If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address for reprints: Waddah B. Al-Refaie, MD, FACS, MedStar Georgetown University Hospital, MedStar-Georgetown Surgical Outcomes Research Center, Georgetown Lombardi Comprehensive Cancer Center, 3800 Reservoir Rd NW, PHC Building, 4th Floor, Washington, DC 20007.
In this study we sought to evaluate whether disparate use of transcatheter aortic valve replacement (TAVR) among non-White patients has decreased over time, and if unequal access to TAVR is driven by unequal access to high-volume hospitals.
Methods
From 2013 to 2017, we used the State Inpatient Database across 8 states (Ariz, Colo, Fla, Md, NC, NM, Nev, Wash) to identify 51,232 Medicare beneficiaries who underwent TAVR versus surgical aortic valve replacement. Hospitals were categorized as low- (<50 per year), medium- (50-100 per year), or high-volume (>100 per year) according to total valve procedures (TAVR + surgical aortic valve replacement). Multivariable logistic regression models with interactions were performed to determine the effect of race, time, and hospital volume on the utilization of TAVR.
Results
Non-White patients were less likely to receive TAVR than White patients (odds ratio [OR], 0.77; 95% CI, 0.71-0.83). However, utilization of TAVR increased over time (OR, 1.73; 95% CI, 1.73-1.80) for the total population, with non-White patients’ TAVR use growing faster than for White patients (OR, 1.06; 95% CI, 1.00-1.12), time × race interaction, P = .034. Further, an adjusted volume-stratified time trend analysis showed that utilization of TAVR at high volume hospitals increased faster for non-White patients versus White patients by 8.6% per year (OR, 1.09; 95% CI, 1.01-1.16) whereas use at low- and medium-volume hospitals did not contribute to any decreasing utilization gap.
Conclusions
This analysis shows initial low rates of TAVR utilization among non-White patients followed by accelerated use over time, relative to White patients. This narrowing gap was driven by increased TAVR utilization by non-White patients at high-volume hospitals.
Non-White patients were less likely to undergo TAVR than their White counterparts, however, this gap is narrowing because of increased TAVR use by non-White patients at high-volume hospitals.
Inequitable use of TAVR among non-White patients has been previously documented. However, dissemination of this new technology has rapidly expanded to new patient populations at a greater number of hospitals. Although non-White patients historically have had limited access to high-volume hospitals, it is unclear how increased TAVR dissemination would affect trends in its disparate use.
Transcatheter aortic valve replacement (TAVR) was first approved by the US Food and Drug Administration in 2011
and has since revolutionized the treatment of aortic valve disease. Broadening indications for TAVR have rapidly enabled more patients to avoid the early morbidity of a surgical aortic valve replacement (SAVR), and expanded the population of patients amenable to valve replacement.
US Food and Drug Administration FDA approves expanded indication for two transcatheter heart valves for patients at intermediate risk for death or complications associated with open-heart surgery.
Analysis of the Transcatheter Valve Therapy Registry has shown that among the 70,221 patients older than the age of 65 who underwent TAVR from 2011 to 2016, only 3.8% were Black and 3.4% were Hispanic—a significant under-representation compared with their proportion of the population.
it is unclear if this inequity has driven disparate access to TAVR.
In this study we used the State Inpatient Database (SID) from 2013 to 2017 to evaluate whether disparate access to TAVR among non-White patients has decreased over time as the availability of this new technology has expanded. We also sought to quantify the effect of hospital volume on racial disparities to determine if inequitable access to high-volume hospitals is a driver of inequitable TAVR utilization and whether expanded availability resulted in a change over time. A description of this background and objective can also be viewed in video form (Video 1). We hypothesized that ethno-racial inequity has decreased over time as TAVR use has expanded, and that this decreasing disparity is driven by increased utilization at low-volume hospitals.
Methods
Patient Population and Data Sources
Data were collected using SID from 2013 to 2017 from 8 ethno-racially and geographically diverse states (Ariz, Colo, Fla, Md, NC, NM, Nev, Wash). State inpatient data was chosen for its ability to allow linkage with the American Hospital Association Yearly Survey and the Area Resource Health File. Together these 3 merged databases allow for the analysis of patient-level, hospital-level, and county-level data among a set of large and diverse states that represent nearly 20% of the US population. The institutional review board of Lifespan-Rhode Island Hospital approved this study with waived consent (00000396; approved October 31, 2018).
Medicare beneficiaries were included for analysis to isolate those with insurance coverage while minimizing any unmeasured confounding effects of payer status on access to surgery. Patients with aortic valve insufficiency (International Classification of Diseases [ICD] Ninth Revision [-9]: 396.3 and ICD 10th revision [-10]: I35.1, I06.1) were also excluded to ensure a more uniform cohort of those eligible for intervention, though we did not exclude patients on the basis of secondary diagnoses. ICD codes were used to identify patients who underwent TAVR (ICD-9: 35.05, 35.06; and ICD-10: 02RF38Z, 02RF38H) or SAVR (ICD-9: 35.21, 35.22; and ICD-10: 02RF07Z, 02RF08Z, 02RF0JZ, 02RF0KZ). We did not query ICD codes to include or exclude patients on the basis of concomitant procedures. Inclusion and exclusion criteria are shown in a consort diagram (Figure 1), with our final analytic cohort including 51,232 patients; 87.04% were White (n = 43,796) and 12.96% were non-White (n = 6522).
Figure 1Consolidated Standards of Reporting Trials (CONSORT) diagram showing patients included for analysis after applying inclusion and exclusion criteria. SID, State Inpatient Database; TAVR, transcatheter aortic valve replacement; SAVR, surgical aortic valve replacement.
Our main outcome of interest was the rate of TAVR utilization over time. To measure this, we identified the total number TAVRs performed and also calculated the proportion of TAVR compared with total aortic valve procedures (TAVR/TAVR + SAVR). The major independent predictors we included were time, race/ethnicity, and hospital volume. To analyze patients according to race/ethnicity we defined 2 groups, White and non-White, with the non-White category consisting of Black, Hispanic, Asian/Pacific Islander, American Indian, and patients listed as other, on the basis of SID data. Hospital volume was computed as a yearly average according to hospital and was categorized as low- (<50 per year), medium- (50-100 per year), or high-volume (>100 per year) according to total valve procedures (TAVR + SAVR) with cutoffs on the basis of existing literature.
Additional relevant covariates were selected to adjust for patient characteristics (age, sex, admission type, median income, Charlson Comorbidity Index score), hospital characteristics (hospital volume, teaching status), and location characteristics (hospital state, local percent of White population, and provider density). Provider density was defined by the Area Health Resources Files as health professional (physician, physician assistant, nurse practitioner, etc) and computed as the number of providers in the county of the patient's residence per 1000 population. Charlson Comorbidity Index score was included as the standardized comorbidity score available with the SID and has also been previously used for cardiac surgery populations.
Continuous variables are summarized as mean and SD and compared between groups using t test/Wilcoxon rank sum test depending on the distribution of the data, with median and interquartile range reported where applicable. Categorical variables were aggregated as frequencies and percentages and compared using the χ2/Fisher exact test. For modeling, logistic regression was used to assess the overall time trend and disparity between race/ethnicity and hospital volume. A 2-way interaction term was included to assess whether any disparate use of TAVR according to race/ethnicity changed over time and a separate interaction term was included for race/ethnicity and hospital volume to determine if disparity was dependent on volume category. A 3-way interaction was then performed for time, race/ethnicity, and hospital volume to evaluate whether the time trend difference between race/ethnicities relied on hospital volume. The absolute trend and trend differences with 95% CI were computed. Analysis was performed using SAS 9.4 (SAS Institute Inc).
Results
Patient Characteristics
Between 2013 and 2017, 51,232 Medicare beneficiaries underwent either TAVR (39.2%) or SAVR (60.8%). Their mean age was 76.8 ± 8.5 years, 59.7% were male, and 43,796 patients were White (87.0%) compared with 6522 non-White patients (13.0%). Collectively, of all procedures (TAVR and SAVR), 16.1%, 28.7%, and 55.3% were performed at low-, medium-, and high-volume hospitals, respectively; 25.7% were performed at teaching hospitals, and 77.2% of admissions were elective (Table 1). For TAVR procedures only, 5.4%, 28.1%, and 66.5% were performed at low-, medium-, and high-volume hospitals, respectively.
Table 1Descriptive statistics of overall Medicare patients (N = 51,232) who underwent TAVR or SAVR from 2013 to 2017
Value
Median age (IQR), y
77.0 (71.0-83.0)
Sex
Male
30,593 (59.7)
Female
20,632 (40.3)
Race
White
43,796 (87.0)
Black
1974 (3.92)
Hispanic
3396 (6.8)
Asian/Pacific islander
392 (0.78)
American Indian
187 (0.37)
Other
573 (1.14)
Charlson Comorbidity Index scores
0
6957 (13.6)
1-2
25,723 (50.2)
3-4
15,557 (30.4)
≥5
2995 (5.85)
Admission type
Elective
39,513 (77.2)
Other
11,670 (22.8)
Annual hospital volume of TAVR + SAVR
Median (IQR)
123.2 (72.4-218.4)
Hospital volume of TAVR + SAVR according to category
Low (<50 procedures per y)
8231 (16.1)
Medium (50-100)
14,695 (28.7)
High (>100)
28,306 (55.3)
Teaching hospital
Yes
12,971 (25.8)
No
37,383 (74.2)
County-level percent White population 2010
Median (IQR)
75.2 (69.6-83.0)
County-level primary provider density per 1000
Median (IQR)
0.80 (0.60-0.90)
State
Arizona
6079 (11.9)
Colorado
3294 (6.4)
Florida
22,439 (43.8)
Maryland
3573 (7.00)
North Carolina
6635 (13.0)
New Mexico
849 (1.66)
Nevada
1867 (3.64)
Washington
6478 (12.6)
Procedure type
SAVR
31,131 (60.8)
TAVR
20,101 (39.2)
Data are presented as n (%) except where otherwise noted.
White patients were older than non-White patients (77.1 vs 74.7 years), more likely to be male (60.6% vs 54.3%), less multimorbid (Charlson Comorbidity Index score ≥5: 5.6% vs 6.8%), more likely to be admitted electively (78.8% vs 68.2%), and underwent more TAVR (40.0% vs 35.0%); all P values < .001 (Table 2). Examining differences in TAVR and SAVR use according to race for each state individually, we noted substantial differences between each state's White population based on US census data,
and the percentage of White patients who underwent TAVR and SAVR. Results are outlined in Table E1. These data demonstrate considerably greater TAVR use among White patients compared with their percent of each state's population (range, 27.2-37.7 percentage point difference) as well as greater SAVR use (range, 20.7-36.8 percentage point difference).
Table 2Bivariate analysis of Medicare patients (N = 50,318) who underwent TAVR or SAVR from 2013 to 2017 according to race
White (n = 43,796)
Non-White (n = 6522)
P value
Median age (IQR)
77.0 (71.0-83.0)
76.0 (69.0-82.0)
<.001
Sex
Male
26,544 (60.6)
3538 (54.3)
<.001
Female
17,246 (39.4)
2983 (45.7)
Median income in quartiles
Quartile 1 (lowest)
9651 (22.5)
2223 (35.0)
<.001
Quartile 2
11,530 (26.9)
1553 (24.4)
Quartile 3
11,580 (27.0)
1489 (23.4)
Quartile 4 (highest)
10,142 (23.6)
1091 (17.2)
Charlson Comorbidity Index scores
0
6128 (14.0)
701 (10.8)
<.001
1-2
22,101 (50.5)
3209 (49.2)
3-4
13,093 (29.9)
2170 (33.3)
≥5
2474 (5.65)
442 (6.78)
Admission type
Elective
34,468 (78.8)
4448 (68.2)
<.001
Other
9284 (21.2)
2071 (31.8)
Annual hospital volume of TAVR + SAVR in terciles
Low
7122 (16.3)
1044 (16.0)
.001
Medium
12,527 (28.6)
2009 (30.8)
High
24,147 (55.1)
3469 (53.2)
Teaching hospital
Yes
10,471 (24.3)
2289 (36.1)
<.001
No
32,646 (75.7)
4046 (63.9)
County level percent White population 2010
Median (IQR)
77.4 (71.3-84.3)
73.5 (63.6-74.3)
<.001
County level primary provider density per 1000
Median (IQR)
0.8 (0.6-0.9)
0.8 (0.6-0.9)
<.001
State
Arizona
5380 (12.3)
682 (10.5)
<.001
Colorado
3004 (6.86)
240 (3.68)
Florida
18,500 (42.2)
3668 (56.2)
Maryland
2800 (6.39)
501 (7.68)
North Carolina
5862 (13.4)
702 (10.8)
New Mexico
662 (1.51)
175 (2.68)
Nevada
1560 (3.56)
289 (4.43)
Washington
6028 (13.8)
265 (4.06)
Procedure type
SAVR
26,300 (60.1)
4250 (65.2)
<.001
TAVR
17,496 (40.0)
2272 (34.8)
Data are presented as n (%) except where otherwise noted.
Logistic regressions were applied to estimate the odds of receiving TAVR versus SAVR and demonstrate the trend of TAVR utilization over time. The odds of receiving TAVR was 5.12 and 6.80 times higher at medium- and high-volume hospitals, respectively, relative to low-volume hospitals. Across all hospital volumes, the utilization of TAVR increased over time (odds ratio [OR], 1.73; 95% CI, 1.73-1.80). Non-White patients were less likely to receive TAVR compared with White patients (OR, 0.77; 95% CI, 0.71-0.83). Other factors predictive of greater TAVR use included older age, female sex, greater comorbidity, elective status, having surgery at a teaching hospital, and county/state level factors (Table 3).
Table 3Logistic regression of factors predicting TAVR over SAVR among Medicare patients from 2013 to 2017
A 2-way interaction was performed to assess whether any disparate use of TAVR according to race/ethnicity changed over time, followed by a 2-way interaction to assess whether the disparity was dependent of hospital volume. Although the overall use of TAVR increased between 2013 and 2017, interacting race/ethnicity and time showed the trend increased for White (OR, 1.75; 95% CI, 1.72-1.79) and non-White (OR, 1.86; 95% CI, 1.77-1.79) patients, with non-White patient use of TAVR increasing at a 6.00% higher rate than for White patients (OR, 1.06; 95% CI, 1.00-1.12; Table 4).
Table 4Two-way interaction between race/ethnicity versus time on the basis of logistic regression estimates
A separate 2-way interaction was performed to evaluate if racial-ethnic disparity was mediated by hospital volume. In low-volume hospitals, the odds of receiving TAVR was higher for non-White patients compared with White patients by 19% (95% CI, 0.92-1.55). Medium-volume hospitals showed 18% (95% CI, 1.04-1.34) higher odds for non-White compared with White patients. At high-volume hospitals, however, the odds of receiving TAVR was 43% (95% CI, 0.52-0.63) lower for non-White versus White patients (Table 5).
Table 5Two-way interaction demonstrating odds ratio (95% CI) between race/ethnicity versus hospital volume on the basis of logistic regression estimates
To further investigate differences according to hospital volume, we performed a marginal analysis for all racial/ethnic categories instead of categorizing White versus non-White. At low-volume hospitals, Black, Hispanic, and “other” race/ethnicities all had higher odds of receiving TAVR compared with White patients, however all 95% confidence intervals crossed 1.00. The difference that we found in medium-volume hospitals in our 2-way interaction described previously was likely driven by Hispanic patients. The statistically significant difference that we found in medium-volume hospitals in our 2-way interaction described previously was likely driven by Hispanic patients. This population was 25% more likely than White patients to receive TAVR at medium-volume hospitals, whereas all other race categories were less likely than White patients. At high-volume hospitals, all race/ethnicity categories were less likely than White patients to receive TAVR. Full results are listed in Table E2.
A 3-way interaction was then performed to evaluate any time trend differences between race/ethnicity and hospital volume. Results showed that TAVR use increased faster for White patients at low- (OR, 0.95; 95% CI, 0.71-1.28) and medium-volume hospitals (OR, 0.91; 95% CI, 0.83-1.00). At high-volume hospitals, however, TAVR rates increased faster for non-White patients (OR, 1.09; 95% CI, 1.01-1.16; Figure 2; Table 6).
Figure 2Transcatheter aortic valve replacement use increased faster (higher β) for White patients (solid line) versus non-White patients (dotted line) at low- and medium-volume hospitals. However, non-White patients saw a greater increase at high-volume hospitals.
These findings redemonstrate that non-White patients continue to suffer disparate access to TAVR compared with their share of the US population. However, with rapidly expanding use of this new technology, non-White patients’ TAVR use increased faster than White patients, signaling a narrowing of the racial/ethnic gap. We have also shown that non-White patients are less likely to receive TAVR at high-volume hospitals compared with their White counterparts, however, this gap is also narrowing. These trends show that any improvement in racial inequity is likely being driven by decreasingly disparate TAVR utilization at high-volume hospitals, as opposed to broadening use at low- and medium-volume hospitals.
The findings that non-White patients have greater comorbidity, undergo more nonelective operations, and are treated more commonly in teaching hospitals, is all consistent with previous research.
Although it is unsurprising that patients undergoing TAVR experience the same structural differences as patients undergoing treatment for other cardiovascular disease processes, it does pose the question, how does the availability of new technology affect the existing disparity? Previous studies have also explored this link between the dissemination of new surgical technologies and access disparity among vulnerable populations, an issue that remains relevant across various surgical subspecialties.
Insurance status and race represent independent predictors of undergoing laparoscopic surgery for appendicitis: secondary data analysis of 145,546 patients.
J Am Coll Surg.2004; 199 (discussion: 575-7): 567-575
The rapidly expanding use of thoracic endovascular aortic repair over open repair provides a recent example of a revolutionary cardiovascular technology. Johnston and colleagues
reported that counter to their hypothesis, racial/ethnic minorities and patients with lower socioeconomic status were more likely to receive thoracic endovascular aortic repair over traditional open repair despite a previously described baseline disparity. Even after controlling for baseline comorbidity and treatment indication, they reasoned that greater disease severity and aneurysm morphology might not have been fully captured in their statistical controls, leading vulnerable populations to preferentially undergo the less invasive therapy. A similar phenomenon might be at play in our TAVR population, with racial/ethnic minorities historically presenting with more advanced disease processes
Comparative analysis of autogenous infrainguinal bypass grafts in African Americans and Caucasians: the association of race with graft function and limb salvage.
Outcome comparison of African-American and Caucasian patients with severe aortic stenosis subjected to transcatheter aortic valve replacement: a single-center experience.
Despite our results showing decreasing disparity over time at high-volume hospitals, it is also true that these institutions began with the greatest disparities, and any progress might represent some reversion to the mean. Historical disparity in those undergoing SAVR,
provides reason to be cautious that any progress toward decreasing the racial/ethnic gap might be reversed as TAVR use continues to expand to low-risk populations and structural causes of disparity remain unaddressed.
Our results also mirrored those of a similar investigation into drug-eluting stent (DES) versus bare-metal stent use and differences according to race/ethnicity. Hannan and colleagues
described an existing disparity with racial-ethnic minorities receiving a DES less frequently, despite it being considered the optimal treatment at the time. As overall use of the DES expanded, the ethno-racial disparity diminished in medium- and high-volume hospitals, though persisted in the lowest volume hospitals, leading the authors to suggest directing patients to high-volume hospitals could decrease disparity.
Hospital volume represents a potential driving factor of racial inequity for TAVR use as well. The initial approval of TAVR by the US Food and Drug Administration established procedural volume criteria,
Between 2013 and 2017, the number of sites performing TAVR in the United States increased from 277 to 554, with low-volume sites (<50 TAVRs annually) representing 39% of sites and performing 14% of cases by 2017.
Although it has previously been shown that market competition is one driver of TAVR adoption, it is not clear whether or not this would exacerbate or alleviate racial disparities,
as TAVR use increases across low-, medium-, and high-volume hospitals. The volume-outcome relationship was redemonstrated in 2019, when Vemulapalli and colleagues
also revealed Black and Hispanic patients were more likely to undergo TAVR in the lowest quartile hospitals according to volume. The idea that vulnerable populations disproportionately receive care at low-volume hospitals has been the focus of study across various cardiovascular interventions,
and is confirmed with our finding that at high-volume hospitals, racial/ethnic minorities were 43% less likely to undergo TAVR than White patients. Although the effect on outcomes remains beyond the scope of this present study, access to hospitals performing TAVR does appear to be influenced by race/ethnicity. Similar to the investigation into DES use,
our findings suggest that any decreased racial/ethnic inequity is being driven by progress at high-volume hospitals, providing a high-yield target for future policy and further investigation. Although it is unclear how the continued rapid growth of TAVR will be distributed among low- versus high-volume hospitals, it is important to continue monitoring these trends.
This study has several limitations. First, the analysis was limited to only 8 states. Although these states are large, geographically and ethno-racially diverse, and represent nearly 20% of the US population, they might not be representative of the country as a whole. Second, although using SID did allow for linkage of patient-level, hospital-level, and county-level data through linkage to American Hospital Association and Area Health Resources Files databases, it does not capture granular information on indications for intervention or outcomes measures. Although this prevented us from stratifying patients on the basis of concomitant procedures, we were able to instead focus on top-line numbers of TAVR versus SAVR to describe the broad dissemination patterns of the new technology in its early years. The SID also only captures those that have made it to the point of intervention, without the ability to assess referral pathways, access to specialists, or social/cultural factors that might influence patients’ health decisions or act as barriers to receiving care. Finally, use of TAVR has continued to expand, with increased use among intermediate- and low-risk populations, as well as increased penetration into medium- and low-volume hospitals. Future work will be necessary to continue examining long-term trends, as the population undergoing TAVR continues to evolve.
Conclusions
This multistate evaluation is representative of a large, ethno-racially heterogeneous patient population of >50,000 patients who underwent TAVR or SAVR, and allows for an early examination of how TAVR is being disseminated. We showed that despite existing racial/ethnic disparity, TAVR use grew faster among vulnerable populations than among their White counterparts, a trend driven by increasing use in high-volume hospitals, and visualized in our Graphical Abstract (Figure 3). Future research is needed to confirm long-term trends and verify continued progress as the new technology expands to intermediate- and low-risk populations and is available at more medium- and low-volume hospitals.
Figure 3Did ethno-racial disparities in access to transcatheter aortic valve replacement change over time?
The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest.
Table E1Percentage of population who underwent TAVR and SAVR broken down by race and state, with the percentage of White population listed for each state
FDA approves expanded indication for two transcatheter heart valves for patients at intermediate risk for death or complications associated with open-heart surgery.
Insurance status and race represent independent predictors of undergoing laparoscopic surgery for appendicitis: secondary data analysis of 145,546 patients.
J Am Coll Surg.2004; 199 (discussion: 575-7): 567-575
Comparative analysis of autogenous infrainguinal bypass grafts in African Americans and Caucasians: the association of race with graft function and limb salvage.
Outcome comparison of African-American and Caucasian patients with severe aortic stenosis subjected to transcatheter aortic valve replacement: a single-center experience.
This study was supported by a grant from the Georgetown-Howard Universities Center for Clinical and Translational Science and The Lee Folger Foundation.
00304-7&id=fx1.jpg){kind=link}
00304-7&id=fx2.jpg){kind=link}
00304-7&id=gr1.jpg){kind=link}
00304-7&id=gr2.jpg){kind=link}
00304-7&id=gr3.jpg){kind=link}
00304-7&id=fx4.jpg){kind=link}