Author + information
- Received October 6, 1998
- Revision received April 2, 1999
- Accepted May 14, 1999
- Published online September 1, 1999.
- Lee Goldman, MD, MPH, FACC∗,* (, )
- Pamela Coxson, PhD∗,
- Maria G.M Hunink, MD, PhD†,
- Paula A Goldman, MPH‡,
- Anna N.A Tosteson, ScD§,
- Murray Mittleman, DrPh, MDCM∥,
- Lawrence Williams, MS¶ and
- Milton C Weinstein, PhD‡
- ↵*Reprint requests and correspondence: Dr. Lee Goldman, Department of Medicine, University of California, San Francisco, 505 Parnassus Avenue, San Francisco, California 94143-0120
This study was undertaken to project the population-wide effect of full implementation of the Adult Treatment Panel (ATP) II guidelines of the National Cholesterol Education Program (NCEP).
The ATP II has proposed guidelines for cholesterol reduction, but the long-term epidemiologic influence of its components has not been fully examined.
We used a calibrated, validated simulation of the U.S. population, aged 35 to 84 years to estimate the potential for the NCEP guidelines, under varying assumptions, to reduce coronary heart disease morbidity and mortality and overall mortality from the years 2000 to 2020.
Primary prevention would yield only about half of the benefits of secondary prevention despite requiring nearly twice as many person-years of treatment. The projected increase in quality-adjusted years of life per year of treatment for secondary prevention was 3- to 12-fold higher than for primary prevention. To yield population-wide epidemiologic benefits equivalent to NCEP recommendations for secondary prevention, primary prevention would require a nearly sixfold increase in the number of persons treated compared with NCEP recommendations. All benefits of universal success of the NCEP primary prevention “screen and treat” guidelines could be achieved by a 11 mg/dl (8%) population-wide reduction in low-density lipoprotein cholesterol levels among persons without preexisting coronary heart disease.
The NCEP guidelines for targeted primary prevention can be a useful component of a rational public health strategy, but only as a complement to the more appealing strategies of secondary prevention and “across-the-board” programs to lower all cholesterol levels.
The National Cholesterol Education Program (NCEP) (1,2)clinical strategy guidelines for screening and treating elevated cholesterol levels are rapidly becoming the standard of care in the U.S. We and others (3–8)have previously presented cost-effectiveness analyses indicating that cholesterol reduction using medications becomes progressively more attractive as the overall risk of coronary heart disease events increases, and that secondary prevention is nearly always more attractive than primary prevention from a cost-effectiveness perspective. However, little attention has been paid to the population-wide epidemiologic considerations. Because many more people would be potentially eligible for primary prevention of a first coronary heart disease diagnosis than for secondary prevention after their initial diagnosis, we analyzed the epidemiologic effect of the three different components of the NCEP guidelines: secondary prevention, primary prevention in higher-risk persons and primary prevention in moderate-risk persons. We also analyzed the potential incremental benefit of extending “diet-only” interventions to more individuals, and our analyses considered alternative scenarios.
The Coronary Heart Disease Policy Model (3,9–11)is a Markov state-transition, computer-based model with three integrated submodels: Demographic–Epidemiologic, Bridge and Disease History. The model is based on the assumption that categorized data from observational cohort studies (12)are consistent with experimental clinical trials (13–17)and can estimate the risk of coronary heart disease events, and that ultimately such events influence mortality (18).
The Demographic–Epidemiologic submodel applies to the U.S. population aged 35 to 84 years without coronary heart disease. Each year, a new cohort of 35-year-old individuals (19,20)enters, and persons turning age 85 exit. To exit before age 85, persons must die or develop coronary heart disease, at which time they enter the Bridge submodel.
The Demographic–Epidemiologic submodel assesses each individual’s risk of developing coronary heart disease and death from all causes based on risk strata defined by age (35 to 84 years in annual intervals), gender, smoking status (yes or no), diastolic blood pressure (≤94, 95–104 or ≥105 mm Hg), serum cholesterol (≤240 mg/dl [6.21 mmol/L], 240–299 mg/dl [6.21–7.75 mmol/L] or >299 mg/dl [7.75 mmol/L]), high-density lipoprotein (HDL) cholesterol (<35 mg/dl [0.91 mmol/L], 35–49 mg/dl [0.91–1.27 mmol/L] or >49 mg/dl [1.27 mmol/L]) and low-density lipoprotein (LDL) cholesterol (<160 mg/dl [4.16 mmol/L], 160–189 mg/dl [4.16–4.91 mmol/L] or >189 mg/dl [4.9 mmol/L]). The initial multivariate distributions of risk factors, conditional on age and gender, and transitions with aging were modeled using data from the Second National Health and Nutrition Examination Survey (NHANES II) (21)and updated to reflect 1986 population averages based on observed trends (22,23)and data from NHANES III (24).
To predict the annual probability of a coronary heart disease event or noncoronary heart disease death on the basis of risk factors, multiple logistic risk functions that controlled for all risk factors in the model simultaneously and that allowed for interaction terms between risk factors and age were estimated using 36-year follow-up data from the Framingham Heart Study (25). The analysis assumed that serum cholesterol did not affect noncoronary mortality (13,14), but we also considered possible adverse effects in sensitivity analyses. We assumed that the effect of risk factor reduction on coronary mortality was mediated by its influence on coronary events (13–17). Coronary mortality and overall mortality were derived from U.S. vital statistics (26).
The Bridge submodel characterizes participants as having angina, myocardial infarction or cardiac arrest during the first 30 days after they develop coronary heart disease, and applies mortality probabilities and 30-day resource costs to persons in each group (9). The Disease History submodel tracks subsequent coronary heart disease events (myocardial infarction, coronary revascularization and cardiac arrest), case-fatality rates and resource costs in persons with prevalent coronary heart disease who survive the Bridge submodel or who enter the model at age 35 with preexisting coronary heart disease.
Implementation of NCEP guidelines
The guidelines of the Adult Treatment Panel II of the NCEP for dietary and drug treatment of LDL cholesterol call for secondary prevention to reduce LDL to 100 mg/dl, aggressive primary prevention with a goal of 130 mg/dl for some patients (“goal 130”) and less demanding primary prevention with a goal of 160 mg/dl for other patients. We assumed all persons would meet the stated goals either with diet or medications as needed beginning at age 35 and continuing to age 84. The target population for secondary prevention includes all persons with clinical evidence of coronary heart disease, with subjective guidelines for including persons with coronary heart disease and low LDL levels (100 to 129 mg/dl). The more aggressive primary prevention (which we term “high-risk”) targets the population free of coronary heart disease but with medium (160–189 mg/dl) or high (>189 mg/dl) LDL and a net of two or more additional coronary heart disease risk factors, where age (≥45 for men and ≥55 for women), smoking, hypertension, low HDL (<35 mg/dl), family history of premature coronary heart disease and diabetes mellitus are each positive risk factors, and high HDL (>60 mg/dl) is a negative (protective) factor. The less demanding primary prevention (which we term “medium-risk”) targets persons with high LDL and at most one additional risk factor. Additional categories of persons are identified as candidates for “diet-only” intervention.
In our simulations, the target population for secondary prevention was assumed to be all persons with coronary heart disease. For primary prevention, simulations included all NCEP risk factors, except that we could not specifically identify persons with a family history of premature coronary heart disease or diabetes mellitus. The NCEP guidelines were simulated by decreasing the mean LDL level for each cell in the model that corresponded to the target population to the specified goal for that population.
Outcome measures included myocardial infarction, coronary heart disease death, years of life and quality-adjusted years of life. Quality of life adjustments using time trade-off utility weights were calculated for persons with a history of coronary heart disease (10,27). All simulations were for the 21-year period from 2000 to 2020.
Variations on model assumptions
In alternative simulations, an association between cholesterol and noncoronary heart disease death was estimated based on the natural logarithm of serum cholesterol using pooled data from several studies (28,29). We also implemented a modified version of the NCEP guidelines in which high-risk and medium-risk persons free of coronary heart disease were treated with as aggressive an LDL goal as the population with coronary heart disease.
In other analyses, we assumed that the NCEP guidelines for dietary intervention alone, i.e., without follow-up drug intervention, would be implemented for: 1) persons with medium LDL (160–190 mg/dl) and at most one additional coronary heart disease risk factor; and 2) persons with low LDL (130–160 mg/dl) and at least two additional coronary heart disease risk factors, and that a 10% reduction in LDL could be realized. Our model does not have a separate subcategory for LDL between 130 and 160 mg/dl, but we estimated from NHANES data that the population with two coronary heart disease risk factors and LDL in this range is roughly half (41% for men, 52% for women) of the population of persons with LDL less than 160 mg/dl, and made our estimates accordingly.
If the Adult Treatment Panel II (ATP) recommendations of the NCEP were followed to achieve the desired cholesterol goals over the 21-year period from the year 2000 to 2020 in adults 35 to 84-years-old, there would be about 500 million person-years of treatment, with an average of about 11.5 million (48%) women and 12.5 million (52%) men treated each year. About two-thirds of the years of treatment are estimated to be for primary prevention, while secondary prevention would be projected to account for only about one-third of the overall person-years of treatment.
Impact of treatment
In women, the 26% of treatment years for secondary prevention are projected to yield 59% of the overall decrease in myocardial infarctions, 78% of the decrease in coronary heart disease deaths, 80% of the increase in years of life and 77% of the increase in quality-adjusted years of life (Table 1). The projected increase in quality-adjusted years of life per year of treatment was more than 5.5-fold higher for secondary prevention than for high-risk primary prevention and nearly 12-fold higher for secondary prevention compared with medium-risk primary prevention.
In men, the 42% of the person-years of treatment allocated toward secondary prevention was estimated to yield 60% of the decrease in myocardial infarctions, 76% of the decrease in coronary heart disease deaths, 78% of the increase in years of life and 75% of the increase in quality-adjusted years of life (Table 2). Overall, secondary prevention yielded projected increases in quality-adjusted years of life per person-year of treatment that were more than threefold higher than for high-risk primary prevention and nearly sixfold higher than for medium-risk primary prevention.
Across all ages among both men and women, secondary prevention provided about 60% to 80% of the additional years of life and quality-adjusted years of life after the institution of treatment estimated to be obtained from following the ATP II guidelines (Fig. 1). For the population as a whole, however, the projected percent increase in expected quality-adjusted years of life after the institution of treatment was generally small, and was more than 4% only among men over age 65 and women over age 75 (Fig. 2).
The percent change in years of life increases for age ranges 45 to 54 and older due to the cumulative effect of lives saved by intervention in younger ages. For example, by 2020, 75- to 84-year-old men and women have estimated increases in years of life of 3.3% and 2.3%, respectively. By 2020, 65- to 74-year-old men have a projected 1.6% additional years of life. All other age ranges have less than a 1% increase. In contrast, for the absolute number of myocardial infarctions and coronary heart disease deaths, the cumulative effect is in the reverse direction, because persons who would now survive to an older age are at risk for myocardial infarctions and coronary heart disease death, even if at a reduced rate. Thus, although all age ranges have a projected decrease in rates of myocardial infarction and coronary heart disease death, the absolute number of these events is projected to increase in ages 55 to 64 and 65 to 74.
Absolute benefit of treatment
In the secondary treatment of women, the projected additional quality-adjusted years of life per person-year of treatment ranged from 8 per 1,000 for women ages 35 to 44 to nearly 140 per 1,000 for women ages 75 to 84 (Table 3). In comparison, for medium-risk primary prevention, the estimated yield was about 2 quality-adjusted years of life per 1,000 years of treatment for women ages 35 to 44, rising to 28 per 1,000 years of treatment for women ages 75 to 84. Trends in men were analogous but less dramatic (Table 3).
These estimates of differential effects are not surprising given the very different baseline risks. The annualized risk of coronary heart disease death in women who already have coronary heart disease is 10% to 130% higher than those of high-risk and medium-risk women without coronary heart disease who are 30 years older (Fig. 3). Similarly, men who already have coronary heart disease also have risks that are 15% to 50% higher than those of high-risk men without coronary heart disease who are 30 years older (Fig. 4). Annualized risks for myocardial infarction follow similar patterns.
Aggressiveness of treatment
The greater influence of secondary prevention is related in part to the more aggressive goal of LDL cholesterol reduction for secondary prevention (100 mg/dl) than for high-risk primary prevention (130 mg/dl) or medium-risk primary prevention (160 mg/dl). However, even if all three groups were treated to reduce their LDL cholesterol levels to 100 mg/dl, about 50% of the reduction in coronary heart disease deaths in both women and men would be due to secondary prevention, although only about 4% of the years of treatment of women and 7% of the years of treatment in men would be for secondary prevention. To achieve the aggressive goals, about 68 million medium-risk women and 47 million medium-risk men would need to be treated annually, and the estimated total number of treated persons would be increased by nearly sixfold compared with NCEP recommendations.
Alternative primary prevention strategies
The influence of the Adult Treatment Panel II recommendations for primary prevention of coronary heart disease would be the equivalent of the results achieved by an 8% (11 mg/dl) reduction in LDL cholesterol in persons without coronary heart disease. Alternatively, an 8-mm Hg decrease in diastolic blood pressure or an 18% decrease in smoking rates in persons without coronary heart disease would save about the same number of quality-adjusted years of life.
Influence of including “diet-only” subjects
If “diet-only” individuals are included in primary prevention, the number of person-years of treatment roughly doubles. Even in the unlikely event that dietary interventions would reduce LDL cholesterol by 10%, at least half of the decrease in myocardial infarctions as well as two-thirds of the decrease in coronary heart disease deaths and the increases in years of life and quality-adjusted years of life would be attributable to secondary prevention, although only 15% of the person-years of treatment would be for secondary prevention.
Possible noncoronary heart disease effects of cholesterol reduction
If the noncoronary heart disease death rate increased because of adverse effects of cholesterol lowering, the benefits of cholesterol reduction would be blunted by an average of about 11,000 more noncoronary deaths per year. Nevertheless, years of life and quality-adjusted years of life were still estimated to increase in every age range for both men and women by about 80% to 90% of what would be predicted without adverse effects on noncoronary mortality. The incremental years of life gained by cholesterol reduction would be less blunted in secondary prevention than in primary prevention, and estimates of the relative benefit of interventions would be even more in favor of secondary prevention.
Altering HDL criteria
Of persons with HDL >50 mg/dl, about 60% of men and 45% of women have HDL levels <60 mg/dl, and hence do not meet NCEP ATP II criteria for a protective HDL level. However, even if we assumed that all people with HDL >50 mg/dl would be eligible for treatment under the NCEP ATP II guidelines, the proportion of population-wide benefit attributable to primary prevention would increase by only about one percentage point in men and two percentage points in women, and conclusions regarding the relative influence of primary versus secondary prevention were not appreciably affected by our use of an HDL cutoff of 50 mg/dl.
We do not dispute the role of LDL cholesterol as a risk factor, HDL cholesterol as a protective factor, or the benefits of medications for improving lipid levels and reducing coronary heart disease events, coronary heart disease mortality or overall mortality (1,13–18,30). Of note is that the CHD Policy Model has been accurate in previous predictions (31,32), but we cannot guarantee future predictions. Our model could not include diabetes or a positive family history of premature coronary heart disease, and its specifications required that we model an HDL threshold of >50 mg/dl, rather than >60 mg/dl, as a protective factor. These limitations led to some underestimation of the number of persons eligible for primary prevention, but a sensitivity analysis suggests that any effect on our estimates of the population-wide influence of primary prevention is very minor. Furthermore, our estimate of the number of Americans eligible for secondary prevention is also on the lower side of the various estimates (33), and any increase in persons eligible for secondary prevention would only reinforce our conclusions.
Our analyses emphasize that under a wide variety of assumptions, 60% to 75% of the population-wide epidemiologic benefit of cholesterol reduction potentially achievable from the ATP II guidelines would be in secondary prevention. These analyses demonstrate that the segment of the population for which cholesterol reduction is the most cost-effective on a per-person basis (3–8)is also the segment that is likely to realize about two-thirds of overall benefits.
Our estimates were remarkably consistent, were similar for women and men, and were of reasonably similar relative effect across various age groups. Even much more aggressive treatment goals or the inclusion of a “diet-only” intervention in a much larger number of persons would have only a small effect on the relative epidemiologic importance of secondary as compared with primary prevention despite the inclusion of far greater numbers of persons in a primary prevention intervention.
The guidelines of the ATP of the NCEP (1)emphasize a case-finding strategy, sometimes called “screen and treat,” analogous to the very successful approach to diagnosing and treating hypertension (34). This approach to hypertension is widely credited with achieving substantial populationwide reductions in blood pressure and stroke (35)and is clearly worth the cost (36), mostly because screening can be performed inexpensively during routine health visits.
For population-wide “screen and treat” programs for cholesterol reduction, however, the challenge is more complicated. Effective classification may require multiple blood tests, and these tests must be repeated to assess treatment efficacy. Furthermore, current medications are generally more costly on an annual basis than medications to control hypertension, and it is not surprising that cost-effectiveness estimates for primary prevention have generally been far less favorable for hypercholesterolemia than for hypertension.
Prior concerns regarding the ATP II guidelines have focused on the potential adverse effects of lowering cholesterol (37), the current cost of medications (38)and the practicality of screening, classifying and appropriately treating the entire population (39), but we are less concerned with these three issues. First, recent data suggest that fears regarding the adverse effects of lowering cholesterol were at least exaggerated if not unfounded (13–18,30). Second, concerns about cost-effectiveness can be addressed by lowering the cost of medications, which represent most of the costs of implementing the ATP II guidelines. Third, although recent surveys estimate that only 1 in 12 adults is being screened annually (39)and that more than 50% of hypercholesterolemic U.S. citizens remain unaware of their elevated cholesterol levels (40), such logistic hurdles could be overcome if the recommended strategy is truly the most appropriate route to solving a major public health problem. In addition, cholesterol reduction likely reduces stroke (41), a noncoronary heart disease benefit that is often overlooked in economic analyses.
However, even if the ATP II guidelines were followed perfectly nationwide, about two-thirds of the influence would come from secondary prevention, where costs per benefit are lower (41)and the likelihood of implementation by physicians and compliance by patients is far higher. Although not directly modeled, the stroke benefit of cholesterol reduction would also likely be largest in persons with more advanced atherosclerosis and other evidence for coronary heart disease, thereby being consistent with our estimate of the relatively greater influence of secondary as compared with primary prevention. Second, the population-wide influence of perfect implementation of the ATP II guidelines for primary prevention in persons without coronary heart disease would be roughly equivalent to an 8% (11 mg/dl) reduction in LDL cholesterol in these same persons. Reductions of this magnitude (about 13–14 mg/dl in total cholesterol) have been achieved over the last decade or so (42–44), without an appreciable reliance on medications or systematic changes in the food supply. Of course, it is unlikely that any “screen and treat” strategy will ever be successful to its full potential epidemiologic influence (45,46). Interventions across the population, such as immunization, fluoridation or even dietary changes (2,46), as suggested by the Population Strategy Panel of the NCEP (47)and by others, are as or are more likely to succeed compared with more complicated approaches that require targeted screening, treatment and compliance.
A successful public health program for the prevention of coronary heart disease via lipid lowering should emphasize secondary prevention with diet and medications. Primary prevention with approaches such as dietary changes that do not require individual measurement of cholesterol levels (2,48)are also extremely appealing. Targeted “screen and treat” primary prevention based on the ATP II guidelines can also be very useful and can be cost-effective for especially high-risk patients without clinical coronary heart disease, but should not be misinterpreted as the principal strategy for maximizing public health effects (49,50). Because a substantial proportion of coronary heart disease deaths occur in previously asymptomatic persons, the development of cost-effective primary prevention strategies must remain a high public health priority.
☆ Supported by Grant RO1 H5-06258 from the Agency for Health Care Policy and Research, Rockville, MD.
- Adult Treatment Plan
- high-density lipoprotein
- low-density lipoprotein
- National Cholesterol Education Program
- National Health and Nutrition Examination Survey
- Received October 6, 1998.
- Revision received April 2, 1999.
- Accepted May 14, 1999.
- American College of Cardiology
- Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults
- Goldman L,
- Gordon D.J,
- Rifkind B.M,
- et al.
- Pedersen T.R,
- Kjekshus J,
- Berg K,
- et al.
- Pharoah P.D,
- Hollingworth W
- Stinnett A.A,
- Mittleman M.A,
- Weinstein M.C,
- et al.
- Wilson P.W.F,
- D’Agostino R.B,
- Levy D,
- et al.
- Long-Term Intervention with with Pravastatin in Ischaemic Disease (LIPID) Study Group
- Gould A.L,
- Rossouw J.E,
- Santanella N.C,
- Heyse J.F,
- Furburg C.D
- ↵U.S. Bureau of the Census. Projections of the Population of the United States by Age, Sex, and Race: 1982–2050. Current Population Reports. Population Estimates and Projections. Washington, DC: U.S. Government Printing Office; Series P-25. No. 922, 1982.
- U.S. Bureau of the Census. U.S. Population Estimates by Age, Sex, Race and Hispanic Origin: 1989. Current Population Reports, Population Estimates and Projections. Washington, DC: U.S. Department of Commerce, Bureau of the Census; Series P-25, No. 1057, 1990.
- ↵National Center for Health Statistics. National Health and Nutrition Examination Survey, 1976–1980 [public use data tapes]. Hyattsville, MD: U.S. Department of Health and Human Services; 1990. Tape Nos. 5305 and 5301 and catalog No. 5411.
- ↵National Center for Health Statistics. Vital and Health Statistics. Smoking and Other Tobacco Use: United States, 1987. Hyattsville, MD: U.S. Department of Health and Human Services, Public Health Service, National Center for Health Statistics; National Health Survey Series 10, No. 169, PHS 89-1597, 1989.
- National Health Interview Survey, United States, 1986. Hyattsville, MD: U.S. Department of Health and Human Services, Public Health Service, National Center for Health Statistics; Maryland National Health Survey Series 10, No. 164, PHS 87-1592, 1987.
- ↵Department of Health and Human Services, 1987. The Framingham Study: an epidemiological investigation of cardiovascular disease. Some risk factors related to the annual incidence of cardiovascular disease and death using pooled repeated biennial measurements: Framingham Heart Study, 30-year follow-up. Section 34, NIH Publication 87-2703. Bethesda, MD: National Heart, Lung, and Blood Institute.
- National Center for Health Statistics
- Fryback D.G,
- Dasbach E.J,
- Klein R,
- et al.
- Jacobs D,
- Blackburn H,
- Higgins M,
- et al.
- Sytkowski P.A,
- D’Agostino R.B,
- Belandger A.J,
- Kannel W.B
- Edelson J.T,
- Weinstein M.C,
- Tosteson A.N.A,
- et al.
- Hulley S.B,
- Walsh J.M,
- Newman T.B
- Stafford R.S,
- Blumenthal D,
- Pasternak R.C
- Goldman L,
- Weinstein M.C,
- Williams L.W
- Expert Panel on Population Strategies for Blood Cholesterol Reduction
- Tosteson A.N.A,
- Weinstein M.C,
- Hunink M.G.M,
- et al.