Applying a Novel Measure of Community-Level Healthcare Access to Assess Breast Cancer Care Timeliness

Abstract

Background: Geographic disparities in breast cancer outcomes exist. Few studies have examined community- and health system-level factors associated with care timeliness, an important measure of care quality.

Methods: The Carolina Breast Cancer Study is a population-based cohort of 2,998 women with invasive breast cancer (2008-2013). Using latent class modeling, patients' census tracts of residence were characterized by healthcare accessibility and affordability. Centers for Medicare and Medicaid Services ratings were used to classify hospitals as low- or high-quality. Six timeliness outcomes were assessed: (i) lacking prediagnostic regular care, (ii) being underscreened, (iii) late-stage diagnosis, (iv) delayed treatment initiation, (v) prolonged treatment duration, and (vi) lacking receipt of Oncotype DX genomic testing. Associations of geographic accessibility, healthcare affordability, and hospital-level quality with care timeliness were evaluated with relative frequency differences (RFD) and 95% confidence intervals (CI).

Results: Compared with "high-accessibility, high-affordability" census tracts, patients residing in "low-accessibility, low-affordability" areas were more likely to be underscreened (RFD = 18.7%, CI, 13.0, 24.3), have late-stage diagnosis (RFD = 6.2%, CI, 2.4, 10.1), and experience prolonged treatment (RFD = 6.9%, CI, 1.4, 12.3). "High-accessibility, low-affordability" areas had the highest frequency of treatment delay (RFD = 9.3%, CI, 3.9, 14.7). Initial surgery at a high-quality facility was associated with less delayed treatment (RFD = -3.9%, CI, -7.5, -0.4) and prolonged treatment (RFD = -5.9%, CI, -9.9, -1.9).

Conclusions: Community- and health system-level factors were associated with timely breast cancer care.

Impact: Policy efforts to improve access in communities should consider multiple dimensions of access, including geospatial accessibility and affordability.

Figure 1.. Map of healthcare access latent class membership for census tracts in North Carolina, USA.

Census tracts were classified using publicly available data representing multiple dimensions of healthcare access. Areas with low geographic accessibility are shown in blue and areas with high geographic accessibility are shown in red. Darker shade represents low healthcare affordability; lighter shade represents high healthcare affordability.

Figure 2.. Frequency of care delay outcomes in relation to community healthcare access.

Community access groups were defined by a latent class model in which North Carolina, USA census tracts were classified using publicly available data representing multiple dimensions of healthcare access. Community access groups were then evaluated in relation to six patient-level outcomes A) lacking pre-diagnostic regular care (N=2,998) B) being under-screened (N=2,045) C) late-stage diagnosis (N=2,995), D) delayed treatment (N=2,974), E) prolonged treatment (N=2,487), and F) not Oncotype DX tested (N=1,615). All regression models were adjusted for age. The delayed and prolonged treatment models were also adjusted for stage at diagnosis and ER/HER2 status.

Figure 3.. Race-stratified frequency of care delay outcomes in relation to community healthcare access.

Community access groups were defined by a latent class model in which North Carolina, USA census tracts were classified using publicly available data representing multiple dimensions of healthcare access. Community access groups were then evaluated in relation to six patient-level outcomes A) lacking pre-diagnostic regular care (N=2,998) B) being under-screened (N=2,045) C) late-stage diagnosis (N=2,995), D) delayed treatment (N=2,974), E) prolonged treatment (N=2,487), and F) not Oncotype DX tested (N=1,615). All regression models were adjusted for age. The delayed and prolonged treatment models were also adjusted for stage at diagnosis and ER/HER2 status. Results are presented stratified for Black and non-Black patients.