Risk Factors, Risk Prediction, and the Apolipoprotein B–Apolipoprotein A-I Ratio

In this issue, van der Steeg and colleagues report the results of a nested case–control study demonstrating that the apolipoprotein B–apolipoprotein A-I (apo B–apo A-I) ratio is a risk factor for future coronary events, independent of low-density lipoprotein cholesterol level, high-density lipoprotein cholesterol level, and other traditional risk factors (1). The study also suggests that the apo B–apo A-I ratio negligibly improves overall risk prediction compared with conventional coronary risk factors and the Framingham risk score. This latter finding contradicts an emerging literature that enthusiastically endorses the apo B–apo A-I ratio as an improved measure of risk (2–5). It is a good reminder that individual risk factors for important diseases are rarely useful for risk prediction and clinical decision making. We review the reasons why.

The risk factor approach to medicine emerged in an era when chronic diseases replaced infectious diseases as the largest burden to the public's health (6, 7); the term “risk factor” first appeared in the medical literature in Annals of Internal Medicine in 1961 (8). Whereas infectious diseases have only 1 necessary external causal agent, chronic medical diseases have many contributing causes and result from complex interactions between genetic factors and environmental exposures over years. Risk factors are measures of these many contributors. Identification of new risk factors is useful because however weak the association, they may point to novel causal pathways, to vulnerable subgroups, or to new targets for intervention. If risk factors are widely distributed in the population, they may still contribute to a significant number of cases of disease, and new and established risk factors together may determine the risk for disease.

Any clinician knows, however, that an individual risk factor (such as an elevated low-density lipoprotein cholesterol level) does little to distinguish patients who will develop an adverse …