Abstract 5106: Sub-Clinical Electrocardiographic Abnormalities Predict Mortality among Primary Prevention Exercise Test Patients with a Qualitatively Normal Electrocardiogram
It is unknown if measurement of subtle electrocardiographic abnormalities in exercise stress test patients with apparently normal resting electrocardiograms improves estimation of prognosis. We performed an observational prospective study of 18,964 consecutive primary prevention patients who had a qualitatively normal electrocardiogram (ECG) and who underwent treadmill exercise stress testing for evaluation of suspected CAD. Eleven sub-clinical quantitative ECG measures related to heart rate, conduction, left ventricular mass, and repolarization were collected digitally. The primary outcome was all-cause mortality. As we are unaware of any external cohorts with this type of data we used out-of-bagging to effectively perform 100 external validations.
Results: During a median follow-up of 10.5 years 1,585 patients died. Faster ventricular rate, longer PR interval, greater ST segment deviation, longer QT interval, and left-most QRS axis were independently associated with mortality. A composite ECG score based on these variables was independently associated with mortality (75th vs. 25th percentile HR 1.36, 95% CI [1.25 to 1.49], P<.0001). Addition of this ECG score to Cox models of established risk factors, improved predictive discrimination (increase in c-index 0.04, 95% CI [0.01 to 0.07]) and reclassification of risk (relative integrated discrimination improvement 3.0%, P=.0000005). Sub-clinical ECG abnormalities in primary prevention patients referred for stress testing are strongly predictive of death. Use of these abnormalities for estimation of long-term prognosis should be considered.