Surgeon Frequency for Multiarterial Grafting and 19-Year Outcomes After Coronary Artery Bypass Grafting in Over One Million Medicare Beneficiaries

J. Michael. Schaffer¹, E. Shih², J. Squers³, J. Banwait⁴, S. Hale⁵, M. Mack¹, J. DiMaio^3
1Baylor Scott and White - The Heart Hospital, Plano, Plano, Texas ²Baylor Scott and White - The Heart Hospital, Plano, Richardson, Texas ³Baylor Scott and White - The Heart Hospital, Plano, Dallas, Texas ⁴Baylor Scott and White Research Institute, Plano, Texas ⁵Baylor Scott & White Research Institute, Plano, Texas

Purpose: Retrospective studies have shown a survival benefit for multi-arterial(MAG) over single-arterial grafting(SAG) during CABG. In contrast, a randomized controlled trial comparing SAG versus MAG found no difference in 10-year survival. Analyzing outcomes by surgeon preference for MAG may account for unmeasured confounding variables which potentially bias traditional retrospective analyses.

Methods: We identified Medicare beneficiaries undergoing first-time, isolated CABG with ≥1 arterial grafts from 2001-2019. Number of arterial and venous conduits were doubly adjudicated by ICD and CPT codes. Risk-adjustment was performed with overlap propensity score weighting. Kaplan-Meier analysis compared survival among beneficiaries undergoing SAG versus MAG and then among beneficiaries undergoing CABG by frequent SAG surgeons(≥95%ile SAG rate) versus frequent MAG surgeons(≥95%ile MAG rate). A non-parametric bootstrap procedure estimated median survival; 95% confidence intervals are shown in brackets.

Implementation of surgeon grafting preference as an instrumental variable assumes that practice pattern variations reflect surgeon treatment preferences which induce(from a patient’s perspective) randomization in treatment status. These analyses assume: (1)between-surgeon variation in treatment(certain surgeons prefer MAG, others SAG), (2)patient-to-surgeon assignment is unrelated to the surgeon’s treatment preference, and (3)surgeon’s use of MAG or SAG is independent of alternative treatments that affect outcomes(e.g. a surgeon’s preference for coronary endarterectomy during CABG).

Results: We identified 1,275,606 patients undergoing isolated non-redo CABG with ≥1 arterial graft and >2 total grafts, of which 1,131,741(88.7%) underwent SAG and 143,865(11.3%) underwent MAG. The annual prevalence of MAG decreased from 2001(11.9%) to 2015(7.9%) then increased to 10.1% by 2019. Among 4,164 surgeons performing ≥10 annual CABG, 383 frequently performed SAG (≥95% SAG rate, 66,149 total CABG performed, of which over 98% were SAG), and 236 frequently performed MAG (≥38% MAG rate, 66,218 total CABGs performed, of which 56% were MAG).

MAG recipients were younger, more likely male, lived in less deprived neighborhoods, had a lower prevalence of CHF, CKD, and COPD, more likely underwent off-pump CABG, and received more grafts compared to SAG recipients. Risk-adjustment achieved balance for measured demographics, preexisting comorbidities, and operative characteristics.

Risk-adjusted median survival was 11.48[11.40,11.55] years after MAG versus 11.07[11.03,11.09] years after SAG, a difference of 4.9[4.4,5.6] months(Figure-1). In contrast, risk-adjusted median survival was 10.20[10.09-10.30] years in beneficiaries undergoing CABG by frequent MAG surgeons versus 10.19[10.06-10.31] years in beneficiaries undergoing CABG by frequent SAG surgeons, a difference of 0.01[-1.6,2.1] months(Figure-2).

Repeated sensitivity analyses varying MAG utilization rates to define frequent MAG surgeons yielded similar results.

Conclusion: Medicare beneficiaries undergoing MAG had superior risk-adjusted survival compared to those undergoing SAG. To address potential bias underlying a surgeon’s decision to perform MAG over SAG, surgeon preference for MAG or SAG during CABG was exploited as an instrumental variable to better measure the treatment effect of conduit strategy. Beneficiaries had similar risk-adjusted survival whether they underwent CABG by frequent MAG or frequent SAG surgeons. We hypothesize that unmeasured confounding variables underlying a surgeon’s decision to perform MAG or SAG have substantial impact on survival, potentially explaining the discordance between retrospective and randomized data regarding long-term survival after MAG.

Identify the source of the funding for this research project: Data acquisition and effort of JKB was supported by a philanthropic gift of Satish and Yasmin Gupta to Baylor Scott & White–The Heart Hospital, Plano, TX