The value of exercise electrocardiography testing in the identification of coronary restenosis: a probability analysis

We studied by means of probability analysis the role of exercise ECG in identifying coronary restenosis. A total of 213 patients were independently evaluated by clinical history, conventional assessment of the exercise ECG (‘yes or no’ statement), D score (a discriminant function derived from exercise ECG), and coronariography, 5.4 ± 2.8 months after successful coronary angioplasty. The initial probability of restenosis (30%), that is, the prevalence of the condition, was radically changed by the result of clinical history (77% for patients with angina vs. 17% for those without angina). By contrast, ECG binary assessment, due to its low accuracy (70% vs 82% of clinical history, P < 0.005), was unable to significantly change the established probabilities after symptomatic evaluation. Finally, D score, which greatly enhanced specificity (92% vs. 76% of bivariate assessment, P < 0.0001), proved to be useful in changing the probability (from 32% to 76% or to 25%) of patients (n = 34) with a discordant result (no angina/positive exercise ECG). When this stepwise approach was tested in 46 new patients, predicted and observed probabilities were actually very similar. We conclude that exercise ECG has a very limited role in identifying coronary restenosis if positive responses are not adjusted with a weighted score which takes into account other exercise derived factors.