Conduit choice and volume-outcome relationships in multi-arterial coronary artery bypass grafting among Medicare beneficiaries in the United States

Abstract

Introduction:

Multi-arterial grafting (MAG) is under-utilized. We evaluated the association between surgeon experience and conduit utilization outcomes following MAG.

Methods:

Using US Centers for Medicare and Medicaid data, we identified 29,268 patients ≥ 65 years undergoing isolated, primary, non-emergency MAG, of which 15,682 met the inclusion criteria. Propensity score matching was performed on 26 baseline patient characteristics. Individual surgeons were grouped into tertiles based on conduit use. Low volume:<3 radial arteries or 2 bilateral internal thoracic artery (BITA) grafts annually; high-volume:>10 radial arteries or 4 BITAs annually. The primary outcome was major adverse cardiac events (MACE): death, myocardial infarction or repeat revascularization at 4-years. Outcomes were compared in a multivariable Cox-proportional hazard model adjusting for the individual surgeon’s case volume of each conduit.

Results:

Among MAG recipients, 5,784 (20%) received radial artery grafts and 9,989 (34%) received BITA grafts. Radial artery recipients were younger (70.6 vs. 71.2years) and more likely diabetic (46.9% vs 43.2%) than BITA recipients. Among 5,778 matched pairs, at 4-years, the incidence of MACE was numerically lower in the radial group (14.7% vs 15.7%, p=0.05), but there was no difference in all-cause mortality (10.8%radial vs 11.5%BITA, p=0.06). Among BITA recipients, surgeon experience was associated with MACE only among the lowest vs the highest volume tertile (Adjusted HR:1.15, 95%CI: 1.01=1.33, p=0.046). There was no association between MACE and surgeon volume for radial artery graft recipients.

Conclusions:

Radial artery and BITA grafting demonstrated similar mid-term outcomes, while there was a surgeon volume effect for BITA use.

Introduction

A analysis of over 1 million patients undergoing coronary artery bypass grafting (CABG) demonstrated a late mortality benefit to multi-arterial grafting compared to single arterial grafting in patients up to the age of 70 years¹. However, the value of multi-arterial grafting and choice of second arterial conduit among older patients is poorly defined. Identifying the optimal revascularization approach in this population is critical, given the rising prevalence of coronary artery disease among the aging population along with increasing life expectancy.

While both American and European revascularization guidelines give Class 1 recommendations for the use of the radial artery over the saphenous vein graft in patients appropriate for multi-arterial grafting, usage of a bilateral internal thoracic artery(BITA) graft only has a Class IIa recommendation^2,3. Data supporting this recommendation are conflicting: while the Arterial Revascularization Trial(ART) did not show any difference in the primary outcome of all-cause mortality at 10 years between patients undergoing BITA grafting versus single internal thoracic artery(ITA) grafting,⁴ the Radial Artery Patency and Clinical Outcomes trial found that the 10-year patency rate of the radial artery was significantly better than that of the free right ITA: patients who received a radial artery graft had lower rates of mortality at 10 years compared to those undergoing right ITA at 10 years(HR:0.53, 95%%CI:0.30–0.95, p=0.03)⁵. Moreover, a network meta-analysis by Gaudino and colleagues found equivalent outcomes among patients undergoing multi-arterial grafting receiving a radial graft versus those receiving BITA, and that the use of either was better than single arterial grafting with saphenous veins⁶. Finally, in a recent analysis of the Society of Thoracic Surgeons Adult Cardiac Surgery Database, Schwann and colleagues found that early comparative outcomes between the use of a radial graft versus BITA were dependent on surgeon experience and that a strong volume outcome relationship existed¹. Taken together, studies directly comparing ITA to the radial artery as the second arterial conduit at the population level are limited, and the volume-outcome relationship has not been defined for mid-term event free survival. Thus, we sought to determine, at the population level, whether radial artery use at the time of multi-arterial CABG reduces the mid-term incidence of major adverse cardiac events and all-cause mortality when compared to BITA grafting.

Patients and Methods

Study overview

We conducted a retrospective cohort study comparing outcomes up to 4-years in patients ≥ 65 years undergoing first-time isolated CABG with multi-arterial grafting with either a radial artery or BITA using the Centers for Medicaid and Medicare Services administrative data between 2015 and 2022. International Classification of Diseases, 10^th Revision, Clinical Modification(ICD-10) diagnosis and procedure codes(Supplemental Table 1) were used to identify 29,268 patients undergoing isolated coronary artery bypass grafting with at least two arterial grafts. The ICD-9 or ICD-10 diagnosis codes documented in the index hospitalization record and all inpatient admissions within the prior two years were used to identify baseline comorbidities(Supplemental Table 2). After further exclusion of patients with any prior history of cardiac surgery, emergency admission status, and patients with >4 bypass grafts (Medicare does not reliably distinguish the number of bypass grafts after 4), and patients without Medicare continuous enrollment or at least 1-year of Medicare coverage, we identified 15,682 patients undergoing isolated first time CABG with an ITA who received either a second ITA(n=9,989) or radial artery(n=5,784)(Supplemental Figure 1). Validation against clinical records were performed at two large academic cardiac institutions. The positive predictive value for both the BITA and radial artery coding algorithm was 100%. Surgeons were ranked according to ascending annual conduit use and grouped into tertiles as follows: Low-volume radial < 3 arteries or < 2 BITA grafts annually; mid-volume radial 4–10 arteries or 3–4 BITA grafts annually; and high-volume radial > 10 arteries or > 4 BITA grafts annually. The Centers for Medicaid and Medicare Service Data Protection Review Board and the Institutional Review Board at Cedars-Sinai Medical Center approved this study with a waiver of informed consent (STUDY00001188, approved on 11/15/2024).

Study endpoints

The primary endpoint was the composite of major adverse cardiac events, defined as time to first event of: death, acute myocardial infarction, or repeat revascularization (with either CABG or percutaneous coronary intervention(PCI) during a median follow-up time of 2.5 years(interquartile range [IQR]: 1–4 years). The secondary outcome was all-cause mortality. Other outcomes included 30-day mortality, in-hospital stroke, wound complications and sternal complications requiring readmission, index hospitalization length of stay, and 1-year sternal complications. Deaths were identified from the Master Beneficiary Summary File, and non-fatal secondary endpoints were defined using ICD-10 diagnosis and procedure codes(Supplemental Table 3). Stroke was defined as any cerebrovascular accidents documented during the index hospitalization that were not present on admission, and any subsequent hospital admission where the principal diagnosis was hemorrhagic or ischemic stroke. We censored patients without one of the above events of interest on December 31, 2022.

Statistical analysis

Trends in utilization of BITA vs radial artery at the time of multi-arterial grafting were assessed using linear regression. Baseline characteristics were first compared in the overall sample between those receiving BITA vs radial artery. Student’s t-test was used for normally distributed continuous variables, Wilcoxon rank-sum test for non-normally distributed continuous variables, and the Chi-square test for categorical variables. Propensity score matching was performed to account for baseline differences in patient characteristics between BITA and radial artery groups to reduce effects of confounding. The propensity score for each patient was estimated using a multivariable logistic regression model in which the intervention performed was regressed on 26 clinical characteristics thought to influence the choice of intervention or that were prognostically important for the outcome. These included patient comorbidities, procedural characteristics (choice of on vs off pump, number of grafts) and year of procedure. Subjects were matched on the logit of the propensity score using 1:1 greedy nearest-neighbour matching with a caliper distance of 0.2 times the standard deviation of the logit of the propensity score.⁴ Success of matching was assessed by computing the standardized difference of each covariate with a cut-off of 0.1 or less to denote acceptable balance.⁵ Early events were compared between the two cohorts using the McNemar test for binary outcomes, a paired t-test for normally distributed variables, and the Wilcoxon signed rank test for non-normally distributed continuous variables.

For the primary outcome of major adverse cardiac events and secondary outcome of all-cause mortality, Kaplan-Meier survival curves were constructed in the matched sample using a stratified log-rank test stratified on the matched pairs to test the equality of the estimated survival curves.⁷ In addition, hazard ratios were estimated using a Cox-proportional hazards model, which incorporated a robust, sandwich-type variance estimator to account for the matched nature of the data.⁸

For the individual components of major adverse cardiac events that were non-fatal (i.e. new readmission for myocardial infarction and repeat revascularization), cumulative incidence functions were used to estimate the incidence of these events after accounting for death as a competing risk. In the matched sample, equality of cumulative incidence function was assessed using a Fine-Gray sub distribution hazard model in which the sub distribution hazard of the outcome was regressed on a single variable denoting treatment status, with a robust variance estimator to account for the matched nature of the sample.⁶

Subgroup analysis

To determine if the volume of BITA or radial grafting at the surgeon level influenced outcome, the primary and secondary outcomes were compared in a multivariable Cox-proportional hazard model and regressed on annual BITA or radial volume tertiles(low, medium, and high) for patients undergoing BITA or radial grafting, respectively.

In a sensitivity analysis, we included only patients who received in situ BITA vs radial artery grafting in a propensity matched analysis for the primary outcome.

Falsification endpoint analysis

To assess for the risk of residual potential unmeasured confounders after propensity score matching, the association between treatment allocation (BITA vs radial artery) and the incidence of a composite falsification endpoint of hospitalization for urinary tract infection or pneumonia was compared.⁹ This endpoint was chosen as it is unlikely to be related to treatment assignment. The cumulative incidence of this endpoint was evaluated with death as a competing risk.

Statistical significance was assumed for p <0.05. The primary outcome was tested initially, while there was no adjustment for multiplicity for secondary or early outcomes. All analyses were conducted with RStudio(version 1.3.959, RStudio: Integrated Development for R. RStudio, PBC, Boston, MA).

Results

Primary matched analysis

In total, there were 9,898 patients in the BITA group and 5,784 patients in radial artery group. Over the course of the study period, there was an increase in the proportion of patients receiving the radial artery from 23% in 2016 to 43% in 2022(Figure 1, p<0.05) at the time of multi-arterial grafting CABG(Figure 1). Overall, there were few significant differences in important baseline characteristics between the BITA and radial artery groups before propensity matching(Table 1). Those who received radial arteries were slightly younger, more often female, had a lower incidence of previous myocardial infarction, and less likely to require urgent surgery. Propensity matching on 26 baseline covariates (including year of procedure) yielded 5,778 pairs of patients(i.e., 99% of radial patients were matched to BITA), who were well-matched with standardized mean differences<0.10 for all covariates.

Figure 1.

National trends in utilization of the radial artery as the second arterial graft among Medicare patients undergoing multi-arterial grafting for isolated CABG.

Table 1.

Baseline characteristics between bilateral internal thoracic artery (BITA) and radial artery groups before and after propensity score matching. Descriptive statistics are reported as count(%) or either media[Q1,Q3] or mean(standard deviation). SMD, standardized mean difference.

Variables	Before PS Matching				After PS Matching
	BITA (n=9898)	Radial (n=5784)	P Value	SMD	BITA (n=5778)	Radial (n=5778)	P Value	SMD
Age	71.21 (5.00)	70.64 (4.72)	<0.001	0.116	70.70 (4.74)	70.64 (4.72)	0.524	0.012
Female Sex	1957 (19.8)	1004 (17.4)	<0.001	0.062	1020 (17.7)	1003 (17.4)	0.695	0.008
Race			<0.001	0.105			0.912	0.027
Asian	142 (1.4)	98 (1.7)			93 (1.6)	98 (1.7)
Black	359 (3.6)	124 (2.1)			108 (1.9)	124 (2.1)
Hispanic	69 (0.7)	40 (0.7)			40 (0.7)	40 (0.7)
North American Native	35 (0.4)	42 (0.7)			29 (0.5)	36 (0.6)
Other	219 (2.2)	134 (2.3)			137 (2.4)	134 (2.3)
Unknown	372 (3.8)	237 (4.1)			237 (4.1)	237 (4.1)
White	8702 (87.9)	5109 (88.3)			5134 (88.9)	5109 (88.4)
Urgency	2744 (27.7)	1277 (22.1)	<0.001	0.131	1306 (22.6)	1276 (22.1)	0.517	0.012
CHADS2-VASC	4.00 [3.00, 5.00]	4.00 [3.00, 5.00]	0.139	0.035	4.00 [3.00, 5.00]	4.00 [3.00, 5.00]	0.608	0.001
Atrial fibrillation	1899 (19.2)	965 (16.7)	<0.001	0.065	969 (16.8)	963 (16.7)	0.901	0.003
Cancer	2187 (22.1)	1198 (20.7)	0.044	0.034	1174 (20.3)	1197 (20.7)	0.612	0.01
Congestive heart failure	2640 (26.7)	1594 (27.6)	0.235	0.02	1572 (27.2)	1592 (27.6)	0.692	0.008
Chronic kidney disease	2572 (26.0)	1436 (24.8)	0.113	0.027	1434 (24.8)	1436 (24.9)	0.983	0.001
Cardiomyopathy	1663 (16.8)	964 (16.7)	0.845	0.004	970 (16.8)	963 (16.7)	0.881	0.003
Coagulopathy	1036 (10.5)	573 (9.9)	0.277	0.019	568 (9.8)	573 (9.9)	0.901	0.003
Dementia	430 (4.3)	214 (3.7)	0.055	0.033	213 (3.7)	214 (3.7)	1	0.001
COPD	1924 (19.4)	1143 (19.8)	0.637	0.008	1113 (19.3)	1141 (19.7)	0.526	0.012
Cerebrovascular disease	2639 (26.7)	1488 (25.7)	0.206	0.021	1504 (26.0)	1486 (25.7)	0.718	0.007
Dialysis	221 (2.2)	50 (0.9)	<0.001	0.111	118 (2.0)	50 (0.9)	<0.001	0.098
Diabetes	4277 (43.2)	2710 (46.9)	<0.001	0.073	2665 (46.1)	2706 (46.8)	0.456	0.014
Hyperlipidemia	8810 (89.0)	5136 (88.8)	0.704	0.007	5128 (88.8)	5132 (88.8)	0.93	0.002
Hypertension	9320 (94.2)	5476 (94.7)	0.19	0.022	5456 (94.4)	5470 (94.7)	0.594	0.011
Liver disease	1132 (11.4)	689 (11.9)	0.384	0.015	654 (11.3)	687 (11.9)	0.353	0.018
Previous MI	3489 (35.2)	1948 (33.7)	0.048	0.033	1951 (33.8)	1945 (33.7)	0.922	0.002
Previous PCI	2128 (21.5)	1254 (21.7)	0.806	0.004	1227 (21.2)	1253 (21.7)	0.571	0.011
History of pacemaker	318 (3.2)	166 (2.9)	0.25	0.02	174 (3.0)	166 (2.9)	0.7	0.008
Peripheral vascular disease	2714 (27.4)	1593 (27.5)	0.884	0.003	1587 (27.5)	1591 (27.5)	0.95	0.002
Stroke	444 (4.5)	259 (4.5)	1	<0.001	251 (4.3)	259 (4.5)	0.751	0.007
Thromboembolism	2514 (25.4)	1354 (23.4)	0.006	0.046	1382 (23.9)	1351 (23.4)	0.511	0.013
Year of surgery			<0.001	0.334			0.003	0.087
2015	350 (3.5)	108 (1.9)			97 (1.7)	108 (1.9)
2016	1437 (14.5)	446 (7.7)			516 (8.9)	446 (7.7)
2017	1371 (13.9)	535 (9.2)			609 (10.5)	535 (9.3)
2018	1358 (13.7)	720 (12.4)			685 (11.9)	720 (12.5)
2019	1391 (14.1)	899 (15.5)			834 (14.4)	899 (15.6)
2020	1134 (11.5)	844 (14.6)			749 (13.0)	844 (14.6)
2021	1317 (13.3)	1040 (18.0)			1022 (17.7)	1037 (17.9)
2022	1540 (15.6)	1192 (20.6)			1266 (21.9)	1189 (20.6)
Number of bypass grafts	3.00 [3.00, 4.00]	3.00 [3.00, 4.00]	0.733	0.004	3.00 [3.00, 4.00]	3.00 [3.00, 4.00]	1	<0.001

Among propensity matched patients, there was no difference in early 30-day mortality between the BITA and radial artery patients (1.2% vs 1.2%, p=0.99). The rate of new in-hospital stroke was also similar between BITA and radial patients (1.0% vs 0.9%, p=1.00). There was no difference in index length of hospitalization (median 6.0 days, IQR: 5.0–9.0 in both groups, p=0.43). Rates of 1-year sternal complications were higher in the BITA group (1.7% vs 1.1%, p=0.009). Similarly, 1-year mortality was higher in the BITA group (3.7% vs 3.0%, p=0.03). Early outcomes before and after propensity matching are provided in Supplemental Table 4.

At 4-year follow-up, the primary outcome of major adverse cardiac events was lower in the radial artery group (14.7% vs 15.7%, HR:0.88, 95%CI:0.77–1.00, p=0.050, Figure 2). There was no difference in 4-year mortality between the BITA and radial groups (11.5% vs 10.8%, HR:0.90, 95%CI:0.79–1.02, p=0.06, Figure 3).

Figure 2.

Cumulative incidence of MACE at 4 years in patients receiving a radial artery as their second arterial graft vs patients receiving a second ITA after propensity score matching (14.7% vs 15.7%, HR:0.88, 95% CI:0.77–1.00, p=0.050).

Figure 3.

Cumulative incidence of death at 4 years in patients receiving a radial artery as their second arterial graft vs patients receiving a second ITA after propensity score matching (11.5% vs 10.8%, HR:0.90, 95% CI:0.79–1.02, p=0.061).

When the components of our composite outcome were analyzed separately with death as a competing risk, we found that the cumulative incidence of acute myocardial infarction was 2.8% amongst both groups (sub-distribution HR: 0.98, 95%CI:0.76–1.27, p=0.89). There was no difference in the cumulative incidence of repeat revascularization with PCI or CABG between the BITA and radial group at 4-years (3.2% vs 3.1%, sub-distribution HR:0.96, 95%CI:0.76–1.22, p=0.75). Hazard ratios for outcomes before and after propensity matching are provided in Supplemental Table 4.

In a subgroup analysis of patients undergoing BITA grafting only (n=9,898), receipt of BITA from low volume BITA surgeons(<2 cases per year) was associated with higher incidence of major adverse cardiac events at 4-years (Figure 4, HR:1.15, 95%CI:1.01–1.33, p=0.046) when compared to high volume surgeons (≥9 BITA cases per year). There was no difference in the incidence of major adverse cardiac events between medium volume surgeons (3–8 BITA cases per year) when compared to high volume surgeons (HR:1.13, 95%CI:0.99–1.29, p=0.07). There was no difference in the incidence of death at 4-years between low volume BITA(Supplemental Figure 2, HR:1.14, 95%CI:0.98–1.35, p=0.09) and medium volume BITA (HR:1.11, 95%CI:0.96–1.30, p=0.16) when compared to the referent group of high volume BITA surgeons. In patients undergoing radial artery grafting (n=5,784), there was no difference in the incidence of 4-year major adverse cardiac events between low volume radial surgeons (<3 radial cases per year, HR:0.99, 95%CI:0.81–1.21, p=0.95) and medium volume radial surgeons (3–14 radial cases per year, HR:0.88, 95%CI:0.70–1.10, p=0.25) when compared to high volume surgeons (≥15 radial cases per year, Figure 5). Similarly, there was no difference in the incidence of mortality at 4-years between low volume radial surgeons(Supplemental Figure 3, HR:1.12, 95%CI:0.89–1.41, p=0.34) and medium volume radial surgeons(HR: 0.92, 95%CI:0.70–1.21, p=0.55) when compared to high volume radial surgeons. In a subgroup analysis, when in situ BITA was compared to the radial artery, there was no difference in the incidence of major adverse cardiac events(MACE) at 4-years (Supplemental Figure 4, 15.1% vs 14.8%, HR:0.95, 95%CI:0.84–1.09).

Figure 4.

Freedom from MACE in patients receiving a second ITA graft only at 4 years stratified by surgeon volume. Low volume surgeons were associated with a higher incidence of MACE at 4-years (HR:1.15, 95% CI:1.01–1.33, p=0.046) compared to high volume surgeons.

Figure 5.

Freedom from MACE among patients receiving a radial artery graft as their second arterial graft only at 4 years stratified by surgeon volume. There was no difference in the incidence of MACE at 4 years when stratified by surgeon volume.

In a falsification endpoint analysis, there was no difference in the 4-year cumulative incidence of the urinary tract infection or pneumonia in the matched cohort comparing BITA and radial patients (Supplemental Figure 5, 5.0% vs 5.0%%, HR:0.95, 95%CI:0.0.79–1.13, p=0.53) suggesting little residual confounding.

In a sensitivity analysis, patients undergoing multi-arterial grafting by low, medium, and high volume surgeons were compared to patients undergoing single arterial grafting irrespective of the surgeon. Among patients treated by low volume multi-arterial grafting surgeons, there was no difference in the primary outcome of major adverse cardiac events when compared to patients receiving single arterial grafts (HR: 0.93, 95CI%:0.88–1.01, p=0.053).

Comment

The use of multiple arterial conduits has been shown to improve survival in patients up to the age of 70-years, yet optimal arterial conduit selection in older patients remain unknown¹. In this analysis of patients > 65 years undergoing multi-arterial coronary artery bypass grafting, the second most used arterial conduit was an additional ITA. However, radial artery usage doubled throughout the study period. Radial artery usage was associated with less sternal complications and a marginal benefit in terms of freedom from major adverse cardiac events, but no difference in survival when compared to the use of BITA at 4-years. Importantly, we found a relationship between surgeon volume and mid-term outcomes in the BITA group, but not in the radial group. This suggests that the benefits of multi-arterial grafting may be lost when using BITA graft by low-volume surgeons.

While current revascularization guidelines recommend usage of the radial artery (Class I) over the saphenous vein as the second conduit of choice in appropriate patients undergoing multi-arterial grafting, the use of BITA has a class IIa recommendation over the saphenous vein^2,3. The class difference in recommendation relates to the differential strength of evidence supporting either conduit: while there is up to 10-years of randomized evidence to support the radial artery over the saphenous vein from a patient level meta-analysis by Gaudino and colleagues,⁷ evidence from ART showed no difference between a BITA grafting strategy compared to saphenous vein grafting at 10-years⁴. As such, evidence to support the survival benefits of BITA is limited to observational studies. The ART did provide several important insights into using two ITA grafts: the cross-over rate from the second ITA arm to the single ITA and saphenous vein group was high (14%), highlighting concerns regarding the impact of surgeon experience on treatment received for patients in this trial⁴. Findings from our analysis support that a volume-outcome relationship exists for multi-arterial grafting with BITA graft that does not appear to exist with radial artery grafting. Compared to BITA, the use of the radial artery resulted in no worse and potentially slightly better outcomes. This may also be related to the ease of use of a radial artery graft compared to the second ITA in terms of targets reached and possible configurations.

Furthermore, in our analysis, the use of BITA was associated with higher rates of sternal complications at 1-year, a highly morbid complication that may influence event free survival. In their assessment of the impact of ITA harvesting strategy on sternal wound complication in the ART, Benedetto and colleagues showed that careful patient selection and conduit harvesting technique can help mitigate the increased risk for sternal complications in BITA patients⁸. Even then, the use of BITA may be more technically complex than the use of a radial artery. In particular, an in situ second ITA may not reach all distal CABG targets, which can influence graft configuration and potentially affect long-term outcomes. In fact, in the Radial Artery Patency and Clinical Outcomes trial, the use of the second ITA as a free graft was associated with worse event free survival compared to the radial artery.⁵ Additionally, the recent publication of a post-hoc analysis of patients undergoing CABG in the Cardiovascular Outcomes for People Using Anticoagulation Strategies trial showed a significantly higher (p=0.008) graft failure rate for those with BITA (26.8%) at 1-year when compared to those with a radial artery graft (9.9%) and venous graft (10.4%), p=0.008.⁹ For this reason, we performed a sensitivity analysis comparing in situ BITA grafting versus radial artery grafting. However, we found that the modest event-free survival benefit seen with the radial artery was no longer significant.

While CABG is the most common operation performed in cardiac surgery, multi-arterial grafting rates remain relatively low, increasing from 10.9% in 2016 to 14.3% in 2021. BITA use has similarly remained low during this period, accounting for only 6.7% of all isolated CABG procedures in 2021.¹⁰ The limited use of multi-arterial grafting at the national level, combined with our findings of a volume-outcome relationship particularly in the usage of BITA graft compared to a radial graft as the second arterial conduit among patients undergoing multi-arterial grafting strengthen existing arguments for CABG subspecialisation.¹¹ Sub-specialization in coronary revascularization may lead to improvements in outcomes following CABG, which until now, have remained relatively stagnant.^11,12

This study must be interpreted in the context of some important limitations. First, while propensity matching was used to account for known confounders, unmeasured or unknown confounders may still exist. Treatment selection and allocation to receiving BITA or radial artery is not known, and we reiterate that the low use of BITA in the low volume surgeon group may be related to unforeseen conduit harvesting issue forcing the surgeon to use a second ITA. Moreover, our assessment of surgeon volume is limited to cases done within the Medicare population-it is possible that low-volume surgeons may have performed a greater number of cases in a younger patient population. Nonetheless, in this analysis, patients were quite similar prior to matching in that all patients in the treatment group were matched to a control patient. Furthermore, we performed a falsification endpoint analysis which was not different between the two groups. Also, the two-year lookback period used may miss prior cardiac surgery occurring before this period, which could lead to misidentification of some re-operative cases. There are limitations in terms of data granularity from the Medicare dataset, specifically, we do not have information on pre-operative angiogram and thus the extent of coronary artery disease or pre-operative echocardiographic data such as left ventricular ejection fraction or data on graft configuration or conduit harvesting techniques are unknown. However, in our analysis, we did match on the number of bypass grafts, a surrogate for the extent of coronary artery disease. This analysis relied on administrative coding for ascertainment for non-fatal outcomes such as acute myocardial infarction rather than a biochemical definition of myocardial infarction. Furthermore, we do not have any information on graft patency, only data around repeat revascularization with CABG or PCI. Also, we were unable to determine the proportion of BITA patients who were unsuitable for radial artery use due to unmeasured factors, coupled with the preferential use of radial conduits for critically stenosed targets (≥70%), may have influenced observed rates of repeat revascularization. Finally, our dataset did not include discharge medications. As such, we do not have information on optimal medical therapy after CABG between the two groups.

Conclusion

In summary, the use of the radial artery at the time of multi-arterial grafting was associated with less sternal wound complications at 1-year, while overall mid-term outcomes were comparable between conduit types. A surgeon volume effect was observed only for BITA use. Consistent with current guideline recommendations, the radial artery may be the preferred conduit in older patients undergoing multi-arterial grafting.