A risk scoring system for prediction of coronary heart disease based on multivariate analysis: development and validation.

Considering the multifactorial disposition and the need of primordial prevention of coronary heart disease (CHD), a risk scoring system for the prediction of CHD was devised at Govt Medical College, Nagpur, India. In this pair-matched case-control study of 154 cases and 154 age and sex matched controls, socioeconomic status, physical inactivity, family history of CHD, type A personality characteristic, cigarette smoking, alcohol consumption, body mass index, diabetes mellitus, hypertension, total serum cholesterol and oral contraceptive use (in women) were studied for association with CHD. The additive risk scoring system based on the results of conditional multiple logistic regression identified five factors, namely, socioeconomic status, physical inactivity, diabetes mellitus, hypertension and toal serum cholesterol with statistical weights of 3,5,2,4 and 5 respectively. On back-validation using receiver operating characteristic (ROC) curve, a total score of 12 was found to be the cut off point above which there was increased risk of CHD. The overall predictive accuracy of this system-equivalent to the area under the ROC curve-was 0.7962 (95% Confidence Interval 0.7468-0.8455). On prospective validation using a separate group of 140 cases and 140 controls, the predictive accuracy was found to be 0.6964 (95% Confidence Interval 0.6341-0.7587). Future studies need to assess the risk scoring system in population based studies.