Development and internal validation of a prognostic model for mortality of patients with abdominal aortic aneurysms treated with endovascular aneurysm repair

© 2020 Hogrefe AG. Summary: Background: Morbidity and mortality associated with elective endovascular aneurysm repair (EVAR) for abdominal aortic aneurysms (AAA) must be balanced against the impending risk of aneurysm rupture and the estimated remaining lifetime. The aim of this study is to develop and validate a prognostic model for mortality of patients with AAA treated with EVAR. Methods: This retrospective observational study included 251 consecutive patients treated with EVAR for asymptomatic AAA between January 2001 and December 2012 at the University Hospital in Bern, Switzerland. Pre-selection of variables was based on a literature review; least absolute shrinkage and selection operator technique was used for the final variable selection. A Firth's bias reduced Cox proportional hazard model was developed and validated using 10,000 bootstrap samples to predict survival after EVAR. Results: The median follow-up time was 5.3 years (range 0.1 to 15.9). At the study closing date 95% of follow-up information was available. The mortality rates were 31.9% at 5 years and 50.5% at the study closing date, respectively. Identified predictors for overall mortality after EVAR were age, hazard ratio (HR) = 2.24 per 10-year increase (95% CI 1.64 to 3.09), the presence of chronic obstructive pulmonary disease (COPD), HR = 2.22 (95% CI 1.48 to 3.31), and lower estimated glomerular filtration rate, HR = 1.24 per 10 ml/min/1.73 m2 decrease (95% CI 1.12 to 1.39). The model showed good discrimination ability, Harrell's C = 0.722 (95% CI 0.667 to 0.778) and was very robust in the bootstrap in-sample validation Harrell's C = 0.726 (95% CI 0.662 to 0.788). Conclusion: Higher age, the presence of COPD and impaired kidney function are independent predictors for impaired survival after EVAR. The expected remaining lifetime should be considered in patients with AAA. This prognostic model can help improving patient care; however, external validation is needed prior to clinical implementation.