External validation of the Endovascular aneurysm repair Risk Assessment model in predicting survival, reinterventions, and endoleaks after endovascular aneurysm repair

Objective The Endovascular aneurysm repair Risk Assessment (ERA) model predicts survival (early death, 3-year survival, and 5-year survival), reinterventions, and endoleaks after elective endovascular aneurysm repair. We externally validated the ERA model in our cohort of patients. Methods This was a retrospective validation study of 433 consecutive patients with an asymptomatic abdominal aortic aneurysm treated with endovascular aneurysm repair in three hospitals (Amsterdam, The Netherlands) between 1997 and 2010. The area under the receiver operating characteristic curve was used as measure of accuracy (>0.70 was considered as sufficiently accurate). Results The early death rate was 1% (3 of 433; 95% confidence interval [CI], 0%-2%), the 5-year survival rate was 65% (95% CI, 61%-70%), the 5-year reintervention rate was 18% (95% CI, 14-78%), and the 5-year rate of type I, II, or III endoleak was 25% (95% CI, 20%-29%). The areas under the curve varied between 0.64 and 0.66 for predictions of survival and between 0.47 and 0.61 for reinterventions and endoleaks. Conclusions The predictions of survival, reinterventions, and endoleaks made by the ERA model were not sufficiently accurate to be used in our clinical practice.