Limited Access to Aortic Valve Procedures in Socioeconomically Disadvantaged Areas

Abstract

Background

To explore how differences in local socioeconomic deprivation impact access to aortic valve procedures and the treatment of aortic valve disease, in comparison to other open and minimally invasive surgical procedures.

Methods and Results

Procedure volume data were obtained from the Healthcare Cost and Utilization Project from 18 states from 2016 to 2019 and merged with area deprivation index data, an index of zip code‐level socioeconomic distress. We estimate the relationship between local deprivation ranking and differences in volumes of aortic valve replacement, which include transcatheter aortic valve replacement (TAVR) and surgical aortic valve replacement (SAVR), versus coronary artery bypass graft surgery and laparoscopic colectomy (LC). All regressions control for state and year fixed effects and an array of zip code‐level characteristics. TAVR procedures have increased over time across all zip codes. The rate of increase is negatively correlated with deprivation ranking, regardless of the higher share of hospitalizations per population in high deprivation areas. Distributional analysis further supports these findings, showing that lower area deprivation index areas account for a disproportionately large share of SAVR, TAVR, and LC procedures in our sample relative to their share of all hospitalizations in our sample. By comparison, the cumulative distribution of coronary artery bypass graft procedures was nearly identical to that of total hospitalizations, suggesting that this procedure is equitably distributed. Regressions show high area deprivation index areas have lower prevalence of SAVR (β=−15.1%, [95% CI, −26.8 to −3.5]), TAVR (β=−9.1%, [95% CI, −18.0 to −0.2]), and LC (β=−19.9%, [95% CI, −35.4 to −4.4]), with no statistical difference in the prevalence of coronary artery bypass graft (β=−2.5%, [95% CI, −12.7 to 7.6]), a widespread and commonly performed procedure. In the population aged ≥80 years, results show high area deprivation index areas have a lower prevalence of TAVR (β=−11.9%, [95% CI, −18.7 to −5.2]) but not SAVR (β=−0.8%, [95% CI, 8.1 to 6.3]), LC (β=−3.5%, [95% CI, −13.4 to −6.4]), or coronary artery bypass graft (β=5.2%, [95% CI, −1.1 to 1.1]).

Conclusions

People living in high deprivation areas have less access to life‐saving technologies, such as SAVR, and even moreso to device‐intensive minimally invasive procedures such as TAVR and LC.

Nonstandard Abbreviations and Acronyms

ADI

area deprivation index

AS

aortic stenosis

AVR

aortic valve replacement

LC

laparoscopic colectomy

SAVR

surgical aortic valve replacement

TAVR

transcatheter aortic valve replacement

Clinical Perspective.

What Is New?

• This study highlights the existence of significant procedural inequities in access to advanced cardiovascular treatments based on the socioeconomic disadvantage status of a geographic area.

• Inequities persist not only for newer technologies like transcatheter aortic valve replacement but also for long‐established procedures like surgical aortic valve replacement and laparoscopic colectomy.

• Despite the growth and adoption of transcatheter aortic valve replacement from 2016 to 2019, its distribution remained as uneven as at the start of the study period.

What Are the Clinical Implications?

• Clinicians should recognize that patients from socioeconomically disadvantaged areas may face barriers to life‐saving treatments, potentially impacting their outcomes.

• The current distribution of procedures suggests a need for targeted policy changes to ensure equitable access to cutting‐edge cardiac procedures irrespective of patients' geographical socioeconomic status.

• Understanding the root causes of these disparities, whether volume restrictions, hospital regulations, transportation costs, or other factors, is essential for tailoring interventions and ensuring all patients receive the best possible care.

The Social Determinants of Health (SDoH) have a significant impact on health through various channels, including individual, social, and economic factors, as well as barriers to accessing medical care. Despite the availability of medical interventions, SDoH may affect the quality and type of treatment received, leading to an unequal distribution of high‐quality care and adverse health outcomes among disadvantaged communities. ¹ , ² , ³ Many of these determinants are directly linked to individuals' personal socioeconomic status as well as that of their community. ⁴

In our research, we analyzed the correlation between the level of socioeconomic disadvantage in a specific zip code (high versus low) and the number of residents (living in the zip code) receiving aortic valve replacement (AVR) procedures, including both surgical aortic valve replacement (SAVR) and transcatheter aortic valve replacement (TAVR). TAVR is a less invasive procedure aimed at treating aortic stenosis (AS), a progressive condition in which the heart's aortic valve opening narrows, restricting blood flow to the aorta. One of the most prevalent forms of cardiovascular disease in the Western world, AS can only be durably treated using a surgical or transcatheter valve replacement. TAVR has been shown to have a similar or lower rate of morbidity and mortality, as well as a reduced length of inpatient hospital stay, compared with traditional SAVR. ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵

To explore the underlying mechanisms linking area deprivation and access to AVR, with a focus on TAVR, we further studied the association between socioeconomic deprivation and procedure access for 2 other commonly performed procedures: coronary artery bypass graft (CABG) and laparoscopic colectomy (LC). CABGs were selected as a comparative measure for all AVRs as they are the most frequently performed inpatient cardiac surgery and are more broadly available; thus, we expect fewer geographic barriers to access. LC surgeries, a minimally invasive alternative to open colectomy, were chosen as a comparator for TAVRs because both procedures have similar prevalence and are less invasive alternatives to open surgery, despite targeting different organs and disease states. Studies suggest that LC is associated with lower mortality and morbidity, quicker recovery, and shorter hospital stays compared with open colectomy. ¹⁶ , ¹⁷ , ¹⁸ Moreover, LC is an established procedure that has been approved for over 2 decades, whereas TAVR was only approved in 2011 and continues to experience year over year growth. ¹⁹

For the purpose of investigating the relationship between socioeconomic disadvantage and access to medical procedures, we curated data from multiple sources and aggregated them to the zip code‐level. The extent of socioeconomic disadvantage was measured using the area deprivation index (ADI), which is a validated marker of neighborhood disadvantage; residence within a community with high ADI is a predictor of an individual's risk for poor health outcomes. However, existing evidence has not assessed how ADI is associated with access to transcatheter procedures like TAVR. The current study aims to fill this gap in evidence and understand the correlation between socioeconomic disadvantage (high ADI score) and access to TAVR.

Previous studies have documented socioeconomic and racial disparities in patient TAVR utilization. ²⁰ , ²¹ , ²² Most closely aligned with our study, Nathan et al ²³ examined the association between zip code‐level socioeconomic conditions and TAVR rates within the Medicare population in major metropolitan areas, finding that zip codes with greater socioeconomic disadvantage and higher rates of minorities had lower TAVR rates. We expand on this research by contrasting TAVR rates to those for other procedures, utilizing ADI measure to succinctly define socioeconomic deprivation and expanding the sample to include non‐Medicare populations living outside of major metropolitan areas (for a subsample of states).

METHODS

All data are publicly available and can be procured from the Healthcare Cost and Utilization Project website.

Sample Definition

We collected 2016 to 2019 procedure volume data from the Healthcare Cost and Utilization Project State Inpatient Database data for the 18 states that published patient zip code information: Arizona, Delaware, Florida, Indiana, Iowa, Kentucky, Maryland, Michigan, Minnesota, Missouri, New Jersey, North Carolina, Oregon, Rhode Island, South Dakota, Vermont, Washington, and Wisconsin. The data contain the universe of community hospital discharges in those states in those years, regardless of patient coverage.

ADI data were obtained from the Neighborhood Atlas website. ²⁴ ADI is a composite measure consisting of a multidimensional evaluation of a region's socioeconomic conditions and its link to health outcomes. ²⁵ It is a composite measure that assigns a rank to each zip code in the United States, ranging from 0 (least disadvantaged) to 100 (most disadvantaged). Census block level ADI data were averaged to the zip code‐level and merged with inpatient hospitalization Healthcare Cost and Utilization Project data by patient zip code to construct our analysis sample.

Zip code‐level American Community Survey Census economic and demographic data were also collected to be used as additional controls in our regression modeling.

All data used in this analysis were deidentified and accessed in compliance with the Health Insurance Portability and Accountability Act. As a retrospective analysis of a deidentified database, this research did not require institutional review board approval or informed consent.

Key Measures

Our analysis outcome measures are procedure volumes. We measured the number of CABG, SAVR, TAVR, and LC inpatient procedures performed in each sample zip code over the total number of inpatient hospitalizations in the area, by year.

We examined the relationship between ADI score and procedure volumes by binning zip codes by ADI score range, to compare the difference between low and high deprivation areas. Our key independent variable is a zip code‐year level indicator variable, HighADI (high level of disadvantage), which receives a value when the zip code is at or above a score of 66 in the national ADI ranking, and is set to zero when the zip code is below a score of 33 (indicating low levels of disadvantage). In our regression analysis, we have omitted zip codes with ADI scores in the mid‐range between the top and bottom terciles of scores. Implementing this strategy resulted in omitting 41% of zip code‐years in our sample, or approximately 49% of the population and 49% of hospitalizations. We note that our results are robust to using wider (25–75) or narrower (40–60) mid‐ranges in our analysis.

Analytic Approach

To estimate the relationship between an area's social deprivation and the extent of procedural utilization, we constructed 2 alternative outcome variables. The first, an indicator variable that receives value when a procedure is performed at least once in an area (procedure availability), and the second, the log of procedural volume (procedure prevalence). The first measure proxies the availability of this procedure for area residents, while the second measures its prevalence conditional on the procedure being available.

For a given procedure, we estimated the procedure availability regression using the following linear probability model, where $i$ and $t$ index zip code and year, respectively:

$$1\left[\text{Volum}{\mathrm{e}}_{\mathrm{it}}>0\right]={\beta }_0+{\beta }_1\text{HighAD}{\mathrm{I}}_{\mathrm{it}}+{\beta }_2{X}_{\mathrm{it}}+\text{State}X\mathrm{yea}{\mathrm{r}}_{\mathrm{it}}+{ϵ}_{\mathit{it}}$$

Here, ${\beta }_1$ is the coefficient of interest, measuring the average percentage point difference in the probability that at least one procedure is performed in high ADI zip codes compared with low ADI zip codes. We control for numerous zip code‐specific economic and demographic variables captured by the vector $X_{\mathrm{it}}$, including population size (logged), total number of hospitalizations (logged), percent of the population that is White, percent Black, median household income, median house value, labor force participation, unemployment rate, share of college educated, median family size, share married, and median age. We further control for regional temporal shocks using state‐year fixed effects, $\text{State}X\mathrm{yea}{\mathrm{r}}_{\mathrm{it}}$. Finally, ${ϵ}_{\mathrm{it}}$ denotes the unobservable term. In estimation, we cluster errors at the state level. We note that our results are qualitatively similar when using a nonlinear probit regression model instead of a linear probability model. Results are available upon request.

For a given procedure, the procedure prevalence regression is identically modeled using the following linear regression:

$$\log \left(\text{Volum}{\mathrm{e}}_{\mathrm{it}}\right)={\gamma }_0+{\gamma }_1\text{HighAD}{I}_{\mathrm{it}}+{\gamma }_2{X}_{\mathrm{it}}+\text{State}X\mathrm{yea}{\mathrm{r}}_{\mathrm{it}}+{ϵ}_{\mathrm{it}}$$

where zip codes with zero procedural volume, $\text{Volum}{\mathrm{e}}_{\mathrm{it}}=0$, are excluded from the sample by construction. Here, ${\gamma }_1$ is the coefficient of interest, measuring the average percent change in the rate of procedural volume (divided by the number of total hospitalizations in the area) in high ADI zip codes compared with low ADI zip codes, among zip codes that have any positive procedural volume. We note that our results would remain unchanged when indexing procedure volumes by population instead of hospitalizations, as we control for log population size and log total number of hospitalizations in our model.

We repeat the regression analysis on the subsample of procedures performed on patients 80 years of age or older. During our sample period, guideline recommendations for the choice of TAVR over SAVR were based on surgical risk, without reference to patient age. However, the current American College of Cardiology and American Heart Association guidelines recommend TAVR for patients over 80 years of age with severe AS. ²⁶ Patients in this age group have a higher likelihood of receiving TAVR at any hospital compared with younger patients, as the benefit from a less invasive intervention increases with age. ²⁷

RESULTS

Summary Statistics and Frequency of Procedures

In this study, we analyzed 25 721 observations of 6623 zip codes in 18 US states, which were obtained from various data sources and aggregated at the zip code‐level. We used the ADI to quantify socioeconomic disadvantage, ranking all zip codes from 0 (least disadvantaged) to 100 (most disadvantaged). Table 1 presents the summary statistics of our sample.

Table 1.

Summary Statistics

	High ADI	Low ADI	P value of difference
Zip code population	6324	15 173	<0.0001
	(10199)	(15184)
Median age, y	42.54	43.17	<0.0001
	(9.29)	(10.66)
Median family size	3.00	2.96	<0.0001
	(0.58)	(0.69)
Male (%)	50.07	49.13	<0.0001
	(6.91)	(7.93)
Married (%)	51.49	55.00	<0.0001
	(13.50)	(14.74)
Race: Black (%)	6.84	5.73	<0.0001
	(15.71)	(10.52)
Race: White (%)	85.79	82.22	<0.0001
	(22.11)	(18.24)
College educated or above	15.16	45.94	<0.0001
	(8.14)	(18.37)
Median household income	43 767	88 754	<0.0001
	(15165)	(38075)
Labor force participation (%)	56.17	63.06	<0.0001
	(12.98)	(14.04)
Unemployment rate (%)	6.61	4.47	<0.0001
	(16.31)	(7.66)
Hospitalizations: Any	13 329	9922	<0.0001
	(15145)	(22404)
Hospitalizations: Diabetes	3559.41	1938.57	<0.0001
	(6189.07)	(4855.35)
Hospitalizations: Hypertension	7017.10	4793.19	<0.0001
	(8680.66)	(11404.83)
Number of zip codes	4935	1688
Number of observations (zip code X year)	19 136	6585

SD in parentheses. Hospitalizations measured per 10 000 residents. ADI indicates area deprivation index; High ADI, high deprivation; and Low ADI, low deprivation.

As highlighted in the Table 1, a statistically significant difference was identified between high and low ADI areas in all demographic and economic variables accounted for in our analysis. High ADI areas, which represent areas with higher levels of socioeconomic deprivation, had a smaller average population (6324 people [SD=10 199] compared with 15 173 [SD=15 184] in low ADI areas), a lower average age, a larger average family size, a higher proportion of men, a lower rate of married individuals, and a lower representation of individuals from a non‐White, non‐Black racial background than low ADI areas. Furthermore, high ADI areas had a lower rate of college‐educated individuals, a lower average median household income, a lower average labor force participation rate, and a higher average unemployment rate. Finally, high ADI areas had a higher rate of inpatient hospitalizations compared with low ADI areas, as well as a significantly higher prevalence of diabetes and hypertension‐related inpatient hospitalizations per 100 000 people, suggesting individuals in high ADI areas are overall sicker than those in low ADI areas.

Table 2 compares procedural volume for our selected procedures across low ADI and high ADI areas. Regardless of how volume is normalized compared with the population, we found that CABG is the most commonly performed procedure, followed by SAVR, TAVR, and finally LC. When normalizing area volume by area population (ie, dividing volume count by population count), we found that high ADI areas have a higher volume of CABG, with 88.24 procedures per 100 000 residents (SD=200.36), compared with 63.86 (SD=275.51) in low ADI areas. For SAVR and LC, the difference in volumes across low ADI and high ADI areas was not statistically significant, while the volume of TAVR was lower on average in high ADI areas, at 19.71 procedures per 100 000 residents (SD=79.81) compared with an average of 24.43 (SD=176.43) in low ADI areas. Moving from means to medians, we find that in high ADI areas, the median volume of SAVR, TAVR, and LC is zero, meaning that at least 50% of high ADI areas had zero patients undergoing these procedures.

Table 2.

Frequency of Procedures by Low and High ADI

	High ADI					Low ADI					P value of mean difference
	Mean	SD	25th Percentile	Median	75th Percentile	Mean	SD	25th Percentile	Median	75th Percentile
Volume (per 100 000 people)
CABG	88.24	200.36	0.00	55.90	111.75	63.86	275.71	6.87	38.43	65.21	<0.0001
SAVR	24.97	89.23	0.00	0.00	25.97	24.89	260.85	0.00	11.79	24.95	0.981
TAVR	19.71	79.81	0.00	0.00	18.01	24.43	176.43	0.00	10.73	26.05	0.037
LC	12.72	57.04	0.00	0.00	9.66	15.15	122.63	0.00	6.44	17.05	0.122
Volume (per 1000 hospitalization)
CABG	7.00	16.91	0.00	4.44	9.05	5.94	11.06	0.55	4.62	7.39	<0.0001
SAVR	2.21	12.79	0.00	0.00	2.03	2.35	6.87	0.00	1.39	2.84	0.264
TAVR	1.66	10.12	0.00	0.00	1.37	2.50	15.36	0.00	1.25	2.89	<0.0001
LC	1.08	8.89	0.00	0.00	0.70	1.48	3.81	0.00	0.75	1.97	<0.0001

SD in parentheses. ADI indicates area deprivation index; CABG, coronary artery bypass grafting; High ADI, high deprivation; LC, laparoscopic colectomy; Low ADI, Low deprivation; SAVR, surgical aortic valve replacement; and TAVR, transcatheter aortic valve replacement.

Differences in procedure volumes were more pronounced and statistically significant when volume was normalized by the number of area hospitalizations, which considers both general access to care in the zip code, as well as a measure of resident morbidity. As before, CABG volumes were higher in high ADI areas compared with low ADI areas. However, the rates of SAVR, TAVR, and LC were lower in high ADI areas. In other words, when area morbidity and access to care were accounted for using procedural volume normalization, we found that SAVR, TAVR, and LC volumes were lower in high ADI areas compared with low ADI areas.

Procedure Volumes Over Time and ADI Rank

Figure 1 presents in binned scatterplots, for each year of our sample, the average number of CABG, SAVR, TAVR, and LC procedures performed (per 1000 hospitalizations) within each decile of the area ADI rank distribution. Within each bin, we weighted the average volume by the total number of hospitalizations in the area. We further plotted the quadratic fit of each binned scatterplot. The figure highlights 2 important findings. First, TAVR volumes have steadily increased year over year across all areas regardless of ADI ranking, from a sample average of 1.16 (SD.147) procedures per 1000 hospitalizations in 2016 to 2.03 (SD=1.89) procedures in 2019. Over the same period, SAVR volumes decreased slightly year over year, LC volumes remained virtually unchanged, while CABG procedure volume increased only slightly, from 5.24 (SD=3.38) in 2016 to 5.48 (SD=3.30) in 2019. Second, SAVR, TAVR, and LC volumes decreased in ADI score, while CABG volumes did not exhibit a clear linear correlation with ADI score. In 2016, the average SAVR, TAVR, and LC volumes in the top ADI rank decile were 43%, 68%, and 62% lower than in the bottom decile, respectively. Similar differences were exhibited in 2017 through 2019, despite the increase in overall TAVR volumes over our sample period: in 2019, the average TAVR volume in the top ADI rank decile was still 67% lower than in the bottom decile.

Figure 1. Transcatheter aortic valve replacement and laparoscopic colectomy volumes, by area deprivation index rank and year.

Weighted average coronary artery bypass grafting, surgical aortic valve replacement, transcatheter aortic valve replacement, and laparoscopic colectomy volumes per 1000 hospitalizations, by area deprivation index rank distribution deciles and year. Average weighted by total number of hospitalizations in the zip code. Trend line plot the linear fit of each respective binscatter. ADI indicates area deprivation index; CABG, coronary artery bypass grafting; LC, laparoscopic colectomy; SAVR, surgical aortic valve replacement; and TAVR, transcatheter aortic valve replacement.

Figure 2 highlights the unequal distribution of TAVR procedures across zip codes by plotting the cumulative share of TAVR procedures and total hospitalizations when areas are sorted by ADI rank. To do this, we summed procedures and populations across all zip codes by ADI, and calculated the cumulative share of these sums out of the total sample when sorted by ADI rank. In our sample, low ADI areas accounted for a smaller share of hospitalizations than high ADI areas: zip codes with an ADI rank of 50 or lower account for 39.7% of all sample hospitalizations. At the same time, the same zip codes accounted for 48.4% of all sample TAVR procedures, and 44.9% of all SAVR procedures. The cumulative distribution of LC procedures exhibited the same pattern as that of TAVR, with zip codes with an ADI rank of 50 or lower accounting for 48.0% of all LC procedures in our sample. By comparison, the cumulative distribution of CABG procedures was nearly identical to that of total hospitalizations.

Figure 2. Cumulative share of hospitalizations and procedures by area deprivation index rank.

Cumulative shares of procedures (red) and all hospitalizations (blue) of the sample, by area area deprivation index rank. ADI indicates area deprivation index; CABG, coronary artery bypass grafting; LC, laparoscopic colectomy; SAVR, surgical aortic valve replacement; and TAVR, transcatheter aortic valve replacement.

Regression Analysis: Relationship Between Procedure Volume and ADI Rank

Table 3 presents our main estimation results. For each of the 4 study procedures, we estimated the probability that any resident in the area underwent the procedure (procedure availability), and the log volume of procedures performed on area residents, if any (procedure prevalence), based on the regression specified in the analytic approach section. For CABG procedures, our estimated high ADI coefficients were not statistically different from zero in both the extensive and intensive regressions.

Table 3.

Estimation Results: Relationship Between Procedure Volume and ADI Rank

	CABG	SAVR	TAVR	LC
Procedure availability: $1\left[\text{Volum}{\mathrm{e}}_{\mathrm{it}}>0\right]$
High ADI	0.0170	−0.0386**	−0.0338**	−0.0774***
	(0.0152)	(0.0141)	(0.0148)	(0.0135)
Observations	25 570	25 570	25 570	25 570
R‐squared	0.485	0.474	0.469	0.447
Procedure prevalence: $\log \left(\text{Volum}{\mathrm{e}}_{\mathrm{it}}\right)$
High ADI	−0.0253	−0.151**	−0.0908**	−0.199**
	(0.0482)	(0.0553)	(0.0421)	(0.0734)
Observations	17 339	12 239	11 109	9655
R‐squared	0.478	0.661	0.670	0.656

State cluster‐robust standard errors in parentheses, ***P<0.01, **P<0.05, *P<0.1. All regressions control for state‐year fixed effects, log population size, log total number of hospitalizations, percent White, percent Black, median household income, median house value, labor force participation, unemployment rate, share of college educated, median family size, share married, and median age. ADI indicates area deprivation index; CABG, coronary artery bypass grafting; High ADI, high deprivation; LC, laparoscopic colectomy; Low ADI, low deprivation; Procedure availability, the probability that any area resident underwent the procedure, Prevalence: log volume of the outcome, conditional on having any procedures performed on area residents; SAVR, surgical aortic valve replacement; and TAVR, transcatheter aortic valve replacement.

For AVR, both surgical and transcatheter, we found that high ADI areas suffer from both reduced access and reduced procedure volumes. High ADI areas were 3.9% (SE=1.4) and 3.4% (SE=1.5) less likely to have any SAVR or TAVR procedure performed compared with low ADI areas, respectively. In areas where such procedures were performed, both SAVR and TAVR volumes were lower on average, measuring at 15.1% (SE=5.5) and 9.1% (SE=4.2), respectively. We found similar results for LC, where high ADI areas were 7.8% (SE=1.4) less likely to have any LC procedures performed and had volumes that were 20% (SE=7.3) lower on average in the high ADI areas with any access to the procedure.

Table 4 presents our estimation results when we limit procedure counts to patients aged ≥80 years. Like our main estimation results, where we studied procedure utilization across all ages, we found no statistically significant difference in CABG procedure access or volumes in high ADI areas in comparison to low ADI areas. For SAVR procedures, we found a 3.7% (SE=1.3) decrease in the procedure availability in high ADI areas, but no decrease in procedure prevalence conditional on availability (unlike the general population). The largest differences in comparing our estimates for the general population to the ≥80 population were measured in access to‐ and volume of TAVR procedures. High ADI areas were 7.3% (SE=1.2) less likely to have any TAVR procedures in the population aged ≥80 years, more than double the difference measured in the general population. When TAVR procedures were performed on patients aged ≥80 years, procedure volumes were on average 12% (SE=3.2) lower when the area was a high ADI area, a larger difference than that measured in the general population. For LC procedures, difference in access to procedures was smaller in those aged ≥80 years, measured in a 3.0% decrease in access in high ADI areas. In areas where LC procedures were performed, we found no statistically significant difference in LC volumes between high and low ADI areas among those aged ≥80 years, though the number of area‐periods in which such volumes were positive was a comparatively small 1300 observations.

Table 4.

Estimation Results: Relationship Between Procedure Volume and ADI Rank in the Population Aged ≥80 Years

	CABG	SAVR	TAVR	LC
Procedure availability: $1\left[\text{Volum}{\mathrm{e}}_{\mathrm{it}}>0\right]$
High ADI	−0.0304	−0.0370**	−0.0731***	−0.0299***
	(0.0184)	(0.0129)	(0.0116)	(0.0102)
Observations	24 037	24 037	24 037	24 037
R‐squared	0.306	0.176	0.417	0.083
Procedure prevalence: $\log \left(\text{Volum}{\mathrm{e}}_{\mathrm{it}}\right)$
High ADI	0.0517	−0.00887	−0.119***	−0.0350
	(0.0297)	(0.0342)	(0.0319)	(0.0468)
Observations	6314	2959	8494	1300
R‐squared	0.802	0.884	0.690	0.942

Using procedure counts for patients aged ≥80. State cluster‐robust standard errors in parentheses, ***P<0.01, **P<0.05, *P<0.1. All regressions control for state‐year fixed effects, log population size, log total number of hospitalizations, percent White, percent Black, median household income, median house value, labor force participation, unemployment rate, share of college educated, median family size, share married, and median age. ADI indicates area deprivation index; CABG, coronary artery bypass grafting; High ADI, high deprivation; LC, laparoscopic colectomy; Low ADI, low deprivation; Procedure availability, the probability that any area resident underwent the procedure, Procedure Prevalence: log volume of the outcome, conditional on having any procedures performed on area residents.

SAVR, surgical aortic valve replacement; and TAVR, transcatheter aortic valve replacement.

DISCUSSION

The present study aimed to examine the relationship between the level of socioeconomic disadvantage in a specific area, measured as high or low ADI, and access to surgical and transcatheter‐based AVR procedures. Throughout the analysis, access to AVR procedures was compared with access to 2 other commonly performed surgical procedures, CABG and LC. For each procedure, we studied both procedure availability (whether residents from the area had any access to the procedure), and procedure prevalence availability (the volume of procedures performed, given availability). We found no association between CABG procedure access or volume and local socioeconomic disadvantage. For both SAVR and TAVR, we found that residents of high ADI areas suffered from both reduced access and reduced procedure volumes per hospitalization compared with low ADI areas. In other words, even when AVR were available locally, residents from disadvantaged areas were less likely to undergo the procedure. We found similar results for LC, where residents of high ADI areas suffered from both reduced access and reduced procedure volumes. For patients aged ≥80 years, a population favored for TAVR over the surgical approach, we found that patients in high ADI areas suffered from greater reduced access and reduced procedure volumes to TAVR compared with the general population, while patients with SAVR (and LC) suffered significantly smaller access problems, and where CABG patients suffered no access issues.

Distributional analysis supports that lower ADI areas account for a disproportionately large share of TAVR, SAVR, and LC procedures in our sample relative to their share of all hospitalizations in our sample. By comparison, the cumulative distribution of CABG procedures was nearly identical to that of total hospitalizations, suggesting that this procedure is equitably distributed.

One explanation as to why CABG is widely available across areas of varying socioeconomic disadvantage ranking, while AVR procedures are inequitably distributed could be the larger footprint of CABG‐performing hospitals compared with AVR‐performing hospitals. In every year in our sample, the number of hospitals where CABG procedures were performed was more than double the number of hospitals where TAVR were performed. For TAVR procedures, this could be explained by the fact that TAVR procedures were restricted to certain hospitals based on national coverage determination requirements established by Centers for Medicare & Medicaid Services. The national coverage determination set minimum requirements for hospitals to start a TAVR program, based on surgeon procedure experience, hospital infrastructure, and most importantly, procedure volumes. During most of our sample period, hospitals wishing to start a TAVR program were required to perform 400 percutaneous coronary intervention cases/year and 20 surgical AVRs per year. ²⁸ These requirements were lessened slightly in 2019. ²⁹ Hospitals that lacked the necessary procedure volumes to start a TAVR program could not offer the procedure, limiting access for patients who live in the surrounding areas. This can result in inequities in the health care system, as patients in some areas have access to the latest medical advancements while others do not. Moreover, TAVR availability has been shown to bring more patients with AS to treatment, ³⁰ and with barriers to TAVRs in deprived and disadvantaged areas, this could potentially impact access to overall AS care.

It is important to note the positive intended role of the restrictive Centers for Medicare & Medicaire Services coverage determinations, especially during the early stages of TAVR adoption. Limiting the number of hospitals performing TAVR aimed to ensure high‐quality outcomes by requiring providers to have sufficient skills, experience, and facilities to safely perform this novel procedure. The emphasis on restricting the procedure to high‐volume medical centers, when TAVR was a new procedure, was based on the correlation between the frequency of procedures performed and improved patient outcomes.

Overall, evidence indicates that many patients with AS do not undergo AVR. In a large meta‐analysis, 41.6% of patients with severe symptomatic AS did not undergo AVR. ³⁰ Similarly, a study of the US‐based Optum Integrated Claims‐Clinical Database found that from 2011 to 2016, the rate of AVR among patients with symptomatic severe AS increased from 20.1% to 37.1%, indicating that more than 60% of patients remain untreated in US practice. ³¹ Another US hospital‐based study showed that over an 18‐year period, the proportion of patients with an indication for AVR who did not receive AVR has remained substantial despite the rapid growth of AVR volumes: less than half (48%) of patients with an indication or potential indication for AVR received AVR. ³²

Our findings suggest that procedural inequity is present even in long‐standing procedures like LC. The argument that the adoption process of a new technology may initially affect the well‐insured and affluent population, but will eventually become more equitable over time, did not hold true for this 2‐decade old, minimally invasive procedure. Over the 4 years of our sample period, the argument has not held true for TAVR either: we found that TAVR has remained as inequitably distributed in 2019 as it was in 2016, despite significant increases in procedure adoption and volume over this time period.

These disparities related to access and volume in minimally invasive procedures such as TAVR and LC highlight the need for a more equitable and comprehensive approach to healthcare delivery that reduces the underlying factors that contribute to procedural inequity.

Limitations

Our study has several limitations. The data underlying the ADI measure are from 2016 to 2020 five‐year American Community Survey data. As such, the ADI is a measure of a zip code's average socioeconomic deprivation over that period, while our procedure volume measures vary by year. We conjecture that by using the top and bottom thirds of the ADI rank scale, we retain the subsample of zip codes that would naturally have more time‐invariant socioeconomic conditions, whether positive or negative. We further note that any averaging‐out of socioeconomic conditions should bias against us finding a correlation between ADI and procedure volume. Another limitation is that our analysis is limited to 18 states, and thus may not fully reflect national trends. However, these 18 states make up about a third (33.8%) of the US population and are geographically spread across the 4 census regions of the country.

Further research is required to establish which mechanisms govern the reduction in some procedure volumes in high ADI areas. Apart from the hospital national coverage determination requirements, patients in high ADI zip codes may suffer from reduced SAVR, TAVR, and LC access because of higher relative transportation costs, insurance coverage, health literacy, and provider bias, among other potential channels. However, these channels should also affect access to CABG, which exhibits no correlation with ADI. Furthermore, hospitals catering to high ADI patients may also be ones that are less likely to offer TAVR or LC procedures because of profitability concerns or lack of expertise.

Finally, we note that other residual confounders may bias the results of our estimation, even after controlling for observable zip‐code‐level socioeconomic conditions. These residual confounders may be nonlinear or unobservable socioeconomic or population health related zip‐code characteristics.

CONCLUSIONS

People living in high deprivation zip codes have less access to cutting edge life‐saving technologies, such as SAVR, and even more so to minimally invasive procedures such as TAVR and LC.

Sources of Funding

This work was supported by Edwards Lifesciences.

Disclosures

G. David, A. Bergman, C. Gunnarsson, and M. Ryan have received consultancy fees from Edwards Lifesciences. S. Chikermane, C. Thompson, and S. Clancy are employees of Edwards Lifesciences.