Multivariate analysis of risk factors in coronary artery disease--predictive value in case detection in cardio vascular epidemiology
Krishnaswami, S.; Richard, J.; Sathyamurthy, I.; Uthaman, C.B.; Sukumar, I.P.
Indian Journal of Medical Research 79: 439-444
1984
ISSN/ISBN: 0019-5340 PMID: 6746061 Document Number: 221723
Univariate and multivariate analyses were carried out on the role of 10 known risk factors in the production of coronary artery disease (CAD) in 400 patients who were subjected to selective coronary arteriography. In 213 patients with CAD and 187 patients with normal coronary arteries, a computerized multiple regression analysis showed that 31.1% of CAD could be explained by a statistically significant interplay of variables with age, smoking, male sex, diabetes mellitus, family history of coronary artery disease and hypertension being the most important risk factors. High lipids, overweight and pattern of work played a less significant role. The predictive risk score equation obtained by such a regression analysis, when tested on the next 50 patients who had selective coronary arteriography done, correctly classified patients into CAD and normal in 90% of the patients. With a line of demarcation at a risk score of 0.330 (range 0-1), the sensitivity was 97.36%, specificity 67% and predictive value 90%. Lipid levels, overweight and work pattern were apparently less contributory to the occurrence of CAD in this series as compared to factors like age, smoking, diabetes, hypertension and family history of ischemic heart disease. The predictive risk score deserves to be evaluated for purposes of mass screening of large populations, subject however, to the inherent limitations of this approach.