Understanding and addressing racial disparities in health care.

Racial disparities in medical care should be understood within the context of racial inequities in societal institutions. Systematic discrimination is not the aberrant behavior of a few but is often supported by institutional policies and unconscious bias based on negative stereotypes. Effectively addressing disparities in the quality of care requires improved data systems, increased regulatory vigilance, and new initiatives to appropriately train medical professionals and recruit more providers from disadvantaged minority backgrounds. Identifying and implementing effective strategies to eliminate racial inequities in health status and medical care should be made a national priority.

Introduction

National data reveal that over the past 50 years, the health of both black and white persons has improved in the United States as evidenced by increases in life expectancy and declines in infant and adult mortality (National Center for Health Statistics, 1998). However, black persons continue to have higher rates of morbidity and mortality than white persons for most indicators of physical health. Hispanics and American Indians also have elevated disease and death rates for multiple conditions. Although the role of medical care as a determinant of health is somewhat limited, medical care (especially preventive care, early intervention and the appropriate management of chronic disease) can play an important role in health (Bunker, Frazier, and Mosteller, 1995). Thus, racial and ethnic differentials in the quantity and quality of care are a likely contributor to racial disparities in health status. Compared with white persons, black persons and other minorities have lower levels of access to medical care in the United States due to their higher rates of unemployment and under-representation in good-paying jobs that include health insurance as part of the benefit package (Blendon et al., 1989; Trevino et al., 1991).

More striking, and disconcerting to many is the large and growing number of studies that find racial differences in the receipt of major therapeutic procedures for a broad range of conditions even after adjustment for insurance status and severity of disease (Harris, Andrews, and Elixhauser, 1997; Wenneker and Epstein,1989). Especially surprising to many are the racial disparities in contexts where differences in economic status and insurance coverage are minimized such as the Veterans Health Administration System (Whittle et al., 1993) and the Medicare program (McBean and Gornick, 1994). Other research indicates that, although physicians' ability to detect the severity of pain does not differ for Hispanic versus non-Hispanic white patients (Todd, Lee, and Hoffman, 1994), Hispanic patients are markedly less likely than non-Hispanic white patients to receive adequate analgesia (Todd et al., 1993; Cleeland et al., 1997). Recent studies document that these differences in the receipt of therapeutic procedures have adverse effects on the health of minority group members (Peterson et al., 1997; Hannan, van Ryn, and Burke, 1999). How do we make sense of these differences and how do we move forward with an effective policy and research agenda to eliminate these disparities?

Race, Racism, and Discrimination

Many observers are surprised and perplexed by these findings. However, we can only regard these findings as surprising if we take an ahistorical and decontextualized view of the data. In compliance with Article 1, Section 2, and Paragraph 3 of the Constitution of the United States, the very first Census in 1790 enumerated three racial groups: whites, blacks as three-fifths of a person, and only “civilized Indians”—those who paid taxes (Anderson, 1988). New racial categories were added in the late 19th Century and beyond (Chinese in 1870, Japanese in 1890, Mexican in 1930) as the need arose to track new marginalized immigrant groups (Anderson and Feinberg, 1995). Race was and is a social category that captures differential access to power and desirable resources in society (Williams, 1997). Throughout the history of the United States, non-dominant racial groups have, either by law or custom, received inferior treatment in major societal institutions. Medical care is no exception.

Thus, understanding racial disparities in medical care requires an appreciation of the ways in which racism has operated and continues to operate in society. The term “racism” refers to an organized system, rooted in an ideology of inferiority that categorizes, ranks and differentially allocates societal resources to human population groups (Bonilla-Silva, 1996). It may or may not be accompanied by prejudice at the individual level. We will illustrate the complex nature of race, racism, and discrimination in society by considering access to housing and employment.

First, Table 1 indicates that there have been important positive changes in the racial attitudes of white persons towards black persons in recent decades and broad current support for the principle of equality in housing and employment (Schuman et al., 1997). In 1963, 60 percent of white persons agreed with the statement that “White people have a right to keep Negroes out of their neighborhoods if they want to, and Negroes should respect that right.” In 1996, only 13 percent of white persons agreed with that statement, documenting a substantial positive attitudinal shift within the white population. Similarly, in 1944, a majority of white persons (55 percent) indicated that white people should have the first chance at any kind of job but, by 1972, only 3 percent of white persons endorsed that view with 97 percent indicating that black persons should have as good a chance as white persons to get any kind of job.

Table 1

Racial Attitudes of White Persons in the United States: Selected Years, 1944-1996

Question	Year of Survey
	1944	1963	1964	1972	1973	1980	1983	1990	1996
	Percent in Agreement
Access to Housing Principle Item
White persons have a right to keep black persons out of neighborhood.	—	60	—	41	—	34	—	24	13
Implementation Item
They would support a law to let homeowners discriminate if they want to.	—	—	—	—	67	60	54	47	33
Access to Employment Principle Item
White persons should have the first chance at any job.	55	15	—	3	—	—	—	—	—
Implementation Item
Government should ensure no discrimination in jobs.	—	—	38	40	—	—	—	—	28
No interest in issue.	—	—	13	18	—	—	—	—	36

SOURCE: (Schuman, Steeh, Bobo, and Krysan, 1997.)

Second, these positive sentiments were given the force of law. In 1968, the Fair Housing Act (Title VIII) made it illegal to refuse to sell or rent a dwelling to any person because of race. Earlier, the Civil Rights Act of 1964 (Title VII) prohibited employers from firing, refusing to hire or promote, or in any way limiting an employee's compensation or job conditions because of race.

Third, Table 1 also indicates that there is considerably less support for policies that would actually implement equal access to housing and jobs (Schuman et al., 1997). In 1973, 67 percent of a national sample of white persons indicated that they would support a law that would guarantee a homeowner the right to decide for himself whom to sell his house to even if he preferred not to sell to black persons. In 1996, one-third of white persons would still grant a homeowner that right. In a similar vein, there is only weak support for policies to eradicate employment discrimination. In 1964, 38 percent of white persons indicated that the government in Washington should see to it that black people get fair treatment in jobs, and 13 percent indicated that they lacked enough interest in the question to favor one side over another. In 1996, the percentage of white persons supporting Federal intervention to ensure fair treatment in jobs declined to 28 percent, while the percentage expressing no interest in the question increased to 36 percent.

Fourth, national data on stereotypes reveal that white persons continue to view black persons negatively, which presumably would make them undesirable as neighbors and employees. For example, 56 percent of white persons believe that black persons prefer to live off welfare, 51 percent believe that black persons are prone to violence, 29 percent view black persons as unintelligent, and 44 percent view them as lazy (Davis and Smith, 1990). Comparatively, white persons believe that only 4 percent of white persons prefer to live off welfare, 16 percent are prone to violence, 6 percent are unintelligent, and 5 percent are lazy. Instructively, white persons viewed black persons, Hispanics, and Asians more negatively than themselves, but black persons were viewed more negatively than all other groups, and Hispanics twice as negatively as Asians. It is possible that these reported levels of negative stereotypes of minority groups are understated due to social desirability concerns.

Such high levels of the acceptance of negative stereotypes of minority groups is an ominous harbinger of widespread societal discrimination. A large body of psychological research indicates that the endorsement of negative racial stereotypes leads to discrimination against minority groups (Devine, 1995; Hilton and von Hippel, 1996). Moreover, well-learned stereotypes are resistant to disconfirmation (Stangnor and McMillan, 1992), and their activation is an automatic process with individuals spontaneously becoming aware of relevant stereotypes after encountering someone to whom the stereotypes are applicable (Devine, 1989; Hilton and von Hippel, 1996).

In the United States, racial stereotypes have real-life consequences for minority groups' access to housing and employment. Based on negative stereotypes of black persons, the majority of white persons express a strong preference for living in racially segregated neighborhoods (Williams et al., 1999; Bobo and Zubrinsky, 1996), and black persons in search of housing are still systematically steered toward neighborhoods having a greater number of minorities, lower home values, and lower median income (Fix and Struyk, 1993). A review of the data on the persistence of housing discrimination in the United States concluded that, “On any given encounter between a black home-seeker and a realtor, the odds are at least 60 percent that something will happen to limit that black renter or buyer's access to housing units that are available to white persons” (Massey, Gross, and Shibuya, 1994).

Studies of white employers reveal that racial stereotypes are used to deny employment opportunities to black applicants (Kirschenman and Neckerman, 1991; Neckerman and Kirschenman, 1991). Additionally, both U.S.-based and foreign companies explicitly use the racial composition of labor market areas in their decisionmaking process regarding where to locate new plants (Cole and Deskins, 1988). Not surprisingly, a Wall Street Journal analysis of the employment records of more than 35,000 U.S. companies found that black persons had a net job loss of 59,000 jobs during the 1990-1991 economic downturn, compared with net gains of 71,100 for white persons, 55,100 for Asians, and 60,000 for persons of Latin extraction (Sharpe, 1993). These job losses reflected the relocation of employment facilities to areas of lower black concentration. Audit studies of employment discrimination also find racial differences in being allowed to submit an application, in obtaining interviews, and in being offered a job. In these studies, when trained black and white job applicants with identical qualifications applied for jobs, discrimination favored the white over the black applicants in one in every five audits (Fix and Struyk, 1993).

The bottom line is that the policies designed to eliminate racial discrimination in housing and employment have failed. The degree of residential racial segregation in 1990 was virtually identical to what it was when Congress passed the Fair Housing Act in 1968 (Massey, 1996). Similarly, the unemployment rate for black persons has been consistently about twice that of white persons from 1950 to the present (Economic Report of the President, 1998). Thus, the advent of civil rights legislation and changes in the racial attitudes of white persons have not been sufficient to eradicate discrimination. In spite of these changes, there has been remarkable stability over time on multiple dimensions of racial inequality. For example, the median income of black persons was 59 cents for every dollar earned by white persons in 1996—identical to what it was in 1978 (Economic Report of the President, 1998).

Lessons for Racial Disparities in Medical Care

The larger literature on societal discrimination suggests that, although racism has changed over time from a blatant “Jim Crow racism” to a more subtle “laissezfaire racism” (Bobo, Kluegel and Smith, 1997), it persists in contemporary America. As painful as it may be to acknowledge, we must begin with the recognition that discrimination is routine and commonplace in society and likely to be similarly prevalent in medicine. With few exceptions (Smith, 1998; Geiger. 1996; Council on Ethical and Judicial Affairs, 1990), the literature on racial disparities in medical care is reluctant to admit and address racial bias among providers as a critical causal factor. In contrast, the evidence is abundant and clear that racial discrimination is not the aberrant behavior of a few “bad apples” but a widespread societal problem.

It is unlikely that personal discrimination on the part of providers is the sole cause of disparities in health care. In any area of societal evaluation, the causes of racial differences are complex and multi-dimensional, with discrimination being only one of them. Moreover, institutional discrimination is often at least as important as individual discrimination. In the case of racial disparities in medical care, other potential explanations include the geographic maldistribution of medical resources, racial differences in patient preferences, pathophysiology, economic status, insurance coverage, as well as in trust, knowledge, and familiarity with medical procedures (Horner, Oddone, and Matchar, 1995; Smith, 1998). On the surface, patient preferences would be the alternative explanation that would be most consistent with all of the available evidence. However, recent research indicating that patient preferences do not account for these disparities (Hannan, van Ryn, and Burke, 1999) suggests that discrimination remains as a central plausible explanation.

Probably most important, much discrimination today occurs through behaviors that the perpetrator does not subjectively experience as intentional. Much contemporary discriminatory behavior is unconscious, unthinking, and unintentional (Allen, 1995; Johnson, 1988; Lawrence, 1987; Oppenheimer, 1993). As noted earlier, biases based on racial stereotypes occur automatically and without conscious awareness even by persons who do not endorse racist beliefs (Devine, 1989). Recent psychological research indicates that persons who do not see themselves as prejudiced will make health care allocation decisions that adversely affect black persons when other negative characteristics are also present (Murphy-Berman, Berman, and Campbell, 1998). For example, respondents expressed greater resentment, gave lower health care priority scores and were more reluctant to make a financial contribution to the health care costs of patients presented as black and unemployed than as white and unemployed. In real-life medical encounters, the automatic activation of stereotypes may provide the negative characterization that triggers unconscious bias. Most legislation, intervention programs, and policy regarding discrimination have been ineffective because of their focus on purposeful or intentional discrimination (Allen, 1995). Relatedly, our review also suggests that one cannot rely on the stated racial attitudes of white persons or the mere existence of laws prohibiting discrimination to ensure that it does not occur in health care settings.

Policy and Research Directions

Improving Equity in Access to Medical Care

Although this article centrally addresses racial differences in the quality of care, a comprehensive approach to address inequities must begin by ensuring parity in access to care. Effectively addressing health care disparities will require comprehensive efforts by multiple sectors of society to address larger inequities in major societal institutions. There is clearly a need for concerted societal-wide efforts to confront and eliminate discrimination in education, employment, housing, criminal justice, and other areas of society which will improve the socioeconomic status (SES) of disadvantaged minority populations and indirectly provide them greater access to medical care. The United States also needs to make the moral and political commitment to guarantee access to medical care as a fundamental right of citizenship.

However, efforts to ensure equitable access to care must go beyond the elimination of financial barriers. A broad range of system barriers such as long waiting time, complex bureaucratic procedures, and the failure to treat patients with dignity and respect can lead to patient alienation and the avoidance of contact with the health care system unless absolutely necessary. Research has long indicated that poor persons and racial minorities are not viewed as desirable patients and health care providers deliver inferior care to persons of low SES (Duff and Hollingshead, 1968; van Ryn and Burke, 2000). Problems of patient-provider communication are exacerbated among persons of low SES, with higher SES patients receiving better technical and interpersonal care, and more positive communication than their lower SES peers (Hall, Roter, and Katz, 1988).

Efforts to ensure equitable access to care are urgently needed to counteract two forces that could potentially exacerbate racial disparities in access to care. First, there are closures of a growing number of health care facilities with hospitals located in low income and minority communities being more likely to close than those located in other areas (Whiteis, 1992; McLafferty, 1982). Second, and more important, the movement from a fee-for-service (FFS) system to a managed care system is likely to adversely impact access to medical care for minorities and other vulnerable populations (Schlesinger, 1987; Randall, 1994; McClellan, 1999; Harvard Law Review, 1995).

Critics argue that managed care is likely to exacerbate current racial inequalities in access to medical care in multiple ways. The new competitive pressures in the financing and delivery of medical care can reduce profits generated by treating privately insured patients, leading to cutbacks in the provision of uncompensated care (Schlesinger, 1987). In addition, managed care plans often limit access to minority physicians and other doctors who primarily care for disadvantaged populations (Harvard Law Review, 1995). Managed care plans look for cost effective physicians who order few procedures, write limited prescriptions, and make limited referrals. In contrast, given the high morbidity, comorbidity and greater severity of disease at time of diagnosis in minority and low income populations, these patients require intense management of chronic illness which can involve more medical services and thus higher costs. Accordingly, managed care plans are likely to view physicians who work in minority communities as undesirable and may also limit the physicians that their enrollees, including poor patients and residents of poor communities, can consult. The net result could be that many physicians who work in minority communities may have a smaller patient load and some may be even unable to maintain a financially viable practice (Harvard Law Review, 1995).

However, it is not clear if these fears have materialized. A 1994 survey of black physicians at a national conference found that 92 percent believed that managed care plans terminated the contracts of black doctors more often than those of white doctors (Lavizzo-Mourey et al., 1996). In fact, 88 percent had been refused a contract by a managed care organization (MCO) and 71 percent had lost patients to a MCO with which they were not affiliated. At the same time, 71 percent had at least one managed care contract and 75 percent indicated that their practice had grown or remained stable in the previous year. Some observers also note that it is not inevitable that access to medical care will worsen for minority populations under managed care arrangements. Increased competition could provide a financial incentive for some health care providers to provide treatment to segments of the community that they may have earlier viewed as undesirable (Schlesinger, 1987).

Improved Data Systems and Monitoring the Quality of Medical Care

Any concerted effort to address racial bias in the medical arena requires systematic and routine data of its occurrence. As noted, important changes are taking place in the organization and delivery of health care services in the United States and it is critical to monitor the impact of these changes on the health care access and quality for vulnerable populations. Major efforts are currently under way to identify the data elements that should be included in national uniform standardized data sets. It is essential that racial and ethnic status are uniformly and comprehensively assessed in these minimum core data sets. Comprehensive assessment means that data systems should include identifiers for major ethnic subgroups within the standard racial/ethnic categories (Williams, 1996). For example, among Hispanics, it is necessary to distinguish the three largest groups: Mexicans, mainland Puerto Ricans, and Cubans. Given the central role of SES in health and the strong relationship between race/ethnicity, and SES, it is important that indicators of SES are also included in any uniform data set (Krieger, Williams, and Moss, 1997; Williams, 1996).

The pervasiveness of discrimination suggests that racial data should be available for every medical encounter. The National Committee of Vital and Health Statistics, a public advisory body to the Department of Health and Human Services has called for the assessment of race/ethnicity and SES (National Committee of Vital and Health Statistics, 1993) in an enrollment database which could then be linked to data on medical encounters. Years of formal education was suggested as the most practical and convenient SES indicator in this context. The presence of these identifiers in an enrollment database would eliminate the resource intensive ordeal of attempting to request racial data at every medical encounter. Stringent efforts to ensure patient privacy and confidentiality would have to be implemented and the necessary training would have to be provided to health care workers to ensure the uniform collection of racial/ethnic and SES data. The analyses of racial disparities in the HCFA data files illustrate that administrative databases can be very helpful in providing findings that shed light on the nature and magnitude of the problem. However, these analyses have focused only on black-white differences because “black, white and other” were the only racial identifiers present. In the early 1990s, HCFA and the Social Security Administration (SSA) went to considerable effort to improve racial/ethnic identifiers in the SSA and Medicare data systems. What is needed now is the routine analysis and reporting of data for all racial/ethnic groups.

Routine reporting of data would identify which health care institutions, if any, demonstrate racial parity in terms of the delivery of medical care. It is likely that there is considerable variation in racial disparities across multiple settings. This kind of data can be used to establish benchmarks—levels of excellence achieved by industry leaders that could enhance our understanding of these best practices and facilitate their replication (Weissman et al., 1999).

Research is also needed to identify the optimal specific strategies that health care institutions can implement, at least on a periodic basis if not continuously, to detect and respond to patterns of discrimination in medical treatment. It has been suggested that hospitals could develop an anonymous reporting system to facilitate the detection of incidents of systematic patterns of biased medical decisionmaking (Noah, 1998). However, it will be crucial to create an environment that encourages reporting as part of a strategy of constructive problem-solving (Leape, 1997). In general, the five strategies for detecting inappropriate medical treatment, in order of yield and intensity of effort are direct observation, chart review, computer screening, focus groups and voluntary reporting (Leape, 1997).

The availability of data on racial differences in medical care would also facilitate at least some civil rights enforcement efforts. One of the limitations of current antidiscrimination laws is that they often rely on proving subjective discriminatory intent (Allen, 1995). Given that the vast majority of white Americans favor non-discrimination in principle, it is often difficult to prove discriminatory intent. In the legal arena, disparate impact claims require statistical data to document the differential effect of policies on racial groups. It is not a frequently used strategy by individuals seeking relief under civil rights statutes because it usually requires the compilation and analysis of large quantities of data to prove a discriminatory effect. However, routine collection of racial data would facilitate the assessment of racial group disadvantage in the medical arena.

Smith (1998) notes that the emphases on monitoring both clinical and financial information in the managed care environment also provides new opportunities for enhanced civil rights monitoring of health care delivery. He indicates that the addition of racial identifiers to existing data systems would facilitate the creation of report cards that could be used to monitor disparities in health plans, health care institutions, and communities. These report cards would include broadly accepted indicators of health and health care delivery (such as mortality rates and the use of specific screening tests) that have been recommended by various standards organizations for the comparative evaluation of medical care. Smith (1998) indicates that similar reporting requirements in the banking industry have led to a dramatic increase in the number of loans approved for racial minorities. However, in order for report cards to work, they must be a part of a larger system.

Renewed Regulatory Vigilance

Another policy strategy to address the problem of discrimination in medicine would be increased regulatory vigilance. The history of overt discrimination in medical care indicates that legal mandates and Federal regulations were ineffective until the institutional commitment and capacity to enforce them was created (Smith, 1998). Some legal scholars argue that there are existing statutes that are not now being enforced. For example, Title VI of the Civil Rights Act of 1964 is a promising statutory avenue for dealing with discrimination in health care delivery (Noah, 1998). Title VI prohibits any entity that receives Federal financial assistance from discriminating on the basis of race in providing goods or services to the beneficiaries of that Federal program. Since Federal financial assistance includes Medicare and Medicaid funds, this prohibition against discrimination applies to virtually all hospitals, nursing homes, and other health care facilities in the United States. Given that the courts have held that Title VI prohibits both intentional discrimination and disproportionate adverse impact (Noah, 1998), the documentation of adverse impact would provide a powerful strategy for addressing and correcting discrimination in care.

Noah (1998) also notes that many neutral policies that may have a disproportionate impact on racial and ethnic minorities receiving medical treatment could also be addressed if disparate impact were documented. For example, some hospitals admit only those patients whose physicians have staff privileges at the hospital, require substantial deposits before admission for inpatient care, refuse to deliver babies if their mothers had not received a certain amount of prenatal care, and create other barriers to the admission of Medicaid patients. Minority patients are more likely to be disadvantaged by these policies, although the policies are not specifically racial in content or intent. Thus, the application of a disparate impact analysis of these policies could be an effective avenue for attacking them. At the same time, Noah (1998) warns that, because Title VI covers institutional policies that are causing disproportionate impact, they would not cover the behavior of individual physicians who either consciously or unthinkingly discriminate as long as the health care entity can prove that it is not its institutional policy.

The Joint Commission on Accreditation of Health Care Organizations (JCAHO) is another regulatory organization that could play a larger role in addressing the question of racial inequities in health care. It could, for example, insist that health care entities establish procedures for the monitoring and correction of unexplained disparities in the delivery of services as one of the requirements for continued accreditation (Noah, 1998). Historically, the JCAHO has played an important role in encouraging progressive change in the health care industry. A recent example is the efforts to encourage hospitals to develop and implement policies to address communication problems with patients who do not speak English. HCFA might also review its current utilization review methods to see if better methods can be developed to identify and correct observed patterns of racial inequalities and delivery of services to beneficiaries of Medicare and Medicaid (Noah 1998).

Monitoring Managed Care

Increased regulatory efforts must also specifically focus on managed care plans given that managed care creates an environment conducive to discrimination and some have argued that there will be more widespread discrimination under managed care (Harvard Law Review, 1995). Unlike the FFS system where a physician's income increases as the number of services provided increases, in managed care there are often pressures to limit the number and cost of services delivered to members of the prepaid health plan. Many managed health care plans offer financial incentives to providers for limiting referrals to specialists. Often a pool of money is reserved for specialists and diagnostic tests, with the provider sharing in the unexpended portion of this pool (McClellan, 1999). Some managed care systems hold individual physicians financially liable when their patients use a higher than average amount of hospital care irrespective of whether the costs were medically justified (Hillman, 1987). These financial incentives have been shown to shape decisionmaking by providers (Pauly, Hillman, and Kerstein, 1990).

Currently available research does not clearly indicate whether medical care overall is better under the FFS system or under managed care plans. Some evidence suggests that enrollees in managed care plans are more likely to receive cancer screening tests than persons in FFS plans (Potosky et al., 1998). Recent national data from 35,000 non-elderly persons revealed that there were no differences between health maintenance organizations (HMOs) and other types of insurance in the use of inpatient care, emergency room and surgeries (Reschovsky et al., 2000). At the same time, patients in HMOs reported less satisfaction, less trust in physicians, lower ratings of physician visits and more unmet medical need. This study found that, while HMOs provided more primary and preventive services, they provided less specialist care and had higher administrative barriers to care. Other evidence suggests that health care for vulnerable populations, such as the elderly and the chronically ill, is worse in managed care plans (McClellan, 1999; Wallace, Enriquez-Haass, and Markides, 1998).

Several solutions have been proposed to the potential threat to minority health posed by managed care organizations (Harvard Law Review, 1995). First, there is need for more systematic auditing of the services of managed care providers by the Government, consumer groups, and professional medical organizations. Currently, there is little Federal oversight and monitoring of fairly rapid changes in health care delivery. There is also the need to develop new legislation or regulations that can ensure the appropriate inclusion of physicians who practice in minority communities in managed care plans. For example, managed care plans that recruit from minority communities could be required to hire a certain proportion of medical providers who previously practiced there. Similarly, laws could be established to ensure that the categorical exclusion of providers in poor communities does not occur. For example, they could prohibit the exclusion of providers from managed care plans based on the health, racial composition, or SES of their patients. Finally, sanctions can be utilized to prevent MCOs, and health care entities more generally, from engaging in discrimination. Systematic evidence of ongoing discrimination could be met with substantial fines or even the threat of the loss of the right to practice.

Wallace, et al. (1998) emphasize that managed care has the potential to improve the quality of medical care for minority populations. Given the greater bureaucratic oversight present in MCOs, there is the potential for greater coordination of care that could ensure that individuals received appropriate medical care. That is, if the appropriate incentives were built into managed care plans they could help reverse the trends of minorities receiving less aggressive treatment and fewer medical procedures across a broad range of illnesses.

Similarly, organizations such as the National Committee for Quality Assurance (NCQA) that currently provides voluntary accreditation to almost one-half of the HMOs in the United States could add racial inequities in care to the more than 60 standards that currently provide the basis for the evaluation of health plans. NCQA has also managed the development of the Health plans Employer Data and Information Set (HEDIS) which is the principal performance assessment for managed care. It uses a set of standardized measures to compare health plans. Currently HEDIS evaluates what a health plan actually does in key areas of care such as immunization rates and member satisfaction. The extent of racial/ethnic disparities should be added as a performance indicator.

Education and Training

There is also a need for intensive and systematic educational campaigns about the problem of racial inequities in health care. The awareness levels of the public and professional community, especially the medical community, must be raised. Research is needed to identify strategies that are most effective to raise awareness of and increase sensitivity to the issues of race in medical practice. Although education has its limits, it is also instructive to know that educational campaigns can accomplish much. For example, in the case of tobacco there has been a per capita decline in tobacco consumption in the United States over the course of the last century whenever there was a major media campaign on the negative effects of cigarette smoking (Warner, 1985). Efforts are clearly needed to impact the medical school curriculum to ensure that issues of race and sensitivity towards these issues are adequately addressed. As Geiger (1996) indicates, “awareness of the dilemmas associated with race and health care should be a part of every physician's training.”

It is not clear that the current emphases and approaches to cultural sensitivity will address the systemic problem of discrimination. Even model cultural sensitivity programs for medical students and residents (Robins et al., 1998; Zweifler and Gonzalez, 1998) do not address unconscious discrimination. Cultural sensitivity programs may enhance and accentuate negative stereotypes. Some cultural sensitivity training focuses on the distinctive behavioral patterns of subgroups in the population and appears to focus primarily on the “strange” behavior of patients. More research is also needed on provider attitudes and behavior (King, 1996) and especially the identification of the strategies that may be most effective in identifying and reducing unconscious discrimination among medical professionals.

Another critical goal of medical education should be to increase the number of minority professionals. Research clearly indicates that black and Hispanic physicians are much more likely than other physicians to care for the uninsured and those with Medicaid and to practice in urban and rural underserved areas where the percentage of residents of their racial/ethnic group is high (Komaromy et al., 1996). A recent study reported that in order to reach racial and ethnic population parity, the United States needs to double the number of black and Hispanic first-year residents and triple the number of Native American residents (Libby, Zhou, and Kindig, 1997). White first-year residents would need to be reduced by two-fifths and Asians by two-thirds.

Current trends do not suggest that these goals are likely to be reached. There has been relatively little increase in the proportion of physicians from underrepresented minority backgrounds in medicine in the last 30 years. For example, black physicians have increased from 2.5 percent of all U.S. physicians prior to 1968 (Carlisle, Gardner, and Liu, 1998) to 2.9 percent currently (Editorial, 1999). The number of underrepresented minorities (black Americans, Mexican Americans, mainland Puerto Ricans, and American Indians) increased in medical schools starting in the late 1960s. Minority enrollment peaked in 1974 at 10 percent of total enrollment but subsequently declined in the wake of reverse discrimination lawsuits (Carlisle, Gardner, and Liu, 1998; Nickens and Cohen, 1996). To reverse this trend the Association of American Medical Colleges launched a new campaign to have 3000 first-year minority medical students by the year 2000. This project was initially successful with enrollment reaching a high of 2014 under-represented minorities (12 percent of all new entrants) in 1994 (Carlisle, Gardner, and Liu, 1998). In 1996, there was a large drop in applications from underrepresented minorities to medical schools (Editorial, 1999) with more than one-half of all U.S. medical schools experiencing a decline in minority enrollment (Carlisle, Gardner, and Liu, 1998). The decline was largest in public medical schools and in the four States (California, Texas, Louisiana, and Mississippi) where affirmative action programs have been banned.

More research is needed to identify what are the most effective strategies for the recruitment and retention of physicians from disadvantaged backgrounds. However, we should also capitalize on the currently available evidence on effective strategies. Affirmative action programs (Federal initiatives that allowed schools and employers to take into consideration a qualified applicant's race, sex, national origin, or disability) have been successful. It is estimated that affirmative action is responsible for 40 percent of all U.S.-trained physicians from underrepresented minority backgrounds (Editorial, 1999). Although underrepresented minority students tend to have lower test scores than other medical students, they do not differ on clinical performance suggesting that other non-cognitive variables are essential in the recruitment of competent physicians (Tekian, 1997). In addition, despite their current unpopularity, affirmative action programs are defensible on multiple grounds including the societal obligation to meet the health care needs of all segments of the population (Nickens and Cohen, 1996). Moreover, recent research documents that racial prejudice, especially contemporary subtle prejudice, is the single most important source of opposition to affirmative action (Williams et al., 1999). These fingings emphasize the critical need to confront and eradicate societal racism in order to develop effective strategies to overcome America's painful history of exclusion and discrimination.

However, McClellan (1999) notes that simply matching physician race with client race is not enough. He cites an uncomfortable but insightful example from the Tuskegee Syphilis Study. In 1969, the Macon County Medical Society endorsed the continuation of the Tuskegee Syphilis Study. At that time, the Society was virtually all-black. McClellan (1999) argues that social class provides an explanation of those black physicians' behavior. That is, because most of the patients in the Tuskegee study were poor and illiterate, the study was not a threat to middle class physicians, their families, or friends. He argues that educational and professional socialization may lead health care professionals to distance themselves in terms of emotional attachment and self-interest from their groups of origin. Unthinking discrimination is likely to occur whenever medical professionals, irrespective of race, endorse negative societal stereotypes of their patients. Research is needed to identify the extent to which these processes operate and what may be the most effective strategies to counter them.

Rather than focusing only on individual characteristics, efforts to reduce racial/ethnic disparities in the quality of medical care should seek to reform the systems for the delivery of care. Is it possible to design and implement systems for the delivery of medical care that can ensure appropriate behavior irrespective of effect and unconscious stereotypes? A definitive answer to this question awaits the necessary research, but Leape (1997) emphasizes that in medicine, as in other occupations, conditions of work can be managed, tasks and processes can be designed, and workers can be trained in ways that minimize undesirable outcomes.

Conclusion

It is a national embarrassment that there are large and persisting racial differences in health. National data reveal, for example, that black persons had an overall mortality rate that was 1.6 times higher than white persons in 1995—identical to the black/white mortality ratio in 1950 (Williams, 1999). Moreover, for multiple causes of death (heart disease, cancer, diabetes, and cirrhosis of the liver) the racial discrepancy was larger in 1995 than in 1950. These inequities fly in the face of cherished American principles given the public's commitment to principles of equal treatment in society. As a society, we need to make it a national priority to build on the cultural support for egalitarian principles and develop strategies to eradicate racial inequities in medical care.

The United States President and other leaders in the executive branch of Government should use the “bully pulpit” to place this issue on the national agenda. President Clinton has drawn much national and media attention to the problem of health disparities by making it a topic of one of his Saturday morning radio addresses and by the establishment of his Commission on Race. Relatedly, in 1994, President Clinton issued an executive order on environmental justice that could be a model of the kind of effort that is needed (Noah, 1998). This executive order created an interagency working group to address questions of environmental justice and to provide guidelines to Federal agencies for coordinating research, collecting data, and developing effective strategies to address inequities. A similar working group on racial disparities in medical care could monitor and coordinate data from multiple sources and provide a coordinated picture of the nature and prevalence of inequities in multiple health care settings. A commitment to the eradication of racial disparities also requires the creation of the institutional capacity to effectively monitor and enforce all existing laws and regulations.

Our review indicates that racism appears to be a technological hazard in the practice of medicine. Much of it may be unthinking and careless and not deliberately hateful behavior. However, regardless of motive, there are pervasive adverse consequences. As a society, we lack data on effective strategies to reduce racism at both the individual and institutional levels. While there are many books published on the topic and many programs on cultural diversity and tolerance, there is little systematic data available about the conditions under which particular strategies are more or less effective. Given the growing body of evidence that indicates that racism adversely affects health in multiple other ways (Krieger, 1999; Williams, 1999), more systematic efforts to develop and assess the impact of various strategies to reduce racism is warranted. Well-funded research centers for excellence should be established to foster interdisciplinary research on understanding and eliminating racial/ethnic disparities in medical care. Courageous moral and political leadership is also needed to take the necessary steps to apply all of the knowledge that we currently have for reducing racial/ethnic disparities in health, and more generally.