Perioperative Risk Profiles and Volume-Outcome Relationships in Proximal Thoracic Aortic Surgery

Background. Using the national Society of Thoracic Surgeons Adult Cardiac Surgery Database data for thoracic aortic surgical procedures for aortic aneurysm, this study aimed to (1) characterize patients' risk profiles and outcomes, (2) evaluate center volume-outcome relationships across US centers, and (3) identify risk factors for operative mortality. Methods. Between 2011 and 2016, 53,559 operations for ascending aortic aneurysm performed across 1,045 centers in the United States were identified. Logistic regression related baseline characteristics and comorbidities to operative mortality. Ten-fold cross-validation was performed to estimate sensitivity and specificity across a range of the discrimination threshold. Centers were stratified into five strata by average annual case volume. Predicted probability of operative mortality was calculated from the model and was used to evaluate patients' risk profiles across the volume strata. Results. Operative mortality occurred in 3.2% of all cases and in 2.2% of elective cases. Only 24 (2.3%) centers performed >= 50 cases annually, whereas 609 (58.3%) centers performed fewer than five cases annually. Multiple logistic regression, of which the c-index was 0.80, revealed that compared with centers with >= 50 cases, centers with fewer than five cases had an increased risk of mortality (odds ratio, 2.50; 95% confidence interval, 2.08 to 3.01; p < 0.0001). The predicted probability of operative mortality was similar across the volume strata, but the observed mortality rate varied significantly, with lower volume yielding higher operative mortality. Conclusions. Proximal thoracic aortic surgical procedures for aortic aneurysms in the United States are associated with a low operative mortality rate of 2.2% for elective cases. Risk of operative death decreases significantly at an annual center volume of more than 20 to 25 cases per year. (C) 2018 by The Society of Thoracic Surgeons.