Examination of Race and Infrainguinal Bypass Conduit Use in the SVS Vascular Quality Initiative

Abstract

Background:

Vein conduit is known to have better patency than prosthetic for infrainguinal bypass (IIB). Here we explore if racial disparities exist in IIB vein conduit use amid pre-operative patient and systemic factors.

Methods:

Retrospective SVS Vascular Quality Initiative data (VQI) for 23,959 IIB between 2003–17 for occlusive disease were analyzed. For homogeneity, only IIB originating from the common femoral artery were included. Demographics of patients receiving vein vs prosthetic were compared and logistic regression analyses were performed with race and preoperative factors to evaluate for predictors of vein conduit use.

Results:

Adjusted regression models demonstrated black patients were 76% as likely (p<.001) and Hispanic patients 79% as likely (p=.003) to have vein conduit compared to white patients. Factors positively correlating with vein use included vein mapping, more distal bypass target, tissue loss or acute ischemia bypass indications, commercial insurance, and weight. Factors against vein use included advanced age, female gender, ASA class 4, urgent procedure, preoperative mobility limitation, prior CABG or leg bypass, prior smoking, preoperative anticoagulation, and a bypass performed in the Southern US or before 2012. While black and Hispanic patients were less likely to receive vein, they were vein mapped at similar or higher rates than other groups.

Conclusion:

Racial disparities exist in conduit use for IIB, with black and Hispanic patients less likely to receive vein bypasses. However, the contribution of race to conduit selection is small in adjusted and unadjusted models. Overall, pre-operative variables in the VQI poorly predicted vein conduit use for IIB.

Introduction

Peripheral arterial disease is one of the most common medical problems in the United States, affecting 3–10% of the population over age 40.(1, 2) Despite advances in minimally invasive therapies, infrainguinal bypass (IIB) for treatment of arterial occlusive disease remains a commonly performed procedure with a rate of 126 performed per 100,000 people annually in the U.S.(3) Racial disparities with regard to outcomes following IIB have previously been noted, with black and Hispanic patients more likely to have limb loss and less likely to maintain bypass patency.(4–9)

Vein grafts have superior patency compared to prosthetic conduit for infrainguinal bypass.(10) The choice of conduit use (vein versus prosthetic) is often made at a surgeon level due to existing patient factors (such as anatomy and prior intervention history) and it is unknown whether there is an association of race with conduit choice. In other domains of surgery, racial disparities in treatment offerings have been seen.(11–13) In pancreatic cancer care while black patients were as likely as white patients to show resectable disease on imaging, they were less likely to receive surgery and chemotherapy interventions and an underlying medical basis for this disparity was not evident.(11) In dialysis access, there have also been studies suggesting black or African American patients are less likely to receive a first-time arteriovenous fistula (AVF) than patients of other ethnicities, however this racial disparity appears to be due to smaller arm vein diameters in these patients.(12)

The objective of this study is to evaluate if ethnoracial disparities exist in choice of conduit for IIB for peripheral arterial disease, and if present, associations with physician practice versus patient characteristics.

Methods

A retrospective analysis of the Society for Vascular Surgery Vascular Quality Initiative (VQI) registry for IIB was utilized for this study. The VQI was developed initially in New England in 2002 to improve outcomes of Vascular Procedures.(14) It has subsequently spread across the country and internationally with 612 current participating centers. The IIB data set contains over 130 variables which span demographics and comorbidities, along with intraoperative and perioperative variables, with greater procedural granularity than many other surgical quality registries.

To provide homogeneity in patient disease, bypass procedures with a common femoral artery origin performed for occlusive disease from 2003 to 2017 were selected for inclusion. This provided a cohort of 23,959 procedures after exclusion of other bypass indications (i.e. aneurysm), bypass origins other than common femoral artery, and missing race/or ethnicity. Race categories documented in the VQI include: white, black or African American, Asian, American Indian or Alaskan native, native Hawaiian or other pacific islander, more than one race, unknown/other, and ethnicity is documented as Hispanic/non-Hispanic. For preservation of statistical power, only ethnoracial groups of white non-Hispanic (subsequently referred to as white), black/African American, Hispanic, and non-white/non-black/non-Hispanic were considered for analyses reported below. Our outcome of interest was vein conduit use. Only preoperative variables were considered for inclusion as possible predictors of vein conduit use. Insurance status was included among preoperative factors as a potential proxy for available care and socioeconomic status. Similarly, due to the expansion of the VQI nationally over time, a control variable for surgery year was included to account for potential uneven racial and geographic distribution of patients in the quality database over time.

Continuous and count variables were compared by t-test and X² testing, respectively. Univariate and multivariate logistic regression analyses were performed to evaluate for predictors of vein conduit use, with pseudo-R² used as an assessment of the model. A p value of <.05 was considered to be statistically significant. All analyses were performed using Stata software (StataCorp, College Station, Texas) in this IRB-approved (UAB IRB# 300002708) retrospective quality database study.

Results

23,959 LEB were identified and included in the study with distal targets consisting of 7,051 above knee popliteal artery, 8,112 below knee popliteal artery, 1,003 tibioperoneal trunk, and 6,232 with tibial vessels as distal target. Further details of the bypass distal target distributions by ethnoracial and gender groups within the cohort can be found in supplementary figure 1 (Supplement 1). Vein conduit alone was used in 13,010 procedures while 10,857 used non-vein conduit.

The proportion of female patients (32.0% of vein conduit patients vs 37.7% of non-vein conduit, p<.001) and minorities (13.7% of vein conduit patients black vs 15.2% non-vein conduit patients, p=.002) were lower in vein bypass group. Proportions of patients with high operative risk of ASA class≥4 (18.6% vs 25.9%, p<.001) and emergent procedures (2.0% vs 3.8%, p<.001) were lower among vein bypass patients than non-vein conduit patients. History of prior revascularization bypasses were lower in vein bypass patients than non-vein conduit patients, with prior CABG rates of 16.0% vs 28.2%, p<.001and prior leg bypass rates of 24.6% vs 33.2%, p<.001(Table 1). A higher proportion of the vein conduit patients were diabetic (50.0% vein vs 48.4% non-vein, p=.01), received vein mapping preoperatively (71.2% vein vs 44.2% non-vein, p<.001), or had a distal target of below knee popliteal/ tibioperoneal trunk (43.1% vein vs 32.2% non-vein, p<.001) or tibial vessel target (33.3% vein vs 17.3% non-vein, p<.001).

Table 1:

Demographics, Risk Factors and Clinical Presentation

Pre-operative Factor	Non-vein Bypass (n=10,857)	Vein Bypass (n=13,010)	p
Age (mean ± SD)	67.4 ± 10.5	66.0 ± 10.8	<.001
Female gender	37.7%	32.0%	<.001
Race			.002
Black	15.2%	13.7%
Hispanic	4.6%	4.2%
Non-white/Non-black/Non-Hispanic	2.5%	2.4%
Weight (mean ± SD)	80.3 ± 19.1	76.4 ± 20.0	<.001
ASA Class			<.001
2	3.7%	4.3%
3	70.2%	76.7%
4	25.9%	18.6%
5	0.1%	0.1%
Hypertension	89.8%	87.9%	<.001
Any Coronary Artery Disease	36.1%	30.6%	<.001
Dialysis	5.6%	4.9%	.03
Chronic obstructive pulmonary disease	31.6%	27.0%	<.001
Diabetes Mellitus	48.4%	50.0%	.01
Smoking			<.001
Prior	46.3%	41.7%
Current	42.6%	46.6%
Preoperative antiplatelet use	79.8%	79.3%	.35
Preoperative anticoagulant use	17.1%	14.7%	<.001
History of any prior coronary artery bypass graft	28.2%	16.0%	<.001
History of any leg bypass	33.2%	24.6%	<.001
Prior Ipsilateral leg bypass	19.2%	12.3%	<.001
Prior Ipsilateral stent	26.1%	26.7%	.33
Living at home preoperative	96.5%	96.7%	.67
Preoperative ambulatory status			<.001
Ambulatory	71.7%	74.0%
Ambulatory with Assistance	22.9%	20.9%
Wheelchair use	4.9%	4.7%
Bedridden	0.5%	0.4%
Procedure Urgency			<.001
Urgent	17.9%	18.8%
Emergent	3.8%	2.0%
Procedure Indication			<.001
Asymptomatic	0.94%	0.79%
Claudication	32.0%	25.3%
Rest Pain	22.9%	23.5%
Tissue Loss	31.6%	39.4%
Acute Ischemia	12.5%	10.8%
Vein Mapping Obtained	44.2%	71.2%	<.001
Distal Bypass Target
Above knee popliteal	41.2%	19.7%	<.001
Below knee popliteal or Tibioperoneal trunk	32.2%	43.1%	<.001
Tibial	17.3%	33.3%	<.001
Insurer			<.001
Medicare	43.5%	37.3%
Medicaid	6.4%	6.8%
Commercial	28.5%	28.6%
Military/VA	0.7%	0.9%
Non-US Insurance	0.6%	0.9%
Self Pay	1.9%	2.1%
Insurance status not reported	18.5%	23.4%
Geographic Region
North	20.7%	21.3%	.26
South	29.1%	23.2%	<.001
East	44.2%	48.5%	<.001
West	6.0%	7.0%	.004

Of interest, rates of prior procedures that might have already utilized vein conduit tended to be lowest among black patients within the cohort. Prior CABG was present in 14.4% of black patients compared to 23.1% of white patients, 23.9% of Hispanic patients, and 25.1% of non-white/non-black/non-Hispanic patients, p<.001. Similarly, rates of any prior leg bypass were lowest within black patients (26.1%) compared to white (28.9%), Hispanic (30.2%), or non-white/non-black/non-Hispanic (28.8%) patients, p=.007.

Overall, black and Hispanic patients were more likely to receive preoperative vein mapping than white patients (63.1% black vs 61.4% Hispanic vs 58.1% white, p<.001). Broken down by distal target level, there was no significant difference in percentage receiving vein mapping for bypass with an above knee target (53.4% among black patients, 54.2% among Hispanic patients, 52.3% among non-white/non-black/non-Hispanic patients and 51.3% among white patients, p=0.55). Black patients, however, were more likely to receive vein mapping in bypasses with a below knee popliteal (68.2% among black patients vs 62.8% among Hispanic patients, 62.1% among white patients and 54.4% in non-white/non-black/non-Hispanic patients, p=.001) or tibial vessel bypass target (71.7% among black patients vs 65.8% among Hispanic patients vs 66.2% among white patients vs 60.9% in non-white/non-black/non-Hispanic patients , p=.002) (Figure 1).

Figure 1: Receipt of Vein Mapping by Ethnoracial Group.

Percent of patients receiving vein mapping by ethnoracial group and bypass targets within the cohort

Despite being more likely to receive vein mapping, black and Hispanic patients were less likely to receive vein bypasses at the distal target level of the below knee popliteal/tibioperoneal trunk (53.0% black vs 53.5% Hispanic vs 59.9% non-white/non-black/non-Hispanic vs 63.5% white, p<.001) or tibial vessel (67.3% black vs 62.5% Hispanic vs 63.0% non-white/non-black/non-Hispanic vs 71.1% white, p<.001) (Figure 2). At the above knee popliteal artery target level, there was no significant difference in the proportion of vein bypasses received across racial groups (37.0% black vs 36.1% Hispanic vs 34.7% non-white/non-black/non-Hispanic vs 36.3% white, p=0.95). Further details of unadjusted rates of vein conduit use within each bypass target location by ethnoracial and gender groups within the cohort can be found in supplementary figure 2 (Supplement 2).

Figure 2: Vein Conduit by Ethnoracial Group:

Percent of patients receiving vein conduit by ethnoracial group and bypass targets within the cohort

On unadjusted regression analysis of vein use by race, black patients had an odds ratio of 0.88 of receiving vein conduit compared to white patients) (p<.001) and Hispanic patient an odds ratio of 0.90 of receiving vein conduit (p=0.08). In the unadjusted model, race alone explained only .04% of the variation in vein conduit use.

Multivariate logistic regression analyses (from preoperative variables with backwards stepwise regression) showed an OR of 0.76 for vein conduit use in black patients (p<.001) and 0.83 in Hispanic patients (p=0.02). Other factors that positively correlated with vein use included preoperative vein mapping (OR 2.81, p<.001), below knee popliteal/tibioperoneal trunk target (OR 2.93, p<.001), or tibial vessel target (OR 4.71, p<.001), weight, or commercial insurance (Table 2). Patient factors that correlated against vein use included ages above 50, female gender, urgent or emergent procedure, ASA class 4, prior smoking, preoperative anticoagulation, preoperative mobility status of either ambulatory with assistance or bedridden, prior CABG, and prior or ipsilateral infrainguinal bypass. Having a bypass performed in the South, or prior to 2012 also contributed to a lower likelihood of vein conduit utilization. Inclusion or exclusion of insurance status in the model did not change the explained variation in conduit use, however the adjusted model as a whole explained only 15.3% of variation in vein conduit use, suggesting the preoperative capture within our quality database cannot well describe the decision making around conduit use.

Table 2:

Predictors of Vein Conduit Use in Multivariate Analysis

	Odds Ratio	95% Confidence Interval	p
Race (referent white)
Black	0.76	0.69–0.83	<.001
Hispanic	0.83	0.71–0.97	0.02
Distal Bypass Target
Tibial	4.65	4.24–5.10	<.001
Below Knee Popliteal or Tibioperoneal trunk	2.97	2.75–3.21	<.001
Preoperative vein mapping	2.77	2.59–2.96	<.001
Procedure Indication (referent claudication)
Tissue loss	1.28	1.17–1.40	<.001
Acute ischemia	1.17	1.04–1.32	.01
Commercial Insurance (referent Medicare)	1.10	1.02–1.19	.008
Weight (per additional kg)	1.008	1.007–1.01	<.001
Age (referent <40 years)
50–59 years	0.50	0.32–0.80	.004
60–69 years	0.46	0.29–0.74	.001
70–79 years	0.45	0.28–0.71	.001
80–89 years	0.37	0.23–0.59	<.001
>89 years	0.25	0.14–0.44	<.001
Female gender	0.84	0.78–0.91	<.001
Procedure Urgency
Urgent	0.86	0.79–0.94	.001
Emergent	0.61	0.49–0.75	<.001
ASA class 4	0.40	0.19–0.84	.02
Prior Coronary Artery Bypass Graft	0.45	0.41–0.49	<.001
Any prior leg bypass	0.71	0.65–0.79	<.001
Prior ipsilateral leg bypass	0.57	0.51–0.65	<.001
Prior smoking	0.84	0.75–0.94	.003
Preoperative anticoagulation	0.88	0.80–0.96	.006
Preoperative mobility status
Ambulatory with assistance	0.91	0.84–0.99	.03
Bedridden	0.54	0.32–0.93	.03
Bypass performed in Southern region of the country (referent: Eastern region)	0.77	0.71–0.84	<.001
Bypass performed prior to 2012	0.56	0.35–0.91	.02

Discussion

Overall, vein conduit use in infrainguinal bypass was poorly predicted by the preoperative variables captured in the VQI. Our study demonstrated that while racial disparities exist in vein conduit use for infrainguinal bypass, these disparities do not appear to be from lack of preoperative vein mapping in minorities. While the VQI captures whether preoperative vein mapping was obtained, there is unfortunately no assessment of whether the vein was deemed suitable for bypass based on the vein mapping results or intraoperative examination. Smaller diameter leg vein is a known risk factor for loss of patency following infrainguinal bypass.(7, 15–17) We postulate that some of the discrepancy seen with regard to rates of mapping versus actual conduit use across racial groups may be related to differences in vein diameter suitability or length of adequate conduit, as unadjusted rates of vein conduit use were lowest among women with tibial bypasses, particularly in minority women. Prior studies have shown racial differences in caliber of upper extremity veins(12). While no precise definition of vein quality has been defined, size, morphology, and compliance are considered relevant criteria.(18) In addition to vein diameter, presence of pre-existing pathological changes such as vein wall thickening or areas of strictures may represent additional barriers to having sufficient length of vein conduit felt to be suitable for infrainguinal bypass. Consideration of adding a variable to the VQI assessing vein suitability could allow for better retrospective assessment of operative planning in the future.

Expectedly, patients with a history of procedures that might have resulted in lower extremity vein utilization (such as prior CABG and prior leg bypasses) had lower odds of vein bypass; prior ipsilateral stenting procedures were not significantly associated with conduit on univariate or multivariate modeling. While arm vein harvest for procedures other than hemodialysis access might be intentionally avoided in dialysis patients, being on dialysis, while significant in univariate regression, did not have significant associations with vein use on multivariate regression.

Vein conduit was more commonly utilized in all racial groups with progressively more distal bypass targets, however black and Hispanic patients were still less likely than white patients to receive vein bypasses for distal targets below the knee. Atherosclerotic disease distribution and vein quality may both contribute to this. It has also been shown that black and Hispanic patients were more likely to present with infrageniculate disease, while white patients were more likely to have isolated aortoiliac occlusive disease.^{19, 20} More distal disease distribution in turn requires a longer segment of good quality vein to make use of vein conduit feasible in infrainguinal bypass. With a lower proportion of vein bypasses being performed under emergent conditions in our study, differences in conduit use could also relate to black patients presenting with more severe vascular disease and increased risk factors.(19) Due to low sample sizes in this retrospective cohort, we are unable to well characterize the heterogeneous group of Non-White/Non-Black/Non-Hispanic patients and did not find statistically significant disparities in vein conduit use for this group on univariate or multivariate regression.

One of the strongest predictors against vein conduit use within our multivariate regression modeling was patient age, and women were significantly older than men within our cohort. While age does not necessarily equate to physiologic reserve, women minorities were also among the least likely to be independently ambulatory preoperatively. Hispanic patients also had significantly higher rates of ASA class 4 or above within the cohort. Expediency of operation may have been favored by a surgeon over the long-term durability of vein conduit within these potentially frail patient populations.

Regional variation has been shown to contribute to perioperative outcome disparities seen between black and white patients in the VQI.(20) Black and Hispanic patients have been more likely to have limb loss and less likely to maintain bypass patency;(4–9) with increased rates of limb loss persisting in black patients even when socioeconomic status was matched to white counterparts(21) though black and Hispanic patients have been shown to have lower rates of long-term mortality in comparison to white counterparts.(20, 22) Amid the known superior patency of vein grafts compared to prosthetic conduit in infrainguinal bypass,(10) vein conduit utilization if feasible and of adequate size is an element of optimal care. Interestingly, our study suggests a greater adjusted odds of vein conduit utilization in more recent years, however over the years of VQI capture, unadjusted proportions of bypasses performed with vein tended to decline from 70.9% in 2003 to 51.25% in 2017. This may in part be a reflection of shifts in practice related to potential endovascular first approaches to therapy (as prior open or endovascular interventions did not disqualify a patient from inclusion in our cohort) and an increased breadth of percutaneous intervention options in patients at high risk for bypass due to poor conduit. Further non-database evaluation of surgeon practice may be required, as has been done via survey regarding conduit familiarity and barriers to surgeon use among UK surgeons performing coronary artery bypass surgery(23) given the poor correlation of vein conduit utilization with preoperative variables seen in the VQI database.

Conclusion

Racial disparities exist in conduit use for infrainguinal bypass, with black and Hispanic patients less likely to receive vein bypasses. However, the contribution of race and ethnicity to conduit selection is small in both adjusted and unadjusted models. Importantly, minority patients are at least as likely, if not more likely to undergo vein mapping preoperatively, suggesting that unmeasured variables such as vein diameter/quality play a role in these disparities. Further studies to evaluate differences in vein availability based on race and ethnicity may further elucidate our findings.