Outcome after open and endovascular repairs of abdominal aortic aneurysms in matched cohorts using propensity score modeling.

ABSTRACT

OBJECTIVE: The objective of this study was to compare outcomes after open repair (OR) vs endovascular aneurysm repair (EVAR) of infrarenal abdominal aortic aneurysms (AAAs). METHODS: Clinical data of consecutive patients treated for asymptomatic AAA between 2000 and 2011 were reviewed. Patients were stratified into low/normal-risk (comorbidity score ≤ 10) and high-risk (score > 10) categories. The primary end point was all-cause mortality; secondary end points were complications, reinterventions, conversions, and ruptures. Propensity score-based matching was performed to compare outcomes. RESULTS: There were 1534 patients, of whom 207 were women (13%); 641 (42%) were treated with OR and 893 (58%) with EVAR. After propensity score matching, we selected 558 pairs of OR and EVAR (mean age, 73 ± 7.6 years); 158 were women (14%). The 30-day mortality rate was 1.3% after OR and 0.9% after EVAR (P = .56). In multivariable analysis, only high risk was an independent predictor of early mortality (odds ratio, 4.65; 95% confidence interval [CI], 1.20-18; P = .03). The early complication rate was lower for EVAR (13%; odds ratio, 0.5; 95% CI, 0.4-0.8; P < .001) than for OR (24%). Median follow-up was 7.6 years (31 days-13.1 years). The cumulative 5-year survival rate was 72% after EVAR and 81% after OR (hazard ratio, 1.44; 95% CI, 1.19-1.73; P < .001). The 5-year survival was not significantly different in matched cohorts operated on after 2005 (77% vs 81%; P = .57). High risk, advanced age, cancer history, AAA size, and EVAR predicted all-cause mortality. Freedom from reintervention was 74% after EVAR and 88% after OR (hazard ratio, 2.60; 95% CI, 1.92-3.51; P < .001). Freedom from rupture was 99.2% after EVAR and 99.8% after OR (P = .04). In multivariable models, female gender was associated with complications; EVAR was associated with reinterventions (P < .05). CONCLUSIONS: In this retrospective propensity score-matched study, early mortality was similarly low after both EVAR and OR, significantly different from all except one large randomized controlled trial. EVAR had fewer early complications, but it was associated with late all-cause mortality and reinterventions and had a small but definite risk of late rupture. Significantly increased mortality at 5 years was no longer observed when operations were performed after 2005. High risk, advanced age, cancer history, and AAA size predicted late all-cause mortality. This study failed to confirm early or late survival benefit for EVAR vs OR. Improved surveillance, longer follow-up, and analysis of factors affecting late death in prospective studies are warranted.