Defining the threshold surgeon volume associated with improved patient outcomes for carotid endarterectomy

The outcomes for common vascular operations, such as carotid endarterectomy (CEA), are associated with surgeon volume. However, the number of operations associated with an improved stroke or death rate for CEA is not known. The objective of the current study was to define the annual surgeon volume of CEAs that is associated with a lower risk of stroke or death rate. The Nationwide Inpatient Sample was analyzed to identify patients undergoing CEA between 2003 and 2009. Annual surgeon volume was correlated with a composite end point of in-hospital stroke or death. Mixed linear regression analyses were conducted to determine if annual surgeon volume of CEAs is independent predictor of the composite outcome. Receiver operating characteristic curves were constructed from the regression models and used to calculate the Youden Index, which defined the optimal cutoff point of annual surgeon volume of CEAs in predicting in-hospital stroke and death. This cutoff point was further assessed using Chi square analyses to determine whether incremental increases in the annual volume of CEAs were associated with a lower in-hospital stroke or death rate. A total of 104,918 CEA cases with surgeon identifiers were included in the analysis. The crude in-hospital stroke or death rate for CEA was 1.26 %. As expected, the stroke or death rate after CEA was higher for symptomatic patients, compared to asymptomatic patients (6.46 % vs 0.72%; P < .0001). For symptomatic patients, the relationship between surgeon volume and the composite end point was not significant (P = .435). In contrast, there was a strong relationship between surgeon volume and outcomes for asymptomatic patients undergoing CEA with a stroke/death rate of 1.66%, 0.91%, and 0.65% for low-, moderate-, and high-volume surgeons (P < .0001). Multivariate analysis identified age, African-American race, Charlson Comorbidity Index, and surgeon volume as independent predictors of stroke/death after CEA for asymptomatic carotid stenosis. For asymptomatic patients, the optimal cutoff number of CEAs to predict stroke/death rate was 19.4 CEAs per year (sensitivity = 74.9%, specificity = 72.6%, Youden index = 0.475). Analyses of outcomes at different cutoff points of surgeon volume revealed that the rate of crude complications and the adjusted probability of stroke or death was higher with case numbers less than 20 CEAs per year and lower with case numbers of 20 CEA or higher per year. Cutoff points above 20 cases were year di