For any preventive intervention, risk-based treatment thresholds are rational for two reasons. First, persons who have a low risk for a disease are less likely to benefit from treatment, but treating these persons exposes them to adverse effects of therapy. Thus, the ratio of benefit to risk is maximized if only persons with a high likelihood of contracting a disease are treated [3]. Second, more persons in low-risk groups than in high-risk groups must be treated to prevent disease. Therefore, scarce medical resources are used most efficiently if treatment is given only to those persons who are most likely to benefit from it.

The NCEP has implemented the concept of risk-based thresholds by specifying criteria that define a group at high risk for coronary heart disease; hypolipidemic therapy is only recommended for those persons with elevated cholesterol levels who meet these criteria. However, the effects of such interventions as aspirin therapy, which reduce the risk for coronary heart disease but that are not used primarily because of lipid-altering effects, have not been considered during the development of policies on cholesterol. We sought to model the effect of these therapies on the number of adults for whom cholesterol-lowering medications would be recommended if consistent, risk-based treatment thresholds were used.

We estimated the change in the number of adults in the United States who would exceed a risk-based threshold for the recommendation of hypolipidemic therapy after receiving an alternative intervention (an intervention not directed at cholesterol). Treatment was considered only for persons who had low-density-lipoprotein (LDL) cholesterol levels of 160 mg/dL (4.1 mmol/L) or greater (the cholesterol cutoff point chosen by the NCEP) and no history of coronary heart disease.

We used data from the Second National Health and Nutrition Examination Survey (NHANES II) [4] to estimate the proportion of adults in the United States between the ages of 20 and 74 years who exceeded the calculated treatment thresholds. Only persons with valid total cholesterol, high-density lipoprotein (HDL) cholesterol, and triglyceride levels who had fasted for at least 9 hours were included (n = 3284). We used published summary estimates from NHANES III [5] to adjust total and HDL cholesterol levels for changes that had occurred in these variables since NHANES II was done. Many persons in NHANES II were being treated for hypertension, and their blood pressure before treatment could not be determined; these persons were assigned a diastolic blood pressure of 100 mm Hg for calculation of their risk for coronary heart disease. This value was the average diastolic blood pressure of hypertensive persons in the National Health Examination Survey I [6] (a study done from 1960 to 1962 in which only 18% of hypertensive persons received medication to reduce blood pressure). Sampling weights were used in all calculations. Population projections for 1994 from the U.S. Bureau of the Census [7] were used to convert proportions to absolute numbers of adults.

To quantify the 10-year risk for coronary heart disease for each person, we used the Framingham Heart Study equations [8]. Risk factors used in this analysis were total cholesterol level, HDL cholesterol level, cigarette smoking, diabetes, and diastolic blood pressure.

To assess the effects that alternative interventions would have on the NCEP guidelines, we attempted to model the risk thresholds recommended by the NCEP. The guidelines do not specify a quantitative risk threshold (for example, "a 10-year risk for coronary heart disease of 5 in 100") above which treatment is recommended. Instead, they identify a set of categorical risk factors that define a high-risk group. Unfortunately, there is considerable overlap in risk among persons labeled "high-risk" and those labeled "low-risk" according to the guidelines ([9] and Avins AL. Unpublished data). Therefore, a single, consistent risk threshold cannot be identified. We used a population-based approach to translate the categorical method of the NCEP into quantitative risk thresholds.

Our primary source for these data was the work of Sempos and colleagues [10] who estimated that, according to NCEP guidelines, 11% of adults in the United States should be given drug therapy for elevated cholesterol levels. We determined that this proportion of persons would be given hypolipidemic medication if treatment were limited to all adults with LDL cholesterol levels greater than 4.1 mmol/L (160 mg/dL) who had a 10-year risk for coronary heart disease greater than 7 per 100.

Sempos and colleagues [10] also estimated the proportions of adults who would require medication if dietary therapy effectively reduced the level of LDL cholesterol by 5% or 10%. If dietary therapy reduced the level of LDL cholesterol by 5%, 7% of adults would receive drug therapy; this corresponds to a threshold of 10-year risk for coronary heart disease of 12 per 100. If dietary modifications reduced LDL cholesterol levels by 10%, approximately 5% of adults would be given medication; this is equivalent to a threshold of 14 per 100. Because the efficacy of dietary treatment for patients with hypercholesterolemia is uncertain [11-13], we used all three estimates in our analysis: "no diet effect" (no reduction in cholesterol level by dietary therapy), "small diet effect" (5% reduction by dietary therapy), and "large diet effect" (10% reduction by dietary therapy).

To estimate the effects of alternative interventions, we chose three treatments that reduce the risk for coronary heart disease but that are not specifically designed to lower cholesterol levels. On the basis of a meta-analysis of two randomized primary prevention trials done in middle-aged men [14], we assumed that aspirin decreased the risk for coronary heart disease by 33%. Observational data suggest that similar reductions occur in women [15]. In a separate analysis, we estimated that decreasing blood pressure in persons with hypertension would reduce risk for coronary heart disease by an average of 14% [16]. We also estimated that treatment of women 55 years of age or older with estrogen replacement therapy would decrease risk by 35% [17].

We limited our assessment of treatment costs to average wholesale prices for medications only. We assumed that persons treated for high cholesterol levels would receive a 3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibitor (also known as a statin) at a cost of $600 per year [18]. The cost of aspirin therapy (325 mg every other day) was assumed to be $2 per year, and the cost of estrogen therapy was assumed to be $135 per year [18]. No additional costs were assigned for antihypertensive therapy, because treatment is advised for all hypertensive persons to reduce their risk for stroke and renal disease [19]. Therefore, no additional costs are incurred for antihypertensive therapy to prevent coronary heart disease.

The analytic approach was as follows. We applied the Framingham risk equations to the NHANES II data to provide an estimate of the number of adults in the United States who would be candidates for hypolipidemic drug therapy for the three treatment thresholds identified. On the basis of this data, we estimated the total cost of drug therapy. We then assumed that all potential candidates for hypolipidemic drug treatment would be treated with aspirin, and we recalculated the number of hypercholesterolemic adults who still exceeded each treatment threshold. A similar analysis was done assuming that all high-risk persons received aspirin, that all hypertensive persons were treated, and that all women 55 years of age or older received estrogen therapy.

The results of these analyses indicate that the effects of alternative interventions are substantial. The number of adults in the United States with LDL cholesterol levels of 4.1 mmol/L (160 mg/dL) or higher and who exceed a range of thresholds of risk for coronary heart disease is shown in Figure 1. The changes induced by using alternative interventions are indicated by the dashed and dotted lines. For example, before use of an alternative intervention, approximately 15 million adults have a 10-year risk of at least 10 per 100. If all of these persons received prophylaxis with aspirin, only 9.5 million of them would still have a risk at that level. If antihypertensive and estrogen therapies were also used by all appropriate persons, the number of adults exceeding this threshold decreases to 6.9 million.

View larger version (16K): [in this window] [in a new window]   Figure 1. Number of hypercholesterolemic adults who exceed any risk threshold for coronary heart disease. Comparison of risk before use of an alternative intervention (solid line), after aspirin therapy (dashed line), and after aspirin, antihypertensive, and estrogen-replacement therapies (dotted line).

Regardless of the threshold, with the use of alternative interventions, substantially fewer persons would still be offered hypolipidemic drug therapy (Table 1). For example, if a small dietary effect could be achieved, aspirin use alone decreases the risk for coronary heart disease so that approximately 45% of persons for whom drug therapy previously would have been recommended would no longer be candidates for such treatment. The addition of antihypertensive and estrogen replacement therapies results in further reductions: Sixty-one percent of persons would no longer have sufficient risk to qualify for drug therapy. Overall, between 6.2 and 8.1 million adults would not be given cholesterol-lowering medications if the effect of these alternative therapies on risk-based thresholds were considered.

Cost savings associated with aspirin use are approximately $3 billion per year (Table 1). If all three alternative interventions are considered, cost savings range from $3.6 to $4.2 billion per year.

Drug treatment of hypercholesterolemia is one of the most controversial and expensive preventive interventions [20-23]. To provide guidance for the treatment of elevated cholesterol levels, the NCEP established treatment recommendations on the basis of absolute risk for coronary heart disease [1, 2].

Although they were an important advance, the NCEP guidelines did not consider the effect of other interventions that decrease the risk for coronary heart disease. We found that accounting for these alternative interventions would avoid the need for specific lipid-lowering drug treatment in many persons with hypercholesterolemia who would otherwise have been considered to have a risk for coronary heart disease high enough to warrant such therapy. For example, if all high-risk adults with elevated cholesterol levels were to receive aspirin, risk for coronary heart disease would be reduced to a level at which drug treatment would no longer be recommended for 25% to 60% of these persons. If the published estimates of the risk-reducing effects of estrogen replacement and antihypertensive therapies [16, 17, 24] are correct, these interventions would eliminate the need for cholesterol-lowering drugs in approximately 45% to 70% of persons who would otherwise be candidates for medication. Because the exact effects of diet on cholesterol levels are not known, we provide a range of results that depend on the efficacy of dietary therapy. These estimates do not include the few women whose elevated LDL cholesterol levels would fall below the 4.1 mmol/L (160 mg/dL) cutoff point as a result of any lipid-lowering effects of estrogen therapy.

Acknowledging the effects of alternative interventions is similar to using elevated HDL cholesterol levels as a negative risk factor in the NCEP II guidelines. Many persons who are currently considered to have high risk (and therefore to be candidates for medication) actually have a risk for coronary heart disease similar to that of persons labeled "low-risk" after receiving alternative interventions. These persons should not be routinely advised to use cholesterol-lowering drug therapy. As the efficacy of other interventions (such as exercise, vitamins, and weight loss [24]) is more clearly defined, further adjustments to lipid-screening guidelines should be considered. Because lifestyle modification is the preferred first line of therapy [1], successful lifestyle changes should always be considered when assessing the overall cardiac risk of a patient.

Accounting for the effect of alternative interventions results in substantial cost savings while adhering to the risk-based concepts of the NCEP guidelines. For example, the NCEP estimates that between 8.7 and 18.5 million adults should receive hypolipidemic drug treatment [10], at a cost of $5.2 to $11.1 billion annually. If alternative interventions are considered in risk assessment, many patients would no longer be candidates for medication, reducing pharmaceutical costs by approximately $3 to $4 billion per year. By ignoring the effects of alternative interventions, current guidelines recommend large expenditures for the treatment of persons at relatively low risk for coronary heart disease—an inefficient use of health care resources. Future cost-effectiveness analyses of hypolipidemic therapy should consider this effect. It is not necessary to consider the magnitude of the risk-reducing effects of these interventions compared with that of the risk-reducing effects of lowering cholesterol levels. We need only to recognize that these interventions will sufficiently reduce the risk for coronary heart disease in many hypercholesterolemic persons to below a threshold value, although their LDL cholesterol levels may still surpass the 4.1 mmol/L (160 mg/dL) cutoff point.

It is important to recognize that these alternative therapies may have other benefits as well as adverse effects [17, 19, 25-30], all of which would have to be included in a complete analysis. These risks, however, should be viewed in the context of the known and potential risks associated with cholesterol-lowering drugs [20, 31, 32].

Our study has several limitations. First, we assumed that the effects of each intervention that reduced risk for coronary heart disease were independent, although little is known about their potential interactions. One post hoc clinical trial analysis, however, suggests that aspirin and hypolipidemic therapy exert independent effects [33]. Second, the Framingham risk equations may not adequately predict coronary heart disease. Third, we used NHANES II data, which is now 18 years old. Because newer data (NHANES III) are not yet available, we used published estimates to adjust for temporal trends in lipid levels [5]. The proportions estimated by Sempos and colleagues [10], were also derived from an analysis of the NHANES III data set and may not strictly apply to the NHANES II data used in these analyses. Fourth, we did not consider niacin therapy (which is much cheaper than statin therapy) in our calculations. However, statins have a more beneficial short-term effect on mortality, and compliance with niacin therapy is low [34, 35]. Finally, the estimates we present are potential estimates only and may differ according to patient compliance and preference. Although some patients and physicians may favor hypolipidemic drug therapy over therapy with aspirin, antihypertensive agents, or estrogen, we believe that this preference should be considered carefully given the individual and public health implications of the choice.

In summary, our analysis indicates that many adults who are currently candidates for cholesterol-lowering drug treatment because they are at high risk for coronary heart disease might no longer need such therapy after receiving alternative interventions that reduce risk for coronary heart disease. These alternative therapies are much less expensive than most hypolipidemic agents and have a long-term safety record that is more clearly defined. The cost savings of this approach are considerable. In view of these results, it may be time to re-examine current recommendations for the treatment of elevated cholesterol levels to reduce the risk for coronary heart disease.