We welcome the new guidelines of the American College of Physicians [1], which selectively target for cholesterol screening persons at relatively high risk for coronary disease. We have developed a refinement of this approach [2] that counts and weighs recognized coronary risk factors, a method that is more accurate than simply counting risk factors. A Table based on the Framingham risk function and simple enough for ordinary practice can be readily constructed to target for screening those who may have a specific coronary risk and to signal the cholesterol concentration that confers that level of risk in the individual [2].

We believe that LaRosa [3] is unwise to reject this approach. He argues that knowledge of their serum cholesterol levels motivates patients to alter their coronary risk status but ignores evidence to the contrary [4]. He clings to the belief that dietary change is effective in substantially reducing cholesterol levels, relying on anecdotal evidence and intensive short-term studies. He chooses to ignore long-term controlled trials done in more than 10 000 high-risk patients in several countries, which show that acceptable dietary change lowers cholesterol levels by only 2% [5]. The direct and linear relation that he cites between cholesterol and coronary risk in young men is real, but it tells us nothing about the absolute risk of these young men, which is extremely small. The prevention of rare coronary catastrophes in young persons is a laudable aim but would entail long-term drug therapy for the many to benefit the few and an enormous financial outlay per coronary event prevented. LaRosa is correct to deplore the underuse of lipid-lowering drugs for secondary prevention, given the strength of recent evidence.

We argue that the logical solution is to direct resources at secondary prevention and not toward cholesterol screening in patients who have a vanishingly small coronary risk regardless of cholesterol level. Elevated blood cholesterol level is undoubtedly important and even fundamental to the atherogenic process, but knowledge of the cholesterol level in isolation is an extremely poor predictor of coronary risk [3] and of the likely benefit to be gained from intervention. These evidence-based methods do no more than direct effort toward those likely to benefit and away from those who cannot possibly benefit and may even be harmed by well-meaning intervention.