Defining high risk in endovascular aneurysm repair

Background: Long-term survival benefit contrasted with rupture risk should determine which patients are suitable for abdominal aortic aneurysm (AAA) intervention. Our aim was to develop a model capable of predicting long-term survival based on preoperative characteristics. Methods: A prospective cohort study using Cox regression modeling. We aimed to associate preoperative characteristics with long-term mortality, and create a predictive nomogram, which was then externally validated on an independent cohort (697 patients) who underwent endovascular abdominal aortic aneurysm (AAA) repair. Results: We pooled the results of 412 patients undergoing endovascular repair of infrarenal and juxtarenal aneurysm who were high risk (average Glasgow aneurysm scores of 72.8 [SD 10.4]). Despite anatomic differences, there were no statistically significant differences in perioperative or long-term outcomes between infrarenal and juxtarenal aneurysms (log rank test, P=.5). Data from this group (64% infrarenal aneurysms and 36% juxtarenal aneurysms) were randomly and evenly split into a model development and test group. Independent predictors of mortality included in the model are age, aneurysm diameter, history of peripheral artery disease, chronic obstructive pulmonary disease (COPD), or congestive heart failure, requirement for supplemental home oxygen, and use of salicylates. Internal validation reveals good calibration and discriminative ability (c-statistic 0.68 [95% confidence interval 0.65-0.71]). External validation confirms good calibration. Conclusions: In the context of acceptable perioperative results, long-term mortality risk can be predicted in endovascular AAA repair and must be balanced against risk of rupture to determine which patients should be offered treatment. (J Vase Surg 2010;51:1088-95.)