Population selection and performance of the exercise ECG for the identification of coronary artery disease

Abstract

To quantify the effect of population selection on the performance of exercise ECG criteria for the detection of coronary artery disease, the exercise ECGs of 212 clinically normal nonvolunteer subjects, 31 patients with no significant coronary disease at angiography, 153 patients with clinically stable angina, and 184 patients with catheterization-proved coronary disease were examined. Test specificity was examined separately in clinically normal subjects and in patients with angiographically normal coronary arteries, and test sensitivity was determined separately in patients with stable angina and those with catheterization-proved disease. Definition and selection of normal and abnormal study populations had marked effects on test performance. Standard ECG criteria, a simple ST depression magnitude partition of 150 μV. an St segment/heart rate (STHR) index partition of 1.60 μV/beat/min, and an STHR slope partition of 2.40 μV/beat/min, identified coronary disease with comparably high specificities (94% to 97%) in clinically normal subjects, but with significantly lower specificities (68% to 77%, p = 0.002 to 0.0001) in patients with angiographically normal coronary arteries. Although sensitivity was significantly lower in patients with stable angina than in patients with catheterization-proved coronary disease for standard criteria (54% vs 70%, p = 0.004) and for the STHR index (88% vs 95%, p = 0.04), there was no significant difference in the poor sensitivity of the simple ST depression magnitude criteria (51% vs 58%) or in the high sensitivity of the STHR slope (93% vs 96%) in these abnormal patient groups. However, the 51% to 70% test sensitivities of standard criteria and simple ST depression criteria were significantly lower than the 88% to 96% test sensitivities of the STHR index and the STHR slope, both in patients with clinical angina and those with proved coronary disease (p<0.001 for all comparisons). Thus test sensitivity and specificity of both standard and heart rate-adjusted ST depression criteria vary with the definition and selection of normal and abnormal clinical populations used to assess test performance. In these clinically relevant populations, improved performance of the STHR slope and the STHR index relative to standard ST segment depression criteria for the identification of coronary disease does not appear to be a consequence of selection factors.