Residential Racial Segregation in Aortic Stenosis Diagnosis and Transcatheter Aortic Valve Implantation Among Medicare Patients

Background Transcatheter aortic valve implantation (TAVI) rates are lower among Black compared with White individuals. However, it is unclear whether racial residential segregation, which remains common in the United States, contributes to observed disparities in TAVI rates. Objectives The purpose of this study was to evaluate the association between county-level racial segregation, and aortic stenosis (AS) diagnosis, management, and outcomes. Methods We identified Black and White Medicare fee-for-service beneficiaries age ≥65 years living in metropolitan areas of the United States (2016-2019). Using the American Community Survey’s Black-White residential segregation index, a measure of geographic racial distribution, we determined segregation in each beneficiary’s county of residence. Using hierarchical modeling, we determined the association between racial segregation and rates of AS diagnosis, TAVI receipt, and 30-day clinical outcomes (mortality, readmission, stroke). Results There were 29,264,075 beneficiaries, of whom 22% lived in a high-segregation county. Among Black beneficiaries, high-segregation county residence was associated with decreased rates of AS diagnosis (OR: 0.97; 95% CI: 0.96-0.98) and TAVI (OR: 0.89; 95% CI: 0.86-0.93) compared with low-segregation county residence. In contrast, among White beneficiaries, high-segregation county residence was associated with higher rates of AS diagnosis (OR: 1.02; 95% CI: 1.02-1.03) and no differences in TAVI (OR: 1.00; 95% CI: 0.99-1.00). Segregation and race were not independently associated with 30-day mortality. Conclusions Among Black Medicare fee-for-service beneficiaries, living in a high-segregation county was independently associated with decreased rates of AS diagnosis and TAVI, an association not seen among White beneficiaries. Residential racial segregation may contribute to racial disparities seen in AS care.

T he development of transcatheter aortic valve implantation (TAVI) has revolutionized the treatment of aortic stenosis (AS) by offering an alternative to surgical aortic valve replacement (SAVR) with comparable or improved outcomes. 1 Yet, numerous studies have found lower rates of aortic valve replacement (AVR), both TAVI and SAVR, among historically disadvantaged and marginalized populations, particularly Black patients. 2,3This has raised significant concerns regarding the equitable distribution of novel technologies.
Structural racism is a major driver of racial disparities in health outcomes, 4 but little is known about its impact on access to new cardiovascular therapies, such as TAVI.A better understanding of the factors contributing to racial disparities, including the role of structural racism on diagnosis and management of AS, is critical to promote health equity.
One of the most evident forms of structural racism is residential segregation, which disproportionately affects Black individuals in the United States. 57][8][9] In addition, health system segregation persists due in part to residential segregation, with outpatient practices and hospitals that serve a high proportion of Black adults being under-resourced and financially constrained. 10ese factors may have major implications for the diagnosis of common cardiovascular conditions like AS, and the delivery of novel cardiovascular therapies such as TAVI, for Black adults.However, the extent to which the intersection between residential segregation and race is associated with disparities in the diagnosis, management, and outcomes of AS remains unknown.
In this study, we aimed to answer the following questions. 1) Is county-level Black-White residential segregation associated with rates of AS diagnosis and TAVI? 2) Does the association between county-level residential segregation and rates of diagnosis and treatment of AS differ for Black compared with White individuals?3) Are county-level residential segregation and race associated with differences in outcomes after TAVI?METHODS STUDY POPULATION.Using the Medicare Provider Analysis and Review (MedPAR) files and institutional outpatient claims from the Centers for Medicare and Medicaid Services, we identified all Medicare fee-forservice (FFS) beneficiaries enrolled between January 1, 2016 and December 31, 2019.We included individuals who were 65 years or older and had a selfreported race of either Black or White.To control for previously identified urban/rural disparities in TAVI, 11,12 we limited the sample to those living in metropolitan statistical areas, based on county of residence during the study period, defined as urban clusters of at least 50,000 people in accordance with the Office of Management and Budget definition. 13neficiaries who lived in a county for which the American Community Survey's residential segregation index (SI) was unavailable, 14 or who had missing county data were excluded.
Patient demographics including age, sex, selfreported race, and dual Medicare-Medicaid enrollment were obtained from MedPAR files.Beneficiaries were considered dual Medicare-Medicaid enrolled if they were also enrolled in Medicaid for at least 1 month during a given year. 15Using data from the Centers for Medicare and Medicaid Services 2015 to 2019 Chronic Conditions Data Warehouse, we identified the presence of 67 chronic conditions for each individual in the study sample. 16,17The study protocol was reviewed and deemed exempt by the Beth Israel Deaconess Medical Center Review Board, given the use of deidentified patient information.
BLACK-WHITE COUNTY SEGREGATION.Our main exposure of interest was the degree of Black-White segregation in the beneficiaries' county of residence, which we determined using the American Community Survey's residential SI. 14 This validated SI measures how evenly Black and White residents are distributed geographically and can range from 0 (complete integration) to 100 (complete segregation). 18This measurement is only available for counties with a Black population >100 (65% of counties).We dichotomized counties into highsegregation or low/moderate-segregation based on having a SI $60 or <60 respectively, a commonly used threshold. 9,19TCOMES.We identified beneficiaries with a diagnosis of AS using International Classification of Diseases-10th Revision (ICD-10) diagnostic codes (I35.0,I35.2, I06.0, I06.2, Q23.0) in outpatient and inpatient encounters as either a primary or a secondary diagnosis. 20We validated the performance of these codes using a transthoracic echocardiographic report dataset linked to Medicare FFS claims and found a sensitivity of 50% and a specificity of 96% for identifying any AS. 21The diagnostic performance of these codes was similar regardless of county segregation (Supplemental Table 1).For those beneficiaries with more than one AS diagnostic code, only the first code was counted.Using ICD-10 procedure codes, we identified beneficiaries who had undergone an AVR with either TAVI (02RF38H, 02RF38Z, 02RF3JH, 02RF3KH, 02RF37H, 02RF37Z, 02RF3JZ, or 02RF3KZ) or SAVR (02RF07Z, 02RF08Z, 02RF0JZ, 02RF0KZ). 22r those who underwent multiple AVRs, only the first procedure was included in our analysis.
Our primary clinical outcome of interest was the 30-day composite of all-cause mortality, ischemic stroke, and all-cause readmission.Mortality and readmission data was obtained from MedPAR files.
Ischemic stroke was defined by previously used ICD-10 diagnostic codes (I63.X, I65.X, or I66.X). 23Our secondary clinical outcomes were the individual components of the composite.Patients who were alive and disenrolled from Medicare within 30 days AVR were excluded from the outcomes analysis.STATISTICAL ANALYSIS.We followed the guidelines for cohort studies as described in the Strengthening the Reporting of Observational Studies in Epidemiology Statement. 24We described the demographic and clinical characteristics of our study population, stratified by county segregation, using means and SDs, or counts and proportions as appropriate.
We calculated the annual rate of AS diagnosis, TAVI, SAVR, and any AVR per 100,000 patient-years.Clinical outcomes were reported as percentages.
We fit a mixed-effect regression to model the rates of AS diagnosis, TAVI, SAVR, and any AVR among all beneficiaries, as a function of county segregation (high vs low/moderate) controlling for beneficiary demographic and clinical characteristics.To assess whether Black adults in high-segregation counties were less likely to be diagnosed with AS or receive AVR than White adults, we also included an interaction term for race and segregation in the models.To account for potential secular trends in outcomes, we included an ordinal time variable, ranging from 0 (year 2016) to 3 (year 2019) in the models.Similar models were used to assess the association between race, racial segregation, and clinical outcomes among those who underwent TAVI, SAVR, and any AVR.
As sensitivity analyses, we repeated the models describe above using SI as a discrete variable composed of 10-point increments.We also repeated these models limited to the cohort of beneficiaries with an established diagnosis of AS.A 2-sided a of 0.05 was used for all statistical tests.All statistical analyses were performed using SAS (version 9.4, SAS Institute).Any AVR 10.7 10.9 9.9 <0.0001 6.5 6.9 5.9 <0.0001 Values are mean AE SD or %.
AS ¼ aortic stenosis; AVR ¼ aortic valve replacement; SAVR ¼ surgical aortic valve replacement; SI ¼ segregation index; TAVI ¼ transcatheter aortic valve intervention.associated with decreased likelihood of the 30-day composite outcome driven by a decrease in readmission, but no difference in mortality or stroke.Among Black patients, there was no association between county segregation and 30-day outcomes.Among White patients undergoing SAVR, living in a high-segregation county was associated with decreased 30-day mortality and stroke.This association was not present among Black patients receiving SAVR (Table 4).Black beneficiaries were not independently associated with 30-day outcomes (Supplemental Table 3).
There were no significant interactions between Black beneficiaries and high-segregation residence on 30-day outcomes (P > 0.05 for all outcomes).SENSITIVITY ANALYSIS.In adjusted models using SI as a discrete variable composed of 10-point increments, among Black individuals, increasing segregation was associated with decreased rates of AS diagnosis, TAVI and any AVR, but no difference in SAVR (Supplemental Table 4).In adjusted models for the subset of individuals with an established diagnosis of AS, a similar association to that of the overall cohort was observed between segregation and race with TAVI, SAVR, and any AVR (Supplemental Figures 2 and 3).

DISCUSSION
In this national study, we assessed the association of Black-White residential segregation and individuallevel race with AS diagnosis, management, and outcomes among Medicare FFS beneficiaries.We found that, compared to White adults, Black adults overall were 49% less likely to be diagnosed with AS and 65% less likely to undergo TAVI but had similar outcomes once TAVI was performed.Furthermore, we found that in high-segregation counties, compared to low/ moderate-segregation counties, Black adults were less likely to be diagnosed with AS and receive TAVI whereas White adults were more likely to be diagnosed with AS but had no difference in rates of TAVI (Central Illustration).
Black-White disparities in the diagnosis and management of AS have been previously described 11,25 ; however, addressing these disparities effectively requires a more complete understanding of their multifactorial etiology.While racial residential segregation, a common form of systemic racism, has been identified as a fundamental cause of Black-White health disparities, 26 this study is the first to explore its association with AS care.Residence in segregated neighborhoods is closely correlated with poverty and disproportionately affects Black people independent of their income. 278][9]25 Racial segregation has also been found to be associated with decreased access to health care, which may also lead to the decreased detection and treatment of severe AS. 28 Our study found an increased diagnosis of AS  Residential Racial Segregation AS Diagnosis and TAVI J U L Y 2 0 2 3 : 1 0 0 4 1 5 uninsured and lower rate of private health coverageamong Black people may result in decreased rates of disease identification, surveillance, and treatment; however, racial disparities in health care access remain after controlling for differences in insurance coverage. 29For Black individuals that are insured and able to access care, cultural differences and distrust in the medical system may cause this population to be less likely to seek medical care and can shape their medical decision-making. 2,30This distrust may be further increased by physician factors such as lack of culturally competent care and subconscious bias. 31rthermore, competing demands, particularly among those living in high-segregation counties, such as housing instability, food insecurity, and lack of job flexibility, can lead to delayed care or refusal to pursue interventions once offered. 32,33ong health care-system factors, Black individuals have been found to receive lower quality primary care, be less likely to have board-certified primary care providers, and be less likely to have access to high quality diagnostic imaging, 34 which in turn may preclude initial diagnosis of AS.Once diagnosed with valvular disease, Black patients are less likely to have guideline-recommended surveillance echocardiograms 35 and to be referred to a cardiologist or a cardiac surgeon. 32,36,37While Black patients tend to live in closer geographic proximity to higher-quality hospitals than White patients do, they are more likely to receive surgery at low-quality hospitals. 38Furthermore, metropolitan zip codes with high proportion Black patients have been found to have lower rates of TAVI despite having geographic proximity to TAVI-capable hospitals. 39It is likely that these factors are further exacerbated among those living in high-segregation counties as their options may be limited to receiving care at hospitals primarily serving people of color which are more likely to be under resourced. 10These factors likely have a cumulative effect as evidence by our finding of a larger magnitude association between residence in a highsegregation county and TAVI than with AS diagnosis.

FIGURE 1
FIGURE 1 Association Between County Segregation and AS Diagnosis/Management Stratified by Race

among
White individuals in high-segregation counties, and decreased diagnosis of AS and treatment with TAVI among Black individuals in high-segregation.Viewed through the lens of intersectionality, our findings suggests that race and residence in a segregated county intersect in a cumulative way that results in Black adults living in high-segregation counties having the highest risk of decreased access to AS care and resultant underdetection and undertreatment of disease.Our study identifies an intersectional disparity in the diagnosis and management of AS, in which there is an increased diagnosis of AS among White individuals in high-segregation counties, and decreased diagnosis of AS and treatment with TAVI among Black individuals in high-segregation.This suggests Black adults living in high-segregation counties are a population at high risk of decreased access to AS care and resultant underdetection and undertreatment of disease.A useful framework to understand the factors contributing to our findings is proposed by Batchelor et al, 2 who categorized the barriers contributing to AS disparities into patient-related factors, health care/system factors, and disease related factors.Previous studies have found a lower prevalence of AS among Black individuals; however, the absence of a compelling biological mechanism to explain racial differences in the development and progression of AS raises the possibility of bias in these observations and underscores the importance that patient and health care factors may have on AS diagnosis and management. 2,20Among patient-related factors, lower rates of insurance coverage-including higher rate of CENTRAL ILLUSTRATION Association Between Residential Segregation and AS Diagnosis/ Management/Outcomes Sevilla-Cazes J, et al.JACC Adv.2023;2(5):100415.Black people living in high-segregation counties, compared to low/moderate-segregation counties, have a lower likelihood of aortic stenosis diagnosis and transcatheter aortic valve implantation, but no difference in 30-day outcomes.In contrast, White people living in highsegregation counties, compared to low/moderate-segregation counties, have higher likelihood of aortic stenosis diagnosis, no difference in transcatheter aortic valve implantation, and improved 30-day outcomes.AS ¼ aortic stenosis; OR ¼ odds ratio; ref ¼ reference; TAVI ¼ transcatheter aortic valve intervention.Sevilla-Cazes et al JACC: ADVANCES, VOL. 2, NO. 5, 2023 Residential Racial Segregation AS Diagnosis and TAVI J U L Y 2 0 2 3 : 1 0 0 4 1 5 may still be unmeasured confounding.Sixth, many clinical factors which we were unable to measure go into the selection of TAVI vs SAVR.Seventh, we determined exposure to a high-segregation county based on county at the time of Medicare FFS enrollment and are unable to account for the cumulative effects of living in a high-segregation county which can persist after moving to a less segregated county. 6Eight, our cohort included only Medicare FFS beneficiaries; we are unable to assess the role of health insurance, or generalize our findings to individuals with Medicare advantage, private insurance, or the uninsured.Finally, we focused our analysis on Black-White disparities, and we did not assess the mechanisms driving disparities in AS diagnosis and TAVI among other historically disadvantaged populations.CONCLUSIONS In this Medicare FFS population, we found that living in a high-segregation county was independently associated with decreased odds of AS diagnosis and TAVI for Black, but not White, individuals.This association is likely driven by patient-related and health care-system factors that result in decreased access to quality AS care among Black people living in high-segregation counties.Among individuals who have received TAVI, race was not associated with clinical outcomes.Our study adds to evidence that racial residential segregation is an important form of structural racism which may have a significant impact on the diagnosis and management of AS.Devoting resources to address structural racism as a social determinant of health is critical to addressing disparities in cardiovascular care.

TABLE 1
Baseline Characteristics of Patients With an AS Diagnosis Stratified by Race and Segregation Annual rates of AVR per 100,000 patient-years can be found in Table2.
RESULTSA total of 29,264,075 unique Medicare FFS beneficiaries living in metropolitan areas between 2016 and 2019 were included in the analysis.Among all beneficiaries, 6,496,554 (22%) lived in a high-segregation county.Compared to beneficiaries living in low/ moderate-segregation counties, those in highsegregation counties were more likely to be female (56.1% vs 54.8%), Black (14.5% vs 8.1%), dually enrolled in Medicaid (11.5% vs 8.7%), and have higher rates of comorbidities (Supplemental Table2).Black patients with an AS diagnosis had higher rates of dual enrollment in Medicaid than White patients (29.6% vs 8.8%); however, dual enrollment rates among Black patients was similar regardless of county segregation.The demographic and clinical characteristics, stratified by race and county segregation, for those with an AS diagnosis can be found in Table1.DIAGNOSIS OF AS.Among all beneficiaries, 864,512 (3%) had a diagnosis of AS during the study period.high-segregationcounty,comparedtoalow/moderate-segregationcounty, had a higher likelihood of AS diagnosis (OR: 1.02; 95% CI: 1.02-1.03)asseen in Figure1.The interaction between race and county-level segregation with AS diagnosis was significant (P < 0.0001).AORTIC VALVE REPLACEMENT.likelihood of SAVR (OR: 1.01; 95% CI: 1.00-1.02),butnodifference in any AVR (OR: 1.00; 95% CI: 1.00-1.01).In adjusted models stratified by race, Black beneficiaries living in a high-segregation county, compared to a low/moderate-segregation county, had lower rates of TAVI (OR: 0.89; 95% CI: 0.86-0.93)andanyAVR (OR: 0.94; 95% CI: 0.91-0.97),butnodifferences in rates of SAVR (OR: 1.01; 95% CI: 1.00-1.02).In contrast, White beneficiaries living in a highsegregation county, compared to a low/moderatesegregation county, had no differences in rates of TAVI, SAVR, or any AVR (Figure1).The interaction between race and county-level segregation was significant for TAVI, but not for SAVR or any AVR.occurred in 14.6% of those who received TAVI, 17.5% of those who received SAVR, and 15.7% of those who received any AVR (Table 3).After multivariable adjustment among White patients who received TAVI, living in a high-segregation county, compared to a low/moderate-segregation county, was

TABLE 2
Annual Rate of AS Diagnosis and AVR Per 100,000 Patient-Years Values are rates per 100,000 years AS ¼ aortic stenosis; AVR ¼ aortic valve replacement; SAVR ¼ surgical aortic valve replacement; SI ¼ segregation index; TAVI ¼ transcatheter aortic valve intervention.

TABLE 4
Association Between High-Segregation a and 30-Day Outcomes Post-AVR Stratified by Race Models are adjusted for demographic factors and comorbidities.aRef: Low/moderate segregation.aOR¼ adjusted odds ratio; AVR ¼ aortic valve replacement; SAVR ¼ surgical aortic valve replacement; TAVI ¼ transcatheter aortic valve intervention.usingdiagnostic codes which are highly specific but poorly sensitive and it is likely that we misclassified some beneficiaries resulting in a significant underestimation of disease prevalence.Third, we were unable to determine severity of AS at the time of diagnosis.Fourth, while our models were adjusted for demographic and clinical characteristics, there