Diabetic patients undergoing coronary artery bypass grafting (CABG) are at increased risk of adverse outcomes, yet the role of multiple arterial grafting (MAG) remains underutilized and understudied in this population. The objective of this study was to evaluate short-term safety and long-term survival outcomes of MAG compared with single arterial grafting (SAG) in diabetic patients, using a propensity-matched design. This retrospective single center study included 3,164 diabetic patients who underwent isolated CABG between January 1996 and September 2023, stratified by graft strategy: SAG (n = 2,499) and MAG (n = 665). Propensity score matching was applied to balance baseline differences, yielding 662 matched pairs. In the unmatched cohort, MAG was associated with a significantly higher number of grafts and a greater rate of deep sternal wound infection (DSWI) (3.0% vs 1.7%; p = 0.028), while 30-day mortality was comparable (2.7% vs 2.8%; p = 0.939). In the matched cohort, short-term outcomes did not differ significantly between groups (p >0.3 for all). Kaplan–Meier survival estimates showed comparable survival (p = 0.954) at 1 year (92% vs 91%), 5 years (85% vs 83%), 10 years (73% vs 73%), 15 years (57% vs 62%), and 20 years (44% vs 41%). Multivariate Cox regression identified age, female sex, New York Heart Association (NYHA) class, prior cardiac surgery, reduced ventricular function, and disease extent as independent predictors of mortality. In conclusion, these findings suggest that MAG can be safely offered to diabetic patients, though its long-term survival benefit remains uncertain.
Patients with diabetes mellitus represent a particularly high-risk population undergoing coronary artery bypass grafting (CABG), due to diffuse coronary artery disease, endothelial dysfunction, and elevated rates of postoperative morbidity and mortality. Despite advances in surgical techniques and perioperative care, diabetic patients continue to experience inferior long-term outcomes compared to nondiabetic cohorts. As the global burden of diabetes increases, defining optimal revascularization strategies for this group remains a critical priority.
Within the CABG landscape, conduit selection remains a subject of ongoing debate. While single arterial grafting (SAG)—typically using the left internal thoracic artery (LITA) supplemented by saphenous vein grafts—has long been the standard approach, there is accumulating evidence supporting multiple arterial grafting (MAG) for improved long-term patency and survival. ,, However, randomized trials such as ART and post hoc analyses have yielded equivocal results, particularly in diabetic subgroup where technical concerns and sternal wound complications may limit MAG’s adoption. , The anticipated results of the ROMA trial are expected to address unresolved uncertainties, but until then, practice variation persists.
The primary objective of this study was to compare long-term survival between SAG and MAG in diabetic patients undergoing isolated CABG. Secondary objectives included evaluating short-term outcomes such as reoperation for bleeding, tracheostomy, stroke, deep sternal wound infection (DSWI), renal replacement therapy (RRT), and 30-day mortality.
Patients and Methods
Data source
This was a retrospective analysis of prospectively gathered data from a dedicated institutional cardiac surgery registry (Patient analysis and tracking system [PATS]; Dendrite Clinical Systems, Oxford, UK). The registry captures a broad spectrum of clinical variables spanning the preoperative, intraoperative, and postoperative periods, as well as in-hospital complications and long-term survival. The database undergoes regular external validation and contributes annually to the National Adult Cardiac Surgery Audit under the National Institute for Cardiovascular Outcomes Research. Mortality data were cross-verified via internal records and the National Health Service Spine system. Completeness of follow-up data was confirmed for the entire study cohort. The study protocol was approved by the institutional audit board, and individual consent was waived in accordance with the retrospective design and the principles of the Declaration of Helsinki.
Study cohort
The study included all diabetic patients undergoing isolated CABG with SAG or MAG between January 1996 and September 2023 at a single high-volume center. A total of 3,164 patients were identified. Patients were stratified into 2 cohorts according to grafting strategy at the time of surgery: SAG (n = 2,499) and MAG (n = 665). Case selection and surgical planning were determined by a weekly multidisciplinary team involving cardiac surgeons, interventional cardiologists, and cardiac imaging specialists. Operative urgency was assessed and managed via a structured waiting list protocol.
Operative strategy
CABG was performed through a midline sternotomy in all patients. Cardiopulmonary bypass was employed at the discretion of the surgical team, and conduit selection was guided by anatomical considerations and individual surgeon preference. SAG was defined as the use of a single internal thoracic artery graft supplemented by venous conduits. MAG referred to the use of 2 or more arterial conduits, most commonly including bilateral internal thoracic arteries (BITA) and/or radial arteries.
Postoperative care
All patients received standardized postoperative care in a dedicated cardiac intensive care unit. Protocols for ventilation, fluid management, and inotropic support were uniformly applied. Postoperative treatment included antiplatelet therapy, statins, beta-blockers, and angiotensin converting enzyme inhibitors/angiotensin receptor blockers, as clinically indicated. Diabetic care—including insulin or oral hypoglycemic agents—was administered in accordance with prevailing guidelines throughout the study timeline, reflecting updates from local and international standards as they evolved. The use of antiplatelet regimens postoperatively, including dual antiplatelet therapy, was not protocolized and reflected individual surgeon preference and patient-specific considerations.
Statistical analysis
Continuous variables were expressed either as mean ± standard deviation for normally distributed data or as median with interquartile ranges for nonparametric distributions. Categorical variables were reported as absolute counts and percentages. Distribution normality was evaluated using the Lilliefors modification of the Kolmogorov–Smirnov test. Comparisons between groups were made using Student’s t -test or Mann–Whitney U-test for continuous data and Pearson χ² or Fisher’s exact test for categorical variables, as appropriate. To evaluate temporal trends in short-term outcomes, patients were stratified into 3 operative eras: 1996 to 2005, 2006 to 2015, and 2016 to 2023. Within each era, in-hospital adverse events and 30-day mortality were compared between SAG and MAG cohorts.
Survival analysis was performed using the Kaplan–Meier method to estimate cumulative survival probabilities, with log-rank testing applied to compare survival curves between grafting strategies and predefined subgroups. Time-to-event data were reported with 95% confidence intervals at 5-year intervals up to 25 years. Cox proportional hazards regression models were used to identify independent predictors of long-term mortality. Variables with p-values <0.05 in univariable Cox regression were considered for entry into the multivariable model using a stepwise selection approach. This stricter threshold prioritized statistical significance, potentially reducing the number of candidate covariates. Final multivariable models retained predictors that remained statistically significant after adjustment, ensuring robustness while optimizing model fit.
To assess whether left ventricular function modified the association between grafting strategy and survival, patients were stratified by ejection fraction into 2 groups: normal (>50%) and impaired (≤50%). Kaplan–Meier survival curves were generated for each subgroup in both unmatched and propensity-matched cohorts, with survival rates tabulated at 1, 5, 10, 15, 20, and 25 years.
Additionally, temporal trends in survival were explored by dividing the cohort into 3 operative eras: 1996 to 2005, 2006 to 2015, and 2016 to 2023. Within each era, survival outcomes were compared between MAG and SAG in both unmatched and matched populations, providing insight into the evolution of grafting efficacy over time.
Propensity score matching (PSM) was undertaken to address potential baseline imbalances between the SAG and MAG groups. A logistic regression model incorporating all preoperative variables was used to generate propensity scores. A 1:1 nearest-neighbor matching algorithm without re placement was applied using a caliper width of 0.20 standard deviations of the logit of the propensity score. This caliper threshold was chosen to optimize balance while maintaining statistical power and cohort size. Balance between matched groups was assessed through standardized mean differences.
All statistical analyses were performed using SPSS version 29.0.2.0 (IBM Corp., Armonk, NY, USA). Data curation and preliminary calculations were conducted using Microsoft Excel (Microsoft Corp., Redmond, WA, USA).
Results
Demographic data
A total of 3,164 diabetic patients underwent isolated CABG during the study period, comprising 2,499 patients receiving SAG and 665 receiving MAG ( Table 1 ). In the unmatched cohort, MAG patients were younger (60.8 ± 9.3 years vs 66.2 ± 9.0 years; p <0.001), more likely to be male (84.1% vs 76.8%; p <0.001), and had a greater burden of triple-vessel coronary disease (89.9% vs 77.8%; p <0.001). Renal impairment and prior cardiac surgery were more prevalent in the MAG group (p = 0.002 and p = 0.003, respectively).
Table 1
Preoperative demographics of single arterial grafting and multiple arterial grafting cohorts
| Unmatched (n = 3,164) | Matched (n = 1,324) | |||||
|---|---|---|---|---|---|---|
| Variable | SAG (n = 2,499) | MAG (n = 665) | p-value | SAG (n = 662) | MAG (n = 662) | p-value |
| Age | 66.18 ± 9.00 | 60.75 ± 9.26 | 0.738 | 61.51 ± 9.0 | 60.81 ± 9.2 | 0.969 |
| Female gender | 579 (23.2%) | 106 (15.9%) | < 0.001 | 99 (15.0%) | 106 (16.0%) | 0.595 |
| BMI | 29.17 ± 11.26 | 28.92 ± 10.41 | 0.392 | 29.01 ± 11.03 | 28.97 ± 10.12 | 0.642 |
| NYHA | 0.018 | 0.409 | ||||
| I | 479 (19.2%) | 149 (22.4%) | 146 (22.1%) | 148 (22.4%) | ||
| II | 1,141 (45.7%) | 324 (48.7%) | 297 (44.9%) | 322(48.6%) | ||
| III | 708 (28.3%) | 154 (23.2%) | 173 (26.1%) | 154 (23.3%) | ||
| IV | 171 (6.8%) | 38 (5.7%) | 46 (6.9%) | 38 (5.7%) | ||
| Previous cardiac surgery | 17 (0.7%) | 13 (2.0%) | 0.003 | 12 (1.8%) | 12 (1.8%) | 1.00 |
| Hypercholesterolemia | 1,955 (78.2%) | 526 (79.1%) | 0.629 | 518 (78.2%) | 523 (79$) | 0.737 |
| Hypertension | 1,822 (72.9%) | 459 (69.0%) | 0.047 | 464 (70.1%) | 458 (69.2%) | 0.720 |
| Smoking | 0.636 | 0.730 | ||||
| Never | 1,015 (40.6%) | 271(40.8%) | 257 (38.8%) | 270 (40.8%) | ||
| Ex-smoker | 245 (9.8%) | 73 (11.0%) | 333 (50.3%) | 319 (48.2%) | ||
| Current | 1,239 (49.6%) | 321 (48.3%) | 72 (10.9%) | 73 (11.0%) | ||
| Renal impairment | 133 (5.3%) | 16 (2.4%) | 0.002 | 21 (3.2%) | 16 (2.4%) | 0.404 |
| Dialysis | 4 (0.2%) | 1 (0.2%) | 0.955 | 0 (0.0%) | 1 (0.2%) | 0.317 |
| COPD/Asthma | 286 (11.4%) | 64 (9.6%) | 0.183 | 61 (9.2%) | 64 (9.7%) | 0.778 |
| Cerebrovascular disease | 162 (6.5%) | 48 (7.2%) | 0.586 | 41 (6.2%) | 47 (7.1%) | 0.508 |
| Peripheral vascular disease | 305 (12.2%) | 71 (10.7%) | 0.498 | 68 (10.3%) | 70 (10.6%) | 0.857 |
| AF | 70 (2.8%) | 16 (2.4%) | 0.279 | 16 (2.4%) | 16 (2.4%) | 1.00 |
| Extent of CAD | <0.001 | 0.749 | ||||
| 1 | 73 (2.9%) | 10 (1.5%) | 10 (1.5%) | 10 (1.5%) | ||
| 2 | 482 (19.3%) | 57 (8.6%) | 65 (9.8%) | 57 (8.6%) | ||
| 3 | 1,944 (77.8%) | 598 (89.9%) | 587 (88.7%) | 595 (89.9$) | ||
| LMS disease | 585 (23.4%) | 150 (22.6%) | 0.543 | 150 (22.7%) | 149 (22.5%) | 0.948 |
| LVEF | 0.066 | 0.365 | ||||
| Good (> 50%) | 1,754 (70.2%) | 497 (74.7%) | 473 (63.4%) | 494 (70.6%) | ||
| Fair (30% to 50%) | 591 (23.6%) | 131 (19.7%) | 152 (23.0%) | 131 (19.8%) | ||
| Poor (< 30%) | 154 (6.2%) | 37 (5.6%) | 37 (5.6%) | 37(5.6%) | ||
| Logistic EuroSCORE | 3.9 ± 5.4 | 3.7 ± 4.9 | 0.810 | 3.89 ± 5.9 | 3.70 ± 4.9 | 0.406 |
| Follow-up (years) | 9.3 ± 6.7 | 10.8 ± 7.3 | < 0.001 | 10.8 ± 7.4 | 10.8 ± 7.3 | 0.986 |
AF = atrial fibrillation; BMI = body mass index; CAD = coronary artery disease; COPD = chronic obstructive pulmonary disease; LMS = left main stem; LVEF = left ventricular ejection fraction; MAG = multiple arterial grafting; NYHA = New York Heart Association;, SAG = single arterial grafting.
Following propensity score matching, 662 well-balanced patient pairs were generated. In the matched cohort, baseline characteristics were well balanced across all covariates including age, sex, New York Heart Association (NYHA) class, comorbidities, extent of coronary artery disease, and left ventricular function ( Table 1 ). No statistically significant differences remained postmatching.
Intraoperative data
MAG was associated with a higher number of grafts in the unmatched population (3.28 ± 0.56 vs 2.83 ± 0.60; p <0.001), though this difference was attenuated after matching ( Table 2 ). Off-pump CABG (OPCAB) was utilized in approximately half of all procedures, with equal distribution across both groups.
Table 2
Intraoperative data
| Unmatched (n = 3,164) | Matched (n = 1,324) | |||||
|---|---|---|---|---|---|---|
| Variable | SAG (n = 2,499) | MAG (n = 665) | p-value | SAG (n = 662) | MAG (n = 662) | p-value |
| OPCAB | 1,277 (48.9%) | 326 (49.0%) | 0.341 | 324 (48.9%) | 324 (48.9%) | 1.00 |
| Number of grafts | 2.83 ± 0.6 | 3.28 ± 0.56 | <0.001 | 3.31 ± 0.6 | 3.28 ±.6 | 0.387 |
MAG = multiple arterial grafting, OPCAB = off-pump coronary artery bypass grafting; SAG = single arterial grafting.
Short-term outcomes
Short-term postoperative outcomes—including reoperation for bleeding, tracheostomy, transient ischemic attack/stroke, deep sternal wound infection (DSWI), renal replacement therapy (RRT), and 30-day mortality—were comparable between MAG and SAG after matching ( Table 3 ). In the unmatched cohort, MAG was associated with a slightly higher DSWI rate (3.0% vs 1.7%; p = 0.028), but no other significant differences were observed. In the matched cohort, 30-day mortality was identical between groups (2.7% vs 2.7%; p = 1.00).
Table 3
In hospital outcomes and short-term mortality
| Unmatched (n = 3,164) | Matched (n = 1,324) | |||||
|---|---|---|---|---|---|---|
| Variable | SAG (n = 2,499) | MAG (n = 665) | p-value | SAG (n = 662) | MAG (n = 662) | p-value |
| Reoperation for bleeding | 81 (3.2%) | 26 (3.9%) | 0.397 | 20 (3.0%) | 26 (3.9%) | 0.368 |
| Tracheostomy | 47 (1.9%) | 3 (0.5%) | 0.009 | 3 (0.5%) | 3 (0.5%) | 1.00 |
| TIA/CVA | 48 (1.9%) | 15 (2.3%) | 0.583 | 13 (2.0%) | 14 (2.1%) | 0.846 |
| DSWI | 42 (1.7%) | 20 (3%) | 0.028 | 16 (2.4%) | 20 (3.0%) | 0.499 |
| RRT | 107 (4.3%) | 24 (3.6%) | 0.439 | 27 (4.1%) | 24 (3.6%) | 0.668 |
| Death at 30 days | 69 (2.8%) | 18 (2.7%) | 0.939 | 18 (2.7%) | 18 (2.7%) | 1.00 |
CVA = cerebrovascular accident; DSWI = deep sternal wound infection; MAG = multiple arterial grafting; RRT = renal replacement therapy; SAG = single arterial grafting; TIA = transient ischemic attack.
To further contextualize the safety profile of MAG over time, the decade-wise comparison of in-hospital outcomes and 30-day mortality across 3 eras: 1996 to 2005, 2006 to 2015, and 2016 to 2023 ( Table 4 ) demonstrated comparable short-term outcomes for SAG and MAG with no statistically significant differences. Notably, the overall incidence of adverse events declined over time in both groups, reflecting improvements in perioperative care and institutional protocols.
Table 4
In hospital outcomes and short-term mortality in each decade
| (1996 to 2005) Outcome | SAG (n = 688) | MAG (n = 304) | p-value |
|---|---|---|---|
| Reoperation | 34 (4.9%) | 16 (5.3%) | 0.345 |
| Tracheostomy | 12 (1.7%) | 1 (0.3%) | 0.407 |
| TIA/CVA | 18 (2.6%) | 11 (3.6%) | 0.498 |
| DSWI | 29 (4.2%) | 16 (5.3%) | 0.432 |
| RRT | 46 (6.7%) | 17 (5.6%) | 0.633 |
| Death at 30 days | 22 (3.2%) | 13 (4.3%) | 0.526 |
| (2006 to 2015) | |||
| Outcome | SAG (n = 914) | MAG (n = 140) | p-value |
| Reoperation | 32 (3.5%) | 4 (2.9%) | 0.616 |
| Tracheostomy | 27 (3.0%) | 1 (0.7%) | 0.305 |
| TIA/CVA | 10 (1.1%) | 2 (1.4%) | 0.214 |
| DSWI | 7 (0.8%) | 0 (0%) | 0.519 |
| RRT | 42 (4.6%) | 4 (2.9%) | 0.599 |
| Death at 30 days | 32 (3.5%) | 2 (1.4%) | 0.509 |
| (2016 to 2023) | |||
| Outcome | SAG (n = 897) | MAG (n = 221) | p-value |
| Reoperation | 22 (2.5%) | 5 (2.3%) | 0.695 |
| Tracheostomy | 8 (0.9%) | 1 (0.5%) | 0.645 |
| TIA/CVA | 20 (2.2%) | 2 (0.9%) | 0.406 |
| DSWI | 6 (0.7%) | 4 (1.8%) | 0.334 |
| RRT | 19 (2.1%) | 3 (1.4%) | 0.620 |
| Death at 30 days | 15 (1.7%) | 3 (1.4%) | 0.280 |
CVA = cerebrovascular accident; DSWI = deep sternal wound infection; MAG = multiple arterial grafting; RRT = renal replacement therapy; SAG = single arterial grafting; TIA = transient ischemic attack.
Survival outcomes
Mean follow-up duration was significantly longer in the MAG cohort compared to SAG (10.8 ± 7.3 years vs 9.3 ± 6.7 years; p <0.001), allowing for extended longitudinal outcome assessment in patients receiving MAG. Kaplan–Meier survival analysis in the unmatched cohort revealed no statistically significant difference between patients receiving MAG and those receiving SAG, with log-rank p = 0.373 ( Figure 1 ). Survival rates for MAG and SAG were comparable at each time interval: 1 year (92% vs 89%), 5 years (85% vs 81%), 10 years (73% vs 71%), 15 years (58% vs 62%), and 20 years (45% for both groups), as detailed in Table 5 . Although numerical differences existed early on, these were not statistically significant and converged over time, indicating similar long-term survival patterns between grafting strategies in the unmatched population.
