Ten-year Comparison of All-Cause Mortality after Endovascular or Open Repair of Abdominal Aortic Aneurysms: A Propensity Score Analysis

Abstract

Background This study aimed to compare the long-term survival after open (OS) or endovascular (EVAR) repair of abdominal aortic aneurysms (AAAs), exploring baseline factors that could affect long-term outcome. Methods We identified 774 patients (501 EVAR, 273 OS, during 1996-2004) with data on perioperative risk factors including 37 variables assessed with a standardized patient response instrument. Propensity score was used to adjust for baseline differences between the two cohorts. Matched cohorts survival analysis and Cox multivariate regression were performed. Results Median follow-up was 6.95 (interquartile range 4.46-9.27) years. EVAR patients were older [75.0 +/- 7.7 (SD) vs. 71.3 +/- 8.5 years, p < 0.001] and had a higher rate of previous myocardial infarction (39.3 vs. 25.3 %, p < 0.001), pulmonary disease (25.9 vs. 18.3 %, p = 0.020), and history of malignancy (5.0 vs. 1.8 %, p = 0.039). The 30-day mortality was comparable (1.4 % EVAR, 1.5 % OS). Although the unadjusted survival rate was lower (median survival: 7.4 years EVAR, 8.8 OS, p = 0.011) and early (within 4 years) hazard was higher after EVAR (p = 0.003), no difference in survival was observed after propensity score-matching (p = 0.688) or propensity score-adjusted Cox regression (hazard ratio 1.01, 95 % confidence interval 0.82-1.25, p = 0.911, EVAR vs. OS). There was a trend toward higher hazard later in both groups. A multivariate Cox regression identified age, pulmonary disease, stroke, dialysis, oral anticoagulation, cardiac enlargement, and smoking history as variables associated with poor survival. Lipid-lowering medication was found to be protective. Conclusions Over long-term follow-up, survivals after endovascular and open repair of AAA are similar. Baseline patient characteristics are correlated with survival, but whether attention to the modifiable risk factor can alter outcome remains to be defined