Reported food intake and distribution of body fat: a repeated cross-sectional study

Abstract

Background: Body mass, as well as distribution of body fat, are predictors of both diabetes and cardiovascular disease. In Northern Sweden, despite a marked increase in average body mass, prevalence of diabetes was stagnant and myocardial infarctions decreased. A more favourable distribution of body fat is a possible contributing factor. This study investigates the relative importance of individual food items for time trends in waist circumference (WC) and hip circumference (HC) on a population level.

Methods: Independent cross-sectional surveys conducted in 1986, 1990, 1994 and 1999 in the two northernmost counties of Sweden with a common population of 250,000. Randomly selected age stratified samples, altogether 2982 men and 3087 women aged 25-64 years. Questionnaires were completed and anthropometric measurements taken. For each food item, associations between frequency of consumption and waist and hip circumferences were estimated. Partial regression coefficients for every level of reported intake were multiplied with differences in proportion of the population reporting the corresponding levels of intake in 1986 and 1999. The sum of these product terms for every food item was the respective estimated impact on mean circumference.

Results: Time trends in reported food consumption associated with the more favourable gynoid distribution of adipose tissue were increased use of vegetable oil, pasta and 1.5% fat milk. Trends associated with abdominal obesity were increased consumption of beer in men and higher intake of hamburgers and French fried potatoes in women.

Conclusion: Food trends as markers of time trends in body fat distribution have been identified. The method is a complement to conventional approaches to establish associations between food intake and disease risk on a population level.

Figure 1

Study design. ¹Counties of Norrbotten and Västerbotten, population register of inhabitants aged 25–64 on January 1^st of the survey-year. ²Randomly selected samples of 250 in age strata 25–34, 35–44, 45–54, 55–64 years. ³Number of individuals who appeared at the health centers for measurements and completed FFQ. ⁴Number of individuals with complete anthropometrical data and at least 90%

Figure 2

Estimated effect of time trends in reported food intake 1986–1999 on average waist- and hip-circumference in women. 10 items with the largest estimated effect on distribution of body fat in women. Sort order is the sum of effects from largest reduction to highest increase of risk for diabetes. The underlying associations between food intake and waist- and hip-circumferences were adjusted for age, body-mass and survey year (model 1)

Figure 3

Estimated effect of time trends in reported food intake 1986–1999 on average waist- and hip-circumference in men. 10 items with the largest estimated effect on distribution of body fat in men. Sort order is the sum of effects from largest reduction to highest increase of risk for diabetes. The underlying associations between food intake and waist- and hip-circumferences were adjusted for age, body-mass and survey year (model 1)