Abstract 55: Reducing Utilization of Cardiac Stress Testing in the Outpatient Primary Care Setting: a Clinical Prediction Rule

Abstract

Background: Utilization of stress testing with imaging has significantly increased over the past decade. A simple clinical decision rule to guide stress test ordering for diagnosis of diagnose ischemia in the primary care setting could significantly reduce unnecessary utilization and health care costs. Methods: We retrospectively identified 495 outpatients without coronary artery disease (CAD) referred from primary care clinics who underwent stress echocardiography at a single US institution from 2007-2009. Predictors of echocardiographic ischemia were determined by logistical regression. From the predictor variables, a simple clinical decision rule was derived. The rule stratified individuals into 2 groups: subjects who should proceed to stress testing (ST) and subjects who do not require a stress test (NST). A receiver-operator characteristic curve was constructed to determine the rule’s accuracy. Event-free (interim stroke, myocardial infarction, or death) survival between the ST and NST arms was compared by log rank test. The rule was validated in a set of 299 charts from a different US institution. Results: In the derivation cohort, 23 tests (4.6%) demonstrated echocardiographic evidence of ischemia. Univariate predictors included male sex (odds ratio (OR) 2.3), family history of early CAD (OR 2.1), any chest pain (OR 9.1), fatigue or shortness of breath (OR 2.04), and age > 50 years (OR 4.4). All variables were retained for the clinical decision rule. The rule selected 183 (37%) of subjects for NST while maintaining 100% sensitivity for ischemia (c-statistic 0.71, P<0.001). 24 adverse clinical events occurred during follow-up (median: 2.6 years). There were no significant differences in adverse outcomes between ST and NST groups (P=0.35). In the validation cohort, 60% of subjects were select for NST with no differences in event-free survival between groups (P=0.30). Conclusion: Our simple decision rule could significantly reduce the number of stress tests ordered by primary care physicians without compromising event-free survival. Further refinement and validation of this score in both retrospective and prospective studies is warranted.