THE CORRELATION OF EPIDEMIOLOGIC VARIABLES

Data from a sizable hospital-based, case-control study from 1985 to 19 90 permitted us to examine the correlation of variables by degree of i nterrelationship. The variables examined relative to their confounding interrelationships include smoking; consumption of alcohol, coffee, m eat, vegetables, and fruits; body mass index; education, and age. The variables with the broadest impact on others are age and education. Wh ile the association among these variables is generally known, the degr ee of association made possible by our large database is generally not fully appreciated and is of obvious significance when data are adjust ed for one other variable. In some instances, the intercorrelation amo ng different variables is relatively complicated. For instance, cigare tte smoking correlates positively with meat consumption and negatively with intake of fruits and vegetables, which, in turn, correlates with alcohol intake. These interrelationships need to be clearly understoo d before interpreting epidemiologic data for causation. In general, th e correlations are similar for men and women. This study of correlatio n of variables, some expected and some unexpected, should be of value as a source to epidemiologists and provide a useful base.