Author(s): Michels KB, Wolk A
Abstract Share this page
Abstract BACKGROUND: To assess the overall influence of diet on health and disease in epidemiological studies, the habitual diet of the study participants has to be captured as a pattern rather than individual foods or nutrients. The simplest way to describe dietary preferences is to separate foods considered beneficial to health from foods considered to promote disease, and separate individuals on the basis of their regular consumption of these foods. METHODS: We used data from 59 038 women participating in the prospective Mammography Screening Cohort in Sweden to investigate the influence of variety of healthy and less healthy foods on all-cause and cause-specific mortality. RESULTS: Women who followed a healthy diet defined as consumption of a high variety of fruits, vegetables, whole grain breads, cereals, fish, and low fat dairy products had a significantly lower mortality than women who consumed few of these foods (3710 deaths total). Women who reported regularly consuming 16-17 healthy foods had a 42\% lower all-cause mortality (95\% CI: 32-50\%) compared to women reporting consumption of 0-8 healthy foods with any regularity (P for trend <0.0001). For each additional healthy food consumed the risk of death was about 5\% lower (95\% CI: 4-6\%). Cardiovascular mortality was particularly low among women who reported consuming a high variety of healthy foods. A less healthy diet defined as consumption of a high variety of red meats, refined carbohydrates and sugars, and foods high in saturated or trans fats was not directly associated with a higher overall mortality. However, women who reported consuming many less healthy foods were significantly more likely to die from cancer than those who consumed few less healthy foods. CONCLUSIONS: A healthy diet can affect longevity. It appears more important to increase the number of healthy foods regularly consumed than to reduce the number of less healthy foods regularly consumed.
This article was published in Int J Epidemiol
and referenced in