LAUSR

regional anesthesia (LA/RA), and general anesthesia (GA). The interaction between anesthesia technique and shunting approach was evaluated. Multivariate logistic regression analysis was performed adjusting for patients demographics (age, gender, race, ethnicity), symptomatic status, comorbidities (diabetes, hypertension, coronary artery disease, congestive heart failure, chronic kidney disease, prior bypass, endovascular intervention or amputation, degree of stenosis, prior contralateral CEA/carotid artery stenting), restenosis, presence of anatomic high-risk factors, emergency status, type of CEA (conventional vs eversion), patching, and contralateral occlusion. Results: A total of 60,399 CEA cases were included: no shunting (48.4%), RS (47.5%), and SS (4.1%). Shunting was more likely performed under GA compared with RA/LA (55.8% vs 13.3%; P < .001), particularly RS (51.7% vs 8.5%; P < .001). SS was associated with 67% increased odds of in-hospital stroke/death compared with RS regardless of anesthetic technique (adjusted odds ratio, 1.67; 95% confidence interval, 1.23-2.28; P < .01). However, in both RS and SS, the incidence of stroke/death was higher when performed under RA/LA compared with GA (2.4% vs 1.1% and 4.9% vs 2.0%, respectively; P < .05; Fig 1). On multivariable adjustment, the interaction between anesthetic technique and shunting approach was significant (P < .05). Compared with GA, LA/RA was associated with double the risk of in-hospital stroke/death in patients who were RS (adjusted odds ratio, 2.1495% confidence interval, 1.15-3.99; P 1⁄4 .02) or SS (adjusted odds ratio, 2.3595% confidence interval, 1.17-4.73; P 1⁄4 .02; Fig 2). In the SS group, stroke/death was higher in awake patients compared with those monitored via electroencephalography and stump pressure (5.2% vs 2.2% and 2.1%, respectively; P 1⁄4 .03). However, there was no association between the neuromonitoring technique and the incidence of stroke/ death after adjustment. Conclusions: Shunting during CEA is more frequently performed under GA. Whether routine or selective, shunting is more safely performed under GA. The exact cause of this difference is unknown; however, surgeons experience, comfort and technical ability might play an important role.