Abstract
To evaluate the association of ultra-processed food (UPF) consumption with gains in weight and waist circumference, and incident overweight/obesity, in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) cohort. We applied FFQ at baseline and categorized energy intake by degree of processing using the NOVA classification. Height, weight and waist circumference were measured at baseline and after a mean 3·8-year follow-up. We assessed associations, through Poisson regression with robust variance, of UPF consumption with large weight gain (1·68 kg/year) and large waist gain (2·42 cm/year), both being defined as ≥90th percentile in the cohort, and with incident overweight/obesity. Brazil. Civil servants of Brazilian public academic institutions in six cities (n 11 827), aged 35-74 years at baseline (2008-2010). UPF provided a mean 24·6 (sd 9·6) % of ingested energy. After adjustment for smoking, physical activity, adiposity and other factors, fourth (>30·8 %) v. first (<17·8 %) quartile of UPF consumption was associated (relative risk (95 % CI)) with 27 and 33 % greater risk of large weight and waist gains (1·27 (1·07, 1·50) and 1·33 (1·12, 1·58)), respectively. Similarly, those in the fourth consumption quartile presented 20 % greater risk (1·20 (1·03, 1·40)) of incident overweight/obesity and 2 % greater risk (1·02; (0·85, 1·21)) of incident obesity. Approximately 15 % of cases of large weight and waist gains and of incident overweight/obesity could be attributed to consumption of >17·8 % of energy as UPF. Greater UPF consumption predicts large gains in overall and central adiposity and may contribute to the inexorable rise in obesity seen worldwide.
Highlights
MethodsWe characterized ultra-processed food consumption (percentage of total daily energy intake from these foods) as a continuous variable and expressed results for a 15 % increase in total energy consumed (approximately the interquartile range)
Foods and beverages classified as ultra-processed foods accounted for a mean of 24·6 (SD 9·6) % of total daily energy consumption; in natura, minimally processed or culinary ingredients for 64·1 %; and processed foods for 11·3 %
The associations with large weight and waist gains slowly plateaued (P for test of non-linearity = 0·23 and 0·17, respectively), while that with incident overweight/obesity completely flattened (P = 0·03 for non-linearity)
Summary
We characterized ultra-processed food consumption (percentage of total daily energy intake from these foods) as a continuous variable and expressed results for a 15 % increase in total energy consumed (approximately the interquartile range) We categorized such consumption into quartiles based on the overall analytic sample and used the first quartile as a reference. We analysed associations of ultra-processed food intake with a large annual weight or waist circumference gain and with incident overweight/obesity using Poisson regression with robust variance, progressively adjusting for age, sex, colour/race, centre, school achievement, and smoking and physical activity. For weight/waist gain and for incident overweight/obesity, we adjusted for baseline BMI, and for waist gain, we adjusted for baseline waist circumference In this full model, we tested interactions by categories of sex (men and women), race/colour (whites and non-whites) and age All analyses were conducted with the statistical software package SAS version 9.4, except for the adjusted population-attributable fractions which were calculated using the statistical software package Stata version 11.1
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
