|
Items in cart [0] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 125
Assessing Potential Sources of Racial
and Ethnic Disparities in Care:
Patient- and System-Leve! Factors
The literature reviewed earlier in this report demonstrates that evi-
dence of racial and ethnic disparities in healthcare is persuasive and re-
markably consistent across a range of health conditions and procedures,
and cannot be fully explained by differences in access to care, such as
insurance status. Moreover, the literature suggests several sources for
these disparities. This evidence, however, does not suffice for an authori-
tative, comprehensive, unambiguous account of how disparities arise. A
number of uncertainties confound efforts to synthesize what is known
empirically about stereotypes and prejudice, doctor-patient relations,
clinical judgment and patient preferences, as well as the social, institu-
tional, financial, and legal forces that shape the practice of medicine. Yet
an effort at such a synthesis is essential to construct an evidence-based
account of how disparities in care emerge, and of what might be done to
eliminate these disparities.
To begin, this chapter presents a model of how disparities might oc-
cur. This model builds upon the wide foundation of empirical evidence
but makes reasoned inferences when they are necessary to explain ob-
served disparities. The committee makes such inferences when, in our
judgment, they are more probable than not and when practical conse-
quences, in the form of recommended actions to ameliorate known dis-
parities, follow from these inferences. In doing so, the committee ac-
knowledges that gaps in our understanding about causation remain and
that further research has the potential to enhance understanding.
The chapter then presents a review of empirical literature that raises
hypotheses regarding potential sources of racial and ethnic disparities in
125
OCR for page 126
26
UNEQUAL TREATMENT
healthcare. This literature is suggestive of a range of sources of dispari-
ties, some of which lie just beyond the conscious perception of individual
actors (e.g., patients, providers, health systems administrators) in clinical
encounters. They include systemic (e.g., those related to health system
administration, financing, accessibility and geographic location), patient-
level (e.g., the clinical appropriateness of care, patients' attitudes, prefer-
ences, and expectations regarding healthcare), and care process-level (e.g.,
physician biases, stereotyping, and uncertainty) factors. This chapter,
however, will focus on the two former sets of variables. As depicted in
Figure 1-1, these include "patient-level" variables, and variables related
to the operation of healthcare systems and the legal and regulatory con-
texts in which health systems function. Chapter 4 will focus the analysis
on care process variables, including the roles of clinician bias, prejudice,
stereotyping, clinical uncertainty, and patient mistrust. According to the
study committee's conceptualization, racial and ethnic differences in care
may arise from all three sets of variables. Disparities in care, however,
emerge from the characteristics of and the operation of healthcare sys-
tems, as well as the legal and regulatory climate in which care is deliv-
ered, and from the process of care (i.e., factors emerging from the pro-
vider-patient interaction).
The following section presents a guiding framework that depicts the
likely interplay of health systems characteristics, patient-level factors,
and care process variables in fostering racial and ethnic disparities in
healthcare.
A MODEL: SOURCES OF HEALTHCARE DISPARITIES
The Role of Clinical Discretion
An integrated model of how racial and ethnic disparities in care
emerge is presented in Figure 3-1. According to this model, patients
present to healthcare providers with varied healthcare needs, expecta-
tions, and preferences, some of which are socio-culturally determined.
Providers, in turn, possess expectations and beliefs that are shaped both
by their professional training and experience, as well as by their social
experiences and broader societal norms and structures. These encounters
take place within healthcare systems and settings that are broadly influ-
enced by institutional design factors (such as the ease of care access), and
financial forces (such as incentives to providers and patients to limit ser-
vice use and healthcare costs). These systems operate within legal and
cultural contexts that influence how healthcare is delivered and the be-
havior of both patients and providers.
OCR for page 127
PATIENT- AND SYSTEM-LEVEL FACTORS
127
FIGURE 3-1 An integrated model of healthcare disparities.
Central to this model is the role of personal discretion in determining
the care that patients receive. Three sets of actors possess and exercise
discretion: clinical caretakers, utilization managers remote from the bed-
side, and patients themselves. Patients' medical histories, physical exam
findings, and diagnostic test results often present a level of uncertainty to
physicians, and patients vary enormously in their help-seeking behavior,
ways of presenting their symptoms and histories, and responses to medi-
cal recommendations (Bursztajn, 1990~. Clinicians typically have mul-
tiple diagnostic and therapeutic options, and choices from among them
sometimes do not rest firmly on empirical evidence. In addition, physi-
cian perception of clinical signs and symptoms is sometimes incomplete
(Eddy, 1996), while decisions concerning diagnostic and therapeutic in-
OCR for page 128
28
UNEQUAL TREATMENT
tervention are no less ambiguous. Significant variations in the incidence
of many common medical and surgical procedures have been documented
within small geographic areas and between individual practitioners
(Wennberg, 1999~. These variations reflect, in many instances, both the
subjectivity of clinical judgment and the lack of professional consensus
about best practice. Further, the lack of firm scientific support for some
medical decisions both engenders variations in clinical practice and makes
it difficult in many cases to reach evidence-based conclusions concerning
the appropriateness of particular practices.
In and of itself, the discretion exercised by patients, providers, and
utilization managers does not produce racial and ethnic disparities in
healthcare. In most cases, patients and providers are able to work to-
gether in an iterative process to match patients' needs with appropriate
treatment, regardless of race or ethnicity. Discretion and ambiguity, how-
ever, create conditions in which race or ethnicity may become salient in
the process of diagnosis and treatment in ways that make disparities more
likely to occur, as explained below.
The Patient as Discretionary Actor: Subjectivity and Variability
A substantial research literature in psychiatry and psychology, soci-
ology, and anthropology documents large differences in how people ex-
perience, understand, and discuss illness (Goff et al., 1998~. Patients' ex-
perience and reporting of pain and other symptoms have been found to
vary greatly (Bonham, 2001), as has patients' help-seeking behavior rela-
tive to health professionals (Milewa et al., 2000~. The relationships be-
tween such variation and differences in how clinical caretakers go about
the diagnosis and treatment of disease have been less well studied. But
the subjectivity and incompleteness of clinical perception leave room for
differences in patients' experience, understanding, and reporting of symp-
toms to affect professional judgment and action. These differences, more-
over, interact with differences in patients' values and attitudes toward
clinical caretakers to shape patients' healthcare choices. To the extent that
such variation correlates with patients' race and ethnicity, it is therefore a
potential contributor to differences in healthcare use.
Clinical Caretakers as Discretionary Actors: Subjectivity and Uncertainty
Medical care at the dawn of the 21st century has achieved heights
thought improbable even a few decades ago. Advances in diagnostic tech-
niques, scientific understanding of the human genome and underlying
disease processes, and new, high-tech interventions have led to break-
OCR for page 129
PATIENT- AND SYSTEM-LEVEL FACTORS
129
throughs in treating and preventing disease. Despite these gains, how-
ever, many medical decisions must be made in the absence of solid evi-
dence as to the efficacy of diagnostic and therapeutic measures or rigor-
ous scientific understanding of the pathophysiology of disease (Mushlin,
1991~. Efforts to better understand pathophysiology are further compli-
cated by variations in clinical expression in individuals with different ge-
netic, environmental, and cultural backgrounds. In addition, even the
most technologically sophisticated diagnostic interventions (e.g., magnetic
resonance imaging and X-ray and positron tomography) reveal little about
the biochemistry and physiology of the diseases they detect. To add to
this uncertainty, medicine's diagnostic constructs are themselves limited
in their predictive (and thus therapeutic) value by the incompleteness of
the pathophysiologic understandings that undergird them (Bloche, 2001~.
Moreover, healthcare providers' ability to assess patients' clinical
signs and symptoms and gather a relevant medical history is constrained
by a number of factors. As noted above, patients' ability to understand
and describe their presenting concerns varies not only by cultural, lin-
guistic, and other sociodemographic background factors, but may also
vary from day to day. The variability and subjectivity of patients' clinical
presentations is compounded by physicians' differences in perception,
cultural and psychological sensitivity, and conceptual frameworks for
evaluating illness. Similarly, many laboratory tests are open to varying
interpretations. Radiologists sometimes give conflicting readings of the
same X-ray, tomogram, or other scan, and pathologists sometimes report
conflicting interpretations of slides sent for assessment of possible malig-
nancy. Many clinical and laboratory data are likewise open to differing
clinical interpretations by physicians with varying conceptual frame-
works, perceptions, and biases. As will be discussed in the next chapter,
it is reasonable to speculate that the resulting diagnostic subjectivity could
permit clinical uncertainty, racial and ethnic biases, and stereotypes to
influence the process and outcomes of clinical evaluation, resulting in ra-
cial and ethnic disparities in medical diagnosis.
Physicians' decisions regarding appropriate therapeutic interventions
introduce still another level of uncertainty, subjectivity, and variability.
Despite clinical medicine's gains noted above, accurate, evidence-based
prediction of the efficacy of many therapeutic alternatives for most pa-
tients is lacking (Bloche, 1999), and geographic variations in clinical prac-
tice patterns are common (Wennberg, 1999~. In the absence of guidance
from prospective and retrospective clinical studies, physicians base their
therapeutic judgments on such factors as their training, prior clinical suc-
cess and failure, and practice norms among professional peers (Bauchner,
Simpson, and Chessare, 2001~. Inevitably, physicians' subjective under-
OCR for page 130
30
UNEQUAL TREATMENT
standings of their patients' needs play a role; thus psychological sensitiv-
ity, cultural and language competency, and conscious and unconscious
stereotypes and biases may also influence therapeutic decision-making.
Further, uncertainty about treatment options in itself, even absent biases
or stereotypes, can lead to disparate treatment of racial and ethnic minor-
ity groups, as will be discussed in Chapter 4. In addition, organizational,
financial, and legal influences shape therapeutic judgment. Such institu-
tional and policy forces are often geared toward promoting cost-effective
and efficient care, but may disproportionately and negatively affect mi-
nority patients (Bloche, 2001~.
Utilization Managers as Discretionary Actors: Uncertainty at a Distance
Variation and subjectivity in healthcare practice may also emerge at
the level of health systems, particularly in managed care arrangements
where utilization managers are charged with authorizing physicians' and
patients' requests for reimbursement for services. Except where contrac-
tually bound by clinical practice protocols, utilization managers evaluate
the necessity of claims from among a range of diagnostic and therapeutic
alternatives acceptable within one or another subset of the medical com-
munity. This evaluation often occurs on a case-by-case basis, without the
guidance of recorded precedent or other administrative means for pursu-
ing consistency between utilization management decisions in similar
cases. In some cases, doctors and patients who seek pre-approval for
planned treatments or who pursue internal appeals when pre-approval is
denied know little or nothing about their health plans' past pre-approval
practices in similar cases. Health plans that employ clinical practice pro-
tocols as cost management tools sometimes treat these protocols as trade
secrets, not to be disclosed to patients or medical practitioners. The con-
sequence of these administrative arrangements is that there is ample room
(and little visibility) for discretion and inconsistency in the treatment of
clinically similar cases (Bloche, 2001~.
The following sections review available empirical evidence and pre-
sent an analysis of how discretion, subjectivity, and preferences of pa-
tients, providers and utilization managers may contribute to healthcare
disparities. Consistent with the committee's model of sources of racial
and ethnic differences in care, these sources are divided into patient-level
variables (such as preferences, needs, and the clinical appropriateness of
care), and factors related to health systems and the legal and regulatory
context of healthcare. Factors arising from the clinical encounter that may
contribute to disparities are addressed in Chapter 4.
OCR for page 131
PATIENT- AND SYSTEM-LEVEL FACTORS
PATIENT-LEVEL VARIABLES PREFERENCES, MISTRUST,
TREATMENT REFUSAL, BIOLOGICAL DIFFERENCES, AND
OVERUSE OF SERVICES
Patients' Preferences
131
To a great extent, patients' values, fears and hopes, and other psycho-
logical characteristics influence the level and type of care they receive.
Patients' trust and doubts about medical advice, as well as their level of
comfort with the effectiveness and potential unintended effects of inter-
ventions, directly influence their willingness to accept physicians' recom-
mendations. In addition, patients' preferences are influenced by their tol-
erance for pain and discomfort, attitudes about long-term/short-term
tradeoffs, and levels of social and emotional support. These factors also
influence physicians' recommendations, in that the physician may directly
assess or infer patients' attitudes toward particular interventions and may
tailor recommendations accordingly. To the extent that minority patients
express greater reluctance to accept physician recommendations, patients'
preferences have the potential to contribute to healthcare disparities. Evi-
dence that minority patients are more likely than whites to decline inva-
sive and/or high-tech procedures is reviewed below.
For many racial and ethnic minorities, however, preferences for treat-
ment are often difficult to separate from mistrust of health professions
that stems from racial discrimination and the history of segregated and
inferior care for minorities (Byrd and Clayton, this volume). Some re-
searchers have not distinguished between these aspects of minority pa-
tients' historic experiences and preferences for treatment, and have con-
trasted "preferences" and racial discrimination as competing explanations
for healthcare disparities. This account overlooks the interaction between
patients' "preferences" and their experiences of discrimination. As Bloche
(2001) notes, "For many African Americans, doubts about the trustwor-
thiness of physicians and healthcare institutions spring from collective
memory of the Tuskeegee experiments (Brandt, 2000) and other abuses of
black patients by largely white health professionals (Randall, 1996; King,
1998~. This legacy of distrust, which, some argue, contributes to dispari-
ties in healthcare provision by discouraging African Americans from seek-
ing or consenting to state-of-the-art medical services, is thus itself a
byproduct of past racism" (Bloche, 2001, p. 105~.
Minority patients' negative experiences with care providers in the
clinical encounter can also diminish their preferences for robust treatment,
and may thereby contribute to racial disparities. It is reasonable to as-
sume that experiences of real or perceived discrimination in healthcare
settings, as evidenced by providers' overt behavior (e.g., as in the ex-
OCR for page 132
32
UNEQUAL TREATMENT
amples from focus group data presented in Chapter 2) or more subtle,
subjective mistreatment (e.g., healthcare providers' low expectations for
compliance or expressions of low empathy for minority patients) can af-
fect patients' feelings about their clinical relationships and thereby
dampen their interest in vigorous diagnostic and therapeutic measures. It
is therefore necessary to distinguish patient "preferences" from experi-
ences or perceptions of discrimination and not neglect the ways in which
patients' preferences can be shaped by provider behavior. In addition,
patients' preferences for treatment may be limited by the quality and com-
pleteness of information presented by the healthcare provider. Thus,
should providers fail to present minority patients with a full range of treat-
ment options, whether out of prejudice, stereotyping, biases, or uncer-
tainty about the diagnosis or appropriate clinical course of action, patients'
preferences will be limited by the information they are presented. These
dynamics will be addressed in greater detail in Chapter 4.
Minority Patient Preferences Regarding Providers and Racial Concordance
Minority patients' experiences, values, and expectations regarding
healthcare may significantly influence their preferences for the race or
ethnicity of their providers. A growing body of evidence suggests that
racial and ethnic minority patients are generally more satisfied with the
care that they receive from minority physicians. Saha, Komaromy, Koep-
sell, and Bindman (1999), for example, found that African-American pa-
tients with African-American healthcare providers were more likely than
those with non-minority providers to rate their physicians as excellent in
providing healthcare, in treating them with respect, in explaining their
medical problems, in listening to their concerns, and in being accessible.
Hispanic patients who received care from Hispanic physicians did not
rate their physicians as significantly better than Hispanic patients with
non-Hispanic healthcare providers, but were more likely to be satisfied
with their overall healthcare.
Similarly, Cooper-Patrick and her colleagues (Cooper-Patrick et al.,
1999) assessed patients' ratings of the quality of interpersonal care in ra-
cially concordant and racially discordant settings. Using a measure of
physicians' participatory decision-making (PDM) style, the authors sur-
veyed over 1800 adults (including 43% white, 45% African American, and
12% other race or ethnicity) who were seen in 1 of 32 primary care settings
by physicians who were either African American (25% of the physician
sample), white (56%), Asian American (15%), or Latino (3%~. Overall,
African-American patients were found to rate their visits as significantly
less participatory than whites, after adjusting for patient age, gender, edu-
cation, marital status, health status, and length of the patient-physician
OCR for page 133
PATIENT- AND SYSTEM-LEVEL FACTORS
133
relationship. Further, patients in race-concordant relationships rated their
visits as significantly more participatory than patients in race-discordant
relationships.
LaVeist and Nuru-leter (in press) examined predictors of racial con-
cordance between patient and provider and the effect of race concordance
on satisfaction among a sample of white, African-American, and Hispanic
patients. Among all racial and ethnic groups, patients who reported hav-
ing at least some choice in selecting a physician were more likely to have
a race-concordant physician. Having a race-concordant physician was
also associated with higher income for African Americans and not speak-
ing English as a primary language among Hispanics. After adjusting for
patients' age, sex, marital status, income, health insurance status, and
whether the respondent reported having a choice in physician, African-
American patients in race-concordant relationships were found to report
higher satisfaction than African Americans in race-discordant relation-
ships. Further, Hispanic patients in race-concordant relationships re-
ported greater satisfaction than patients from other racial and ethnic
groups in similarly concordant relationships.
While these studies lend important information regarding patients'
perceptions of the interpersonal quality of care, few studies have corrobo-
rated this data with more objective assessments of the patient-provider
interaction in racially concordant and discordant settings. Cooper and
Roter (this volume) describe a study that assessed this relationship using
post-visit surveys and audiotape analysis among a sample of 143 white
and 110 African-American patients seen by 1 of 13 white or 18 African-
American primary care doctors. Cooper and colleagues found that the
average length of visits was shortest among white physicians with Afri-
can-American patients (13.2 minutes), and was longest among African-
American physicians seeing white patients (18.4 minutes). Visits by Afri-
can-American patients were characterized by greater physician verbal
dominance overall, but physician verbal dominance was highest in visits
between white physicians and African-American patients, and lowest
among white patients seen by African-American physicians. In addition,
visits between white physicians and African-American patients were the
least patient-centered, while the African-American physician-white
patient interaction was characterized by the highest levels of patient
centeredness. However, patients in race-concordant relationships rated
their physicians' decision-making styles as more participatory. The find-
ing that African-American physicians were more patient centered and
spent more time with white patients, according to Cooper and Roter, sug-
gests two possibilities. One is that African-American physicians, by vir-
tue of their educational success and professional standing, presumably
have had greater opportunities to develop skills in communicating with
OCR for page 134
34
UNEQUAL TREATMENT
individuals from the dominant culture than white physicians have had to
develop communication skills with individuals from minority cultures.
This suggests that African-American physicians are likely to be bicultural
(i.e., able to function effectively in the dominant culture as well as in mi-
nority cultures) or are acculturated (i.e., have assumed traits of the domi-
nant culture). Secondly, African-American physicians may "overcompen-
sate" for perceived deficiencies of their own group by adopting behaviors
that indicate less respect for themselves or members of their own group
(Cooper and Roter, this volume).
A significant limitation of these studies is the lack of random assign-
ment of patients with physicians, introducing selection factors as a poten-
tial confound. In fact, Saha et al. (1999) found that African-American and
Hispanic patients who had the ability to choose their provider were more
likely to choose a racially or ethnically concordant physician. A signifi-
cant proportion (42%) of Hispanic patients in this study reported select-
ing Hispanic physicians because of linguistic barriers with other provid-
ers. Nonetheless, these studies demonstrate that racial concordance is
associated with greater participatory decision-making, greater patient-
centered care, lower levels of physician verbal dominance, and greater
patient satisfaction. In turn, evidence from other studies indicates that
patient satisfaction is associated with greater patient compliance with
treatment regimens, participation in treatment decisions, and use of pre-
ventive care services (Cooper and Roter, this volume). For racial and eth-
nic minorities, according to Cooper-Patrick and colleagues (1999), racial
concordance may increase the likelihood that they will "share cultural
beliefs, values, and experiences in the society [with their provider], allow-
ing them to communicate more effectively and to feel more comfortable
with one another" (p. 588~.
Little evidence exists, however, to directly demonstrate that the qual-
ity of care provided is better when minority patients and their providers
are of the same racial or ethnic group. Evidence of the efficacy of race-
concordant patient and provider relationships is only indirect, as patient
satisfaction, participation, and patient-centeredness of care are also asso-
ciated with greater adherence to clinical regimens, greater participation in
health screening and preventive medicine, and in some cases, health out-
comes (Cooper-Patrick et al., 1999~. Chen et al. (2001) suggest the oppo-
site that racial disparities persist in care even when minority patients
are treated by minority physicians. The authors performed a retrospec-
tive analysis of data obtained from a sample of Medicare patients hospi-
talized with acute myocardial infarction (MI) to determine whether racial
differences in rates of cardiac catheterization were related to the race of
attending physician. Consistent with other studies, the authors found
that African-American patients were less likely than white patients to re-
OCR for page 135
PATIENT- AND SYSTEM-LEVEL FACTORS
135
ceive catheterization within 60 days after MI. No significant differences
were found, however, between African-American and white attending
physicians in rates of catheterization among these patients. Among pa-
tients treated by African-American physicians, 38 percent of black patients
and 50 percent of white patients received catheterization. Among pa-
tients treated by white physicians, 38 percent of black patients and 46
percent of whites received the procedure. Chen et al. conclude that "ra-
cial discordance between the patient and the physician does not explain
differences between black patients and white patients in the use of cardiac
catheterization" (2001,p. 1447~.
While some newspaper accounts of this study suggested that racial
bias is not a likely source of disparities in care (Associated Press, May 9,
2001), this interpretation appears premature. Several methodological
problems complicate interpretation of the results obtained by Chen et al.
Data on the race of the attending physician were missing for nearly one-
third of the initial patient sample. In addition, African-American patients
were more likely to be cared for in public or teaching hospitals, where
greater barriers exist to receipt of catheterization, such as the availability
of the procedure on-site. The most serious methodological problem, how-
ever, was the determination of the race of the attending physician ("the
clinician who is largely responsible for the care of the patient from the
beginning of the hospital episodes. Upon closer examination, it becomes
apparent that the African-American physicians of these patients tended
to be internists, not cardiologists, when compared with the white attend-
ing physicians. While these physicians may all refer patients for the pro-
cedure, the determination of who receives the procedure is typically made
by the cardiologist. Thus, the authors compare two different physician
pools who cared for these African-American patients post-myocardial in-
farction African-American internists versus white cardiologists to as-
sess differences in utilization of a procedure that is specifically performed
by and managed by cardiologists. Notably, of the nearly 20,000 cardiolo-
gists in the United States during the study period in 1994 and 1995, only
316 (approximately 1.5%) were African American.
Minority Patient Mistrust and Experiences of Discrimination
Some racial and ethnic minorities express greater levels of mistrust of
healthcare providers and the medical establishment than white Ameri-
cans, citing breeches of trust that have previously occurred between mi-
norities and the scientific and medical communities (Swanson and Ward,
1995~. In addition, survey research generally indicates that ethnic minor-
ity patients perceive higher levels of racial discrimination in healthcare
settings than non-minorities. For example, in a survey of 781 African-
OCR for page 149
PATIENT- AND SYSTEM-LEVEL FACTORS
149
factors were considered. In a study of prenatal birth outcomes among
civilian and military women, Barfield et al. (1996) found that rates of pre-
natal care utilization were lower, and rates of low birth weight and fetal
and neonatal mortality higher among African-American women than
white women, but that these racial disparities were lower (but still signifi-
cant) among the military population. In addition, a recent study of VA
systems found modest racial differences in mortality rates among Afri-
can-American and white patients admitted for pneumonia, angina, con-
gestive heart failure, chronic obstructive pulmonary disease, diabetes, or
chronic renal failure, but these differences suggested better survival rates
for minority patients (The et al., 2001~.
Other studies, however, note significant racial differences in VA sys-
tems in rates of procedures such as cardiac catheterization. Peterson et al.
(1994), Mirvis et al. (1994), Whittle et al. (1993), and Mirvis and Graney
(1998) all found African-American VA patients less likely to receive car-
diovascular procedures than white VA patients. Sedlis et al. (1997) found
that therapeutic cardiac procedures (surgery or PTCA) were offered more
frequently for white VA patients (72.9%) than African-American VA pa-
tients (64.3%~. This difference could not be explained by simple clinical
differences between the two groups. Conigliaro et al. (2000) found that
although African-American VA patients were less likely then white VA
patients to undergo CABG and PTCA, when RAND appropriateness cri-
teria were considered, African Americans were still less likely to receive
CABG when deemed "necessary." Oddone et al. (1999) studied racial
differences in rates of carotid artery imaging among patients diagnosed
with transient ischemic attack, ischemic stroke, or amaurosis fugax seen
at one of four VA Medical Centers. After controlling for patients' age, co-
morbid factors, clinical presentation, anticipated operative risk, and hos-
pital, African-American patients were found to be half as likely as whites
to receive carotid imaging.
Evidence for racial and ethnic disparities in care in VA systems is
therefore mixed, but suggests that financial, structural and institutional
factors of these systems, as well as the universally available care for mili-
tary personnel in DoD systems may serve to attenuate some disparities in
care. For example, physicians in both DoD and VA systems are salaried,
eliminating the role of financial incentives to physicians to recommend or
withhold specialized procedures (Okelo et al., 2001~. In addition, other
practices of these health systems related to larger quality improvement
goals may also serve to attenuate disparities. The VA, for example, has
instituted clinical decision support programs for physicians, which pro-
vide automated, time-sensitive and context-sensitive clinical reminders at
the point of care, such as prescription checks and preventive care infor-
mation. These clinical supports rely on a computerized patient record
OCR for page 150
50
UNEQUAL TREATMENT
system that provides patients' medical and social histories, discharge sum-
maries and progress notes, allergies, prior laboratory results, and other
information. Clinical reminder notifications provided through these com-
puterized data systems are largely based on the VA's national clinical
practice guidelines. In addition, clinical care is evaluated relative to per-
formance measures in six domains (quality, functional status, patient sat-
isfaction, access, cost, and healthy communities). Most of the measures
used to assess progress in these domains are based on "best practices"
formally supported by evidence-based medicine (Swift, 2001~. Such prac-
tice guidelines, as will be discussed in Chapter 5, may help to reduce varia-
tions in care due to clinical discretion and/or uncertainty.
Significantly, some evidence also suggests that when patients' race or
ethnicity is unknown (e.g., when treatment decisions are made by a group
of conferring physicians based solely on clinical data), racial and ethnic
disparities in care may be attenuated. Okelo et al. (2001) assessed whether
racial differences in recommendations for cardiac revascularization per-
sisted when patients' race or ethnicity is unknown. The authors described
the treatment decision-making procedures of cardiologists at the Cleve-
land VA Medical Center, who review clinical data of each patient consid-
ered for revascularization absent information about patients' race or
ethnicity. Following this procedure, Okelo et al. found no overall racial
differences in recommendation for revascularization. After adjusting for
patients' age, co-morbidities, location and number of coronary stenoses,
left ventricular function, and previous CABG, the authors found that
white patients were more likely to undergo CABG and African-American
patients were slightly (but not significantly) more likely to undergo PICA.
These findings lead Okelo et al. to conclude that "when only clinical fac-
tors are considered, the rates of recommendations for revascularization
will be similar for white and African-American patients; but the type of
revascularization procedure may differ by ethnicity and may depend, in
part, on clinical factors" (Okelo et al., 2001, p. 698~.
The Managed Care Revolution
Managed healthcare remains the predominant model of cost contain-
ment in an era of continuing escalation of healthcare costs and overall
health expenditures. Most managed care organizations employ various
forms of either supply-side (i.e., incentives to healthcare providers to prac-
tice frugally) or demand-side (i.e., incentives to patients to constrain the
use of services) cost containment strategies, or combinations of both as
part of managed competition strategies (Rice, this volume). As part of
broader efforts to contain costs, improve the quality of care, and increase
market share, some managed care organizations employ standardized
OCR for page 151
PATIENT- AND SYSTEM-LEVEL FACTORS
151
practice protocols and collect data on patient satisfaction and outcomes of
care. As such, managed care offers the potential to help eliminate dispari-
ties in healthcare. In many other areas, however, managed care has intro-
duced new institutional dynamics that may enhance the conditions in
which racial and ethnic disparities in healthcare can occur.
Utilization review and practice guidelines, for example, may be used
by some managed care organizations (MCOs) to ensure that physicians
provide services deemed medically appropriate. In this vein, it may be
assumed that prospective utilization management, when applied in a stan-
dardized fashion, offers the prospect of ensuring that clinical care is con-
sistent across patient groups. As noted earlier in this chapter, however,
the subjectivity and ambiguity of clinical situations make standardized
practice difficult, and guidelines cannot be developed for all clinical con-
tingencies (Bloche, 2001~. As a result, utilization managers must, in many
instances, authorize reimbursement under conditions where considerable
ambiguity and uncertainty exist. Under these conditions, advocacy by
committed clinical caretakers may influence utilization managers' deci-
sions. Typically, such advocacy is more likely to occur where patients
and their providers have an established relationship and where providers
have the time and resources to pursue claims. Minority patients, as noted
below, are less likely than whites to receive care from private physicians
and are less likely to have a regular primary care provider even when
compared to whites at the same insurance level (Lillie-Blanton et al., 2001~.
It is therefore possible that minorities may be less likely to benefit from
the advocacy of their provider. The outcomes of competition for resources
within a plan also hinge on utilization managers' discretion. With the
exception of the studies cited below (see, for example, Lowe et al., 2001),
there has been little research into subjective influences on utilization re-
viewers' decisions in ambiguous cases. Possible influences may include
different degrees of sponsorship and advocacy on behalf of patients from
their provider, which may be associated with patients' socioeconomic sta-
tus, and utilization managers' assumptions about which patients are most
likely to appeal utilization decisions (Bloche, 2001~.
Another supply-side constraint employed by many MCOs is the prac-
tice of cost control via Revolution of financial risk, thereby shifting re-
sponsibility for cost control to practicing physicians. Economic rewards
for frugality and penalties for costly tests, treatments, and referrals have
become common in contemporary clinical practice (Rice, 1997; Rice, this
volume). The result has been increased reliance on the discretion of
gatekeeping clinical caretakers to set limits and manage scarce resources.
As noted in the model depicted in Figure 3-1, such discretion may allow
cognitive, affective, social and cultural factors to influence clinical discre-
tion in racially disparate ways. It may also affect medical resource alloca-
OCR for page 152
52
UNEQUAL TREATMENT
lion decisions, in that physicians' suspicions and fears about which pa-
tients will protest or sue if denied a test or treatment may influence (even
at a subconscious level) the distribution of resources (Bloche, 2001~.
While more research must be conducted to fully test these hypoth-
eses, evidence indicates that low-income and ethnic minority patients are
less likely to have a regular provider, are more likely to be denied claims,
and are less satisfied with many aspects of the care they receive in man-
aged care settings. In a study of low-income African-American, Hispanic,
and white patients enrolled in managed care and fee-for-service plans in
four states, Leigh and colleagues found that for all three groups, those
enrolled in managed care plans were less likely to have a regular provider
than those enrolled in fee-for-service plans (Leigh, Lillie-Blanton, Mar-
tinez, and Collins, 1999~. African-American and Hispanic patients en-
rolled in managed care plans, however, were more likely than whites en-
rolled in MCOs to lack a regular provider, as approximately two of every
five (38% among African Americans and 42% among Hispanics) lacked a
regular provider, compared with 27% of whites enrolled in such plans. In
addition, African-American patients enrolled in managed care plans were
more than twice as likely as African Americans enrolled in fee-for-service
plans to report that they did not obtain needed care. Further, when asked
about "the extent to which your physician cares about you," Hispanic
patients enrolled in managed care plans were nearly twice as likely as
Hispanics enrolled in fee-for-service plans to rate their physicians' level
of concern as "fair" or "poor" (Leigh, Lillie-Blanton, Martinez, and Collins,
1999~.
Similarly, Phillips et al. (2000) used 1996 Medical Expenditure Panel
Survey (MEPS) data to compare the experiences of 22,087 African-Ameri-
can, Hispanic, Asian-American, and non-Hispanic white patients enrolled
in either managed care plans or other types of health systems (e.g., fee-
for-service plans). Overall, survey respondents reported generally high
levels of satisfaction with care, but minorities reported experiencing
greater barriers to care than white patients. In particular, Hispanics expe-
rienced the greatest difficulty of the surveyed groups in obtaining care
(24%), followed by Asian Americans (16%~. Three in ten Hispanics re-
ported lacking a usual source of care, as did two in ten African Americans
and 21% of Asian Americans. Whites were least likely to report these
barriers to care. In addition, and in contrast to Leigh et al. (1999) above,
Phillips et al. found that among all racial and ethnic groups, those en-
rolled in managed care plans were more likely to report having a usual
source of care than those enrolled in non-managed care plans. Minorities
enrolled in managed care plans, however, tended to experience greater
dissatisfaction with their usual source of care than those not enrolled in
managed care plans. Asian Americans enrolled in managed care plans
OCR for page 153
PATIENT- AND SYSTEM-LEVEL FACTORS
153
were 10 times more likely than Asian Americans enrolled in other types
of plans to express dissatisfaction with their usual source of care; Hispan-
ics enrolled in managed care plans were 4 times more likely to express
this belief; while whites enrolled in managed care plans were only 1.5
times more likely than whites enrolled in non-managed care plans to en-
dorse this view.
Research also suggests that managed care organizations' gatekeeper
policies may pose greater barriers to care for minority patients. Lowe et
al. (2001), for example, assessed racial differences in rates of gatekeeper
approval for emergency department (ED) services sought by more than
15,000 African-American and white patients at an urban hospital. Nearly
three-fourths (73%) of the ED visits analyzed were by African-American
patients, and over two-thirds (67%) of visits were by Medicaid beneficia-
ries. Following a triage assessment by ED staff, 4.4% of visits were denied
authorization for services, most commonly because they were deemed
"minor" or non-urgent. African-American patients were more likely to
receive low triage scores upon presentation; however, after adjusting for
patients' age, gender, day and time of ED visit, type of MCO and triage
score, African Americans were nearly 1.5 times more likely to be denied
authorization for care. Patients who were covered by a Medicaid MCO or
those covered by MCOs with mixed Medicaid and commercial patient
populations were also more likely than those covered by purely commer-
cial MCOs to be denied authorization for care. The authors note it un-
likely that the gatekeepers who approved or denied authorization knew
the race or ethnicity of patients presenting in the ED, as they generally did
not know the patients and were not informed by ED staff of the patients'
race. Therefore, these disparities could have emerged from other sources,
such as ED staff's initial triage assessments, advocacy efforts by primary
gatekeepers on behalf of patients (as discussed above), or other unmea-
sured factors (Lowe et al., 2001~.
Finally, some of the most significant support for the hypothesis that
managed care may pose greater barriers to care for racial and ethnic mi-
norities than whites is provided by Tai-Seale and colleagues (Tai-Seale,
Freund, and LoSasso, 2001~. Using a "natural experiment," the authors
assessed the differential effects of mandatory enrollment in managed care
plans on use of clinical services by African-American and white Medi-
caid beneficiaries. A "difference-in-differences" econometric approach
controlled for both time trends in demand for services and for fixed char-
acteristics of beneficiaries that may have affected their use of services.
African-American beneficiaries, including both children and adults, ex-
perienced significant declines in the use of physician services relative to
whites. This relationship was found even when trends in service use un-
related to managed care were controlled by comparing service use to ben-
OCR for page 154
54
UNEQUAL TREATMENT
eficiaries not subject to mandatory enrollment in managed care plans (Tai-
Seale et al., 2001~.
Supply-Side Cost Containment and
Demand for Clinical Services
When patients are well insured, demand for clinical services is not
constrained by demand-side prices. Because of low co-payment and/or
generous insurance coverage of healthcare expenses, these patients will
tend to display a higher demand for clinical services. Such is the case
when previously uninsured or underinsured patients are provided with
better health insurance, as their use of services (and subsequently, their
healthcare costs) increases. In these circumstances, health plans will often
use supply-side constraints to encourage doctors to engage in more frugal
practice. These cost-containment efforts may involve capitation (provid-
ing a set fee for all patients seen in a health system or practice), Revolution
of financial risk to providers, or other practices (Rice, this volume).
Similarly, limitations on the availability of physicians or resources within
hospitals or clinics may also induce supply-side constraints. These sup-
ply-side constraints can engender demand-supply mismatches within
hospitals (Ioskow, 1981) and other clinical institutions, as patients will be
less able to access all desired providers or services.
These demand-supply mismatches have the potential to contribute to
racial and ethnic disparities in care. Excess demand for a hospital's ser-
vices creates multiple internal queues for these services (Harris, 1979~.
Competition for these services within institutions may turn on the ability
of providers to use their influence in advocating for their patients. As
Bloche (2001) observed, "Absent bright-line, easy-to-apply criteria for pri-
oritizing among patients in a queue, the politics of personal influence and
professional hierarchy shape resource allocation. Attending physicians
with the professional stature and/or political skills to push to the head of
the queue in clinically ambiguous situations will do so on behalf of the
patients to whom they feel most committed. Conversely, house staff and
less influential attending physicians will have more difficulty making their
way up the queue" (Bloche, 2001, p. 107~.
As noted above, racial and ethnic minority patients are less likely to
be seen by a private physician, or to have a regular primary care provider,
even when insured at the same level as whites (Lillie-Blanton et al., 2001~.
Moreover, they are more likely to receive care in hospital clinics and other
settings characterized by rapid staff turnover and lack of continuity of
care providers. Under these circumstances, it is reasonable to assume that
physician advocacy on behalf of patients will be less likely, either because
the physician is less familiar with patients that he or she does not regu-
OCR for page 155
PATIENT- AND SYSTEM-LEVEL FACTORS
155
larly treat, or because resource constraints such as capitation prevent phy-
sicians from meeting all patients' demands for services (Rice, this vol-
ume). Therefore, patients cared for by physicians in settings that support
continuity of clinical relationships may have preferred access to services
when demand-supply mismatch conditions exist.
Legal and Regulatory Policy and Healthcare Disparities
A number of legal and regulatory mechanisms exist that, in theory,
may serve to remedy discriminatory healthcare practice. In some cases,
however, these mechanisms are insufficient by themselves to address dis-
criminatory practices, or cannot be implemented without addressing sig-
nificant obstacles. A few of these mechanisms are briefly described be-
low, as a means of providing examples of how legal and regulatory tools,
while well-intended, often fail to address the complexity of racial and eth-
nic discrimination in healthcare.
Medical Tort Law and Clinical Discretion
Medical malpractice law, in some cases, has served as an effective re-
sponse to departures from standards of competent practice. Its application
to the problem of healthcare discrimination, however, has been limited.
In theory, medical malpractice law prescribes a unitary level of
care, regardless of health insurance status or ability to pay. Tort doc-
trine assumes that a "correct" standard of care can be discerned from
physician-experts through the adversary process. Yet, as noted above,
clinical practice patterns and styles vary widely. Without high-quality
data about the efficacy of alternative approaches, physician-experts
cannot provide testimony that distinguishes scientifically between
"correct" and "incorrect" clinical practice variations. So long as the
care at issue in a medical malpractice case adhered to one or another
widely accepted practice variation, it can be defended by resorting to
like-minded physician-experts. Without empirical evidence that the
practice variation at issue is "wrong," the requirement that plaintiffs
shoulder the burden of proof on the issue of negligence in tort cases
poses a high barrier to legal success. The lack of such evidence poses
another obstacle to malpractice plaintiffs. Plaintiffs must shoulder the
burden of proof as to whether the negligence they allege was in fact
the cause of the harm that occurred. In the absence of high-quality
evidence concerning the comparative efficacy of alternative courses of
treatment, proof that a defendant physician's choice of one treatment
over another resulted in harm (or a diminished probability of a favor-
able outcome) is more difficult. Racial disparities in care that fall
OCR for page 156
56
UNEQUAL TREATMENT
within the range of widely accepted clinical practice variations are thus
not easily amenable to correction through the operation of medical
malpractice law (Bloche, 2001~.
Moreover, only a small proportion of arguable errors of clinical judg-
ment arguable based on empirical grounds for preferring one approach
over another result in medical malpractice suits (Weller, 1993~. Even
smaller proportions yield monetary settlements or judgments, and poor
people and members of disadvantaged minority groups are less likely
than other Americans to sue their doctors (Burstin et al., 1993~. Medical
malpractice law is therefore of weak utility as a mechanism to address
racial and ethnic discrimination in healthcare.
Emergency Medical Treatment and Active Labor Act
The federal Emergency Medical Treatment and Active Labor Act
(EMTALA)2 requires federally funded hospitals (e.g., those that partici-
pate in Medicare or Medicaid) that operate emergency rooms to screen
all emergency room patrons for "emergency medical conditions" regard-
less of patients' ability to pay, and to provide stabilizing treatment for
emergency conditions. Further, these hospitals are required to refrain
from discharging patients or transferring them to other facilities (also
known as "patient dumping") on economic grounds. Judicial interpre-
tation of EMTALA, however, has been criticized as having weakened
the law's force as a deterrent to disparate treatment in the emergency
room (Bloche, 2001~. Federal appellate court panels in several circuits
have ruled that the mandatory emergency screening examination re-
quired by EMTALA need not meet national standards of care, but rather,
should conform only to the screening hospital's regular practice. Plain-
tiffs, as a result, commonly experience difficulty pursuing suits alleging
violations of EMTALA, as they must challenge local hospital policy, of-
ten without the assistance of physicians familiar with emergency room
screening practice at the hospital they intend to sue or other evidence of
violations of hospital emergency room procedures. As Bloche (2001)
notes, "the resulting 'code of silence' problem is obvious: avoidance of
the 'code of silence' barrier was a principal reason for the shift from
community to national standards of care in medical malpractice law"
(Bloche, 2001, p. 110~. The difficulties encountered by plaintiffs in suing
to enforce EMTALA may lead to cursory evaluation and transfer or dis-
charge of members of disproportionate numbers of minority patients,
whether because of no or insufficient insurance, racial discrimination,
or unconscious bias. To add to this difficulty, state laws mandating
2EMTALA, 42 U.S.C. §1395dd (1995~.
OCR for page 157
PATIENT- AND SYSTEM-LEVEL FACTORS
157
emergency room screening have generally been construed and applied
with similar permissiveness (Rosenblatt et al., 1997~.
The Unfulfilled Potential of Title VI
Title VI of the Civil Rights Act of 1964 bars discrimination in health-
care and other services by all entities that receive federal funds. Title VI
therefore applies to the vast majority of U.S. hospitals and clinics, given
the large percentage of these care settings that rely significantly or in part
on Medicaid or Medicare reimbursement. Significantly, the law extends
beyond intentional discrimination to prohibit many facially neutral prac-
tices that may result in disparate negative effects on racial and ethnic mi-
norities and other disadvantaged groups. The impact of Title VI in deseg-
regating healthcare and ensuring the equitable treatment of all patients
has been enormous. Despite resistance to desegregation in the early years
following the law's passage, for example, the enforcement of Title VI by
federal investigators, aided by activists and health professionals, resulted
in many previously segregated hospitals opening their doors and wards
to all patients who could pay (Smith, 1999~. Evidence of discrimination in
some sectors of the healthcare industry, however, remained. Discrimina-
tory practices such as denial of admitting privileges to African-American
physicians,3 refusal of admission to patients lacking attending physicians
with staff privileges, high prepayment requirements for black patients,
and discriminatory routing of ambulances continued in some instances
(Smith, 1999~. In these cases, the DHHS Office for Civil Rights (OCR) has
enacted such measures as revising requirements for staff privileges, elimi-
nating prepayment requirements, and requiring changes in ambulance
routes (Rosenbaum et al., 2000~.
Despite these gains, some argue that Title VI has yet to fulfill its po-
tential as a tool to eliminate discrimination in healthcare (Perez, this vol-
ume; Bloche, 2001~. For example, the federal regulations promulgated
pursuant to Title VI did not offer detailed compliance instruction to
healthcare institutions (Rosenbaum, 2000; U.S. Commission on Civil
Rights, 1999), making it difficult for even the well-intended institutions to
assess what practices may run afoul of the law. More significantly, fed-
eral Title VI regulations held that Medicare's payments to physicians do
not constitute "federal financial assistance" under Title VI. This rule
3Some hospitals pursued the facially neutral strategy of refusing to grant privileges to
physicians who were not members of their local medical societies. The difficulty for Afri-
can-American doctors (and their patients) in some localities, was that these medical societies
(which received no "federal financial assistance" and were thus beyond Title VI's reach)
refused admission to blacks (Smith, 1999~.
OCR for page 158
58
UNEQUAL TREATMENT
meant that private physicians were not subject to Title VI, despite the fact
that virtually all other federal payments to private actors are treated by
the regulations as "federal financial assistance," triggering Title VI pro-
tections (Rosenbaum, 2000~. If physicians who accept Medicare were sub-
ject to Title VI, the law would have given DHHS (and private plaintiffs) a
powerful civil rights enforcement tool, applicable not only to racial dis-
parities in the care provided to Medicare patients but also to disparate
treatment of non-Medicare patients by physicians who accept Medicare.
Given that most physicians accept Medicare, and given their important
role as key decisionmakers with respect to use of hospital resources and
services, extending the reach of Title VI to Medicare coverage of physi-
cian services would subject most of the private healthcare sector to Title
VI enforcement.
The reach and effectiveness of Title VI can be improved by address-
ing these gaps. More specific regulatory guidance, based on empirical
research regarding potential disparate impact and means to improve ac-
cess to and quality of care for minority patients, will enable healthcare
institutions to develop more finely crafted policies and will help enforce-
ment efforts by drawing distinctions between allowable and potentially
illegal practices (Bloche, 2001~. More robust DHHS monitoring and en-
forcement, similar in scope to the early efforts of the Department follow-
in~ Passage of Title VI can held to re-establish federal leadership work
J ' 1 1
1 e1 1 · · ~ · ~ 1 · · ~ · ~~ · e1 ~ ~~\ T 1 1 · ~ -
toward the elimination of care disparities tomato, Filmy). In addition, ap-
plication of Title VI to private physicians who accept Medicare would
extend the law's reach to a significant segment of the healthcare industry.
Furthermore, the application of Title VI beyond intentional discrimi-
nation to include policies that may create disparate racial impacts could
be an important tool for civil rights enforcement. Disparate impact could
be assessed using institution-specific statistical evidence of disparities in
healthcare provision. Such evidence may suffice to state a prima facie case
of discrimination, requiring a healthcare provider to justify policies and
practices that result in racially disparate clinical decisions (Barnes and
Weiner, 1999~. Establishing proof of institution-specific disparities and
of causal links between such disparities and particular policies and prac-
tices will pose significant challenges. The possibility of institution-
specific databases sufficiently powerful to serve this probative purpose is
speculative, but the ongoing effort to establish electronic clinical record-
keeping (see Chapter 7) may make such evidence increasingly accessible
to civil rights enforcement authorities.
Despite the promise of this type of data, however, new challenges
have emerged within the last year that will limit private parties' ability to
seek legal relief under Title VI from policies with disparate racial impact
(Perez, this volume). In Alexander v. Sandoval, the U.S. Supreme Court
OCR for page 159
PATIENT- AND SYSTEM-LEVEL FACTORS
159
held that Title VI did not create a private right of action concerning poli-
cies with disparate impact, absent discriminatory intent. This action there-
fore places the greatest burden of civil rights enforcement with U.S.
DHHS, which will shape Title VI's future as a health policy tool through
its civil rights enforcement policies.
SUMMARY
This chapter presents a review of evidence regarding potential sources
of racial and ethnic differences in healthcare, once access-related factors
such as patient education, income, and insurance status are held constant.
Consistent with the committee's definitions of differences, disparities, and
discrimination in care, several sources are identified. Those related to pa-
tients' preferences, needs, and racial or ethnic differences in the clinical
appropriateness of care may contribute to differences in the quality or
intensity of care provided, but these are not sources of healthcare dispari-
ties, as they do not imply undue differential treatment on the basis of race
or ethnicity. Disparities in care, on the other hand, likely emerge from a
range of sources, such as characteristics of healthcare systems and the
legal and regulatory context of healthcare delivery. In the next chapter,
sources of disparities arising from the clinical encounter will be examined
in greater detail.
Finding 3-1: Many sources including health systems, healthcare
providers, patients, and utilization managers may contribute to
racial and ethnic disparities in healthcare.
Evidence suggests that several sources may contribute to healthcare
disparities, including healthcare providers, patients, utilization man-
agers and healthcare systems. In the current era of healthcare deliv-
ery, clinical decision-making increasingly involves this large num-
ber of individuals, who are subject to an array of systems influences
that may contribute to healthcare disparities.
Representative terms from entire chapter:
racial differences
