Abstract
Background
Inflammatory breast cancer (IBC) is a rare and highly aggressive form of primary breast cancer. Little is known regarding risk factors for IBC, specifically the association with socioeconomic position (SEP).
Methods
The association between breast cancer type (IBC vs. non-IBC) with county-level SEP in the Surveillance, Epidemiology, and End Results database for cases diagnosed from 2000–2007 was examined. County-level SEP characteristics included metropolitan vs. non- metropolitan residence, percent below the poverty level, percent less than high school graduate,and an index combining the poverty and high school variables. IBC and non-IBC age-adjusted incidence rates (IRs) were calculated, stratified on SEP and race/ethnicity. The odds of IBC vs. non-IBC given a particular SEP characteristic, adjusting for age and race/ethnicity, was examined through fitting of hierarchical logistic regression models (HLMs).
Results
IRs for IBC generally increased as SEP decreased, while the opposite was found for non-IBC. HLM results showed low SEP is associated with higher odds of IBC: Highest (≥ 20%) vs. lowest (<10%) persons below the poverty level Odds Ratio (95% Confidence Interval) = 1.25 (1.09–1.43); Highest (>28.76%) vs. lowest (≤15.99%) persons less than high school graduate = 1.25 (1.10–1.42); Low SEP as measured by poverty-high school index vs. high SEP = 1.26 (1.11–1.44).
Conclusion
Overall breast cancer has been found to be positively associated with SEP, whereas in this analysis IBC was associated with decreasing SEP.
Impact
Studies focused on understanding the disparity in IBC incidence, as well as interventions to eliminate these differences are needed.
Keywords: Inflammatory Breast Cancer, Socioeconomic Position, Epidemiology, Incidence Rates, Hierarchical Logistic Regression
Introduction
Breast cancer is a heterogeneous disease, characterized by distinct tumor subtypes thought to correspond to different etiologies (1–5). Inflammatory breast cancer (IBC) is a rare and highly aggressive form of primary breast cancer (6–11). Although risk factors for IBC remain largely unknown, some studies have shown different risk factor profiles for IBC as compared to non-IBC cases (12–15).
Breast cancer incidence in the United States is related to socioeconomic position (SEP) (US) (16), being greater among women with higher education and income (17, 18) and among women residing in communities with higher average levels of education and income (19–23). Although some studies have found much of this association can be explained by known breast cancer risk factors (17, 22), a study examining both individual- and community-level SEP revealed that after adjusting for individual SEP and breast cancer risk factors, women living in the highest SEP communities continued to have greater odds of having breast cancer compared to women living in the lowest SEP communities, suggesting community-level effects on breast cancer risk (16). Robert et al. hypothesized that these community effects could independently affect breast cancer risk through various pathways including more access to mammograms (leading to more breast cancer detection) and community norms such as exogenous hormone use, alcohol intake, and diet (16).
Higher incidence of overall breast cancer in urban areas, both in the US and internationally, has been reported for many years (24–28). Residence is also related to SEP, with rural residents in the US generally having lower income, less education, and lower health insurance coverage than their urban counterparts (29).
Given the lack of knowledge regarding factors associated with IBC incidence, and the evidence that some overall breast cancer risk factors may not have the same effect on IBC risk, the aim of this study was to examine the association of county-level SEP measures to IBC and non-IBC incidence in the US Surveillance, Epidemiology, and End Results (SEER) database linked to 2000 US Census county-attribute data.
Materials and Methods
Data Source
The SEER 17 Registries database linked to 2000 US county attributes was utilized for this analysis (30). The population-linked dataset includes all breast cancer cases from 2000–2007 for the following SEER registries: Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San-Francisco-Oakland, Seattle-Puget Sound, Utah, Los-Angeles, San Jose-Monterey, rural Georgia, the Alaska Native Tumor Registry, Greater California, Kentucky, Louisiana, and New Jersey (31). The US SEER database covers approximately 26% of the US population, including 23% of African Americans, 40% of Hispanics, 42% of American Indians and Alaska Natives, 53% of Asians, and 70% of Hawaiian/Pacific Islanders (32).
Individual-Level Measures
The outcome variable for this analysis was diagnosis of a first malignant primary breast cancer (International Classification of Diseases for Oncology (ICD-O-3) = C500-C509) as IBC or non-IBC. In order to be certain all IBC cases were captured, a comprehensive case definition was used where a breast cancer case having any one of the following codes assigned to the SEER variables below was classified as IBC (6, 15, 33–35):
Site and Morphology.Histologic Type ICD-O-3 (2000–2007) = 8530 (“Inflammatory Carcinoma”) (36, 37)
Stage - TNM.Derived AJCC [American Joint Committee on Cancer] T, 6th ed (for cases 2004+) = T4d (“Inflammatory Carcinoma”) (36, 38)
-
Extent of Disease [EOD] - CS.CS extension (for cases 2004+) = 71–73 (36, 39)
-
◦
71: “Diagnosis of inflammatory carcinoma without a clinical description of inflammation, erythema, edema, peau d’orange, etc., of more than 50% of the breast, with or without dermal lymphatic infiltration. Inflammatory carcinoma, NOS.”
-
◦
72: “Diagnosis of inflammatory carcinoma with a clinical description of inflammation, erythema, edema, peau d’orange, etc., of less than or equal to 50% of the breast, with or without dermal lymphatic infiltration.”
-
◦
73: “Diagnosis of inflammatory carcinoma with a clinical description of inflammation, erythema, edema, peau d’orange, etc., of more than 50% of the breast, with or without dermal lymphatic infiltration.”
-
◦
-
Extent of Disease - Historic.EOD 10 - extent (for cases 2000–2003) = 70 (36, 40)
-
◦
70: “Inflammatory carcinoma, including diffuse (beyond that directly overlying the tumor) dermal lymphatic permeation or infiltration”
-
◦
All other histologic types were considered non-IBC. Breast cancers assigned to histologic codes 9590 and greater (lymphomas) were not included.
Age at diagnosis was analyzed as a continuous variable. A merged race and ethnicity variable with the following categories was also examined: Non-Hispanic White (NH White), Black, Hispanic White, Asian/Pacific Islander (API), and American Indian/Alaska Native (AI/AN).
County-Level Measures
Four county-level measures of SEP derived from the 2000 US census were used in this analysis (41). Counties were divided into metropolitan (metro) versus non-metro areas based on their 2003 US Department of Agriculture Rural-Urban Continuum code (RUCC), as has been done in previous cancer studies (42–45). Codes 1–3 were defined as metro counties, while codes 4–9 defined non-metro counties (code definitions below).
Rural-Urban Continuum Code Definitions for 2003 (46, 47):
Metro counties:
1 (Counties in metro areas of 1 million population or more)
2 (Counties in metro areas of 250,000 to 1 million population)
3 (Counties in metro areas of fewer than 250,000 population)
Non-metro counties:
4 (Urban population of 20,000 or more, adjacent to a metro area)
5 (Urban population of 20,000 or more, not adjacent to a metro area)
6 (Urban population of 2,500 to 19,999, adjacent to a metro area)
7 (Urban population of 2,500 to 19,999, not adjacent to a metro area)
8 (Completely rural or less than 2,500 urban population, adjacent to a metro area)
9 (Completely rural or less than 2,500 urban population, not adjacent to a metro area)
Percent of persons within a county living below the federal poverty level was divided into three categories: <10%, 10%-19.99%, ≥20%. This measure has been used in other SEER-based studies of SEP variation and disparities in cancer incidence and outcomes (48, 49). Percent of the population below the poverty level has several advantages as a SEP measure. It is easily understood, being based on readily interpretable variables with a priori cut-points, with areas having ≥ 20% poverty generally considered “poverty areas” in census publications and studies using census data (48–51). Percent below the poverty level also takes into account a family’s size and age structure and is directly tied to the person’s ability to buy a representative basket of goods and services, being updated annually by the US Census Bureau to reflect changes in the Consumer Price Index (48, 52).
Education, specifically completion of high school measured at both the individual and aggregate level, has been previously shown to be positively associated with overall breast cancer incidence (19, 53). In order to examine the county-level effect of high school education on IBC incidence, percent of adults (>25 years of age) in a county who did not graduate from high school was divided into quartiles based on the distribution of the variable across all counties in the US using the SEER county attributes database (54). The quartiles were as follows: ≤15.99%, 16–20.80%, 20.81–28.76%, >28.76%.
To examine the joint effect of poverty and education on the odds IBC vs. non-IBC incidence, a composite measure of these two variables was also created. The percent below high school graduate quartiles were combined with the 3-category poverty variable to create high, middle, and low SEP categories as follows: High SEP = <10.00% poverty and <15.99% less than high school graduate, Low SEP = 1) 10%-19.99% poverty and >28.76% less than high school graduate, or 2) ≥20% poverty and >28.76% less than high school graduate, or 3) ≥20% poverty and 20.81–28.76% less than high school graduate, with the Middle SEP group being all remaining combinations. A similar index was used previously in a study examining overall breast cancer survival in the SEER database (43).
Statistical Analysis
Statistical software developed by the SEER program (SEER*stat version 6.6.2, National Cancer Institute, Bethesda, MD) readily allows for the calculation of incidence rates using the incidence dataset linked to county attribute and population data (55). Female adult (18+ years of age) IBC and non-IBC incidence rates (IRs) for the period from 2000–2007 were calculated. IRs were directly age-adjusted to the 2000 US standard population (56, 57), and further stratified on the following county-level SEP measures: metro vs. non-metro residence, percent below poverty, and percent less than high school graduate, as well as race/ethnicity. Upper and lower 95% confidence intervals (CIs) around the IR ratios were used to test for statistical significance. A 95% IR ratio CI not including 1 was indicative of the rates being statistically significantly different at the 0.05 significance level (58). 95% CIs for the IRs and IR ratios were calculated using the method described in Tiwari et al. (59). While SEER*stat allows for the calculation of incidence rates stratified by county attribute measures, it does not allow for the calculation of incidence rates by a variable created from the merging of two county attributes, such as the poverty-high school index.
In order to directly compare IBC to non-IBC cases’ association with SEP, hierarchical logistic regression models (HLMs) were fit using the SAS glimmix procedure (version 9.1; SAS Institute Inc., Cary, NC) (60). As women who reside in the same county may have similar, unmeasured characteristics, and therefore be more alike than individual women across different counties, it is important to take this intra-class correlation into account as ignoring it may lead to underestimation of the regression coefficient’s standard error (61, 62). A hierarchical modeling structure allows for the examination of both individual- and county-level fixed effects as well as accounting for the effect due to clustering within counties through the addition of a random intercept, leading to more accurate standard error estimates (61–63).
Four separate HLMs were fit, with each model including one county-level SEP measure as well as age at diagnosis and merged race/ethnicity as independent variables. Age at diagnosis and race/ethnicity were included in all models due to their previously reported association with the incidence of different histopathologic types of breast cancer, specifically IBC (1, 6, 7, 15, 33, 64–66). All HLMs modeled the log odds ratio (OR) of IBC. The general model is outlined below where the county intercept becomes a linear combination of a grand mean (α), the county SEP fixed effect (γ), and county random effects (μ), with β1 and β2 being individual fixed effects (63):
Results
346,211 first primary breast cancer cases in women 18+ years of age diagnosed from 2000–2007 were available for this analysis; 39 cases missing county-level information and 2,545 cases missing race/ethnicity were excluded, leaving 343,627 cases. After these exclusions, there were 5,536 IBC and 338,091 non-IBC cases included in this analysis. In the model examining metro vs. non-metro counties, a further 394 cases were excluded due to missing RUCC, leaving 5,525 IBC and 337,708 non-IBC cases for analysis.
Table 1 describes the study population characteristics. The mean age at diagnosis for IBC was 58.0 years vs. 60.7 years for non-IBC. The majority of IBC cases were of NH White race/ethnicity (67.0%), followed by Black (14.6%), Hispanic White (12.7%), API (5.1%), and AI/AN (0.7%). The comparable percentages for non-IBC were 75.0, 9.1, 8.6, 6.9, and 0.4%. 22.6% of IBC cases were classified as high SEP using the poverty-high school index, while 51.4% were middle SEP and 26.0% low SEP. Using the same index, 27.3% of non-IBC cases were considered high SEP, 50.8% middle SEP, and 21.8% low SEP.
Table 1.
Inflammatory Breast Cancer n and (%) |
Non-Inflammatory Breast Cancer n and (%) |
|
---|---|---|
Mean Age at Diagnosis (SD) | 58.0 (14.5) | 60.7 (14.3) |
Race/Ethnicity | ||
Non-Hispanic White | 3,707 (67.0) | 253,442 (75.0) |
Black | 806 (14.6) | 30,780 (9.1) |
Hispanic White | 704 (12.7) | 29,111 (8.6) |
Asian/Pacific Islander | 280 (5.1) | 23,252 (6.9) |
American Indian/Alaska Native | 39 (0.7) | 1,506 (0.4) |
Residence at Diagnosisa | ||
Metro County | 4,966 (89.9) | 304,346 (90.1) |
Non-Metro County | 559 (10.1) | 33,362 (9.9) |
County-Level % Below Poverty | ||
<10.00% | 1,917 (34.6) | 134,331 (39.7) |
10.00–19.99% | 3,122 (56.4) | 179,577 (53.1) |
≥20.00% | 497 (9.0) | 24,183 (7.2) |
County-Level % Less than High School Graduate |
||
≤15.99% | 1,669 (30.2) | 118,092 (34.9) |
16–20.80% | 1,699 (30.7) | 107,849 (31.9) |
20.81–28.76% | 855 (15.4) | 45,123 (13.4) |
>28.76% | 1,313 (23.7) | 67,027 (19.8) |
Poverty-High School Index: | ||
High SEP | 1,250 (22.6) | 92,447 (27.3) |
Middle SEP | 2,845 (51.4) | 171,883 (50.8) |
Low SEP | 1,441 (26.0) | 73,761 (21.8) |
394 cases missing RUCC – Note: no other cases missing
Table 2 provides the age-adjusted IRs (per 100,000) and IR ratios for IBC according to SEP characteristics and race/ethnicity and Table 3 the corresponding results for non-IBC. There was no difference in the overall IBC IRs for metro vs. non-metro residence. When these rates were stratified by race/ethnicity, the only significant difference found was for Blacks (metro IR=3.4, non-metro IR=4.8). The IBC IRs for counties where >10% of the residents were below the poverty level were generally significantly higher for all races combined, NH Whites, and Blacks. No significant difference was seen in IBC poverty stratified rates for Hispanic Whites, API, or AI/AN. The IBC IRs for counties where a higher percentage of the population had not graduated from high school followed a similar pattern as the poverty variable.
Table 2.
Variable | All Races Combined | Non-Hispanic White | Black | |||
---|---|---|---|---|---|---|
Incidence Rate | Rate Ratio | Incidence Rate | Rate Ratio | Incidence Rate | Rate Ratio | |
Metro | 2.3 (2.3–2.4) | 2.3 (2.2–2.4) | 3.4 (3.2–3.7) | |||
Non-Metro | 2.3 (2.1–2.5) | 0.99 (0.90–1.08) | 2.2 (2.0–2.4) | 0.95 (0.85–1.05) | 4.8a (3.7–6.1) | 1.40 (1.06–1.81) |
County-Level % Below Poverty | ||||||
<10.00%d | 2.2 (2.1–2.3) | 1.00 | 2.2 (2.1–2.3) | 1.00 | 2.8 (2.3–3.3) | 1.00 |
10.00–19.99% | 2.4b (2.3–2.5) | 1.12 (1.06–1.19) | 2.4b (2.3–2.5) | 1.07 (1.00–1.15) | 3.7b (3.4–4.0) | 1.31 (1.08–1.60) |
≥20.00% | 2.7b (2.4–2.9) | 1.24 (1.12–1.37) | 2.5 (2.2–2.8) | 1.12 (0.99–1.27) | 3.7b (3.0–4.5) | 1.32 (1.01–1.72) |
County-Level % Less than High School Graduate |
||||||
≤15.99%d | 2.1 (2.0–2.2) | 1.00 | 2.1 (2.0–2.2) | 1.00 | 3.2 (2.8–3.8) | 1.00 |
16–20.80% | 2.3c (2.2–2.4) | 1.09 (1.02–1.17) | 2.4c (2.2–2.5) | 1.11 (1.03–1.21) | 3.4 (2.9–3.9) | 1.04 (0.84–1.28) |
20.81–28.76% | 2.6c (2.4–2.8) | 1.23 (1.13–1.33) | 2.5c (2.3–2.7) | 1.18 (1.06–1.31) | 3.3 (2.9–3.8) | 1.03 (0.84–1.27) |
>28.76% | 2.6c (2.4–2.7) | 1.20 (1.12–1.30) | 2.5c (2.4–2.7) | 1.20 (1.09–1.32) | 4.1c (3.6–4.8) | 1.28 (1.04–1.58) |
Variable | Hispanic White | Asian/Pacific Islander | American Indian/Alaska Native | |||
---|---|---|---|---|---|---|
Incidence Rate | Rate Ratio | Incidence Rate | Rate Ratio | Incidence Rate | Rate Ratio(95%CI) | |
Metro | 2.4 (2.2–2.6) | 1.3 (1.1–1.4) | 1.2 (0.7–1.9) | |||
Non-Metro | 2.6 (1.7–3.7) | 1.05 (0.69–1.55) | 1.4 (0.7–2.4) | 1.08 (0.55–1.92) | 1.6 (0.6–3.5) | 1.35 (0.43–3.47) |
County-Level % Below Poverty | ||||||
<10.00%d | 2.3 (1.9–2.8) | 1.00 | 1.2 (1.0–1.5) | 1.00 | 2.5 (1.4–4.0) | 1.00 |
10.00–19.99% | 2.4 (2.2–2.6) | 1.05 (0.85–1.30) | 1.3 (1.1–1.5) | 1.08 (0.84–1.39) | 1.2 (0.6–2.1) | 0.49 (0.22–1.08) |
≥20.00% | 2.7 (2.1–3.4) | 1.18 (0.87–1.60) | 0.8 (0.2–2.0) | 0.62 (0.16–1.68) | 1.3 (0.5–2.8) | 0.55 (0.19–1.41) |
County–Level % Less than High School Graduate |
||||||
≤15.99%d | 2.0 (1.6–2.5) | 1.00 | 1.2 (1.0–1.5) | 1.00 | 2.3 (1.4–3.7) | 1.00 |
16–20.80% | 2.5 (2.2–2.9) | 1.25 (0.96–1.64) | 1.2 (1.0–1.4) | 0.95 (0.70–1.28) | 1.3 (0.4–2.9) | 0.55 (0.16–1.46) |
20.81–28.76% | 2.4 (1.9–2.9) | 1.17 (0.86–1.59) | 1.6 (0.9–2.6) | 1.29 (0.71–2.22) | 0.9 (0.2–2.4) | 0.40 (0.10–1.23) |
>28.76% | 2.5 (2.2–2.8) | 1.24 (0.97–1.60) | 1.4 (1.1–1.8) | 1.13(0.82–1.56) | 1.3 (0.6–2.6) | 0.57 (0.22–1.35) |
Non-Metro rate significantly different from Metro rate (p<0.05)
Rate significantly different from <10.00% rate (p<0.05)
Rate significantly different from ≤15.99% rate (p<0.05)
Referent Category for Rate Ratios
Table 3.
Variable | All Races Combined | Non-Hispanic White | Black | |||
---|---|---|---|---|---|---|
Incidence Rate | Rate Ratio | Incidence Rate | Rate Ratio | Incidence Rate | Rate Ratio | |
Metro | 144.4 (143.9–144.9) | 158.7 (158.0–159.3) | 136.3 (134.7–137.9) | |||
Non-Metro | 135.5a (134.0–137.0) | 0.94 (0.93–0.95) | 139.1a (137.5–140.7) | 0.88 (0.87–0.89) | 131.6 (125.4–138.0) | (0.92–1.01) |
County–Level % Below Poverty | ||||||
<10.00%d | 151.9 (151.0–152.7) | 1.00 | 159.5 (158.6–160.5) | 1.00 | 135.1 (131.8–138.5) | 1.00 |
10.00–19.99% | 139.8b(139.2–140.5) | 0.92 (0.91–0.93) | 155.7b(154.8–56.6) | 0.98 (0.97–0.98) | 136.0 (134.1–137.9) | 1.01 (0.98–1.04) |
≥20.00% | 129.9b (128.2–131.5) | 0.86 (0.84–0.87) | 139.3b (137.2–141.5) | 0.87 (0.86–0.89) | 137.4 (133.1–141.9) | 1.02(0.98–1.06) |
County-Level % Less than | ||||||
High School Graduate | ||||||
≤15.99%d | 151.2 (150.4–152.1) | 1.00 | 158.6 (157.6–159.6) | 1.00 | 135.9 (132.7–139.2) | 1.00 |
16–20.80% | 146.7c (145.8–147.6) | 0.97 (0.96–0.98) | 158.8 (157.7–159.9) | 1.00 (0.99–1.01) | 137.6 (134.5–140.8) | 1.01 (0.98–1.05) |
20.81–28.76% | 136.7c (135.5–138.0) | 0.90 (0.89–0.91) | 147.1c (145.4–148.7) | 0.93 (0.92–0.94) | 134.8 (132.0–137.7) | 0.99 (0.96–1.02) |
>28.76% | 131.6c (130.6–132.6) | 0.87 (0.86–0.88) | 152.7c (151.2–154.2) | 0.96 (0.95–0.97) | 135.8 (132.5–139.1) | 1.00 (0.97–1.03) |
Variable | Hispanic white | Asian/pasific Islander | American Indian/Alaska Native | |||
---|---|---|---|---|---|---|
Incidence Rate | Rate Ratio | Incidence Rate | Rate Ratio | Incidence Rate | Rate Ratio | |
Metro | 105.0(103.7–106.3) | 108.1(106.6–109.5) | 50.3(46.8–53.9) | |||
Non-Metro | 103.6(97.5–110.0) | 0.99(0.93–1.05) | 117.3a (110.3–124.7) | 1.09(1.02–1.16) | 70.3a (62.2–79.3) | 1.40(1.21–1.61) |
County–Level % Below Poverty | ||||||
<10.00%d | 122.6(119.4–125.8) | 1.00 | 115.1(112.8–117.4) | 1.00 | 106.4 (98.2–115.0) | 1.00 |
10.00–19.99% | 101.5b (100.1–103.0) | 0.83 (0.80–0.85) | 105.4b (103.6–107.2) | 0.92 (0.89–0.94) | 47.3b (43.3–51.6) | 0.44 (0.40–0.50) |
≥20.00% | 98.2b (94.5–102.0) | 0.80 (0.77–0.84) | 79.0b (71.0–87.6) | 0.69 (0.62–0.76) | 52.8b (46.4–59.8) | 0.50 (0.43–0.58) |
County–Level % Less than High School Graduate |
||||||
≤15.99%d | 121.2 (117.7–124.7) | 1.00 | 116.9 (114.3–119.5) | 1.00 | 115.3 (107.4–123.7) | 1.00 |
16–20.80% | 113.2c (110.7–115.8) | 0.93 (0.90–0.97) | 104.5c (102.2–106.8) | 0.89 (0.87–0.92) | 46.7c (40.7–53.2) | 0.40 (0.35–0.47) |
20.81–28.76% | 101.2c (98.0–104.5) | 0.84(0.80–0.87) | 89.3c(83.5–95.3) | 0.76 (0.71–0.82) | 45.3c (38.9–52.5) | 0.39 (0.33–0.46) |
>28.76% | 96.0c (94.2–97.8) | 0.79 (0.77–0.82) | 107.4c (104.7–110.2) | 0.92 (0.89–0.95) | 32.6c (28.2–37.6) | 0.28 (0.24–0.33) |
Non-Metro rate significantly different from Metro rate (p<0.05)
Rate significantly different from <10.00% rate (p<0.05)
Rate significantly different from ≤15.99% rate (p<0.05)
Referent Category for Rate Ratios
Non-IBC IRs for those living in non-metro counties were significantly lower for all races combined and NH Whites, but significantly higher for API and AI/AN race/ethnicity groups. There was no significant difference in the metro and non-metro IRs for Blacks and Hispanic Whites. Non-IBC IRs for those living in counties where >10% of the residents were below the poverty level were significantly lower for all races combined, NH Whites, Hispanic Whites, API, and AI/AN, with no significant difference in rates observed for Blacks. The non-IBC IRs for counties where a higher percentage of the population had not graduated from high school followed a similar pattern as the poverty variable.
Table 4 gives the ORs and 95% confidence intervals (CIs) for the four HLMs. In these models, which account for the clustering of cases within counties, after adjustment for age at diagnosis and race/ethnicity, residing in a high poverty county (≥ 20%), a county in the highest quartile of persons with less than a high school degree, and a low SEP county based on the poverty-high school index was significantly associated with higher odds of IBC (OR (95% CI) = 1.25 (1.09–1.43), 1.25 (1.10–1.42), and 1.26 (1.11–1.44), respectively). Younger age and Black, Hispanic White, and AI/AN race/ethnicity were also significantly associated with higher odds of IBC in all models, while API race/ethnicity was significantly associated with lower odds of IBC.
Table 4.
Model 1 | Model 2 | Model 3 | Model 4 | ||||
---|---|---|---|---|---|---|---|
Variable | OR (95% CI) | Variable | OR (95% CI) | Variable | OR (95% CI) | Variable | OR (95% CI) |
County-Level % Less than High School |
|||||||
County-Level % | Graduate | Poverty-High School | |||||
Below Poverty | ≤15.99%b | 1.00 | Index: | ||||
<10.00%b | 1.00 | 16–20.80% | 1.04 (0.93–1.17) | High SEPb | 1.00 | ||
Non-Metrob | 1.00 | 10.00–19.99% | 1.06 (0.96–1.18) | 20.81–28.76% | 1.11 (0.97–1.28) | Middle SEP | 1.10 (0.98–1.22) |
Metro | 0.90 (0.81–1.00) | ≥20.00% | 1.25 (1.09–1.43) | >28.76% | 1.25 (1.10–1.42) | Low SEP | 1.26 (1.11–1.44) |
Age at Diagnosis | 0.99 (0.99–0.99) | Age at Diagnosis | 0.99 (0.99–0.99) | Age at Diagnosis | 0.99 (0.99–0.99) | Age at Diagnosis | 0.99 (0.99–0.99) |
Race/Ethnicity | Race/Ethnicity | Race/Ethnicity | Race/Ethnicity | ||||
NHWhiteb | 1.00 | NHWhiteb | 1.00 | NHWhiteb | 1.00 | NHWhiteb | 1.00 |
Black | 1.73 (1.59–1.87) | Black | 1.70 (1.56–1.84) | Black | 1.70 (1.57–1.85) | Black | 1.70 (1.56–1.84) |
Hispanic White | 1.46 (1.34–1.59) | Hispanic White | 1.44 (1.32–1.57) | Hispanic White | 1.44 (1.32–1.57) | Hispanic White | 1.44 (1.32–1.57) |
API | 0.79 (0.69–0.89) | API | 0.78 (0.69–0.89) | API | 0.78 (0.69–0.89) | API | 0.78 (0.69–0.89) |
AI/AN | 1.55 (1.06–2.26) | AI/AN | 1.59 (1.14–2.22) | AI/AN | 1.62 (1.16–2.26) | AI/AN | 1.61 (1.15–2.25) |
All variables in each column are mutually adjusted for each other
Referent Category
Discussion
Contrary to the previously reported positive association between urban residence, SEP, and overall breast cancer occurrence (16–28), this study found that living in a high poverty county (≥ 20%), a county with a high percentage of less than high school graduates, and residing in a low SEP county as defined by the poverty-high school index were significantly associated with IBC, even after adjustment for age at diagnosis and race/ethnicity.
Prior studies examining overall breast cancer occurrence have found it to be associated with urban residence and higher SEP (16–28). However, the majority of breast cancers are non-IBC, and thus determining IBC’s association with SEP based on studies of all breast cancer types is difficult. This study used an inclusive definition to separately characterize IBC from all other breast cancer types, and then directly compared IBC to non-IBC through use of HLMs adjusting for age and race/ethnicity, in order to specifically examine SEP and its association with IBC as a distinct breast cancer entity.
Residing in a county with a large percent of persons below the poverty level, less than high school graduates, and in the low SEP group of the poverty-high school index were all associated with IBC in this analysis, suggesting that poverty and education are capturing similar aspects of SEP that affect IBC incidence. As SEP measures, poverty and education both act as a summary measure of a county’s SEP and can be compared over time and across US geographic areas (50). While poverty and education are correlated, as each has been shown to be related to overall breast cancer incidence, they were included as separate SEP measures as well as in a combined index in this analysis (16–23, 67). Metro vs. non-metro area of residence at diagnosis captures various characteristics that can be directly and/or indirectly related to an individual’s health, such as population density, geographic isolation, exposure to agriculture, industrial or commercial complexes, and proximity and access to health care services (68, 69). Furthermore, there are many ways to classify counties based on characteristics such as administrative units, land-use, and economic concepts (68). It is possible the definition of metro vs. non-metro used in this analysis does not capture specific factors that may be related to IBC, and thus explain the lack of association seen.
Although risk factors for IBC remain largely unknown, some studies have shown different risk factor profiles for IBC patients as compared to non-IBC patients. Chang et al. found that high BMI was significantly associated with increased risk of IBC, regardless of menopausal status (12). This is in contrast to overall breast cancer, where higher premenopausal weight has been shown to reduce risk (70). Chang et al. also found IBC patients were more likely to be premenopausal and have younger age at menarche and first birth as compared to non-IBC and non-breast cancer patients (12).
A study conducted in France by Le et al. found that IBC patients had a lower educational level, a higher BMI, a longer cumulative duration of breastfeeding, and included a greater proportion of non-European women as compared to non-IBC patients (13). A recent study of Egyptian breast cancer cases found IBC patients had significantly lower parity than non-IBC patients (14). Furthermore, a 2010 study based in Tunisia reported a rural predominance of IBC among the cases studied, and hypothesized the reduction in IBC seen in that country was due in part to increasing SEP (71).
In this analysis, IBC was associated with younger age and Black race/ethnicity, while API race/ethnicity was associated with lower odds of IBC, as found in previous studies (1, 6, 7, 15, 33, 64–66). White Hispanic and AI/AN race/ethnicity were also found to be significantly associated with IBC. A previous study showed no difference in the age-adjusted IBC IR between Hispanic and non-Hispanic women for cases diagnosed from 1994–1998 reported to the North American Association of Central Cancer Registries (64). However, this study did not classify Hispanic origin as mutually exclusive from other race/ethnicities, and used the more restrictive ICD-O-3 8530 code to define IBC (64). Younger age at IBC diagnosis has been reported for AI/AN women as compared to White women (6, 64), although no studies which directly compared the IRs or proportion of IBC between AI/AN women and other race/ethnicities were located.
The strengths of this study include the use of the US SEER database, 5 mutually exclusive race/ethnicity categories, a comprehensive definition of IBC, and a hierarchical modeling structure. The SEER program is considered the standard for cancer registry data quality worldwide (72). Quality control studies, including case-finding, recoding, and reliability studies are continually conducted by the SEER program to ensure data included in the registries are accurate and collected and recorded in a uniform and timely manner across all registries (72). As IBC is a relatively rare diagnosis, the US SEER database, which covers 26% of the US population residing in varying regions and geographic areas with over-representation of minority groups, allows for the stratification of IBC incidence by SEP and race/ethnicity categories (32).
Previous studies have been limited to reporting IBC rates and proportions for a limited number of race/ethnicity categories, usually for White, Black, & Other, due to small numbers of cases as well as the manner in which this data was recorded by SEER (1, 7, 15, 33, 73). Beginning with the November 2005 SEER data submission, the algorithms for creating the race recode variables within the SEER database were revised, allowing for the examination of incidence for four race categories: White, Black, AI/AN, and API, as well as Hispanic ethnicity (73). The race and Hispanic ethnicity data can also now be merged in order to create mutually exclusive race/ethnicity categories (73). This allowed for the current analysis to report results for 5 mutually exclusive race/ethnicity categories, as opposed to the more limited race/ethnicity analyses in previous IBC studies.
IBC studies have been hampered by lack of a standard case definition (6, 11, 74). Previous studies have used the ICD-O 8530 designation to define IBC (7, 64, 65, 75, 76). ICD-O code 8530 is a pathologic designation requiring plugging of the dermal lymphatics with tumor emboli and does not consider clinical skin changes (6). However, this conservative IBC definition is not consistent with the current AJCC staging manual guidelines, and may underestimate the true incidence of IBC (6, 7, 64, 77). SEER EOD codes are based on a combined clinical and operative/pathologic assessment abstracted from the pathology report, and allow for identification of IBC cases that do not have ICD-O 8530 as the pathologic diagnosis for years prior to 2004 (7). From 2004 forward, SEER includes a variable with derived AJCC staging, which allows for the identification of IBC cases defined as the primary tumor designation of “T4d” (35, 36). The comprehensive IBC definition used in this paper is similar to that used in recent IBC studies (6, 15, 33–35). Using this definition should help ensure less misclassification of IBC cases to the non-IBC group. Finally, use of a hierarchical modeling structure allows for the calculation of more accurate standard errors, thus adding confidence to any significant results found (61–63).
A few limitations should be noted when interpreting the results of this analysis. Though SEER data are broadly representative of the US population, cases recorded in the SEER database are more likely to be foreign born and urban as compared to the US population as measured in the 2000 census (78). There are also a relatively small number of AI/AN IBC cases available for this analysis (n=39), which is reflected in the wide 95% CIs around the IR and OR estimates for this race/ethnicity category. However, due to the US SEER’s large size and over 30 years of follow-up, it is generally considered to accurately represent the overall US cancer population (78).
Another limitation is the lack of individual-level SEP information in the US SEER database and the inherent ecologic bias in interpreting the results of this analysis at the individual-level. Any associations seen between county-level SEP and IBC occurrence may not necessarily hold were individual-level SEP available and used in the analysis (79). Therefore, the results of this analysis are better interpreted at the contextual level, i.e., the effect being measured is that of residing in a county with a particular SEP characteristic, not that of the breast cancer cases’ individual SEP. However, a study comparing census-level SEP measures to individual-level measures found they were similarly associated with individual-level health outcomes (52). Furthermore, the US SEER database linked to US census data provides a unique opportunity to conduct analyses stratified by race/ethnicity and county-level SEP measures on a relatively large number of IBC and non-IBC cases.
Overall breast cancer has been found to be positively associated with SEP, whereas in this analysis IBC was associated with decreasing SEP. One explanation for these results is that women of lower SEP have less access to health care that would lead to early detection and the resultant neglected breast cancer develops into IBC. Some earlier IBC work suggested that it may be a subtype of locally advanced breast cancer rather than a distinct entity (80). However, the majority of recent studies on the epidemiology, clinical and prognostic characteristics, biology, and molecular genetics of IBC suggest is it likely a distinct biologic entity from other breast cancer (10, 12–15, 35, 81–84). Another explanation is that breast cancers occurring in women of lower SEP presenting with skin involvement are misdiagnosed as IBC (85, 86). However, there is little literature, especially in the US, suggesting women of lower SEP are at higher risk for IBC, so it is unlikely clinicians would be more likely to look for and diagnose (or misdiagnose) IBC disproportionately in women of lower SEP.
These results are in keeping with a growing amount of evidence showing IBC likely has a different risk factor profile than other breast cancers and is a distinct biologic entity (10, 12–15, 35, 81–84). Few studies have examined the epidemiology of rarer forms of breast cancer such as IBC, though studies that have suggest the general breast cancer risk profile may not hold for rarer breast cancer subtypes (1, 65, 87). Further investigation into the etiology of IBC is needed in order to elucidate risk factors for the disease that would help guide prevention and screening programs, especially studies which examine individual and community-level associations between multiple SEP measures and IBC incidence. However, these results also indicate studies designed to investigate why the disparity of higher incidence of IBC in lower SEP groups and racial/ethnic minorities is observed, as well as potential interventions to eliminate these differences, are called for. Furthermore, because treatment is especially urgent in IBC, design and implementation of strategies that would promote earlier IBC diagnosis among lower SEP groups and racial/ethnic minorities, which traditionally experience less access to early detection programs, would likely have a direct and favorable impact on their prognosis.
Acknowledgements
The authors thank Dr. William Anderson at the Division of Cancer Epidemiology and Genetics, National Cancer Institute, for providing expertise on IBC and the SEER database, as well as the merged race-ethnicity SEER*stat variable used in this analysis.
Financial Support:This work was supported by a Rackham Merit Fellowship (JAS) from the University of Michigan and the Cancer Epidemiology Education in Special Populations Program of the University of Michigan (CA R25 112383). Additional funding was received from the Avon Foundation (AS and SDM), the Breast Cancer Research Foundation (SDM), and the Debbie Strange-Browne Inflammatory Breast Cancer Foundation (SDM).
Footnotes
The authors have no potential conflicts of interest to disclose.
References
- 1.Anderson WF, Chu KC, Chang S, Sherman ME. Comparison of age-specific incidence rate patterns for different histopathologic types of breast carcinoma. Cancer Epidemiol Biomarkers Prev. 2004 Jul;13(7):1128–1135. [PubMed] [Google Scholar]
- 2.Li CI, Anderson BO, Daling JR, Moe RE. Trends in incidence rates of invasive lobular and ductal breast carcinoma. JAMA. 2003 Mar 19;289(11):1421–1424. doi: 10.1001/jama.289.11.1421. [DOI] [PubMed] [Google Scholar]
- 3.Li CI, Daling JR, Malone KE. Incidence of invasive breast cancer by hormone receptor status from 1992 to 1998. J Clin Oncol. 2003 Jan 1;21(1):28–34. doi: 10.1200/JCO.2003.03.088. [DOI] [PubMed] [Google Scholar]
- 4.Li CI, Malone KE, Daling JR. Differences in breast cancer hormone receptor status and histology by race and ethnicity among women 50 years of age and older. Cancer Epidemiol Biomarkers Prev. 2002 Jul;11(7):601–607. [PubMed] [Google Scholar]
- 5.Sorlie T. Molecular portraits of breast cancer: Tumour subtypes as distinct disease entities. Eur J Cancer. 2004 Dec;40(18):2667–2675. doi: 10.1016/j.ejca.2004.08.021. [DOI] [PubMed] [Google Scholar]
- 6.Anderson WF, Schairer C, Chen BE, Hance KW, Levine PH. Epidemiology of inflammatory breast cancer (IBC) Breast Dis. 22:9–23. doi: 10.3233/bd-2006-22103. 2005–2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Chang S, Parker SL, Pham T, Buzdar AU, Hursting SD. Inflammatory breast carcinoma incidence and survival: The Surveillance, Epidemiology, and End Results program of the National Cancer Institute, 1975–1992. Cancer. 1998 Jun 15;82(12):2366–2372. [PubMed] [Google Scholar]
- 8.Dawood S, Cristofanilli M. Inflammatory breast cancer: A complex disease. CancerNetwork: Oncology. 2008 Dec 1;22(14) April 21, 2010. [Google Scholar]
- 9.Gonzalez-Angulo AM, Hennessy BT, Broglio K, Meric-Bernstam F, Cristofanilli M, Giordano SH, et al. Trends for inflammatory breast cancer: Is survival improving? Oncologist. 2007 Aug;12(8):904–912. doi: 10.1634/theoncologist.12-8-904. [DOI] [PubMed] [Google Scholar]
- 10.Houchens NW, Merajver SD. Molecular determinants of the inflammatory breast cancer phenotype. CancerNetwork: Oncology. 2008 Dec 1;22(14) April 21, 2010. [PubMed] [Google Scholar]
- 11.Levine PH, Veneroso C. The epidemiology of inflammatory breast cancer. Semin Oncol. 2008 Feb;35(1):11–16. doi: 10.1053/j.seminoncol.2007.11.018. [DOI] [PubMed] [Google Scholar]
- 12.Chang S, Buzdar AU, Hursting SD. Inflammatory breast cancer and body mass index. J Clin Oncol. 1998 Dec;16(12):3731–3735. doi: 10.1200/JCO.1998.16.12.3731. [DOI] [PubMed] [Google Scholar]
- 13.Le MG, Arriagada R, Bahi J, Pfeiffer F, Cammoun M, Tabbane F, et al. Are risk factors for breast cancer similar in women with inflammatory breast cancer and in those with non-inflammatory breast cancer? Breast. 2006 Jun;15(3):355–362. doi: 10.1016/j.breast.2005.08.018. [DOI] [PubMed] [Google Scholar]
- 14.Lo AC, Georgopoulos A, Kleer CG, Banerjee M, Omar S, Khaled H, et al. Analysis of RhoC expression and lymphovascular emboli in inflammatory vs non-inflammatory breast cancers in Egyptian patients. Breast. 2009 Feb;18(1):55–59. doi: 10.1016/j.breast.2008.11.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Anderson WF, Chu KC, Chang S. Inflammatory breast carcinoma and noninflammatory locally advanced breast carcinoma: Distinct clinicopathologic entities? J Clin Oncol. 2003 Jun 15;21(12):2254–2259. doi: 10.1200/JCO.2003.07.082. [DOI] [PubMed] [Google Scholar]
- 16.Robert SA, Strombom I, Trentham-Dietz A, Hampton JM, McElroy JA, Newcomb PA, et al. Socioeconomic risk factors for breast cancer: Distinguishing individual- and community-level effects. Epidemiology. 2004 Jul;15(4):442–450. doi: 10.1097/01.ede.0000129512.61698.03. [DOI] [PubMed] [Google Scholar]
- 17.Heck KE, Pamuk ER. Explaining the relation between education and postmenopausal breast cancer. Am J Epidemiol. 1997 Feb 15;145(4):366–372. doi: 10.1093/oxfordjournals.aje.a009114. [DOI] [PubMed] [Google Scholar]
- 18.Kelsey JL, Bernstein L. Epidemiology and prevention of breast cancer. Annu Rev Public Health. 1996;17:47–67. doi: 10.1146/annurev.pu.17.050196.000403. [DOI] [PubMed] [Google Scholar]
- 19.Devesa SS, Diamond EL. Association of breast cancer and cervical cancer incidence with income and education among whites and blacks. J Natl Cancer Inst. 1980 Sep;65(3):515–528. [PubMed] [Google Scholar]
- 20.Krieger N. Social class and the black/white crossover in the age-specific incidence of breast cancer: A study linking census-derived data to population-based registry records. Am J Epidemiol. 1990 May;131(5):804–814. doi: 10.1093/oxfordjournals.aje.a115571. [DOI] [PubMed] [Google Scholar]
- 21.Liu L, Deapen D, Bernstein L. Socioeconomic status and cancers of the female breast and reproductive organs: A comparison across racial/ethnic populations in Los Angeles county, California (United States) Cancer Causes Control. 1998 Aug;9(4):369–80. doi: 10.1023/a:1008811432436. [DOI] [PubMed] [Google Scholar]
- 22.Prehn AW, West DW. Evaluating local differences in breast cancer incidence rates: A census-based methodology (United States) Cancer Causes Control. 1998 Oct;9(5):511–7. doi: 10.1023/a:1008809819218. [DOI] [PubMed] [Google Scholar]
- 23.Yost K, Perkins C, Cohen R, Morris C, Wright W. Socioeconomic status and breast cancer incidence in California for different race/ethnic groups. Cancer Causes Control. 2001 Oct;12(8):703–11. doi: 10.1023/a:1011240019516. [DOI] [PubMed] [Google Scholar]
- 24.Dey S, Soliman AS, Hablas A, Seifeldin IA, Ismail K, Ramadan M, et al. Urban-rural differences in breast cancer incidence by hormone receptor status across 6 years in Egypt. Breast Cancer Res Treat. 2009 Jun 23; doi: 10.1007/s10549-009-0427-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Doll R. Urban and rural factors in the aetiology of cancer. Int J Cancer. 1991 Apr 1;47(6):803–10. doi: 10.1002/ijc.2910470602. [DOI] [PubMed] [Google Scholar]
- 26.Hall SA, Kaufman JS, Millikan RC, Ricketts TC, Herman D, Savitz DA. Urbanization and breast cancer incidence in North Carolina, 1995–1999. Ann Epidemiol. 2005 Nov;15(10):796–803. doi: 10.1016/j.annepidem.2005.02.006. [DOI] [PubMed] [Google Scholar]
- 27.Hausauer AK, Keegan TH, Chang ET, Glaser SL, Howe H, Clarke CA. Recent trends in breast cancer incidence in US white women by county-level urban/rural and poverty status. BMC Med. 2009 Jun 26;7:31. doi: 10.1186/1741-7015-7-31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Nasca PC, Mahoney MC, Wolfgang PE. Population density and cancer incidence differentials in New York state, 1978–82. Cancer Causes Control. 1992 Jan;3(1):7–15. doi: 10.1007/BF00051906. [DOI] [PubMed] [Google Scholar]
- 29.Monroe AC, Ricketts TC, Savitz LA. Cancer in rural versus urban populations: A review. J Rural Health. 1992 Summer;8(3):212–20. doi: 10.1111/j.1748-0361.1992.tb00354.x. [DOI] [PubMed] [Google Scholar]
- 30.Surveillance, Epidemiology, and End results (SEER) program (www.seer.cancer.gov) SEER*Stat database: Incidence - SEER 17 regs research data + Hurricane Katrina impacted Louisiana cases, Nov 2009 sub (1973–2007 varying) - linked to county attributes - total U.S., 1969–2007 counties, National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2010, based on the November 2009 submission.
- 31.SEER data, 1973–2007. Bethesda, MD: National Cancer Institute; 2009. Nov, Available from: http://seer.cancer.gov/data/index.html. [Google Scholar]
- 32.Overview of the SEER program. Bethesda, MD: National Cancer Institute; Available from: http://seer.cancer.gov/about/ [Google Scholar]
- 33.Hance KW, Anderson WF, Devesa SS, Young HA, Levine PH. Trends in inflammatory breast carcinoma incidence and survival: The Surveillance, Epidemiology, and End results program at the National Cancer Institute. J Natl Cancer Inst. 2005 Jul 6;97(13):966–975. doi: 10.1093/jnci/dji172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Schairer C, Brown LM, Mai PL. Inflammatory breast cancer: High risk of contralateral breast cancer compared to comparably staged non-inflammatory breast cancer. Breast Cancer Res Treat. 2011 Mar 9; doi: 10.1007/s10549-010-1324-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Dawood S, Ueno NT, Valero V, Woodward WA, Buchholz TA, Hortobagyi GN, et al. Differences in survival among women with stage III inflammatory and noninflammatory locally advanced breast cancer appear early: A large population-based study. Cancer. 2011 May 1;117(9):1819–1826. doi: 10.1002/cncr.25682. [DOI] [PubMed] [Google Scholar]
- 36.SEER Research Record Description: Cases Diagnosed in 1973–2007, November 2009 Data Submission. Bethesda, MD: National Cancer Institute; Surveillance Epidemiology and End Results Program. Available from: http://seer.cancer.gov/data/documentation.html. [Google Scholar]
- 37.ICD-O-3 SEER Site/Histology Validation List [Updated December 4, 2009] Bethesda, MD: National Cancer Institute; SEER Program Quality Control Section. Available from: http://seer.cancer.gov/icd-o-3/sitetype.icdo3.d20091204.pdf. [Google Scholar]
- 38.Edge SB, Byrd DR, Compton CR, Fritz AG, Greene FL, Trotti A III, editors. American Joint Committee on Cancer. Breast. AJCC Cancer Staging Manual. Seventh ed. New York: Springer Science + Business Media LLC; 2010. p. 354. [Google Scholar]
- 39.Johnson CH, editor. SEER Program Coding and Staging Manual 2004, Revision 1. Bethesda, MD: National Cancer Institute; NIH Publication number 04-5581. [Google Scholar]
- 40.Fritz A, Ries L, editors. SEER Extent SEER of Disease-1988:Codes and Coding Instructions. Third edn. Bethesda, MD: U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health; Statistics Cancer Branch,Program Surveillance of Division Control Cancer Sciences Population Cancer National Institute (1998) Breast: C50.0-C50.6 C50.8-C50.9. Available from: http://www.seer.cancer.gov/manuals/EOD10Dig.pub.pdf. [Google Scholar]
- 41.County attributes. Bethesda, MD: National Cancer Institute; Surveillance, Epidemiology, and End Results Program. Available from: http://seer.cancer.gov/seerstat/variables/countyattribs/index.html. [Google Scholar]
- 42.Crowell RE, Goetz T, Wiggins C, Magana E. Regional disparities in treatment and survival of early stage non-small cell lung cancer. Ethn Dis. 2007 spring;17(2):358–364. [PubMed] [Google Scholar]
- 43.Yu XQ. Socioeconomic disparities in breast cancer survival: Relation to stage at diagnosis, treatment and race. BMC Cancer. 2009 Oct 14;9:364. doi: 10.1186/1471-2407-9-364. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Reid-Arndt SA, Cox CR. Does rurality affect quality of life following treatment for breast cancer? J Rural Health. 2010 Fall;26(4):402–405. doi: 10.1111/j.1748-0361.2010.00295.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Klein J, Ji M, Rea NK, Stoodt G. Differences in male breast cancer stage, tumor size at diagnosis, and survival rate between metropolitan and nonmetropolitan regions. Am J Mens Health. 2011 Mar 16; doi: 10.1177/1557988311400403. [DOI] [PubMed] [Google Scholar]
- 46.United States Department of Agriculture, Economic Research Service. Measuring Rurality: Rural-Urban Continuum Codes. Available from: http://www.ers.usda.gov/briefing/rurality/ruralurbcon/
- 47.Rural-urban continuum codes. Bethesda, MD: National Cancer Institute; Surveillance, Epidemiology, and End Results Program. Available from: http://seer.cancer.gov/seerstat/variables/countyattribs/ruralurban.html. [Google Scholar]
- 48.Singh GK, Miller BA, Hankey BF, Edwards BK. Cancer Incidence, Mortality, Stage, Treatment, and Survival, 1975–1999. NCI Cancer Surveillance Monograph Series, Number 4. Bethesda, MD: National Cancer Institute; 2003. Area Socioeconomic Variations in U.S. NIH Publication No. 03-5417. [Google Scholar]
- 49.Ward E, Jemal A, Cokkinides V, Singh GK, Cardinez C, Ghafoor A, et al. Cancer disparities by race/ethnicity and socioeconomic status. CA Cancer J Clin. 2004 Mar-Apr;54(2):78–93. doi: 10.3322/canjclin.54.2.78. [DOI] [PubMed] [Google Scholar]
- 50.Krieger N, Chen JT, Waterman PD, Soobader MJ, Subramanian SV, Carson R. Geocoding and monitoring of US socioeconomic inequalities in mortality and cancer incidence: Does the choice of area-based measure and geographic level matter?: The public health disparities geocoding project. Am J Epidemiol. 2002 Sep 1;156(5):471–482. doi: 10.1093/aje/kwf068. [DOI] [PubMed] [Google Scholar]
- 51.Bishaw A. Areas With Concentrated Poverty: 1999. Census 2000 Special Reports CENSR-16, July 2005. U.S. Department of Commerce, Economics and Statistics Administration, U.S. Census Bureau. Available from: http://www.census.gov/prod/2005pubs/censr-16.pdf.
- 52.Krieger N. Overcoming the absence of socioeconomic data in medical records: Validation and application of a census-based methodology. Am J Public Health. 1992 May;82(5):703–710. doi: 10.2105/ajph.82.5.703. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Carter CL, Jones DY, Schatzkin A, Brinton LA. A prospective study of reproductive, familial and socioeconomic risk factors for breast cancer using NHANES I data. Public Health Rep. 1989 Jan-Feb;104(1):45–50. [PMC free article] [PubMed] [Google Scholar]
- 54.Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat database: County attributes - total U.S., 1969–2007 counties (www.seer.cancer.gov/seerstat/variables/countyattribs), National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch.
- 55.Surveillance, Epidemiology, and End Results (SEER) program (www.seer.cancer.gov) SEER*Stat database: Incidence - SEER 17 regs research data + Hurricane Katrina impacted Louisiana cases, Nov 2009 sub (2000–2007) <Single ages to 85+, Katrina/Rita population adjustment> - linked to county attributes - total U.S., 1969–2007 counties, National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2010, based on the November 2009 submission.
- 56.SEER*Stat tutorials: Calculating age-adjusted rates. Bethesda, MD: National Cancer Institute; Surveillance, Epidemiology, and End Results Program. Available from: http://www.seer.cancer.gov/seerstat/tutorials/aarates/definition.html. [Google Scholar]
- 57.Standard populations (millions) for age-adjustment. Bethesda, MD: National Cancer Institute; Surveillance, Epidemiology, and End Results Program. Available from: http://seer.cancer.gov/stdpopulations/index.html. [Google Scholar]
- 58.SEER*stat help: P-values for frequency or rate sessions. Bethesda, MD: National Cancer Institute; Surveillance, Epidemiology, and End Results Program. Available from: http://seer.cancer.gov/seerstat/WebHelp/seerstat.htm. [Google Scholar]
- 59.Tiwari RC, Clegg LX, Zou Z. Efficient interval estimation for age-adjusted cancer rates. Stat Methods Med Res. 2006 Dec;15(6):547–569. doi: 10.1177/0962280206070621. [DOI] [PubMed] [Google Scholar]
- 60.SAS Support (support.sas.com) The Glimmix Procedure [Documentation] 2006 Jun; Available from: http://support.sas.com/rnd/app/papers/glimmix.pdf.
- 61.Blakely TA, Woodward AJ. Ecological effects in multi-level studies. J Epidemiol Community Health. 2000 May;54(5):367–374. doi: 10.1136/jech.54.5.367. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Li J, Alterman T, Deddens JA. Analysis of Large Hierarchical Data with Multilevel Logistic Modeling Using PROC GLIMMIX; SAS Users Group International 31 Proceedings Paper 151-31; March 26–29, 2006; San Francisco, CA. Available from: http://www2.sas.com/proceedings/sugi31/151-31.pdf. [Google Scholar]
- 63.Dai J, Li D, Rocke D. Hierarchical Logistic Regression Modeling with SAS GLIMMIX; Irvine, CA. SAS Conference Proceedings: Western Users of SAS Software 2006; September 27–29, 2006; Available from: http://www.lexjansen.com/wuss/2006/Analytics/ANL-Dai.pdf. [Google Scholar]
- 64.Wingo PA, Jamison PM, Young JL, Gargiullo P. Population-based statistics for women diagnosed with inflammatory breast cancer (United States) Cancer Causes Control. 2004 Apr;15(3):321–328. doi: 10.1023/B:CACO.0000024222.61114.18. [DOI] [PubMed] [Google Scholar]
- 65.Anderson WF, Pfeiffer RM, Dores GM, Sherman ME. Comparison of age distribution patterns for different histopathologic types of breast carcinoma. Cancer Epidemiol Biomarkers Prev. 2006 Oct;15(10):1899–1905. doi: 10.1158/1055-9965.EPI-06-0191. [DOI] [PubMed] [Google Scholar]
- 66.Fong M, Henson DE, Devesa SS, Anderson WF. Inter- and intra-ethnic differences for female breast carcinoma incidence in the continental United States and in the state of Hawaii. Breast Cancer Res Treat. 2006 May;97(1):57–65. doi: 10.1007/s10549-005-9088-5. [DOI] [PubMed] [Google Scholar]
- 67.U.S. Census Bureau, Current Population Survey, 2010 Annual Social and Economic Supplement. Table Pov29: Years of school completed by poverty status, sex, age, nativity and citizenship: 2009. Available from: http://www.census.gov/hhes/www/cpstables/032010/pov/new29_100_01.htm.
- 68.Cromartie J, Bucholtz S. Defining the “rural” in rural america. Amber Waves, US Department of Agriculture, Economic Research Service. 2008 Jun;6(3) Available from: http://www.ers.usda.gov/AmberWaves/June08/Features/RuralAmerica.htm. [Google Scholar]
- 69.Jones CA, Parker TS, Ahearn M, Mishra AK, Variyam JN. Health status and health care access of farm and rural populations, EIB-57, U.S. Department of Agriculture, Economic Research Service. 2009 Aug [Google Scholar]
- 70.Swanson CA, Brinton LA, Taylor PR, Licitra LM, Ziegler RG, Schairer C. Body size and breast cancer risk assessed in women participating in the breast cancer detection demonstration project. Am J Epidemiol. 1989 Dec;130(6):1133–1141. doi: 10.1093/oxfordjournals.aje.a115440. [DOI] [PubMed] [Google Scholar]
- 71.Boussen H, Bouzaiene H, Ben Hassouna J, Dhiab T, Khomsi F, Benna F, et al. Inflammatory breast cancer in Tunisia: Epidemiological and clinical trends. Cancer. 2010 Jun 1;116(11 Suppl):2730–5. doi: 10.1002/cncr.25175. [DOI] [PubMed] [Google Scholar]
- 72.hSEER: Surveillance, Epidemiology, and End Results program [September 2005 Brochure] Bethesda, MD: National Cancer Institute; Available from: http://seer.cancer.gov/about/SEER_brochure.pdf. [Google Scholar]
- 73.Race recode changes. Bethesda, MD: National Cancer Institute; Surveillance, Epidemiology, and End Results program. Available from: http://www.seer.cancer.gov/seerstat/variables/seer/race_ethnicity/ [Google Scholar]
- 74.Dawood S, Merajver SD, Viens P, Vermeulen PB, Swain SM, Buchholz TA, et al. International expert panel on inflammatory breast cancer: Consensus statement for standardized diagnosis and treatment. Ann Oncol. 2010 Aug 9; doi: 10.1093/annonc/mdq345. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 75.Levine PH, Steinhorn SC, Ries LG, Aron JL. Inflammatory breast cancer: The experience of the Surveillance, Epidemiology, and End results (SEER) program. J Natl Cancer Inst. 1985 Feb;74(2):291–297. [PubMed] [Google Scholar]
- 76.Anderson WF, Chen BE, Brinton LA, Devesa SS. Qualitative age interactions (or effect modification) suggest different cancer pathways for early-onset and late-onset breast cancers. Cancer Causes Control. 2007 Dec;18(10):1187–1198. doi: 10.1007/s10552-007-9057-x. [DOI] [PubMed] [Google Scholar]
- 77.Taylor SH, Walters R. Potential impact of tumor registry rule changes for recording inflammatory breast cancer. Cancer. 2010 Jun 1;116(11 Suppl):2745–2747. doi: 10.1002/cncr.25174. [DOI] [PubMed] [Google Scholar]
- 78.Yu JB, Gross CP, Wilson LD, Smith BD. NCI SEER public-use data: Applications and limitations in oncology research. Oncology (Williston Park) 2009 Mar;23(3):288–295. [PubMed] [Google Scholar]
- 79.Rothman KJ, Greenland S, Lash TL, editors. Morgenstern H. Ecologic studies. Modern Epidemiology. Third ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2008. pp. 511–529. [Google Scholar]
- 80.Kokal WA, Hill LR, Porudominsky D, Beatty JD, Kemeny MM, Riihimaki DU, et al. Inflammatory breast carcinoma: A distinct entity? J Surg Oncol. 1985 Nov;30(3):152–155. doi: 10.1002/jso.2930300306. [DOI] [PubMed] [Google Scholar]
- 81.Brouwers B, Paridaens R, Lobelle JP, Hendrickx W, Smeets A, Neven P, et al. Clinicopathological features of inflammatory versus noninflammatory locally advanced nonmetastatic breast cancer. Tumour Biol. 2008;29(4):211–216. doi: 10.1159/000152938. [DOI] [PubMed] [Google Scholar]
- 82.Dawood S, Broglio K, Gong Y, Yang WT, Cristofanilli M, Kau SW, et al. Prognostic significance of HER-2 status in women with inflammatory breast cancer. Cancer. 2008 May 1;112(9):1905–1911. doi: 10.1002/cncr.23350. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 83.Kleer CG, Zhang Y, Pan Q, Gallagher G, Wu M, Wu ZF, et al. WISP3 and RhoC guanosine triphosphatase cooperate in the development of inflammatory breast cancer. Breast Cancer Res. 2004;6(2):R110–R115. doi: 10.1186/bcr755. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 84.Turpin E, Bieche I, Bertheau P, Plassa LF, Lerebours F, de Roquancourt A, et al. Increased incidence of ERBB2 overexpression and TP53 mutation in inflammatory breast cancer. Oncogene. 2002 Oct 24;21(49):7593–7597. doi: 10.1038/sj.onc.1205932. [DOI] [PubMed] [Google Scholar]
- 85.Bonnier P, Charpin C, Lejeune C, Romain S, Tubiana N, Beedassy B, et al. Inflammatory carcinomas of the breast: A clinical, pathological, or a clinical and pathological definition? Int J Cancer. 1995 Aug 9;62(4):382–385. doi: 10.1002/ijc.2910620404. [DOI] [PubMed] [Google Scholar]
- 86.Gueth U, Wight E, Schoetzau A, Langer I, Dieterich H, Rochlitz C, et al. Non-inflammatory skin involvement in breast cancer, histologically proven but without the clinical and histological T4 category features. J Surg Oncol. 2007 Mar 15;95(4):291–297. doi: 10.1002/jso.20660. [DOI] [PubMed] [Google Scholar]
- 87.Anderson WF, Chen BE, Jatoi I, Rosenberg PS. Effects of estrogen receptor expression and histopathology on annual hazard rates of death from breast cancer. Breast Cancer Res Treat. 2006 Nov;100(1):121–126. doi: 10.1007/s10549-006-9231-y. [DOI] [PubMed] [Google Scholar]