• Dedication to     Ancel Keys  • Introduction  • Overview  • Yugoslavia  • Italy  • Greece  • Finland  • Netherlands  • U.S.A.  • Japan  • Addendum    - Gate 27    - MN Code    - Main Results    - Anecdotes    - CIA    - Low-Coronary Risk      Male

Study Coordination and Main Results

Main Results of the Seven Countries Study - Feasibility

Prior to the Seven Countries Study no one had seriously attempted to compare cardiovascular disease frequency and risk factors in a systematic manner between defined populations and across a wide range of variation in lifestyle and disease rates. The study demonstrated that with good will, careful planning, and hard work, field studies could be mounted, leadership and collaboration found, funding obtained, and the research carried out despite difficulties.

Cross-sectional differences

The study was the first to establish credible data on prevalence rates of heart attack and stroke in contrasting cultures. Differences were found on the order of five- to ten-fold in coronary heart disease. Moreover, the study was the first to document population differences in the average levels and distributions of risk factors. The more dramatic example was the almost non-overlapping values for serum cholesterol between men of the same age in Japan and eastern Finland during the 1960s. The study also demonstrated large differences in composition of the diet in the otherwise similar, stable, rural, agricultural or pastoral populations, varying between three percent and twenty-two percent for saturated fatty acids and nine percent to forty percent for total fat calories.

Differences on follow-up

The Seven Countries Study was the first to establish credible data on coronary disease incidence rates in contrasting populations, again with differences found on the order of five- to eight-fold. In twenty years' follow-up there were favorable cardiovascular and all-cause death rates in Greece, Japan, and Italy compared with the other areas.

The study was the first to demonstrate the remarkable (ecologic) correlations among average diet and other lifestyle factors, risk factor distributions, and heart disease rates. The strongest correlation was between coronary disease incidence and average saturated fatty acid intake or serum cholesterol levels.

Low coronary death rates in a population were not compensated by an excess of other causes of death. In fact, all-causes death rate reflected rather well the death rates from heart attacks in the first ten years of the study.

The importance of blood pressure and serum cholesterol in individual prediction of coronary risk was borne out in all cultures, whereas there were only weak relationships with physical activity, body weight and smoking habits.

Coronary death rates in the cohorts correlated highly with WHO vital statistics on coronary deaths (r=0.98) and all-causes death rates (r=0.86) nationally, indicating wide generalizability of study findings to regional and national experience.

In longer-term follow-up, only the lower levels of body mass index were associated with average age at death. Otherwise, there was no relationship of weight to longevity throughout the distribution of body mass.

The remarkable finding that cigarette smoking was a minor risk factor for coronary and all-causes death in Italy, Greece and Japan, was in part due to the few coronary events. In other countries, there was a strong individual relationship of smoking to cardiovascular and to non-cardiovascular deaths. The electrocardiogram and resting heart rate were both strongly related to ten-year all-causes death rates, as was timed vital capacity of the lungs.

The variation within individuals in composition of the diet was of the same order as the variation between individuals, demonstrating that diet cannot be correlated with disease under such conditions.

Multivariate analysis

Only two entry risk factors were significantly related to differences in heart attack incidence among cohorts: blood cholesterol accounted for more than forty percent of the variation in coronary death rate, while blood pressure and cholesterol together accounted for sixty percent.

The risk of deaths from all-causes in ten years was least for men above average body weight and skinfold thickness. Both habitual physical activity and resting heart rate were significant predictors of death and coronary death in the European groups. Systolic and diastolic pressure were both significant risk factors for coronary and total deaths when adjusted for all other risks. Prediction of coronary death was far superior to prediction of all-causes death.

Multivariate solutions from experience in one area predicted well the relative individual coronary risk in other areas. There were, however, large differences in absolute risk. The solution for southern Europe, for example, greatly underestimated the absolute risk of coronary deaths in northern Europe. This suggests that measurements were not representative of long-term characteristics of men in the different areas at entry, or that unidentified factors unrelated to those considered in the study contributed to the increased risk of North Americans and northern Europeans, or were protective in southern Europeans and the Japanese.

A major focus of current analyses is on the effects of the very significant average increases in some risk factors, particularly serum cholesterol, in the first ten years follow-up of cohorts, a rise that was more marked in southern than northern Europe, and in younger than older men, and was partly explained by dietary changes.

Population correlations

The Seven Countries Study was the first to compute population (ecologic) correlations between risk factors and disease incidence, demonstrating significant population correlations as well as population thresholds for fatty-artery diseases. It established that population death rates from coronary heart disease can be predicted precisely by knowledge of the average serum cholesterol. It also demonstrated remarkable departures from prediction for such populations as east Finland where heart attack rates were greater than predicted by mean cholesterol values, and the island of Crete, where the rates were less than predicted, thus opening important new questions about causation.

The study was the first to apply partial correlation coefficients derived from relationships between risk and disease found in one country or group of countries to those in another. This showed the universality of risk factors as predictors of an individual's relative risk of coronary disease in contrasting cultures. But the study was also the first to demonstrate the different force of a risk factor in populations than in individuals. For example, it found the slope of individual risk factor relationships to heart attacks approximately twice as great in the United States as in Europe, and in northern as in southern Europe. This difference lessened as the follow-up was extended to thirty years.

Risk factor changes

The study was among the first to demonstrate dramatic changes in a relatively short time, in both directions, in average levels and distributions of risk characteristics for whole populations. This confirms the overwhelming role of culture and environment in determining the differences, and the changes, in disease risk, especially cardiovascular diseases. It was among the first studies to demonstrate the predictive importance of change in average risk characteristics, particularly serum cholesterol and blood pressure, in the population risk of heart attack and stroke.

Long-term prediction

The study was one of the first to examine the relationship between baseline characteristics during health and subsequent longevity, variously defined as survival for twenty five years or to age 75 or 85. In this instance, cigarette smoking was the main contributor, with a major contribution also from blood pressure, but very little from serum cholesterol and body mass index.

The study has consistently illustrated the complexity of the relationship among body weight, body mass, obesity, and disease rates. The shape of the relation was different among cultures, and it was counter-intuitive, in that longterm survival showed no relationship throughout the distribution of body weight, except for the excess risk of being in the lowest class of body mass. Moreover, survival was greater in those who gained weight in middle age than in those who did not gain or who lost weight.

The study was the first to demonstrate in population correlations a strong inverse relationship of monounsaturated fatty acids (olive oil) in the diet to coronary, cancer and all-cause mortality.

The twenty-year follow-up revealed that eighty-one percent of the difference among populations in coronary deaths could be explained by average saturated fatty acid intake. Absolute and relative increase in coronary mortality over time was greatest in the Serbian cohorts, largely explained by major increases in their average serum cholesterol levels.

Public health implications

The major result of the Seven Countries Study has been strengthening of the concept of population causes and mass phenomena involved in the genesis of mass coronary disease, hypertension and stroke. It has contributed to the idea that we are dealing with mass cultural phenomena influencing already widespread individual susceptibility, resulting, where environments are unfavorable, in heavy population rates of disease. This concept has played a central role in the wider population strategy of prevention and health promotion, complementing the traditional medical strategy among high risk individuals. It has stimulated research on population causes, and on community-wide preventive strategies, which now characterize much ongoing research in the epidemiology and prevention of heart and blood vessel diseases. It has provided a sound scientific basis for public health policy on prevention.

continued...