Socioeconomic and Geographic Characteristics of Hospitals Establishing Transcatheter Aortic Valve Replacement Programs, 2012-2018

Abstract

Background:

Despite the benefits of novel therapeutics, inequitable diffusion of new technologies may generate disparities. We examined the growth of transcatheter aortic valve replacement (TAVR) in the United States to understand the characteristics of hospitals that developed TAVR programs and the socioeconomic status of patients these hospitals served.

Methods:

We identified fee-for-service Medicare beneficiaries aged 66 years or older who underwent TAVR between January 1, 2012 and December 31, 2018, and hospitals that developed TAVR programs (defined as performing ≥ 10 TAVRs over the study period). We used linear regression models to compare socioeconomic characteristics of patients treated at hospitals that did and did not establish TAVR programs and described the association between Core Based Statistical Area (CBSA)-level markers of socioeconomic status and TAVR rates.

Results:

Between 2012 and 2018, 583 hospitals developed new TAVR programs, including 572 (98.1%) in metropolitan areas, and 293 (50.3%) in metropolitan areas with pre-existing TAVR programs. Compared with hospitals that did not start TAVR programs, hospitals that did start TAVR programs treated fewer patients with dual eligibility for Medicaid (difference of −2.83%, 95% CI −3.78 to −1.89%, p=<0.01), higher median household incomes (difference $2,447, 95% CI $1348 to $3547, p=0.03), and from areas with lower Distressed Communities Index (DCI) scores (difference −4.02 units, 95% CI −5.43 to −2.61, p=<0.01). After adjusting for the age, clinical comorbidities, race/ethnicity and socioeconomic status, areas with TAVR programs had higher rates of TAVR and TAVR rates per 100,000 Medicare beneficiaries were higher in CBSAs with fewer dual eligible patients, higher median income, and lower DCI scores.

Conclusions:

During the initial growth phase of TAVR programs in the U.S., hospitals serving wealthier patients were more likely to start programs. This pattern of growth has led to inequities in the dispersion of TAVR, with lower rates in poorer communities.

Introduction

Despite the potential benefits associated with the development of novel therapeutics, inequitable diffusion of new technologies preferentially to areas with high socioeconomic status may generate or worsen disparities in care and health inequities [1]. Transcatheter aortic valve replacement (TAVR) has revolutionized the treatment and management of aortic stenosis (AS) since it was approved by the U.S. Food and Drug Administration in 2011, expanding the population of patients eligible for valve replacement to include those at unacceptably high risk of mortality from surgical aortic valve replacement (SAVR) and offering a less morbid alternative to SAVR for patients at lower risk [2,3,4,5].

Prior studies have demonstrated that patients living in rural environments are underrepresented among those patients undergoing TAVR, and there are concerns that geographic, racial and socioeconomic factors may contribute to inequities in access to TAVR [6,7,8,9,10]. In fact, TAVR may be particularly sensitive to propagating inequities in care, given the need for multiple treating physicians, extensive specialized pre-procedural testing, and surgical and interventional site-volume requirements for centers seeking to offer this therapy [11]. To address this gap in knowledge, our study sought to answer four specific questions: 1) What are the characteristics of hospitals that developed TAVR programs and where were they located? 2) What are the socioeconomic characteristics of patients served by hospitals that developed TAVR programs compared with those that did not? 3) Are individuals living in areas with TAVR centers more likely to undergo TAVR? 4) Are area level markers of socioeconomic status associated with number of TAVRs performed per resident? To achieve these aims, we used Medicare claims data to further understand the extent of geographic and socioeconomic disparities in access to TAVR in the United States between 2012 and 2018.

Methods

This study was deemed to be exempt by the institutional review board at the University of Pennsylvania. The data that support the findings of this study will not be made available. The methods used will be made available on reasonable request.

Study Cohort

The Medicare Provider Analysis and Review data files and the Master Beneficiary Summary data files were used to identify Medicare fee-for-service beneficiaries aged 66 years or older who underwent TAVR between January 1, 2012 and December 31, 2018. Patients undergoing TAVR were identified using International Classification of Diseases, Ninth Edition, Clinical Modification (ICD-9CM) procedure codes (35.05 and 35.06) and International Classification of Diseases, Tenth Edition (ICD-10PCS) procedure codes (02RF37Z, 02RF38Z, 02RF3JZ, 02RF3KZ, 02RF37H, 02RF38H, 02RF3JH, 02RF3KH) [8,12].

Since the 2012 Centers for Medicare and Medicaid Services National Coverage Determination for TAVR (in force during the study period) requires hospitals starting a TAVR program to offer on-site cardiac surgery, only hospitals with existing cardiac surgery programs were considered candidate hospitals for the development of a TAVR program. Hospitals were considered to have an existing cardiac surgery program if they coded for ≥ 10 major cardiac surgery procedures in 2011. Hospitals that performed ≥ 10 TAVR procedures in a calendar year were defined as TAVR programs for that year and all subsequent years. We chose 10 procedures to minimize the effect of administrative coding errors at the hospital-level.

Geographic Identification

Patient and hospital ZIP code information was obtained from the Hospital Data Claims and Demographic Data files. Patients and hospitals were assigned to Core Based Statistical Areas (CBSAs) using ZIP code information from United States Department of Housing crosswalk files using 2010 Census geographies [13]. CBSAs are distinct geographic areas consisting of an urban center along with surrounding counties socioeconomically linked to the urban center by commuting, as defined by the United States Office of Management and Budget. The Office of Management and Budget defines metropolitan areas as urban clusters of at least 50,000 people and micropolitan areas as urban clusters of between 10,000 and 50,000 people. ZIP codes that were not linked to metropolitan or micropolitan CBSA were defined as rural. We elected to use CBSA as opposed to hospital referral regions in order to better characterize patients by where they reside, work, and live.

Socioeconomic Identification

Socioeconomic status of Medicare fee-for-service patients was defined using three measures. First, we identified median household income based on ZIP code using the Dartmouth Atlas [14]. Second, we assessed dual-eligibility status for Medicaid using the Medicare Denominator files, which has been used previously as a measure of poverty and socioeconomic disadvantage [15]. Third we identified the Distressed Communities Index (DCI) score for each ZIP code using provided crosswalk files for data between 2012 and 2016 [16]. The DCI combines seven economic indicators (percent of population with high school diploma, housing vacancy rate, percent of adults not working, poverty rate, median income ratio, change in employment and change in business establishments) to generate a single index score, with a range from 0 (least distressed) to 100 (most distressed). The DCI is available at the ZIP code level for areas containing at least 50,000 people (missing in 30% of ZIP codes).

Statistical Analysis

New TAVR programs among candidate hospitals with cardiac surgery were identified in each year between 2012 and 2018. The geographic location, by CBSA classification, was identified for each program, as well as whether each CBSA had an existing TAVR program. All ZIP codes in the United States with at least 10 total Medicare fee-for-service beneficiaries aged 66 years or older between January 1, 2012 and December 31, 2018 were identified. We used an age cut-off of 66 years to ensure a 12-month period to assess comorbidities, which were selected empirically based on clinical judgement [17,18]. Individual ZIP codes were assigned to each CBSA; ZIP codes not assigned to a CBSA (identified as “non-core” in crosswalk files) were defined as rural. Clinical and demographic characteristics, socioeconomic indicators, and TAVR procedures among beneficiaries within each CBSA were identified.

Characteristics of New TAVR Programs

Characteristics of candidate hospitals with cardiac surgery programs that established TAVR programs were compared with hospitals that did not establish TAVR programs using Student’s t-test to compare means and Chi-Square analysis to compare proportions, as appropriate. To understand the association between opening of new TAVR programs and hospital characteristics, we estimated multivariable logistic regression models with the opening of a new TAVR program as the dependent variable. Covariates included hospital characteristics (number of beds, geographic region, for-profit status, teaching status), CBSA designation (metropolitan versus micropolitan), an indicator variable for a pre-existing TAVR program within the same CBSA, and an indicator variable for each year of the study. Rural hospitals were excluded from this analysis as they do not have defined CBSAs, and the presence of a pre-existing TAVR program within the same CBSA could not be defined for these hospitals.

Patient Populations Served by New TAVR Programs

Among candidate TAVR hospitals, we identified the first year that a candidate hospital performed a TAVR and became a TAVR program. Indicators for socioeconomic status (median household income, percentage of beneficiaries dual-eligible for Medicaid, mean DCI) were identified for all inpatients treated in the year prior to TAVR adoption based on ZIP code information. We then used three separate linear regression models (one each for median household income, percentage of beneficiaries dual-eligible for Medicaid services, and mean DCI) to compare socioeconomic characteristics of patients served by newly established TAVR programs with those served at candidate hospitals that did not establish a TAVR program, adjusting for each year of the study period. Hospitals were considered non-adopters for each year prior to establishment of a TAVR program. As a sensitivity analysis to understand the role of requiring a cardiac surgical program, TAVR hospitals were compared to all acute care hospitals that did not develop TAVR programs.

Association Between Presence of a TAVR program, Markers of Socioeconomic Status, and CBSA-Level Rates of TAVR

In order to understand the association between the presence of a TAVR program and the number of TAVRs performed during the study period, we used generalized estimating equations. The dependent variable was the number of TAVRs performed per 100,000 Medicare beneficiaries within each CBSA. Demographic, clinical, racial and socioeconomic characteristics of all beneficiaries within the CBSA were included as covariates (Supplemental Table I), as well as an indicator for the presence of a TAVR program. Each of the three indicators for socioeconomic status (median household income, percentage of beneficiaries dual-eligible for Medicaid, mean DCI) were introduced separately as covariates into the model to avoid issues of co-linearity.

Statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC). All statistical testing was 2-tailed, with p-values<0.05 designated statistically significant.

Results

Within the United States, we identified 37,373 unique ZIP codes with at least 10 total Medicare fee-for-service beneficiaries aged 66 years or older between January 1, 2012 and December 31, 2018. Using crosswalk files, 30,492 (81.6%) ZIP codes were assigned to one of 926 defined CBSAs; the remainder were categorized as rural. Descriptive statistics on beneficiaries at the ZIP code level, stratified by CBSA designation, are presented in Supplemental Table II. The mean number of TAVR performed per 100,000 Medicare beneficiaries was 308.2 (SD 421.1) in Metropolitan Zip codes, 320.2 (SD 539.4) in Micropolitan Zip codes and 307.7 (SD 622.1) in Rural Zip codes. Compared with patients living in the 702 CBSAs without TAVR programs, patients living in the 248 CBSAs with a TAVR program had higher rates of TAVR procedures per 100,000 Medicare beneficiaries (334 versus 295, p<0.01).

By identifying beneficiaries who had undergone TAVR, we found 583 hospitals that had established a TAVR program by December 31, 2018, with 554 new TAVR programs opened between January 1, 2012 and December 31, 2018. Over the 7-year study period, 543 TAVR centers opened in metropolitan areas (98.0%), including 293 in areas with pre-existing TAVR programs, 10 (1.8%) in micropolitan areas, and 1 (0.2%) in a rural area. The number of newly established TAVR programs generally decreased each year between 2012 and 2018 (Figure 1). In all but two years (2012 and 2013), the largest number of TAVR centers opened in metropolitan CBSAs with pre-existing TAVR programs.

Figure 1:

Newly established TAVR programs in the United States in metropolitan, micropolitan and rural areas. No TAVR programs opened in micropolitan areas with a pre-existing TAVR center during the study period.

Characteristics of New TAVR Programs

Characteristics of hospitals that established TAVR programs during the study period as compared with characteristics of candidate hospitals that did not establish TAVR programs are presented in Supplemental Table III. On multivariable modeling, hospitals with > 400 beds compared to those < 100 beds (OR 2.73, 95% CI 1.53 to 4.89, p<0.001) and teaching hospitals (OR 2.47, 95% CI 1.93 to 3.15, p<0.001) were more likely to develop TAVR programs (Table 1). Hospitals located in a metropolitan CBSAs were more likely to establish TAVR programs than those in micropolitan CBSAs (OR 2.70, 95% CI 1.40 to 5.19, p<0.001).

Table 1:

The association between the odds of hospitals starting new TAVR programs and hospital factors among hospitals with cardiac surgery programs.

	OR (95% CI)	p-value
Bed Size (<100 beds as reference)
100-399 beds	0.71 (0.40 to 1.26)	0.24
>= 400 beds	2.73 (1.53 to 4.89)	<0.001
Teaching Hospital (Non-teaching as reference)	2.47 (1.93 to 3.15)	<0.001
Metropolitan CBSA (Micropolitan as reference)	2.70 (1.40 to 5.19)	0.003
Region (West as reference)
Midwest	0.60 (0.46 to 0.79)	<0.001
Northeast	0.86 (0.63 to 1.18)	0.35
South	0.82 (0.64 to 1.05)	0.12
Profit Status (Government as reference)
For-profit	1.53 (1.02 to 2.29)	0.04
Non-profit	3.08 (2.21 to 4.30)	<0.001
Pre-existing TAVR Center Within CBSA	0.75 (0.61 to 0.92)	0.006

Patient Populations Served by New TAVR Programs

Compared with candidate hospitals that did not establish TAVR programs during the study period, hospitals that developed a TAVR program treated fewer patients with dual eligibility for Medicaid (difference of −2.83%, 95% CI −3.78 to −1.89%, p<0.001) (Table 2). TAVR hospitals also treated patients with higher median household incomes (difference $2,447, 95% CI $1,348 to $3,547, p<0.001), and patients from areas with lower DCI (difference −4.02 units, 95% CI −5.43 to −2.60, p<0.01) compared with capable hospitals that did not develop a TAVR program. Similar findings were noted when comparing TAVR hospitals to all hospitals that did not develop a TAVR program (Supplemental Table IV).

Table 2:

Difference in socioeconomic characteristics of patients cared for by hospitals that did and did not start new TAVR programs

	TAVR Program (n = 583)	No TAVR Program (n = 577)	Difference (95% CI)	p-value
Dual Eligibility for Medicaid (%)	12.36	15.19	−2.83 (−3.78 to −1.89)	<0.001
Median Household Income ($)	56,756	54,309	2,447 (1,348 to 3,547)	<0.001
Distressed Communities Index (unit)	42.78	46.80	−4.02 (−5.43 to −2.60)	<0.001

Association Between Presence of a TAVR program, Markers of Socioeconomic Status, and CBSA-Level Rates of TAVR

After adjusting for the age, clinical comorbidities, race/ethnicity and socioeconomic status with dual eligibility for Medicaid services, median household income and DCI, areas with TAVR programs had higher rates of TAVR by 9.94% (95% CI 8.80% to 11.08%), 8.02% (6.87% to 9.17%) and 7.25% (6.08% to 8.41%), respectively (Table 3, Supplemental Tables V, VI and VII).

Table 3:

Association between Core Based Statistical Area-level markers of socioeconomic status and rates of TAVR per 100,000 Medicare beneficiaries among the 926 studied CBSA. Additional co-variates are shown in Supplemental Tables IV, V and VI and include age, male sex, race/ethnicity, region of residence, medical comorbidities and an indicator variable for an existing TAVR program within the CBSA.

	% Difference in No. TAVR per 100,000 Medicare Beneficiaries (95% CI)	p-value
Dual Eligibility for Medicaid (per 1% increase)	−1.19 (−1.34 to −1.04)	<0.001
Presence of a TAVR program (adjusted for dual-eligibility for Medicaid)	9.94 (8.80 to 11.08)	<0.001
Median Household Income (per $1000 decrease)	−0.62 (−0.67 to −0.56)	<0.001
Presence of a TAVR program (adjusted for median household income)	8.02 (6.87 to 9.17)	<0.001
Distressed Communities Index (per 1-unit increase)	−0.35 (−0.38 to −0.32)	<0.001
Presence of a TAVR program (adjusted for DCI	7.25 (6.08 to 8.41)	<0.001

When assessing the association between CBSA-level markers of socioeconomic status and rates of TAVR, for each 1% increase in the percentage of patients dual eligible for Medicaid within a CBSA, the number of TAVR procedures performed per 100,000 Medicare beneficiaries was 1.19% lower (95% CI −1.34% to −1.04%, p<0.01). For each $1000 decrease in the median household income within a CBSA, the number of TAVR procedures performed per 100,000 Medicare beneficiaries was 0.62% lower (95% CI −0.67% to −0.56%, p<0.01). For each 1-unit increase in the DCI, the number of TAVR procedures performed per 100,000 Medicare beneficiaries was 0.35% lower (95% CI −0.38% to −0.32%, p<0.01).

Discussion

In the current study, we evaluated the populations served by the growth of TAVR centers in the United States between 2012 and 2018. We found that hospitals in metropolitan areas were more likely to develop TAVR programs, and this growth was predominantly in areas with pre-existing TAVR programs. Furthermore, hospitals that established TAVR programs served patient populations that were less socioeconomically disadvantaged than patient populations at candidate hospitals that did not establish TAVR programs. Finally, we found that CBSAs with TAVR programs had higher rates of TAVR among residents compared with areas without TAVR programs, and CBSAs with higher percentages of patients dual-eligible for Medicaid, lower median household incomes, and more community distress, as assessed by the DCI, had lower rates of TAVR per 100,000 Medicare beneficiaries. These findings highlight geographic and socioeconomic inequities in the growth and dispersion of a novel, high-cost technology.

Though advances in biotechnology have permitted rapid advances in therapeutic options for patients, the initial growth in the availability in procedures may not be equally afforded to all segments of the population. Despite the presence of a novel therapeutic, only some members of the population may have access to the benefits, thereby generating health inequities. For example, during the expansion of cardiac catheterization laboratories in the late 20^th century, increases in the number of sites offering cardiac catheterization and coronary angioplasty did not improve access to these procedures [19]. More recently, concerns have been raised about the allocation of advanced heart failure therapies among women and racial minorities [20].

Access to TAVR remains an important topic, balancing programmatic requirements to initiate and maintain a program that meets threshold quality standards with ensuring complete geographic access to the procedure across the United States [11,21]. Policies such as the National Coverage Determination (NCD) strive to strike this balance; however, given the extensive requirements necessary to establish a TAVR program, these policies may unintentionally affect access to this procedure among some groups of patients in the United States. There exists a tension between attempting to concentrate procedures within high-volume centers of excellence with maintaining complete geographic access of this procedure to all patients in our country. The results of our study to assess where new TAVR programs are established and the patient populations served by these centers suggest that access to TAVR may indeed lead to inequities to care, with four main implications.

First, while the raw number of TAVR sites has grown, TAVR sites are localized to metropolitan areas. Fewer than 5% of new TAVR programs during this time period opened in non-metropolitan areas (micropolitan and rural), and only 1 program opened in a rural area. Further, though presence of an existing TAVR center within a metropolitan area was associated with lower likelihood of a hospital starting a new TAVR program, the majority of new TAVR centers were nevertheless located in metropolitan areas with a pre-existing TAVR program, leading to a numerical expansion in centers over the study period without increasing geographic access to the procedure to the same extent. While the opening of multiple TAVR programs within a metropolitan area may be justified based on a per capita basis, there are nonetheless large areas of the country without ready geographic access to the procedure.

Second, the hospitals that adopted the TAVR procedure served more economically advantaged patient populations. When comparing the patient populations served by hospitals that developed TAVR programs with the patient populations served by hospitals capable of developing TAVR programs but did not, hospitals that developed TAVR programs cared for patients who were less socioeconomically disadvantaged. We found that on average, candidate hospitals that did develop TAVR programs took care of fewer patients who were dual-eligible for Medicaid services by 3%, which is meaningful since almost 20% of the U.S. population uses Medicaid services [22]. Similar statistically and policy-relevant differences were observed in the DCI, whereby hospitals developing TAVR programs cared for patients with a 3-point lower score, which represents a 10% relative difference given a median DCI of 43 in metropolitan areas. Thus, the initial exposure of the procedure within a geographic region benefited the most economically advantaged members.

These findings are likely due to a combination of volume and structural requirements imposed by the NCD that serve as an impediment to rural, low-volume centers, as well as market forces within the health care industry which reinforce strategic decisions by health care executives that generate financial rewards [23]. Due to cost-of-living adjustments in Medicare reimbursements, hospitals serving economically disadvantaged populations may have lower TAVR reimbursement, which may render this procedure financially non-viable for these hospitals and the procedure unavailable for disadvantaged patients [24]. The initial development of a program for a novel therapeutic likely requires significant financial and resource capital, which may not be available at all hospitals, especially those that are safety-net providers or serve a larger Medicaid population where reimbursements may not be as lucrative. For TAVR, a procedure necessitating multiple specialty providers and expensive equipment, these financial forces may significantly impact procedural availability. The combination of these factors may exacerbate pre-existing structural barriers and inequities in health for socioeconomically disadvantaged patients, highlighting the importance of considering these inequities when designing coverage policies and reimbursement strategies.

Third, the presence of a TAVR program led to a significant difference in the number of TAVRs performed per 100,000 Medicare beneficiaries residing within that area. Despite adjusting for clinical comorbidities, race/ethnicity and socioeconomic status, areas with TAVR programs had an almost 10% higher rate of TAVRs performed. Ultimately, the presence of a TAVR program led to a meaningful difference in access to the procedure. Notably, though few programs were opened in micropolitan areas during the study period, rates of TAVR were higher in micropolitan areas as compared to metropolitan areas, after adjusting for clinical characteristics and the presence of an existing TAVR program. Micropolitan areas were smaller, had more community distress and lower median household income, though they did have fewer patients dual eligible for Medicaid. It is possible that these represent small, tightly knit communities where referral networks are comprised of fewer physicians. With education of a relatively small number of referring physicians, robust referral patterns may be generated for the few eligible patients residing in micropolitan areas. Conversely, though metropolitan areas may develop many TAVR programs, outreach and referral patterns may be more difficult to coordinate among the many physicians caring for patient with aortic stenosis.

Fourth, as a result of where TAVR programs have opened since initial FDA approval in 2011, and the types of patients served by these hospitals, the initial growth and development of TAVR has favored areas with wealthier and more privileged patients. CBSAs and ZIP codes with more socioeconomically disadvantaged patients had lower rates of TAVR when compared to areas with more advantaged patients, despite adjusting for age, clinical characteristics and geographic region. Though the prevalence of aortic stenosis may vary in different geographic regions, age has been consistently and significantly associated with the prevalence of aortic stenosis within a population [25,26,27]. Thus, our age-adjusted results likely reflect reduced access to TAVR in areas with patients who are socioeconomically disadvantaged. Further, the geographic availability of TAVR does appear to affect the rates of this procedure, as CBSA with TAVR programs had higher rates of TAVR when compared to CBSA without TAVR programs. Though prior studies have studied the driving distances between patients and TAVR centers [28,29], as we found in this paper, there may be socioeconomic barriers beyond geographic proximity – including referral patterns and the ability to navigate complex care systems – which affect patients’ access to TAVR services.

Since TAVR has demonstrated superiority over existing therapeutic options for aortic stenosis, particularly in patients at unacceptably high risk for SAVR, the limited availability of this procedure for disadvantaged subsets of the population likely contributed to inequities in health. Strategies to address these inequities, whether they result from differences in referral for TAVR, patient willingness to undergo TAVR or provider willingness to perform TAVR are imperative to mitigate resulting healthcare disparities and are important areas of future inquiry. Governing and regulating bodies may consider the equitable diffusion of a technology when approving and outlining the use of novel technologies, considering the barriers in access to care for disadvantaged populations. Notably, the receipt of TAVR services may be only one, small facet of a much larger problem of inadequate access to primary care and diagnostic cardiology services among disadvantaged populations, which are important avenues of future inquiry.

Limitations

Our study has several limitations. First, our study was performed among Medicare fee-for-service beneficiaries and may not be generalizable to patients with Medicare Advantage, which comprise between 25-40% of Medicare enrollees over the study period [30]. Second, the use of an administrative dataset precludes a granular understanding of the reasons why certain hospitals with cardiac surgery programs did or did not establish a TAVR program, which may include reasons such as local specialist expertise, nuanced local market forces or strategic planning decisions. We were not able to account for potential local referral patterns within geographic regions, which may affect access to TAVR among socioeconomically advantaged and disadvantaged populations. Despite the reasons behind adoption or non-adoption of TAVR, geographic and socioeconomic inequities in the types of patients with access to TAVR were observed. Third, as previously mentioned, the true prevalence of aortic stenosis within a population may be poorly understood as routine echocardiographic screening is not performed on a population scale. However, age is strongly correlated with the prevalence of aortic stenosis within a population, and all patient-level analyses in this study were age-adjusted. Fourth, we chose a broader definition of candidate hospitals that would have been identified using minimum volume requirements imposed by the NCD in order to identify all possible hospitals that could develop a TAVR program given the presence of infrastructure to support cardiac surgery. Finally, we used CBSA to define geographic regions. While studies have previously used Dartmouth Atlas Hospital Referral Regions, these areas are explicitly defined by the presence of one or more tertiary medical centers, and thus may not reflect geographic inequities in healthcare access and differences in hospital characteristics. By contrast, CBSAs are defined by commuting and socioeconomic patterns, and may be a better reflection of communities in which patients live and work.

Conclusion

Despite the growth of TAVR programs in the United States over the past decade, this growth has been inequitable – occurring predominantly in metropolitan areas with pre-existing TAVR programs. The presence of a TAVR program was associated with a difference in the number of TAVR performed for residents in that area, accounting for clinical comorbidities, race/ethnicity and socioeconomic status. Further, hospitals serving socioeconomically disadvantaged patients were less likely to start TAVR programs, and rates of TAVR were lower in socioeconomically disadvantaged areas. The unequal introduction of this novel technology appears to be one factor leading to the generation of health care inequities among vulnerable groups.

List of Abbreviations

AS

aortic stenosis

CBSA

core based statistical area

CI

confidence interval

DCI

Distressed Communities Index

NCD

National Coverage Determination

SAVR

surgical aortic valve replacement

TAVR

transcatheter aortic valve replacement