Evolution of the exercise electrocardiogram

Abstract

T he exercise electrocardiogram, also known as the exercise tolerance test (ETT), remains the most widely accessible and widely used technique for the investigation of suspected coronary artery disease and for the assessment of its severity. It has been estimated that 6 to 8 million treadmill tests are performed each year in the United States, but clinical usefulness of the simple ETT is limited by poor sensitivity of standard test criteria for the detection of coronary disease, a high proportion of equivocal test responses, imperfect specificity for coronary obstruction in patients with other forms of heart disease, poor predictive value of positive tests in populations with a low likelihood of disease, and imperfect identification of prognostically important coronary obstruction. As a consequence of these limitations, the standard ETT has served most faithfully over the past several decades as the method against which nearly all newer diagnostic imaging techniques can be favorably compared. Clinical confidence in the usefulness of the ETT has been undermined further by the inevitable arithmetic of Bayesian analysis that has focused more attention on the derivative problem of poor positive predictive value than on the primary problem of poor sensitivity, even though it is evident that similar predictive limitations affect all imperfect tests. Moreover, inferences regarding poor specificity of the ETT have been extrapolated unfairly from studies of catheterized patients, often with cardiomyopathy or valvular heart disease but no coronary obstruction, to general populations. Indeed, the poor sensitivity and specificity characteristics of standard ETT criteria derived from a half century of study in increasingly complex, selected populations have made it overly easy to challenge the cost effectiveness and even the routine clinical value of the simple ETT. We believe that these seemingly immutable negative aspects of exercise electrocardiography require fundamental reconsideration, however, because the ETT is evolutionary and not fixed in mid-century application. It is time to incorporate improvements in treadmill testing into clinical practice, clinical comparisons, and also into our perceptions of the clinical value of the test itself. We propose that the sensitivity and specificity of the ETT can be usefully increased in clinically important populations by appropriate application of newer methods of