Analysis of historical variables, risk factors and the resting electrocardiogram as an aid in the clinical diagnosis of recurrent chest pain.

A prospective study on 184 consecutive patients presenting with the chief complaint of recurrent chest pain (RP) for diagnostic coronary arteriography (CA) was conducted utilizing a simple questionnaire of historical, physical and electrocardiographic variables. A linear logistic regression analysis yielded a final data set of 13 variables. Concurrently, staff cardiologists who obtained the questionnaire data through direct questioning rendered a clinical diagnosis of either angina (coronary artery disease [CAD]) or noncardiac chest pain. Utilization of the regression analysis increased diagnostic accuracy from 69 to 86% (p less than 0.0003); sensitivity from 83 to 88% (NS) and specificity from 49 to 84% (p less than 0.0001). The best predictive variables for the presence or absence of obstructive CAD documented by CA were in order of decreasing value: age, electrocardiogram, pain aggravated by sex, sex (gender), pain aggravated by movement, diabetes mellitus, pain described as prickling, pain described as burning, pain relieved by rest, pain with radiation to both arms, associated nausea, associated dyspnea, and a history of a lipid disorder. Four variables were predictive of normal coronary anatomy (NCA), pain aggravated by movement, prickling, nausea, and dyspnea. Although this set of predictor variables may not apply equally well to all populations of cardiac patients, the availability and relative simplicity of the program allow for adding or deleting variables and thus provide for considerable potential in the diagnostic assessment of RP. An inexpensive pocket computer can utilize the coefficients generated by the logistic regression program to calculate the probability of CAD as the cause of RP.