Abstract
Purpose:
The impact of racial residential segregation on breast cancer treatment disparities is unclear. We examined whether racial segregation is associated with adjuvant treatment receipt and patient knowledge of disease.
Methods:
We surveyed a population-based sample of women in Northern California with stage I-III breast cancer diagnosed in 2010–2011 (participation rate=68.5%). For black, Hispanic and white women, we measured black and Hispanic segregation using the location quotient of racial residential segregation (LQ), a proportional measure of the size of a minority group in the census tract compared to the larger metropolitan statistical area. We categorized LQ values for blacks and Hispanics into quartiles, with Quartile 1 representing a lower relative level of segregation than Quartile 4. We used multivariable logistical regression to assess the odds of receiving guideline recommended adjuvant therapy and patient knowledge of tumor characteristics by relative residential segregation.
Results:
We observed greater residential segregation for black vs. Hispanic patients (p<.05). Overall, there were no treatment differences by Hispanic LQ or black LQ, except for black LQ Quartile 3 (vs. 1) where we observed higher odds of hormonal therapy. Knowledge of disease did not vary by black LQ, but patients in Hispanic LQ Quartile 3 (vs. Quartile 1) had less tumor knowledge.
Conclusions:
We did not find clear associations for racial residential segregation and treatment or cancer knowledge in Northern California, an area with low levels of segregation. Additional research should assess the impact of segregation on breast cancer treatment disparities in a variety of geographical locations.
Keywords: breast cancer, disparities, segregation, patient knowledge, treatment receipt
Introduction:
Racial and ethnic disparities in breast cancer treatment are well documented and contribute to worse outcomes for racial and ethnic minorities[1,2]. Mortality rates for black (vs. white) women have remained substantially higher over time, with a widening of disparities in breast cancer mortality in recent decades[3,4].
Racial residential segregation, defined as the degree to which two or more groups live separately from one another, has been suggested to be a fundamental cause of racial and ethnic disparities by causing racial differences in socioeconomic status, access to services and environments harmful for health[5,6]. Growing evidence suggests that racial residential segregation (referred to as segregation hereafter) of racial and ethnic minorities may be associated with racial disparities in the diagnosis, treatment, and mortality rates of breast cancer and other cancers[7–10].
Prior literature on segregation and breast cancer disparities has focused primarily on black-white and Hispanic-white segregation, and studies have shown mixed results. 11−13This may be due to variability in geographical areas used in analysis, measures of segregation, and geographic locations of patient cohorts [11–13]. However, a new segregation measure, location quotient of residential segregation (LQ), has increasingly been used in health disparities literature. It is a measure of the relative concentration (or ratio) of individuals in a minority group in a large geographical unit (metropolitan statistical area[MSA]) compared to a small geographical unit (census tract)[6]. The strength of this measure is that it allows for comparison across different geographical units and assesses differences in the racial composition of individual’s neighborhood compared to the larger MSA.(REF 6,12 below) The primary benefit of LQ compared to traditional measures of segregation such as, Isolation Index (II) and Dissimilarity Index (DI), is that it is a relative measure comparing two geographic units as opposed to an analysis of one concrete geographic unit allowing for assessment of relative deprivation. Recent research has suggested that societal inequities, including health inequity, is best studied by assessing relative differences between groups or individuals as opposed to assessing in isolation[14]. Additionally, traditional measures of segregation such as Index of Dissimilarity, have historically been used to examine larger population levels such as cities or MSAs. Specifically, these measures have sought to identify the likelihood of different racial/ethnic groups interacting within a larger geographical unit. The use of LQ, as a small area metric, allows for comparison of the immediate racial/ethnic composition of the environment in which an individual lives compared to the surrounding area.[15] A recent analysis using LQ determined that greater levels of racial and ethnic minority segregation had an independent adverse association with breast cancer mortality[12]. However, to our knowledge, no prior study has assessed LQ and its association with intermediate factors to mortality, including patient knowledge of disease and adjuvant treatment receipt, or whether relative changes in LQ are associated with breast cancer outcomes.
Many patients lack knowledge of their own tumor characteristics, and less knowledge is associated with less receipt of guideline-recommended adjuvant treatments[16, 17]. Black and Hispanic women, for instance, are less likely to know their own tumor stage, estrogen receptor (ER) status and human epidermal growth factor receptor 2 (HER2) status than white women[17]. One prior study found that greater black and Hispanic segregation was associated with less receipt of adequate care, this study was a claims-based analysis, focused on older patients only, and did not define adequate treatment based on tumor characteristics[11].
In this study, we examined whether relative increases of segregation of racial and ethnic minorities, measured by LQ, was associated with women’s knowledge about their breast cancers and with adjuvant receipt of hormonal therapy, radiation, and chemotherapy among women with early-stage breast cancer. We hypothesized that segregation may contribute to disparities in both treatment and cancer knowledge (Figure 1).
Figure 1.
Conceptual Model of relationship between racial residential segregation and treatment receipt and patient knowledge.
Methods:
Study Design:
Figure 1 represents our conceptual model for how segregation of racial and ethnic minorities may contribute to disparities in treatment and patient knowledge. This model suggests that reduced access to medical services and neighborhood characteristics in geographical areas with racial and ethnic minority segregation leads to reduced cancer care treatment and limited patient knowledge. To address our question, we completed a secondary analysis of data previously collected as part of telephone-based survey, which primarily examined surgeon and hospital selection and treatment receipt for a population-based sample of breast cancer patients residing in Northern California.76 The details of our survey have been previously published [16–18]. In brief, we mailed study invitation letters to eligible participants with breast cancers reported to the California Cancer Registry (CCR) and then contacted them by phone to schedule interviews. All surveys were conducted by telephone using computer-assisted telephone interview software and participants received $20. Interviews were conducted in English and Spanish.
Participants:
The study cohort and exclusion criteria have been previously reported [16,17]. A total of 1118 white, black or Hispanic women from regions 1/8 (San Francisco/Santa Cruz) and region 3 (Sacramento) of the California Cancer Registry (CCR) were identified. All 1118 patients (Supplemental Figure 1), received a diagnosis of stage 0-III breast cancer in 2010–2011 and received primary breast surgery. Among the 1118 patients identified, 231 refused participation, 317 could not be reached, and 68 were deceased or ill (response rate=47.8%; participation rate among patients whose contact information was available=68.5%). Among the 502 survey respondents, 2 self-identified as Asian American and were excluded because our focus was on the care of black and Hispanic patients. Of these remaining 500 study participants, all participants with stage 0 (n=86) disease were excluded from the present analysis due to variability in treatment recommendations. An additional 28 participants were excluded due to inability to geocode their addresses. The final cohort included 386 study participants who self-identified as non-Hispanic white (n=163), non-Hispanic black (n=124), or Hispanic (n=99). The racial and ethnic demographic data for each individual participant’s block group, census tract, county and metropolitan statistical area (MSA) were obtained from the 2010 U.S. Census. We obtained study approvals from the CCR, the California Health and Human Services Agency Committee for the Protection of Human Subjects, and Harvard Medical School’s Committee on Human Studies.
Variables:
Our primary outcomes of interest were breast cancer knowledge and receipt of guideline-recommended adjuvant treatment. Breast cancer knowledge was defined as patient knowledge of her own breast cancer characteristics (stage, grade, ER and HER2 status) and was measured as an ordinal score of 0–4 (number of answers correct). As per prior definitions, answers were considered ‘correct’ if a participant’s answer matched the characteristic according to the CCR or if the CCR result was ‘unknown,’ ‘not performed,’ or missing. One woman who reported having stage IV disease was considered to correctly report stage in case her cancer (originally diagnosed as stage 0-III in the CCR) recurred[16].
Adjuvant treatment receipt was defined as either self-reported or CCR report of various treatments for breast cancer. For this analysis, patients were divided into treatment cohorts according to their tumor characteristics, so that receipt of adjuvant chemotherapy, radiation and hormonal therapy could be examined. The chemotherapy cohort (n=173) was defined as study participants with the following disease criteria: stage IIB (any subtype), stage III (any subtype), tumor >1 cm if ER and progesterone receptor (PR)-negative, or tumor >1 cm if HER2-positive. The radiation therapy cohort (n=236) included those who underwent breast conserving surgery or unilateral or bilateral mastectomy for stage III disease. The hormonal therapy cohort (n=295) included participants with stage I-III, ER or PR-positive cancers. Study participants could be included in more than one treatment group, depending on disease characteristics.
Quantitative Variable:
Our independent variable of interest was location quotient of racial residential segregation (LQ)[6]. LQ is a relative measure of two proportions that compares the proportion of minority group m in census tract relative to the proportion of minority group m in the larger MSA for comparison across different geographical units, allowing for assessment of differences in the racial composition of individual’s neighborhood compared to the larger MSA [6,12].
The formula is calculated by the following:
LQim=(xim/Xi) / (Ym/Y)
LQim=the value of ith census tract in a MSA for minority group m
xim=the number of individuals from minority group m living in the ith census tract;
Xi=the total number of residents in the ith track
Ym=the total number of individuals from minority group m in the MSA
Y=total number of residents in the MSA
The LQ ranges from zero to infinity, with values <1 indicating that the proportion of minority group m in the census tract is less than the proportion of the same group in the larger MSA. A LQ >1 indicates that the proportion of minority group m in the census tract is more than the proportion of the same group in the larger MSA. For instance, an LQ=2 describes that the census tract proportion of minority group m is 2 times the proportion of the same group in the MSA[6]. For each patient, black LQ and Hispanic LQ, were calculated and analyzed using quartiles, with Quartile 1 representing the lowest relative level of segregation in our analysis. Because black or Hispanic LQ values range from zero to infinity, the use of quartiles allows for the assessment of how relative increases in black or Hispanic segregation, impacts the outcomes of interest. We selected quartiles to evaluate LQ because prior cited cut-offs in the literature have not been extensively validated for both black and Hispanic segregation and were not easily applied to our population of patients who resided in lower-segregated areas [6]. The use of quartiles allowed for analysis to be conducted with the intent of evaluation of how incremental increases in segregation within our study population impacted patient knowledge and treatment receipt. The categories for black LQ quartiles in our analysis were the following: Quartile 1 (.03-.32), Quartile 2 (.33-.96), Quartile 3 (.97–1.74), and Quartile 4 (1.75–6.59). The cut-offs for Hispanic LQ quartiles were the following: Quartile 1 (0.12–0.58), Quartile 2 (0.59–0.92), Quartile 3 (0.93–1.37), Quartile 4 (1.38–3.44). We mapped the LQ quartiles assigned to each geocode of residence for the included CCR areas. For each black LQ and Hispanic LQ quartile, analyses were conducted to examine associations of race/ethnicity, adjuvant treatment receipt and patient knowledge with different levels of segregation.
Control variables of interest included tumor characteristics as well as variables previously shown to impact treatment and knowledge, including race/ethnicity, age, educational attainment, tumor stage, health literacy, and co-morbidities, all categorized similarly to past analyses [18,19]. The comorbidity variable was determined based on the Charlson index and by adding the number of self-reported medical conditions [20–21]. Variables were defined and categorized as per Table 1.
Table 1.
Characteristics of Entire Cohort and by Adjuvant Therapy Eligibility
| Entire cohort N=386 | Hormonal therapy eligible N=295 | Radiation therapy eligible N=236 | Chemotherapy eligible N=173 | |
|---|---|---|---|---|
| Age* (years) | ||||
| <50 | 103(26.68) | 75(25.42) | 56(23.73) | 57(32.95) |
| 50–59 | 114(29.53) | 81(27.46) | 70(29.66) | 58(33.53) |
| 60+ | 169(43.78) | 129 (47.12) | 110 (46.61) | 58(33.53) |
| Race‡ | ||||
| Non-Hispanic white | 163(42.23) | 128(43.39) | 101(42.80) | 66(38.15) |
| Non-Hispanic black | 124(32.12) | 92(31.19) | 69(29.24) | 65(37.57) |
| Hispanic | 99(25.65) | 75(25.42) | 66(27.97) | 42(24.28) |
| Educational Attainment‡ | ||||
| High school or less | 129(33.42) | 98(33.22) | 77(32.63) | 53(30.64) |
| Some college | 121(31.35) | 93(31.53) | 79(33.47) | 54(31.21) |
| College graduate | 136(35.35) | 104(35.25) | 80(33.90) | 66(38.15) |
| Annual Household Income ($) ‡ | ||||
| <20,000 | 80(20.73) | 54(18.31) | 48(20.34) | 40(23.12) |
| 20,000–39,999 | 72(18.65) | 54(18.31) | 41(17.37) | 34(19.65) |
| 40,000–59,999 | 57(14.77) | 46(15.59) | 37(15.68) | 21(12.14) |
| >60,000 | 138(35.75) | 107(36.27) | 81 (34.32) | 65(37.57) |
| Unknown | 39(10.10) | 34(11.53) | 29(12.29) | 13(7.51) |
| Marital Status‡ | ||||
| Unmarried/Unknown | 178(46.11) | 137(46.44) | 111(47.03) | 79(45.66) |
| Married | 208(53.89) | 158(53.56) | 125(52.97) | 94(54.34) |
| Insurance Status‡ | ||||
| None | 23(5.96) | 14(4.75) | 12(5.08) | 13(7.51) |
| Veteran’s Admin. | 5(1.30) | 3(1.02) | 2(0.85) | 2(1.16) |
| Medicaid | 26(6.73) | 20(6.77) | 13(5.50) | 15(8.67) |
| Medicare | 95(22.61) | 78(26.44) | 64(27.12) | 28(16.18) |
| Medicare-Medicaid | 17(4.40) | 11(3.73) | 10(4.24) | 7(4.05) |
| Private | 219(56.74) | 169(57.29) | 134(56.78) | 107(61.85) |
| Other | 1(0.26) | 0(0.00) | 1(0.42) | 1(0.58)13(7.51) |
| Tumor Stage* | ||||
| I | 186(48.19) | 151(51.19) | 125(52.97) | 43(24.86) |
| II | 154(39.90) | 109(36.95) | 68(28.31) | 84(48.55) |
| III | 46(11.92) | 35(11.86) | 43(18.22) | 46(26.59) |
| Tumor Size* | ||||
| <1cm | 96(24.87) | 81(27.46) | 64(27.12) | 18(10.40) |
| >1cm | 290(75.13) | 214(72.54) | 172(72.88) | 155(89.60) |
| Health Literacy Score, Mean (Sd)§ | 1.85(1.08) | 1.75(1.01) | 1.80(1.06) | 1.66(0.95) |
| Comorbidity Score‡ | ||||
| 0 | 235(60.88) | 176(59.66) | 145(61.44) | 110(63.58) |
| 1 | 103(26.68) | 81(27.46) | 66(27.97) | 43(24.86) |
| 2+ | 48(12.44) | 38 (12.88) | 25 (10.59) | 20(11.56) |
| Black Location Quotient | ||||
| Quartile 1 (Q1) | 96(24.87) | 70(25.08) | 69(29.24) | 42(24.28) |
| Quartile 2 (Q2) | 96(24.87) | 74(25.08) | 57(24.15) | 48(27.75) |
| Quartile 3 (Q3) | 97(25.13) | 82(27.80) | 59(25.00) | 31(17.92) |
| Quartile 4 (Q4) | 97(25.13) | 69(23.39) | 51(21.61) | 52(30.06) |
| Hispanic Location Quotient | ||||
| Quartile 1 (Q1) | 96(24.87) | 75(25.42) | 64(27.12) | 44(25.43) |
| Quartile 2 (Q2) | 97(25.13) | 76(25.76) | 56(23.73) | 40(23.12) |
| Quartile 3 (Q3) | 96(24.87) | 68(23.05) | 60(25.42) | 50(28.90) |
| Quartile 4 (Q4) | 97(25.13 | 76(25.76) | 56(23.73) | 39(22.54) |
| Knowledge Composite Score | ||||
| 0 | 54(13.99) | 44(14.92) | 28(11.86) | 27(15.61) |
| 1 | 107(27.72) | 92(23.83) | 63(26.69) | 40(23.12) |
| 2 | 104(26.94) | 73(24.74) | 67(28.39) | 42(24.28) |
| 3 | 94(24.35) | 67(22.71) | 60(25.42) | 50(28.90) |
| 4 | 27(6.99) | 19(6.44) | 18(7.63) | 14(8.09) |
Self-Report: race/ethnicity, education attainment, marital status, insurance, and comorbidities
Registry Report: age, tumor stage, and tumor size.
Health literacy scores were calculated using a series of three questions: (1) “How confident are you filling out medical forms?”(2) “How often do you have someone help you read hospital materials?”(3) “How often do you have problems learning about your medical conditions?” All responses used a five item Likert scale with 1 representing the most confidence and fewest problems. [19]
Statistical methods:
We used chi-squared tests and Kruskal-Wallis tests to assess differences in baseline participant and tumor variables by each relevant treatment group. We also examined differences in black LQ and Hispanic LQ by study participant race/ethnicity using chi-squared tests. An additional analysis was done evaluating differences in patient knowledge of each tumor characteristic (stage, grade, ER and HER2 status) by black LQ and Hispanic LQ.
Next, we performed multivariable logistic regression for adjuvant treatment receipt and ordinal logistic regression for patient knowledge (summing outcome of answering more tumor characteristics questions correctly, 0–4), first with a base model which assessed the odds of each outcome of interest (adjuvant treatment receipt and patient knowledge), adjusting for race/ethnicity, age, education, tumor size, health literacy, and comorbidity. We did not include insurance in models because few were uninsured (n=23). In a second set of models, we added black and Hispanic LQ separately, to assess the association between (a) segregation and treatment receipt for each treatment-eligible cohort and (b) patient knowledge with black LQ and Hispanic LQ Quartile 1, the lowest relative level of segregation, as the reference group. In a sensitivity analysis, we repeated models comparing Quartile 4 for black LQ and Hispanic LQ with the remaining quartiles, given that this quartile included women with the most evident segregation relative to the larger MSA. With the use of Quartile 4 as the reference group, this sensitivity analysis allowed for assessment of the association between relative decreases in segregation and the outcome variables of interests. Ultimately, the comparison of analysis with Quartile 1 and 4 as the reference groups allowed for assessment for how both increases and decreases in racial and ethnic segregation impacted the outcomes of interest.
Results:
The baseline characteristics of each treatment-eligible cohort are summarized in Table 1. Overall, the study cohort was 42% non-Hispanic white, 32% non-Hispanic black, and 26% Hispanic. Most participants were aged >50 years (Table 1). Within the adjuvant chemotherapy-eligible cohort, approximately 82% received chemotherapy. Among those eligible for hormonal therapy, 83% received hormonal therapy, and 92% of those eligible for radiation received treatment.
Racial Segregation:
Overall, the average black LQ value was larger than for Hispanic LQ segregation (1.21 vs. 1.06) indicating overall higher relative levels of black segregation compared to Hispanic segregation (Table 2). There were significant differences in the proportion of non-Hispanic white, non-Hispanic black, and Hispanic participants in each quartile of black LQ but only within Quartile 1 for Hispanic LQ (all p<.05) (Figures 2a and 2b). The black LQ and Hispanic LQ quartiles for the census tracts in Northern California in which our study participants resided are shown in Figures 3a and 3b.
Table 2.
Interquartile Range of Black Location Quotient & Hispanic Location Quotient.
| Segregation Variables | ||||||||
|---|---|---|---|---|---|---|---|---|
| Mean±SD | Median(IQR) | Min | Max | 1st Quartile | 2nd Quartile | 3rd Quartile | 4th Quartile | |
| Black LQ | 1.213+1.15 | 0.96 (0.32–1.74) | 0.03 | 6.59 | 0.03–0.32 | 0.33–0.96 | 0.97–1.74 | 1.75–6.59 |
| Hispanic LQ | 1.06+0.66 | 0.92 (0.58–1.37) | 0.12 | 3.34 | 0.12–0.58 | 0.58–0.92 | 0.92–1.37 | 1.38–3.44 |
Note: Black LQ and Hispanic LQ represent black location quotient and Hispanic location quotient, respectively. The values represent the proportion of each minority group in a census tract compared to the proportion in the Metropolitan Statistical Area (MSA). Higher black LQ and Hispanic LQ values represent higher levels of segregation.6
Abbreviations: SD, standard deviation, IQR, inter-quartile range.
Figure 2.
a. Black Location Quotient by Race/Ethnicity
b. Hispanic Location Quotient by Race/Ethnicity
Note: Unadjusted proportion of study cohort participants by race/ethnicity within each quartile of black LQ and Hispanic LQ. P-values by X2 testing comparing the proportion of patients of each race/ethnicity within quartile. Quartile 1 represents lowest segregation and Quartile 4 represents highest segregation. Blue bars represent Non-Hispanic White participants, orange bars represent Non-Hispanic black participants, and gray bars represent Hispanic participants.
Abbreviations: LQ, location quotient
Figure 3.
a. Black Location Quotient of Study Participant Census Tracts
b. Hispanic Location Quotient of Study Participant Census Tracts.
Note: The census tracts included are within the San Francisco, Santa Cruz and Sacramento regions in the California Cancer Registry.
Adjusted Results:
Adjuvant Treatment Receipt:
In the base model for adjuvant hormonal therapy and in the model adjusting for racial segregation, there were significant differences in the adjusted odds of treatment by age, race/ethnicity, and tumor stage which persisted after adjusting for LQ (Table 3). Black women were less likely to receive hormonal therapy compared to whites (odds ratio[OR]=0.38; 95% confidence interval [CI]=0.15–0.95). In addition, individuals age >60 had significantly lower odds of hormonal therapy receipt than those age <50 (OR=0.25; 95% CI=0.09–0.69) and those with stage II (vs. I) disease had higher odds of hormonal therapy. Participants with a black LQ in Quartile 3, were more likely to receive adjuvant hormonal therapy than those in Quartile 1 (OR=4.06; 95% CI 1.26–12.93). For the radiation-eligible cohort, there were no significant differences in treatment by subgroup or for black LQ or Hispanic LQ. In the chemotherapy-eligible cohort, age ≥60 (vs. <50) and having some college (vs. no high school diploma) were associated with lower odds of chemotherapy, while higher stage (II-III vs. I) was associated with higher odds of treatment. There were no differences observed in chemotherapy receipt by race/ethnicity or by Hispanic or black LQ. In our sensitivity analysis with Quartile 4 vs. other groups, there were no significant differences in adjuvant treatment receipt by black LQ or Hispanic LQ (data not shown).
Table 3.
Unadjusted Percentage and Adjusted OR for Knowledge (n=386) and for Adjuvant Treatment Receipt Among Each Treatment-Eligible Group.
| Patient Knowledge | Treatment Receipt | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Having more answers correct N=386 | HT Receipti N=244 | pi | Adjusted OR for HT Receiptxx | RT Receipti N=216 | Pi | Adjusted OR for RT Receiptxx | CT Receipti N=142 | Pi | Adjusted OR for CT Receiptxx | |
| Age* (years) | ||||||||||
| <50 | 1.00 | 67(89.33) | 0.003 | 1.00 | 49(87.50) | 0.464 | 1.00 | 53(92.98) | 0.0009 | 1.00 |
| 50–59 | 0.60(0.36–0.99) | 73(90.12) | 1.22(0.39–3.85) | 65(92.86) | 1.61(0.42–6.13) | 50(86.21) | 0.46(0.12–1.82) | |||
| ≥60 | 0.26(0.156–0.43) | 104(74.82) | 0.25(0.09–0.69) | 102(92.73) | 1.12(0.31–4.08) | 39(67.24) | 0.08(0.02–0.33) | |||
| Race‡ | 0.791 | |||||||||
| Non-Hispanic white | 1.00 | 109(85.16) | 0.238 | 1.00 | 91(90.10) | 1.00 | 57(86.36) | 0.360 | 1.00 | |
| Non-Hispanic black | 0.49(0.30–0.82) | 71(77.17) | 0.38(0.15–0.95) | 64(92.75) | 1.12(0.30–4.21) | 50(76.92) | 0.70(0.18–2.65) | |||
| Hispanic | 0.59(0.35–1.00) | 64(85.33) | 0.99(0.37–2.67) | 61(92.42) | 1.25(0.35–4.48) | 35(83.33) | 0.45(0.11–1.86) | |||
| Education Attainment‡ | ||||||||||
| High school or less | 1.00 | 77(78.57) | 0.062 | 1.00 | 73(94.81) | 0.229 | 1.00 | 45(84.91) | 0.595 | 1.00 |
| Some college | 3.17(1.92–5.25) | 84(90.32) | 2.37(0.91–6.14) | 69(87.34) | 0.39(0.10–1.48) | 42(77.78) | 0.23(0.06–0.87) | |||
| College graduate | 2.23(1.35–3.70) | 83(79.81) | 0.57(0.24–1.37) | 74(92.50) | 0.72(0.16–3.19) | 55(83.33) | 0.54(0.16–1.84) | |||
| Tumor Stage* | ||||||||||
| I | 1.00 | 115(76.16) | 0.0008 | 1.00 | 115(92.00) | .0880 | 1.00 | 29(67.44) | 0.0097 | 1.00 |
| II | 1.37(0.92–2.05) | 102(93.58) | 5.99(2.34–15.34) | 65(95.59) | 1.98(0.49–8.03) | 71(84.52) | 4.66(.53–14.17) | |||
| III | 1.14(0.63–2.09) | 27(77.14) | 1.06(0.39–2.86) | 36(82.72) | 0.53(0.17–1.66) | 42(91.30) | 7.13(1.74–29.31) | |||
| Health Literacy Score, Mean (continuous)§ | 1.74 | 0.329 | 0.96(0.66–1.40) | 1.81 | 0.689 | 0.94(0.54–1.63) | 1.69 | 0.515 | 1.34(0.719–2.49) | |
| Comorbidity Score‡ | ||||||||||
| 0 | 1.00 | 148(84.09) | 1.00 | 131(90.34) | 0.614 | 1.00 | 90(81.82) | 0.564 | 1.00 | |
| 1 | 0.78(0.50–1.21) | 65(80.25) | 0.736 | 1.40(0.63–3.11) | 61(92.42) | 1.46(0.46–4.60) | 37(86.05) | 4.07(1.08–15.36) | ||
| 2+ | 0.95(0.52–1.74) | 31(81.58) | 1.97(0.66–5.89) | 24(96.00) | 2.62(0.30–23.24) | 15(75.00) | 1.88(0.44–0.81) | |||
| Black Location Quotient | ||||||||||
| Quartile 1 | 1.00 | 57(81.43) | 0.170 | 1.00 | 62(89.86) | 0.749 | 1.00 | 35(83.33) | 0.667 | 1.00 |
| Quartile 2 | 1.11(0.65–1.90) | 62(83.78) | 1.27(0.46–3.47) | 51(89.47) | 0.98(0.27–3.58) | 40(83.33) | 0.77(0.20–2.94) | |||
| Quartile 3 | 1.41(0.79–2.52) | 73(89.02) | 4.06(1.26–12.93) | 55(93.22) | 1.52(0.33–7.05) | 27(87.10) | 0.78(0.15–4.14) | |||
| Quartile 4 | 0.96(0.52–1.78) | 52(75.36) | 1.04(0.35–3.13) | 48(94.12) | 1.86(0.32–10.68) | 40(76.92) | 0.61(0.13–2.85) | |||
| Hispanic Location Quotient | ||||||||||
| Quartile 1 | 1.00 | 66(88.00) | 0.454 | 1.00 | 58(90.63) | 0.810 | 1.00 | 36(81.82) | 0.248 | 1.00 |
| Quartile 2 | 0.83(0.48–1.45) | 63(82.89) | 0.49(0.16–1.52) | 51(91.07) | 0.83(0.19–3.57) | 35(87.50) | 2.66(0.62–11.36) | |||
| Quartile 3 | 0.53(0.30–0.96) | 53(77.94) | 0.51(0.16–1.6) | 54(90.00) | 0.71(0.17–3.04) | 43(86.000 | 2.63(0.62–11.15) | |||
| Quartile 4 | 0.76(0.42–1.36) | 62(81.58) | 0.78(0.25–2.48) | 53(94.64) | 1.20(0.23–6.30) | 28(71.79) | 0.89(0.22–3.64) | |||
Note: All data are presented as whole number, percentage or odds ratio (OR,95% CI). Values in bold represent p<.05.
Abbreviations: HT, hormonal therapy, RT, radiation therapy, CT, chemotherapy, OR, odds ratio
Percentages represent the unadjusted proportion of treatment eligible cohort that received treatment by each variable listed. P values were calculated by X2 for the categorical variables and by Kruskal-Wallis test for health literacy.
Using multivariable logistic regression, adjusting for all variables in the table.
Patient Knowledge:
We observed significant differences in summed knowledge scores for tumor characteristics by age, race, education and health literacy in the base ordinal logistic regression model (Table 3). Knowledge of disease did not vary significantly by black LQ. However, participants with Hispanic LQ in Quartile 3 answered significantly fewer questions correctly regarding their tumor characteristics compared to Quartile 1 (OR=0.53; 95% CI=0.30–0.96). When we examined associations of knowledge of specific tumor variables, a higher proportion of women reported correct ER in Quartile 1 of black LQ (68%) compared with those residing in Quartile 4 (45%, p=0.02) (Supplemental Figures 2a and 2b). In addition, the proportion of women reporting correct ER was significantly different by Hispanic LQ quartiles (76% in Quartile 1 vs. 40% in Quartile 4, p<0.0001). We also observed significant differences by quartiles of Hispanic LQ for knowledge of stage, with those residing in Hispanic Quartile 1 having the highest proportion of women reporting correct disease stage (69% in Hispanic LQ Quartile 1 vs. 48% in Quartile 4 areas, p<.05). Knowledge of HER2 and grade were not associated with Hispanic or black LQ quartiles (Supplemental Figures 3a and 3b). In our sensitivity analysis with comparisons to Quartile 4, there were no significant differences in patient knowledge by black or Hispanic LQ.
Discussion:
In this population-based analysis of women with early stage breast cancer, we found limited associations between relative increases in segregation and treatment receipt as well as overall patient knowledge, but potential associations of segregation with knowledge of individual tumor variables such as ER and stage. In contrast to our hypothesis, despite observing differences in the degree of segregation for black and Hispanic women, there was no consistent association between women who lived in areas with relatively higher levels of black or Hispanic segregation for treatment receipt of adjuvant chemotherapy, hormonal therapy, or chemotherapy.
Several studies have suggested that segregation contributes to neighborhood variation in health outcomes and health care access[22–26]. Segregation, as the result of state and federal policies, remains relevant to neighborhood composition and social polarization, with African Americans living in the most segregated communities followed by Hispanics [26]. In areas with higher levels of segregation, there are lower levels of economic and education opportunities, higher rates of concentrated poverty, and reduced access to medical services[5,27]. Thus, segregation may serve as a contributor to racial and ethnic heath disparities, although the extent to which segregation may contribute to cancer treatment disparities is not well understood.
Our results are not consistent with a prior study suggesting that as black and Hispanic segregation increased, black and white women living in these areas were less likely to receive adequate breast cancer care [11]. However, another study demonstrated that the black/white and Hispanic/white disparity in breast cancer stage (early vs. late) at diagnosis was largest in lower-income, less segregated areas and smallest in areas with more segregation[28]. Conflicting findings are due to differences in segregation measures used, criteria regarding treatment receipt, cohort variation and variability of geographical distribution. Further, these studies in conjunction with our findings suggest that segregation may have variable effects along the spectrum of diagnosis to treatment (e.g. targeted outreach for breast cancer screening in segregated neighborhoods may reduce disparities in stage at diagnosis but may not impact treatment receipt).
Our study represents the first analysis to examine potential associations between relative increases segregation and two intermediate contributors to racial disparities in breast cancer mortality, patient knowledge of disease and receipt of adjuvant therapy. Although we found no clear associations, it will be important to assess if knowledge of one’s disease is associated with segregation in other populations.
Our study has several limitations. First, all study participants had geographical addresses in Northern California, which has had historically lower levels of racial segregation compared to other metropolitan areas, such as Detroit or Chicago, potentially limiting our ability to assess the effects of more extreme levels of segregation[29]. For instance, it has been previously reported that amongst the largest 25 cities in the United States, cities in Northern California, specifically San Francisco and San Jose, are among the cities were the mortality rates between non-Hispanic Black and non-Hispanic White women were not significantly different, which may be secondary to the relatively small percentage of non-Hispanic Blacks in these cities and/or the low levels of segregation[30]. Further, not all census tracts in Northern California were represented by our study population (Figure 2). Second, some subgroups were small and the proportion of women receiving adjuvant treatment was high, and we may have had limited power to detect meaningful associations between segregation and outcome variables. Third, we only included individuals who selfidentified as black, white or Hispanic, so generalizability to other racial and ethnic groups is limited. Fourth, our study focused on care during 2010–2011; and do not reflect care with more recently approved treatments. Lastly, LQ is a relatively new measure of segregation used in the health services literature. As a relative measure, it only considers one minority group at two different geographical units, which may limit the ability to assess segregation patterns including multiple racial/ethnic groups or to detect true segregation within one given geographical area. However, by assessing LQ with quartiles, we were able to assess trends in relative increase and decrease in segregation.
In conclusion, we did not find clear associations between racial residential segregation and receipt of recommended treatment or patient’s knowledge of one’s illness. To clarify the impact of racial segregation further, future studies should focus on associations between residential racial segregation and racial differences in breast cancer mortality and should include varying measures of segregation, different geographic locations, and assessment of potential mediating factors that contribute to persistent racial and ethnic disparities in cancer mortality.
Supplementary Material
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Funding: Susan G. Komen (NLK, RAF), American Cancer Society Mentored Research Scholar Grant (RAF), National Cancer Institute (NLK [K24CA181510])
Disclosures: The authors declare that they have no conflict of interest.
Ethical approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Data availability: The datasets during and/or analyzed during the current study are available from the corresponding author on reasonable request.
References
- 1.Shavers VL, Brown ML. Racial and ethnic disparities in the receipt of cancer treatment. J Natl Cancer Inst. 2002; 94(5):334–357. [DOI] [PubMed] [Google Scholar]
- 2.Joslyn SA. Racial differences in treatment and survival of early-stage breast carcinoma. Cancer. 2002; 95(8): 1759–1766. [DOI] [PubMed] [Google Scholar]
- 3.DeSantis CE, Fedewa SA, Sauer AG, et al. Breast cancer statistics, 2015: convergence of incidence rates between black and white women. CA: Cancer J Clin. 2015; 66(1):31–42. [DOI] [PubMed] [Google Scholar]
- 4.Smigal C, Jemal A, Ward E, et al. Trends in Breast Cancer by race and ethnicity: update 2006. CA: Cancer J Clin. 2006; 56(3):168–183. [DOI] [PubMed] [Google Scholar]
- 5.Williams DR, Collins C. Racial residential segregation: a fundamental cause of racial disparities in health. Public Health Reports. 2001; 116(5):404–416. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Sudano JJ, Perzynski A, Wong DW, et al. Neighborhood racial residential segregation and changes in health or death among older adults. Health and Place. 2013; 19:80–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hayanga Ja, Kaiser HE, Sinha R, et al. Residential segregation and access to surgical care by minority populations in US counties. J Am Coll Surg. 2009; 208(6): 1017–1022. [DOI] [PubMed] [Google Scholar]
- 8.Johnson AM, Hines RB, Bayakly R. The effects of residential segregation and neighborhood characteristics on surgery and survival in patients with early-stage non-small cell lung cancer. Cancer Epidemiol Biomarkers Prev. 2016; 25(5): 750–758. [DOI] [PubMed] [Google Scholar]
- 9.Hayanga AJ, Zeliadt SB, Backhus LM. Residential Segregation and Lung Cancer Mortality in the United States. JAMA Surgery. 2013;148(1):37–42. [DOI] [PubMed] [Google Scholar]
- 10.Dai D Black residential segregation, disparities in spatial access to health care facilities, and late-stage breast cancer diagnosis in metropolitan Detroit. Health and Place 2010;16(5): 1038–1052. [DOI] [PubMed] [Google Scholar]
- 11.Haas JS, Earle CC, Orav JE, et al. Racial segregation and disparities in breast cancer care and mortality. Cancer. 2008;113(8): 2166–2172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Pruitt SL, Craddock SJ, Tiro JA, et al. Black, white, and Hispanic breast cancer patients in Texas, 1995 to 2009. Cancer. 2015; 121(11):1845–1855. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.White K, Haas JS, Williams DR. Elucidating the role of place in health care disparities: the example of racial/ethnic residential segregation. Health Serv Res. 2012; 47: 1278–1299. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wilkinson R, Pickett K. The problems of relative deprivation: Why some societies do better than others. Social Science and Medicine. 2007; 65:1965–1978. [DOI] [PubMed] [Google Scholar]
- 15.Bemanian A, Beyer, K. Measures Matter:The local exposure/isolation metrics and relationships between local-level segregation and breast cancer survival. Cancer Epidemiol Biomarkers. 2017; 26(4): 516–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Freedman RA, Kouri EM, West DW, et al. Association of breast cancer knowledge with receipt of guideline-recommended breast cancer treatment. J Oncol Practice. 2016; 12(6);e613–e625. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Freedman RA, Kouri EM, West DW, et al. Racial/ethnic disparities in knowledge about one’s breast cancer characteristics. Cancer. 2015; 121(5):724–732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Freedman RA, Kouri EM, West DW, et al. Racial/ethnic differences in patient’s selection of surgeons and hospitals for breast cancer surgery. Cancer. 2015; 1(2):222–230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Chew LD, Griffin JM, Partin MR, et al. Validation of screening questions for limited health literacy in a large VA outpatient population. J Gen Intern Med. 2008; 23: 561–566. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ayanian JZ, Chrischilles EA, Fletcher RH, et al. Understanding cancer treatment and outcomes: the Cancer Care Outcomes Research and Surveillance Consortium. J Clin Oncol. 2004;22:2992–2996. [DOI] [PubMed] [Google Scholar]
- 21.Katz JN, Chang LC, Sangha O, Fossel AH, Bates DW. Can comorbidity be measured by questionnaire rather than medical record review? Med Care. 1996; 34(1):73–84. [DOI] [PubMed] [Google Scholar]
- 22.Kershaw K, Albrecht S, Carnethon M. Racial and ethnic residential segregation, the neighborhood socioeconomic environment, and obesity among blacks and Mexican Americans. Am J Epidemiol. 2013;177(4):299–309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Dimick J, Ruhter J, Sarrazin M, et al. Black patients more likely than whites to undergo surgery at low-quality hospitals in segregated regions. Health Aff. 2013; 32(6):1046–1053. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Dinwiddie G, Gaskin D, Chan K, et al. Residential segregation, geographic proximity and type of services used: evidence for racial/ethnic disparities in mental health. Soc Sci Med. 2013; 30:67–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kershaw KN, Robinson WR, Gordo-Larsen P, et al. Association of changes in neighborhood-level racial residential segregation with changes in blood pressure among black adults: the CARDIA study. JAMA Intern Med. 2017;177(7):9961002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.U.S. Census Bureau, Housing and Household Economic Statistic Division. Racial and ethnic residential segregation in the United States: 1980–2000. https://www.census.gov/hhes/www/housing/housing_patterns/pdftoc.html
- 27.Massey DS, Denton NA. American apartheid: segregation and the making of the underclass. Harvard University Press; Cambridge, Mass; London, England: 1993. [Google Scholar]
- 28.Haas S, Earle CC, Orav JE, et al. Racial segregation and disparities in cancer stage for seniors. J Gen Intern Med. 2008; 23(5):699–705. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Massey DS. Residential Segregation and Neighborhood conditions in U.S. Metropolitan Areas, in Smelser NJ, Wilson WJ, Mitchell F(ed): America Becoming: racial rends and their consequences. Washington D.C., National Academy Press, 2001, pp 391–434. [Google Scholar]
- 30.Hunt BR, Whitman S, Hurlbert MS. Increasing Black:White disparities in breast cancer mortality in the 50 largest cities in the United States. Cancr Epidemiology. 2016. 45:169–173. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.