With Limited RCTs of Arterial Grafting, Researchers Take a Different Tack
A new study comparing multiple and single arterial grafting highlights how physician preference can bias observational data.

LOS ANGELES, CA—To combat some of the limitations of observational analyses in cardiac surgery, factoring in instrumental variables like surgeon preference could potentially alleviate some confounding bias, according to the authors of a new study.
Their analysis sought to offer a fresh perspective by moving beyond the traditional as-treated methodology when comparing strategies for arterial grafting in CABG.
“I do feel strongly that we continue to underestimate the magnitude of the treatment effect associated with unmeasured confounding variables,” Justin Schaffer, MD (Baylor Scott & White The Heart Hospital, Plano, TX), who presented the data last week at the 2025 Society of Thoracic Surgeons (STS) Annual Meeting, told TCTMD. “The majority of studies in cardiac surgery are retrospective analyses that are subject to those biases.”
Instrumental variable analyses, by design, allow researchers to approximate randomization in a nonrandomized setting. “[It’s] a tool we can apply to observational data that mimics the randomizer function from a randomized control trial, thereby inducing a pseudo randomization of patients such that confounding variables are, in theory, equally distributed among treatment arms,” he explained.
The approach requires several assumptions—in this particular study, researchers considered three: that some CABG surgeons prefer multiple arterial grafting (MAG) while others prefer single arterial grafting (SAG), that patients are assigned to surgeons irrespective of this preference, and that surgeon preference is unrelated to alternative treatments such as coronary endarterectomy or off-pump surgery.
This statistical methodology “represents the cutting edge, or the best we can do to try to handle that,” Schaffer argued.
Surgeon Preference Affects Results
For their study, Schaffer and colleagues looked at data from more than 1 million Medicare beneficiaries who underwent isolated CABG with multiple grafts, including at least one arterial, between 2001 and 2019. Compared with the 1,131,741 patients treated with SAG, the 143,865 who received MAG tended to be younger, more often male, from areas with a lower area deprivation index, and had fewer comorbidities. MAG surgeries were more often elective and off-pump. On average, those in the MAG cohort received 2.3 arterial grafts out of 3.8 and patients treated with SAG received one arterial graft out of 3.41.
The researchers found a “nice and important progressive decline in the frequency of all venous grafting” over the study period, Schaffer said, adding that the trends observed with MAG and SAG are in line with what has been seen in previous STS database analyses.
In their traditional as-treated analysis, the researchers found a survival advantage with MAG over SAG over 12 years in a propensity-matched analysis of more than 100,000 pairs (HR 0.87; 95% CI 0.85-0.89). However, there was no longer a survival benefit when comparing procedures performed by surgeon “zealots” in the 95th percentile of MAG and SAG use, respectively (HR 0.96; 95% CI 0.82-1.12).
“These data are concordant with the ART trial, which found no difference in survival between single and bilateral IMAs,” Schaffer said.
Ultimately, the findings underscore the importance of surgeons needing to be “more skeptical of observational analyses due to the presence of unmeasured confounding variables and appreciate that unmeasured confounders can have a substantial treatment effect biasing the results of observational studies,” he concluded. “Our study noted substantial differences between as-treated and surgeon preference analyses, which could explain the observed differences between randomized data and observational analyses regarding the clinical question of MAG versus SAG.”
Next Steps
To TCTMD, Mario Gaudino, MD, PhD (Weill Cornell Medicine, New York, NY), said he did not agree with the push to perform more instrumental variable analyses on observational datasets, but rather continued to call for more randomized trials in cardiac surgery. “Those studies are actually not the right answer,” he said.
“I disagree that surgeon preference is a good instrumental variable,” Gaudino continued, citing that all cardiac surgeons can and do perform both MAG and SAG, “and they make the decision to do one or the other based on the patient characteristics. So, I think the concept of instrumental variable applied to surgeon preferences is flawed and wrong.”
Only randomized data, like those coming in 2027 from the ROMA trial that he’s leading, will give a clear answer regarding which arterial grafting strategy is best, according to Gaudino.
“We have been doing nonrandomized comparison of single versus multiple arterial grafting for five decades, and the randomized trial that we have done does not support the result of the nonrandomized data at least so far,” he said. “I don’t think we need the nonrandomized comparisons because surgeons always have a reason why they do one operation or the other. . . . You are just looking at different patients that have different outcomes, but the difference in outcome is not related to the intervention. And that is exactly the mistake that we have done for five decades, and I hope that we will stop doing it.”
Regardless of the type of analysis, panelist and incoming STS President Joseph F. Sabik III, MD (University Hospitals Cleveland Medical Center, OH), urged surgeons to move the field of coronary surgery forward through increasing the use of arterial grafting. “We can argue about statistical methodology or the patient populations, but arterial grafts don’t develop atherosclerosis like vein grafts, and vein grafts have been shown to be just as good as balloon angioplasty,” he stressed during the session.
Moreover, he said, “I think we should adopt a policy where we want to start with MAG and defend why we don’t use them.” Sabik encouraged surgeons to continue to “innovate so that we have a purpose” and “stop defending the saphenous vein graft because [of the notion]: ‘We don’t have studies that show it’s not as good.’”
Still, Schaffer argued that “we should be, at our baseline, data driven. Cardiologists have interventions that are much less substantial and it’s a lot easier for them to enroll their patients in randomized controlled trials,” he said. “Nevertheless, it is our duty to chase the data to try to prove which treatment modality is better than another and not just use our anecdotal experience to guide that decision.
“Our best option out there is a randomized controlled trial, and we should continue to chase that and base our decisions on those studies,” Schaffer stressed.
Yael L. Maxwell is Senior Medical Journalist for TCTMD and Section Editor of TCTMD's Fellows Forum. She served as the inaugural…
Read Full BioSources
Schaffer JM. Multi-arterial grafting (MAG) and survival after CABG: an instrumental variable analysis. Presented at: STS 2025. January 24, 2025. Los Angeles, CA.
Disclosures
- Schaffer reports no relevant conflicts of interest.
- Sabik reports serving on the advisory board for and as a speaker for Medtronic.
- Gaudino reports being the PI for ROMA.
Comments