Abstract
OBJECTIVES: To develop a model to predict risk of in-hospital bleeding following endovascular peripheral vascular intervention. BACKGROUND: Peri-procedural bleeding is a common, potentially preventable complication of catheter-based peripheral vascular procedures and is associated with increased mortality. We used the National Cardiovascular Data Registry (NCDR) Peripheral Vascular Interventions (PVI) Registry to develop a novel risk-prediction model to identify patients who may derive the greatest benefit from application of strategies to prevent bleeding. METHODS: We examined all patients undergoing lower extremity PVI at 76 NCDR PVI hospitals from 2014 to 2017. Patients with acute limb ischemia (n = 1600) were excluded. Major bleeding was defined as overt bleeding with a hemoglobin (Hb) drop of ≥ 3 g/dl, any Hb decline of ≥ 4 g/dl, or a blood transfusion in patients with pre-procedure Hb ≥ 8 g/dl. Hierarchical multivariable logistic regression was used to develop a risk model to predict major bleeding. Model validation was performed using 1000 bootstrapped replicates of the population after sampling with replacement. RESULTS: Among 25,382 eligible patients, 1017 (4.0%) developed major bleeding. Predictors of bleeding included age, female sex, critical limb ischemia, non-femoral access, prior heart failure, and pre-procedure hemoglobin. The model demonstrated good discrimination (optimism corrected c-statistic = 0.67), calibration (corrected slope = 0.98, intercept of -0.04) and range of predicted risk (1%-18%). CONCLUSIONS: Post-procedural PVI bleeding risk can be predicted based upon pre- and peri-procedural patient characteristics. Further studies are needed to determine whether this model can be utilized to improve procedural safety through developing and targeting bleeding avoidance strategies.