Combining clinical and thallium data optimizes preoperative assessment of cardiac risk before major vascular surgery

Study Objective: To determine whether clinical markers and preoperative dipyridamole-thallium imaging are both useful in predicting ischemic events after vascular surgery. Design: Retrospective, observational study. Setting: University medical center. Patients: Two hundred fifty-four consecutive patients referred to a nuclear cardiology laboratory before surgery. Forty-four patients had surgery cancelled or postponed after clinical evaluation and dipyridamole-thallium imaging. Surgery was not confirmed for ten. Two hundred patients receiving prompt vascular surgery were the study group. Measurements and Main Results: Thirty patients (15%) had early postoperative cardiac ischemic events, with cardiac death in 6 (3%) and nonfatal myocardial infarction in 9 (4.5%). Logistic regression identified five clinical predictors (Q waves, history of ventricular ectopic activity, diabetes, advanced age, angina) and two dipyridamole-thallium predictors of postoperative events. Of patients with none of the clinical variables (n = 64), only 2 (3.1%; 95% CI, 0% to 8%) had ischemic events with no cardiac deaths. Ten of twenty (50%; 95% CI, 29% to 71%) patients with three or more clinical markers and events. Eighteen of one hundred sixteen (15.5%; 95% CI, 7% to 21%) patients with either 1 or 2 clinical predictors had events. Within this group, 2 of 62 (3.2%; 95% CI, 0% to 8%) patients without thallium redistribution had events compared with 16 events in 54 patients (29.6%; 95% CI, 16% to 44%) with thallium redistribution. The multivariate model using both clinical and thallium variables showed significantly higher specificity at equivalent sensitivity levels than models using either clinical or thallium variables alone. Conclusions: Preoperative dipyridamole-thallium imaging appears most useful to stratify vascular patients determined to be at intermediate risk by clinical evaluation. For patients with one or two clinical predictors, thallium redistribution correlates with substantial change in probability of events. For nearly half the patients, however, thallium imaging may have been unnecessary because of very high or low cardiac risk predicted by clinical information alone.