Early and long-term outcomes of bilateral internal mammary artery versus radial artery grafting: A national cohort study Authors: Juan A Montero, Santiago Cubas, Maximiliano Hernandez, Carolina Sosa, Gerardo Soca, Víctor Dayan Published: 2026 ABSTRACT AbstractObjectiveTo compare early and long-term outcomes of bilateral internal mammary artery (BIMA) grafting versus left internal mammary artery (LIMA) plus radial artery (RA) grafting in patients undergoing isolated coronary artery bypass grafting (CABG).MethodsThis retrospective national cohort study included patients undergoing isolated CABG with multiarterial grafting in Uruguay between 2002 and 2022. Patients undergoing BIMA grafting or LIMA+RA grafting were identified from a mandatory national cardiac surgery database. Propensity score matching with multiple imputation was used to balance baseline characteristics between the 2 groups. Long-term survival was assessed using Kaplan-Meier analysis, Cox proportional hazards models, and restricted mean survival time (RMST). Competing-risk regression (Fine-Gray) was used to evaluate repeat percutaneous coronary intervention. Effect modification by sex and diabetes was explored.ResultsAmong 2882 patients (2132 BIMA and 750 LIMA+RA), 4203 matched (with replacement) patients were analyzed. At 10 years, RMST did not differ between groups (9.06 years for BIMA vs 8.98 years for LIMA+RA; P = .36), and BIMA grafting was associated with improved long-term survival (hazard ratio, 0.73; 95% confidence interval, 0.55-0.93; P = .03) which was not supported by sensitivity analyses. There was no interaction with sex or with diabetes. BIMA grafting was associated with lower operative mortality but with a higher rate of mechanical ventilation and longer duration of inotrope use. No between-group differences were observed in major ischemic events, deep sternal wound infection, or repeat revascularization.ConclusionsIn this national cohort, BIMA grafting improved early outcomes but did not confer a consistent overall long-term survival advantage compared with LIMA+RA grafting. A lack of support from sensitivity analyses implies the need for an individualized approach.Key Words: coronary artery bypass grafting, bilateral internal mammary artery, radial artery, multiarterial grafting, long-term survival, sex differences, clinical outcomes Multiarterial grafting. Central Message Bilateral internal mammary artery (IMA) grafting improves early outcomes but does not confer a consistent overall long-term survival advantage compared with left IMA + radial artery grafting after coronary artery bypass grafting surgery. Perspective Although multiarterial grafting is associated with favorable early outcomes, the optimal second arterial conduit remains a topic of debate. In this national 2-decade cohort, bilateral internal mammary artery grafting did not consistently improve long-term survival overall but was associated with a short-term survival benefit, supporting a patient-specific approach to arterial conduit selection. Coronary artery bypass grafting (CABG) has consistently proven to be one of the most effective approaches to revascularization in patients with obstructive coronary artery disease (CAD). The use of multiarterial grafting (MAG) with one or both internal mammary arteries (IMAs) and the radial artery (RA) has been associated with better early outcomes and a lower incidence of long-term cardiovascular complications.1 Since the 1950s, interest has grown in using the IMA to supply blood to ischemic areas of the heart. The first successful use of the right IMA (RIMA) dates to the 1960s, while use of the RA was first reported by Carpentier in 1973.1,2 Seventy years later, although grafting of the LIMA to the left anterior descending artery is the undisputed gold standard, the question of the optimal second conduit remains a subject of ongoing debate. CABG remains a cornerstone in the management of advanced CAD. Although the saphenous vein (SV) remains the most frequently used conduit, increasing evidence supports the use of arterial grafts, particularly the RIMA and the RA, as alternatives associated with improved long-term outcomes.1,3, 4, 5, 6, 7, 8, 9 Observational studies and secondary analyses of randomized trials have consistently suggested that greater arterial revascularization may lead to enhanced graft patency and reduced mortality4,10; however, these findings have not been conclusively validated by randomized controlled trials (RCTs). The Arterial Revascularization Trial (ART), the largest RCT reported to date, failed to demonstrate a survival benefit for BIMA grafting over single IMA use at 10 years.11 In contrast, a comprehensive network meta-analysis involving approximately 150,000 patients indicated that both BIMA and LIMA+RA grafting strategies were associated with similar and significant reductions in long-term mortality compared to LIMA+SV grafting.12 These findings have renewed interest in defining the optimal second arterial conduit for CABG, particularly in the context of long-term survival and clinical outcomes. Methods This retrospective national cohort study was based on the adult cardiac surgery database of the National Resources Fund (NRF) of Uruguay. The study was approved on October 30, 2023, by the Institutional Review Board of the British Hospital, Montevideo, Uruguay, which waived the need for individual patient consent. The NRF is the governmental agency in charge of financing 100% of cardiac surgery procedures in Uruguay. The NRF database includes mandatory data entry by trained personnel using standardized protocols. Vital status was determined through deterministic linkage with the Ministry of Health's National Mortality Registry, which ensures complete follow-up for all patients within the country. Patients Patients were included who underwent isolated CABG between January 1, 2002, and December 31, 2022. Patients requiring resuscitation prior to surgery, emergency procedures, or reoperation and those who underwent only a single bypass were excluded. Only those patients who underwent either BIMA or LIMA+RA were included in the final analysis. The primary outcome was survival. Subgroup analysis was performed according to diabetes and sex. The secondary outcome was a composite of operative mortality, postoperative stroke, deep sternal wound infection (DSWI), and postoperative acute myocardial infarction (AMI). Individual components of the secondary composite outcome were assessed independently, along with prolonged inotropic support, prolonged mechanical ventilation, respiratory infection, and postoperative atrial fibrillation. The need for a new percutaneous coronary intervention (PCI) was assessed as a secondary outcome. Definitions Operative mortality was defined as in-hospital death occurring any time during the index hospitalization or within 30 days of CABG after discharge. Postoperative stroke was defined as stroke occurring during the first 30 days after surgery. Prolonged inotropic support was defined as support for >24 hours after surgery; prolonged mechanical ventilation, as support provided for >12 hours after surgery. Postoperative atrial fibrillation was defined as atrial fibrillation occurring before hospital discharge. Complete revascularization was defined as at least 1 bypass in a major coronary system (left anterior descending, circumflex, or right coronary) with >50% stenosis. Statistics Continuous variables are reported as mean with standard deviation. Categorical variables are presented as absolute value (%) and compared using Fisher's exact test. To address potential confounding in the comparison of patients undergoing CABG with RIMA grafting versus RA grafting, we used a propensity score matching strategy. Missing baseline covariate data were handled using multiple imputation by chained equations under the assumption that data were missing at random. One imputed dataset was used for the matching process. A logistic regression model was used to estimate the propensity score for receiving a RIMA graft. The model included the following covariates selected a priori based on clinical relevance: age, sex, weight, height, diabetes mellitus, prior stroke, dyslipidemia, chronic obstructive pulmonary disease, chronic renal failure, peripheral vascular disease, smoking status, hypertension, prior AMI, left ventricular ejection fraction (LVEF), serum creatinine, and New York Heart Association functional class III/IV. Propensity score matching was performed using 1:3 nearest-neighbor matching with replacement, allowing matching of each control patient (RA group) to multiple treated patients (RIMA group) if selected as the closest match. A caliper width of 0.2 times the standard deviation of the logit of the propensity score was applied to restrict matches to those within a clinically meaningful range of similarity. Covariate balance between matched groups was assessed using standardized mean difference, with a value <0.1 considered to indicate acceptable balance (Figure 1). Raw outcomes in the original unmatched cohorts were additionally reported as a supplementary analysis to provide unweighted estimates of event rates (Table E1). Post-treatment variables such as complete revascularization, cross-clamp time and number of bypasses were not included in the matching algorithm but were adjusted for in outcome regression models. Cox proportional hazards and logistic regression models were fitted using cluster-robust standard errors at the patient level to account for the nonindependence of observations introduced by propensity score matching with replacement. Balance diagnostics after matching with multiple imputation, replacement, and a caliper width of 0.2 times the standard deviation of the logit of the propensity score. SMD, Standardized mean difference; COPD, chronic obstructive pulmonary disease; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association. Survival was assessed with the Kaplan-Meier method. We compared survival between patients using both a Cox proportional hazards model and restricted mean survival time (RMST) analysis. The proportional hazards assumption was tested using scaled Schoenfeld residuals and was not violated. Nevertheless, RMST also was used to provide an absolute and clinically interpretable measure of average survival time up to 5 years and 10 years of follow-up. Cox regression and logistic regression were performed in the matched cohort to adjust for imbalances (sex, CRF, previous stroke, LVEF, smoker status, and year of surgery). The primary analysis included adjustment for clinically relevant covariates to account for residual imbalance after matching. An additional unadjusted model (adjusting only for year of surgery) was performed as a sensitivity analysis. Interaction analyses were performed using the Wald test for the interaction term in the clustered Cox model. The results are reported as hazard ratio (HR) with 95% confidence interval (CI), representing the relative hazard for RIMA relative to RA. To assess the time to new PCI following CABG, while accounting for the competing risk of death, we used a competing-risk regression model based on the Fine and Gray subdistribution HR (sHR) approach. The primary event, defined as the occurrence of PCI and death prior to PCI, was treated as a competing event. Patients who experienced neither event during follow-up were censored at their last observation. Results are presented as sHR with corresponding 95% CI and P value. We generated cumulative incidence function curves to visually compare the cumulative incidence of PCI between groups while accounting for the competing risk of death. Results During the study period, 2882 patients underwent isolated CABG with MAG. Of these, 2132 received BIMA grafts and 750 received LIMA+RA grafts. The median follow-up was 18.2 years (95% CI, 18.1-18.3 years) for the RA group and 9.1 years (95% CI, 8.7-9.4 years) for the BIMA group. In the original cohort, the proportion of females and the incidence of diabetes were higher in the RA group, while the rate previous AMI was lower (Table 1). RIMA patients also were younger and had better LVEF. After propensity score matching, we were able to generate comparable groups of 2071 patients in the BIMA group and 2132 patients in the RA group (Figure 1). The 2 groups were similar except for a higher proportion of females (13.2% vs 10.5%) and higher rates of stroke (1.7% vs 0.8%) and smoking (38.9% vs 35.1%) in the RIMA group and higher incidences of CRF (4.0% vs 2.5%) and peripheral vascular disease (6.2% vs 4.6%) and a slightly better LVEF (54.9% vs 53.9%) in patients who underwent RA grafting. Table 1Demographic and procedural characteristics of patients who received multiarterial revascularizationVariableOverall cohortPropensity score–matched cohortAll patients (N = 2882)RIMA (N = 2132)Radial (N = 750)P valueAll patients (N = 4203)RIMA (N = 2071)Radial (N = 2132)SMDFemale sex, n (%)453 (15.7)284 (13.3)169 (22.5)<.001498 (11.9)274 (13.2)224 (10.5)0.08Age, y, mean (SD)57.0 (8.8)56.2 (8.5)59.3 (9.1)<.00156.3 (8.9)56.2 (8.5)56.4 (9.2)0.02Hypertension, n (%)1807 (62.7)1348 (63.2)459 (61.2).3232705 (64.4)1309 (63.2)1396 (65.5)0.05Diabetes, n (%)777 (27.0)487 (22.8)290 (38.7)<.001937 (22.3)472 (22.8)465 (21.8)0.02Stroke, n (%)52 (1.8)36 (1.7)16 (2.1).43153 (1.3)35 (1.7)18 (0.8)0.08Dyslipidemia, n (%)1660 (57.6)1240 (58.2)420 (56.0).3032414 (57.4)1203 (58.1)1211 (56.8)0.03CRF, n (%)74 (2.6)54 (2.5)20 (2.7).842136 (3.2)51 (2.5)85 (4.0)0.08PVD, n (%)149 (5.2)103 (4.8)46 (6.1).166228 (5.4)96 (4.6)132 (6.2)0.06Smoker, n (%)1106 (38.4)823 (38.6)283 (37.7).6741554 (37.0)805 (38.9)749 (35.1)0.07AMI, n (%)704 (24.4)622 (29.2)82 (10.9)<.001792 (18.8)370 (17.9)422 (19.8)0.06LVEF, %, mean (SD)52.3 (14.4)53.9 (11.6)47.7 (19.6)<.00154.5 (11.7)53.9 (11.4)54.9 (11.4)0.09Creatinine, mg/dL, mean (SD)1.05 (0.74)1.05 (0.74)1.06 (0.73).5231.05 (0.69)1.05 (0.75)1.06 (0.64)0.01OPCAB, n (%)2093 (72.6)1559 (73.1)534 (71.2).3093165 (75.3)1514 (73.1)1615 (77.4)0.09RIMA, Right internal mammary artery; SMD, standardized mean difference; CRF, chronic renal failure; PVD, peripheral vascular disease; AMI, acute myocardial infarction; LVEF, left ventricular ejection fraction; OPCAB, on-pump coronary artery bypass. Demographic and procedural characteristics of patients who received multiarterial revascularization RIMA, Right internal mammary artery; SMD, standardized mean difference; CRF, chronic renal failure; PVD, peripheral vascular disease; AMI, acute myocardial infarction; LVEF, left ventricular ejection fraction; OPCAB, on-pump coronary artery bypass. The annual total numbers of each procedure changed from 2002 to 2022 (Figure 2). During the study period, a marked temporal variation in arterial grafting strategies was observed. Annual volume of bilateral internal mammary artery (BIMA) and right atrial (RA) grafting. Survival At 10 years, the difference in RMST between the 2 groups was not statistically significant (difference: 0.141 years; 95% CI, −0.020 to 0.303 years; P = .086). However, in a sensitivity analysis at 5 years, a statistically significant difference favoring the BIMA group was observed (difference: 0.082 years; 95% CI, 0.015-0.149 years; P = .017). The use of BIMA was associated with a significantly lower risk of long-term mortality compared with RA grafting (HR, 0.73; 95% CI, 0.55-0.93; P = .03) (Figure 3, A). In a sensitivity analysis without additional covariate adjustment, the association between BIMA and survival remained directionally consistent but did not reach statistical significance (HR, 0.78; 95% CI, 0.58-1.04; P = .095) (Figure 3, B). No interaction was found with diabetes (P = .333) or sex (P = .439). Survival after coronary artery bypass grafting CABG in patients who received bilateral internal mammary artery (BIMA; red) grafts or left internal mammary artery (LIMA) + right atrial (RA; blue) grafts. A, Survival after adjustment for unbalanced covariates. B, Survival without adjustment except for year of surgery. Shaded areas represent 95% confidence interval (CI). HR, Hazard ratio. Postoperative Outcomes The risk for the composite outcome was similar in the 2 groups. Operative mortality was lower with BIMA (odds ratio, 0.38; 95% CI, 0.148-0.954; P = .040). No between-group differences were found in postoperative AMI, stroke; or DSWI; however, the risk of prolonged mechanical ventilation and prolonged inotropic support was higher with BIMA (Table 2). The operative and postoperative outcomes in patients who underwent multiarterial revascularization modeled with year of surgery are also shown (Table E2). Table 2Operative and postoperative outcomes in patients who underwent multiarterial revascularizationVariableAll patients, n (%)RIMA, n (%)Radial, n (%)P valueaOR (95%CI)P valueComposite outcome270 (6.4)117 (5.7)153 (7.2).0440.70 (0.36-1.34).281Operative mortality137 (3.3)46 (2.2)91 (4.3)<.0010.38 (0.148-0.954).040DSWI23 (0.6)16 (0.8)7 (0.3).0512.24 (0.313-16.07).421Post-AMI98 (2.3)50 (2.4)48 (2.3).7261.50 (0.71-3.15).284Stroke29 (0.7)17 (0.8)12 (0.6).3121.51 (0.34-6.77).590Inotropes for >24 h1554 (37.0)804 (38.8)750 (35.2).0141.43 (1.06-1.94)<.019Prolonged MV1076 (25.6)596 (28.8)480 (22.5)<.0011.41 (1.03-2.08)<.036Respiratory infection358 (8.5)209 (10.1)149 (7.00)<.0011.34 (0.76-2.37).306POAF507 (12.1)288 (13.9)219 (10.3)<.0011.23 (0.82-1.85).314Multivariate analysis was performed using the following covariates: age, diabetes, previous AMI, left ventricular ejection fraction, sex, and complete revascularization. RIMA, Right internal mammary artery; aOR, adjusted odds ratio; CI, confidence interval; DSWI, deep sternal wound infection; AMI, acute myocardial infarction; MV, mechanical ventilation; POAF, postoperative atrial fibrillation. Operative and postoperative outcomes in patients who underwent multiarterial revascularization Multivariate analysis was performed using the following covariates: age, diabetes, previous AMI, left ventricular ejection fraction, sex, and complete revascularization. RIMA, Right internal mammary artery; aOR, adjusted odds ratio; CI, confidence interval; DSWI, deep sternal wound infection; AMI, acute myocardial infarction; MV, mechanical ventilation; POAF, postoperative atrial fibrillation. Freedom From New Revascularization The median time to revascularization was 17.9 years in the RA group and 11.3 years in the BIMA group. In a competing-risks analysis using the Fine-Gray sHR model, the associations between BIMA and RA and the incidence of PCI was evaluated, accounting for death as a competing event. Among the 4203 patients included in this analysis, 516 experienced the event of interest (ie, PCI), 1089 experienced a competing event, and 2598 were censored during follow-up. None of the revascularization strategies was significantly associated with the cumulative incidence of PCI (sHR, 0.88; 95% CI, 0.74-1.05; P = .15). Model-based cumulative incidence curves derived from the Fine-Gray model showed a slightly lower, but largely overlapping, incidence of PCI over time among patients with BIMA grafts compared to those with LIMA+RA grafts, consistent with the regression results (Figure 4). No interaction was found with either sex or diabetes. Cumulative incidence of post-CABG PCI in patients receiving bilateral internal mammary artery (BIMA) grafts vs left internal mammary artery (LIMA) + right atrial (RA) grafts. Discussion Current evidence consistently suggests that both BIMA grafting and RA grafting demonstrate superior long-term performance compared with SV grafting1,3, 4, 5, 6, 7, 8,13; however, to date only a limited number of RCTs have directly compared RITA and RA conduits, underscoring the ongoing need for adequately powered trials to guide optimal second arterial conduit selection. We recently reported the largest and longest follow-up experience in Latin America.13 The RAPCO trial represents the first RCT to suggest a potential clinical advantage of the RA over both the BIMA and the SV as a second conduit in CABG.14 RA use was associated with a lower incidence of 15-year major adverse cardiac events and a trend toward improved survival, with no significant difference in the rate of AMI or repeat revascularization. Despite its randomized design and long-term follow-up, the generalizability of RAPCO is limited by its highly selected, relatively low-risk patient population, performance in specialized centers, modest sample size with limited statistical power for several endpoints, and heterogeneity in target vessel selection and graft configuration. Similarly, the ART demonstrated no difference in 10-year mortality between BIMA and single IMA grafting strategies, despite greater arterial graft use in the BIMA group, but found a higher incidence of sternal wound complications with BIMA grafting.11 These findings challenged the presumed long-term survival advantage of BIMA grafting. A subsequent post hoc analysis suggested a stepwise improvement in outcomes with increasing arterial graft use, with total arterial revascularization showing the greatest benefit; however, these observations remain hypothesis-generating and require confirmation in randomized studies.4 In our 2-decade observational cohort, we observed distinct and opposing temporal trends in arterial conduit use for CABG (Figure 2). BIMA use increased steadily from the mid-2000s, reaching a peak around 2015-2016, followed by a progressive decline, particularly after 2016 and again after 2019. In contrast, RA use was higher in the early study years, subsequently declined, and then demonstrated a modest but consistent increase in the later period. These opposing trends likely reflect a dynamic shift in MAG strategies over time. Notably, the decline in BIMA use temporally coincided with publication of the 5- and 10-year ART results, which reported no survival benefit and an increased risk of sternal complications with BIMA.4,11 These findings may have contributed to a more selective use of BIMA grafts in contemporary practice, while the stabilization and gradual increase in RA use suggest its adoption as an alternative arterial conduit with a more favorable perceived safety profile. This temporal shift also explains the shorter follow-up observed in the BIMA group (9.1 years vs 18.2 years) and highlights the influence of emerging randomized evidence on real-world surgical decision making. While there was difference in 10-year survival between the 2 groups, after adjusting for baseline differences, the Cox model showed an association between BIMA use and better survival. The attenuation of the treatment effect in unadjusted analyses suggests that the observed association may be sensitive to model specification, highlighting the importance of cautious interpretation and individual decision making. The discrepancy between the Cox and RMST analyses reflects that these methods estimate different quantities. While the Cox model provides a relative effect (ie, HR), RMST captures an absolute difference in survival time. In our study, the small and nonsignificant RMST difference suggests limited clinical relevance despite a statistically significant HR. In addition, given the violation of the proportional hazards assumption for some covariates, RMST offers a robust complementary measure. The discrepancy between truncation times with RMST suggests that the observed benefit may be more evident in the early follow-up period, while estimates at longer horizons are more uncertain, likely reflecting reduced numbers at risk and increased censoring. Additionally, the use of BIMA grafts peaked around 2015, while the use of RA grafts peaked around 2005. This 10-year difference introduced residual confounding variables, such as improved postoperative care and increased surgical experience, which may have contributed to the early survival advantage of BIMA grafting. No interaction was observed with diabetes or sex. Evidence supporting a sex-specific effect of MAG remains suggestive but inconclusive. Observational studies have reported sex-related differences in outcomes, including higher rates of DSWI and altered mortality patterns in women, which may explain in part the historically lower use of BIMA grafts in female patients.15, 16, 17 Importantly, in all published randomized CABG trials, the treatment effect of MAG differed by sex and tended to be larger in women. In the ART trial, the HR for the MAG effect was 1.00 (95% CI, 0.84-1.18) in men versus 0.78 (95% CI, 0.53-1.13) in women; however, women accounted for only 14% of the study population, and the P value for interaction was not statistically significant.4,11 In the RADIAL analyses, sex emerged as a significant treatment effect modifier, suggesting a greater benefit from MAG for women.18 Similarly, in RAPCO, subgroup analysis showed a greater benefit of RA use in women compared to men,14 although female patients represented only 19.1% of the cohort. Collectively, these findings suggest a potentially larger benefit of MAG in women; however, existing trials are underpowered to formally test this hypothesis owing to the consistent underrepresentation of female patients. Previous evidence also supports a volume–outcome relationship in BIMA grafting. Centers with higher rates of BIMA use have reported lower long-term mortality and DSWI rates, as well as improved survival among BIMA recipients.3 In the ART trial, only 83.6% of patients randomized to BIMA ultimately received the assigned strategy, compared with 96.1% in the conventional CABG group.4,11 This high crossover rate underscores the greater technical complexity of BIMA grafting, even among surgeons with documented expertise, and highlights the potential importance of surgical experience and institutional volume in optimizing outcomes. In the present study, BIMA use was associated with a significantly lower composite outcome compared with the RA strategy, driven primarily by a reduction in operative mortality. Importantly, no significant differences were observed in major postoperative complications such as AMI, stroke, and DSWI, supporting the overall safety of BIMA use in appropriately selected patients. Nevertheless, BIMA grafting was associated with higher rates of certain postoperative morbidities, including prolonged mechanical ventilation and inotropic support. These findings likely reflect the greater operative complexity and physiologic burden associated with BIMA harvesting and underscore the need for careful patient selection and perioperative management. Additionally, previous studies have shown a higher risk of atelectasis in patients undergoing BIMA harvest, probably because of lung collapse and manipulation.19 This could explain the longer mechanical ventilation time in this group of patients. The apparent paradox of lower operative mortality with BIMA despite a higher rate of perioperative morbidity may reflect residual confounding. In particular, selection bias and differences in surgeon experience or institutional practice patterns may have influenced both conduit selection and outcomes. This argument was recently raised by Tam and colleagues,20 who found a surgeon–volume effect for BIMA grafting not evident in RA grafting. Given the observational nature of our study and the absence of surgeon-level data, these factors could not be directly accounted for and should be considered when interpreting the results. With respect to late outcomes, BIMA use was not associated with a significantly increased sHR for subsequent PCI. Taken together, our findings suggest that BIMA grafting may confer an early survival advantage without increasing the risk of major ischemic or infectious complications, albeit at the expense of higher perioperative morbidity. These results reinforce the concept that the choice of a second arterial conduit should be individualized, balancing the potential benefits of BIMA grafting against its increased technical complexity, particularly in patients at higher risk for respiratory complications. Beyond operative strategy, long-term outcomes after CABG are strongly influenced by the quality of postoperative secondary prevention. Optimization of antiplatelet therapy, lipid-lowering treatment, glycemic control, and blood pressure management has a central role in determining graft patency and long-term survival, irrespective of the conduit strategy used. As advances in medical therapy continue to improve outcomes across all revascularization approaches, the relative impact of conduit selection may become less pronounced. In the contemporary era of coronary revascularization, when percutaneous techniques are highly developed and surgery is increasingly reserved for patients with extensive multivessel disease, the relative superiority of individual arterial conduits as the second graft may be of limited practical relevance. Future research may need to focus less on direct conduit comparisons and more on defining the incremental value of maximizing arterial revascularization strategies and achieving more extensive, if not fully arterial, coronary revascularization. Limitations This study has several limitations that should be considered when interpreting the findings. First, its retrospective and observational design introduces the potential for residual confounding despite the use of propensity score matching and regression adjustment. Although a comprehensive set of baseline variables was included, unmeasured confounders still might have influenced treatment allocation and outcomes. Second, the use of propensity score matching with replacement introduced nonindependence of observations. Although this was addressed through the use of cluster-robust standard errors, some degree of model dependence cannot be excluded, as suggested by the attenuation of treatment effects in sensitivity analyses. Third, some important procedural details were not available in the national registry. Specifically, data on graft configuration (eg, pedicled vs free or composite RIMA), conduit-to-target vessel assignment, and degree of target vessel stenosis were not captured. These technical factors may influence both early and long-term outcomes and could contribute to heterogeneity within treatment groups. Fourth, surgeon-level variables, including experience, case volume, and operative strategy, were not available. These factors may have influenced both conduit selection and outcomes and may partly explain the observed differences in early mortality and perioperative morbidity. Finally, follow-up duration differed between groups owing to temporal changes in surgical practice, which may affect long-term comparisons despite the use of complementary methods such as Cox regression and restricted mean survival time. Conclusions In this large national cohort, BIMA grafting was associated with improved early outcomes compared with an RA-based strategy, without an increase in major ischemic or infectious complications. However, no consistent long-term survival advantage was observed across analytical approaches. While adjusted models suggested a potential survival benefit, this finding was not robust in sensitivity analyses and was not confirmed by RMST at 10 years. These results suggest that any long-term survival benefit of BIMA grafting is likely modest and context-dependent. Accordingly, arterial conduit selection should be individualized, taking into account patient characteristics, surgical complexity, and institutional experience. Webcast You can watch a Webcast of this AATS meeting presentation by going to: https://www.aats.org/resources/long-term-outcomes-after-bilat-12210. Conflict of Interest Statement The authors reported no conflicts of interest. The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest. Appendix E1 Table E1Postoperative outcomes in the original cohort of patients before propensity score matchingVariableRIMA, n (%)RA, n (%)P valueComposite outcome123 (5.8)55 (7.3).149Operative mortality49 (2.3)30 (4.0).020DSWI16 (0.8)4 (0.5).719Post-AMI53 (2.5)25 (3.3).271Stroke17 (0.8)4 (0.5).630Inotropes for >24 h831 (39.0)268 (35.7).126Prolonged MV169 (22.5)616 (28.9)<.001Respiratory infection212 (9.9)49 (6.5).006POAF299 (14.0)103 (13.7).891RIMA, Right internal mammary artery; RA, radial artery; DSWI, deep sternal wound infection; AMI, acute myocardial infarction; MV, mechanical ventilation; POAF, postoperative atrial fibrillation. Postoperative outcomes in the original cohort of patients before propensity score matching RIMA, Right internal mammary artery; RA, radial artery; DSWI, deep sternal wound infection; AMI, acute myocardial infarction; MV, mechanical ventilation; POAF, postoperative atrial fibrillation. Table E2Operative and postoperative outcomes in patients who underwent multiarterial revascularization modeled with year of surgeryVariableAll patients, n (%)RIMA, n (%)RA, n (%)P valueaOR (95% CI)P valueComposite outcome270 (6.4)117 (5.7)153 (7.2).0440.70 (0.53-0.91).009Operative mortality137 (3.3)46 (2.2)91 (4.3)<.0010.38 (0.26-0.55)<.001DSWI23 (0.6)16 (0.8)7 (0.3).0512.24 (0.89-6.26).100Post-AMI98 (2.3)50 (2.4)48 (2.3).7261.50 (0.95-2.37).082Stroke29 (0.7)17 (0.8)12 (0.6).3121.51 (0.69-3.41).452Inotropes for >24 h1554 (37.0)804 (38.8)750 (35.2).0141.43 (1.25-1.65)<.001Prolonged MV1076 (25.6)596 (28.8)480 (22.5)<.0011.41 (1.23-1.62)<.001Respiratory infection358 (8.5)209 (10.1)149 (7.00)<.0011.34 (1.06-1.70).014POAF507 (12.1)288 (13.9)219 (10.3)<.0011.23 (1.01-1.51).044Multivariate analysis was performed using the following covariates: age, diabetes, previous AMI, left ventricular ejection fraction, sex, and complete revascularization. RIMA, Right internal mammary artery; RA, radial artery; aOR, adjusted odds ratio; CI, confidence interval; DSWI, deep sternal wound infection; AMI, acute myocardial infarction; MV, mechanical ventilation; POAF, postoperative atrial fibrillation. Operative and postoperative outcomes in patients who underwent multiarterial revascularization modeled with year of surgery Multivariate analysis was performed using the following covariates: age, diabetes, previous AMI, left ventricular ejection fraction, sex, and complete revascularization. RIMA, Right internal mammary artery; RA, radial artery; aOR, adjusted odds ratio; CI, confidence interval; DSWI, deep sternal wound infection; AMI, acute myocardial infarction; MV, mechanical ventilation; POAF, postoperative atrial fibrillation. References