First, this policy ignores what is known about the pathophysiology of atherosclerosis itself—that is, that accumulation of extracellular cholesterol in pools underneath the endothelium is a key factor in the activity of an atherosclerotic plaque. It is this extracellular cholesterol pool that, through its effect on endothelial function, promotes local vascular spasm and clotting, as well as fissuring and rupturing of the endothelial cap [9, 10]. When total blood cholesterol levels are low enough, atherogenesis does not occur, even in the presence of other risk factors such as hypertension, smoking, and diabetes [11, 12]. In other words, an elevated circulating cholesterol level is the sine qua non for atherogenesis. A "normal" cholesterol level, then, is the level below which atherogenesis does not occur, when other risk factors are present. From both animal experiments and population studies, that total cholesterol level appears to be about 160 mg/dL (4 mmol/L) [1, 13]. No one would suggest that all persons with levels greater than 160 mg/dL should receive drug therapy, whether they have other risk factors or not. However, the knowledge that the cholesterol level is in a range that promotes atherogenesis is important for both physicians and patients in providing the motivation to change diet, adopt exercise habits, lose weight, stop smoking, and better manage blood pressure.

The issue is not the expense of cholesterol testing itself. Such testing is now reliable and inexpensive. Even though the NCEP recommends that such screening be part of a physician encounter, screening is greatly underused in both sexes and across the age spectrum [14]. Moreover, the implication that cholesterol lowering by diet is ineffective is not really tenable. It is true that in many clinical trials, dietary interventions have had only a modest effect. Many of the drug trials of coronary prevention, however, selected persons who were resistant to diet, and the recommended dietary interventions in those trials were themselves modest [15, 16]. The average effect of diet, moreover, is often not the effect experienced by individual patients. Many of us have seen patients who, through a combination of dietary change, weight reduction, and the institution of a regular exercise program, have dramatically lowered their cholesterol levels. The wide range of responsiveness can be seen in a recent study of 32 older men and women, in whom changes in the total cholesterol level resulting from the NCEP step diet ranged from +5% to –40%, although the average decrease was –20% [17].

Dietary change is not an all-or-none phenomenon. The more an individual person lowers cholesterol and saturated fat intake, the better the cholesterol lowering is likely to be. The striking interpopulation differences in cholesterol levels observed in epidemiologic studies [18] are a function not primarily of population genetics but of differences in the saturated fat and cholesterol content of national diets. This is most dramatically shown in migratory studies. Japanese persons living in Japan have much lower cholesterol levels than do Americans. When Japanese persons live in Hawaii or California, however, their diets become progressively more westernized, and their cholesterol levels and coronary risk, in turn, become indistinguishable from those of others in the same locales [19, 20]. The authors of the College's guidelines [7] themselves concede the effect of diet in endorsing dietary advice for the population at large.

The authors make much of the fact that little evidence justifies cholesterol interventions in younger and older populations. However, cholesterol levels determined in young men in their early twenties are directly and linearly related to coronary risk 40 years later [21]. Interestingly, persons with the lowest cholesterol levels have the lowest mortality from all causes; that is, a low cholesterol level is not associated with an increased risk for noncardiovascular mortality.

It is true that clinical trials have not been done in young adults; because of the low rate of clinical heart disease in this population, such trials are not likely ever to be done. It is also true, of course, that for the same reason, current NCEP guidelines do not recommend drug use in such persons unless their risk, as a result of high cholesterol levels and the presence of other risk factors, is high. To dismiss the need to know blood cholesterol levels, however, is to assume that no cholesterol level, no matter how high, requires intervention in young adults.

The situation in older persons, as a result of recently completed trials, is different. Subgroup analyses in several secondary prevention trials have indicated that the benefits in persons older than 65 years of age is almost identical to the benefits seen in younger persons [2, 22, 23]. In a recently completed primary prevention trial, the West of Scotland study [24], persons older than 55 years of age had benefits similar to those in younger participants.

Similarly, studies that included women (and separately analyzed the data for women) have shown the same benefit as that seen in men [2, 23, 25, 26], although the numbers have been too small to provide definitive information on mortality. Nevertheless, all of these studies point in the same direction. It is reasonable to ask how many more studies are needed. No clinical trial data exist for any of the major racial or ethnic minority groups in the United States. Shall we also withhold cholesterol screening for them? Surely, some rational extrapolation is permitted.

Knowledge of the cholesterol level has been shown, in a controlled trial, to lead to dietary-induced cholesterol lowering [27]. Such dietary interventions, moreover, have been shown in three recent studies (the Lifestyle Heart Trial, the St Thomas' Atherosclerosis Regression Study, and the Heidelberg study) [28-30] to reduce coronary disease, as seen on serial angiography. The NCEP handles the issue of screening more rationally, not by withholding cholesterol measurements from those persons without two other risk factors, but by using risk to temper the levels at which drug interventions are introduced.

The authors of the College guidelines [7] also point out that recent evidence found using more powerful cholesterol-lowering drugs indicates that the time between the institution of therapy and benefits on morbidity and mortality end points is much shorter than was once believed. They propose, therefore, that there is time to wait before the initiation of preventive interventions [2, 5]. This policy, however, condemns us to risk waiting until a catastrophic vascular event, which either is fatal or greatly increases the chance of subsequent mortality, occurs before we intervene with "preventive" measures. This approach is exactly analogous to proposing that we wait to measure blood pressure until the patient develops transient ischemic episodes or, worse, has a stroke. In approximately 25% of patients with coronary atherosclerosis, sudden death is the first manifestation of disease [31]. Even those who do not die, however, may be crippled by myocardial damage. A catastrophic first event is, of course, a reliable and perhaps even cost-effective way to identify those most susceptible to hypercholesterolemia, but it is hardly a suitable way to practice preventive medicine.

The authors of the College guidelines [7] imply that cholesterol lowering itself is potentially dangerous, a putative cause of noncardiovascular mortality. Recently completed large clinical trials of substantial cholesterol lowering, in persons with and those without overt coronary disease, however, showed not only a 20% to 30% decline in total mortality but also no increase in noncardiovascular mortality [2, 22]. A recent meta-analysis [32] showed that the increases in noncardiovascular mortality seen in some trials are completely unrelated to the degree of cholesterol lowering but may be related to the use of fibrates (particularly clofibrate), dextrothyroxin, or high-dose estrogen (in men). Even in that study, it is necessary to add all causes of noncardiovascular mortality to find any significant association, a procedure of dubious scientific validity.

The relation between spontaneously occurring low cholesterol levels and noncardiovascular mortality is more likely one of effect and cause. A low cholesterol level is the result of chronic disease, particularly in older persons. In a probability sample of the U.S. population [33], the association was almost entirely limited to persons older than 70 years of age. Moreover, in a recent study of 7309 middle-aged Japanese-American men in Honolulu, Hawaii [34], the association was not seen at all. Suicide was directly related to cholesterol levels and to heavy alcohol consumption [34].

Finally, the decrease in the average cholesterol level in the United States during the past 30 years, 20 mg/dL (0.5 mmol/L) [35], has been accompanied by a decrease, not an increase, in noncardiovascular mortality [36]. It is time to put the concern that cholesterol lowering promotes noncardiovascular mortality in a more rational perspective. This concern should not affect public policy.

The College's proposed guidelines [7] are the result of an attempt to promote what is called "evidence-based" clinical decision making. The term evidence-based suggests a laudable goal. As practiced here, however, evidence-based decisions (alleged to be devoid of the biases and prejudices of experts in the field) reflect an over-reliance on selected epidemiologic and clinical trial data. In other words, they ignore whole categories of important evidence. Such an approach is not as exact as these authors might hope. For example, such analyses are often based exclusively on effects on mortality. They ignore the benefits on morbidity provided by an intervention, particularly in a younger population, in which death is less likely in any case.

Such evidence-based analyses, moreover, ignore molecular and pathophysiologic data that link quantitative and qualitative changes in lipoproteins to vascular wall biology [37]. Over-reliance on epidemiologic and clinical trial data has, in the past, led to some seriously mistaken conclusions. For years, the notion has been promulgated that it is unnecessary to measure triglyceride levels because triglycerides do not independently predict coronary risk [38]. Such a conclusion ignores completely the critical role that triglycerides play in altering the structure and function of other lipoproteins, rendering them more atherogenic [39, 40]. As a result, knowledge of triglyceride levels and, in selected cases, intervention to lower them, should be a part of the management of hyperlipidemia [8].

The College's guidelines [7] also promote the remarkable point of view that practicing physicians may misuse cholesterol screening information and will overuse cholesterol-lowering medications. In doing so, the guidelines ignore available data to the contrary. Even in patients with established coronary disease, for whom there is almost no argument about the benefits of cholesterol-lowering drugs, cholesterol intervention is still underused [14]. Recently presented survey data in the United States indicate that only 27% of patients with coronary heart disease are being treated with cholesterol-lowering diets and only 29% are being treated with drugs (Cholesterol Awareness Surveys, 1995. Presented in Bethesda, Maryland, 4 December 1995, at the National Heart, Lung, and Blood Institute. Manuscript in preparation.) Such undertreatment is even more extensive in Canada, where guidelines similar to those proposed here have been promulgated for some time. There, only 22% of persons at high risk for coronary heart disease are treated with diet and only 8% are treated with drugs [41], even though unequivocal evidence shows that cholesterol lowering dramatically decreases coronary morbidity and mortality and total mortality in patients with established coronary disease [1, 2]. In 30 years of medical practice, I cannot remember an instance in which a prominent national advisory body has recommended that inexpensively obtained and prognostically important information about a patient be withheld for fear of the consequences of the physician's actions.

Guidelines developed in this way have met with considerable skepticism and resistance from practicing physicians. A recent discussion in Annals [42] reviewed the basis for that resistance, including the irrelevancy of group data to individual responses and the differences between individual and societal interests. Policy analysts need to pay more attention to these discrepancies as they attempt to formulate practice guidelines that will have relevance to real patient care and are acceptable to physicians and their patients.

In summary, the guidelines on cholesterol screening proposed in this issue [7] are wrong and misguided. They ignore the value of inexpensively obtained prognostic information for adults of all ages. They are equivalent to recommending that blood pressure or body weight be left unmeasured because the results might lead to inappropriate drug use. They assume a screening and drug-prescribing behavior on the part of physicians that is the opposite of what has actually been documented. Unfortunately, physicians still miss screening opportunities and do not treat high cholesterol levels often enough, even in the face of established coronary disease, which is an unequivocal indication. Finally, these guidelines ignore substantial evidence that an elevated blood cholesterol level is the most important single factor in the atherogenic process.

It has been suggested that in the United States, the attention given to cholesterol is fomented by over-eager pharmaceutical companies interested in seeing their products prescribed for large numbers of citizens who do not really need them. Remarkably, however, neither the authors of the College guidelines nor other critics mention the continued promotion of agricultural products high in saturated fat, which have led to high rates of death, both from atherosclerosis and probably from many forms of cancer. A major effort to change the western agricultural economy and the western diet would simultaneously decrease the risk for atherosclerotic disease and the need for drug prescription in many patients with hypercholesterolemia. In my view, this would be a much more rational and, ultimately, more cost-effective approach than the policy recommended in this issue.