Food groups as predictors for short-term weight changes in men and women of the EPIC-Potsdam cohort.

This study examined the effect of food group intake on subsequent 2-y weight change. Food-frequency questionnaire-based food intake data of 17,369 nonsmoking subjects of the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam cohort were examined in their relation to a subsequent weight change. Dietary data, collected from 1994 to 1998, were grouped into 24 food groups. Weight change per year follow-up was the outcome of interest; large weight gain was defined as > or =2 kg; small weight gain as > or =1 kg to <2 kg; large weight loss as < or = -2 kg; small weight loss as < or = -1 kg to > -2 kg and weight maintenance as +/- 1 kg. For each food group, a separate polytomous logistic regression model with stable weight as the reference group was constructed, controlling for age, body mass index, previous weight change, and behavioral and lifestyle factors. Odds ratios (OR) and 95% confidence intervals (CI) estimated the increase in risk associated with each 100 g/d increment in food group intake. In women, consumption of high energy, high fat food groups significantly predicted large weight gain, e.g., fats (OR = 1.75; 95% CI, 1.01-3.06), sauces (OR = 2.12; 95% CI, 1.17-3.82) and meat (OR = 1.36; 95% CI, 1.04-1.79), and the consumption of cereals predicted large weight loss (OR = 1.43; 95% CI, 1.09-1.88). In men, intake of high energy, high sugar foods, i.e., sweets, was significantly predictive of large weight gain (OR = 1.48; 95% CI, 1.03-2.13). Our data show that a diet rich in high fat and high energy foods predicts short-term weight gain even if controlled for many potential confounding factors.