[Probability of coronary lesions evaluated by the discriminant analysis of clinical and ergometric parameters]

It has been suggested that multivariate analysis can be used to improve the diagnostic value of the conventional exercise tests. In order to evaluate the usefulness of this method in clinical practice, we have conducted a retrospective study of 558 male subjects without history of infarction who had undergone an exercise test and, less than 90 days later, a coronary arteriography. All exercise tests were performed according to Bruce's procedure and with an equipment which provided continuous averaging of ST segment values on V5, V2 and VF leads. The prevalence of significant coronary lesions (more than 50% luminal narrowing) was 56%. The values of 12 clinical and electrocardiographic parameters at rest and during exercise were subjected to a step-by-step multivariate analysis. Compared with the quantitative analysis of ST alone, the multivariate analysis increased the sensitivity (68% vs 59%, p less than 0.05) and specificity (83% vs 76%, NS) of the tests and the percentage of well-classified subjects (74.6% vs 66.8%, p less than 0.01). The best combination was obtained with the first 5 parameters of the final classification, viz.: (i) duration of exercise; (ii) clinical history of angina; (iii) anginal pain during exercise; (iv) age, and (v) maximum heart rate. The validity of the method was demonstrated on a recent series of 200 consecutive patients where the prevalence of coronary lesions was 58%. The discriminant function score enabled 74% of these patients to be correctly classified and improved the sensitivity of their exercise test.(ABSTRACT TRUNCATED AT 250 WORDS)