Choice of coronary heart disease risk factor variables in a cross-sectional study of white South Africans

This report explores the possibility of redefining risk factors so as to improve their observed associations with prevalent coronary heart disease (CHD). A large cross-sectional community study of 5,895 white males and females aged 25-64 years yielded 240 cases of confirmed angina pectoris and 361 of confirmed myocardial infarction. Odds ratios for CHD end-points by level of risk factors when risk factors were expressed in the conventional manner (e.g. total cholesterol, systolic and diastolic blood pressure or current smoking) were often low and not statistically significant. Redefinition of risk factor variables in a manner that improved their specificity or compensated to some extent for the decreased risk factor exposure as a result of a CHD event (e.g. stopping smoking after a myocardial infarct) improved the strength of association. In this study, the most useful cholesterol variable was total cholesterol minus high-density lipoprotein cholesterol; for blood pressure the most useful variable was a blood pressure of 160/95 mmHg or above and/or being on anti-hypertensive treatment; and for smoking the most useful variable was the total duration of smoking (previous and current). Strong associations with CHD end-points were also found for conventionally expressed serum uric acid, diabetes prevalence (females) and family history of CHD. The study suggests that appropriate redefinition of risk factor variables and CHD end-points in cross-sectional studies yields associations similar in strength and direction to those found in prospective studies.