Content uploaded by Chyke Doubeni
Author content
All content in this area was uploaded by Chyke Doubeni on Dec 04, 2015
Content may be subject to copyright.
88 Journal of the National Cancer Institute Monographs, No. 35, 2005
R ACIAL DISPARITIES IN CANCER CARE AND SURVIVAL
Disparities and Survival Among Breast Cancer Patients
Terry S. Field , Diana S. M. Buist , Chyke Doubeni , Shelley Enger , Hassan
Fouayzi , Gene Hart , Eli J. Korner , Lois Lamerato , Donald J. Bachman , Jennifer
Ellis , Lisa Herrinton , Mark C. Hornbrook , Rick Krajenta , Liyan Liu, Janice Yao
Background: Although rates of survival for women with
breast cancer have improved, the survival disparity between
African American and white women in the United States
has increased. Purpose: To determine whether this survival
disparity persists in an insured population with access to
medical care. Methods: In this retrospective cohort study,
we extracted data from the tumor registries of six nonprofi t,
integrated health care delivery systems affi liated with the
Cancer Research Network and assessed the survival of
African American ( n = 2276) and white ( n = 18 879) female
enrollees who were diagnosed with invasive breast cancer
from January 1, 1993, through December 31, 1998. Cox pro-
portional hazards regression was used to estimate the death
rate among African American women relative to that of
white women after adjustment for potential explanatory
factors including stage at diagnosis, tumor characteristics,
and treatment. Results: Five-year survival was lower for
African American women (73.8%) than for white women
(81.6%). African American women were less likely to have
tumor characteristics with good prognosis. Controlling for
age at diagnosis, stage, grade, tumor size, and estrogen and
progesterone receptor status, the adjusted hazard rate ratio
for African American women was 1.34 (95% confi dence
interval = 1.22 to 1.46). Similar risks were found among
women ages 20 – 49 and 50 and older. Controlling for treat-
ment slightly lowered the hazard rate ratio to 1.31 (95%
confi dence interval = 1.20 to 1.43). Conclusions: Among
women with invasive breast cancer, being insured and
having access to medical care does not eliminate the survival
disparity for African American women. [J Natl Cancer Inst
Monogr 2005;35:88 – 95]
Although rates of survival for women with breast cancer have
improved during the last two decades, the survival disparity be-
tween African American and white women has increased ( 1 – 3 ) .
A number of potential explanatory factors have been investigated
including the stage of disease at the time of diagnosis ( 4 – 7 ) ;
access to health insurance and medical care ( 8 , 9 ) ; variability in
the aggressiveness of treatment ( 10 – 15 ) ; environmental, socio-
economic, and psychosocial factors ( 16 – 19 ), and differences in
underlying tumor biology ( 20 – 26 ) .
Few studies have been able to assess outcomes in popula-
tions with apparent equal access to medical care. Several
studies based in the U.S. Department of Defense healthcare
system have found a higher risk of death for African American
women with breast cancer despite being treated in an equal
access system with standardized treatment approaches ( 27 , 28 ) .
A study of survival among women with breast cancer enrolled
in a single managed care organization found differences be-
tween African American and white women that were mitigated
by control for treatment and eliminated by controlling for indi-
cators of socioeconomic level ( 29 ) . However, this study was
based in only one health care delivery site. Because clinical
trials ensure equal treatment, they are another source of equal
access to care. However, trials of breast cancer therapy have
enrolled few African American women ( 30 – 32 ) , and partici-
pants have been matched on specifi c cancer characteristics,
limiting confi dence that the trial outcomes are representative
of the range of breast cancers found among African American
women.
To contribute further to the evidence on possible asso-
ciations between access to health insurance and medical care
and survival disparities, we conducted a study based on data
from the “ virtual tumor registry ” of the Cancer Research
Network (CRN) ( 33 ) . The CRN is a consortium of research
organizations affi liated with nonprofi t integrated health care
delivery systems and the National Cancer Institute. The CRN
has instituted a virtual tumor registry that consists of data sets
located at the individual health care sites with a standard
variable set and coding scheme, allowing merges of data for
specifi c studies.
Our objectives in this study were to determine whether dis-
parities in survival between African American and white women
with breast cancer would persist in a large, insured, multiracial
population without obvious barriers to access to medical care,
and to identify any factors associated with racial disparities that
could serve as points of possible interventions at the healthcare
system level.
Affi liations of authors: Meyers Primary Care Institute, University of Massa-
chusetts Medical School, Fallon Foundation, and Fallon Community Health
Plan, Worcester, MA (TSF, CD, HF); Center for Health Studies, Group Health
Cooperative, Seattle, WA (DSMB, GH); Center for Research and Evaluation,
Kaiser Permanente Southern California, Pasadena, CA (SE, JY); Clinical Re-
search Unit, Kaiser Permanente Colorado, Denver, CO (EJK, JE); Center for
Health Services Research, Henry Ford Health System, Detroit, MI (LL, RK);
Center for Health Research, Northwest/Hawaii, Kaiser Permanente Northwest,
Portland, OR (DJB, MCH); Division of Research, Kaiser Permanente Northern
California, Oakland, CA (LH, LL).
Correspondence to: Terry S. Field, DSc, Meyers Primary Care Institute, 630
Plantation St., Worcester, MA 01605 (e-mail: tfi eld@meyersprimary.org ).
See “ Notes ” following “ References. ”
DOI: 10.1093/jncimonographs/lgi044
© The Author 2005. Published by Oxford University Press. All rights reserved.
For Permissions, please e-mail: journals.permissions@oxfordjournals.org.
by guest on August 19, 2015http://jncimono.oxfordjournals.org/Downloaded from
Journal of the National Cancer Institute Monographs, No. 35, 2005 89
M ETHODS
Setting and Population
The CRN consists of the research programs, enrollee pop-
ulations, and databases of 11 integrated health care organiza tions
that are members of the HMO Research Network. The health
care delivery systems participating in the CRN are Group
Health Cooperative, Harvard Pilgrim Health Care, Henry Ford
Health System/Health Alliance Plan, HealthPartners Research
Foundation, the Meyers Primary Care Institute of the Fallon
Healthcare System/University of Massachusetts, and Kaiser
Permanente in six regions: Colorado, Georgia, Hawaii, Northwest
(Oregon and Washington), Northern California, and Southern
California. The 11 health plans have nearly 10 million enrollees.
The CRN conducts collaborative research on variations in cancer
prevention and treatment policies and practices.
The six CRN sites participating in this study are predomi-
nantly staff-model managed care organizations with tumor regis-
tries that capture cancer diagnoses for their full population of
enrollees (Henry Ford Health System in Detroit, MI; Group
Health Cooperative in Seattle, WA; and Kaiser Permanente in
four regions: Colorado, Northwest [Oregon and Washington],
Northern California and Southern California). Collectively, these
six sites provided care to over 7 million enrollees in 1998. Study
subjects were health plan members who were diagnosed with
incident invasive breast cancer between January 1, 1993, and
December 31, 1998, while they were enrolled in one of these
managed care organizations. The study was limited to female
subjects who were age 20 years or older at the time of the
breast cancer diagnosis.
The protocol for this study was reviewed and approved by
the institutional review boards of all participating institutions.
D e fi nitions
The tumor registries at these sites contain data elements con-
sistent with the Surveillance, Epidemiology, and End Results
(SEER) program’s registries. We limited our study to invasive
cancers, defi ned using the behavior and SEER summary stage
variables. Cancer type was established by combining the regis-
tries’ primary site and morphology variables. These variables
were coded using International Classifi cation of Diseases for
Oncology , 3rd edition (ICD-03). Cancers that did not constitute
solid tumors were identifi ed and excluded based on morphology
codes; breast cancer was identifi ed from the remaining diagnoses
based on site codes. We used morphology codes to categorize
breast cancer into comedo, ductal, ductal/lobular, infl ammatory,
lobular, medullary, mucinous, Paget’s, papillary, phyllodes, tubu-
lar, or other histology. The tumor registries coded tumor grade as
well differentiated, moderately differentiated, poorly differenti-
ated, not differentiated, or unknown. Registries at fi ve sites also
coded the stage of cancer at diagnosis using the American Joint
Committee on Cancer (AJCC), 5th edition staging scheme, which
we categorized as stages I, IIA, IIB, III, IV, and unknown/un-
staged. Tumor size was available for most women and was cate-
gorized as less than 2.0 cm, 2.0 to less than 5.0 cm, 5.0 cm or
larger, diffuse/other, or unknown. The sites’ tumor registries had
information on the estrogen receptor test results for a portion of
their breast cancer patients, categorized as positive, negative, test
not done, or unknown. We categorized women with missing
receptor status as having unknown receptor status. Results of
progesterone receptor tests were less commonly available. All
site registries included data on the receipt of surgery, radiation
therapy, chemotherapy, and hormone therapy during the initial
treatment phase.
We linked all enrollees identifi ed in the cancer registry data to
enrollment fi les to extract a complete enrollment history. At the
time of the data extraction, all cancer registry and enrollment
data at each site were current as of December 31, 2003. Follow-
up began on the diagnosis date and ended on the earliest among
the dates of death, disenrollment, or December 31, 2003. We
calculated the length of enrollment in the health plan before the
date of the breast cancer diagnosis and categorized it as less
than 30 days, 30 to less than 180 days, 180 days to less than
1 year, 1 year to less than 4 years, 4 years to less than 10 years, or
10 years or longer.
Demographic characteristics were drawn from the tumor
registry data. Only non-Hispanic white and African American
women were eligible for the study. To exclude Hispanic women,
the race and ethnicity variables were combined and women
were identifi ed as Hispanic according to the Hispanic variable,
including those with a Spanish surname ( 34 ) . Age at the time of
diagnosis was directly extracted from the registry data.
Endpoints
We evaluated three end points: death from any cause ( n =
4942), death from breast cancer ( n = 1491), and death from any
cancer ( n = 1777). Follow-up for all-cause mortality in this
population is very thorough because it is directly associated with
continued enrollment in the health plan. However, cause of death
data in the tumor registries are not as consistently collected and
were missing or inconclusive for 50% of the deaths. Cause of
death was coded in several ways (ICD-9CM, ICD-10CM, and
ICDO); therefore, codes were manually searched to identify
deaths from cancer. Women who disenrolled from their health
plan during the follow-up period for reasons other than death
were censored at the time of disenrollment ( n = 3843, 18% of
the cohort).
Statistical Analyses
We began analyses by determining the distribution of age, the
tumor characteristics, and the previous enrollment history for
African American and white women and calculating 5-year sur-
vival according to race. We conducted subanalyses for AJCC
stage from the sites with available data.
We divided the cohort by age at diagnosis into those ages
20 – 49 years and those aged 50 years or older to assess the possi-
bility of age-specifi c racial disparities in survival. Age 50 was
chosen as the split point as a proxy for menopausal status ( 7 , 35 , 36 ) .
We calculated the frequency distributions for tumor characteris-
tics for African American and white women in each age group.
We fi t Cox proportional hazards regression models in the
entire cohort and separately in the two age groups to determine
the extent to which stage at diagnosis and tumor characteristics
explained any disparities in survival. The proportional hazards
assumption for race in the Cox regression models was visually
evaluated by review of the Schoenfeld residuals ( 37 ) and graph-
ical assessment of the Cox proportional hazards functions ( 38 ) .
The assumption was not met during the initial 84 days after
by guest on August 19, 2015http://jncimono.oxfordjournals.org/Downloaded from
90 Journal of the National Cancer Institute Monographs, No. 35, 2005
diagnosis, but from that point in time the hazards remained
proportional. We fi t all Cox models beginning at day 85 after
diagnosis and described outcomes during the initial 84 days
separately. We tested interaction terms between each group of
tumor characteristic variables and race and between CRN site,
year of diagnosis, and the treatment variables and race, and we
found none to be signifi cant.
We investigated the possibility that treatment explained any
differences in survival through several approaches. We constructed
Cox regression models that added control for variables indicating
receipt of each of the four major types of therapy in the entire co-
hort and the two age groups and ran a Cox regression model in the
subgroup of women age 50 years and older who were diagnosed
at local stage with small tumors and who received surgery, radia-
tion therapy, and either chemotherapy or hormone therapy.
We also assessed the frequency distribution of tumor charac-
teristics by race for women diagnosed at local stage and ran Cox
regression models predicting mortality within each age group.
All analyses were conducted using SAS version 9.1.
R ESULTS
Characteristics of the Study Population
From 1993 through 1998, 2276 African American women and
18 879 white women were diagnosed with invasive breast cancer
at the six CRN sites. Unadjusted 5-year survival was lower for
African American women (73.8%) than for white women
(81.6%) ( Table 1 ). During the 5 years following the breast cancer
diagnosis, 13% of African American women and 12% of white
women disenrolled from their health plan for reasons other than
death. Fewer African American women had tumor characteris tics
associated with good prognosis: They had lower rates of being
diagnosed at local stage or in AJCC stages I and IIA, having
well-differentiated tumors, positive estrogen receptor status, and
small tumor size. Information on progesterone receptor status
was unavailable for the majority of women. Length of enrollment
prior to diagnosis was very similar for the two groups.
Age is closely associated with breast cancer survival; most
studies have found that younger, premenopausal women with
breast cancer are more likely to have tumor characteristics with
adverse prognoses ( 39 – 43 ) . African American women are gener-
ally diagnosed with breast cancer at a younger age than white
women ( 21 , 44 , 45 ) . Among the African American women in our
cohort, 35.1% were diagnosed while younger than 50 year of age,
compared to 20.9% of white women ( Table 1 ). Five-year survival
was lower for African American women than for white women in
both age groups. Among African American and white women,
younger women had less favorable tumor characteristics than
older women. African American women had less favorable tumor
characteristics than white women regardless of age. The distribu-
tions of stage at diagnosis, tumor grade, and estrogen receptor
status for older African American women were similar to that of
younger white women.
During the fi rst 84 days following diagnosis the hazard curves
for death for African American and white women were entwined
and crossed, and this time period was excluded from the multi-
variate analyses. During that period, 1.0% of the African
American women ( N = 22) and 0.8% of white women ( N = 157)
died, and 20 African American women and 103 white women
disenrolled from their health plans.
Multivariable Analyses of Survival
Within the remaining cohort of 2234 African American and
18 619 white women, we conducted Cox regression models for
the complete cohort and separately by age group ( Table 2 ).
African American women had increased risks of death compared
to white women regardless of age. Inclusion of potential con-
founding variables related to tumor characteristics lowered the
hazard rate ratio for African American women under age 50
years but had only a small effect among those age 50 years and
older. We conducted analyses in the subset of women enrolled in
the fi ve sites whose tumor registries included AJCC stage,
replacing SEER summary stage variables with AJCC stage. In-
clusion of AJCC stage lowered the additional risk of death
associated with being African American in the older age group:
for the entire cohort the hazard rate ratio was 1.24 (95% con-
fi dence interval [CI] = 1.12 to 1.39), for women under age 50
years the hazard rate ratio was 1.37 (95% CI = 1.10 to 1.71), and
for those age 50 years and older the hazard rate ratio was 1.20
(95% CI = 1.05 to 1.36). Despite the unfavorable tumor charac-
teristics of African American women, controlling for these
characteristics did not fully explain their higher risk of death.
Previous studies have documented that cancer-related deaths
are often misclassifi ed ( 46 – 48 ) . Therefore, our primary analyses
focused on death from any cause. However, to assess the possibil-
ity that our results were caused by increased death rates from
non – breast cancer conditions among African American women,
we conducted Cox regression models controlling for tumor char-
acteristics with death from breast cancer as the outcome and with
death from any cancer as the outcome. The adjusted risk of death
from breast cancer for African American women was 1.48 (95%
CI = 1.26 to 1.73); the adjusted risk of death caused by any cancer
for African American women was 1.49 (95% CI = 1.28 to 1.74).
We used a variety of approaches to assess the possibility that
differences in treatment might explain any racial disparities in
survival. Investigators have suggested that assessing survival by
stage at diagnosis provides a way for testing whether aggressive-
ness of treatment is a partial explanation for survival disparities
( 49 ) ; for women diagnosed at local stage, there are available
treatments with high success rates so that inadequate treatment
would be associated with large differences in outcome; for women
diagnosed at distant stage treatment is not highly successful so
survival disparities would be smaller. To assess this possibility,
we conducted Cox regression models predicting mortality sepa-
rately by stage at diagnosis as well as in subgroups of women
with positive tumor characteristics, including estrogen receptor –
positive status and those with well-differentiated tumors. Among
women diagnosed at local stage the hazard rate ratio for African
American women was 1.42 (95% CI = 1.23 to 1.64), at regional
stage 1.22 (95% CI = 1.07 to 1.39), and at distant stage 1.41 (95%
CI = 1.09 to 1.84). Among women with estrogen receptor –
positive tumors, the hazard rate ratio for African American
women was 1.30 (95% CI = 1.12 to 1.49). Among women with
well-differentiated tumors the survival disparity was eliminated —
the hazard rate ratio for African American women was 1.04 (95%
CI = 0.73 to 1.49), but this group only included 237 African
American women, and the confi dence interval was wide.
Although the treatment data available from the tumor reg-
istries were of limited specifi city, we were able to identify
whether any type of surgery, radiation therapy, chemotherapy,
or hormone therapy was provided to these women ( Table 3 ).
by guest on August 19, 2015http://jncimono.oxfordjournals.org/Downloaded from
Journal of the National Cancer Institute Monographs, No. 35, 2005 91
African American women in both age groups diagnosed at local
or regional stage were signifi cantly less likely than white women
( P <.0001) to receive surgery, radiation therapy, and hormone
therapy. Inclusion of treatment variables in the Cox regression
models lowered the hazard rate ratios: Among the full cohort, the
adjusted risk of dying for African American women was 1.31
(95% CI = 1.20 to 1.43); among women under age 50 years, the
risk was 1.33 (95% CI = 1.12 to 1.58), and among women age
50 years and older, the risk was 1.28 (95% CI = 1.16 to 1.43).
To determine whether African American and white women
with similar tumor characteristics who received similar treatment
had parallel outcomes, we assessed survival in the subgroup of
women who were age 50 years and older, diagnosed at local
stage, with tumors smaller than 2.0 cm, who received surgery,
radiation therapy, and either chemotherapy or hormone therapy.
Within this group the hazard rate ratio associated with being
African American was 1.35 (95% CI = 0.81 to 2.22). This was a
much smaller group that included only 136 African American
women, and the confi dence interval was wide.
Characteristics of Women Diagnosed at Local Stage
Prior studies have found that African American women are
diagnosed with breast cancer at later stages than white women. To
assess the extent to which aggressive efforts to shift the time of
diagnosis to a point when the cancers of African American women
Table 1. Characteristics of African American and white women diagnosed with invasive breast cancer in 1993 to 1998
African American Women White Women
All ( N = 2276) Age 20–49 ( N = 800) Age 50+ ( N = 1476) All ( N = 18 879) Age 20–49 ( N = 3944) Age 50+ ( N = 14 935)
% % % % % %
5-year survival 73.8 73.6 73.9 81.6 84.2 80.9
Age
20 – 44 20.3 57.9 10.1 48.5
45 – 54 28.9 42.1 21.8 23.2 51.5 15.7
55 – 64 24.2 37.6 25.0 31.7
65 – 74 17.8 27.4 24.9 31.5
75 – 84 7.7 11.9 13.9 17.5
85+ 1.1 1.8 2.8 3.6
SEER stage
Local 56.2 50.9 59.1 68.5 58.0 71.2
Regional 36.2 41.3 33.4 27.2 37.6 24.4
Distant 5.4 4.8 5.7 3.3 3.2 3.4
Unknown 2.3 3.1 1.8 1.0 1.2 1.0
Cancer grade
Well differentiated 10.5 6.4 12.7 18.3 11.5 20.0
Moderately differentiated 29.7 25.5 31.9 36.3 33.9 37.0
Poorly differentiated 41.0 50.8 35.8 25.0 37.7 21.7
Not differentiated 2.7 3.0 2.5 1.7 2.4 1.6
Unknown 16.1 14.4 17.1 18.6 14.5 19.7
Estrogen receptor
Positive 45.6 39.4 49.0 59.9 52.8 61.8
Negative 27.7 34.5 24.0 14.4 22.2 12.3
Not assessed 5.9 5.0 6.4 4.7 4.2 4.8
Unknown 20.8 21.1 20.6 21.1 20.8 21.1
Progesterone receptor
Positive 23.7 19.8 25.9 25.5 24.5 25.7
Negative 16.6 20.8 14.4 10.3 11.9 9.8
Not assessed 12.1 10.1 13.1 7.0 6.2 7.2
Unknown 47.6 49.4 46.6 57.3 57.4 57.2
Tumor size
<2.0 centimeters 37.1 32.0 39.8 57.4 47.3 60.1
2.0 to <5.0 centimeters 33.3 36.5 31.6 28.5 35.1 26.7
5.0+ centimeters 7.7 9.1 7.0 4.5 6.6 4.2
Diffuse/other 1.7 1.6 1.8 1.1 1.2 <0.1
Unknown 20.1 20.8 19.8 8.1 9.7 7.9
Enrollment before diagnosis
<30 days 1.1 1.9 1.0 1.1 2.1 1.0
30 – 180 days 4.1 5.5 3.4 4.2 7.2 3.4
180 – 365 days 4.3 5.9 3.4 3.1 5.1 2.6
1 – 4 years 15.0 19.1 12.7 15.2 23.0 13.1
4 – 10 years 25.1 33.9 20.3 25.3 31.1 23.8
10+ years 50.5 33.8 59.6 51.1 31.6 56.2
AJCC stage (5 sites) N = 1589 N = 542 N = 1047 N = 11 090 N = 2282 N = 8808
I 36.6 27.5 41.3 52.5 40.7 55.6
IIA 28.3 32.1 26.4 22.6 27.0 21.5
IIB 15.4 18.6 13.8 11.4 17.1 9.9
III 9.7 12.2 8.4 6.2 9.3 5.3
IV 6.0 5.7 6.1 3.3 2.8 3.5
Unknown/unstaged 4.0 3.9 4.1 4.0 3.2 4.2
by guest on August 19, 2015http://jncimono.oxfordjournals.org/Downloaded from
92 Journal of the National Cancer Institute Monographs, No. 35, 2005
were localized would lessen any survival disparities, we examined
tumor characteristics of women diagnosed at local stage ( Table 4 ).
Even with cancers diagnosed at local stage, African American
women were signifi cantly more likely to have tumor characteristics
with poor prognosis than white women in each age group ( P <.0001).
In parallel with our fi ndings across all stages, older African Ameri-
can women with breast cancer diagnosed at local stage had tumor
characteristics similar to those of younger white women.
D ISCUSSION
African American women with invasive breast cancer in this
study had worse survival than white women, despite having
health insurance and apparent equal access to medical care. This
was true for women younger and older than age 50 years, diag-
nosed at local, regional, or distant stage. African American
women were less likely to have tumor characteristics with good
prognosis; however, controlling for these characteristics did not
eliminate the survival disparity. The only subgroup in which
African American women did not have a survival disadvantage
was women with well-differentiated tumors. This group included
very few African American women, and the hazard rate ratio had
a wide confi dence interval. Several additional prognostic tumor
characteristics have recently been found to differ in frequency
among racial groups including high nuclear grade, necrosis, and
alterations in p53 and c- me ( 26 ) . These factors were not captured
in the tumor registries, and the extent to which they may explain
survival disparities is not yet fully understood. The possibility
that differences in tumor characteristics explain survival differ-
ences is still an open question.
Differences in overall health status and rates of comorbid
conditions might explain the differences found in all-cause mor-
tality. However, analyses that focused specifi cally on mortality
from breast or other cancers found even larger hazard ratios for
African American women.
We used several approaches to assess the extent to which
differences in treatment aggressiveness accounted for the racial
survival disparity. Although the rates of provision of the major
treatment modalities did differ by race, control for receipt of
these treatments did not eliminate the disparity. African Ameri-
can women remained at higher risk of death even in the subgroup
of women diagnosed at local stage with small tumors who re-
ceived surgery, radiation, and either chemotherapy or hormone
therapy. This does not eliminate the possibility that treatment dif-
ferences are an explanation. The treatment data available in the
tumor registries were limited and did not include information on
aspects of quality, delays in the timing of care, or whether treat-
ment regimens were actually completed, or details on the specifi c
types of surgery, radiation, or chemotherapy agents. Racial dif-
ferences in these aspects of treatment may exist despite insured
status and may result from either patient or health care system
issues or from differences in tumor biology.
All of the women in this study had health insurance and
were enrolled in managed care plans. Approximately 75% of the
Table 2. Proportional hazards models assessing mortality for African American and white women with invasive breast cancer *
All Women Age 20 – 49 Age 50+
Hazard Rate Ratio (95% CI) Hazard Rate Ratio (95% CI) Hazard Rate Ratio (95% CI)
Model including only race
African American 1.39 (1.28 to 1.51) 1.68 (1.43 to 1.97) 1.36 (1.24 to 1.51)
Model including confounders
African American 1.34 (1.22 to 1.46) 1.35 (1.14 to 1.60) 1.31 (1.18 to 1.46)
Age (continuous) 1.04 (1.04 to 1.05) 0.98 (0.97 to 0.99) 1.06 (1.06 to 1.06)
SEER stage
Local 1 (referent) 1 (referent) 1 (referent)
Regional 1.91 (1.79 to 2.04) 2.90 (2.46 to 3.43) 1.77 (1.64 to 1.90)
Distant 9.09 (8.17 to 10.12) 13.40 (10.32 to 17.41) 8.67 (7.70 to 9.77)
Ustaged/unknown 2.33 (1.88 to 2.89) 3.90 (2.45 to 6.21) 1.99 (1.56 to 2.53)
Cancer grade
Well differentiated 1 (referent) 1 (referent) 1 (referent)
Moderately differentiated 1.14 (1.03 to 1.26) 2.17 (1.37 to 3.45) 1.10 (0.99 to 1.23)
Poorly differentiated 1.69 (1.52 to 1.88) 2.88 (1.82 to 4.56) 1.65 (1.47 to 1.84)
Not differentiated 1.80 (1.46 to 2.23) 3.70 (2.10 to 6.52) 1.66 (1.30 to 2.12)
Unknown 1.33 (1.19 to 1.50) 2.52 (1.55 to 4.09) 1.28 (1.13 to 1.44)
Estrogen receptor
Positive 1 (referent) 1 (referent) 1 (referent)
Negative 1.48 (1.35 to 1.62) 1.78 (1.46 to 2.16) 1.41 (1.27 to 1.57)
Not assessed 1.37 (1.17 to 1.60) 1.63 (1.09 to 2.44) 1.36 (1.15 to 1.61)
Unknown 1.27 (1.13 to 1.43) 1.36 (1.01 to 1.83) 1.26 (1.11 to 1.44)
Progesterone receptor
Positive 1 (referent) 1 (referent) 1 (referent)
Negative 1.10 (0.98 to 1.23) 0.97 (0.75 to 1.28) 1.14 (1.01 to 1.29)
Not assessed 0.91 (0.78 to 1.05) 0.90 (0.61 to 1.32) 0.89 (0.76 to 1.05)
Unknown 0.97 (0.83 to 1.13) 1.15 (0.77 to 1.70) 0.92 (0.78 to 1.09)
Tumor size
<2.0 centimeters 1 (referent) 1 (referent) 1 (referent)
2.0 to <5.0 centimeters 1.50 (1.40 to 1.61) 1.60 (1.32 to 1.94) 1.46 (1.35 to 1.58)
≥5.0 centimeters 2.49 (2.23 to 2.78) 2.39 (1.87 to 3.06) 2.48 (2.18 to 2.81)
Diffuse/other 3.33 (2.64 to 4.20) 2.71 (1.64 to 4.46) 3.86 (2.96 to 5.04)
Unknown 1.38 (1.21 to 1.57) 1.64 (1.21 to 2.23) 1.32 (1.14 to 1.52)
* Excluding those who disenrolled or died in less than 85 days after diagnosis; controlling for age at diagnosis to breast cancer histology to enrollment history to
year of diagnosis to and health plan.
by guest on August 19, 2015http://jncimono.oxfordjournals.org/Downloaded from
Journal of the National Cancer Institute Monographs, No. 35, 2005 93
women in both racial groups had been enrolled in the same plan
for more than 4 years, and similar percentages of both groups of
women remained enrolled after the breast cancer diagnosis. This
implies similar access to care. However, a number of additional
factors may affect access. Each organization has a variety of plan
alternatives with varying copays and other fi nancial hurdles that
could affect access for enrollees with limited incomes. Medical
offi ces and facilities providing cancer-specifi c treatments may
not be located near the areas where African American enrollees
reside; problems with travel have been found to affect receipt
of some breast cancer treatments ( 50 , 51 ) . The not-for-profi t
managed care plans involved in this study serve predominantly
working and middle-class populations. This indicates that their
African American and white enrollees are not likely to be either
impoverished or wealthy. However, there may be substantial
differences in the level of fi nancial resources between the two
groups. The out-of-pocket expenses involved in cancer care
( 52 – 54 ) , and the frequent need for time taken from work for both
the patients and their care givers ( 55 ) , could also affect the timing
and thoroughness of the women’s participation in treatment
regimens. A previous study set in one managed care plan found
that control for median household income based on census data
eliminated an apparent survival disparity between African
American and white women ( 29 ) . These data were not available
for the cohort examined in this study. The ways in which fi nan-
cial resources affect access to care and treatment in an insured
population are not well understood and require additional
research.
Health insurance and fi nancial resources are not the only fac-
tors that may affect access to care. Attempts to understand health
disparities among racial and ethnic groups have also highlighted
the role of patient – physician communication during medical
visits. Several studies have found that the communication pat-
terns of physicians differ between white and African American
Table 3. Initial treatment provided to African American and white women with breast cancer by age and SEER summary stage
African American Women White Women
All Age 20 – 49 Age 50+ All Age 20 – 49 Age 50+
% % % % % %
Diagnosed at local stage N = 1279 N = 407 N = 872 N = 12 923 N = 2287 N = 10 636
Surgery 94.0 93.1 94.4 97.9 97.0 98.1
Radiation 41.4 41.5 41.4 48.3 49.0 48.1
Chemotherapy 30.6 57.5 18.0 18.7 53.3 11.2
Hormone therapy 37.7 20.2 45.9 49.5 31.4 53.3
Diagnosed at regional stage N = 823 N = 330 N = 493 N = 5131 N = 1483 N = 3648
Surgery 92.6 92.1 92.9 97.3 96.9 97.5
Radiation 31.4 32.4 30.6 36.3 38.9 35.2
Chemotherapy 74.9 93.9 62.1 69.3 94.9 58.9
Hormone therapy 35.2 20.9 44.8 53.4 33.1 61.6
Diagnosed at distant stage N = 122 N = 38 N = 84 N = 630 N = 126 N = 504
Surgery 64.8 73.7 60.7 64.0 77.8 60.5
Radiation 22.1 34.2 16.7 28.1 31.8 27.2
Chemotherapy 63.9 86.8 53.6 54.4 88.1 46.0
Hormone therapy 32.0 7.9 42.9 50.2 35.7 53.8
Table 4. Characteristics of breast cancers diagnosed at local stage
African American Women White Women
All ( N = 1279) Age 20 – 49 ( N = 407) Age 50+ ( N = 872) All ( N = 12 923) Age 20–49 ( N = 2287) Age 50+ ( N = 10 636)
% % % % % %
Grade
Well differentiated 14.3 8.4 17.1 22.3 15.6 23.7
Moderately differentiated 32.3 27.5 34.5 37.0 34.5 37.5
Poorly differentiated 34.0 45.2 28.8 20.5 33.0 17.8
Not differentiated 2.4 2.2 2.4 1.5 2.2 1.4
Unknown grade 17.0 16.7 17.2 18.7 14.7 19.6
Estrogen receptor status
Positive 47.8 37.8 52.4 61.4 53.5 63.1
Negative 26.2 34.6 22.3 13.0 21.3 11.2
Not done 5.3 4.4 5.7 4.8 4.4 4.9
Unknown 20.7 23.1 19.6 20.8 20.8 20.8
Progesterone receptor status
Positive 24.2 18.2 27.1 25.9 24.2 26.2
Negative 15.8 21.4 13.2 9.4 11.6 9.0
Not done 12.0 11.1 12.5 7.0 6.0 7.2
Unknown 47.9 49.4 47.3 57.7 58.2 57.6
Tumor size
<2 centimeters 49.7 42.5 53.0 68.2 59.1 70.2
2 to <5 centimeters 28.5 32.4 26.6 23.0 29.8 21.5
≥5 centimeters 4.0 5.7 3.2 1.9 2.7 1.7
Unknown size 17.9 19.4 17.2 6.9 8.4 6.6
by guest on August 19, 2015http://jncimono.oxfordjournals.org/Downloaded from
94 Journal of the National Cancer Institute Monographs, No. 35, 2005
patients ( 56 ) and that African American patients are more likely
to perceive racism, report mistrust of the medical care system,
and be less satisfi ed with their care ( 57 ) . In focus groups with
cardiac patients, investigators found that African American
patients wanted to build a trusting relationship with physicians
before agreeing to invasive procedures and complained that this
trust was lacking ( 58 ) . Research is needed on communication
between physicians and African American cancer patients and its
effect on treatment and outcomes.
This study was based in six large, integrated health care deliv-
ery systems. One of its intentions was to identify potential inter-
ventions for reducing any racial disparities detected. No clear
directions emerged. There is a possibility that increasing mam-
mography screening among African American women would
shift more diagnoses to earlier stages. In parallel with white
women, survival rates among African American women were
directly associated with stage at diagnosis (e.g., 87.8% of those
diagnosed in AJCC stage I survived for at least 5 years, compared
to 83.6% diagnosed in stage IIA, 71.8% in stage IIB, 48.5% in
stage III, and 9.4% in stage IV). However, survival disparities in
comparison to white women persisted. A recent cost/benefi t anal-
ysis of interventions for African American women with breast
cancer concluded that improvements in treatment are a more
cost-effective approach than increased screening ( 59 ) . More
information is required about the details of treatment provided to
African American women and race-specifi c rates of mammo-
graphic screening in these managed care plans before a similar
analysis can be conducted for this setting of care.
Having health insurance and apparently equal access to
medical care did not eliminate disparities in breast cancer sur-
vival for this group of African American women. This may
have resulted from differences in treatment, more aggressive
pathology, less effective follow-up, or the interaction between
African American women and the health care system. If we are
to reduce the health disparities for African American cancer
patients in the U.S. health care system, we must focus more
research on all aspects of access to medical care with attention
to the details of how socioeconomic factors and patients’ interac-
tions with the health care system affect their care.
R EFERENCES
(1) Chevarley F, White E. Recent trends in breast cancer mortality among white
and black US women. Am J Public Health 1997 ; 87 : 775 – 81.
(2) Brawley OW. Disaggregating the effects of race and poverty on breast can-
cer outcomes. J Natl Cancer Inst 2002 ; 94 : 471 – 3.
(3) Chu KC, Tarone RE, Brawley OW. Breast cancer trends of black women
compared with white women. Arch Fam Med 1999 ; 8 : 521 – 8.
(4) Moorman P, Jones B, Millikan R, Hall I, Newman B. Race, anthropo-
metric factors, and stage at diagnosis of breast cancer. Am J Epidemiol
2001 ; 153 : 284 – 91.
(5) Young H, Ries LG, Pollack ES. Cancer patient survival among ethnic groups
in the United States. J Natl Cancer Inst 1984 ; 73 : 341 – 52.
(6) Eley JW, Hill HA, Chen VW, Austin DF, Wesley MN, Muss HB, et al. Racial
differences in survival from breast cancer: Results of the National Cancer In-
stitute black/white cancer survival study. J Natl Cancer Inst 1995 ; 87 : 1686 – 93.
(7) Joslyn SA, West MM. Racial difference in breast carcinoma survival. Cancer
2000 ; 88 : 114 – 23.
(8) Bradley CJ, Given CW, Roberts C. Race, socioeconomic status, and breast
cancer treatment and survival. J Natl Cancer Inst 2002 ; 94 : 90 – 6.
(9) Ayanian JZ, Kohler BA, Abe T, Epstein AM. The relation between health
insurance coverage and clinical outcomes among women with breast cancer.
N Engl J Med 1993 ; 329 : 326 – 31.
(10) McWhorter WP, Mayer WJ. Black/white differences in type of initial
breast cancer treatment and implications for survival. Am J Public Health
1987 ; 77 : 1515 – 7.
(11) Breen N, Wesley MN, Merrill RM, Johnson K. The relationship of socioeco-
nomic status and access to minimum expected therapy among female breast
cancer patients in the National Cancer Institute black/white cancer survival
study. Ethn Dis 1999 ; 9 : 111 – 25.
(12) Shavers VL, Brown ML. Racial and ethnic disparities in the receipt of can-
cer treatment. J Natl Cancer Inst 2002 ; 94 : 334 – 57.
(13) Shavers VL, Harlan LC, Stevens JL. Racial/ethnic variation in clinical pre-
sentation, treatment, and survival among breast cancer patients under age
35. Cancer 2003 ; 97 : 134 – 47.
(14) Mandelblatt JS, Kerner JF, Hadley J, Hwang Y, Eggert L, Johnson LE, et al.
Variations in breast carcinoma treatment in older Medicare benefi ciaries:
Is it black and white? Cancer 2002 ; 95 : 1401 – 14.
(15) Li CI, Malone KE, Daling JR. Differences in breast cancer stage, treat -
ment, and survival by race and ethnicity. Arch Intern Med 2003 ; 163 :
49 – 56.
(16) Reynolds P, Hurley S, Torres M, Jackson J, Boyd P, Chen VW. Use of cop-
ing strategies and breast cancer survival: results from the black/white cancer
survival study. Am J Epidemiol 2000 ; 152 : 940 – 9.
(17) Soler-Vila H, Kasl SV, Jones BA. Prognostic signifi cance of psychoso-
cial factors in African American and white breast cancer patients. Cancer
2003 ; 98 : 1299 – 308.
(18) Newman LA, Mason J, Cote D, Vin Y, Carolin K, Bouwman D, et al.
African American ethnicity, socioeconomic status, and breast cancer sur-
vival. Cancer 2002 ; 94 : 2844 – 54.
(19) Gordon NH. Socioeconomic factors and breast cancer in black and white
Americans. Cancer Metastasis Rev 2003 ; 22 : 55 – 65.
(20) Middleton LP, Chen V, Perkins GH, Pin V, Page D. Histopathology of breast
cancer among African American women. Cancer 2003 ; 97 : 253 – 7.
(21) Aziz H, Hussain F, Sohn C, Mediavillo R, Saitta A, Hussain A, et al. Early
onset of breast carcinoma in African American women with poor prognostic
factors. Am J Clin Oncol 1999 ; 22 : 436 – 40.
(22) Chu KC, Anderson WF, Fritz A, Ries LA, Brawley OW. Frequency distri-
butions of breast cancer characteristics classifi ed by estrogen receptor and
progesterone receptor status for eight racial/ethnic groups. Cancer 2001 ;
92 : 37 – 45.
(23) Furberg H, Millikan R, Dressler L, Newman B, Geradts J. Tumor charac-
teristics in African American and white women. Breast Cancer Res Treat
2001 ; 68 : 33 – 43.
(24) Henson DE, Chu KC, Levine PH. Histologic grade, stage, and survival in
breast carcinoma. Comparison of African American and Caucasian women.
Cancer 2003 ; 98 : 908 – 17.
(25) Cunningham JE, Butler WM. Racial disparities in female breast cancer in
South Carolina: clinical evidence for a biological basis. Breast Cancer Res
Treat 2004 ; 88 : 161 – 76.
(26) Jones BA, Kasl SV, Howe CL, Lachman M, Dubrow R, Curnen MM,
et al. African American/white differences in breast carcinoma: p53 altera-
tions and other tumor characteristics. Cancer 2004 ; 101 : 1293 – 1301.
(27) Wojcik BE, Spinks MK, Optenberg SA. Breast carcinoma survival analysis
for African American and white women in an equal-access health care sys-
tem. Cancer 1998 ; 82 : 1310 – 18.
(28) Jatoi I, Becher H, Leake CR. Widening disparity in survival between white
and African American patients with breast carcinoma treated in the U.S.
Department of Defense healthcare system. Cancer 2003 ; 98 : 894 – 9.
(29) Yood MU, Johnson CC, Blount A, Abrams J, Wolman E, McCarthy BD,
et al. Race and differences in breast cancer survival in a managed care popu-
lation. J Natl Cancer Inst 1999 ; 91 : 1487 – 91.
(30) Tejeda HA, Green SB, Trimble EL, Ford L, High JL, Ungerleider RS, et al.
Representation of African Americans, Hispanics and whites in National
Cancer Institute cancer treatment trials. J Natl Cancer Inst 1996 ; 88 : 812 – 6.
(31) Murthy VH, Krumbolz HM, Gross CP. Participation in cancer clinical trials:
race-, sex-, and age-based disparities. JAMA 2004 ; 291 : 2720 – 6.
(32) Gross CP, Filardo G, Mayne ST, Krumholz HM. The impact of socioeco-
nomic status and race on trial participation for older women with breast
cancer. Cancer 2005 ; 103 : 483 – 91 .
(33) Hornbrook MC, Hart G, Ellis JL, Bachman DJ, Ansell G, Greene SM, et al.
Building a virtual cancer research organization. J Natl Cancer Inst Monogr
2005 ; 35 : 12 – 25.
by guest on August 19, 2015http://jncimono.oxfordjournals.org/Downloaded from
Journal of the National Cancer Institute Monographs, No. 35, 2005 95
(34) Stewart SL, Swallen KC, Glaser SL, Horn-Ross PL, West DW. Comparison
of methods for classifying Hispanic ethnicity in a population-based cancer
registry. Am J Epidemiol 1999 ; 149 : 1063 – 71.
(35) Morabia A, Flandre P. Misclassifi cation bias related to defi nition of menopausal
status in case-control studies of breast cancer. Int J Epidemiol 1992 ; 21 : 222 – 7.
(36) Anderson WF, Chatterjee N, Ershler WB, Brawley OW. Estrogen receptor
breast cancer phenotypes in the Surveillance, Epidemiology, and End Re-
sults database. Breast Cancer Res Treat 2002 ; 76 : 27 – 36.
(37) Schoenfeld D. Partial residuals for the proportional hazards regression
model. Biometrika 1982 ; 69 : 239 – 41.
(38) Grambsch PM, Therneau TM. Proportional hazards tests and diagnostics
based on weighted residuals. Biometrika 1994 ; 81 : 515 – 26.
(39) Chung M, Chang HR, Bland I, Wanebo H. Younger women with breast can-
cer have a poorer prognosis than older women. Cancer 1996 ; 77 : 97 – 103.
(40) Bonnier P, Romain S, Charpin C, Lejeune C, Tubiana N, Martin PM, et al.
Age as a prognostic factor in breast cancer: relationship to pathologic and
biologic features. Int J Cancer 1995 ; 62 : 138 – 44.
(41) Yildirim E, Dalgic T, Berberoglu U. Prognostic signifi cance of young age in
breast cancer. J Surg Oncol 2000 ; 74 : 267 – 72.
(42) Gajdos C, Tartter PI, Bleiweiss IJ, Bodian C, Brower ST. Stage 0 to stage III
breast cancer in young women. J Am Coll Surg 2000 ; 190 : 523 – 9.
(43) Maggard MA, O’Connell JB, Lane KE, Liu JH, Etzioni DA, Ko CY. Do young
breast cancer patients have worse outcomes? J Surg Res 2003 ; 113 : 109 – 13.
(44) Newman LA, Alfonso AE. Age-related differences in breast cancer stage at
diagnosis between black and white patients in an urban community hospital.
Ann Surg Oncol 1997 ; 4 : 655 – 62.
(45) Elmore JG. Breast cancer tumor characteristics in black and white women.
Cancer 1998 ; 83 : 2509 – 15.
(46) Rutqvist LE. Validity of certifi ed causes of death in breast carcinoma pa-
tients. Acta Radiol Oncol 1985 ; 24 : 385 – 90.
(47) Hoel DG, Ron E, Carter R, Mabuchi K. Infl uence of death certifi cate errors
on cancer mortality trends. J Natl Cancer Inst 1993 ; 85 : 1063 – 68.
(48) Feuer EJ, Merrill RM, Hankey BF. Cancer surveillance series: interpreting
trends in prostate cancer – part II: cause of death misclassifi cation and the recent
rise and fall in prostate cancer mortality. J Natl Cancer Inst 1999 ; 91 : 1025 – 32.
(49) Chu KC, Lamar CA, Freeman HA. Racial disparities in breast carcinoma
survival rates. Cancer 2003 ; 97 : 2853 – 60.
(50) Athas WF, Adams-Cameron M, Hunt WC, Amir-Fazli A, Key CR. Travel
distance to radiation therapy and receipt of radiotherapy following breast-
conserving surgery. J Natl Cancer Inst 2000 ; 92 : 269 – 71.
(51) Lamont EB, Hayreh D, Pickett KE, Dignam JJ, List MA, Stenson KM, et al.
Is patient travel distance associated with survival on phase II clinical trials
in oncology? J Natl Cancer Inst 2003 ; 95 : 1370 – 5.
(52) Given BA, Given CW, Stommel M. Family and out-of-pocket costs for
women with breast cancer. Cancer Pract 1994 ; 2 : 187 – 93.
(53) Moore K. Out-of-pocket expenditures of outpatients receiving chemotherapy.
Oncol Nurs Forum 1998 ; 25 : 1615 – 22.
(54) Moore KA. Breast cancer patients’ out-of-pocket expenses. Cancer Nursing
1999 ; 22 : 389 – 96.
(55) Chirikos TN, Russell-Jacobs A, Jacobsen PB. Functional impairment
and the economic consequences of female breast cancer. Women Health
2002 ; 36 : 1 – 20.
(56) Johnson RL, Roter D, Powe NR, Cooper LA. Patient race/ethnicity
and quality of patient-physician communication during medical visits.
Am J Public Health 2004 ; 94 : 2084 – 90.
(57) LaVeist TA, Nickerson KJ, Bowie JV. Attitudes about racism, medical mis-
trust, and satisfaction with care among African American and white cardiac
patients. Med Care Res Rev 2000 ; 57 : 146 – 61.
(58) Collins TC, Clark JA, Petersen LA, Kressin NR. Racial differences in
how patients perceive physician communication regarding cardiac testing.
Med Care 2002 ; 40 : I -27 – 34.
(59) Mandelblatt JS, Schechter CB, Yabroff R, Lawrence W, Dignam J,
Muennig P, et al. Benefi ts and costs of interventions to improve breast can-
cer outcomes in African American women. J Clin Oncol 2004 ; 22 : 2554 – 66.
N OTES
This study was carried out under the auspices of the Cancer Research Network,
supported by a grant from the National Cancer Institute, CA79689. Dr. Buist’s
time was supported by CRTG-02 – 024 – 01-CCE, a grant from the American
Cancer Society.
The overall goal of the CRN is to increase the effectiveness of preventive,
curative and supportive interventions that span the natural history of major can-
cers among diverse populations and health systems, through a program of col-
laborative research. This overarching aim of the CRN, coupled with the expertise
of the investigative team, and geographically-dispersed population base, fosters
effi cient and effective research on variations in cancer prevention and treatment
policies and practices.
by guest on August 19, 2015http://jncimono.oxfordjournals.org/Downloaded from