African-American women have increased breast cancer mortality compared with white women. Diagnostic and treatment gaps may contribute to this disparity.
African-American women have increased breast cancer mortality compared with white women. Diagnostic and treatment gaps may contribute to this disparity.
In this retrospective, longitudinal cohort study, Southern US health plan claims data and linked medical charts were used to identify racial disparities in the diagnoses, treatment, and mortality of commercially insured women with newly diagnosed breast cancer. White women (n = 476) and African-American women (n = 99) with newly diagnosed breast cancer were identified by breast cancer claims codes (International Classification of Diseases, Ninth Revision, Clinical Modification codes 174, 233.0, 238.3, and 239.3) between January 2000 and December 2004. Race, diagnoses (breast cancer stage, estrogen/progesterone receptor [ER/PR]-positive status), treatment (breast-conserving surgery, antiestrogen therapy, and chemotherapy interruption or reduction), and all-cause mortality were assessed from medical charts. Multivariate regression analyses were adjusted for age, geography, and socioeconomic status to test the association of race with diagnoses/treatment.
White women were older (P < .001) and had higher rates of diagnosis at stage 0/I (55.2% vs 38.4%; P < .05) than African-American women. More white women had positive ER/PR status (75% vs 56% African-American; P = .001) and received antiestrogen therapy if they were positive (37.2% vs 27.3% African-American; P < .001). White women received slightly more breast-conserving surgery and chemotherapy dose modification than African-American women (P value nonsignificant). African-American women had a higher mortality rate (8.1%) than white women (3.6%; P = .06). In adjusted analyses, African-American women were diagnosed at later stages (odds ratio, 1.71; P = .02), and white women received more antiestrogen therapy (odds ratio, 2.1; P = .03).
Disparities in medical care among patients with newly diagnosed breast cancer were evident between African-American women and white women despite health plan insurance coverage. Interventions that address the gaps identified are needed. Cancer 2010. © 2010 American Cancer Society.
The importance of minimizing racial disparities in healthcare is evidenced by the number of broad-based initiatives in the United States. Such initiatives include efforts by the Agency for Healthcare Research and Quality,1 the US. Department of Health and Human Services Initiative on Racial and Ethnic Disparities,2 and the American Society of Clinical Oncology Advisory Group on Health Disparities.3 Estimates are that the breast cancer mortality rate is 33% higher in African-American women than in white women.4-7 Disparities that may increase the risk of mortality in African-American women include a later disease stage at diagnosis,8, 9 higher rates of negative estrogen receptor (ER) status,7, 10, 11 decreased chemotherapy dose intensity because of an interruption or reduction in dosage,12 and a higher prevalence of comorbid conditions13, 14 compared with white women.
Gaps in adequate preventive and primary treatment activities often are evident when comparing African-American women with white women who have newly diagnosed breast cancer.13, 15 African-American women may be less likely than white women to receive indicated adjuvant therapies, including radiation therapy, chemotherapy, and antiestrogen therapy.7, 12, 13, 15, 16 Although racial disparities in breast cancer diagnosis and treatment have been documented previously, the majority of previous studies did not specify the type of insurance coverage used8, 11 or were conducted mostly in Medicaid/Medicare populations,12-14 and both of those conditions ignore a segment of the population with racial differences but equivalent commercial insurance coverage.
In this retrospective cohort analysis, we identified gaps in medical care between African-American women and white women with newly diagnosed breast cancer who were covered by the same health plan in the southeastern United States. We hypothesized that race was associated with differences in stage of cancer at diagnosis, use of antiestrogen therapy, use of breast-conserving surgery, chemotherapy dose interruption or reduction, and rates of mortality before and after adjustment for age, socioeconomic status (SES), and the inclusion of measures, as appropriate for geography, body mass index (BMI), stage of cancer at diagnosis, ER/progesterone receptor (PR)-positive status, use of blood growth factor, and presence of anemia or neutropenia.
Patient data were obtained from an administrative medical claims database that was created using data from a large commercial health plan located in the southeastern United States. Claims data were used to identify linked medical charts and to measure age, use of pharmaceutical agents, and comorbid medical conditions present at the time of breast cancer diagnosis; all other measures presented in these analyses were extracted from medical chart data. A limited data set, as defined by the Health Insurance Portability and Accountability Act (HIPAA) of 1996, was used for these analyses. HealthCore, Inc. (Wilmington, Del) had all HIPAA-required business associate and data-use agreements in place before conducting the research. A HIPAA waiver of authorization was obtained from a central institutional review board allowing the use of Protected Health Information to obtain medical charts.
Medical claims between January 1, 2000 and December 31, 2004 (intake period) for women in the HealthCore Integrated Research Database were reviewed to identify women who had at least 2 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9) diagnosis codes (primary or secondary) for breast cancer (ICD-9 codes 174, 233.0x, 238.3x, and 239.3x). The date of the first medical claim for breast cancer was defined as the “index diagnosis date.” Patients were required to have continuous plan enrollment for 12 months before and 6 months after the index diagnosis date to allow exclusion based on prior cancer diagnosis or prior cancer treatment, and they had to complete follow-up of treatment patterns for at least 6 months after index diagnosis date. We targeted patients who had newly diagnosed breast cancer for this study and excluded women who had either a medical claim for any type of cancer within 180 days before the index diagnosis date or a medical or pharmacy claim for cancer treatment (ie, surgical, radiation, or chemotherapy) in the 90 days before the index diagnosis date. Patients were followed longitudinally from their index diagnosis date until the earlier date of their end of eligibility or August 31, 2005 (mean follow-up, 27.4 months; standard deviation, 16.1 months).
Once patients with newly diagnosed breast cancer were identified, the physician(s) who provided surgery or chemotherapeutic care was identified by the physician's tax identification codes from claims data. The physician with the highest amount of claim dates for treatment of each patient was contacted and recruited through the HealthCore Integrated Research Network. Chart abstraction was performed by nurses who were trained in standardized data-collection techniques. Data from the medical charts were extracted through completion of a prepared computerized data-extraction tool. We targeted high-minority areas of the southeastern state covered by the health plan. We used the patient and physician office zip codes that matched regions that had ≥50% minorities according to 2000 census tract data so that we could measure diagnostic, treatment pattern, and mortality differences between white women and African-American women. No greater than 8% of charts were obtained from any single practice to prevent skewing data toward practice patterns from a particular group.
Of the 3017 women who were identified from administrative claims data with newly diagnosed breast cancer during the study intake period as eligible for medical chart abstraction, we obtained medical charts from 766 patients. In total, 170 patients (22.2%) were missing data on race after chart abstraction. Of those patients who had race data (n = 596 women), 79.8% were white (n = 476), 16.6% were African-American (n = 99), 1.5% were Asian (n = 9), 0.5% were Latino (n = 3), and 1.5% were of “other” race (n = 9). Because of small sample sizes for the other racial cohorts, analytic comparisons were made only between white patients and African-American patients.
The information obtained from the medical charts included patient race, cancer stage at diagnosis, positive ER/PR status, cancer treatment received (ie, breast-conserving surgery, chemotherapy dose interruption or reduction, radiation therapy, antiestrogen therapy), and mortality. We also collected data on complications and/or treatment received because of illness or therapy and on notation of follow-up care in the medical chart. Race was not always available in the medical chart, as mentioned above. We summarized demographic characteristics for patients with other or undocumented race (text only) but otherwise excluded them from additional analyses. The exclusion of patients based on unknown race was based on several factors: These women may differ inherently from women with known race inasmuch as they may have more missing data overall, they may have been treated differently overall, or they may have other characteristics that confound our ability to assess their risk accurately. All of the multivariate models conducted in our analyses included 476 white women and 99 African-American women with newly diagnosed breast cancer.
The main outcomes for these analyses were 1) breast cancer stage at diagnosis, 2) use of antiestrogen therapy, 3) use of breast-conserving surgery or lumpectomy, 4) chemotherapy dose modification (ie, an interruption or reduction), and 5) rates of all-cause mortality. Breast cancer stage at diagnosis was defined as stages 0, I, II, III, and IV according to the 2003 National Comprehensive Cancer Network guidelines.17 All other outcomes were assessed based on medical chart review from the date of first chemotherapy, antiestrogen treatment, or surgery to the end of medical chart notation. Antiestrogen therapy was defined as the use of hormone therapy (tamoxifen, fulvestrant, toremifene) or aromatase inhibitors (anastrazole, exemestane, letrozole).
Age at index diagnosis date, medical comorbidities (hypertension, osteoarthritis, coronary artery disease, diabetes mellitus, renal disease), and concurrent medication use (contraceptives/hormone-replacement therapy and antihypertensive, antihyperlipidemic, diuretic, and corticosteroid medications) at baseline were collected using claims data for the chart-abstracted cohort. BMI (in kg/m2) was calculated using weight (kg) and height (m), which were collected from medical chart review. Patient SES was based on average household income using zip code estimates from 2000, which, with a Georgia-specific cost-of-living adjustment from the Bureau of Labor Statistics (available at: http://www.bls.gov accessed November 2008), were adjusted to 2006 figures and divided into quintiles. Rural-Urban Commuting Area (RUCA) codes were defined using patient's household zip codes and were used in multivariate modeling to adjust for geographic area (7% rural, 16% suburban, 77% urban).
We collected data on cancer-specific treatment received (surgery, radiation therapy, chemotherapy, and hormone therapy), adverse events, supportive care, and oncologist follow-up communication with primary care physicians (PCPs). Because we abstracted oncologist's medical charts, data were more specific for therapy provided by oncologists (eg, chemotherapy) than for therapy provided by radiologists. Chemotherapy dosing for the first 8 cycles was recorded as intended, and dosing delays and reductions that actually were received and documented were captured. Because we were unable to determine what the protocol-specified regimen was in most cases, we designated the initial dose as the “intended” dose. A chemotherapy dose delay was defined as a difference ≥7 days between intended and actual treatment time, and a dose reduction was defined as a dose difference ≥15% between the intended dose and the actual dose received. Chemotherapy dose reductions were compared by categorical age (ages <40 years, 40 to 49 years, and ≥50 years), race, stage at diagnosis, SES quintile, and BMI. We measured adverse events and supportive care that began during cancer treatment as follows: nausea/vomiting or antiemetic treatment, menopausal symptoms or hormone-replacement therapy, pain medication or management, hair loss, fever or infection, erythrocyte growth factor administration, depression or antidepressant medication, anxiety or anxiolytic medication, intravenous fluids, and referral for mental healthcare. The use of white blood cell growth factor was infrequent and was not included in these analyses. We also assessed notation of follow-up communication with the PCP.
All analyses in which race was the exposure variable were compared using only white patients and African-American patients. In addition, the receipt of a chemotherapy dose modification (delay or reduction) was used to compare categorical age, race, stage at diagnosis, and SES. Continuous measures (age, BMI) were compared using standard t tests. Analyses of categorical measures were conducted using Pearson chi-square tests. Multivariate regression modeling was conducted to measure the association of race with stage of cancer at diagnosis, use of antiestrogen therapy, breast-conserving surgery, and mortality rates only if race was associated with these outcomes in unadjusted bivariate analyses. Covariates that were included in every model were age and SES quintiles, which were treated as linear covariates (the comparison group was the fifth, or highest, quintile). Geography (RUCA; the comparison group was urban) was included in the regression model that analyzed stage of cancer at diagnosis. Stage of cancer at diagnosis (the comparison group was stage 0) was included as a covariate in modeling the use of antiestrogen therapy, the rate of chemotherapy dose modification, and rates of mortality. Use of erythrocyte growth factor, BMI, and the presence of anemia or neutropenia were included as additional covariates in modeling chemotherapy dose modification. Stage of cancer at diagnosis was modeled using ordinal logistic regression modeling. All other outcomes were binary and were modeled using binary logistic regression. Statistical analyses were performed using SAS, version 9.1 (SAS Institute Inc., Cary, NC).
This work was supported by a grant from Amgen Inc., Thousand Oaks, California. Funding was provided specifically for the collection of treatment data for patients newly diagnosed with breast cancer from claims data and medical chart review for women from the same US health plan. Collaboration with Amgen Inc. and the health plan was overseen by HealthCore, Inc. and was ongoing until project completion. Although preparation of this article was funded by Amgen Inc. with contribution from health plan representatives, the data reported here have not been modified in any way based on funding source or health plan collaboration.
African-American women were significantly younger than white women (P = .001) and had twice the rate of hypertension (P < .001) (Table 1). African-American women also were more obese (mean, BMI, 32.8 kg/m2) than white women (mean, BMI, 27.8 kg/m2; P < .01).
|Characteristic||No. of Patients (%)*|
|White Women||African-American Women|
|No. of patients||476 (62)||99 (13)|
|Age: Mean±SD, y||52.9 ± 11||48.9 ± 11†|
|BMI: Mean±SD, kg/m2||27.8 ± 7||32.8 ± 8.6‡|
|Baseline comorbid conditions|
|Hypertension||103 (21.6)||46 (46.5)†|
|Osteoarthritis||48 (10.1)||9 (9.1)|
|Cardiovascular disease||22 (4.6)||4 (4)|
|Diabetes mellitus||9 (1.9)||3 (3)|
|Renal disease||4 (0.8)||5 (5.1)‡|
|Baseline concurrent medications|
|Contraceptives/HRT||186 (39.1)||20 (20.2)†|
|Antihypertensives||132 (27.7)||40 (40.4)‡|
|Antihyperlipidemics||68 (14.3)||10 (10.1)|
|Diuretics||60 (12.6)||20 (20.2)|
|Corticosteroids||59 (12.4)||9 (9.1)|
Women with other or undocumented race were assessed for baseline characteristics only (data not shown). These women were 52 ± 9.9 years of age on average. Hypertension was their most prevalent comorbid condition (26.2%), and all other medical conditions ranged in prevalence from 1.6% (renal disease) to 5.8% (osteoarthritis). Rates of medication use were as follows: 36.6% received contraceptives/hormone replacement therapy, 26.2% received antihypertensive medications, 13.1% received antihyperlipidemic medications, 10.5% received diuretics, and 6.3% received corticosteroids.
White women (55.2%) were more likely than African-American women (38.4%) to be diagnosed with stage 0 or stage I disease, and approximately twice as many African-American women (6.1%) compared with white women (3.6%) were diagnosed with stage IV disease (P < 0.05) (Fig. 1).
Positive ER/PR status was documented in >70% of white women and African-American women. Fewer African-American women had positive ER/PR status than white women (56% vs 75%; P = .001). Among women who had positive ER/PR status, African-American women were less likely to receive aromatase inhibitors (P < .001) or tamoxifen (P<.001) (Table 2).
|Parameter||No. of Patients (%)|
|White Women, N=476||African-American Women, N=99|
|Chemotherapy||294 (61.8)||75 (75.8)*|
|Surgery||407 (85.5)||90 (90.9)|
|Breast-conserving surgery||332 (69.8)||66 (66.7)|
|Radiation therapy||302 (63.4)||67 (67.7)|
|Antiestrogen therapy||308 (64.7)||48 (48.5)‡|
|ER/PR measured||354 (74.4)||71 (71.7)|
|ER/PR-positive patients†||267 (75.4)||40 (56.3)‡|
|Treatment of ER/PR-positive patients§|
|Aromatase inhibitor use∥||94 (35.3)||9 (22.5)‡|
|Tamoxifen use||120 (45.1)||16 (40.0)‡|
|Adverse events and supportive care|
|Nausea/vomiting or antiemetic treatment||203 (42.6)||33 (33.3)|
|Menopausal symptoms or hormone-replacement therapy||175 (36.8)||18 (18.2)‡|
|Pain, opioids, or referral to a pain management provider||147 (30.9)||28 (28.3)|
|Hair loss||121 (25.4)||22 (22.2)|
|Fever or infection||97 (20.4)||16 (16.2)|
|Erythrocyte growth factor||153 (32.1)||56 (56.6)‡|
|Depression or antidepressant treatment||89 (18.7)||11 (11.1)|
|Anxiety or anxiolytic therapy||89 (18.7)||9 (9.1)*|
|Intravenous fluids||53 (11.1)||11 (11.1)|
|Referral to a mental health professional||1 (0.2)||0 (0)|
|Follow-up communication with primary care physician|
|Notation and summary of communication||226 (47.5)||45 (45.5)|
|Notation, unknown follow-up communication||27 (5.7)||10 (10.1)|
|No notation or follow-up communication||115 (24.2)||21 (21.2)|
|Unable to determine||108 (22.7)||23 (23.2)|
In adjusted analyses, the odds that African-American women would be diagnosed with a later stage of cancer were nearly twice those of white women (odds ratio [OR], 1.71; 95% confidence interval [CI], 1.09-2.67; P = .02) after adjusting for age, geography according to RUCA code, and SES (Table 3).
|Variable||Stage at Diagnosis||Antiestrogen Therapy||Chemotherapy Dose Modification|
|OR||95% CI||P||OR||95% CI||P||OR||95% CI||P|
A higher percentage of African-American women received chemotherapy compared with white women (P = .002). The use of antiestrogen therapy was more frequent in white women than in African-American women (P < .001). Surgery (P = .15) and radiation therapy (P = .42) were slightly more frequent and breast-conserving surgery was slightly less frequent (P >.50) among African-American women compared with white women (see Table 2).
With the exception of erythrocyte growth factor, white women were more likely to receive supportive care or to have adverse events from illness or therapy. Although depression or antidepressant treatment (17.4%) and anxiety or anxiolytic therapy (17%) were prevalent in the total cohort, only 1 woman was referred to a mental health professional. Slightly less than 50% of all women of both races had notation and documentation of follow-up with their PCP in their medical chart (see Table 2).
Of the 458 patients who received chemotherapy (294 white women, 75 African-American women, 89 women of other/undocumented race), 19% had a medical chart notation of dose interruption or reduction, ie, modification. An analysis of only African-American women or white women indicated that older women (P = .03) and white women (P = .07) had a higher percentage of chemotherapy dose delay or dose reduction compared with younger or African-American women, respectively (see Table 4).
|Parameter||No. of Patients (%)||Total No.*|
|Dose delay or dose reduction†||89 (19.4)||458|
|African American||9 (12)||75|
|Stage at diagnosis|
|SES quintile: Median [range]|
|First $33,800 [$19,367-$38,182]||18 (19.4)||93|
|Second: $45,383 [$38,290-$49,405]||10 (10.9)||92|
|Third: $55,360 [$49,459-$59,817]||16 (16.8)||95|
|Fourth: $68,132 [$60,188-$74,049]||25 (29.1)||86|
|Fifth: $84,282 [$74,394-$122,701]||20 (22.2)||90|
The mortality rate from any cause was higher in African-American women (8%) compared with white women (4%; P = .06). Breast cancer progression was the cause of 56% of all documented deaths and caused a higher percentage of deaths in white women (59%) than in African-American women (50%). There was no correlation between survival status and SES (Table 5).
|Variable||No. of Patients (%)|
|All Causes||Breast Cancer||Other||Unknown|
|White||422 (88.7)||17 (3.6)*||10 (58.8)||3 (17.6)||4 (23.5)||37 (7.8)|
|African-American||82 (82.8)||8 (8.1)*||4 (50)||0 (0)||4 (50)||9 (9.1)|
|SES quintile: Median [range]|
|First: $33,800 [$19,367-$38,182], n = 148||132 (89.2)||4 (2.7)||2 (50)||0 (0)||2 (50)||12 (8.1)|
|Second: $45,383 [$38,290-$49,405], n = 151||136 (90.1)||7 (4.6)||4 (57.1)||2 (28.6)||1 (14.3)||8 (5.3)|
|Third: $55,360 [$49,459-$59,817], n = 152||114 (75)||9 (5.9)||5 (55.6)||1 (11.1)||3 (33.3)||29 (19.1)|
|Fourth: $68,132 [$60,188-$74,049], n = 148||117 (79.1)||5 (3.4)||4 (80)||0 (0)||1 (20)||26 (17.6)|
|Fifth: $84,282 [$74,394-$122,701], n = 149||127 (85.2)||6 (4.7)||4 (66.7)||0 (0)||2 (33.3)||16 (10.7)|
White women were twice as likely to receive antiestrogen therapy compared with African-American women (OR, 2.09; 95% CI, 1.07-4.06; P = .03) after adjusting for age, stage, SES, and positive ER/PR status (Table 3).
For every 10-year increment in age, the odds of chemotherapy dose delay or dose reduction increased by 48%. No association was observed between chemotherapy modification and either SES quintiles or BMI; however, the presence of neutropenia (OR, 1.96; P = .04) was associated with dosing modifications.
In the current analyses, we hypothesized that race was associated with differences in stage of cancer at diagnosis, use of antiestrogen therapy, use of breast-conserving surgery, chemotherapy dose interruption or reduction, and rates of mortality before and after adjusting for age, SES, BMI, geography, and other relevant diagnostic and treatment factors. In adjusted analyses, we observed that race was a factor in stage of cancer at diagnosis, antiestrogen therapy, chemotherapy dose modification, and rates of mortality. After adjusting for covariates, stage at diagnosis and antiestrogen therapy still were associated significantly with race. Our findings are distinguishable from previous studies in this area, because we compared African-American women and white women who had similar commercial health insurance. We noted that African-American women were diagnosed at younger ages and with later stage disease than white women. Although African-American women were younger, their all-cause mortality was double that of white women. In this study, we also uncovered several apparent gaps in documentation, including recording measurement of ER/PR status, staging from pathology reports, dosing and intent of chemotherapy, and communication between the patient's oncology treatment provider and her PCP.
Comorbid medical conditions have been identified previously as a risk factor for increased complications in African-American women in a cohort with 47% commercial health insurance, 28% Medicare coverage only, and 18% Medicaid insurance.13 In our sample, African-American women had higher rates of hypertension and renal disease compared with white women. We also observed that more white women had positive ER/PR status in our sample compared with African-American women, a finding that supports the current literature.7, 10, 11 In addition, white women were more likely to receive antiestrogen therapy after adjusting for age, SES, and geography.
We observed that African-American women were younger and were diagnosed with a later disease stage than white women, findings that also are supported by previous studies.8, 9 Moorman et al8 suggested that obesity and SES may contribute to differences in disease stage at diagnosis. BMI was higher in African-American women versus white women in our study. We observed no differences in disease stage at diagnosis, treatment, or mortality based on SES. Our mortality data are supported by previous findings,4-7 but the small sample size precluded investigation of the impact covariates may have had on the association between mortality and race.
Our findings suggest that the identification of racially based disparities in medical care among patients who have newly diagnosed breast cancer can catalyze and refine the development of culturally sensitive interventions that could improve outcomes in African-American women. Health plan reminders, call centers, and community outreach programs are potential avenues for patient education. In addition, health plans are well positioned to proactively collaborate with oncologists to improve communication regarding early notification of diagnoses and support for patients in initial treatment.
Documentation gaps in the medical charts that we reviewed represented opportunities to improve quality of care, potentially through the use of health plan-provided flow sheets. These tools could improve documentation of race, which is an important factor in these analyses and was missing in >20% of all charts that we reviewed. Diagnostic testing rates and medical chart documentation of test results must improve for healthcare to improve, particularly for African-American women. We demonstrated that <50% of the charts in this study had any documentation of communication between oncologists and PCPs, which could be improved with secure electronic communication systems or integrated electronic health records. Future analyses in other health plans should include more focus on chemotherapy dose intensity, especially in relation to other patient factors, such as the use of supportive therapy. This will require adequate documentation of dosing and intent of treatment (curative vs palliative). Such analyses could investigate the potentially greater opportunity for white women to have a curative goal because of a lower disease stage at diagnosis and to evaluate any impact on the use of supportive care.
Our data indicate that most patients who were depressed or anxious did not receive mental health support. The development of screening programs to identify and refer breast cancer patients to mental health providers is underused and necessary. The health plan that was involved in this study developed a resource list to address mental health and support groups for patients that was well received by both patients and providers.18 Family involvement is another avenue by which health plans may help to optimize care and promote compliance.
Our analyses had a few limitations. Claims data generally do not have complete capture because of coordination of benefits, do not contain services received through the public health system, and may contain coding inaccuracies and errors. These limitations are inherent in the use of retrospective claims data, and our stringent definition of “newly diagnosed” precluded any active cancer diagnosis or treatment ongoing on or around the time frame of interest, which strengthened our analyses. Our chart selection methodology focused on high-volume oncology offices, because chart abstraction studies are limited by budget and time factors. We collected data from large practices for efficiency and to facilitate the analysis of chemotherapy dose modifications. However, we could not determine whether the initial dose was the protocol-specified dose and may have missed dose reductions that were included in the “intended” dose. We did not obtain greater than 8% of charts from any single practice so that the data would not be skewed by a particular provider's group practice patterns. In addition, the acquisition of mortality data relied on chart notation and was limited by physician's continued follow-up of patients. There also were strengths in our analyses. We concurrently measured risk factors for breast cancer, diagnostic patterns, treatment patterns, adverse events, and supportive care received. We also measured patient outcomes and the development of adverse events, which we compared across racial differences. The most notable and compelling strength of these data were that all women had equivalent health plan coverage and access to care.
Among commercially insured patients who were newly diagnosed with breast cancer, we observed that disparities in care existed between African-American women and white women. We determined that opportunities to address these disparities include facilitating diagnosis at earlier stages, improving ER/PR testing rates and documentation of testing/results, increasing the percentage of ER/PR-positive patients who receive antiestrogen therapy, and aggressively managing comorbidities. Health plan-initiated interventions are needed for patient education, support of adequate care, and continued communication regarding breast cancer diagnosis and treatment.18 Currently, we are evaluating such efforts in the health plan that was used for this study. We encourage health plans and healthcare providers to offer culturally sensitive support and education to women who are diagnosed with breast cancer and to encourage active participation in their treatment decisions.
Supported by a grant from Amgen Inc. (Thousand Oaks, Calif).