Stamler J, Stamler R, Neaton JD, Wentworth D, Daviglus ML, Garside D, Dyer AR, Liu K, Greenland P. Low Risk-Factor Profile and Long-term Cardiovascular and Noncardiovascular Mortality and Life ExpectancyFindings for 5 Large Cohorts of Young Adult and Middle-Aged Men and Women. JAMA. 1999;282(21):2012-2018. doi:10.1001/jama.282.21.2012
Author Affiliations: Department of Preventive Medicine, Northwestern University Medical School, Chicago, Ill (Drs Stamler, Daviglus, Dyer, Liu, and Greenland, Ms Stamler, and Mr Garside); and Division of Biostatistics, School of Public Health, University of Minnesota, Minneapolis (Dr Neaton and Ms Wentworth). Ms Stamler died February 28, 1998.
Context Three major coronary risk factors—serum cholesterol level, blood
pressure, and smoking—increase incidence of coronary heart disease (CHD)
and related end points. In previous investigations, risks for low-risk reference
groups were estimated statistically because samples contained too few such
people to measure risk.
Objective To measure long-term mortality rates for individuals with favorable
levels for all 3 major risk factors, compared with others.
Design Two prospective studies, involving 5 cohorts based on age and sex, that
enrolled persons with a range of risk factors. Low risk was defined as serum
cholesterol level less than 5.17 mmol/L (<200 mg/dL), blood pressure less
than or equal to120/80 mm Hg, and no current cigarette smoking. All persons
with a history of diabetes, myocardial infarction (MI), or, in 3 of 5 cohorts,
electrocardiogram (ECG) abnormalities, were excluded.
Setting and Participants In 18 US cities, a total of 72,144 men aged 35 through 39 years and
270,671 men aged 40 through 57 years screened (1973-1975) for the Multiple
Risk Factor Intervention Trial (MRFIT); in Chicago, a total of 10,025 men
aged 18 through 39 years, 7490 men aged 40 through 59 years, and 6229 women
aged 40 through 59 years screened (1967-1973) for the Chicago Heart Association
Detection Project in Industry (CHA) (N = 366,559).
Main Outcome Measures Cause-specific mortality during 16 (MRFIT) and 22 (CHA) years, relative
risks (RRs) of death, and estimated greater life expectancy, comparing low-risk
subcohorts vs others by age strata.
Results Low-risk persons comprised only 4.8% to 9.9% of the cohorts. All 5 low-risk
groups experienced significantly and markedly lower CHD and cardiovascular
disease death rates than those who had elevated cholesterol level, or blood
pressure, or smoked. For example, age-adjusted RRs of CHD mortality ranged
from 0.08 for CHA men aged 18 to 39 years to 0.23 for CHA men aged 40 through
59 years. The age-adjusted relative risks (RRs) for all cardiovascular disease
mortality ranged from 0.15 for MRFIT men aged 35 through 39 years to 0.28
for CHA men aged 40 through 59 years. The age-adjusted RR for all-cause mortality
rate ranged from 0.42 for CHA men aged 40 through 59 years to 0.60 for CHA
women aged 40 through 59 years. Estimated greater life expectancy for low-risk
groups ranged from 5.8 years for CHA women aged 40 through 59 years to 9.5
years for CHA men aged 18 through 39 years.
Conclusions Based on these very large cohort studies, for individuals with favorable
levels of cholesterol and blood pressure who do not smoke and do not have
diabetes, MI, or ECG abnormalities, long-term mortality is much lower and
longevity is much greater. A substantial increase in the proportion of the
population at lifetime low risk could contribute decisively to ending the
Long-term, population-based, prospective studies have amassed extensive
data on relationships of major coronary-cardiovascular risk factors—particularly
serum cholesterol level, blood pressure, and cigarette smoking—with
incidence of coronary heart disease (CHD), stroke, and cardiovascular disease
(CVD), to mortality from these causes and all causes and longevity.1- 7
These relationships have been well summarized as " . . . strong, continuous,
graded, consistent, independent, predictive, and etiologically significant
for those with and without coronary heart disease."7
The judgment on etiologic significance is based on the consistent results
of many epidemiological studies and on concordant findings from clinical and
postmortem investigations and animal experimentation. This judgment is reinforced
by data from randomized controlled trials demonstrating that sustained lowering
of high blood pressure or elevated serum cholesterol level produces sizable
reductions in CHD-CVD incidence and in cause-specific and all-cause mortality.7- 14
These positive results have been obtained repeatedly, even though trials have
been undertaken in middle-aged and older people after decades of exposure
to these adverse traits. Extensive data also document that smoking cessation
has similar favorable effects.15,16
Most epidemiologic research on the impact of major risk factors deals
with the predictive value of higher levels of such factors. In assessments
of their combined impact, risks of those with favorable status for all 3 major
risk factors have been estimated statistically, for example, by extrapolation
down multiple logistic smoothed curves.3 This
was necessary because in the population samples studied, numbering in the
hundreds or thousands, too few people had low levels of all major risk factors—hence
too few CVD events—to permit direct measurement of risk.
Large, long-term studies permit measured estimates based on actual observed
mortality. In this article, we use data on 5 cohorts from 2 studies, the Multiple
Risk Factor Intervention Trial (MRFIT) and the Chicago Heart Association Detection
Project in Industry (CHA): 2 cohorts of young adult men, 2 cohorts of middle-aged
men, and 1 cohort of middle-aged women—366,559 people all together.
Published reports on the MRFIT and CHA cohorts detail their baseline
screening methods.5,6,17- 19
We provide a summary of these here.
All together, 361,662 men aged 35 through 57 years were screened in
1973-1975 at 22 centers in 18 US cities for recruitment for MRFIT. The 342,815
men with complete baseline risk factor data are the focus here, stratified
into 2 cohorts: those aged 35 through 39 years (n = 72,144) and 40 through
57 years (n = 270,671). Trial eligibility was based on a man's major risk
factor profile; therefore, initial screening included measurements only of
blood pressure and serum cholesterol level; current smoking (by questionnaire),
including number of cigarettes per day; and conditions for exclusion, ie,
drug treatment for diabetes and previous hospitalization for myocardial infarction
(MI). Blood pressure was measured according to a standardized protocol by
trained certified staff, using a mercury sphygmomanometer, with the man seated.
Diastolic blood pressure (DBP) was measured at the fifth Korotkoff sound.
Three readings per individual were taken; the average of the second and third
systolic blood pressure (SBP) measurements was used for analyses. Serum total
cholesterol level was determined, in 15 standardized local laboratories, by
the Lieberman-Burchard color reaction and use of serum calibrators to yield
values equivalent to Abell-Kendall reference values.17- 19
Vital status of the men is ascertained periodically through the US National
Death Index. Prior to 1979, Social Security Administration records were used.
With a mean follow-up of 16 years, 38,265 deaths have been identified; cause
of death is known for 98.9% of decedents. Underlying cause of death was coded
by a nosologist using the International Classification of
Diseases, Ninth Revision (ICD-9).20
Employees of 84 Chicago-area companies and organizations, about 75,000
people, were invited to participate. The response rate was 55%. Screening
was done by 2 trained and standardized 4-person field teams who collected
demographic information, medical history, and medical treatment data; information
on past and present smoking status; 1 measurement of height, weight, heart
rate, and supine blood pressure; resting electrocardiogram (ECG); and venipuncture
for blood chemistry measurements. Serum total cholesterol level was determined
by the Levine and Zak method.6,17
The criteria of the Pooling Project3 were used
to code ECG abnormalities. Three cohorts are the focus here: men aged 18 through
39 years, men aged 40 through 59 years, and women aged 40 through 59 years.
Methods of follow-up to ascertain vital status include local procedures
and use of Social Security Administration and National Death Index records.
With a mean follow-up of 22 years, vital status has been determined for more
than 99% of the cohorts. For each decedent, underlying cause of death was
coded by a trained staff professional, using the International
Classification of Diseases, Eighth Revision (ICD-8 ).21
Criteria for defining a person as low risk were all of the following
at baseline: serum cholesterol level less than 5.17 mmol/L (<200 mg/dL),
SBP/DBP of 120/80 mm Hg or lower; not a current smoker; no history of diabetes
or MI; and, for the 3 CHA cohorts, no ECG abnormalities.
Deaths from all CHDs were defined for MRFIT cohorts as ICD-9 codes 410 through 414 and 429.9, for CHA cohorts as ICD-8 codes 410 through 414; MI, code 410; stroke, codes 430 through
438; all CVD, codes 390 through 459; all cancers, codes 140 through 209; violence,
for MRFIT cohorts ICD-9 codes 800 through 999, for
CHA cohorts ICD-8 codes E800 through E999 exclusive
of codes E930 through E936. Coders were blinded to baseline data.
To focus on risk for persons with favorable levels of serum cholesterol,
blood pressure, and no tobacco use (all 3 combined), compared with persons
with adverse levels of 1 or more of these, persons with histories of diabetes
or MI were excluded (all 5 cohorts), as were persons with ECG abnormalities
(the 3 CHA cohorts). Mortality rates for low-risk and other persons were age-adjusted
by the direct method to the age distribution of all persons in an age stratum.
Cox proportional hazards regression was used to calculate age-adjusted relative
risks (RRs) and their 95% confidence intervals (CIs) for low-risk compared
with other persons.
Cox multivariate proportional hazards regression was used to calculate
coefficients for the relation of baseline major risk factors to all-cause
mortality for each cohort. Coefficients were used to estimate number of years
of greater life expectancy for each low-risk subcohort compared with other
persons of the same cohort.17 Thus, the coefficient
for the relationship of SBP to all-cause mortality in the Cox multivariate
analyses for CHA men aged 18 through 39 years is 0.0116. Average SBP for the
942 low-risk men was 116.0 mm Hg; for the 9083 other men, 136.0 mm Hg; by
exponentiation, estimated RR of death is e−0.0116×20
= e−0.232 = 0.793. To estimate impact of this lower SBP on
life expectancy, we used the concomitant Cox coefficient for the relationship
of age to all-cause mortality, 0.0703. The product for SBP exponentiation,
0.0116 × 20 = 0.232, is also obtained when the coefficient for age,
0.0703, is multiplied by 3.3, which indicates that SBP of 116 mm Hg vs 136
mm Hg is equivalent to being, on average, 3.3 years younger: eg, age 26.7
years rather than age 30 years. From US life tables,22
male expectation of life at age 30 years is 44.1 years; at age 26.7 years,
47.2 years: ie, 3.1 years estimated greater life expectancy is attributable
to SBP 116 mm Hg vs SBP 136 mm Hg. Similar calculations yield data on impact
on life expectancy of favorable status of the low-risk subcohort for serum
cholesterol level and smoking compared with the other subcohort. These 3 estimates
are summed to give the overall estimate presented here.
The proportion of persons meeting low-risk criteria was small: for young
adult men, 9.9% (MRFIT) and 9.4% (CHA); for middle-aged men, 6.0% (MRFIT)
and 4.8% (CHA); and for middle-aged women, 6.8% (CHA) (Table 1 and Table 2).
In accordance with low-risk criteria, average blood pressure and serum cholesterol
levels were much lower for low-risk subcohorts compared with other persons.
Body mass index was lower for CHA low-risk subcohorts compared with others
The CHD mortality rate was much lower for low-risk subcohorts than for others, by 86% to 92% for
low-risk young adult men (<40 years) and 77% to 79% for low-risk middle-aged
subcohorts (Table 3). Findings
were similar for death attributed to acute MI.
For low-risk subcohorts, CHD death accounted for a much smaller proportion
of all death than for others (Table 3).
This finding was especially prominent for low-risk young adult men, with CHD
mortality only 6% to 8% of all mortality vs 25% to 29% for others.
All CVD mortality was much lower for low-risk subcohorts than for others
by 72% to 85% (Table 3).
There were no stroke deaths in the 2 young adult low-risk subcohorts.
For the 2 middle-aged, male low-risk subcohorts, stroke mortality was lower
than for others by 52% to 76%. Mortality from cancers was consistently lower
for low-risk subcohorts compared with others: by 44% to 56% for the 4 male
low-risk subcohorts and 17% for the female low-risk subcohort. No results
significantly supported the hypothesis that low serum cholesterol level is
associated with greater risk of violent death. For the 2 young adult cohorts,
mortality from all other causes was similar for low-risk men and others. For
the 3 middle-aged cohorts, RR was lower for low-risk groups than others by
36% to 86%(Table 4).
Mortality from all causes was consistently and markedly lower for low-risk
groups vs others: by 50% to 58% for men and 40% for women (Table 5). Estimated greater life expectancy for low-risk subcohorts
vs others ranged from 5.8 years to 9.5 years.
Large sample sizes and long follow-up of the 5 MRFIT and CHA cohorts
enabled measurement of actual cause-specific and all-cause mortality experience
of adults assessed to be low risk at baseline. Results were consistent qualitatively
and quantitatively for all 5 cohorts, young adult and middle-aged, male and
female, free at baseline of a history of diabetes and MI, and of ECG abnormalities
(CHA cohorts). Only a small minority (<10%) met all criteria for low risk—serum
cholesterol level under 5.17 mmol/L (<200 mg/dL), SBP/DBP of 120/80 mm
Hg or less, and no cigarette smoking. During long-term follow-up, low-risk
subcohorts, compared with others, consistently experienced significantly and
markedly lower CHD death rates by 77% to 92%, and CHD mortality was a much
smaller proportion of all-cause mortality. Findings for stroke and for all
CVD paralleled those for CHD. There was no evidence of significant countervailing
non-CVD mortality for low-risk subcohorts; rather, their cancer mortality
was consistently lower. Consequently, compared with others, all-cause mortality
was markedly lower for low-risk persons (by 40% to 58%), and their estimated
longevity was much greater (by 5.8 to 9.5 years).
These findings directly confirm earlier statistical estimates of the
benefits of low-risk status. For example, in the national cooperative Pooling
Project, risk of a first major coronary event was estimated by multiple logistic
regression to be lower by 70% for middle-aged men in the lowest quintile of
risk, compared with all other men.3 Concordantly,
recent data from the Framingham Study estimate CHD risk to be considerably
reduced for low-risk men and women compared with all men and women.23 Results for the 5 MRFIT and CHA cohorts go beyond
such estimates in several respects: (1) they are actual observations, not
extrapolations from regression analyses; (2) they are not only for middle-aged
men and women, but also young adult men; (3) they demonstrate the favorable
impact of low-risk status not only on CHD incidence, but also CHD mortality,
risk of fatal stroke, all CVD, all cancers, and all causes, without any significant
evidence of countervailing mortality risks; (4) they indicate that low-risk
status is associated with greater life expectancy by several years; and (5)
additional analyses (reported previously) on the CHA cohorts show further
that low risk in middle age is associated with lower average annual costs
for medical care in older age.24
These data on the benefits of low risk are almost certainly underestimates
due to both misclassification of individuals with a single measurement of
the 3 major risk factors25 and lack of data
on the fourth independent major risk factor, adverse dietary pattern.1,2,5,6,8,17,19,26
Data on participant exercise habits were also missing. In this regard, the
2 diet-dependent major risk factors (serum cholesterol level and blood pressure)
may be viewed not only as etiologically significant traits, but also as markers
of other lifestyle characteristics contributing to favorable outcomes for
low-risk subcohorts. This inference is supported by the CHA data showing lower
average body mass index for low-risk individuals vs others.
The data here challenge the view that the major risk factors " . . .
explain at most half of all myocardial infarctions." 27
Despite underestimation, favorable status for all 3 major risk factors consistently
predicted long-term CHD and MI mortality rates that are lower by much more
than 50%: by 86% to 92% for young adult men and 77% to 79% for middle-aged
persons compared with others. Consequently, for low-risk subcohorts, in contrast
to others, MI and CHD mortality rates were not at epidemic levels, were not
main causes of death, and did not account for a large proportion of all deaths.
Available data indicate that this favorable status for low-risk persons holds
for both African Americans and whites, and for those of lower and higher socioeconomic
These findings are relevant for the national effort to end the CHD-CVD
epidemic. They lend strong support to the concept2
that a strategy based on identifying, evaluating, and treating people with
risk factors is not enough. A population-wide strategy is critical to prevent
and reduce the magnitude of all the major risk factors, first and foremost
by safe nutritional-hygienic means, so that a substantial increase is achieved
in the proportion of people in the population who, throughout life, have favorable
levels for all the major risk factors and so are at low risk. For upcoming
generations, this means encouraging favorable behaviors beginning in early
childhood in regard to eating, drinking, exercising, and smoking. For others
(particularly older children, teenagers, and young adults), this strategy
emphasizes efforts to preserve favorable risk factor status for those who
still have none of the major risk factors.
Genetic makeup undoubtedly influenced propensity to fall into low-risk
categories. However, as shown by multiple data sets on groups such as American
Seventh Day Adventists, Chinese, Greeks, Italians, Japanese, and South Africans,
adult population average serum cholesterol level lower than 5.17 mmol/L (<200
mg/dL) is widely prevalent.1,2
For the US population as a whole in the 1990s, mean serum cholesterol level
has fallen almost to the national health goal for the year 2000 of no more
than 5.17 mmol/L (200 mg/dL).29 Similarly,
extensive data are available on isolated populations around the world with
average adult SBP/DBP of 120/80 mm Hg or less, with little or no blood pressure
rise during adulthood, and with little or no hypertension30:
favorable blood pressure patterns that are not due to unusual genetic makeup,
since with migration and adoption of modern lifestyles these populations too
develop adverse blood pressure levels.
Therefore, lifestyle also clearly influences who will fall into the
low risk-factor group. Since the 1960s, nutritional recommendations have been
available for prevention of dyslipidemia in the form of advice to decrease
intake of dietary total fat, saturated fat, cholesterol; partially replace
saturated fat with monounsaturated and polyunsaturated fat; increase intake
of dietary fiber, especially water-soluble fiber; and prevent or reduce overweight.1,2,8,19,26,29,31
Average serum cholesterol levels of the adult population have decreased from
approximately 6.21 mmol/L (240 mg/dL) to less than 5.30 mmol/L (205 mg/dL).29 More recently, lifestyle recommendations have been
set down for prevention of adverse blood pressure levels. These initially
involved avoidance of high salt intake, inadequate potassium intake, excess
alcohol use, overweight, and sedentary habits,30- 32
and have been expanded to include high intake of fruits and vegetables, fat-free
and low-fat protein sources, and low intake of lipid-rich foods (ie, reduced
dietary total fat, saturated fat, and cholesterol).7,33,34
National survey data indicate that average blood pressure levels of Americans
and rates of high blood pressure are lower as a result of improved lifestyles,
independent of effects of antihypertensive drug treatment.35
All these data support the concept that lifestyles, particularly nutritional
habits, interdigitate with polygenic propensities (widespread in the population)
to influence average serum lipid and blood pressure levels of both individuals
and the overall population. Adverse levels are not fixed consequences of the
genome; they are widely amenable to prevention by safe nutritional-hygienic
means, with resultant sizable increases in the proportion of the population
at low risk.
In summary, data from large, population-based, prospective studies indicate
that lifetime favorable status in regard to all 3 major CHD-CVD risk factors
(serum cholesterol level, blood pressure, and smoking) leads to low mortality
rates from CHD, CVD, and all causes and increased life expectancy. The extensive
findings support a strategic emphasis on population-wide primary prevention
of all major risk factors as a key component of the effort to end the CHD-CVD
epidemic. Research advances have supplied the scientific information to make
implementation of this strategic component widely feasible. The challenge
is to mobilize the societal will and resources to realize these goals in all
population strata to help end the CHD-CVD epidemic early in the next century.