Regional Market Competition is Associated with Aneurysm Diameter at the Time of EVAR

Abstract

Background:

Local market competition has been previously associated with more aggressive surgical decision-making. For example, more local competition for organs is associated with acceptance of lower quality kidney offers in transplant surgery. We hypothesized that market competition would be associated with the size of an abdominal aortic aneurysm (AAA) at the time of elective endovascular aneurysm repair (EVAR).

Methods:

We included all elective EVARs reported in the Vascular Quality Initiative database (2012–2018). Small AAAs were defined as a maximum diameter <5.5 cm in men or <5.0 cm in women. We calculated the Herfindahl-Hirschman Index (HHI), a measure of physician market concentration (higher HHI = less market competition), for each US census region. Multilevel regression was used to examine the association between the size of AAA at EVAR and HHI, clustering by region.

Results:

Of 37,914 EVARs performed, 15,379 (40.6%) were for small AAAs. There was significant variation in proportion of EVARs performed for small AAAs across regions (P < 0.001). The South had both the highest proportion of EVARs for small AAAs (44.2%) as well as the highest market competition (HHI 50), whereas the West had the lowest proportion of EVARs for small AAAs (35.0%) and the lowest market competition (HHI 107). Adjusting for patient characteristics, each 10 unit increase in HHI was associated with a 0.1 mm larger maximum AAA diameter at the time of EVAR (95% CI 0.04–0.24 mm, P = 0.005).

Conclusions:

Physician market concentration is independently associated with AAA diameter at time of elective EVAR. These data suggest that physician decision-making for EVAR is impacted by market competition.

INTRODUCTION

There has been recent recognition of high rates of endovascular aneurysm repair (EVAR) performed for small diameter abdominal aortic aneurysms (AAAs).^1,2 Although retrospective data suggest that patients with small AAAs may have lower complications and perioperative mortality than those who underwent EVAR for a larger AAA,¹ multiple prior studies have found no survival benefit to early intervention compared with surveillance for patients with small AAAs.^3–5 As a result, the Society for Vascular Surgery (SVS) clinical practice guidelines recommend elective repair of fusiform AAAs at ≥5.5 cm for men or ≥5.0 cm for women.⁶ Despite these guidelines, a recent study found that 41% of patients undergoing EVAR in the Vascular Quality Initiative (VQI) had small AAAs.¹ Similarly, a study of center-level compliance with the SVS guidelines on size of AAAs at time of repair found that compliance ranged from 36% to 92% within centers participating in the VQI. Given these findings, we sought to identify other factors in the decision to intervene on a small AAA.

In other surgical fields, market competition has been shown to influence surgeon and hospital decision-making. In the field of kidney transplantation, areas with higher competition for deceased donor kidneys have higher use of low-quality kidneys and associated worse transplant outcomes.⁷ Similarly, in minimally invasive surgery, higher market competition is associated with higher odds of undergoing robotic surgery, suggesting that increased market competition leads hospitals to acquire robotic technology.⁸ There is evidence of regional differences in patient selection for EVAR as well,⁹ but it is not clear whether market competition plays a role in the regional differences of patient selection.

We hypothesized that market competition is associated with the size of AAAs at the time of elective EVAR. We studied VQI and US geographic regional competition quantified by the Herfindahl-Hirschman Index (HHI), a measure of market concentration, to evaluate the association of market competition with small AAA repair.

METHODS

Data Source

We used data from the VQI, a collaboration of the Society for Vascular Surgery Patient Safety Organization, regional quality improvement groups, and a commercial technology partner.¹⁰ This study was acknowledged as exempt by the Johns Hopkins Medicine Institutional Review Board (IRB00212076).

Study Population

We included all elective EVAR from 2012 to 2018. Patients with ruptured AAAs (n = 2,426) and those with no preoperative maximum AAA diameter reported (n = 900) were excluded. Patients whose EVAR was performed at institutions outside of geographically defined VQI regions were also excluded (n = 180), as their data could not be used to examine regional market competition.

Exposures

We defined small AAAs as a maximum diameter <5.5 cm in men or <5.0 cm in women based on the SVS guidelines for AAA repair.⁶ Patient characteristics including symptomatic AAA, sex, race, primary insurer, body mass index, American Society of Anesthesiologists class, smoking history, and comorbid diseases including coronary artery disease, hypertension, diabetes mellitus, congestive heart failure, chronic obstructive pulmonary disease, end-stage kidney disease with dialysis dependence, and family history of AAAs were also examined.

Regional Market Competition

Deidentified numbered VQI regions (Supplemental Fig. 1) were combined to create US geographical regions per VQI staff as follows: the West included VQI regions 6, 10, 22, 24, and 28. The Midwest included VQI regions 13, 14, 18, and 23. The South included VQI regions 3, 5, 7, 8, and 25. The Northeast included VQI regions 11, 12, 19, and 21.

For each VQI region and US region, the Herfindahl-Hirshman Index (HHI) was calculated. The HHI is a measure of local market competition, where higher HHI reflects more monopoly and less competition, whereas lower HHI reflects higher market competition.^7,8 The HHI for each market is calculated as the sum of squares of the market share for each firm in the market and ranges from 0 to 10,000. For example, a market with 2 firms where one has a 5% share of the market and the other firm has a 95% share of the market has an HHI of 5² + 95² = 9,050, reflecting a monopoly with very little competition. In contrast, a market with 100 firms where each firm has a 1% share of the market has an HHI of 100 × (1²) = 100, indicating a high level of competition in this market.

Statistical Analysis

The primary outcome of our study was EVAR performed for small AAAs. Characteristics and regional differences were compared for small versus usual size AAA repair using chi-square and Wilcoxon rank-sum tests. Multilevel logistic regression was used to examine the association of EVAR performed for small AAAs and the HHI, adjusting for patient characteristics chosen a priori and clustering by region for both VQI regions and for US regions. Multilevel linear regression was then used to examine the association between the size of AAAs at the time of EVAR and HHI, adjusting for the same patient characteristics and with regional clustering. For VQI regions where a wider range of HHI was found, the HHI was scaled to 100 units for multivariable analyses, whereas for US regions with a more narrow range of HHI, a scale per 10 units of HHI was used. All analyses were performed using Stata/SE version 15.0 (College Station, TX) with P < 0.05 denoting statistical significance.

RESULTS

Study Population

Of 37,914 EVAR reported during the study period, 15,379 (40.6%) were performed for small AAAs. Patients who underwent EVAR for small AAAs were younger (median age 72 vs. 75 years, P < 0.001), less frequently women (10.3% vs. 25.1%, P < 0.001), less frequently symptomatic (7.4% vs.10.4%, P < 0.001), and more frequently had concurrent iliac artery aneurysms (29.6% vs. 22.9%, P < 0.001) than patients who underwent EVAR for usual size indications (Table I). 12,532 EVAR were performed in the Northeast (33.0%), 12,095 were performed in the South (31.9%), 7,747 were performed in the Midwest (20.4%), and 5,542 were performed in the West (14.6%).

Table I.

Patient characteristics associated EVAR performed for small AAAs^a

Characteristic	Usual size indication (n = 22,535)	Small AAA (n = 15,379)	P-value
Age, median (interquartile range [IQR]) years	75 (68, 81)	72 (66, 78)	<0.001
Female sex	5,665 (25.1%)	1,587 (10.3%)	<0.001
White race	20,142 (89.4%)	13,925 (90.5%)	<0.001
BMI, median (IQR)	27 (24, 31)	28 (24, 31)	<0.001
Primary insurer			<0.001
Medicare	13,345 (59.2%)	9,346 (60.8%)
Medicaid	451 (2.0%)	336 (2.2%)
Commercial	6,670 (29.6%)	4,757 (30.9%)
Military/VA	286 (1.3%)	147 (1.0%)
Non-US insurance	650 (2.9%)	107 (0.7%)
Self-pay	188 (0.8%)	103 (0.7%)
Missing	945 (4.2%)	583 (3.8%)
US region			<0.001
West	3,602 (16.0%)	1,940 (12.6%)
Midwest	4,608 (20.4%)	3,139 (20.4%)
South	6,751 (30.0%)	5,342 (34.7%)
Northeast	7,574 (33.6%)	4,958 (32.2%)
Comorbid diseases
Preoperative smoking history			0.35
Never smoker	3,032 (13.5%)	2,026 (13.2%)
Prior smoker	12,335 (54.7%)	8,359 (54.4%)
Current smoker	7,149 (31.7%)	4,984 (32.4%)
Missing	19 (0.1%)	10 (0.1%)
CAD	6,633 (29.5%)	4,515 (29.4%)	0.88
Hypertension	18,834 (83.6%)	12,779 (83.1%)	0.21
Diabetes mellitus			0.25
None	17,933 (79.6%)	12,190 (79.3%)
Diet	1,020 (4.5%)	650 (4.2%)
Noninsulin medication	2,722 (12.1%)	1,919 (12.5%)
Insulin	847 (3.8%)	608 (4.0%)
Missing	13 (0.1%)	12 (0.1%)
CHF	2,998 (13.3%)	1,721 (11.2%)	<0.001
COPD	7,826 (34.7%)	4,888 (31.8%)	<0.001
ESRD	281 (1.2%)	149 (1.0%)	0.012
Family history of AAA	1,797 (8.0%)	1,283 (8.4%)	0.21
Aneurysm characteristics
Preoperative maximum diameter, median (IQR) mm	59 (56, 65)	50 (45, 52)	<0.001
Symptomatic	2,339 (10.4%)	1,137 (7.4%)	<0.001
Concurrent iliac aneurysm	5,052 (22.9%)	4,452 (29.6%)	<0.001

^aSmall AAA defined as <5.0 cm in women or <5.5 cm in men.

VQI Regions

There was wide variation in percentage of EVAR performed for small AAAs across VQI regions, ranging from 14.3% to 52.3%. There was also a wide range of HHI across VQI regions, from 126 to 1,426 (Fig. 1). Adjusting for patient age, sex, race, primary insurer, symptomatic AAAs, and concurrent iliac artery aneurysm, there was no significant association between EVAR performed for small AAAs and competition within the VQI region (per 100 units HHI, adjusted odds ratio 1.00, 95% confidence interval [CI] 0.97–1.03, P = 0.79) (Table II). Similarly, adjusting for patient characteristics, there was no significant association between size of AAA at time of EVAR and competition within the VQI region (per 100 units higher HHI, 0.02 mm smaller maximum AAA diameter, 95% CI-0.1 mm to0.2 mm, P = 0.74).

Fig. 1.

VQI regional percentage of EVAR performed for small AAAs and market competition described by Herfindahl-Hirschman Indices. There was wide variation in percentage of EVAR performed for small AAAs (bars) and HHI (diamonds) across VQI regions.

Table II.

Multivariable logistic regression assessing the association of EVAR performed for small AAA and VQI region competition as measured by the Herfindahl-Hirschman index

Characteristic	Adjusted odds ratio	95% confidence interval
Herfindahl-Hirschman Index, per 100 points	1.00	0.97–1.03
Age, per year	0.97	0.97–0.97
Female sex	0.37	0.34–0.39
White race	1.05	0.97–1.13
Primary insurer (reference, Medicare)
Medicaid	0.80	0.68–0.93
Commercial	0.86	0.81–0.90
Military/VA	0.58	0.47–0.71
Non-US insurance	0.23	0.15–0.36
Self-pay	0.54	0.42–0.70
Symptomatic AAA	0.70	0.64–0.76
Concurrent iliac artery aneurysm	1.26	1.20–1.33

US Regions

After grouping VQI regions into US geographic regions, the highest market competition was located in the South (HHI 50) and the lowest market competition was located in the West (HHI 107) (Fig. 2). The highest percentage of EVAR performed for small AAAs was in the South at 44.2%, followed by the Midwest (40.5%), Northeast (39.6%), and West (35.0%) (P < 0.001). Adjusting for patient age, sex, race, primary insurer, symptomatic AAA, and concurrent iliac artery aneurysm, there was a significantly higher odds of EVAR performed for small AAAs among US regions with higher market competition (per 10 units HHI, adjusted odds ratio1.03, 95% CI 1.01–1.05, P = 0.03) (Table III). Similarly, adjusting for patient characteristics, there was a lower mean maximum AAA diameter associated with higher market competition within the US region (per 10 units higher HHI, 0.1 mm lower maximum AAA diameter, 95% CI 0.04 to0.24 mm, P = 0.005).

Fig. 2.

US regional percentage of EVAR performed for small AAAs and market competition described by Herfindahl-Hirschman Index. Adjusting for patient characteristics, there was a higher odds of EVAR performed for small AAA among US regions with higher regional market competition (P = 0.03).

Table III.

Multivariable logistic regression assessing the association of EVAR performed for small AAA and US region competition as measured by the Herfindahl-Hirschman Index

Characteristic	Adjusted odds ratio	95% confidence interval
Herfindahl-Hirschman Index, per 10 points	0.97	0.95–0.99
Age, per year	0.97	0.97–0.97
Female sex	0.37	0.34–0.39
White race	1.03	0.96–1.11
Primary insurer (reference, Medicare)
Medicaid	0.80	0.68–0.93
Commercial	0.85	0.81–0.89
Military/VA	0.56	0.45–0.68
Non-US insurance	0.24	0.19–0.30
Self-pay	0.55	0.42–0.70
Symptomatic AAA	0.71	0.65–0.76
Concurrent iliac artery aneurysm	1.24	1.18–1.30

DISCUSSION

In this study of elective EVAR performed in the VQI, 40.6% of EVAR were performed for AAAs smaller than 5.5 cm in men or smaller than 5.0 cm in women. There was a wide variation in the percentage EVAR performed for small AAAs across VQI regions, and this variation remained when VQI regions were grouped into US regions. All US regions had high competition as measured by the HHI, a description of market concentration. There was a significant association of EVAR performed for a small AAA with higher US regional competition, suggesting that physician decision-making in elective EVAR may be influenced by market competition.

The high proportion of EVAR performed for small AAAs is consistent with a recent report from Jones et al.¹ They found that 41% of EVAR reported in the VQI were performed for small AAA, and that patients who underwent EVAR for small AAAs were younger, had fewer comorbid diseases, and had lower perioperative morbidity and mortality than those who underwent EVAR for larger diameter AAAs. We similarly found that younger age was associated with EVAR for small AAAs, as was male sex. The lower OR of EVAR performed for small AAAs in women is consistent with data from a recent study by Ramkumar et al.¹¹ that showed lower utilization of EVAR than open AAA repair in women. While there is limited information regarding specific indication for intervention for small AAAs in our study, the high utilization of EVAR for small AAAs merits further investigation as it is outside of practice guidelines published by the SVS,⁶ which are based on the low rate of rupture of small AAAs.¹²

Our finding that US regional competition was associated with EVAR performed for small AAAs reaffirms prior evidence that market competition influences health care decision-making. Sethi et al.¹³ have previously reported that more hospital competition within a hospital market was associated with an increased odds of EVAR compared with open AAA repair. That study used the Nationwide Inpatient Sample and linked Hospital Market Structure data, which does not include clinically granular data on the size of AAAs at time of repair. In contrast, the VQI does include data on size of AAAs at the time of repair, making our study a logical next step of that earlier work. Taken together, the data suggest that market competition drives both higher utilization of EVAR for AAA repair overall, and an “earlier” utilization of EVAR at smaller AAA diameter.

We did not find an association of competition within VQI regions and EVAR for small AAAs. We presume this is due to the lack of geographic meaning of VQI regions (Supplemental Fig. 1). For example, New York, New Jersey, and Connecticut are in 3 different VQI regions, although the New York City metropolitan area includes all 3. Interestingly, in another national study of EVAR performed between 2001 and 2006, there was no correlation between EVAR utilization and the number of physicians in a market, but there was higher EVAR utilization in areas with higher AAA incidence.¹⁴ While that study found substantial state-wide variation in the use of EVAR, the number of physicians, the number of vascular surgeons, hospital beds, teaching hospitals, or trauma centers did not predict high EVAR utilization. The discrepancies in those data compared with our study may suggest that the influence of competition on size of AAAs at EVAR is a more modern phenomenon or could be related to our use of a formal market competition measure (HHI) compared with more general measures of volume.

The concept that market competition appears to drive EVAR performed for smaller AAA suggests there may be an opportunity for directed educational interventions to improve outlier performance. For example, regional, center, and/or physician-based performance reports could be used in an attempt to improve compliance with practice guidelines within high-competition markets. The use of peer-benchmarked performance reporting for hospitals has been used to improve outcomes and decrease costs in a variety of specialties in the past,^15–17 and recent data suggests that physician-level interventions can similarly impact outlier utilization habits.^18,19 The VQI already distributes center-level performance reports for various vascular metrics.¹⁰ While there is no accounting for the “rogue surgeon” who performs procedures outside of guidelines and may be influenced by fee-for-service reimbursement, we believe that adding a metric of compliance with SVS guidelines for AAA size at the time of repair to the VQI EVAR reports may be useful to improve guideline compliance.

Our study has several limitations. First, it is important to note that some small AAAs may require repair below guideline-based thresholds.⁶ For example, rapidly expanding aneurysms, saccular aneurysms, and symptomatic aneurysms deserve repair earlier than a standard uncomplicated infrarenal AAA. While we adjusted for symptomatic status, the VQI does not contain information about aneurysm morphology or growth rate. There are devices on the market where the primary indication for repair is iliac artery aneurysm size, which may result in the concomitant repair of a small AAA below threshold. Consistent with this notion, we found a significant association of concurrent iliac aneurysm with small AAA repair, which we included in our multivariable model. Second, our findings are limited to institutions that participate in the VQI. We cannot comment on institutions performing EVAR in the same markets that do not participate in the VQI. There may be variation regionally in how many institutions participate in the VQI as well, which may introduce a sampling bias in considering regional differences. Finally, we are limited in our definition of regions. While we found that VQI region competition was not associated with EVAR for small AAAs, VQI regions are not meaningful hospital markets. We were able to group VQI regions into US geographic regions and did find an association between EVAR for small AAA and competition, but these are very large regions. Future work examining the association of small AAA repair with market competition would ideally use a better definition of local markets such as metropolitan areas²⁰ and would account for epidemiologic differences in population and disease density. Unfortunately, we did not have this information available in our deidentified VQI data set.

CONCLUSIONS

We have demonstrated that US regional market competition is associated with AAA size at the time of EVAR, with repair at smaller diameter in markets with higher competition. Specifically, in US regions with higher competition, there are higher odds of EVAR performed for AAAs smaller than 5.5 cm in men or 5.0 cm in women. These findings suggest that physician decision-making regarding AAA repair may be driven by market competition. There may be a role for directed educational interventions in markets with outlier EVAR utilization to improve compliance with published AAA management guidelines.