Determinants of Breast Cancer Screening Adherence During the COVID-19 Pandemic in a Cohort at Increased Inherited Cancer Risk in the United States

Abstract

Background

We examined neighborhood characteristics concerning breast cancer screening annual adherence during the COVID-19 pandemic.

Methods

We analyzed 6673 female patients aged 40 or older at increased inherited cancer risk in 2 large health care systems (NYU Langone Health [NYULH] and the University of Utah Health [UHealth]). Multinomial models were used to identify predictors of mammogram screening groups (non-adherent, pre-pandemic adherent, pandemic period adherent) in comparison to adherent females. Potential determinants included sociodemographic characteristics and neighborhood factors.

Results

Comparing each cancer group in reference to the adherent group, a reduced likelihood of being non-adherent was associated with older age (OR: 0.97, 95% CI: 0.95, 0.99), a greater number of relatives with cancer (OR: 0.80, 95% CI: 0.75, 0.86), and being seen at NYULH study site (OR: 0.42, 95% CI: 0.29, 0.60). More relatives with cancer were correlated with a lesser likelihood of being pandemic period adherent (OR: 0.89, 95% CI: 0.81, 0.97). A lower likelihood of being pre-pandemic adherent was seen in areas with less education (OR: 0.77, 95% CI: 0.62, 0.96) and NYULH study site (OR: 0.35, 95% CI: 0.22, 0.55). Finally, greater neighborhood deprivation (OR: 1.47, 95% CI: 1.08, 2.01) was associated with being non-adherent.

Conclusion

Breast screening during the COVID-19 pandemic was associated with being older, having more relatives with cancer, residing in areas with less educational attainment, and being seen at NYULH; non-adherence was linked with greater neighborhood deprivation. These findings may mitigate risk of clinically important screening delays at times of disruptions in a population at greater risk for breast cancer.

Plain Language Summary

Breast Cancer Screening Adherence in the US During COVID-19: We examined predictors of breast cancer screening adherence during COVID-19 at two large healthcare systems. Adherence was associated with older age, having more relatives with a cancer history, and living in areas with less educational attainment. Nonadherence was associated with greater neighborhood deprivation.

Keywords

COVID-19 mammograms cancer screening inherited risk racial disparities electronic health records structural barriers neighborhood barriers education health inequities

Highlights

• Adherence to American Cancer Society guidelines for breast screening mammograms in a higher-risk (family history of cancer) population during the COVID-19 pandemic was associated with older age, having more relatives with a history of cancer, living in areas with less educational attainment (some college or below), and being seen at NYULH (vs being seen at UHealth). Non-adherence was associated with living in areas with greater neighborhood deprivation.

• Some underrepresented groups face social and neighborhood barriers to health care access. Predictors explored in this analysis were age, race/ethnicity, language preference, number of relatives with a history of cancer, neighborhood deprivation, residential segregation, racial and economic polarization, racial isolation, and educational isolation.

• Lower average days between breast screening mammograms were associated with more relatives with cancer and being seen at the NYULH study site (vs being seen at UHealth)

Introduction

Despite persistent health inequities in the United States,¹ there have been strides to induce fairness in the system so that people of all backgrounds have equitable access to health care.^2,3 Barriers to health care are due to structural (eg, living in disadvantaged areas, longer distances to seeking care) and neighborhood factors (eg, segregation) that increase the risk of chronic illnesses such as cancer and may be linked to non-adherence to cancer prevention and screening practices.^4-6 Additionally, medical mistrust among racially minoritized individuals stems from unethical studies that were conducted in the past, which may also be a contributing factor for a person not sharing or lacking knowledge of their family history (eg, worsening health outcomes).^7,8 These factors were not resolved leading up to the pandemic, which may have exacerbated disparities that already existed.⁹ Exploring how these factors interacted with adherence to cancer screening practices in a population at higher cancer risk during the COVID-19 pandemic is an important area of exploration.

Knowledge of family history of cancer gives patients context for understanding their health since family members share similar genetic makeup and life habits, which are all associated with future risk of disease.^10,11 Breast cancer is 1 of the most prevalent cancers in the United States, with an estimated 43 600 women dying from this type of cancer in 2021.¹² With the adoption of clinical decision support (CDS) algorithms, patients can be identified for tailored cancer risk management strategies (eg, genetic testing) based on their family history of cancer recorded in the electronic health record (EHR).¹³ Genetic testing among those who have not had cancer themselves but have a family history of cancer can be carried out to uncover specific gene mutations that are linked with higher cancer risk. For example, carriers of pathogenic variants in the gene BRCA1, regardless of age, are more likely to have an aggressive subtype of breast cancer called triple-negative breast cancer which accounts for 15% to 20% of all breast cancer subtypes.¹⁴ Furthermore, patient characteristics also play an important role, where it was seen that Black women have an increased prevalence of triple-negative tumors compared to non-Black women.¹⁵ Being among those at greater inherited risk for breast cancer increases the importance of adherence to preventative strategies. Based on US Preventive Services Taskforce (UPSTF) guidelines, women in the US from ages 50 to 74 who are at average risk are recommended to get biennial screening mammography.¹⁶ Based on a patient’s risk factors (eg, family history of breast cancer), it may be beneficial for them to start these screenings at an earlier age, such as in their 40s.¹⁶ Additionally, other guidelines, such as those from the American Cancer Society, recommend that even patients who are at average risk (eg, no personal, family, gene mutation history) receive annual breast screening mammograms starting from the age of 40.¹⁷ The guidelines are more consistent that patients at increased risk of breast cancer should start screening earlier, in their 40s or even earlier.^17,18 Adherence to these guidelines can help detect early stages of breast cancer thus leading to the treatment having a higher probability of being successful.¹⁷

Unfortunately, hospital policies, personal decisions, and mitigation strategies during the COVID-19 pandemic may have delayed women’s annual breast screening mammograms.^19,20 US cancer screening rates, including breast screening participation, fell in the first phases of the pandemic (eg, March 2020 to December 2020).²¹ A major reduction in breast screening was caused by the recommendation of the American Society of Breast Surgeons (ASBrS) and the American College of Radiology (ACR) to delay breast screening exams (eg, screening mammography, ultrasound) on March 26, 2020.²² The effect of these delays may be expected to have a minor effect on breast cancer mortality in the 5-10 years following this service interruption.²³ However, there is limited research on the disruption to annual breast screening mammograms for high-risk patients during the COVID-19 pandemic.

There also may have been increased roadblocks faced by women who belong to underserved communities in obtaining the necessary preventative care that they need, due to the pandemic impact on health systems in communities with high levels of COVID-19 morbidity and mortality. This can be seen with people living in areas with higher rates of poverty (eg, independently associated) being correlated with a lower likelihood of receiving screening in the COVID-19-affected year.²⁴ Blacks and Latinx had a higher presence working in contexts that had potential direct contact with COVID-19 (eg, health care, supermarkets, hospitality services, public transportation).^25,26 Often these essential workers were scheduled to work longer hours and may not have had the time, energy, or access to seek preventive care during the height of the pandemic.²⁷ Additionally, there was an alarming decline in overall life expectancy for Black (2.7 years) and Latinx (1.9 years) individuals, which was greater when compared to people who are White (0.8 years).^9,28,29

These observed inequities may be due to social (eg, race/ethnicity, place of being born, area of work, sites of education), structural, and neighborhood-level determinants of health, which may contribute to delays in getting breast screening mammograms.^8,30-32 These determinants may result in deferred education about the importance of screening.⁸ Communities and populations that are medically underserved (eg, African American and Latinx women) experience delays in the starting and conclusion of treatment after diagnosis of breast cancer.^33-36 A previous study reported that educational achievement is strongly correlated with health literacy, which is important to understanding one’s overall health.³⁷ Living in a deprived neighborhood was associated with lower breast cancer survival even after controlling for multiple factors (eg, race, age, breast cancer risk factors, tumor, access to care).³⁸ This means that there are unaccounted-for factors due to structural and neighborhood disadvantages. Furthermore, people from racial/ethnic subgroups and those with lower socioeconomic status are largely affected by inequalities in access to health care. This suggests that neighborhood factors (eg, living in disadvantaged areas) that are interlinked with racism play a contributing role in delaying health care and treatment, which may lead to worse health outcomes.^6,7,38-42 Lower socioeconomic status results in difficulty attaining tangible resources such as health insurance.⁸ This may contribute to Black women being among the highest rates of mortality from breast cancer.⁴³ However, more research is needed regarding the impact of these factors among populations with higher risk for cancer.

In this study, using data collected by 2 National Cancer Institute-designated comprehensive cancer centers from 2 large health care systems in the United States, we aimed to identify predictors of adherence to annual breast screening mammograms during the COVID-19 pandemic among patients with increased hereditary risk (eg, meeting criteria of family members with hereditary cancer) for cancer. American Cancer Society¹⁷ guidelines are used in this analysis because these guidelines were in clinical use in both health care systems, during the study timeframe. We also investigated whether neighborhood factors (eg, additional data joined to EHR) contributed to a lack of adherence to annual breast screening mammograms such as experiencing greater delays because of social, structural, and neighborhood-level barriers. Based on previous research that was described above, unexplained factors are associated with lower adherence to breast screening mammograms amongst female patients. Findings from this research could inform health strategies and policies for future pandemics or health emergencies for high-risk patients.

Materials and Methods

Patient Selection

This analysis used data from the Broadening the Reach, Impact, and Delivery of Genetic Services (BRIDGE) study, which was conducted at 2 sites: NYU Langone Health (NYULH) and the University of Utah (UHealth).⁴⁴ NYULH has served over 8 million patients annually across 300 ambulatory sites and affiliated hospitals (2018 to 2021). UHealth served over 1 million patients across 6 states in the same time frame, operating 5 hospitals and 12 community health care centers. These sites both utilize the Epic© EHR system but have different health system organization structures and patient demographics.

During the early periods of the pandemic, these institutions placed lockdowns at similar times (NYULH: March 22, 2020,⁴⁵ UHealth: March 27, 2020⁴⁶). In the early days of the pandemic (around mid-April) New York County, where NYULH is located, experienced a drastic increase in COVID-related deaths (about 100 deaths for 7-day average).⁴⁷ In contrast, Salt Lake County, where UHealth is centered, experienced much lower COVID-related deaths (<1 deaths for 7-day average) within the same period.⁴⁸ Around mid-June 2020, the majority of cases/deaths of COVID-19 in New York State (8% mortality) were focused in New York City.⁴⁹ Meanwhile, Utah experienced lower mortality due to COVID-19 (1% mortality).⁵⁰ In New York City, the death rates were highest among people of color (Black: 201.6 per 100 000, Latinx: 204.6 per 100 000) compared to Whites (101.3 per 100 000).⁵¹ Utah depicted a similar story where Latinx individuals were 14.2%⁵² of the population but had a greater relative proportion of cases (n = 6315) compared to Whites (78.0% of the population; n = 5125).⁵³

Patients in the BRIDGE trial were identified by a CDS algorithm as meeting National Comprehensive Cancer Network (NCCN) family history-based criteria for genetic evaluation of inherited cancer syndromes (eg, a higher-risk population).¹³ In addition, eligibility criteria for the BRIDGE trial required individuals to be aged 25 to 60 years old, have had a primary care appointment in the NYULH or UHealth systems from 2018 to 2021, have not had a prior diagnosis of cancer other than non-melanoma skin cancer, and also have not received genetic counseling or testing associated with the family history of cancer. Among these patients, the data for this analysis was drawn from the EHR from 2018 to 2021 and was primarily comprised of patient demographics (eg, age, sex, race/ethnicity, language preference) and breast cancer screening appointments before and during the COVID-19 pandemic (eg, had opportunities to be screened due to being active primary care patients throughout the period). For this analysis, the data was restricted to women who were eligible for a breast screening mammogram at the start of the study period; female patients ≥40 years old in 2018 (43 years old at the time of data extraction).¹⁷ The breast screening mammogram dates ranged from May 1, 2018, to May 28, 2021. Patients in the analytical sample had complete data on all variables used in this analysis. NYULH and UHealth Institutional Review Boards approved this study. The University of Utah Institutional Review Board in Salt Lake City, Utah, approved this protocol (IRB_00115509) on 5/31/19, with an amendment on 5/18/20, acting as a single IRB for the 2 institutions.

Outcome Measures

Breast cancer Screening

We identified patients in key periods of the COVID-19 pandemic (before and during the pandemic) to examine their annual breast screening mammogram rates. The pre-pandemic screening dates correspond to the time before the lockdowns in the respective health system cities (NYULH: before March 22, 2020,⁴⁵ UHealth: before March 27, 2020⁴⁶). We determined whether patients underwent breast screening mammograms before and after the start of the pandemic and the days between consecutive breast screening mammograms. This was done to highlight the importance of examining pre-pandemic adherence vs pandemic-only adherence (eg, the impact of the pandemic). At both sites, there was a time between the initial lockdowns (NYULH: March 22, 2020,⁴⁵ UHealth: March 27, 2020⁴⁶) and the resumption (June 2020) of outpatient services (eg, breast screening mammograms). Few patients had breast screening mammograms during this period (NYULH: 76 patients, UHealth: 27 patients). Breast screening mammogram intervals shorter than 400 days were coded as 1 (adherent), and intervals equal to or exceeding 400 days were coded as zero (nonadherent). Using this indicator, we categorized patients into 4 groups: adherent (female patients who adhered to annual breast screening mammograms in both periods, where the pandemic did not affect screening), pre-pandemic adherent (female patients who adhered to annual breast screening mammograms during pre-pandemic periods and not during pandemic periods, where the pandemic may have disrupted these patients), pandemic period adherent (female patients who adhered to annual breast screening mammograms during pandemic periods and not during pre-pandemic periods, where the pandemic may have encouraged these patients to start preventive care regularly), and nonadherent (female patients who did not adhere to annual breast screening mammograms in both periods, where not following annual breast screening mammograms was not affected by the pandemic). We also computed the average time between breast cancer screenings for each patient by calculating the time between consecutive breast screening mammograms and then computed the average.

Sociodemographic Characteristics

Information from the EHR included the patient’s age (measured continuously), race/ethnicity (White, Black, Latinx, Other), preferred language (English or non-English), and the number of relatives with a history of cancer. The “Other” race consisted of Asian, American Indian/Alaska Native, Native Hawaiian/Pacific Islander, and Unknown/Did not disclose. For “non-English”, this group consisted of people whose preferred language was Spanish or “Other”.

Neighborhood Characteristics

Patients’ zip codes were linked to census tract-level measures using 5-year (2017-2021) American Community Survey estimates of neighborhood deprivation, residential segregation, and educational isolation from the ndi⁵⁴ package in R.⁵⁵ We merged the neighborhood variables with the breast screening mammograms dataset in 4 steps. First, for each neighborhood variable, we merged the census tract from the crosswalk available by the U.S. Department of Housing and Urban Development (HUD’s) Office of Policy Development and Research.⁵⁶ Second, since 1 zip code can contain multiple census tracts, we multiplied the neighborhood variable by the census tract weight. Third, within each zip code, a summation was applied to obtain the weighted variable for each zip code to have the correct neighborhood values. Fourth, we joined the neighborhood variables to the analytic dataset by zip code.

Neighborhood Deprivation

The Powell-Wiley Neighborhood Deprivation Index (NDI) quantified neighborhood deprivation.^57,58 This index was derived through a factor analysis of 13 measures encompassing wealth, income, education, occupation, and housing conditions.⁵⁹ Higher values of neighborhood deprivation represent more deprivation of the given neighborhood, with a range of −3.6 to 2.8.

Residential Segregation

The Dissimilarity Index (DI) is a measure of racial segregation that examines the evenness between 2 racial groups in a given area.⁶⁰ Based on the ndi package,⁵⁴ this variable can range from zero to 1, with values closer to 1 representing the proportion of the racial/ethnic subgroup (eg, non-Hispanic or non-Latinx Black) members that would have to change their location of living to obtain an even distribution compared to the given reference group (eg, non-Hispanic or non-Latinx White).

Racial and Economic Polarization

The Index of Concentration at the Extremes (ICE) was initially designed to measure residential segregation.^61,62 It is computed by taking the white non-Hispanic population in the 80th income percentile vs the Black alone (including Hispanics) populations in the 20^th income percentile in a given city. This variable can range from −1 (most deprived) to 1 (most privileged). The middle ground of zero can represent 2 possible outcomes: there are residents in the given area who aren’t in either category or there is an equal distribution of the categories. Ultimately a value of zero for ICE demonstrates that a given area is not dominated by concentrations of either of the 2 groups.

Racial Isolation

The Racial Isolation Index (RII) measures the extent to which a community is racially isolated from other groups (eg, how isolated are non-Hispanic or Latinx, Black or African).⁶³ Greater values (closer to 1) of RII are correlated with a higher prevalence of being Hispanic or Latinx, Black or African American alone in a given area (more isolation). In other words, values that are closer to zero represent lower proportions of being Hispanic or Latinx, Black or African American that are separated from other races (less isolation), meaning that Hispanic or Latinx, Black or African American are more integrated with other races in their given community.

Educational Isolation

The Educational Isolation Index (EII) is the proportion of a given area’s educational attainment.⁶⁴ This variable measures the proportion of which a given area is educated, ranging from zero to 1, with lower value areas indicating greater education attainment and higher value areas indicating less education attainment (eg, reference group: people with some college or below - less than high school graduate, high school graduate (includes equivalency), and some college or associate’s degree).

Statistical Analysis

We calculated descriptive statistics for all variables to describe the study sample. We performed bivariate analyses using the Kruskal-Wallis rank sum and Pearson’s Chi-squared tests to examine descriptive and neighborhood factors across the study sites. Multivariable models⁶⁵ were used to identify demographic and neighborhood predictors of cancer screening adherence and the average time between screening dates during the COVID-19 pandemic.⁶⁶ Multilevel models were explored, but due to the small interclass correlation coefficient (<0.01), a zip code-level effect was not needed. When utilizing a multinomial logistic regression to understand adjusted associations between the breast screening categories, we compared each group in reference to the adherent group. This allowed us to represent what variables are associated with higher or lower probabilities of being in each of the categories when compared to the adherent group. When investigating average days between breast screening mammograms, a linear regression was used to explain the adjusted associations. All continuous variables in the multivariable models were standardized with a mean of zero and a standard deviation of 1. Regression coefficients with their corresponding 95% confidence intervals are presented. All statistical analyses were conducted using R,⁵⁵ and statistical significance was assessed at P < 0.05.

Results

Patient Characteristics

Out of the 6673 female patients meeting eligibility criteria (eg, at higher-risk for cancer) (Table 1), the majority were White (65.47%), had EHR-recorded preference for English (95.50%), and had an average age of 52.07 years (SD = 5.74 years). Patients in the sample had an average of 1.03 breast screening mammograms recorded in the EHR (SD = 1.11 breast screening mammograms) and an average of 458.23 days (SD = 158.00 days) between breast screening mammograms among those with at least 2 mammograms (n = 2163; 32.41%). Patients had an average of 2.26 relatives with cancer (SD = 1.54 relatives) documented in their family history. Patients in the overall sample had a mean score of −0.38 (SD = 0.71) for NDI, which represents an overall distribution of patients living in areas with low neighborhood deprivation. The mean DI score was 0.39 (SD = 0.13), indicating that, on average, individuals in this study lived in areas with low-to-moderate levels of residential segregation. ICE had a mean score of 0.26 (SD = 0.18) signifying an overall distribution of patients living in an area with a low to moderate distribution of privilege. With the RII, there was a mean score of 0.08 (SD = 0.15), which represented less racial isolation, on average, across the areas where these patients resided. Finally, EII had a mean score of 0.53 (SD = 0.18), which represents moderate educational isolation across the areas where these patients lived.

Table 1.

Descriptive Statistics by Breast Cancer Screening Categories.

Characteristic	N	Overall N = 6673	Adherent N = 314	Nonadherent N = 5607	Pandemic Period Adherent N = 317	Pre-Pandemic Adherent N = 435	P-value
Number of b reast screening mammograms , mean (SD)	6673	1.03 (1.11)	3.24 (0.57)	0.70 (0.84)	2.46 (0.56)	2.67 (0.56)	<0.001
Average time between mammograms (Days), mean (SD)	2163	458.23 (158.00)	339.57 (65.63)	559.44 (145.86)	359.05 (82.00)	360.91 (99.36)	<0.001
Age, mean (SD)	6673	52.07 (5.74)	53.15 (5.47)	51.92 (5.76)	52.42 (5.62)	53.09 (5.57)	<0.001
Race/Ethnicity, no. (%)	6673						0.001
White		4369 (65.47%)	220 (70.06%)	3674 (65.53%)	183 (57.73%)	292 (67.13%)
Black		522 (7.82%)	27 (8.60%)	420 (7.49%)	45 (14.20%)	30 (6.90%)
Latinx		692 (10.37%)	28 (8.92%)	582 (10.38%)	40 (12.62%)	42 (9.66%)
Other		1090 (16.33%)	39 (12.42%)	931 (16.60%)	49 (15.46%)	71 (16.32%)
Language, no. (%)	6673						0.049
English		6373 (95.50%)	306 (97.45%)	5361 (95.61%)	296 (93.38%)	410 (94.25%)
Non-English		300 (4.50%)	8 (2.55%)	246 (4.39%)	21 (6.62%)	25 (5.75%)
Number of relatives with cancer, mean (SD)	5240	2.26 (1.54)	2.82 (1.86)	2.18 (1.48)	2.39 (1.69)	2.61 (1.69)	<0.001
Neighborhood deprivation mean (SD)	6673	−0.38 (0.71)	−0.49 (0.72)	−0.37 (0.70)	−0.27 (0.77)	−0.40 (0.71)	0.001
Residential segregation, mean (SD)	6673	0.39 (0.13)	0.41 (0.13)	0.39 (0.13)	0.38 (0.12)	0.39 (0.12)	0.007
Racial and economic polarization, mean (SD)	6673	0.26 (0.18)	0.29 (0.18)	0.26 (0.18)	0.23 (0.19)	0.26 (0.18)	0.001
Racial isolation, mean (SD)	6673	0.08 (0.15)	0.09 (0.14)	0.08 (0.14)	0.13 (0.19)	0.09 (0.15)	<0.001
Educational isolation, mean (SD)	6673	0.53 (0.18)	0.50 (0.16)	0.53 (0.18)	0.53 (0.17)	0.50 (0.17)	<0.001

P-value by Kruskal-Wallis rank sum or Pearson’s Chi-squared tests.

Patient Characteristics by Breast Cancer Screening Categories

In the patient characteristics stratified by breast cancer screening groups (Table 1), female patients who adhered to breast cancer screening guidelines before and during the COVID-19 pandemic (ie, throughout both periods) had the highest proportion of White participants (70.06%) compared to the other breast cancer screening categories (nonadherent: 65.53%, pandemic period adherent: 57.73%, pre-pandemic adherent: 67.13%, P = 0.001). Adherent patients had the highest proportion who were English speaking (97.45%) compared to the other breast cancer screening practice groups (nonadherent: 95.61%, pandemic period adherent: 93.38%, pre-pandemic adherent: 94.25%, P = 0.049).

The average age between the breast cancer screening groups varied by 1 year across the categories (adherent: 53.15 years, nonadherent: 51.92 years, pandemic period adherent: 52.42 years, pre-pandemic adherent: 53.09 years, P < 0.001). Women who were adherent in both periods had the highest average number of breast screening mammograms (3.24 breast screening mammograms) compared to the other groups (non-adherent: 0.70 breast screening mammogram, pandemic period adherent: 2.46 breast screening mammograms, pre-pandemic adherent 2.67 breast screening mammograms, P < 0.001). Adherent patients in both periods had the lowest average days between mammograms (339.57 days) compared to the other breast cancer screening groups (nonadherent: 559.44 days, pandemic period adherent: 359.05 days, pre-pandemic adherent: 360.91 days, P < 0.001). We also observed that these same adherent patients had the highest average number of relatives with cancer (2.82 relatives) compared to the other breast cancer screening groups (nonadherent: 2.18 relatives, pandemic period adherent: 2.39 relatives, pre-pandemic adherent: 2.61 relatives, P < 0.001).

We also observed varying scores across the structural and neighborhood factors between the breast cancer screening groups. Adherent patients had a mean NDI score of −0.49, which was the lowest compared to the other groups (nonadherent: −0.37, pandemic period adherent: −0.27, pre-pandemic adherent: −0.40, P = 0.001). This implies that adherent patients tended to live in areas with the least neighborhood deprivation. Adherent patients had the highest mean score for DI (adherent: 0.41) compared to the other breast cancer screening groups (nonadherent: 0.39, pandemic period adherent: 0.38, pre-pandemic adherent: 0.39, P = 0.007), which means a higher prevalence of living in residentially segregated neighborhoods. Adherent patients also had the highest mean ICE score (0.29), signifying living in more privileged areas compared to the other groups (nonadherent: 0.26, pandemic period adherent: 0.23, pre-pandemic adherent: 0.26, P = 0.001). Adherent patients had the second highest mean RII score (0.09) compared to the other breast cancer screening groups, which indicates that these patients are more racially isolated (nonadherent: 0.08, pandemic period adherent: 0.13, pre-pandemic adherent: 0.09, P < 0.001). Lastly, adherent patients were tied for the lowest mean EII score (0.50), which means they lived in an area with more educational attainment compared to the other breast cancer screening groups (nonadherent: 0.53, pandemic period adherent: 0.53, pre-pandemic adherent: 0.50, P < 0.001).

Associations with Breast Cancer Screening Adherence

Predictors of breast cancer screening adherence based on multinomial regression models are presented (Table 2). Compared to those in the adherent group, patients with older age (OR: 0.97, 95% CI: 0.95, 0.99) were less likely to be in the nonadherent group. Compared to the adherent group, having greater number of relatives with cancer was associated with a lower likelihood of being in the nonadherent group (OR: 0.80, 95% CI: 0.75, 0.86). Additionally, also compared to the adherent group, having more relatives with cancer was associated with a lower likelihood of being in the pandemic period adherent group (OR: 0.89, 95% CI: 0.81, 0.97). We also observed that, compared to the adherent group, higher scores for NDI (OR: 1.47, 95% CI: 1.08, 2.01) were associated with an increased likelihood of being in the nonadherent group. Compared to the adherent group, those living in areas with higher scores of EII were associated with a lower likelihood of being in the pre-pandemic adherent group (OR: 0.77, 95% CI: 0.62, 0.96). Lastly, compared to the adherent group, patients being seen at NYULH were less likely to be in the nonadherent group (OR: 0.42, 95% CI: 0.29, 0.60) or pre-pandemic adherent group (OR: 0.35, 95% CI: 0.22, 0.55).

Table 2.

Multinomial Regression for Breast Cancer Adherence During the COVID-19 Pandemic (Ref = Adherent Females).

Characteristic	Non-adherent OR (95% CI)	Pandemic Period Adherent OR (95% CI)	Pre-pandemic Adherent OR (95% CI)
Age	0.97 (0.95, 0.99)	0.98 (0.95, 1.01)	1.00 (0.97, 1.03)
Race/Ethnicity
White	Reference	Reference	Reference
Black	0.95 (0.57, 1.57)	1.42 (0.75, 2.67)	0.70 (0.36, 1.38)
Latinx	0.87 (0.57, 1.34)	1.27 (0.73, 2.21)	0.91 (0.52, 1.58)
Other	1.44 (0.99, 2.11)	1.31 (0.79, 2.16)	1.42 (0.89, 2.26)
Language
English	Reference	Reference	Reference
Non-English	1.13 (0.53, 2.38)	1.87 (0.78, 4.50)	1.64 (0.68, 3.99)
Number of relatives with cancer	0.80 (0.75, 0.86)	0.89 (0.81, 0.97)	0.94 (0.87, 1.03)
Neighborhood deprivation	1.47 (1.08, 2.01)	0.95 (0.63, 1.42)	1.40 (0.94, 2.10)
Residential segregation	0.99 (0.87, 1.13)	0.84 (0.70, 1.00)	0.95 (0.81, 1.12)
Racial and economic polarization	1.28 (0.86, 1.89)	0.75 (0.45, 1.27)	1.14 (0.69, 1.90)
Racial isolation	1.11 (0.91, 1.36)	0.98 (0.76, 1.27)	1.18 (0.91, 1.53)
Educational isolation	0.91 (0.77, 1.08)	0.92 (0.72, 1.16)	0.77 (0.62, 0.96)
Site
UHealth	Reference	Reference	Reference
NYULH	0.42 (0.29, 0.60)	1.10 (0.66, 1.83)	0.35 (0.22, 0.55)

UHealth: University of Utah Health, NYULH: NYU Langone Health.

We also presented (Table 3) adjusted associations for the average days between breast screening mammograms, patient demographics, and neighborhood factors. We observed that higher values for the number of relatives with cancer was associated with fewer days between breast screening mammograms (b = −10.23 days, 95% CI: −16.61, −3.85). Lastly, fewer days between breast screening mammograms were related to being a patient at NYULH (b = −51.76 days, 95% CI: −71.04, −32.49).

Table 3.

Linear Regression for Breast Cancer Adherence for Average Days Between Mammograms During the COVID-19 Pandemic.

Characteristic	Beta (95% CI)
Age	−3.04 (−10.41, 4.33)
Race/Ethnicity
White	Reference
Black	−6.16 (−35.44, 23.13)
Latinx	9.90 (−14.53, 34.32)
Other	−12.57 (−33.50, 8.36)
Language
English	Reference
Non-English	−17.84 (−54.60, 18.91)
Number of relatives with cancer	−10.23 (-16.61, -3.85)
Neighborhood deprivation	−9.44 (−27.52, 8.63)
Residential segregation	2.66 (−4.78, 10.10)
Racial and Economic polarization	−7.57 (−30.50, 15.33)
Racial isolation	1.97 (−9.24, 13.18)
Educational isolation	−2.95 (−12.83, 6.93)
Site
UHealth	Reference
NYULH	−51.76 (-71.04, -32.49)

UHealth: University of Utah Health, NYULH: NYU Langone Health.

Discussion

In this multisite study, we examined the association of sociodemographic and neighborhood factors with breast cancer screening adherence among high-risk patients in primary care practices in 2 large health care systems, who met guideline-based criteria for cancer genetics evaluation based on their family history of cancer. With this unique population, there is novelty and implications of the study addressing this high-risk population. We observed multiple factors associated with adherence to American Cancer Society guidelines for annual breast screening mammograms starting at age 40 during the COVID-19 pandemic. This included being of older age, having more relatives with a history of cancer, living in areas with less educational attainment, and being a patient at NYULH when adjusting for all variables in this analysis. Higher non-adherence was seen in areas with greater neighborhood deprivation after adjusting for all predictors. Additionally, overall mammography use is lower in the state of Utah than in New York, which may explain the lower likelihood of adherence being associated with UHealth. These findings suggest that patients may have social and neighborhood factors that contributed to their cancer screening practices during the COVID-19 pandemic.

This is 1 of a few studies that have examined adherence to breast cancer screening mammograms before and during COVID-19, focused on higher-risk populations. Multiple studies found that women of older age, at average risk, were associated with higher adherence to breast screening in the years leading up to the pandemic.^67,68 During this analysis of potential predictors in a high-risk population, we observed that older age was associated with being adherent to mammography screening guidelines during the pandemic. We also observed that having more relatives with a history of cancer was associated with adherence, and study site (ie, Utah). Also, patients with higher risk may have been prioritized for screening during the shutdown and that possibly continued until the backload of patients needing screening was addressed.

From the previous literature, racial differences exist in the knowledge of family history.¹¹ Having relatives with certain chronic conditions can increase the likelihood of the person having the disease themselves.⁶⁹ Consequently, these individuals may possess greater knowledge regarding screening practices, prevention, and family history, and may receive stronger recommendations from providers to screen. This may explain the association that was observed between more relatives with cancer and adherence to breast cancer screening practices.^70,71 In a prior study, higher levels of primary care utilization over a 10-year period were associated with cancer screening adherence for each race.⁷² In our analysis, we did not find any associations between race and adherence to breast screening mammograms. Results from previous studies suggest that language preference is also associated with adherence.⁷³ In this analysis, we didn’t find any relationship between language preference and breast screening mammogram practices, although few patients preferred a language other than English.

A prior study explored the possibility of neighborhood disadvantage, racial discrimination, and lack of social support playing a role in the breast cancer stage at diagnosis and survival.⁷⁴ Although the current analysis was not a survival analysis on breast cancer, mammograms are an important strategy for the early detection and improved survival of breast cancer. Furthermore, barriers to health care access are often a result of residential segregation and neighborhood factors that have historically placed minoritized individuals in disadvantaged areas.⁷⁵ Socioeconomic status (SES) plays a role in adherence to annual breast screening mammograms.⁶⁸ We observed people living in an area that was deprived were 1.5 times more likely to be non-adherent to annual mammography. In a previous study, it was observed that limited health literacy and lower education attainment were associated with lower genetic knowledge.³⁷ We observed that areas with less education were associated with a lower likelihood of being in the pre-pandemic period compared to adherent females, this may be a result of older adults living in areas with low educational attainment trying to break generational cycles. ^76-78

These results are represented within the limitations of our study. In terms of the analytical sample, this population was restricted to a specific population of female patients at increased inherited cancer risk in 2 health care systems, which may limit the generalizability of the findings to other populations or settings. Moreover, this analysis relies on self-reported data, which may be subject to recall bias or inaccuracies in reporting. Additionally, the cross-sectional study design precludes the determination of causality between the factors examined and breast cancer screening adherence. Cancer screening guidelines recommend large time gaps (eg, annual, biennial) between screenings, which results in data points that are sparse. Additionally, different groups of people may have had differential gaps in data as a result of having breast screening mammograms at places other than our study sites (eg, due to more frequent changes in job status, changes in health insurance status, seeking care at another health system). This may explain why patients who were nonadherent during the pandemic were the majority of people who had ≤1 breast screening mammogram, which also means the linear regression is comprised of people with 2 or more dates. Having 2 breast screening mammograms in a 3-year period may be hard to obtain for some patients. This study also did not account for potential confounding factors that may influence breast cancer screening adherence, such as individual-level health behaviors or comorbidities. It is also important to note that the data could not be stratified by site due to small sample sizes, thus using the site acted as a proxy for site differences (eg, dates for mammogram facilities’ closure and reopening). After the start of the pandemic, there was a decrease in the amount of breast screening mammograms, and screening started to rebound to pre-pandemic levels around 19 weeks after March 11, 2020.²⁰

Although this study was conducted within the bounds of limitations, this academic work had numerous strengths. By supplementing EHR data with social and neighborhood factors, we were able to observe that there are community-based factors that contribute to adherence to breast screening mammograms during the COVID-19 pandemic in this higher-risk population. We also displayed a method on how to reweight the neighborhood factors. This method can provide an interpretable framework that can be applied to other research investigating neighborhood factors associated with different health outcomes. Even with this forward momentum to better understand latent racial, health, and community-based disparities, we reflect on the future work that can be done. Our recommendations include analyzing a wider timeframe (eg, 5 to 10 years before COVID-19 to the present day) with more information about patients’ cancer screening adherence. Since this analysis was solely conducted on a sample of female patients in 2 health care systems, we would also advise that this analysis be expanded to more diverse populations. Also, since patients can be linked by zip code, there are areas of exploration to merge more neighborhood characteristics that may influence breast cancer adherence (eg, access to health care facilities, transportation, social support networks, air and water quality).^79-82 The impact of health care system characteristics are also important to explore. Lastly, researchers could explore the potential impact that telehealth and other remote health care services may have contributed to breast cancer screening adherence during the pandemic.

Conclusion

Different factors were associated with adherence to cancer screening guidelines during the COVID-19 pandemic, among patients at increased risk for hereditary cancer. In this study, adherence to annual breast screening mammograms was associated with being of older age, having more family relatives with cancer, living in areas with less educational attainment (some college or below), and being seen at NYULH, whereas non-adherence was associated with living in areas with greater neighborhood deprivation when comparing the cancer groups in reference to adherent females. These findings provide information to guide the development of possible interventions for these individuals. Such intervention approaches would need to address structural barriers such as neighborhood deprivation. For example, intervention approaches might address improving housing security for low-income families and improving access to place-based resources such as jobs,⁸³ which would support individuals in obtaining needed screenings. Future studies would be able to expand on neighborhood and social factors that play a role in the frequency of current practices regarding adherence to breast screening mammograms.

Footnotes

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Kensaku Kawamoto reports honoraria, consulting, sponsored research, licensing, or co-development outside the submitted work in the past 3 years with Hitachi, Pfizer, RTI International, the University of California at San Francisco, Indiana University, the Korean Society of Medical Informatics, the University of Nebraska, NORC at the University of Chicago, the Regenstrief Foundation, Elsevier, the University of Pennsylvania, MD Aware, Security Risk Solutions, Custom Clinical Decision Support, and Yale University in the area of health information technology. Kensaku Kawamoto was also an unpaid board member of the non-profit Health Level 7 International health IT standard development organization, he is an unpaid member of the U.S. Health Information Technology Advisory Committee, and he has helped develop a number of health IT tools which may be commercialized to enable wider impact. None of these relationships have direct relevance to the manuscript but are reported in the interest of full disclosure.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the Broadening the Reach, Impact, and Delivery of Genetic Services study (3U01CA232826-03S2).

Ethical Statement

Name of the Board that Reviewed their Study

University of Utah Institutional Review Board.

ORCID iDs

Adrian Harris

Kensaku Kawamoto

Data Availability Statement

The dataset supporting the conclusions of this article can be obtained by emailing a request to the corresponding author (ah5588@nyu.edu).

Appendix

References

Hammonds

Reverby

. Toward a historically informed analysis of racial health disparities since 1619. Am J Public Health. 2019;109(10):1348-1349. doi:10.2105/AJPH.2019.305262.

Civil rights division | title VI of the civil rights act of 1964 42 U.S.C. § 2000d et seq, 2015. Accessed October 27, 2023. https://www.justice.gov/crt/fcs/TitleVI-Overview

United States . National archives and records administration: Office of the federal register. An act entitled the patient protection and affordable care act. 2010. Published online. https://www.govinfo.gov/app/details/PLAW-111publ148. Accessed October 27, 2023.

Dean

Subramanian

Williams

Armstrong

Charles

Kawachi

. The role of social capital in African–American women’s use of mammography. Soc Sci Med. 2014;104:148-156. doi:10.1016/j.socscimed.2013.11.057.

Molina

Plascak

Patrick

Bishop

Coronado

Beresford

SAA

. Neighborhood predictors of mammography barriers among US-based latinas. J Racial Ethn Health Disparities. 2017;4(2):233-242. doi:10.1007/s40615-016-0222-3.

Phelan

Link

Tehranifar

. Social conditions as fundamental causes of health inequalities: theory, evidence, and policy implications. J Health Soc Behav. 2010;51(1_suppl):S28-S40. doi:10.1177/0022146510383498.

Powell

Richmond

Mohottige

Yen

Joslyn

Corbie-Smith

. Medical mistrust, racism, and delays in preventive health screening among african-American men. Behav Med. 2019;45(2):102-117. doi:10.1080/08964289.2019.1585327.

Henderson

O’Meara

Haas

, et al. The role of social determinants of health in self-reported access to health care among women undergoing screening mammography. J Womens Health. 2020;29(11):1437-1446. doi:10.1089/jwh.2019.8267.

Isasi

Naylor

Skorton

Grabowski

Hernández

Rice

. Patients, families, and communities COVID-19 impact assessment: lessons learned and compelling needs. NAM Perspect. 2021;2021:202111c. doi:10.31478/202111c.

10.

Link

Phelan

. Social conditions as fundamental causes of disease. J Health Soc Behav. 1995;Spec No:80-94. doi:10.2307/2626958.

11.

Valdez

Yoon

Qureshi

Green

Khoury

. Family history in public health practice: a genomic tool for disease prevention and health promotion. Annu Rev Public Health. 2010;31(1):69-87. doi:10.1146/annurev.publhealth.012809.103621.

12.

Cancer Facts & Figures 2021. https://www.cancer.org/research/cancer-facts-statistics/all-cancer-facts-figures/cancer-facts-figures-2021.html Accessed March 29, 2024.

13.

Bradshaw

Kawamoto

Kaphingst

, et al. GARDE: a standards-based clinical decision support platform for identifying population health management cohorts. J Am Med Inform Assoc. 2022;29(5):928-936. doi:10.1093/jamia/ocac028.

14.

Lee

Alexander

Schnitt

, et al. Clinical outcome of triple negative breast cancer in BRCA1 mutation carriers and noncarriers. Cancer. 2011;117(14):3093-3100. doi:10.1002/cncr.25911.

15.

Stead

Lash

Sobieraj

, et al. Triple-negative breast cancers are increased in black women regardless of age or body mass index. Breast Cancer Res. 2009;11(2):R18. doi:10.1186/bcr2242.

16.

Recommendation: breast cancer: screening | United States preventive services Taskforce. https://www.uspreventiveservicestaskforce.org/uspstf/recommendation/breast-cancer-screening Accessed March 29, 2024.

17.

ACS Breast Cancer Screening Guidelines. https://www.cancer.org/cancer/types/breast-cancer/screening-tests-and-early-detection/american-cancer-society-recommendations-for-the-early-detection-of-breast-cancer.html Accessed March 29, 2024.

18.

Lee

Monticciolo

Moy

. Screening guidelines update for average-risk and high-risk women. Am J Roentgenol. 2020;214(2):316-323. doi:10.2214/AJR.19.22205.

19.

Czeisler

MÉ

Marynak

Clarke

KEN

, et al. Delay or avoidance of medical care because of COVID-19–related concerns — United States, une 2020. Morb Mortal Wkly Rep. 2020;69(36):1250-1257. doi:10.15585/mmwr.mm6936a4.

20.

Song

Bergman

Chen

, et al. Disruptions in preventive care: mammograms during the COVID-19 pandemic. Health Serv Res. 2021;56(1):95-101. doi:10.1111/1475-6773.13596.

21.

Burus

Lei

Huang

, et al. Undiagnosed cancer cases in the US during the first 10 Months of the COVID-19 pandemic. JAMA Oncol. 2024;10:500-507. doi:10.1001/jamaoncol.2023.6969.

22.

News article | ASBrS. https://www.breastsurgeons.org/news/?id=45 Accessed March 29, 2024.

23.

Nguyen

Ambinder

Myers

Oluyemi

. Addressing disparities related to access of multimodality breast imaging services before and during the COVID-19 pandemic. Acad Radiol. 2022;29(12):1852-1860. doi:10.1016/j.acra.2022.03.017.

24.

Miller

Meneveau

Rochman

, et al. Impact of the COVID-19 pandemic on breast cancer screening volumes and patient screening behaviors. Breast Cancer Res Treat. 2021;189(1):237-246. doi:10.1007/s10549-021-06252-1.

25.

Selden

Berdahl

. COVID-19 and racial/ethnic disparities in health risk, employment, and household composition. Health Aff. 2020;39(9):1624-1632. doi:10.1377/hlthaff.2020.00897.

26.

Jason

Wilson

Catoe

Brown

Gonzalez

. The impact of the COVID-19 pandemic on black and hispanic Americans’ work outcomes: a scoping review. J Racial Ethn Health Disparities. 2024;11:1157-1172. doi:10.1007/s40615-023-01594-6. Published online April 28.

27.

Hibel

Boyer

Buhler-Wassmann

Shaw

. The psychological and economic toll of the COVID-19 pandemic on Latina mothers in primarily low-income essential worker families. Traumatology. 2021;27(1):40-47. doi:10.1037/trm0000293.

28.

Arias

. Provisional Life Expectancy Estimates for January through June; 2020. Published online 2019.

29.

NVSS - life expectancy. https://www.cdc.gov/nchs/nvss/life-expectancy.htm Accessed January 31, 2024.

30.

Dean

Gehlert

Neuhouser

, et al. Social factors matter in cancer risk and survivorship. Cancer Causes Control. 2018;29(7):611-618. doi:10.1007/s10552-018-1043-y.

31.

Williams

Mohammed

Shields

. Understanding and effectively addressing breast cancer in African American women: unpacking the social context. Cancer. 2016;122(14):2138-2149. doi:10.1002/cncr.29935.

32.

Braveman

Egerter

Williams

. The social determinants of health: coming of age. Annu Rev Public Health. 2011;32(1):381-398. doi:10.1146/annurev-publhealth-031210-101218.

33.

Richards

Westcombe

Love

Littlejohns

Ramirez

. Influence of delay on survival in patients with breast cancer: a systematic review. Lancet. 1999;353(9159):1119-1126. doi:10.1016/S0140-6736(99)02143-1.

34.

Dennis

Gardner

Lim

. Analysis of survival and recurrence vs. patient and doctor delay in treatment of breast cancer. Cancer. 1975;35(3):714-720. doi:10.1002/1097-0142(197503)35:3<714::AID-CNCR2820350326>3.0.CO;2-V.

35.

Gwyn

Bondy

Cohen

, et al. Racial differences in diagnosis, treatment, and clinical delays in a population-based study of patients with newly diagnosed breast carcinoma. Cancer. 2004;100(8):1595-1604. doi:10.1002/cncr.20169.

36.

Press

Carrasquillo

Sciacca

Giardina

EGV

. Racial/ethnic disparities in time to follow-up after an abnormal mammogram. J Womens Health. 2008;17(6):923-930. doi:10.1089/jwh.2007.0402.

37.

Kaphingst

Blanchard

Milam

Pokharel

Elrick

Goodman

. Relationships between health literacy and genomics-related knowledge, self-efficacy, perceived importance, and communication in a medically underserved population. J Health Commun. 2016;21(Suppl 1):58-68. doi:10.1080/10810730.2016.1144661.

38.

Goel

Hernandez

Thompson

, et al. Neighborhood disadvantage and breast cancer–specific survival. JAMA Netw Open. 2023;6(4):e238908. doi:10.1001/jamanetworkopen.2023.8908.

39.

Ross

Mirowsky

. Neighborhood disadvantage, disorder, and health. J Health Soc Behav. 2001;42(3):258-276.

40.

Hill

Ross

Angel

. Neighborhood disorder, psychophysiological distress, and health. J Health Soc Behav. 2005;46(2):170-186. doi:10.1177/002214650504600204.

41.

Shen

Fuemmeler

Sheppard

, et al. Neighborhood disadvantage and biological aging biomarkers among breast cancer patients. Sci Rep. 2022;12(1):11006. doi:10.1038/s41598-022-15260-0.

42.

Smith

Zhao

Wang

, et al. Neighborhood characteristics influence DNA methylation of genes involved in stress response and inflammation: the Multi-Ethnic Study of Atherosclerosis. Epigenetics. 2017;12(8):662-673. doi:10.1080/15592294.2017.1341026.

43.

Breast cancer death rates are highest for black women—again | breast cancer facts & figures, 2022-2024. https://www.cancer.org/research/acs-research-news/breast-cancer-death-rates-are-highest-for-black-women-again.html Accessed March 29, 2024.

44.

Kaphingst

Kohlmann

Chambers

, et al. Comparing models of delivery for cancer genetics services among patients receiving primary care who meet criteria for genetic evaluation in two healthcare systems: BRIDGE randomized controlled trial. BMC Health Serv Res. 2021;21(1):542. doi:10.1186/s12913-021-06489-y.

45.

NEWS10 ABC . COVID timeline: 11-23 March 2020; 2021. https://www.news10.com/news/covid-one-year-later-march-13-23-2020-timeline/ Accessed September 8, 2023.

46.

Stevens

Harkins

. Utah governor asks, Salt Lake City mayor orders residents to stay home to slow the spread of the coronavirus. The Salt Lake Tribune; 2020. https://www.sltrib.com/news/2020/03/27/utah-governor-issue-new/ Accessed August 24, 2023.

47.

New York County . New York coronavirus cases and deaths. USAFact; 2024. https://usafacts.org/visualizations/coronavirus-covid-19-spread-map/state/new-york/county/new-york-county/. Accessed February 26, 2024.

48.

Salt Lake County, Utah Coronavirus Cases and Deaths. Bellevue: USAFacts; 2024. https://usafacts.org/visualizations/coronavirus-covid-19-spread-map/state/utah/county/salt-lake-county/ Accessed February 26, 2024.

49.

COVID-19 Testing Tracker . Department of Health. https://coronavirus.health.ny.gov/covid-19-testing-tracker Accessed February 26, 2024.

50.

Overview. https://coronavirus-dashboard.utah.gov/pandash_overview.html Accessed February 26, 2024.

51.

covid-19-deaths-race-ethnicity-04162020-1.pdf. https://www.nyc.gov/assets/doh/downloads/pdf/imm/covid-19-deaths-race-ethnicity-04162020-1.pdf Accessed February 26, 2024.

52.

U.S. Census Bureau QuickFacts: Utah. https://www.census.gov/quickfacts/fact/table/UT/PST045223 Accessed February 26, 2024.

53.

COVID-19 data. https://coronavirus-dashboard.utah.gov/overview.html Accessed February 26, 2024.

54.

Buller

. NCI. Ndi: neighborhood deprivation indices; 2023. https://cran.r-project.org/web/packages/ndi/index.html Accessed August 31, 2023.

55.

R Core Team . R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing; 2022. https://www.R-project.org/

56.

HUD USPS ZIP code crosswalk files | HUD USER. Accessed December 9, 2023. https://www.huduser.gov/portal/datasets/usps_crosswalk.html

57.

Andrews

Tamura

Claudel

, et al. Geospatial analysis of neighborhood deprivation index (NDI) for the United States by county. J Maps. 2020;16(1):101-112. doi:10.1080/17445647.2020.1750066.

58.

Slotman

Stinchcomb

Powell-Wiley

, et al. Environmental data and methods from the Accumulating Data to Optimally Predict Obesity Treatment (ADOPT) core measures environmental working group. Data Brief. 2022;41:108002. doi:10.1016/j.dib.2022.108002.

59.

Diez Roux

Mair

. Neighborhoods and health. Ann N Y Acad Sci. 2010;1186(1):125-145. doi:10.1111/j.1749-6632.2009.05333.x.

60.

Duncan

. A methodological analysis of segregation indexes. Am Sociol Rev. 1955;20(2):210-217. doi:10.2307/2088328.

61.

Feldman

Waterman

Coull

Krieger

. Spatial social polarisation: using the Index of Concentration at the Extremes jointly for income and race/ethnicity to analyse risk of hypertension. J Epidemiol Community Health. 2015;69(12):1199-1207. doi:10.1136/jech-2015-205728.

62.

Krieger

Waterman

Spasojevic

Maduro

Van Wye

. Public health monitoring of privilege and deprivation with the index of concentration at the Extremes. Am J Public Health. 2016;106(2):256-263. doi:10.2105/AJPH.2015.302955.

63.

Anthopolos

James

Gelfand

Miranda

. A spatial measure of neighborhood level racial isolation applied to low birthweight, preterm birth, and birthweight in North Carolina. Spat Spatiotemporal Epidemiol. 2011;2(4):235-246. doi:10.1016/j.sste.2011.06.002.

64.

IJERPH | free full-text | assessing disparity using measures of racial and educational isolation. https://www.mdpi.com/1660-4601/18/17/9384 Accessed December 10, 2023.

65.

Hidalgo

Goodman

. Multivariate or multivariable regression? Am J Public Health. 2013;103(1):39-40. doi:10.2105/AJPH.2012.300897.

66.

Gelman

Hill

. 2006, Data analysis using regression and multilevel/hierarchical models. Cambridge: Higher Education from Cambridge University Press. doi:10.1017/CBO9780511790942

67.

Gathirua-Mwangi

Cohee

Tarver

, et al. Factors associated with adherence to mammography screening among insured women differ by income levels. Wom Health Issues. 2018;28(5):462-469. doi:10.1016/j.whi.2018.06.001.

68.

Katz

Tengekyon

Kahan

Calderon-Margalit

. Patient and physician characteristics affect adherence to screening mammography: a population-based cohort study. PLoS One. 2018;13(3):e0194409. doi:10.1371/journal.pone.0194409.

69.

Ashida

Goodman

Stafford

Lachance

Kaphingst

. Perceived familiarity with and importance of family health history among a medically underserved population. J Community Genet. 2012;3(4):285-295. doi:10.1007/s12687-012-0097-x.

70.

Kaphingst

Bather

Daly

Chavez-Yenter

Vega

Kohlmann

. Interest in cancer predisposition testing and carrier screening offered as part of routine healthcare among an ethnically diverse sample of young women. Front Genet. 2022;13:866062. https://www.frontiersin.org/articles/10.3389/fgene.2022.866062 Accessed December 10, 2023.

71.

Kaphingst

Peterson

Zhao

, et al. Cancer communication research in the era of genomics and precision medicine: a scoping review. Genet Med. 2019;21(8):1691-1698. doi:10.1038/s41436-018-0402-0.

72.

Flores

López

Miles

, et al. Impact of primary care physician interaction on longitudinal adherence to screening mammography across different racial/ethnic groups. J Am Coll Radiol. 2019;16(7):908-914. doi:10.1016/j.jacr.2018.12.020.

73.

Vang

Margolies

Jandorf

. Screening mammogram adherence in medically underserved women: does language preference matter? J Cancer Educ. 2022;37(4):1076-1082. doi:10.1007/s13187-020-01922-y.

74.

Coughlin

. Social determinants of breast cancer risk, stage, and survival. Breast Cancer Res Treat. 2019;177(3):537-548. doi:10.1007/s10549-019-05340-7.

75.

Massey

Denton

. The dimensions of residential segregation. Soc Forces. 1988;67(2):281-315. doi:10.1093/sf/67.2.281.

76.

Bather

Goodman

Kaphingst

. Racial segregation and genomics-related knowledge, self-efficacy, perceived importance, and communication among medically underserved patients. Genet Med Open. 2024;2:100844. doi:10.1016/j.gimo.2023.100844.

77.

Hovick

Yamasaki

Burton-Chase

Peterson

. Patterns of family health history communication among older african American adults. J Health Commun. 2015;20(1):80-87. doi:10.1080/10810730.2014.908984.

78.

Jones

Moss

Wright

Rosemberg

Killion

. “Maybe this generation here could help the next generation”: older african American women’s perceptions on information sharing to improve health in younger generations. Res Gerontol Nurs. 2018;11(1):39-47. doi:10.3928/19404921-20171129-01.

79.

CDC . Neighborhood Characteristics | CDC. Atlanta: Centers for Disease Control and Prevention; 2023. https://www.cdc.gov/dhdsp/health_equity/neighborhood.htm. Accessed December 20, 2023.

80.

ZIP Code-Level Estimates from a Local Health Survey . Added value and limitations - pmc. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7392987/ Accessed May 30, 2024.

81.

Transportation Maps and Geospatial Data . Bureau of transportation statistics. https://www.bts.gov/maps Accessed May 30, 2024.

82.

Social Determinants of Health (SDOH) and PLACES Data | PLACES: Local Data for Better Health. Atlanta: Centers for Disease Control and Prevention; 2023. https://www.cdc.gov/places/social-determinants-of-health-and-places-data/index.html Accessed May 30, 2024.

83.

Arcaya

Ellen

Steil

. Neighborhoods and health: interventions at the neighborhood level could help advance health equity. Health Aff. 2024;43(2):156-163. doi:10.1377/hlthaff.2023.01037.