Food intake patterns and 25-year mortality from coronary heart disease: Cross-cultural correlations in the Seven Countries Study
- Cite this article as:
- Menotti, A., Kromhout, D., Blackburn, H. et al. Eur J Epidemiol (1999) 15: 507. doi:10.1023/A:1007529206050
In the Seven Countries Study, associations between the intake of food-groups and 25-year mortality from coronary heart disease (CHD, defined as sudden coronary death or fatal myocardial infarction) were investigated. Baseline surveys were carried out between 1958 and 1964. A number of individual characteristics were measured in 12,763 middle-aged men belonging to 16 cohorts in seven countries (USA, Finland, The Netherlands, Italy, former Yugoslavia, Greece and Japan). Dietary information was collected in sub-samples using the weighed record method. Vital status of all participants was verified at regular intervals during 25 years of follow-up and the underlying cause of death was adjudicated. Eighteen different food-groups and combinations were considered for comparison among cohorts. Large differences in food-group consumption were seen, with high consumption of dairy products in Northern Europe, meat in the USA, vegetables, legumes, fish, and wine in Southern Europe, and cereals, soy products, and fish in Japan. Population death rates from CHD showed large differences, ranging from 268 per 1000 in East Finland to 25 per 1000 in Crete, Greece. Animal food-groups were directly correlated, and vegetable food-groups (except potatoes) as well as fish and alcohol were inversely correlated with CHD mortality. Univariate analysis showed significant positive correlation coefficients for butter (R = 0.887), meat (R = 0.645), pastries (R = 0.752), and milk (R = 0.600) consumption, and significant negative correlation coefficients for legumes (R = −0.822), oils (R = −0.571), and alcohol (R = −0.609) consumption. Combined vegetable foods (excluding alcohol) were inversely correlated (R = −0.519), whereas combined animal foods (excluding fish) were directly correlated (R = 0.798) with CHD death rates. Multivariate stepwise analysis selected butter, lard+margarine and meat as significant predictors and produced an R2 of 0.922. These findings were confirmed by factor analysis. These cross-cultural analyses are consistent with the hypothesis that dietary patterns are important determinants of differences in population CHD death rates, and confirm the opposite effects on apparent risk of animal and vegetable foods.