Objective: To assess the role of serum lipid levels as screening tests in adults.
Design: Pooled analysis of clinical trials, supplemented by analysis of data from the Framingham Heart Study, to estimate the effect of cholesterol reduction in patient groups stratified by cardiac risk.
Study selection: Published randomized controlled trials of cholesterol reduction, meta-analyses of such trials, prospective cohort studies of the association between cholesterol levels and morbidity and death related to coronary heart disease, and cost-effectiveness analyses of cholesterol reduction.
Data analysis: Two-stage logistic regression on cardiac risk factors and outcomes in the Framingham Heart Study. The first stage predicted the risk for death from coronary heart disease using standard risk factors but not cholesterol; the second stage predicted the risk for death from coronary heart disease and all causes as functions of age and cholesterol level, stratified by the risk predicted from the first stage.
Results: Randomized clinical trials show that cholesterol reduction confers survival benefits in patients with symptomatic coronary disease. In asymptomatic middle-aged men, who are at lower risk for death from coronary disease, cholesterol reduction prevents coronary disease but has not been shown to prolong life. The risk model based on analysis of the data from the Framingham Heart Study is consistent with the randomized trial data and shows that in the demographic groups excluded from trials, the hypothetical benefits of cholesterol reduction are greatest when the underlying risk for coronary disease is greatest.
Conclusions: Screening with total cholesterol levels is most likely to be useful when done in populations at high short-term risk for dying of coronary heart disease, such as survivors of myocardial infarction and middle-aged men with multiple cardiac risk factors. In these populations, cholesterol reduction appears to be both effective and cost-effective. In other populations, the benefits of reduction are much smaller or are uncertain.