sharpton warned in 1998 that “health will be the new civil rights battlefront”; that same year, President Clinton remarked in a radio address delivered during Black History Month that “nowhere are the divisions of race and ethnicity more sharply drawn than in the health of our people” and speculated that one of the causes might be “discrimination in the delivery of health services.” In the March/April, 2005 issue of Health Affairs, Senator Ted Kennedy urged that “greater resources should be given to the hhs Office for Civil Rights.” And, in an especially alarmist tone, Marian Wright Edelman of the Children’s Defense Fund told the 2005 graduating class of Colgate University that “the new racism that is seeping across our country is wrapped up . . . in racial disparities in health.”

This social justice perspective continues to frame the issue of minority health. For example, introducing the Health Care Equity and Accountability Act in 2003, Senator Tom Daschle cited the need to correct doctors’ “bias,” “stereotyping,” and “discrimination.” The American Medical Association felt moved to reaffirm its “long-standing policy of zero tolerance [against] racially or culturally biased health care.”⁶ The American Public Health Association “call[ed] on the President and the Congress of the United States to recognize and promote legal redress for discrimination in health and health care.”⁷ On the research front, the National Institutes of Health are funding research on “the effect of racial and ethnic discrimination on health care delivery.”⁸ In some medical schools, “racial sensitivity” training is now required. And in 2005, New Jersey was the first state to pass a law requiring doctors to receive “cultural competency” training as a condition of obtaining or renewing their licenses to practice medicine. Unfortunately, these institutional mandates and practices legitimate the “biased-doctor model” of health disparities at the expense of other factors.

Limits of the iom report

The most rigorous studies reviewed by the iom sought to control for confounding clinical or economic variables, such as concurrent illness, supplemental insurance, or patients’ refusal to undergo procedures. But because most of the studies were retrospective and relied upon chart review or large Medicare administrative databases, many such variables could not be captured.

Some of these are generally recorded, such as comorbid conditions and severity of disease at the time care is sought. Others are often missing from administrative databases — for instance, ekg subtleties, position of occlusion in carotid and coronary vessels, coronary ejection fraction, and pulmonary function test performance — even though they figure importantly in physician decision making. Moreover — and this is key — these unrecorded variables do vary by race and ethnicity. Note, for example, the well-documented frequency with which coronary angiograms of black patients show less anatomical suitability for intervention — either lesions in the vessels are too diffuse for angioplasty, or the patients have a higher incidence of normal-appearing vessels, despite the clinical appearance of having suffered acute myocardial infarction. An examination of records, therefore, could suggest a racial bias in treatment simply because coronary angiograms are less often given to black patients, and the records themselves do not indicate the reasons for those treatment decisions.

Consequently, the panel concluded that treatment differences occur everywhere and that they are manifest for all kinds of care.⁹ But this conclusion was in error, as other studies indicate. Baicker and colleagues at Dartmouth College, for example, have shown important regional inconsistencies in treatment. One region might display wide race disparities in some procedures, such as hip replacement or back surgery, smaller discrepancies in bypass, and almost no gap in mammograms.¹⁰ Does that mean that doctors in the region who perform hip replacements are biased, but cardiac-care doctors are not? Or is it possible that there are other, benign reasons for those statistical disparities?

Missing variables are not the entire story, however. Other kinds of evidence are necessary to bestow a fuller picture of the dynamics involved in treatment differences and permit more confidence in the iom’s claims about bias on the part of providers. For example, we need many more prospective studies that ask doctors and patients about how they make decisions to offer and to accept, respectively, particular treatments.

Researchers at the Philadelphia Veterans Affairs Medical Center found that more blacks than whites (61 percent versus 29 percent) maintained the folk belief that the spread of lung cancer was accelerated when the tumor was exposed to air during surgery and would oppose surgery because of this.¹¹ A related study conducted at Detroit’s Henry Ford Health System found that black patients with operable lung cancer refused surgery over three times more often than whites (19 percent versus 5 percent), though both whites and blacks were offered the surgery at similar rates.¹²

But tests and procedures are not always offered at the same rate. Asking doctors why they did not order a particular test could yield explanations such as the one offered by Dr. Gary Curhan, writing in the Journal of the American Medical Association (jama) in 2005 about work-ups for first-time kidney stones: “If the patient is uninterested in making long-term lifestyle changes or taking medication, then I do not proceed with an evaluation [for a first stone.]” In other words, the physician decides to undertake an expensive work-up only if a patient is invested in cooperating with the diet and other lifestyle changes needed to improve his condition — otherwise, he treats the patient’s symptoms without looking for a specific cause.

These are just two of countless examples showing how a patient’s characteristics can influence the care he receives. In a sweeping book, The Status Syndrome: How Social Standing Affects Our Health and Longevity (Bloomsbury Publishing, 2005), epidemiologist and physician Sir Michael Marmot documents the importance of factors that are not readily measured by disparity researchers — in part because their accounting requires time-consuming, face-to-face interviewing. For example, Marmot emphasizes the importance of personal autonomy and control over one’s life circumstances. With respect to treatment per se, it is not surprising that patients with chaotic lifestyles — an often inevitable aspect of living in or near poverty irrespective of race — are not going to be good candidates for ongoing care requiring complex regimens.

A third kind of study valuable for understanding race-related factors in treatment compares care provided by white and black doctors to white and black patients. For example, evidence that doctors of both races treat black patients similarly, say, in terms of rate of referral for catheterization — even if both refer black patients less often than they do white patients — would cause us to question a charge of bias. We are aware of only one study that has analyzed data with this question in mind. Jersey Chen and colleagues at Yale University analyzed data from the Cooperative Cardiovascular Project.¹³ They evaluated 40,000 Medicare beneficiaries hospitalized for acute myocardial infarction in 1994 and 1995 to determine whether differences between black patients and white patients in the use of cardiac catheterization within sixty days after acute myocardial infarction varied according to the race of their attending physicians. Black patients had significantly lower rates of cardiac catheterization than white patients, regardless of whether their attending physician was white (38.4 percent rate of catheterization for black patients versus 45.7 percent for whites) or black (38.2 percent versus 49.6 percent).

There was no significant interaction between the race of the patients and the race of the physicians in the use of cardiac catheterization, strongly suggesting that racial bias was not at issue. Critics of the Chen study, however, have suggested that the predominantly white cardiologists to whom the black internists referred their patients exhibited racial bias by undertreating the black patients.¹⁴ To this Chen and colleagues reply by noting this would mean that black attending physicians concurred with and supported racially biased decisions — a scenario they believe unlikely.¹⁵ Moreover, the adjusted mortality rate among black patients was lower than or similar to that among white patients for up to three years after the infarction, suggesting that the care received by the patients, even if it was different, was equally effective.

Is geography destiny?

With most health care delivered locally — and with racial and ethnic groups not evenly scattered about the country — it is imperative that researchers account for geography in evaluations of health disparities. When they do, they discover that geographic residence often explains race-related differences in treatment better than even income or education. One of the most striking limitations of the iom report is the absence of such an analysis.

Consider the concept of the “hospital referral region,” or hrr. The Dartmouth Atlas of Health Care defines an hrr as a geographic area served by a major hospital equipped with comprehensive surgical capacity, also known as a tertiary care hospital. In the United States there are 306 hrrs, yet only 36 of them have a nationally representative mix of residents. Among the rest, a number have black population rates that are three to six times the national average of 13 percent. Because health care varies a great deal depending on where people live, and because blacks are overrepresented in regions of the United States that are burdened with poorer health facilities, disparities are destined to be, at least in part, a function of residence.

Medicare datasets do not include geographic identifiers, so geographic data are often lost to researchers who rely on these sources. Consequently, as Amitabh Chandra and Jon Skinner of Dartmouth College have observed, many disparity evaluations do not sufficiently control for geographic variation among patients.¹⁶ This can produce misleading findings.

For example, assume black patients from city x and city y receive exactly the same care as white patients from the same places. In city x, all patients receive suboptimal care; in city y, all patients receive excellent care.

Now compare the care of all black residents of cities x and y with the care of whites from both cities. If the proportion of black residents in the two cities is not identical, there will appear to be racial differences in treatment even though blacks and whites living in the same place receive the same care. Thus, if minority patients are not randomly distributed throughout locations — only 6 percent of poor whites live in high-poverty neighborhoods, whereas 22 percent of Hispanics and 34 percent of blacks do — geographic differences in utilization and health outcomes are going to appear, analytically, as racial disparities. And researchers who fail to control for location effects will interpret geographic health disparities as racial disparities.

As a rule, the quality of care received by blacks is inversely related to the concentration of black residents in the local population. For example, Baicker, Chandra, and Skinner found that the frequency of annual eye exams in black diabetic patients covered by Medicare declined as the number of blacks in the local population increased.¹⁷ Along these lines, blacks who lived in predominantly white hrrs received the same or slightly better eye care than whites. Angus Deaton of Princeton University and Darren Lubotsky of the University of Illinois have found that at both the regional and the metropolitan statistical area levels, white and black mortality rates are higher in areas where blacks make up a larger portion of the total population. Similarly, the Dartmouth group found significantly higher risk-adjusted mortality following acute myocardial infarction in U.S. hospitals that disproportionately serve black patients.¹⁸ In her study, Amber Barnato and colleagues found that 1,000 of 4,690 hospitals nationwide accounted for treating 85 percent of the black Medicare patients in 1994–95.¹⁹

The effects of location on health disparities have also been studied using infant mortality rates. Jeannette Rogowski and colleagues at rand used the rich Vermont–Oxford network dataset to examine the effects of hospital quality on the mortality rates of very low-birthweight babies, controlling for condition of the baby at birth (via Apgar scores) as well as other characteristics such as gestational age, race, method of delivery, birth defects, and prenatal care.²⁰ The authors found that black babies were more likely to be born in hospitals that primarily served minority areas (57 percent for black births as compared with 18 percent for white births).

Thus, at a minimum, black and white babies are not being delivered at the same kinds of hospitals. The characteristics of the hospitals serving these two populations also varied systematically. Black babies were significantly more likely to be born in government-run hospitals that served a relatively high proportion of Medicaid patients and where doctors spent less time with patients due to high patient volume (and for other reasons as well). Further, the hospitals where black babies were born were significantly less likely to have neonatal intensive care units or to perform neonatal cardiac surgery.

In the Rogowski analysis of 28-day infant mortality rates, these hospital characteristics proved to be a significant source of variation in the survival chances between white and black babies. Babies born in minority-serving hospitals were 30 percent more likely to die in the first 28 days than those born in hospitals that served few minorities (less than 15 percent of patients), and this effect was quantitatively similar for both white and black babies.

Although not nearly as important as the minority-serving versus majority-serving distinction, many other hospital characteristics that differed by race also proved significant in determining mortality. For instance, having a neonatal intensive care unit that performed cardiac surgery reduced infant mortality by 14 percent, and being born in a government-run hospital raised mortality rates by 7 percent relative to a private, not-for-profit hospital and by 24 percent relative to a for-profit hospital. Again, these results included controls for condition at birth, prenatal care, maternal income and education levels, and gestational age.

Thus, by focusing on race we miss a very important cause of health-care difference: geography. Where a person lives, irrespective of race, has a much larger effect on how the medical system treats him.

Variation among hospitals is another factor for which disparity studies often do not control. Indeed, the studies below describe a pervasive trend: Hospitals that treat greater numbers of minority patients generally offer poorer-quality service than those that treat fewer minorities.

In general, hospitals that perform a low volume of surgical procedures such as coronary bypass, gall bladder removal, or valve replacement have higher mortality rates for the given procedures than those that perform more. A 2002 study by John Birkmeyer and others in the New England Journal of Medicine showed that black patients were more likely to be treated at low-volume hospitals and more likely to die for that reason. The crucial importance of volume has been underscored by the Leapfrog Group (a coalition of more than 80 large public and private insurance purchasers), which urges both patients and payers to select hospitals that perform a certain minimum threshold number of procedures per year.

Elizabeth Bradley of Yale and colleagues found that hospital-to-hospital differences made a considerable impact on treatment differentials in the case of suspected heart attack. The cohorts included 37,143 patients receiving angioplasty at 434 hospitals and 73,032 patients receiving fibrinolytic therapy (medicine to dissolve blood clots in coronary arteries) in 1,052 hospitals. Their findings, published in 2004 in JAMA: “A substantial portion of the racial and ethnic disparity in time to treatment is accounted for by the hospital to which a patient is admitted, in contrast to differential treatment by race and ethnicity inside the hospital.”

A nationwide study of all Medicare patients treated in 4,690 hospitals between 1994 and 1995 for acute myocardial infarction revealed a similar finding. On average, black patients went to hospitals that used evidence-based medical treatments (that is, state-of-the-art practices) less frequently and had worse mortality rates (but higher rates of cardiac procedures, suggesting better-quality surgical than medical care). “Incorporating the hospital effect altered the finding of racial disparity analyses and explained more of the disparities than race,” wrote Amber Barnato of the University of Pittsburgh and her coauthors.

Once again, we find that minority patients receive different treatments than whites primarily because they attend lower-quality hospitals — a pattern that helps exonerate physicians from the charge of systematic bias in their treatment of patients. Most likely, this is a function of minorities’ disproportionate poverty or near-poverty status. Studies comparing similarly disadvantaged blacks and groups of whites (such as those clustered in poverty in Appalachia and rural Maine) would underscore the primacy of social capital.

Debunking the “biased-doctor” model

A central assumption that underlies the biased-doctor model is that black patients are treated worse than white patients when served by the same (white) doctor. But research published in 2004 in the New England Journal of Medicine by Peter Bach and colleagues at Manhattan’s Memorial Sloan-Kettering Cancer Center and the Center for the Study of Health Care Change in Washington has produced findings that cast doubt on that assumption. The authors showed that white and black patients, on average, do not even visit the same population of physicians — making the idea of preferential treatment by individual doctors a far less compelling explanation for disparities in health. They showed, too, that a higher proportion of the doctors that black patients tend to see may not be in a position to provide optimal care.

The research team examined more than 150,000 visits by black and white Medicare recipients to 4,355 primary-care physicians nationwide in 2001. It found that the vast majority of visits by black patients — 80 percent — were made to a small group of physicians — 22 percent of all those in the study. Is it possible, the researchers asked, that doctors who disproportionately treat black patients are different from other doctors? Do their clinical qualifications and their resources differ?

The answer is yes. Physicians of any race in the study who disproportionately treated black patients were less likely to have passed a demanding certification exam in their specialty than the physicians treating white patients. More important, they were more likely to answer “not always” when asked whether they had access to high-quality colleague-specialists, such as cardiologists or gastroenterologists, to whom they could refer their patients or to nonemergency hospital services, diagnostic imaging, and ancillary services, such as home health aid.

These patterns reflect geographic distribution. Primary-care physicians who lack board certification and who encounter obstacles to specialized services are more likely to practice in areas where blacks receive their care — namely, poorer neighborhoods, as measured by the median income. Bach and his colleagues suggest that these differences play a considerable role in racial disparities in health care and health status. They make a connection between well-established facts: that physicians who are not board-certified are less likely to follow screening recommendations and more likely to manage symptoms rather than pursue diagnosis. Thus, rates of screening for breast and cervical cancer or high blood pressure are lower among black patients than white, and black patients are more likely to receive a diagnosis when their diseases are at an advanced stage.

Limited access to specialty services similarly puts black patients at a disadvantage. The Bach study is the first to examine physicians’ access to specialty care and nonemergency hospital admissions in light of the race of the patients they treat. That capacities of doctors who treat black patients may account for some part of the health gap was considered in a 2002 study by researchers at the Harvard School of Public Health. The study found that physicians working for Medicare managed-care plans in which black patients were heavily enrolled provided lower-quality care to all patients. Specifically, their patients were less likely to receive the four clinical services the authors measured — mammography, eye exam for diabetics, beta-blocker after myocardial infarction, and follow-up after hospitalization for mental illness.²¹

A report in the American Journal of Public Health in 2000 found that blacks in a sample of almost 30,000 patients in New York State undergoing cardiovascular surgery in 1996 had poorer access to high-quality surgeons than did whites. Even among patients at the same hospital, whites were treated by better-performing surgeons, a phenomenon that may reflect some selection of patients by surgeons based on insurance coverage.²² Donald Gemson of the Columbia University School of Public Health and colleagues showed that foreign-trained physicians and doctors not board-certified were more likely to treat black patients in New York City than to treat whites. They also found that practitioners whose caseload was more than 50 percent black or Hispanic were less likely to follow nationally recognized treatment guidelines, such as recommending mammograms or flu vaccinations for the elderly.²³ Kevin Heslin of Charles R. Drew University and his team showed a correlation between physicians’ experience in treating hiv and the race of their hiv patients, with hiv-positive black patients more likely to be treated by physicians less experienced with the disease.²⁴

At the Center for Studying Health System Change in Washington, D.C., J. Lee Hargraves and colleagues used the Community Tracking Study Physician Survey, a nationally representative study of American physicians, to assess their abilities to obtain medically necessary services for their patients.²⁵Physicians were asked how often they could arrange referrals to specialists and inpatient admissions for their patients. According to the survey, black physicians were more likely to report difficulties admitting patients to hospitals than white physicians, and Hispanic physicians were more likely to report having a poor specialty-referral network than white physicians.

It is important to recognize that many of the physicians working in black communities are hardworking, committed individuals who earn considerably less than other doctors. As Bach’s team notes, they deliver more charity care than doctors who mostly treat white patients and derive a higher volume of their practice revenue from Medicaid, a program whose fees are notoriously low. They are often solo practitioners who scramble to make good referrals for their patients but are stymied by a dearth of well-trained colleagues and by limited access to professional networks with advanced diagnostic techniques.

Class trumps race

To return to the question we posed at the beginning — would a white patient and a black patient arriving at the emergency room receive the same care? — we see that the question itself (at least as it is commonly understood) is flawed. The question presumes that black and white patients frequent the same health-care services, carry the same insurance coverage, and have identical health conditions — yet the data reveal that often they do not.

The most obvious and influential causes of these disparities reside in the differing health resources available to blacks and whites, including the quality of the physicians who treat them. These features place the emphasis on aspects of the health-care system in generating race-related differentials in treatment and far less so on clinically unjustifiable differences in treatment of white and minority patients by a given physician.

Meanwhile, true physician “bias” is very difficult to measure and define (since rational inferences are not the same as genuine prejudice). The Institute of Medicine panel might well have come to that conclusion itself had Congress directed it to evaluate the relative contributions of geographic, demographic, social, and economic factors in explaining discrepancies in care and outcomes. With that charge, the panel might well have come to a similar conclusion about the contribution of bias and the dubious value of emphasizing its role in maintaining the care gap and trying to combat it.

But if physicians cannot fairly be accused of bias, does this not just shift the charge of bias to the health-care system? In other words, do black patients receive poorer care because they are black or because they have disproportionately lower incomes and social capital (for example, less capacity for negotiating complex systems) than whites — and are thus disproportionately mired in systems that are underfinanced?

The most recent report from the Agency for Healthcare Research and Quality suggests this is so. It examines, separately, quality by race and quality by income.²⁶ It says that “remote rural populations” receive poor care, and “many racial and ethnic minorities and persons of lower socioeconomic positions” receive suboptimal care.

A better test of the class-trumps-race hypothesis would be to compare the quality of care received by poor whites clustered in a particular geographic area (for example, Appalachian populations) to that received by poor blacks who are clustered for example, in southeast Washington, D.C. If, after accounting for regional differences in practice or in health-care financing, comparable (and suboptimal) care were demonstrated, this would provide powerful support for the idea that systems serving poor people, irrespective of race, provide lower-quality care. Until such data are published — surprisingly we could find no reports on care of low-income whites versus low-income minorities — the allegation of racial bias in the system is unsupported.²⁷

Fortunately, policymakers are attuned to the quality problem and are grappling with it on several fronts, including the promotion and spread of information technology, performance enhancement of medical systems, outcome-based reimbursement to providers, and provider incentives (including malpractice reform, tax breaks, and assertion of market mechanisms that, among other things, reward physicians for the time they spend with patients). They also recognize that low-income patients benefit from a strong safety net provided by the federally funded community health-care system (guaranteeing a usual source of care); grassroots outreach through black churches, social clubs, and worksites; patient “navigators” to help negotiate the system; language services; and efforts to get more good doctors into distressed neighborhoods. Seemingly simple innovations, such as clinic night hours, could be a great boon to patients with hourly-wage employment who risk a loss of income, or even their jobs, by taking time off from work for doctors’ appointments.

Much has been made of the need for greater sensitivity in the doctor-patient relationship. Common sense dictates that patients benefit when they trust their physicians and interact with them productively. But the remedies for unsatisfactory doctor-patient relationships do not reside in racial sensitivity training for health-care professionals or the specter of Title vi litigation — all of which have been advocated.

Ultimately, improvement in the quality of care and self-care would elevate the status of minority health appreciably. But the greater public-health good would be served by applying these goals to all underserved people rather than focusing on minorities. By focusing on those with the worst health, the targets of intervention will still turn out to be poor minority groups, but they will include lower-class whites as well. For example, establishing screening (for cancer, diabetes, or hypertension) or wellness-education programs in benighted areas such as southeast Washington, D.C., or the Watts neighborhood of Los Angeles would benefit all residents and shrink overall racial differentials in health outcome because they would disproportionately target minorities.

Perhaps one of the most important factors in health disparities — self-care — does not depend much on health systems, except, perhaps, as vehicles for education. It is behaviors such as smoking, excessive alcohol use, unhealthy dietary patterns, and lack of exercise that figure so prominently in the development and course of chronic disease. In this arena, too, the influence of class outstrips race.

Words such as “prejudice,” “bias” and “discrimination” are charged and divisive. Nonetheless, many medical schools, health philanthropies, policymakers, and politicians are proceeding as if physician “bias” were an established fact.

Civil rights advocates talk about the lingering shadow cast by troubled race relations on the health-care system. Yet, paradoxically, health campaigns that seek to educate about the alleged bias of physicians will only inflame the mistrust that some minority patients already harbor. Concentrating on improving the health of all underserved Americans is the fairest and most efficient public health agenda.

¹ Olivia Carter-Pokras and Claudia Baquet, “What is a ‘Health Disparity?’” Public Health Reports 117 (2002), 427.

² Consortium of Social Science Associations, “cossa: Washington Update” 23:2 (January 26, 2004).

³ “Health Disparities Report at Center of Controversy; Department Altered Scientists’ Conclusions to Fit ‘Political Goals,’ Lawmakers Say,” Black Issues in Higher Education (February 12, 2004).

⁴ Congressional Black Caucus Foundation, “National Medical Association Appalled Over Distorted hhs Disparities Report” (February 10, 2004).

⁵ Leonard S. Rubenstein and Gretchen Borchelt, “Administration’s Reality Gap on Health Disparities,” Center for American Progress (September 20, 2004).

⁶ American Medical Association, “Improving Immunization: Addressing Racial and Ethnic Populations,” Roadmaps for Clinical Practice Series (June 2005).

⁷ American Public Health Association, “Research and Intervention on Racism as a Fundamental Cause of Ethnic Disparities in Health” (2005).

⁸ U.S. Department of Health and Human Services, “The Effect of Racial and Ethnic Discrimination/Bias on Health Care Delivery,” pa–05–006 (expiration date: January 3, 2008).

⁹ Paul D. Stein et al., “Venous Thromboembolic Disease: Comparison of the Diagnostic Process in Blacks and Whites,” Archives of Internal Medicine 163: 15 (2003). This national hospital discharge survey found no difference by race in terms of venous ultrasound, contrast venography, radioisotope lung scan, or duration of hospital stay; at the same time, age-adjusted rates of deep venous thrombosis (dvt) and pulmonary embolus (pe) were the same in blacks and whites.

¹⁰ Katherine Baicker et al., “Who You Are and Where You Live: How Race and Geography Affect the Treatment of Medicare Beneficiaries,” Health Affairs, October 7, 2004.

¹¹ Mitchell L. Margolis et al., “Racial Differences Pertaining to a Belief About Lung Cancer Surgery: Results of a Multi-Center Study,” Annals of Internal Medicine 139:7 (2003).

¹² Jennifer McCann et al., “Evaluation of the Causes for Racial Disparity in Surgical Treatment of Early Stage Lung Cancer,” Chest 128:5 (2005).

¹³ Jersey Chen et al., “Racial Differences in the Use of Cardiac Catheterization After Acute Myocardial Infarction,” New England Journal of Medicine 344:19 (2001).

¹⁴ Donald A. Barr, “Racial Differences in the Use of Cardiac Catheterization,” New England Journal of Medicine 839:19 (2001).

¹⁵ Chen et al., “Racial Differences in the Use of Cardiac Catheterization After Acute Myocardial Infarction.”

¹⁶ Chandra and Skinner, “Geography and Racial Health Disparities,” nber.Working Paper 9513, National Bureau of Economic Research, Inc., (2003).

¹⁷ Baicker et al., “Geographic Variation in Health Care and the Problem of Measuring Racial Disparities,” Perspectives in Biology and Medicine,  48:1, supp.  S42–53 (2005).

¹⁸ Skinner et al., “Mortality After Acute Myocardial Infarction In Hospitals that Disproportionately Treat African Americans,” Circulation 112:2634–41 (2005).

¹⁹ Amber Barnato et al., “Hospital-Level Racial Disparities in Acute Myocardial Infarction Treatment and Outcomes,” Medical Care  43: 308–19 (2005).

²⁰ Leo Morales et al., “Mortality Among Very Low Birthweight Infants in Hospitals Serving Minority Populations,” American Journal of Public Health 95 (2005).

²¹ Eric C. Schneider, Alan M. Zaslavsky, and Arnold M. Epstein, “Racial Disparities in the Quality of Care for Enrollees in Medicare Managed Care,” Journal of the American Medical Association 287:10 (2002).

²² Barbara Rothenberg et al., “Explaining Disparities in Access to High-Quality Cardiac Surgeons,” Annals of Thoracic Surgery 78 (2004).

²³ Donald Gemson, Jack Elinson, and Peter Messeri, “Differences in Physician Prevention Practice Patterns for White and Minority Patients,” Journal of Community Health 13:1 (1988).

²⁴ Kevin C. Heslin et al., “Racial and Ethnic Differences in Access to Physicians with hiv-Related Expertise,” Journal of General Internal Medicine 20:3 (2005).

²⁵ J. Lee Hargraves, Jeffrey J. Stoddard, and Sally Trude, “Minority Physicians’ Experiences Obtaining Referrals to Specialists and Hospital Admissions,” Journal of General Medicine 3:3 (2001).

²⁶ U.S. Department of Health and Human Services, Agency for Healthcare Research and Quality, National Healthcare Disparities Report 2003, chapter 3.

²⁷ Gilbert H. Friedell, M.D. (Director Emeritus, Markey Cancer Center, Lexington, Ky.), in discussion with authors (September 28, 2005).