Comparing the efficacy of shunting approaches and cerebral monitoring during carotid endarterectomy using a national database

Objective: Carotid endarterectomy practice patterns, including the use of shunts and cerebral monitoring techniques, are typically surgeon-dependent and differ greatly on a national level. Prior literature evaluating these techniques is often underpowered for detecting variations in low-frequency outcomes. The purpose of this study was to evaluate current carotid endarterectomy practice patterns and to allow comparison across surgical approaches using a large national database. Methods: We divided carotid cases entered into the Vascular Quality Initiative database between October 2012 and April 2015 into routine shunting, selective shunting, and never shunting cohorts, excluding endarterectomies performed with concomitant procedures and those with incomplete information on the use of a shunt. The selective group was subdivided into cases with awake, electroencephalography, and stump pressure monitoring. We evaluated differences in practice patterns and compared rates of stroke, death, return to the operating room, reperfusion injury, and re-exploration after closure across these groups. Multivariate logistic regression models adjusting for risk factors were used to identify predictors of each outcome. Results: Between October 2012 and April 2015, there were a total of 28,457 endarterectomies included in our analysis, of which 14,128 involved routine shunting, 1740 involved never shunting, and 12,489 involved selective shunting. Of the selective cases, 6144 involved electroencephalography monitoring, 2310 involved stump pressure monitoring, and 2052 involved awake monitoring. Unadjusted rates of in-hospital death and stroke were 0.30% (95% confidence interval [CI], 0.21-0.39) and 0.78% (95% CI, 0.64-0.93) for routine shunting and 0.22% (95% CI, 0.14-0.31) and 0.91% (95% CI, 0.75-1.08) for selective shunting, respectively. The unadjusted rate of in-hospital death was lower in the awake monitoring group than in the routine shunting group (0.05% vs 0.30%; P =.037). After adjustment for patient risk factors, the multivariate models showed no difference in rates of any primary outcomes among the groups, although there was a shorter postoperative length of stay for the awake monitoring group compared with the routine shunting group (1.55 days vs 2.00 days, respectively; P <.01). Conclusions: Analysis of the Vascular Quality Initiative registry shows equivalent unadjusted rates of in-hospital death and stroke across different approaches to shunting and cerebral monitoring with the exception of the awake monitoring group, which has lower unadjusted mortality compared with the routine shunting group. In the risk-adjusted analysis, however, there are no differences across any of the groups. Given the clinical equivalence of approaches to shunting and cerebral monitoring, further work should evaluate the relative cost of these techniques.