Abstract 20153: Predicting the Need for Transfusions in Endovascular Aneurysm Repair (EVAR)

Purpose: The majority of patients undergoing EVAR do not require blood transfusion, yet blood type and screen (T&S) is routinely performed. Identifying patients in whom T&S can be avoided presents a substantial cost saving opportunity. Hypothesis: We hypothesized that intraoperative blood transfusions can be predicted preoperatively. Methods: Using the Vascular Study Group of New England database from 2003-2014, we performed a retrospective review of 4700 patients who underwent EVAR. The cohort was split randomly into a training (60%) and validation (40%) set. A backwards logistic regression analysis was performed to identify predictors of intraoperative blood transfusion in the training set. The model was then tested in the validation set to estimate the receiver operating curves (ROC) and goodness of fit. Results: Preoperative hemoglobin, urgency (elective, symptomatic, or ruptured), age, maximal anterior-posterior AAA diameter, female gender, and history of CHF (asymptomatic, mild, moderate, or severe) were all significant predictors of intraoperative blood transfusions. The c-statistic for our model was .82 in the training set and .84 in the validation set, and the Hosmer-Lemenshow goodness-of-fit statistic was 0.99. Conclusions: Intraoperative blood transfusions can be routinely predicted preoperatively. Avoidance of T&S in low risk populations provides a substantial cost-saving opportunity.