A Catchment and Location-Allocation Analysis of Mammography Access in Delaware, US: Implications for disparities in geographic access to breast cancer screening

Background: Despite a 40% reduction in breast cancer mortality over the last 30 years, not all groups have benefited equally from these gains. A consistent link between later stage of diagnosis and disparities in breast cancer mortality has been observed by race, socioeconomic status, and rurality. Therefore, ensuring equitable geographic access to screening mammography represents an important priority for reducing breast cancer disparities. This study conducted a catchment and location-allocation analysis of mammography access in Delaware, a state that is representative of the US in terms of race and urban-rural characteristics and experiences an elevated burden from breast cancer. Methods: A catchment analysis using the ArcGIS Pro Service Area analytic tool characterized the geographic distribution of mammography sites and Breast Imaging Centers of Excellence (BICOEs). Poisson regression analyses identified census tract-level correlates of access. Next, the ArcGIS Pro Location-Allocation analytic tool identified candidate locations for the placement of additional mammography sites in Delaware according to several sets of breast cancer screening guidelines. Results: The catchment analysis showed that for each standard deviation increase in the number of Black women in a census tract, there were 64% (95% CI, 0.18–0.66) fewer mammography units and 85% (95% CI, 0.04–0.48) fewer BICOEs. The more rural counties in the state accounted for 41 % of the population but only 22% of the BICOEs. The results of the location-allocation analysis depended on which set of screening guidelines were adopted, which included increasing mammography sites in communities with a greater proportion of younger Black women and in rural areas. Conclusions: The results of this study illustrate how catchment and location-allocation analytic tools can be leveraged to guide the equitable selection of new mammography facility locations as part of a larger strategy to close breast cancer disparities.


Introduction
Breast cancer is the leading cause of cancer incidence and the second leading cause of cancer mortality among US women. 1 Advances in early detection and treatment are largely believed to have contributed to the 40% reduction in breast cancer mortality observed over the last 30 years, 1 but not all groups have bene ted equally from these advances. Persistent breast cancer disparities have been observed by race, socioeconomic status (SES), and geographic area. Black women have a 40% higher breast cancer mortality rate relative to White women despite similar incidence rates between the racial groups. 1 This mortality rate grows to 86% higher for younger Black vs. White women, 1 owing to the greater risk that Black women have of being diagnosed with advanced-stage breast cancer before age 50. 2 Other research found 14% lower breast cancer ve-year survival rates for low-SES patients relative to their more advantaged peers. 3 Approximately two-thirds of this disparity was attributable to conditions at presentation, including later stage at diagnosis. Finally, geographic characteristics including neighborhood measures of disadvantage (e.g., SES, segregation) 4-10 and rurality [11][12][13][14] have been associated with later stage at diagnosis and poorer breast cancer survival.
Given the consistent link between later stage at diagnosis and disparities in breast cancer outcomes across multiple population subgroups, ensuring equitable access to screening mammography represents an important goal of breast cancer prevention and early detection. Screening mammography has been a central component of breast cancer programs in the US over the last 30 + years. 15 A review of the evidence shows that screening mammography can reduce breast cancer mortality by at least 40% when completed on an annual basis beginning at age 40. 16 More recent studies have helped to establish that the bene ts of screening are independent of treatment advances. 17,18 Screening mammography decreases mortality by detecting tumors at a smaller size and an earlier stage, when therapy is more effective. 19 Decisions about how to allocate mammography resources to ensure equitable geographic access are contingent on which set of screening guidelines are adopted. Multiple US medical organizations have issued screening guidelines that vary along several dimensions, including the recommended age of initiation and screening interval. 20 On one end of the spectrum, the American College of Radiology (ACR) 21 recommends that women of average risk for breast cancer should initiate annual screening mammography at age 40 to maximize life-years gained. The National Comprehensive Cancer Network (NCCN) 22 and the American Society of Breast Surgeons (ASBrS) 23 have issued similar recommendations.
On the other end of the spectrum, the US Preventive Services Task Force (USPSTF) published a "B" recommendation (i.e., "moderate to substantial net bene t") for women of average risk to initiate biennial screening mammography at age 50. 24 The USPSTF issued a lower level "C" recommendation (i.e., "small net bene t") for women ages 40-49, citing the need to balance the bene ts of screening against the potential harms of overdiagnosis and overtreatment. 24 Other organizations, such as the American Cancer Society (ACS), have issued recommendations that fall somewhere between the ACR and USPSTF guidelines (i.e., annual mammography initiated at age 45 before transitioning to biennial screening beginning at age 55). 25 Ensuring equitable access to mammography facilities under the ACR relative to the USPSTF guidelines would likely require signi cantly greater mammography screening capacity given the earlier age of initiation and shorter screening interval, particularly in rural and other disadvantaged areas where access is typically limited.
Beyond the existing screening guidelines, increasing awareness of racial differences in the age distributions of breast cancer incidence and mortality has called for action to advance health equity.
While some organizations, such as the ACR and ASBrS, have called for formal lifetime breast cancer risk assessment by the age of 30 for Ashkenazi Jewish and Black/African American women to identify those who would bene t from risk reduction strategies including earlier screening with mammography and/or MRI, some have suggested establishing race-based imaging guidelines. 26,27 Race-based guidelines refer to screening schedules based on a patient's race. It has been argued that the USPSTF guidelines contribute to racial disparities and should be speci cally modi ed to recommend screening initiation at age 40 for Black women. 26 As noted, the ACR and other organizations do currently recommend screening for all women beginning at age 40, regardless of race. However, under the Affordable Care Act, private insurers and Medicaid are only required to cover preventive services recommended by the USPSTF at the B grade or higher. 28 Thus, the USPSTF recommendations may impact access to services. In addition, evidence has shown that the current USPSTF guidelines have led to a decrease in clinicians recommending mammography to younger Black women. 29 The USPSTF is currently in the process of updating their guidelines and is explicitly considering the key question of whether the bene ts and harms of screening differ by population characteristics, including race. 30 Addressing this speci c question, a recent simulation modeling study evaluated how the USPSTF screening mammography guidelines could be made more equitable for Black women in the US. 31 Simulation modeling was required because Black women have been historically underrepresented in screening trials, precluding analyses strati ed by race. 32 The authors reported that initiating biennial screening for Black women beginning at age 40 would achieve the same bene ts of biennial screening beginning at age 50 observed for White women, which could reduce the Black-White difference in breast cancer mortality by 57%. 31 It should be noted that race-based approaches to medicine have been critiqued on multiple ethical and pragmatic grounds. [33][34][35] Nevertheless, if the USPSTF was to utilize a race-based approach when updating their screening guidelines to address the breast cancer disparity observed for younger Black women, instead of adopting more inclusive guidelines recommended by the ACR/NCCN/ASBrS, this could have implications for the allocation of mammography resources.
In addition to allocating mammography facilities on the basis of screening guidelines, more recent evidence has pointed to the importance of mammography facility quality. 36,37 Quality measures for mammography facilities include academic setting, mammograms being read exclusively by breast imaging specialists, and the availability of digital mammography. 36 Designation as a Breast Imaging Center of Excellence (BICOE) by the ACR 38 has been used in prior research to understand the link between comprehensive assessments of breast imaging facilities and racial disparities in stage at diagnosis. 37 Importantly, community-based programs designed to equitably improve access to high-quality screening mammography facilities have been shown to meaningfully reduce disparities in breast cancer mortality. 39,40 This study had two objectives related to evaluating and improving equitable access to screening mammography facilities in Delaware. We focused our analyses on Delaware because it is broadly representative of the US in terms of race and urban-rural characteristics 41,42 and has among the highest incidence rates in the US for breast cancer among younger Black women 43 and triple negative breast cancer, 44 an aggressive subtype of breast cancer that is more likely to present at a younger age. 45 In addition, Delaware has the cancer care infrastructure necessary to implement population-level prevention programs and a track record of eliminating other cancer disparities with improved screening programs. 46 The rst objective was to conduct a statewide catchment analysis of mammography access. The catchment analysis described the location of mammography facilities, including BICOE-accredited facilities, and evaluated whether these facilities were patterned by sociodemographic characteristics. The second objective was to conduct a location-allocation analysis to identify candidate locations for the establishment of new mammography facilities to optimize equitable access according to the existing ACR guidelines and to the USPSTF guidelines with or without race-based considerations.

Data sources
Census tract measures of population size; number of women aged 40-49, 50-74, and older than 74; percentage of women who are Black; area deprivation; and percentage of households with at least one vehicle were obtained from the U.S. Census Bureau's American Community Survey 5-year estimates. 47 Area deprivation was operationalized as a Z-score composite of education, employment, income and poverty, and household composition, where a higher score indicates greater deprivation. 48 The per tract number of bus stops were obtained from the Delaware open data portal. 49 Mammography facilities were compiled from two sources, the U.S. Food and Drug Administration certi ed facility list 50 and the American College of Radiology's accredited facility list, 51 the latter resource also identifying whether a facility was a BICOE. We retrieved all sites in Delaware as well as sites within border-adjacent ZIP codes in Maryland and Pennsylvania recognizing that Delaware residents may cross state lines for mammography services. For sites in Delaware, the number of active mammography units per site was obtained from the Delaware Department of Health and Social Services. This information was used to estimate site capacity, or how many screening mammograms a site could perform each year. Using a calculation by Young and colleagues, 52 and the de nition of maximum capacity of three mammograms per mammography unit per business hour, 53 capacity of a facility with one mammography unit was estimated at an average of 4,500 screenings per unit per year. For sites with more than one mammography unit this value was multiplied by the number of units at that site. Lastly, census tract and county boundary de nition shape les were downloaded from the U.S. Census Bureau. 54 Across Delaware's three counties, we explored heterogeneity by tracts inside or outside of New Castle County (i.e., Kent and Sussex counties), as New Castle County is more urban and contains the relatively densely populated city of Wilmington, while Kent and Sussex counties tend to be more rural (see Supplemental

Statistical analysis
First, we used descriptive statistics to summarize statewide and county-speci c census tract measures of population, transportation (a proxy for accessibility), and mammography sites. We then performed a catchment analysis using the ArcGIS Pro Service Area analytic tool to identify areas within 15, 30, and greater than 30 minutes driving time from each existing mammography site, with driving time serving as an indicator of geographic access. This Service Area analytic tool calculates the maximum driving distance from a point that can be traveled along a road network, 55 representing a service catchment area.
As part of this catchment analysis, Poisson regression models predicted the independent census tract correlates (enumerated earlier) of the number of mammography facilities, units, BICOE facilities, and BICOE units statewide and separately for New Castle County and Kent and Sussex Counties. All independent variables were standardized for modeling. These ecological models included the census tract population as an offset to account for population differences and the estimates may be interpreted as relative risks per standard deviation change with corresponding 95% con dence intervals (CIs).
Next, we used the ArcGIS Pro Location-Allocation analytic tool to identify candidate locations for the placement of additional mammography sites in Delaware. This Location-Allocation tool uses heuristic procedures to identify locations for services based on location-speci c demand. 56 Within this analysis, the services are mammography screenings, and demand represents the people eligible to receive these services according to screening guidelines. Three competing speci cations of demand were used in these analyses. The primary speci cation was based on the USPSTF recommendation of biennial mammography screenings for women aged 50 to 74 years. 24 The second speci cation was based on a simulation study by Chapman and colleagues that recommends initiating biennial screening in Black women at age 40, in addition to the USPSTF's recommendation of biennial screening for all women aged 50-74 years. 31 The third speci cation was based on the ACR recommendation of annual mammograms for women age 40 and older. 21 These three speci cations of demand were calculated for each census tract and represented the number of women who would be eligible to receive screening mammography each year. Population-weighted centroids were calculated for each census tract and represented the location of the women residing in that census tract (i.e., location of demand). New mammography sites could be placed anywhere within the census tracts, with candidate locations created using the ArcGIS Pro Fishnet tool. To identify locations of new mammography sites that would ll in the gaps in demand that the current mammography sites are unable to reach, the location-allocation analysis took into consideration the locations and the capacities of the existing mammography sites. This was achieved using the Maximize Capacitated Coverage problem type, which selects candidate sites such that the maximum amount of demand is served without exceeding the capacity of the sites. 57 Candidate sites were assumed to have a single mammography unit with the same capacity as existing facilities with one unit. For all demand speci cations, the location-allocation analyses were run three times, allowing for the addition of one, three, or ve new mammography sites. Driving time from demand points to mammography sites was used to determine appropriate location allocation, and a cut-off of 20-minutes driving time was speci ed as the maximum amount of time an individual would likely travel to a site.
Finally, we used the Location-Allocation tool to identify existing sites that might bene t from a conversion to a BICOE. This analysis used the primary USPSTF demand speci cation and all other parameters used in the previously described location-allocation analyses, with one difference: only BICOE sites and their capacities were used as the existing locations, while all non-BICOE sites were speci ed as candidate locations.
All analyses were conducted in R version 3.6. 3

Catchment Analysis
The majority of Delaware's population lives in the northernmost county, New Castle County, which encompasses 130 (61%) census tracts (Supplemental Fig. 1). Results of the service area analysis are illustrated in Fig. 1, showing Table 1). Fourth, we also noted opposing associations for the transportation predictors, where a greater proportion of households with at least one vehicle was associated with a decreased rate of mammography facilities and units, and a greater number of bus stops was associated with an increased rate of mammography facilities and units.

Location-allocation analysis
Results of the location-allocation analysis using the USPSTF demand speci cation, the race-based speci cation, and the ACR speci cation are depicted in Figs

Discussion
In a catchment analysis of mammography access in Delaware, the state with among the highest rates of breast cancer among younger Black women in the US, we observed what initially appeared to be adequate access to screening. In New Castle County, the most populous county in the state, 98% of the population lived within a 15-minute drive of a mammography facility. In the other, more rural two counties in the state, 78% and 98% of the population lived within a 15-minute and 30-minute drive of a facility, respectively. Across all racial groups, we observed a positive relationship between the number of younger women (i.e., 40-49 years) and the number of mammography facilities/units and BICOEs statewide. We did not observe signi cant associations between the number of women currently eligible for screening mammography under the current USPSTF guidelines and measures of mammography access, with the exception of a signi cantly decreased number of units relative to the number of women 50-74 years in New Castle County census tracts.
When mammography access was considered through a health equity lens, we found preliminary evidence suggestive of disparities related to race and rurality. For every standard deviation increase in the number of Black women in a census tract, there were 64% fewer mammography units statewide. In New Castle County, the county with the largest Black population in the state, we observed 84% fewer units for every standard deviation increase in the number of Black women in a census tract. This nding was even stronger for BICOE units: for every standard deviation increase in the number of Black women in a census tract, there were 85% and 98% fewer BICOE units statewide and in New Castle County, respectively (with similar results observed for facilities). Fewer mammography facilities and units in predominantly Black census tracts points to a potential disparity in screening access. Regarding disparities by SES, we did not nd a signi cant association between area deprivation and the number of mammography facilities or units in New Castle County, Kent and Sussex Counties, or statewide. Regarding disparities by rurality, the number of statewide facilities and units were proportional to the population size for New Castle County and Kent and Sussex Counties. However, while 100% of the census tracts in New Castle County were within a 30-minute drive of a mammography facility, two census tracts in the southern part of Delaware had drive times greater than 30 minutes. In addition, the more rural counties in the state accounted for 41% of the population but only 22% of the BICOEs.
The results of the location-allocation analysis using the USPSTF demand speci cation highlighted the opportunity to increase access in the more rural, southern part of the state. When adding ve additional mammography sites, four were proposed for the southern part of the state and one in New Castle County.
When ve existing non-BICOE mammography facilities were considered for conversion to BICOE sites, four were identi ed in the southern part of the state and one in Wilmington, the largest city in the state.
This nding is consistent with other research, which has found that among the greatest disparities in the geographic access to mammography facilities exist in small towns and rural areas. 12,52,58 When the results of these analyses are considered for the USPSTF guidelines with race-based screening demand speci cations, three additional sites were proposed for the southern part of the state and two additional sites were proposed for New Castle County in areas that have larger numbers of younger Black and other minority women. Finally, under the ACR demand speci cation, all ve new mammography sites were proposed for the southern, more rural part of the state.
These results illustrate that decisions about allocating mammography screening resources are impacted by which set of screening guidelines are adopted. Adopting ACR guidelines, which recommend all women initiate annual screening mammography beginning at age 40, would lead to a greater focus on improving access in rural areas. The USPSTF guidelines would lead to a similar allocation, albeit with a small shift in allocation to more populous areas. Adopting the USPSTF guidelines inclusive of a race-based approach to screening would lead to a greater allocation of mammography resources to more populous and racially diverse geographic areas.
This study, which to our knowledge represents the rst location-allocation analysis of geographic access to screening mammography under multiple screening guideline demand speci cations, highlights the potential need to increase access to screening mammography for younger Black women and in rural areas. This study has several limitations. First, our analyses focused only on Delaware and ndings may not apply to other states or geographic regions. 56 Delaware has notably elevated rates of breast cancer among Black women under age 50, 43 including rates of more aggressive subtypes of breast cancer, 44 and therefore represents an important state in its own right to assess. Other states with similar pro les that may warrant similar assessments include those that overlap with the lower Mississippi Delta Region. 59 Beyond racial disparities, this study did not examine mammography access for other high-risk groups (e.g., Ashkenazi Jewish women). 60 Second, drive time represented our proxy for accessibility. For women accessing mammography facilities via other means (e.g., public transportation) and for whom other barriers limit access (e.g., hours of operation, insurance, childcare), 61 our analysis may not fully capture these complex patterns. For example, while ownership of a vehicle was more limited in the urban areas of New Castle County, the number of bus stops was greater; one federally quali ed health center in Wilmington previously noted that over 50% of their patients rely on busses for transit. 62 Therefore, future studies of access should consider the time it would take to reach a mammography site via public transportation, as well as measures of other types of barriers, and mammography facility capacity. This research could inform the development of other interventions designed to close disparities in access to screening mammography, such as community outreach and transportation.
A third limitation of this study was the use of BICOE designation as a quality measure. Prior research found that breast cancer diagnoses made at BICOE-designated facilities are less likely to be a later stage, 37 but it remains unclear what explains this association. BICOE designation requires ACR accreditation in mammography and stereotactic biopsy, breast ultrasound and ultrasound-guided biopsy, and breast MRI and MRI biopsy or the ability to refer the patient for MRI/MRI biopsy to another facility with a referral relationship. Therefore, an ACR accredited mammography unit at a BICOE facility is not necessarily of higher quality than an ACR accredited unit at a non-BICOE facility. It may not be necessary, let alone feasible, to convert a mammography facility to a BICOE to improve access to mammography. There is also not an established relationship between BICOE-designated facility and radiologist characteristics. Separate research reported a relationship between radiologist characteristics (i.e., quali cations, a liation, and experience) and false-negative rates (i.e., missed breast cancer detection), particularly for racial/ethnic minorities and lower-income women. [63][64][65] To conclude, drawing on the de nition that health disparities represent potentially avoidable differences in disease outcomes, 66 ensuring equitable geographic access to high-quality screening mammography facilities could help to close breast cancer disparities observed by race and rurality. However, making decisions about how to allocate mammography resources to ensure equitable access is contingent on which set of breast cancer screening guidelines are adopted, among other considerations (e.g., certi cate of need). Given a set of guidelines, catchment and location-allocation analyses can guide the selection of locations for new mammography facilities and represent important methodological tools that can be leveraged in service of health equity. Future studies should collect additional data on access, quality, and capacity across geographic areas and population subgroups to facilitate the generation of more nely tuned and potentially impactful recommendations for the allocation of mammography facilities.

Declarations
Ethics approval and consent to participate: Not applicable. This study did not involve the use of any animal or human data or tissue.