BREAST CANCER DISPARITIES IN THE CONTEXT OF RESIDENTIAL RACIAL SEGREGATION: EVIDENCE FROM Kirsten Beyer, Yuhong Zhou, Kevin Matthews, Amin Bemanian, Kelly Hoormann, Purushottam Laud, Ann Nattinger WI Breast Cancer Task Force March 16, 2016
OVERVIEW Introduction (Spatial) epidemiology of breast cancer (in southeastern Wisconsin) Breast cancer survival disparities Residential racial segregation Segregation and cancer survival New index of racial bias in mortgage lending Exploratory spatial data analysis Survival analysis: individual and neighborhood factors, & Future Directions
Wisconsin was one of three states (Mississippi, Oklahoma) where breast cancer death rates did NOT decrease among Black women from 2003-2012. Despite an estimated screening rate of 88% among Black women in Wisconsin, the proportion of tumors diagnosed late among Black women was 49% - the highest rate in the nation among all states examined.
160 Female Breast Cancer Incidence in MCW Cancer Center Catchment Area by Race/Ethnicity, 2006-2010 (Age-Adjusted Rate with 95% Confidence Interval) 140 135.5 120 118.4 100 80 92.5 93.9 80.2 60 40 20 0 White Hispanic Black AIAN API
Breast Cancer Incidence and Mortality by County in Wisconsin Source: Wisconsin Cancer Reporting System (http://www.cancer-rates.info/wi/index.php)
FOCUS: BREAST CANCER SURVIVAL DISPARITIES Nationally, wide gaps in breast cancer survival rates by race persist and may be growing. Not only is this gap significant and persistent it exists despite the availability of effective early detection and treatment therapies that are known to lengthen survival among diverse population groups. Black/African Americans are known to have the shortest survival among all racial groups for most cancers. Nationally, only 68% of Black women diagnosed with breast cancer are alive 10 years post-diagnosis, as compared to 84% of White women.
Introduction SEGREGATION AND SURVIVAL Studies have linked measures of segregation with cancer survival. Schootman et al. (2009) found that the percentage of the census tract population of African American race was in combination with tumor grade and stage at diagnosis able to explain why African American women were more likely to develop breast cancer metastases than White women. Russell et al (2011) found a small but significant relationship between neighborhood residential composition and survival in Georgia. Pruitt found segregation to be associated with poorer mortality among both Black and Hispanic women in Texas. Whitman et al. (2012) found that the citywide segregation level was associated with the degree of Black to White disparity in breast cancer mortality rates, among the 25 largest cities in the United States. Potential pathways linking processes and patterns of residential racial segregation to cancer survival include spatial access to quality health care, exposure to stressors that promote cancer progression or hinder recovery through increased production of stress hormones or reduced immune function, and local health behavioral norms. The influence of environmental sources of stress on cancer survival and survival disparities have not been widely examined.
SEGREGATION AND HOUSING DISCRIMINATION The Great Migration of 1910-1930 brought many African Americans north in search of jobs. Housing availability resulted in African American populations clustering in the central city of Milwaukee. Meanwhile, manufacturing jobs were pulled from the central city with cheap land and low taxes to suburbs. Whites began to move out of the central city to the burgeoning suburban areas and white dominated suburban towns formed. The central city became predominantly non-white. Dis-investment in urban cores ensued and many cities experienced inner city blight. http://www.lib.niu.edu/1979/ii790704.html Discriminatory housing policies, including redlining, perpetuated segregation. The Fair Housing Act (1968) and a number of other policies were put in place in order to address inequality. The Home Mortgage Disclosure Act (1975) was enacted to collect data and monitor mortgage loan practices.
OBJECTIVES 1. Examine breast cancer survival disparities among Hispanic, Non- Hispanic Black, and Non-Hispanic White women in southeastern Wisconsin 2. Develop and test a new method for characterizing spatially continuous patterns of racial bias in mortgage lending, as a measure of institutional racism. 3. Determine whether this new measure of racial bias in mortgage lending, as well as a composite socioeconomic status index, are associated with breast cancer survival.
STUDY AREA The study area is defined as the two southeastern Wisconsin metropolitan statistical areas (Milwaukee- Waukesha-West Allis, Racine) The City of Milwaukee is the metropolitan center of this region with approximately 600,000 residents, of whom 39% are non-hispanic Black/African American, 17% are Hispanic/Latino, and 38% are non-hispanic White. The Milwaukee County population experiences lower socioeconomic status than the state population. Milwaukee residents have lower levels of education, higher rates of unemployment, lower median household incomes, higher rates of poverty, and lower home ownership rates. Milwaukee has consistently been called one of the most segregated cities in the United States. Back in time 60 years : America s most segregated city Toronto Star, Jan 25, 2016 Employment Rate Black 58% White 88% Poverty rate Black 39% White 8% Residents living in extreme-poverty neighborhoods Black 32.9% White 1.6% Adult male incarceration rate Black 11.9% White 0.9% Median household income Black $26,036 White $62,100 Average poverty rate of school attended by a student who is Black 78.1% White 24.2% Source: American Civil Liberties Union
A SPATIALLY CONTINUOUS INDEX OF RACIAL BIAS IN MORTGAGE LENDING Home Mortgage Disclosure Act (1975) database Census tract Age, sex, race, ethnicity, and income of applicant Loan purpose, amount A continuously defined measure of racial bias in mortgage lending, using adaptive spatial filtering Black applicant to White applicant odds ratio of denial of mortgage application, controlling for the sex of the applicant and the loan to income ratio Filter threshold guided by number of total applicants, number of White applicants, and number of Black applicants Model-based index estimated for each filter and interpolated
SPATIAL AND STATISTICAL ANALYSIS Descriptive statistics Kaplan Meier curve Five-year survival (Kaplan Meier) map, using adaptive spatial filtering Cox proportional hazards regression map, using adaptive spatial filtering Hazard ratio of being inside versus outside of the filter, adjusting for age and stage at diagnosis Cox proportional hazards regression models Individual characteristics Neighborhood characteristics Population density SES index (primary component from analysis of multiple census variables) Individual Census SES variables Index of racial bias in mortgage lending
STUDY POPULATION Frequency Percent Race and Ethnicity Black 1,010 10.70% Hispanic 286 3.00% White 8,172 85.3% Age Groups 18-44 years 1,255 13.30% 45-54 years 2,270 24.00% 55-64 years 2,214 23.40% 65-74 years 1,799 19.00% 75+ years 1,930 20.40% Stage at Diagnosis Local 5,981 63.20% Regional 2,993 31.60% Distant 494 5.2%
0.40 0.50 0.60 0.70 0.80 0.90 1.00 Breast Cancer Survival by Race and Ethnicity All Causes of Death in Southeastern MSAs 2002-2011 0 1000 2000 3000 4000 Months of Survival White/Caucasian Hispanic Black/African American
Five-Year (All Causes) Breast Cancer Survival (Kaplan Meier Method) Local (All Causes) Breast Cancer Survival (Cox proportional hazards model, adjusted for age and stage of diagnosis)
NEIGHBORHOOD SES AND POPULATION DENSITY
RACIAL BIAS IN MORTGAGE LENDING: A CONTINUOUSLY DEFINED MEASURE The odds of being denied a mortgage application if you are Black, as compared to if you are White, while controlling for the sex of the applicant and the loan amount to income ratio.
MODEL RESULTS: INDIVIDUAL CHARACTERISTICS HR [95% CI] Race and Ethnicity White Referent Black 1.602 * 1.387 1.849 Hispanic 1.581 * 1.146 2.180 Stage at Diagnosis Local Referent Regional 1.984 * 1.771 2.222 Distant 9.813 * 8.549 11.264 Age Groups 18-44 years Referent 45-54 years 1.033 0.830 1.285 55-64 years 1.309 * 1.059 1.616 65-74 years 1.993 * 1.619 2.454 75+ years 4.604 * 3.816 5.556 * P-value <0.05 Number of subjects=9,468 (excluding 131 obs. with Unstaged/Missing stage) Number of deaths=1,560
MODEL RESULTS: SUBGROUP ANALYSES WITH NEIGHBORHOOD CHARACTERISTICS Black/African American Group White/Caucasian Group HR [95% CI] HR [95% CI] Stage at Diagnosis Age Groups Population Density 1.003* ~ 1.000 ~ 1.005 ~ 0.998 ~ 1.001 ~ 1.000 1.005* 1.001 1.007 0.999 1.002 + Racial Bias Index (continuous) 1.140 * 1.035 1.256 1.009 0.983 1.036 + Racial Bias Index (boolean >2) 1.650 * 1.198 2.273 0.985 0.868 1.117 + SES Composite Index 0.954 0.869 1.046 1.001 0.951 1.055 + Percent Female-Headed Household 0.985 * 0.972 0.999 0.997 0.989 1.005 * P-value <0.05 Number of subjects=1,010 (excluding Number of subjects=8,172 (excluding 2 obs. with Unstaged/Missing stage 49 obs. with Unstaged/Missing stage Number of deaths=228 Number of deaths=1,293
CONCLUSIONS AND FUTURE DIRECTIONS Non-Hispanic Black and Hispanic women experience poorer breast cancer survival than White women in southeastern Wisconsin. Racial bias in mortgage lending is associated with breast cancer survival among Black women. The experience of the Hispanic community was not emphasized in this work, but will be examined as research continues, including with a measure that emphasizes the odds of denial of a Hispanic applicant, as compared to a non-hispanic White applicant. Race and ethnicity are complex and socially constructed. There are multiple variables for race in the dataset, for applicant and for co-applicant. More work is needed to examine how the patterns of denial appear when measured using different definitions. Patterns of mortgage denial, including by race, are not static through time or space. Additional work will focus on spatial and temporal patterns of mortgage denial. Milwaukee is a highly segregated metropolitan area, and this preliminary work indicates that patterns of segregation may have public health implications. Future work should be done in other places to confirm these findings, and community engaged approaches should be employed to move these findings toward intervention and population health improvement.
ACKNOWLEDGMENTS Thanks to the Wisconsin Cancer Reporting System for access to breast cancer data for Southeastern Wisconsin. This work is supported in part by the Research and Education Program Fund, a component of the Advancing a Healthier Wisconsin endowment at the Medical College of Wisconsin, and in part by the Medical College of Wisconsin Cancer Center, Population Sciences Program through a seed grant: Spatial Patterning of Segregation and Survival in Southeastern Wisconsin.
FORTHCOMING Beyer KMM, Zhou Y, Matthews K, Hoormann K, Bemanian A, Laud PW, Nattinger AB. 2016. Breast and Colorectal Cancer Survival Disparities in Southeastern Wisconsin, Wisconsin Medical Journal, 115:1, p17-21. Beyer KMM, Zhou Y, Matthews K, Bemanian A, Laud PW, Nattinger AB. New spatially continuous indices of redlining and racial bias in mortgage lending: links to survival after breast cancer diagnosis and implications for health disparities research (in review).
Father James Groppi, Roman Catholic priest and noted civil rights activist, with supporters at a demonstration for open housing in Milwaukee in the 1960s ( http://www.themakingofmilwaukee.com) KIRSTEN BEYER KBEYER@MC W.EDU