Geographic Access to Transcatheter Aortic Valve Replacement Centers in the United States: Insights From the Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy Registry

Key Points

Question  What is the geographic access to transcatheter aortic valve replacement (TAVR) centers in the United States?

Findings  In this observational study, 1 232 568 of 47 527 537 individuals 65 years and older lived in a zip code with a TAVR center, and 43 789 169 (92.1%) lived in a hospital referral region containing a TAVR center. For those who underwent successful TAVR, median driving time to implanting center was 35.0 minutes (interquartile range, 20.0-70.0 minutes).

Meaning  In the context of the US health care system where specialized care is centralized, TAVR services have significant penetration.

Importance  Geographic access to transcatheter aortic replacement (TAVR) centers varies in the United States as a result of controlled expansion through minimum volume requirements.

Objective  To describe the current geographic access to TAVR centers in the United States.

Design, Setting, and Participants  Observational study from June 1, 2015, to June 30, 2017. United States census data were used to describe access to TAVR center. Google Maps and the Society of Thoracic Surgeons American College of Cardiology Transcatheter Valve Therapy Registry were used to describe characteristics of patients undergoing successful TAVR according to proximity to implanting center. The study analyzed 47 527 537 individuals 65 years and older in the United States and 31 098 patients who underwent successful transfemoral TAVR, were linked to fee-for-service Medicare, and had a measurable driving time.

Main Outcomes and Measures  Median driving distance to a TAVR center.

Results  Among 40 537 zip codes in the United States, 490 (1.2%) contained a TAVR center, and among 305 hospital referral regions (HRR), 234 (76.7%) contained a TAVR center. Of the 31 749 patients who underwent successful transfemoral TAVR and were linked to fee-for-service Medicare, 31 098 had a measurable driving time. Mean (SD) age was 82.4 (6.9) years, 14 697 patients (47.3%) were women, and 7422 (23.87%) lived in a rural area. This translated to 1 232 568 of 47 527 537 individuals (2.6%) 65 years and older living in a zip code with a TAVR center and 43 789 169 (92.1%) living in an HRR with a TAVR center. Among 31 749 patients who underwent successful transfemoral TAVR and were linked to fee-for-service Medicare, 31 098 had a measurable driving time. All of these patients (100.0%) underwent their procedure in a TAVR center within their HRR, with 1350 (4.3%) undergoing TAVR in a center within their home zip code. Median driving time to implanting TAVR center was 35.0 minutes (IQR, 20.0-70.0 minutes), ranging from 2.0 minutes to 18 hours and 48 minutes.

Conclusions and Relevance  Most US individuals 65 years and older live in an HRR with a TAVR center. Among patients undergoing successful transfemoral TAVR, median driving time to implanting center was 35.0 minutes. Within the context of the US health care system, where certain advanced procedures and specialized care are centralized, TAVR services have significant penetration. More studies are required to evaluate the effect of geographic location of TAVR sites on access to TAVR procedures among individuals with an indication for a TAVR within the US population.

Higher transcatheter aortic valve replacement (TAVR) procedural volumes are associated with better clinical outcomes, a finding that has been confirmed in an analysis¹ reflecting current techniques and the expansion of TAVR to patients at intermediate surgical risk. Consequently, in an attempt to ensure quality, national coverage decision limits TAVR to institutions meeting minimal TAVR, percutaneous coronary intervention, and surgical aortic valve replacement volume requirements.² However, these volume requisites necessarily exclude some hospitals from offering TAVR services. As a result of these national coverage decisions and individual hospital strategic priorities, geographic access to TAVR sites varies and is generally clustered around urban centers.³ Understanding the importance of the trade-off between maintaining high procedural volumes to ensure quality and geographic accessibility to TAVR centers is important from a public health perspective.⁴ The objective of the study was to describe the geographic availability of TAVR centers in the United States.

This study included nationwide successful transfemoral TAVR cases (June 1, 2015, to June 30, 2017) within the Society of Thoracic Surgeons (STS)/American College of Cardiology (ACC) Transcatheter Valve Therapy (TVT) Registry⁵ linked to US Centers for Medicare and Medicaid Services (CMS) administrative claims. Only patients for whom device deployment was successful, according to a standardized definition,⁶ were considered. Valve-in-valve procedures and aborted procedures were excluded. Because the CMS national coverage decision requires reporting of cases to the TVT Registry, it is expected to include virtually all the commercial procedures performed in the United States. This study was approved by the institutional review board of Duke University and by the Chesapeake Research Review. A waiver of consent was granted for this retrospective analysis using anonymized data.

Geographic distribution of TAVR centers among members of the US population 65 years and older was examined using US census data by determining the proportion of individuals whose zip code matched (1) the zip code of a TAVR center from the TVT Registry and (2) a zip code in a hospital referral region (HRR) containing at least 1 TAVR center. Hospital referral regions represent multi–zip code regional health care markets containing at least 1 hospital with facilities to perform major cardiovascular procedures or neurosurgery (eMethods in the Supplement). Rural status of the patients was determined according to their zip code using the Federal Office of Rural Health Policy data files.⁷

Driving time between TAVR patient residency and their implanting TAVR center was determined using Google Maps on a Sunday at 5 pm (time between TVT center and centroid of patient’s zip code identified through linked CMS claims). Driving times greater than 99th percentile (>20 hours) were thought to likely reflect actual air or train travel and therefore were excluded.

Baseline characteristics are presented as mean (SD) or median (interquartile range [IQR]) for continuous variables, as appropriate, and as counts and percentages for discrete variables. Continuous and discrete variables are compared using Wilcoxon rank sum and χ² tests, respectively. Two-sided P values less than .05 were considered statistically significant. Patient driving times were grouped into quartiles for analysis. Analyses were done at the Duke Clinical Research Institute using SAS, version 9.4.

Among 40 537 zip codes in the United States, 490 (1.2%) contained a TAVR center, and among 305 HRRs, 234 (76.7%) contained a TAVR center. Using US census data, this translated to 1 232 568 of 47 527 537 individuals (2.6%) aged 65 years and older living in a zip code with a TAVR center, and 43 789 169 (92.1%) living in an HRR with a TAVR center (Figure).

Among 31 749 patients who underwent successful transfemoral TAVR and were linked to fee-for-service Medicare, 31 098 had a measurable driving time. All of them (100.0%) underwent their procedure in a TAVR center within their HRR, and 1350 (4.3%) underwent the procedure in a TAVR center within their home zip code. Median driving time to implanting TAVR center was 35.0 minutes (IQR, 20.0-70.0 minutes), ranging from 2.0 minutes to 18 hours and 48 minutes. Mean (SD) age was 82.4 (6.9) years, 14 697 patients (47.3%) were women, and 7422 (23.87%) lived in a rural area (Table 1). Baseline characteristics based on quartiles of driving time are presented in Table 1 and are notable for a significantly larger percentage of rural patients in the highest driving time quartile.

Median time to a TVT site in the Midwest and in the South regions was significantly longer than for the other regions (Midwest: 40 minutes vs 34 minutes, respectively; P < .001; South: 37 minutes vs 34 minutes, respectively; P < .001) (Table 2). The Northeast region reported a significantly shorter median driving time than all other regions (31 minutes vs 37 minutes, respectively; P < .001). Proportion of TAVRs done at urban TVT sites, university hospitals, and larger hospitals increased with higher driving times.

In this study of geographic access to TAVR among individuals in the United States 65 years and older, 43 789 169 (92.1%) live in an HRR containing a TAVR center, and all successful transfemoral procedures were performed within a TAVR center in the patient’s HRR. Additionally, access to TAVR centers varies across US regions. Patients who underwent their procedure in the Midwest or in the South had longer driving times to their TAVR center than in other regions. With the expected growth of the volume of TAVR procedures owing to population aging and to expansion to low-risk patients, our findings have important implications for public health resource allocation and national coverage decisions.

In an attempt to preserve quality, a minimum of 20 annual institutional TAVR procedures (or more than 40 procedures over a 2-year period) is a condition for CMS reimbursement,⁸ justified by the inverse volume-mortality association.¹ Because more than 70% of transfemoral TAVR procedures are performed in sites located within urban areas, as shown in our study, access to postprocedural care at the implanting TAVR center may be limited for the approximately 20% of the US population living in rural areas.⁹ Access to care is a complex construct involving geography, socioeconomic status, ethnicity, race, insurance status, patient preferences, and physician-related factors.^10,11 In this analysis, we describe geographic access to TAVR as a function of HRR, which has been used to investigate regional variation in access to and quality of care.¹²

This study has several limitations. Evaluating whether centralization of TAVR services restricts access for patients living remotely would have required data on the prevalence of TAVR-eligible patients (with severe symptomatic aortic stenosis) who were not referred to a TAVR center. These data are not available in any nationwide data set. Additionally, driving time was used as a marker of geographic proximity but may not be reflective of transportation time using public transportation. Driving time on a Sunday evening at 5 pm was reported, which might not reflect driving times during busy periods, especially in urban areas. Finally, we assumed patients were living in the centroid of their zip code because complete addresses were not available.

Most of the US population 65 years and older live within a hospital referral region with a TAVR center. Among patients undergoing successful transfemoral TAVR, median driving time to implanting center was 35.0 minutes. Driving times among patients undergoing successful transfemoral TAVR vary significantly by region of the country. These results suggest that within the context of the US health care system, where certain advanced procedures and specialized care are centralized, TAVR services have significant penetration. More studies are required to evaluate the effect of geographic location of TAVR sites on access to TAVR procedures among individuals with an indication for a TAVR within the US population.

Corresponding Author: Sreekanth Vemulapalli, MD, Duke University Hospital, 2301 Erwin Rd, PO Box 3301, Durham, NC 27705 (sreekanth.vemulapalli@duke.edu).

Accepted for Publication: March 24, 2020.

Published Online: June 10, 2020. doi:10.1001/jamacardio.2020.1725

Author Contributions: Drs Marquis-Gravel and Vemulapalli had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Marquis-Gravel, Cox, Thourani, Vemulapalli.

Acquisition, analysis, or interpretation of data: Marquis-Gravel, Stebbins, Kosinski, Harrison, Hughes, Gleason, Kirtane, Carroll, Mack, Vemulapalli.

Drafting of the manuscript: Marquis-Gravel.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Marquis-Gravel, Stebbins, Kosinski.

Obtained funding: Vemulapalli.

Administrative, technical, or material support: Cox, Gleason.

Supervision: Kosinski, Harrison, Hughes, Thourani, Vemulapalli.

Conflict of Interest Disclosures: Dr Marquis-Gravel received a training grant from the Canadian Institute of Health Research, personal fees from Servier, and honoraria from Novartis. Dr Gleason reported grants from Medtronic and other from Abbott outside the submitted work. Dr Kirtane reported institutional funding to Columbia University and/or the Cardiovascular Research Foundation from Medtronic, Boston Scientific, Abbott Vascular, Abiomed, CSI, Philips, and ReCor Medical. Dr Carroll reported nonfinancial support from Edwards Lifesciences and from Medtronic during the conduct of the study. Dr Mack reported nonfinancial support from Edwards Lifsciences, Medtronic, and Abbott outside the submitted work. Dr Vemulapalli received grants/contracts from the American College of Cardiology, Society of Thoracic Surgeons, National Institutes of Health, Patient-Centered Outcomes Research Institute, US Food and Drug Administration (NEST), HeartFlow, Abbott Vascular, Boston Scientific and consulting/advisory board from Boston Scientific, Janssen, HeartFlow, Premier, Novella, and Zafgen.

Funding/Support: This research was supported by the American College of Cardiology Foundation’s National Cardiovascular Data Registry and The Society of Thoracic Surgeons National Database.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed in this article represent those of the authors and do not necessarily represent the official views of American College of Cardiology Foundation’s National Cardiovascular Data Registry and The Society of Thoracic Surgeons National Database. Dr Kirtane is Associate Editor of JAMA Cardiology, but she was not involved in any of the decisions regarding review of the manuscript or its acceptance.