While transcatheter aortic valve replacement (TAVR) is a rapidly expanding minimally invasive alternative to surgical aortic valve replacement (SAVR), its access might be limited due to proximity to a TAVR center. Among Medicare beneficiaries, real-world driving times from residential zip codes to TAVR and SAVR center zip codes were computed using the Google Distance Matrix Application Programming Interface. Zip code-level sociodemographic correlates of driving times more than 1 hour to TAVR and SAVR centers were computed using generalized linear mixed-effects models. Of 29,089 US residential zip codes, 407 (1.4%) had a TAVR center and 639 (2.2%) a SAVR center. The median driving time to the nearest zip code with a TAVR center (59 min [IQR, 30–96]) was longer compared with SAVR center (44 min [IQR, 24–73]), and driving times were longer in Western and Southern regions compared with the Northeast. A higher proportion of beneficiaries drive over 1 hour to nearest TAVR center (24.3%) compared with SAVR center (13.1%). Zip codes with a higher median age, a higher ratio of Hispanic to White individuals, and outside metropolitan areas were more likely to have driving times longer than 1 hour to the nearest TAVR centers. In conclusion, access to TAVR is consistently lower compared with SAVR centers, particularly in the Western and Southern US. The geographic barrier to access care, particularly among socioeconomically disadvantaged rural communities, requires evaluating the selection process for sites that provide care.
Transcatheter aortic valve replacement (TAVR) has rapidly expanded in its use to treat patients with severe aortic stenosis (AS) following its approval by the US Food and Drug Administration (FDA) in 2011. , TAVR’s geographic expansion across the US has been deliberate and regulated, with specific guidelines for the minimum volume requirements established by the Centers for Medicare and Medicaid Services (CMS) in 2020 Although the goal of this expansion is to ensure high-quality care for the emerging procedure, the frequent care needs both pre- and postprocedure may create a barrier for patients if they cannot access care in their geographic proximity. , On the other hand, there are documented disparities in TAVR across race and sex groups, with lower use and worse outcomes for women and those of Black race and Hispanic ethnicity. ,,, However, it is unclear whether limited access derives the underrepresentation in these groups. Moreover, ensuring equitable access to TAVR remains critical as there have been increased wait times in lower-volume, nonmetropolitan regions with substantial increases in TAVR volumes overall. ,,,
Surgical aortic valve replacement (SAVR) represents the established invasive treatment of aortic valve diseases, whereas TAVR is a minimally-invasive alternative with growing indications. TAVR and SAVR are both viable treatment strategies across all surgical risk categories, with the choice between them guided by patient-specific factors including anatomy, comorbid health conditions, life expectancy, and patient preference. , Nevertheless, access to TAVR may be more limited than SAVR due to CMS requirements, which restrict TAVR to high-volume centers with specialized infrastructure and mandatory participation in national outcome registries. Therefore, SAVR can be a benchmark for evaluating access to aortic valve replacement procedures.
In this study, we compared the geographic accessibility to TAVR centers with SAVR centers in the US based on driving times and distances from residential zip codes to the nearest treatment center. Additionally, we evaluated the association between sociodemographic features and geographic accessibility to TAVR and SAVR centers.
Methods
Data sources
Using the CMS procedures dataset, we identified the zip codes where TAVR and SAVR centers were located in 2020. We identified all TAVR centers providing care to Medicare fee-for-service beneficiaries based on their submission of reimbursement claims with CPT codes 33361–33369. Similarly, we identified all SAVR centers using physician claims with CPT codes 33405–33406 and 33410–33412. We defined the 4 regions of the US (Northeast, Midwest, South, and West) according to the states within the US Census Bureau geographic levels. The study focused on the continental US, including the District of Columbia, and we did not include Alaska, Hawaii, or other islands and territories. We obtained county-level population totals from the US Census Bureau in 2020 to generate figures, as mapping package used in RStudio required county-level data as opposed to zip-code level data. Furthermore, we extracted demographic and socioeconomic distribution at the zip-code level from the 2020 Social Determinants of Health Database, provided by the US Department of Health and Human Services. Finally, we used the Rural-Urban Commuting Area Codes made available by the US Department of Agriculture to determine the urbanicity of zip codes. Please note the Institutional Review Board deemed the study to be nonhuman subject research with regard to proper ethical oversight, and this study was conducted from 2022-2024.
Study outcomes
The study outcomes were the minimum driving time and minimum driving distance from each residential zip code to its nearest zip codes with TAVR/SAVR centers. We first identified every residential zip code in the US using the “zipcodeR” package in R, excluding zip codes of post office (PO) boxes, government and military organizations, and private companies. The ZipcodeR package is a comprehensive R library that provides detailed ZIP code-level data in the U.S., including geographic coordinates, administrative boundaries (city, county, state, FIPS codes), time zones, and support for crosswalks to larger regions. It is widely used for spatial analysis, data enrichment, and regional segmentation. Next, we approximated the driving distances by calculating the Haversine straight-line spherical distance between the centroids of every residential and TAVR/SAVR center zip code. We then input the residential-center zip code pairs as queries to the Google Distance Matrix Application Programming Interface (API) to estimate the real-world driving times and distances, as determined by Google Maps. The Distance Matrix API returns the distance and travel time between each origin and destinations, allowing identification of minimum distances and times from each origin to destination. To improve the efficiency of querying the Google Distance Matrix API, we extracted and paired the 3 TAVR/SAVR center zip codes with the minimum Haversine distances from each residential zip code. These pairs, consisting of each residential zip code and the 3 closest TAVR/SAVR center zip codes, were processed through Google Distance Matrix API to determine the shortest real-world driving time and distance for each residential zip code. To replicate a patient’s experience of needing care at any hour of the day, the time of day and traffic pattern were not specified on the API request, as there is no set time when the procedures are performed or follow-up care is delivered.
Study covariates
We identified zip code-level covariates, including demographics, socioeconomic features, and region of the zip code. We used demographic data representing all individuals residing in each zip code, rather than data limited to patients who underwent the procedure. Our analyses were conducted at the provider level and incorporated population characteristics obtained from Census-based surveys, to reflect the broader community context rather than the characteristics of those who had undergone the procedure. Age was defined as a continuous variable and a percentage of total individuals within each zip code that were either 18–44 years, 45–64 years, and above 65 years. Sex was noted as a percentage of total individuals per zip code that were men or women. Race and ethnicity were defined as a percentage of total individuals in each zip code who were non-Hispanic White, non-Hispanic Black, Hispanic, and others, with the latter including non-Hispanic Asian, non-Hispanic American Indian and Alaska Native, and non-Hispanic Native Hawaiian and Other Pacific Islander. Household income was defined as a continuous variable representing the median residential zip code-level income, and as a percentage of total individuals per zip code with annual income less than $20,000, $20,000–$50,000, and greater than $50,000. Regions were categorized by zip code into the Northeast, Midwest, South, or West. Zip code urbanicity was recorded as a categorical variable from 1 (most urban) to 10 (most rural) and grouped into metropolitan (categories 1–3), micropolitan (categories 4–6), and rural regions (categories 7–10).
Statistical analyses
Driving times and distance are reported as median and interquartile ranges (IQR) across US census regions. Zip code-level sociodemographic characteristics were reported as median and interquartile ranges for continuous variables and numbers and percentages for categorical variables across all zip codes. The median driving times and distances were compared across sociodemographic subgroups and regions using the Kruskal-Wallis rank-sum test.
We categorized residential zip codes based on minimum driving times and distances to TAVR/SAVR centers into 4 groups of 0–30, 31–60, 61–90, and longer than 90 minutes, and 0–20, 21–50, 50–100, and longer than 100 miles, respectively. To assess the zip code-level correlates of poor geographic accessibility to TAVR/SAVR centers, we used generalized linear mixed-effects logistic regression models. For these models, driving time over 60 minutes was defined as the dependent variable, zip code-level attributes of age, sex, race and ethnicity, income, and urbanicity as fixed effects, and the state as a random effect. Zip code-level attributes were included in the model as continuous ratios for age, sex, race, and income subgroups, whereas urbanicity was included as a categorical variable with 3 levels: metropolitan, micropolitan, and rural. The ratios comprised the ratio of individuals aged 65 or above to individuals under 65 years to define the input for age, the ratio of women to men for sex, the ratio of Black, Hispanic, and other race/ethnicity to White race population for race/ethnicity, and the ratio of individuals earning under $50,000 to over $50,000 annually for household income. Collinearity among model covariates (race, income, and urbanicity) was assessed using variance inflation factors (VIFs). The statistical analyses were 2-sided, and the significance level was set at 0.05. All analyses were done in R (version 4.2.2).
Ethical statement
Publicly available and de-identified data was used for this study, precluding the need to obtain informed consent.
Results
Availability of TAVR and SAVR across the US
There were 29,089 US residential zip codes with Medicare beneficiaries, of which 407 zip codes had TAVR centers, and 639 zip codes had SAVR centers. Of note, 112 zip codes had only TAVR centers, 344 zip codes had only SAVR centers, and 295 zip codes had both TAVR and SAVR centers. Individuals in zip codes with neither TAVR nor SAVR centers had a lower median population (n = 3,572 [1,054–14,741]) compared with individuals in zip codes with TAVR centers only (n = 27,716 [18,190–36,614]), SAVR centers only (n = 29,562 [19,356–41,222]), and both TAVR and SAVR centers (n = 25,852 [16,922–35,608]). The proportion of White individuals living in zip codes with no TAVR nor SAVR (77.0%) was greater than that of zip codes with SAVR only (64.5%), TAVR only (69.1%), and both TAVR and SAVR (65.4%). The median income for individuals in zip codes with TAVR only was approximately $10,000 higher than that of individuals in zip codes with no TAVR and SAVR, SAVR only, and both TAVR and SAVR ( Table 1 ).
Table 1
Demographic features for zip codes with transcatheter aortic valve replacement and surgical aortic valve replacement centers
| No TAVR nor SAVR | SAVR only | TAVR only | TAVR and SAVR | |
|---|---|---|---|---|
| Number of zip codes | 28,338 | 344 | 112 | 295 |
| Population, median [IQR] | 3,572 [1,054–14,741] | 29,562 [19,356–41,222] | 27,716 [18,190–36,614] | 25,852 [16,922–35,608] |
| Age, median [IQR] | 42 [37–48] | 37 [34–42] | 39 [36–44] | 37 [32–41] |
| Women, (%) | 50.4 | 51.2 | 51.5 | 51.6 |
| Race | ||||
| White (%) | 77.0 | 64.5 | 69.1 | 65.4 |
| Black (%) | 7.6 | 7.1 | 6.5 | 8.0 |
| Hispanic (%) | 3.6 | 10.0 | 7.3 | 8.7 |
| Others (%) | 11.8 | 18.4 | 17.1 | 17.9 |
| Income (USD), median [IQR] | 58,481 [47,118–74,583] | 59,320 [44,890–79,190] | 71,668 [51,030–89,717] | 58,294 [45,981–81,428] |
Abbreviations: IQR = interquartile range; SAVR = surgical aortic valve replacement; TAVR = transcatheter aortic valve replacement.
Florida has the highest number of TAVR centers (n = 69), followed by Pennsylvania (n = 68) and New York (n = 63). California has the highest number of SAVR centers (n = 155), followed by Florida (n = 128), and Texas (n = 95). Wyoming (TAVR, n = 0; SAVR: n = 2) and New Mexico (TAVR: n = 1; SAVR: n = 3) had the lowest number of centers ( Figure 1 ).
Central illustration. Number of (A) transcatheter aortic valve replacement and (B) surgical aortic valve replacement centers across the United States. Driving time to zip codes with (C) transcatheter aortic valve replacement and (D) surgical aortic valve replacement centers across the United States. Abbreviations: SAVR, surgical aortic valve replacement; TAVR, transcatheter aortic valve replacement.
Geographical variations in driving times and distances
The median driving time for individuals to the nearest zip code with a TAVR center was 59 min (IQR 30–96), which was significantly higher than to the nearest SAVR center (44 min [IQR 24–73]; p < 0.001). Median driving time for individuals to the nearest zip code with a TAVR/SAVR center was lowest in the District of Columbia (TAVR: 11 min [8–15], SAVR: 11 min [8–15]) and highest in North Dakota for TAVR centers (170 min [99–234]) and Montana for SAVR centers (166 min [110–220]; Figure 1 ).
Across regions, the median driving time for individuals to the nearest zip codes with TAVR and SAVR centers was lowest in the Northeast (TAVR: 38 min [22–66], SAVR: 33 min [19–55]) and highest in the West (TAVR: 91 min [40–158], SAVR: 65 min [25–123]; Figure 1 ). The median driving time for individuals in the Midwest was 53 min [30–80] to the nearest zip code with a TAVR center and 42 min [24–65] to the nearest zip code with a SAVR center, compared with the South (TAVR: 64 min [34–97], SAVR: 47 min [26–70]). Within all regions, the driving time for individuals to the nearest zip code with a TAVR center was significantly longer than the driving time to the nearest zip code with a SAVR center (p < 0.05; Figure 2 , Table S1 ).
Driving time to nearest transcatheter aortic valve replacement and surgical aortic valve replacement centers across the United States census regions. Abbreviations: SAVR, surgical aortic valve replacement; TAVR, transcatheter aortic valve replacement.
Correlates of increased driving time and distance
Among adults over 65 years of age, 24.3% of individuals lived over 60 minutes away from a TAVR center, compared with 13.1% of individuals for a SAVR center. Moreover, 22% and 11% of men and 21% and 12% of women live over 60 minutes away from a TAVR and SAVR center, respectively. Further, a higher proportion of White individuals would need to drive over 60 minutes to the nearest TAVR (25.2%) or SAVR (13.8%) centers, compared with individuals in other racial and ethnic groups, and is not influenced by region. Finally, individuals in zip codes with a shorter driving time (0–30 minutes) had a higher median income ($71,842 [$54,792–$94,730] for TAVR and $69,521 [$53,348–$91,024] for SAVR centers), compared with individuals in zip codes with a longer driving time (61–90 minutes; $53,750 [$44,688–$64,719] for TAVR and $51,795 [$42,862–$62,125] for SAVR centers; Table 2 ).
Table 2
Driving durations to the nearest transcatheter aortic valve replacement and surgical aortic valve replacement centers across demographic and socioeconomic subgroups
| Duration (minutes) | ||||||||
|---|---|---|---|---|---|---|---|---|
| 0–30 | 31–60 | 61–90 | >90 | |||||
| TAVR | SAVR | TAVR | SAVR | TAVR | SAVR | TAVR | SAVR | |
| Age | ||||||||
| Median [IQR] | 39 [35–43] | 39 [35–44] | 42 [38–47] | 43 [38–48] | 44 [38–49] | 44 [39–50] | 44 [38–51] | 45 [39–53] |
| 18–44 years (%) | 56 | 68 | 23 | 22 | 10 | 6 | 11 | 4 |
| 45–64 years (%) | 53 | 64 | 26 | 25 | 11 | 7 | 11 | 4 |
| >65 years (%) | 50 | 61 | 26 | 26 | 12 | 8 | 12 | 5 |
| Men, (%) | 53 | 65 | 25 | 24 | 11 | 7 | 11 | 4 |
| Race | ||||||||
| White (%) | 47 | 59 | 28 | 28 | 13 | 8 | 12 | 5 |
| Black (%) | 64 | 74 | 20 | 20 | 9 | 5 | 7 | 1 |
| Hispanic (%) | 59 | 74 | 19 | 17 | 7 | 4 | 15 | 5 |
| Asian (%) | 76 | 86 | 16 | 12 | 4 | 2 | 4 | 1 |
| American Indian (%) | 35 | 42 | 18 | 21 | 17 | 15 | 31 | 22 |
| Native American (%) | 60 | 73 | 19 | 18 | 10 | 5 | 12 | 3 |
| Income (USD) | ||||||||
| Median [IQR] | 71,842 [54,792–94,730] | 69,521 [53,348–91,024] | 63,069 [51,607–78,698] | 59,152 [48,534–72,684] | 53,750 [44,688–64,719] | 51,795 [42,862–62,125] | 51,386 [42,066–61,602] | 51,607 [42,045–61,458] |
| <$20,000 | 63 | 75 | 10 | 24 | 11 | 8 | 25 | 15 |
| $20,000–$50,000 | 46 | 54 | 20 | 24 | 19 | 14 | 22 | 8 |
| >$50,000 | 56 | 66 | 28 | 25 | 11 | 5 | 10 | 4 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
