[Skip to Navigation]
Sign In
JAMA Classics
September 17, 2008

The Multiple Risk Factor Intervention Trial (MRFIT)—Importance Then and Now

Author Affiliations

Author Affiliations: Department of Preventive Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois (Dr Stamler); Department of Biostatistics, School of Public Health, University of Minnesota, Minneapolis (Dr Neaton).

Author Affiliations: Department of Preventive Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois (Dr Stamler); Department of Biostatistics, School of Public Health, University of Minnesota, Minneapolis (Dr Neaton).

JAMA. 2008;300(11):1343-1345. doi:10.1001/jama.300.11.1343

Is Relationship Between Serum Cholesterol and Risk of Premature Death From Coronary Heart Disease Continuous and Graded? Findings in 356 222 Primary Screenees of the Multiple Risk Factor Intervention Trial (MRFIT)


The 356 222 men aged 35 to 57 years, who were free of a history of hospitalization for myocardial infarction, screened by the Multiple Risk Factor Intervention Trial (MRFIT) in its recruitment effort, constitute the largest cohort with standardized serum cholesterol measurements and long-term mortality follow-up. For each five-year age group, the relationship between serum cholesterol and coronary heart disease (CHD) death rate was continuous, graded, and strong. For the entire group aged 35 to 57 years at entry, the age-adjusted risks of CHD death in cholesterol quintiles 2 through 5 (182 to 202, 203 to 220, 221 to 244, and ≥245 mg/dL [4.71 to 5.22, 5.25 to 5.69, 5.72 to 6.31, and ≥6.34 mmol/L]) relative to the lowest quintile were 1.29, 1.73, 2.21, and 3.42. Of all CHD deaths, 46% were estimated to be excess deaths attributable to serum cholesterol levels 180 mg/dL or greater (≥4.65 mmol/L), with almost half the excess deaths in serum cholesterol quintiles 2 through 4. The pattern of a continuous, graded, strong relationship between serum cholesterol and six-year age-adjusted CHD death rate prevailed for nonhypertensive nonsmokers, nonhypertensive smokers, hypertensive nonsmokers, and hypertensive smokers. These data of high precision show that the relationship between serum cholesterol and CHD is not a threshold one, with increased risk confined to the two highest quintiles, but rather is a continuously graded one that powerfully affects risk for the great majority of middle-aged American men.

See PDF for full text of the original JAMA article.


In 1986 in JAMA, we reported findings of the 6-year follow-up of the large cohort screened for the Multiple Risk Factor Intervention Trial (MRFIT).1 The article, challenging existing dogmas about the relationship between cholesterol and coronary heart disease (CHD), generated much interest. Now, the 25-year results are available with almost 7% (23 382) of the men deceased due to CHD.2-4 The 1986 findings and conclusions regarding the relationship of a single measurement of serum cholesterol to premature CHD mortality are verified in depth by the 25-year data with extraordinary precision due to large size of the cohort, long follow-up, and a large number of CHD deaths.

The main finding of our report was that the relationship between serum cholesterol and CHD mortality is continuous, graded, and strong; ie, CHD risk is progressively higher at every cholesterol level from 160 mg/dL and higher levels, with no threshold. This finding prevails with 5-, 10-, 15-, 20-, and 25-year follow-up and for the first, second, third, fourth, and fifth 5-year follow-up periods. These robust results, controlled for age, systolic blood pressure, number of cigarettes smoked per day, diabetes status, race and ethnicity, and study geographic site, prevailed over the 25-year follow-up with only modest attenuation in quantitative strength of relative risk from higher serum cholesterol levels and with increase over time in absolute excess risk from higher serum cholesterol levels as the CHD death rate increased annually.

These findings also held for men of every age 35 to 39, 40 to 44, 45 to 49, 50 to 54, and 55 to 57 years); race and ethnicity (African American, Asian American, Hispanic American, and non-Hispanic white American); lower and higher income strata across the 22 MRFIT centers in 18 US cities; cigarette smokers and nonsmokers; normotensive, prehypertensive, and hypertensive participants; nondiabetic and diabetic participants; and for men stratified into 6 subgroups based on blood pressure and cigarette smoking status; also for the separate cohort of 5362 men with a history of prior myocardial infarction—37% of whom died from CHD.

Data Regarding Other Cohorts

The 1986 findings and conclusions have also been validated for women and men by data from many other prospective studies, eg, on young adult and middle-aged Chicago residents (>35 000 individuals observed for >30 years)2 and on 61 cohorts worldwide (≈900 000 individuals) observed for an average of 13 years (33 744 CHD deaths [3.7%]) in an Oxford University meta-analysis.5 For the 61 cohorts combined, the relationship between a single serum cholesterol level and CHD mortality, ie, continuous, graded, and strong, was quantitatively similar for men in the MRFIT cohort and men and women of all the other cohorts. The cholesterol and CHD relationship prevailed across geographic locations (on 4 continents), at all blood pressure levels, for smokers and nonsmokers, and across body mass index (BMI) strata (MRFIT lacked BMI data). This further information set is especially relevant given the worldwide obesity epidemic, the consequent unprecedented high prevalence rates of overweight/obesity, and its adverse effects on serum cholesterol and other metabolic CHD risk factors (eg, blood pressure, glycemia/diabetes). With an apparent focus on the potential for CHD prevention and control, the Oxford University report highlighted relative risks with estimated “usual” serum cholesterol lower by approximately 40 mg/dL: CHD risk approximately one-half lower in early middle age (40-49 years), one-third lower in later middle age (50-69 years), and one-sixth lower in older age (70-89 years). Although relative risk was less extreme with older age, absolute excess risk was greater.5

Clearly, these fundamental findings quantitating the relationship between serum cholesterol and CHD are generalizable populationwide. The depth, breadth, and consistency of these findings reflect the fact, recognized throughout the research, public health, medical, and public policy communities, that this is an etiologically significant relationship.

Myths and Dogmas Refuted

The 1986 JAMA article1 helped refute several dogmas and myths that were once influential, eg, that the relationship between serum cholesterol and CHD is a threshold one—with greater risk only at cholesterol levels equal to or greater than 240, 250, or 260 mg/dL; that proper cut point for abnormal serum cholesterol is therefore 240, 250, 260, or even 300 mg/dL; and that serum cholesterol and the other readily measured major CHD risk factors (blood pressure, smoking, diabetes, overweight/obesity) account for no more than 50% of CHD events.6

In presenting MRFIT data rebutting these notions and explicitly rejecting them, the 1986 article1 helped strengthen scientific foundations for the efforts to prevent, control, and eradicate the CHD epidemic. The MRFIT data were powerful underpinnings for the clinical serum cholesterol classification of the National Cholesterol Education Program2,7,8: for adults, favorable levels are denoted as less than 200 mg/dL; borderline high levels as 200 to 239 mg/dL; and high levels as 240 mg/dL or greater.

Low Coronary Heart Disease Risk

The 1986 MRFIT report for the first time put forward the concept of optimal or low CHD risk, gave a first set of criteria for its definition, emphasized its rarity among US adults, and also emphasized its benefits.1 For participants in the MRFIT cohort without a history of myocardial infarction (N = 356 222), low risk was defined as all of the 5 following criteria: optimal level of serum cholesterol and systolic and diastolic blood pressure, nonsmoking status, and no history of treatment for diabetes. Only 2% of the men in the MRFIT cohort met these criteria, only 6 of these men died from CHD during the 6-year follow-up, and the CHD death rate was 87% lower than for the rest of the cohort.

Correspondingly, among the Chicago Heart Association middle-aged cohort only 2% (men) and 5% (women) were at low risk based on all 6 of the following criteria2: serum cholesterol lower than 200 mg/dL, systolic blood pressure 120 mm Hg or lower, diastolic blood pressure 80 mm Hg or lower, no smoking, no diabetes, and BMI lower than 25.0. For both of these subcohorts and the similarly defined MRFIT low-risk subcohort, the 25- to 30-year CHD mortality rate was lower by 69% to 82% compared with the corresponding rate for all other individuals; the all-cause mortality rate was lower by 52% to 59%; and estimated longevity was greater by 6 to 7 years.2

From 1986 to the present, the findings on low risk have informed public policy on the strategy for ending the CHD epidemic: because for low-risk individuals, CHD ceases to be epidemic and because relatively few individuals are at low risk, vital strategic challenges, tasks, and priorities for medical care and public health are to achieve steady, progressive, and sustained increases year by year in the proportion of all population strata at low risk.

“Disturbances of Human Culture”: Causes of Disease Epidemics

Data are extensive regarding what needs to be done to help most adults become low risk for CHD. The essentials derive from a basic law of medicine and public health: epidemics are, as set down by Virchow, due to “ . . . disturbances of human culture.”9 The first and foremost of the crucial disturbances producing epidemic rates of major CHD risk factors and CHD is populationwide adverse dietary patterns, along with cigarette smoking and sedentary lifestyle at work and leisure. The diets—high in caloric density, total fat, cholesterol, and saturated and trans fats (from fat- and cholesterol-laden red meats, dairy products, egg yolks, visible fats, and commercial baked goods); high in salt and processed sugars; for some, excessive in alcohol intake; and for all too many, relatively inadequate/low in key micro- and macronutrients from vegetables, fruits, whole grains, and legumes (eg, calcium, iron, magnesium, phosphorus, potassium, antioxidant and other vitamins, fiber, vegetable protein, and mono- and polyunsaturated fats)—account for the epidemic occurrences of adverse levels of serum cholesterol, blood pressure, and other metabolic CHD risk factors.

In 2008, the role of high dietary cholesterol intake needs to be emphasized for several reasons: first, high cholesterol intake significantly influences serum cholesterol level2,10-12; second, high cholesterol intake relates independently to CHD risk over and above its role in increasing serum cholesterol levels2; third, feeding cholesterol (eg, from egg yolks) is and has been since 190813 the sine qua non for experimental production of atherosclerosis in laboratory animals, including nonhuman primates (ie, for replicating the human lesion underlying the CHD epidemic); and fourth, sustained commercial propaganda seeks to obfuscate these facts.

Public Policy and Popular Response

The role of adverse eating patterns as key causes of the CHD epidemic (especially the diet and serum cholesterol relationship) was largely delineated in the 1950s and 1960s and recommendations to modify and improve lifestyles were addressed both to the whole population (generally at risk) and to the sizable strata at higher risk (the 2-pronged strategy).1 The population responded by showing substantial improvements (albeit that fell short of national goals): eg, average intake levels of total fat (as percentage of total kilocalories) decreased from approximately 40% to 45% down to 32%; cholesterol from approximately 700 to 320 mg per day; and percentage of kilocalories from saturated fats from approximately 17% to 12%. Predictably (based on metabolic ward data that enabled precise estimates),10-12 population average serum cholesterol level declined considerably despite the countervailing influence of the obesity epidemic. From a 1950s/1960s level of approximately 235 to 240 mg/dL, it reached 200 mg/dL by the year 2000, a decline that predated mass statin use and that was undoubtedly due to improved dietary composition. A national health goal was achieved, but regrettably, little heralded. Over the decades from 1960 to 1990, the proportion of the population at low CHD risk also increased modestly but then decreased (Paul D. Sorlie, PhD, and Teri A. Manolio, MD, PhD, personal communication, July 2008); this remains a major clinical and public health concern.


Much progress has been made since 1948 when one of us (J.S.) began a research career studying classic texts averring that dietary factors had no influence on human serum cholesterol levels. Over the decades, epidemiologic, metabolic ward, animal experimental, clinical trial, anthropologic, and other research modalities have produced extensive concordant knowledge on the disturbances of human culture, first and foremost dietary—that caused epidemic CHD. The MRFIT findings have made an extraordinary contribution. They demonstrate the power of large numbers and hard clinical end points to illuminate public policy. Collected with years-long support from the National Heart, Lung, and Blood Institute, the MRFIT results have been and will continue to be an important national resource for informing public health policy.

The crucial scientific findings to end the CHD epidemic are now available. The challenge and task is to apply them in all appropriate patient contacts and across all population strata to extend the progress to date. One key for achieving this is priority emphasis on primordial prevention, ie, family dedication to favorable lifestyles (nutrition, exercise, nonsmoking) as norms of human behavior. This especially applies to the future mother so that when she conceives and throughout pregnancy, her exposures and those of her fetus are optimal and become primary and lifelong habits for the newborn infant and preschool child. Healthy eating patterns such as the DASH diet are available as models for this crucial aspect of disease.2,14,15

Back to top
Article Information

Corresponding Author: Jeremiah Stamler, MD, Department of Preventive Medicine, Feinberg School of Medicine, Northwestern University, 680 N Lake Shore Dr, Ste 1102-D335, Chicago, IL 60611-4402 (j-stamler@northwestern.edu).

Financial Disclosure: None reported.

Funding/Support: This study for the 25-year follow-up of the MRFIT cohort was supported by the sustained contract and grant support from the National Heart, Lung, and Blood Institute, which includes research grant RO1-HL68140.

Additional Information: We express appreciation to the many colleagues who participated in the MRFIT research effort for their years-long contributions; some of these colleagues are cited by name in the 1986 article.1 This article was completed on July 30, 2008, the day the New York Times published the obituary of Julius B. Richmond, MD, a vigorous antismoking advocate and founder of Project Head Start, who in 1979 as US Surgeon General put forward the first set of health goals for the United States. We dedicate this Commentary to Dr Richmond and hope it will help individuals throughout the United States make further improvements in lifestyles and nutrition; eg, to achieve the further health goal of a national adult average serum cholesterol level of 180 mg/dL by 2020—without drug treatment.

Stamler J, Wentworth D, Neaton JD. Is relationship between serum cholesterol and risk of premature death from coronary heart disease continuous and graded?  JAMA. 1986;256(20):2823-28283773199PubMedGoogle ScholarCrossref
Stamler J, Neaton JD, Garside DB, Daviglus M. Current status: six established major risk factors—and low risk. In: Marmot M, Elliott P, eds. Coronary Heart Disease Epidemiology: From Aetiology to Public Health. 2nd ed. London, England: Oxford University Press; 2005:32-70
Stamler J, Neaton JD, Garside DB, Daviglus ML. The major adult cardiovascular diseases: a global historical perspective. In: Lauer R, Burns TL, Daniels RS, eds. Pediatric Prevention of Atherosclerotic Cardiovascular Disease. London, England: Oxford University Press; 2006:27-48
Stamler J, Daviglus ML, Garside DB,  et al.  Low-risk cardiovascular status: impact on cardiovascular mortality and longevity. In: Lauer R, Burns TL, Daniels RS, eds. Pediatric Prevention of Atherosclerotic Cardiovascular Disease. London, England: Oxford University Press; 2006:49-60
Prospective Studies Collaboration. Lewington S, Whitlock G, Clarke R,  et al.  Blood cholesterol and vascular mortality by age, sex, and blood pressure.  Lancet. 2007;370(9602):1829-183918061058PubMedGoogle ScholarCrossref
Stamler J. Low risk—and the “no more than 50%” myth/dogma.  Arch Intern Med. 2007;167(6):537-53917389284PubMedGoogle ScholarCrossref
 Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults: the Expert Panel.  Arch Intern Med. 1988;148(1):36-693422148PubMedGoogle ScholarCrossref
National Cholesterol Education Program; Expert Panel on Population Strategies for Blood Cholesterol Reduction.  Report of the Expert Panel on Population Strategies for Blood Cholesterol Reduction. Bethesda, MD: US Department of Health and Human Services, Public Health Service; 1990. National Institutes of Health Publication 90-3046
Ackerknecht EH. Rudolf Virchow: Doctor, Statesman, Anthropologist. Madison: University of Wisconsin Press; 1953
Hegsted DM, Austman LM, Johnson JA, Dallal GE. Dietary fat and serum lipids.  Am J Clin Nutr. 1993;57(6):875-8838503356PubMedGoogle Scholar
Clarke R, Frost C, Collins R,  et al.  Dietary lipids and blood cholesterol.  BMJ. 1997;314(7074):112-1179006469PubMedGoogle ScholarCrossref
Food and Nutrition Board.  Cholesterol. In: Food and Nutrition Board. Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids. Washington, DC: National Academies Press; 2005:542-588
Anitschkow N. Experimental arteriosclerosis in animals. In: Cowdry EV, ed. Arteriosclerosis: A Survey of the Problem. New York, NY: Macmillan; 1933:271-322
Sacks FM, Svetkey LP, Vollmer WM,  et al.  Effects on blood pressure of reduced sodium and the Dietary Approaches to Stop Hypertension (DASH) diet.  N Engl J Med. 2001;344(1):3-1011136953PubMedGoogle ScholarCrossref
Appel LJ, Sacks FM, Carey VJ,  et al.  Effects of protein, monounsaturated fat, and carbohydrate intake on blood pressure and serum lipids: results of the OmniHeart randomized trial.  JAMA. 2005;294(19):2455-246416287956PubMedGoogle ScholarCrossref