Prediction of postoperative myocardial infarction after suprainguinal bypass using the Vascular Quality Initiative Cardiac Risk Index

Abstract

BackgroundThe Vascular Quality Initiative (VQI) Cardiac Risk Index (CRI) was developed to estimate the risk of postoperative myocardial infarction (POMI) for noncardiac vascular procedures. Whereas suprainguinal bypass carried the second highest odds of POMI, the performance of the all-procedure model declined when it was applied to the suprainguinal subset. We sought to improve the VQI CRI for application in this high-risk group undergoing open revascularization for aortoiliac occlusive disease. MethodsThe VQI Suprainguinal Bypass Registry was queried for elective procedures performed between January 2010 and March 2017. Logistic regression was used to create a model for estimating the risk of in-hospital POMI with preoperative variables including demographics, comorbidities, and planned inflow source. After adjustment for overfitting, internal validation was performed using both bootstrapping and 10-fold cross-validation methods. ResultsAmong 8157 procedures, the incidence of POMI was 3.2% (n = 258). After bootstrapping variable selection, age, graft inflow, preoperative stress test, American Society of Anesthesiologists class, indication for procedure, and coronary artery disease were chosen for inclusion as predictors in the final risk model. The final model demonstrated good discrimination (area under the curve = 0.725). On internal validation, the model discriminated well (area under the curve = 0.713), with good calibration (plot intercept = 0.0006 and slope = 1.001). Using this model, POMI risk estimates ranged from 0.6% to 30.4%. ConclusionsWhereas the incidence of POMI among all suprainguinal bypasses was 3%, model-based estimates ranged 50-fold, from 0.6% to 30.4%. This underscores the heterogeneity of this cohort and the need for patient-specific risk estimation. Although some of the strongest predictors were nonmodifiable (eg, age), the model provided specific estimates according to graft inflow and stress testing. This supraspecific VQI CRI module risk predictor may enhance preoperative counseling by influencing the decision to pursue preoperative stress testing and ultimately the type of revascularization strategy chosen.