Logistic model estimation of death attributable to risk factors for cardiovascular disease in Evans County, Georgia.

Linear logistic analysis of the relationship of cardiovascular disease risk factors to an overall measure of health effect, ten-year mortality, revealed significant associations of death with systolic blood pressure, age, sex, diabetes mellitus, smoking, cholesterol, obesity (Quetelet index), race and social index. Attributable risk and population attributable risk estimates were derived from the model by changing actual variable values to target values. The results confirmed systolic hypertension and smoking as major public health problems and diabetes mellitus as a powerful risk factor for death. The small detrimental effect of cholesterol on probability of death limits the potential for an overall beneficial effect of preventive intervention. In fact, drug intervention in two reported trials of cholesterol reduction had negative overall effects. Demonstration of the association of a characteristic with a specific disease state does not alone justify attempts to eliminate the high risk state from a population. An overall detrimental health effect must be documented by suitable studies before trials of preventive intervention are undertaken or recommendations made to the public.