Predicting Mortality of Ruptured Abdominal Aortic Aneurysms in the Era of Endovascular Repair

Abstract

Previous risk prediction models of mortality after ruptured abdominal aortic aneurysm (rAAA) repair have been limited by imprecision, complexity, or inclusion of variables not available in the preoperative setting. Most importantly, these prediction models have been derived and validated before the adoption of endovascular aneurysm repair (EVAR) as a treatment for rAAA. We sought to derive and validate a new risk-prediction tool using only easily obtainable preoperative variables in patients with rAAA who are being considered for repair in the endovascular era. We used the Vascular Study Group of New England (VSGNE) database to identify all patients who underwent repair of RAAA (2006-2015). Variables were entered into a multivariable logistic regression model to identify independent predictors of 30-day mortality. Linear regression was then used to develop an equation to predict risk of 30-day mortality. During the study period, 649 patients underwent repair of rAAA; of these, 247 (38.1%) underwent EVAR and 402 (61.9%) underwent an open repair. The overall mortality associated with rAAA was 30.7% (open, 33.4% and EVAR, 26.2%). On multivariate modeling, the primary determinants of 30-day mortality were advanced age (>76 vs. ≤76years, odds ratio [OR]=2.91 and CI: 2.0-4.24), elevated creatinine (>1.5mg/dL vs. ≤1.5mg/dL, OR=1.57 and CI: 1.05-2.34), and lowest systolic blood pressure (SBP) (BP<70mm Hg vs. ≥70mm Hg, OR=2.65 and CI: 1.79-3.92). The logistic regression model had an area under a c-statistic of 0.69. The corresponding linear model used to provide a point estimate of 30-day mortality (%) was % mortality = 14 + 22 * (age >76) + 9 * (creatinine >1.5) + 20 * (bp <70) Using this model, patients can be stratified into different groups, each with a specific estimated risk of 30-day mortality ranging from a low of 14% to a high of 65%. In the endovascular era where both open and endovascular treatment are offered for the treatment of rAAA three variables, easily obtained in an emergency setting, accurately predict 30-day mortality for patients operated on for rAAA. This simple risk prediction tool could be used as a point of care decision aid to help the clinician in counseling patients and their families on treatment of those presenting with rAAA.