Does having insurance affect differences in clinical presentation between Hispanic and non-Hispanic white women with breast cancer?
Hispanic women with breast cancer present differently than do non-Hispanic white (NHW) women. Lack of access to care has been offered as an explanation for these differences. In this study breast cancer presentation was examined in Hispanic women in a comprehensive, equal-access health care system.
Hispanic and NHW breast cancer cases registered between 1995 and 2004 in the Kaiser Permanente of Colorado Tumor Registry were compared by age at diagnosis, stage, tumor grade, size, and receptor status. Multivariate logistic regression was performed to generate age-adjusted odds ratios by ethnicity and each tumor characteristic.
A total of 139 Hispanic women and 2118 NHW women with breast cancer were identified. Hispanic women had a mean average age at diagnosis of 56 years compared with 61 years for NHW women (P < .0001). Use of mammographic screening services in the prior 2 years was similar by ethnicity. Relative to NHW women, Hispanic women had age-adjusted odds ratios of 2.70 (95% confidence interval [CI]: 1.26–5.77) for having stage IV disease, 2.25 (95% CI: 1.39–3.67) for having poorly differentiated tumors, 2.16 (95% CI: 1.26–3.69) for having a tumor greater than 5 cm, and 1.88 (95% CI: 1.24–2.81) for having estrogen receptor-negative tumors.
Despite equal access to health care services, differences persist in the size, stage, and grade of breast cancer for Hispanic women compared with NHW women. The results of the study suggest a biologic/genetic basis for these differences. Cancer 2007. © 2007 American Cancer Society.
The previous literature has shown that Hispanic women diagnosed with breast cancer present differently than non-Hispanic white (NHW) women. These differences include earlier mean age at diagnosis,1–9 later stage of disease at diagnosis,1–4, 10–17 and the presence of tumor characteristics associated with poor prognosis such as large tumor size,2, 3, 9, 13, 14 high tumor grade,3, 13–15 and tumors less likely to have estrogen or progesterone receptors.1–3, 13, 14, 18 Explanations for these differences have included biologic/genetic factors, psychosocial/cultural factors (eg, education level, poverty level, marital status, cultural beliefs), and structural factors (eg, geography, neighborhood composition, health insurance status, and access to health care services). A few studies have looked at the relation between insurance status and breast cancer outcomes, such as stage at diagnosis and survival,7, 16, 19, 20 but few have examined whether differences persist between Hispanic women and non-Hispanic white women in health care systems with equal access to care. Zaloznik21 examined breast cancer stage at diagnosis in NHW and Hispanic women in the Department of Defense health care system, where all women have the same ability to access health care. They found that even within this single system Hispanic females tended to present with more advanced stage disease than did Caucasians, at younger ages, and with larger tumors. The conclusion reached was that “despite the ease of access to health care and free healthcare, there still appears to be a tendency for ethnic differences in stage at diagnosis of breast cancer between Hispanic and Caucasian women.”21 In 2003, Wojcik et al.15 studied a group of women diagnosed and treated at Brooke Army Medical Center in San Antonio, Texas. Over a 10-year period from 1987 to 1997, there were 719 white women and 67 Hispanic women with breast cancer. Despite equal proportions of women in these 2 groups reporting annual mammographic screening, they also observed differences in age at diagnosis, stage, and grade.15 The purpose of this study is to determine if these differences in age, stage, and tumor characteristics persist for Hispanic women despite equal access to screening and care in a nonprofit managed care organization.
MATERIALS AND METHODS
This was a population-based cross-sectional study of Hispanic white women and NHW women enrolled in Kaiser Permanente Colorado (KPCO) who had a diagnosis of primary breast cancer between the years 1995–2004. KPCO is a nonprofit, closed-panel Managed Care Organization serving nearly 410,000 members in the Denver Metropolitan area. Women were excluded from this study if they were under age 20 or had a previous diagnosis of cancer other than nonmelanoma skin cancer.
KPCO maintains a centralized Tumor Registry of all cancer patients diagnosed since 1987. New cases are registered using specific cancer-related ICD-9 codes from 3 sources: pathology, external claims, and internal appointments. In the KPCO data systems, patient characteristics include ethnicity, age at diagnosis, socioeconomic status (SES), recent access to the health care system, and history of recent screening. Hispanics are largely Mexican Americans in this region. SES within this analysis is defined by percent living in poverty by block group census based on the 2000 US census. This variable has 3 levels, those living in block groups with less than 10% living in poverty (not poor), those in block groups with 10% to 20% living in poverty (moderately poor), and those in block groups with greater than 20% living in poverty (poor). From existing electronic data sources at KPCO we captured information on access and prior screening. Recent access to health care was defined as a visit to a primary care provider (family practice, internal medicine, or obstetrics/gynecology) in a period of less than 2 years or greater than 2 years from the time of diagnosis. Recent mammographic screening was defined as having had mammography within 2 years before the date of diagnosis and not having had a mammogram in this period (not including the mammogram that immediately preceded the breast cancer diagnosis). Data on access and mammography were only available in the electronic database from 1998 forward.
Tumor characteristics included AJCC stage, tumor grade, tumor size, estrogen receptor (ER) status, progesterone status (PR) status, and histology. The tumor registry codes tumor grade as well differentiated, moderately differentiated, poorly differentiated, undifferentiated and unknown. Tumor size was categorized as less than 2.0 cm, 2.0 to less than 5.0 cm, 5.0 cm or larger, and unknown. Her2 status of the tumors was not registered in the database.
Descriptive statistics including frequencies and chi-square results were generated for each variable with bivariate analysis assessing differences by ethnicity. Subgroup analysis was done on 2 age groups: those 20 to 49 years of age and those 50 years of age or older (age 50 years is picked as a proxy for menopause). We also performed a subgroup analysis on members by enrollment status.
Logistic regression was performed to evaluate the association between ethnicity and the various tumor characteristics and multivariate modeling was used to assess confounding. We conducted analyses both on the subgroup with 3 or more years of KPCO membership and on the entire population. We retained age, SES, and length of membership in the final model. Interaction terms between race and age and race and SES were assessed, but none were significant. Analysis was conducted using SAS v. 9.1 (Cary, NC). The protocol for this study was reviewed and approved by both the Kaiser Permanente Institution Review Board (IRB) and the Colorado Multiple IRB.
There were 139 Hispanic women and 2118 NHW women diagnosed with breast cancer in the KPCO population in the study period (Table 1). This ratio (1:15) is slightly lower than that in Colorado in this period (1:12),22 likely reflecting the differences in health insurance by ethnicity. Mean age at diagnosis for Hispanic women was 56 years and for NHW women was 61 years (P < .001). Diagnosis under age 50 years occurred in 31% of Hispanics compared with 20% of NHWs (P = .002). Significant differences in SES were also found, with greater numbers of Hispanic (H) women living in poverty (28% H vs 18% NHW living in ‘moderately poor’ neighborhoods and 13% H vs 5.5% NHW living in ‘poor’ neighborhoods, P < .001). To determine whether the 2 groups of women were similar in terms of access to the system and history of recent mammogram screening before the current breast cancer diagnosis, we looked at health services usage only for those members who had been members for 3 or more years. (As data on access to care and mammography have only been available electronically since 1997, we could not capture these data for those diagnosed before mid-1999.) This subgroup analysis included 53% of Hispanics and 63% of NHWs in the overall study. In this subgroup there were no statistically significant differences between Hispanic members and NHW members in recent visit to a primary care provider (84% H, 85% NHWs, P = .11) or in receipt of mammography within 2 years before their diagnosis with breast cancer (70.3% H, 76.1% NHWs, P = .52) (Table 1).
Table 1. Distribution of Selected Patient Characteristics Among Hispanic and Non-Hispanic White Women Diagnosed With Breast Cancer, 1995–2004, Kaiser Permanente Colorado
|Age at diagnosis|
| 20–49 y||43 (30.9)||418 (19.7)|| |
| 50+y||96 (69.1)||1700 (80.3)||.002|
|Mean age at diagnosis. y||56||61||<.0001|
| <10% living in poverty||80 (57.6)||1604 (75.7)|| |
| 10–20% living in poverty||39 (28.1)||389 (18.4)|| |
| >20% living in poverty||18 (13.0)||117 (5.5)|| |
| Unknown||2 (1.4)||8 (0.4)||<.001|
|Membership ≥3 y|
| Yes||74 (53.2)||1334 (63.0)|| |
| No||65 (46.8)||784 (37.0||.022|
|Recent access of system*||(n = 74)||(n = 1334)|| |
| ≤2 y||62 (83.8)||1130 (84.7)|| |
| Greater than 2 y||2 (2.7)||96 (7.2)|| |
| Unknown||10 (13.5)||108 (8.1)||.106|
|Screening mammogram (apart from that tied to the breast cancer diagnosis)*|
| ≤2 y||52 (70.3)||1015 (76.1)|| |
| Greater than 2 y||1 (1.4)||14 (1.1)|| |
| Unknown||21 (28.4)||305 (22.9)||.524|
Statistically significant differences were found between the 2 groups in AJCC stage at diagnosis, grade of tumor at diagnosis, tumor size at diagnosis, as well as in percent of tumors found to be ER/PR negative (all P < .02) (Table 2). Compared with NHW women, the crude odds ratio (OR) for having a stage IV tumor was 2.63 (95% confidence interval [CI]: 1.25–5.53) for Hispanic women (Table 3). Adjustment for SES, age, and length of membership did not greatly alter the odds ratio (adjusted OR = 2.70; 95% CI: 1.26–5.77). Younger Hispanic women were less likely than NHW women to have early-stage disease and in later years were significantly more likely to have later-stage disease (OR = 3.48; 95% CI: 1.55–7.81). Hispanic women had a higher likelihood of having a poorly differentiated tumor (OR = 2.25; 95% CI: 1.39–3.67). Hispanic women were more likely to be diagnosed with a tumor of 5 cm or larger (OR = 2.16; 95% CI: 1.23–3.69) and of having a tumor that was ER-negative (OR = 1.88; 95% CI: 1.24–2.81). Adjustments for SES, age, and length of membership did not substantially change these ORs.
Table 2. Distribution of Selected Tumor Characteristics Among Hispanic and Non-Hispanic White Women Diagnosed With Breast Cancer, 1995–2004, Kaiser Permanente Colorado
| Stage I||79 (56.8)||1200 (56.7)|| ||23 (53.5)||165 (39.5)|| ||56 (58.3)||1035 (61.0)|| |
| Stage II||37 (26.7)||690 (32.6)|| ||13 (30.2)||205 (49.0)|| ||24 (25.0)||485 (28.5)|| |
| Stage III||13 (9.4)||135 (6.4)|| ||5 (11.6)||38 (9.1)|| ||8 (8.3)||97 (5.7)|| |
| Stage IV||9 (6.5)||52 (2.5)|| ||1 (2.3)||8 (1.9)|| ||8 (8.3)||44 (2.6)|| |
| Unknown||1 (0.7)||41 (2.0)||.02||1 (2.3)||2 (0.5)||.13||0 (0.0)||39 (2.3)||.007|
| Well-differentiated||29 (20.9)||655 (30.9)|| ||4 (9.3)||97 (23.2)|| ||25 (26.0)||558 (32.8)|| |
| Moderately differentiated||49 (35.3)||827 (39.1)|| ||13 (30.2)||138 (33.0)|| ||36 (37.5)||689 (40.5)|| |
| Poorly differentiated||52 (37.4)||445 (21.0)|| ||22 (51.2)||148 (35.4)|| ||30 (31.3)||297 (17.5)|| |
| Undifferentiated||2 (1.4)||14 (0.7)|| ||1 (2.3)||5 (1.2)|| ||1 (1.0)||9 (0.5)|| |
| Unknown||7 (5.0)||177 (8.3)||<.001||3 (7.0)||30 (7.2)||.16||4 (4.2)||147 (8.6)||.009|
| <2 cm||83 (60.0)||1331 (62.8)|| ||21 (48.8)||211 (50.5)|| ||62 (64.6)||1120 (65.9)|| |
| 2 to <5 cm||33 (23.8)||601 (28.3)|| ||14 (32.6)||167 (42.0)|| ||19 (19.8)||434 (25.5)|| |
| 5+ cm||18 (13.0)||134 (6.5)|| ||7 (16.2)||35 (8.4)||—||11 (11.5)||99 (5.8)|| |
| Unknown||5 (3.6)||52 (2.5)||.02||1 (2.3)||6 (1.4)||.31||4 (4.2)||47 (2.8)||.09|
| Positive||89 (64.0)||1567 (74.0)|| ||26 (60.5)||297 (71.0)|| ||63 (65.6)||1270 (74.8)|| |
| Negative||36 (26.0)||338 (16.0)|| ||16 (37.2)||96 (23.0)|| ||20 (20.8)||242 (14.2)|| |
| Not done||10 (7.2)||136 (6.4)|| ||1 (2.3)||15 (3.6)|| ||9 (9.4)||121 (7.2)|| |
| Unknown||4 (2.9)||77 (3.6)||.02||0 (0.0)||10 (2.4)||.17||4 (4.2)||67 (3.9)||.22|
| Positive||79 (56.7)||1344 (63.5)|| ||26 (53.5)||269 (64.4)|| ||56 (58.2)||1075 (63.2)|| |
| Negative||44 (31.7)||554 (26.2)|| ||19 (44.2)||124 (29.7)|| ||25 (26.0)||430 (25.3)|| |
| Not done||10 (7.2)||138 (6.5)|| ||1 (2.3)||15 (3.6)|| ||9 (9.4)||125 (.2)|| |
| Unknown||6 (4.3)||82 (3.9)||.47||0 (0.0)||10 (2.4)||.21||6 (6.3)||72 (4.2)||.62|
Table 3. Risk of Selected Tumor Characteristics for Hispanic Compared With Non-Hispanic White Women Diagnosed With Breast Cancer, 1995–2004, Kaiser Permanente Colorado*
| Stage I||1.0||1.0||1.0||1.0||1.0||1.0|
| Stage II||0.81 (0.55–1.22)||0.72 (0.48–1.10)||0.48 (0.23–0.98)||0.32 (0.15–0.70)||0.93 (0.57–1.53)||0.91 (0.55–1.49)|
| Stage III||1.46 (0.79–2.7)||1.27 (0.67–2.39)||0.99 (0.35–2.77)||0.64 (0.21–1.90)||1.55 (0.72–3.34)||1.58 (0.73–3.44)|
| Stage IV||2.63 (1.25–5.53)||2.70 (1.26–5.77)||1.07 (0.13–9.13)||0.64 (0.07–5.76)||3.41 (1.53–7.54)||3.48 (1.55–7.81)|
| Moderately differentiated||1.34 (0.84–2.14)||1.26 (0.78–2.03)||2.28 (0.72–7.22)||1.96 (0.61–6.36)||1.17 (0.69–1.97)||1.12 (0.66–1.90)|
| Poorly differentiated||2.64 (1.65–4.22)||2.25 (1.39–3.67)||3.60 (1.21–10.7)||2.83 (0.90–8.93)||2.25 (1.30–3.90)||2.17 (1.25–3.78)|
| Undifferentiated||3.23 (0.70–14.87)||2.40 (0.46–12.4)||4.85 (0.45–51.8)||6.7 (0.41–118.3)||2.48 (0.30–20.3)||2.16 (0.26–18.3)|
| <2 cm||1.0||1.0||1.0||1.0||1.0||1.0|
| 2 to <5 cm||0.88 (0.58–1.33)||0.81 (0.53–1.23)||0.84 (0.42–1.71)||0.74 (0.35–1.56)||0.81 (0.48–1.37)||0.80 (0.47–1.46)|
| 5+ cm||2.16 (1.26–3.69)||1.96 (1.13–3.41)||2.01 (0.8–5.08)||1.52 (0.55–4.17)||2.04 (1.04–4.00)||1.95 (0.99–1.49)|
| Negative||1.88 (1.24–2.81)||1.74 (1.16–2.67)||1.9 (0.98–3.70)||1.60 (0.78–3.26)||1.67 (0.99–2.81)||1.61 (0.95–2.73)|
| Negative||1.35 (0.92–1.98)||1.33 (0.90–1.96)||1.80 (0.94–3.41)||1.53 (0.78–3.03)||1.12 (0.69–1.81)||1.12 (0.69–1.83)|
The results of this study confirm those of many previous studies that breast cancer presents differently in Hispanic women: at a younger age, at a later stage of disease, with higher-grade tumors, larger tumors, and with more tumors that are ER/PR negative. Age, SES, and length of membership in the health plan did not confound these associations. Most important, our study shows that these differences were apparent even among a group of Hispanic women with equal access to care and similar health care utilization.
Explanations for these differences have included biologic/genetic, socioeconomic/cultural, and structural factors. Some authors have suggested that the finding of earlier mean age at diagnosis in Hispanic women is simply explained by the age distribution in this population (greater numbers of younger women). Others suggest that biologic/genetic differences have a role. Authors of a study done in Mexico that included 29,075 cases of breast cancer stated that their findings of earlier mean age at diagnosis might be explained by genetic factors.2 Ethnic differences in the effects of obesity and reproductive factors lend support to an explanation involving biologic factors. For example, Baumgartner et al.,23 in a study of 150 Hispanic and 466 NHW women in New Mexico, showed that among Hispanics the highest quartile for body mass index (BMI) and waist circumference was significantly associated with reduced tumor size. For NHW women, obesity-related measures were associated with increased tumor size. In a study that evaluated reproductive risk factors and breast cancer, higher parity was associated with reduced risk of breast cancer for NHW, but not for Hispanics (P < .008).24 Additionally, a study done by the Centers for Disease Control (CDC) in Atlanta found that early age at menarche was associated with a decreased risk of breast cancer in Hispanic women.25 These differences in hormonally related effects may in part explain why Hispanic women with breast cancer present differently.
Explanations for why Hispanic women present with later stage at diagnosis are many. Some suggest that cultural beliefs that Hispanic women have about cancer, comfort with exposing their bodies to examination, and religious beliefs that are fatalistic in nature may be responsible for delays in seeking care.26–29 Others have suggested that socioeconomic factors have a greater effect. For example, being unmarried, having lower education levels/income, and living in neighborhoods with more poverty have been shown to be associated with later stage of disease.6, 10, 13, 14, 21, 30–35 This may be due to the effects of poor social support and lifestyle issues that create higher risk or due to the association between poverty and insurance status and access to care, both of which might contribute to delays in seeking care. In a study by Miller et al.13 statistical adjustment for sociodemographic factors accounted for an 80% reduction in ORs for advanced stage in Hispanic women. O'Malley and Li14 looked at the relation between SES and survival in whites, blacks, and Hispanics, finding significant associations between several socioeconomic variables and stage of disease. Several other studies in African American women have shown similar decreases in ORs for late-stage disease after adjustment for sociodemographic variables.15, 31, 32 In our study, differences in stage persisted after adjustment for SES, and, importantly, differences were not related to prior preventive health care utilization, including previous use of screening mammography.
Access to care is critical to being able to get mammography screening and to ensuring accurate and timely diagnosis of disease, such as breast cancer. Some studies have looked at whether the findings of later-stage disease and larger tumor size at diagnosis in Hispanic women is due to differential receipt of mammography. Jacobellis et al.36 looked at mammography screening and stage of disease in black, Hispanic, and white women in Colorado. In their study, correcting for screening did not completely reduce stage differentials among black and Hispanic women. Wojcik et al.15 looked at black, white, and Hispanic women in a comprehensive health care system at the Brooke Army Medical Center in San Antonio, Texas. They found that although receipt of mammography was similar, stage of disease differentials persisted for Hispanics compared with whites (24% of Hispanics diagnosed with stage III/IV disease vs 17% of whites, P = .03). In yet another study, the authors found that a history of mammography accounted for less than 10% of the racial differences in stage of breast cancer at diagnosis.37
Other studies have looked at the effect of insurance status on stage of disease at diagnosis. These have shown differentials between the uninsured, those with Medicaid, and those with private insurance.19–21 Lee-Feldstein et al.19 showed that uninsured or publicly insured patients with breast cancer were less likely to have early-stage disease at diagnosis. Roetzheim et al.16 found that, despite type of insurance, there was a separate association between stage of disease and Hispanic ethnicity. However, few studies have been done in populations of women with the same access to care. In a study done in the Department of Defense health care system, a comprehensive, equal-access system, Hispanic women tended to present with more advanced stage of disease, although the difference was not statistically significant.21
In this study we again examined the differences between Hispanic and NHW women in an equal-access health care system. We found that Hispanic women presented at earlier mean age, with later stage, larger size tumors, tumors with poorer differentiation, and less likely to have ERs. Although having insurance increases the likelihood of receiving health care, we looked further at 2 other measures of accessing care, recent mammography and recent visit to a primary care provider (PCP). In order to do this, we restricted our study population to those who had been members for 3 or more years. Hispanic women did not differ significantly from NHW women in terms of having received recent mammography or in having recently seen a PCP. When we conducted our analysis of tumor characteristics in this subgroup, our findings were similar to those of the full group in terms of stage, grade, tumor size, and ER/PR status (data not shown).
This study had several limitations. Information on other risk factors that have been associated with breast cancer outcomes, such as BMI, reproductive history, marital status family history of breast cancer, or mammographic density was not collected; and the determination of ethnicity via medical record review uses surname recognition only, which is historically subject to some errors. However, any resulting ethnic misclassification would cause a bias toward the null, so our findings would be conservative. Other tumor measures such as Her2 status or markers of proliferation or inflammation would have been helpful to assess. Lastly, group level socioeconomic measures may not be the best measure of individual status. However, some investigators support the increased use of neighborhood-based measures because they may capture aspects of a person's living conditions that may be missed at the individual level.37–39
Despite equal access to health care services, differences persist in the presentation of Hispanic women with breast cancer compared with NHW women. The results of this study, in our opinion, lend further support to the evidence for a biologic/genetic basis for these differences. The persistent findings of earlier mean age at diagnosis, advanced stage, poorer grade, larger tumor size, and fewer cases with ERs may suggest that true biologic differences exist in breast cancer by ethnicity. Future research should more carefully explore differences in clinical presentation as well as biologic differences in tumor genotypes and phenotypes, as different strategies for breast cancer prevention may then be warranted for Hispanic women.
We thank Kaiser Permanente of Colorado, Kim Bischoff with the KPCO Tumor Registry, Terry Field with the Cancer Research Network, and the Colorado Central Cancer Registry for assistance.