点击显示 收起
Department of Nutrition, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115, E-mail: frank.hu{at}channing.harvard.edu
Dear Sir:
Although dietary protein has been the focus of controversy regarding several popular diets, scientific data on the effects of protein intake on the development of chronic disease are limited. International studies suggest a positive correlation between animal protein intake and ischemic heart disease rates across countries (1). Therefore, we tested this hypothesis directly in the Nurses' Health Study (2). By analyzing repeated measures of dietary data over 14 y of follow-up, we firmly rejected the hypothesis that high protein intakes increase the risk of ischemic heart disease. In contrast, our data suggest a modest inverse association for both animal and vegetable protein intake. This finding is compatible with results of metabolic studies indicating improved blood lipid profiles when protein replaces carbohydrates (3). In the conclusion, we cautioned the application of these findings to public dietary advice because a high dietary protein intake is often accompanied by high saturated fat and cholesterol intakes.
Campbell questioned the validity of our findings because they contradict the results of international correlation studies on animal product consumption and disease rates. However, international correlations such as those cited by Campbell are intractably confounded by other dietary and lifestyle factors associated with economic affluence in different countries; differences in physical activity, body fat, and smoking are particularly important. Also, the food disappearance data used in most of the calculations may be more indicative of food wastage within a country than actual consumption. Correlational studies conducted within a country can usually provide more credible data than international comparisons because of relatively homogeneous populations and the possibility of collecting data on potential confounding variables at individual levels. A survey of 65 counties in rural China, however, did not find a clear association between animal product consumption and risk of heart disease or major cancers (4).
Prospective cohort studies of individuals, in which diet is assessed before the occurrence of disease, are typically considered to be the strongest nonrandomized design because it is possible to control for other known risk factors. One common misperception is that the dietary experience of a single population in a typical prospective cohort study is too homogeneous to detect associations with disease risk. In the Nurses' Health Study, however, we identified several important dietary factors for risk of ischemic heart disease, including trans fatty acids (5), the ratio of polyunsaturated to saturated fat (5), -linolenic acid (6), cereal fiber (7), nuts (8), whole-grain products (9), and fruit and vegetables (Kumudi unpublished observations, 2000).
Because high protein intakes are associated with other dietary variables and lifestyle factors, we conducted careful statistical analyses to adjust for these variables. However, the multivariate relative risks were similar to the age-adjusted ones, suggesting that confounding by other dietary variables and lifestyle factors was likely to be minor. Also, in stratified analyses according to levels of smoking and exercise and intakes of dietary fat and fiber, the modest inverse association with dietary protein persisted. Campbell suggests that the independent effects of various nutrients cannot be teased out because of their high correlations. This assertion is not substantiated by our analyses showing opposite associations with risk of coronary heart disease for different types of dietary fat that are intercorrelated (5). The large sample size, the long follow-up, and the multiple dietary measurements made in the Nurses' Health Study provide high power to examine independent effects of many individual nutrients. Although we agree that overall dietary patterns are also important in determining disease risk (10), we believe that identification of associations with individual nutrients should be the first step because it is the specific compounds or groups of compounds that are fundamentally related to the pathophysiology of the disease. Specific components of diet can be modified, and individuals and the food industry are actively doing so. Understanding the health effects of specific dietary changes, which Campbell refers to as "reductionism," is therefore an important undertaking.
REFERENCES