Breast Cancer Incidence, Hormone Receptor Status, Historical Redlining, and Current Neighborhood Characteristics in Massachusetts, 2005-2015

Abstract

Background: Scant research has analyzed contemporary US cancer incidence rates in relation to historical redlining (ie, 1930s US federally imposed residential segregation), implemented via the color-coded federal Home Owners' Loan Corporation (HOLC) maps.

Methods: We analyzed Massachusetts Cancer Registry data for all patients with primary invasive breast cancer (BC) diagnosed in 2005-2015 among women in the 28 Massachusetts municipalities with digitized 1930s HOLC maps. Multilevel Poisson regression estimated BC incidence rate ratios (IRR), overall and by tumor estrogen receptor (ER-positive, ER-negative) and progesterone receptor (PR-positive, PR-negative) status, in relation to HOLC grade and contemporary census tract (CT) social characteristics.

Results: Net of age and racialized group, the extremes of BC incidence were detected by combinations of HOLC grade and contemporary CT racialized economic segregation. Compared with CTs with the best HOLC grade (A + B) and most privileged contemporary characteristics (T1), for all, ER-positive and PR-positive BC, incidence was highest in T1 and mixed HOLC grade CTs (eg, IRRER+; Mixed-T1 = 1.10, 95% confidence interval [CI] = 1.01 to 1.21) and lowest in CTs with most concentrated racialized economic deprivation (T3) and no HOLC grade (eg, IRRER+; No Grade-T3 = 0.85, 95% CI = 0.75 to 0.95). For ER-negative and PR-negative BC, incidence was highest in CTs with the most contemporary deprivation, but the best HOLC grade (eg, IRRER-; A+B-T3 = 1.27, 95% CI = 0.93 to 1.75) and lowest in T1 and worst HOLC-graded CTs (eg, IRRER-; D-T1 = 0.84, 95% CI = 0.56 to 1.25).

Conclusion: Breast cancer risk may be shaped by combined histories of redlining and present-day CT characteristics.

Figure 1.

HOLC grade (1937-1938) by ICE for racialized economic segregation tercile (2005-2015) trajectories for Massachusetts CTs (n = 474), 2010 boundaries, in municipalities (n = 28) with 1930s HOLC maps. HOLC grade operational definition: A/Green = CTs whose land area is [(100% A) OR (≥50% A and <100% A)]; B/Blue = CTs whose land area is [(100% B) OR (≥50% B and <100% B)]; C/Yellow = CTs whose land area is [(100% C) OR (≥50% C and <100% C)]; D/Red = CTs whose land area is [(100% D) OR (≥50% D and <100% D)]; Mixed = no category with ≥50%; No grade assigned = ≥50% of land area is unknown. CT ICE measures were calculated for each year in the study period: 2005-2015. CT ICE measures are based on ACS 5-year annual estimates centered in the years 2008 (2006-2010), 2009 (2007-2011), 2010 (2008-2012), 2011 (2009-2013), 2012 (2010-2014), 2013 (2011-2015), 2014 (2012-2016), and 2015 (2013-2017). CT ICE measures for the years 2005-2007 are based on linear interpolation using decennial census data for the year 2000 and the ACS 5-year estimate for 2008 (2006-2010) as anchors. For these maps, CTs were assigned to the ICE tercile they most commonly (mode) were categorized as during the years in the study period. All decennial census and ACS data are normalized to 2010 CT boundaries. ACS = American Community Survey; CT = census tract; HOLC = Home Owners’ Loan Corporation; ICE = Index of Concentration at the Extremes; M = mixed; NG = no grade; T1 = tercile 1 (highest concentration non-Hispanic White high-income households); T2 = tercile 2; T3 = tercile 3.

Figure 2.

Boxplots of current CT characteristics (2005-2015; 2010 boundaries) by HOLC grade (1937-1938): 28 Massachusetts municipalities with HOLC maps (n = 474 CTs). Results for (A) percentage of persons living below the US federal poverty line, (B) percentage of persons foreign-born, (C) ICE for racialized segregation, (D) ICE for economic segregation, (E) ICE for racialized economic segregation, and (F) ICE for home ownership. Each boxplot within each panel visualizes—in order of the features arranged from the minimum to the maximum of the y-axis—the smallest value greater than the lower quartile minus 1.5 times the interquartile range, the lower quartile, the median value (bold line), the mean value (asterisk), the upper quartile, and the largest value less than the upper quartile plus 1.5 times the interquartile range for each variable. For each ICE variable, 1 = highest concentration of the most privileged group (100% of population); -1 = highest concentration of the most deprived group (100% of population). HOLC grade operational definition: A/Green = CTs whose land area is [(100% A) OR (≥50% A and <100% A)]; B/Blue = CTs whose land area is [(100% B) OR (≥50% B and <100% B)]; C/Yellow = CTs whose land area is [(100% C) OR (≥50% C and <100% C)]; D/Red = CTs whose land area is [(100% D) OR (≥50% D and <100% D)]; Mixed = no category with ≥50%; No grade assigned = ≥50% of land area is unknown. CT ICE, poverty, and foreign-born measures were calculated for each year in the study period: 2005-2015. CT measures are based on ACS 5-year annual estimates centered in the years 2008 (2006-2010), 2009 (2007-2011), 2010 (2008-2012), 2011 (2009-2013), 2012 (2010-2014), 2013 (2011-2015), 2014 (2012-2016), and 2015 (2013-2017). CT ICE measures for the years 2005-2007 are based on linear interpolation using decennial census data for the year 2000 and the ACS 5-year estimate for 2008 (2006-2010) as anchors. All decennial census and ACS data are normalized to 2010 CT boundaries. ACS = American Community Survey; CT = census tract; HOLC = Home Owners’ Loan Corporation; ICE = Index of Concentration at the Extremes.

Figure 3.

Incidence rate ratios and 95% confidence intervals, 2005-2015, of primary invasive breast cancer, overall and by tumor hormone receptor status, for combinations of historical HOLC grade and contemporary ICE for racialized economic segregation tercile: imputed Massachusetts Cancer Registry (n = 15 689 patients) and decennial Census and ACS data (n = 10 765 212 residents). Results for (A) all breast cancer combined, (B) ER-positive breast cancer, (C) PR-positive breast cancer, (D) ER-negative breast cancer, and (E) PR-negative breast cancer. HOLC grade operational definition: A/Green = CTs whose land area is [(100% A) OR (≥50% A and <100% A)]; B/Blue = CTs whose land area is [(100% B) OR (≥50% B and <100% B)]; C/Yellow = CTs whose land area is [(100% C) OR (≥50% C and <100% C)]; D/Red = CTs whose land area is [(100% D) OR (≥50% D and <100% D)]; Mixed = no category with ≥50%; No grade assigned = ≥50% of land area is unknown. CT ICE measures were calculated for each year in the study period: 2005-2015. CT measures are based on ACS 5-year annual estimates centered in the years 2008 (2006-2010), 2009 (2007-2011), 2010 (2008-2012), 2011 (2009-2013), 2012 (2010-2014), 2013 (2011-2015), 2014 (2012-2016), and 2015 (2013-2017). CT ICE measures for the years 2005-2007 are based on linear interpolation using decennial census data for the year 2000 and the ACS 5-year estimate for 2008 (2006-2010) as anchors. All decennial census and ACS data are normalized to 2010 CT boundaries. ACS = American Community Survey; CI = confidence interval; CT = census tract; ER- = estrogen receptor–negative breast cancer; ER+ = estrogen receptor–positive breast cancer; HOLC = Home Owners’ Loan Corporation; ICE = Index of Concentration at the Extremes; M = mixed; NG = no grade; PR- = progesterone receptor–negative breast cancer; PR+ = progesterone receptor–positive breast cancer; T1 = tercile 1 (highest concentration non-Hispanic White high-income households); T2 = tercile 2; T3 = tercile 3.

Figure 4.

Measures of multiplicative and additive interaction and 95% confidence intervals, 2005-2015, for interactions between historical HOLC grade and contemporary ICE for racialized economic segregation tercile in relation to incidence rates of primary invasive breast cancer, overall and by tumor hormone receptor status: imputed Massachusetts Cancer Registry (n = 15 689 patients) and decennial Census and American Community Survey data (n = 10 765 212 residents). Results regarding (A) multiplicative interaction and (B) additive interaction. HOLC grade operational definition: A/Green = CTs whose land area is [(100% A) OR (≥50% A and <100% A)]; B/Blue = CTs whose land area is [(100% B) OR (≥50% B and <100% B)]; C/Yellow = CTs whose land area is [(100% C) OR (≥50% C and <100% C)]; D/Red = CTs whose land area is [(100% D) OR (≥50% D and <100% D)]; Mixed = no category with ≥50%; No grade assigned = ≥50% of land area is unknown. CT ICE measures were calculated for each year in the study period: 2005-2015. CT measures are based on ACS 5-year annual estimates centered in the years 2008 (2006-2010), 2009 (2007-2011), 2010 (2008-2012), 2011 (2009-2013), 2012 (2010-2014), 2013 (2011-2015), 2014 (2012-2016), and 2015 (2013-2017). CT ICE measures for the years 2005-2007 are based on linear interpolation using decennial census data for the year 2000 and the ACS 5-year estimate for 2008 (2006-2010) as anchors. All decennial census and ACS data are normalized to 2010 CT boundaries. ACS = American Community Survey; CI = confidence interval; CT = census tract; ER = estrogen receptor; HOLC = Home Owners’ Loan Corporation; ICE = Index of Concentration at the Extremes; M = mixed; NG = no grade; PR = progesterone receptor RERI = relative excess risk due to interaction; T1 = tercile 1 (highest concentration non-Hispanic White high-income households); T2 = tercile 2; T3 = tercile 3.