Ultraprocessed food consumption and risk of overweight and obesity:

the University of Navarra Follow-Up (SUN) cohort study1,2

Raquel de Deus Mendonça,3,4,6 Adriano Marçal Pimenta,3,5 Alfredo Gea,3,7,8 Carmen de la Fuente-Arrillaga,3,7,8
Miguel Angel Martinez-Gonzalez,3,7–9 Aline Cristine Souza Lopes,4 and Maira Bes-Rastrollo3,7,8*
3
University of Navarra, Department of Preventive Medicine and Public Health, Pamplona, Spain; Departments of 4Nutrition and 5Maternal-Child Nursing and
Public Health, School of Nursing, Federal University of Minas Gerais, Belo Horizonte, Brazil; 6CAPES Coordination for the Improvement of Higher
Education Personnel Foundation, Ministry of Education of Brazil, Brası́lia, Brazil; 7Navarra Health Research Institute, Pamplona, Spain; 8Biomedical
Research Center Network in Physiopathology of Obesity and Nutrition, Carlos III Health Institute, Madrid, Spain; and 9Department of Nutrition, Harvard
T.H. Chan School of Public Health, Boston, MA

ABSTRACT were obese, and in the Eastern Mediterranean w25% of women

Background: Ultraprocessed food consumption has increased in and 15% of men were obese (1).
the past decade. Evidence suggests a positive association between Changes in the food system continuously promote obesity.
ultraprocessed food consumption and the incidence of overweight There is now a greater availability of ready-to-eat or -heat foods
and obesity. However, few prospective studies to our knowledge known as ultraprocessed foods, which are products that have
have investigated this potential relation in adults. little, if any, whole foods and are manufactured with substances
Objective: We evaluated the association between ultraprocessed extracted from foods or synthesized in laboratories (dyes, fla-
food consumption and the risk of overweight and obesity in a prospec- vorings, and other additives) (2). They have high amounts of fat,
tive Spanish cohort, the SUN (University of Navarra Follow-Up) study. sugar, and salt and a high energy density and low fiber content;
Design: We included 8451 middle-aged Spanish university gradu- they are extremely palatable foods that are aggressively adver-
ates who were initially not overweight or obese and followed up for tised (3) and contain a large diversity of chemical additives.
a median of 8.9 y. The consumption of ultraprocessed foods (de- Examples of ultraprocessed foods include breakfast cereals,
fined as food and drink products ready to eat, drink, or heat and reconstituted meat products, soft drinks, and ready-to-eat
made predominantly or entirely from processed items extracted or foodstuffs (2).
refined from whole foods or synthesized in the laboratory) was Several prospective studies have been carried out to assess the
assessed with the use of a validated semiquantitative 136-item relation between dietary components and obesity. An analysis of
food-frequency questionnaire. Cox proportional hazards models 3 American cohorts, the Nurses’ Health Study, Nurses’ Health
were used to estimate adjusted HRs and 95% CIs for incident
Study II, and Health Professionals Follow-Up Study, showed
overweight and obesity.
that the consumption of foodstuffs such as sweets, desserts,
Results: A total of 1939 incident cases of overweight and obesity
processed meats, fries, and sugar-sweetened beverages were
were identified during follow-up. After adjustment for potential
strongly associated with weight gain in US adults (4). However,
confounders, participants in the highest quartile of ultraprocessed
the relation between the consumption of foods aggregated
food consumption were at a higher risk of developing overweight or
according to their degree of processing (i.e., ultraprocessed foods)
obesity (adjusted HR: 1.26; 95% CI: 1.10, 1.45; P-trend = 0.001)
and excess weight has only recently been analyzed.
than those in the lowest quartile of consumption.
A cross-sectional times series study in 15 Latin American
Conclusions: Ultraprocessed food consumption was associated
with a higher risk of overweight and obesity in a prospective cohort countries revealed an association between sales of ultraprocessed
of Spanish middle-aged adult university graduates. Further longitu- foods and changes in body weight in 12 countries from 2000 to
dinal studies are needed to confirm our results. This trial was reg- 2009 (5). Cross-sectional studies have demonstrated an associ-
istered at clinicaltrials.gov as NCT02669602. Am J Clin Nutr ation between the consumption of these foods and the risk of
2016;104:1433–40. metabolic syndrome in adolescents (6) and obesity in adolescents

Keywords: obesity, overweight, ultra-processed food, food- 1

Supported by Carlos III Health Institute and the European Regional De-
processing industry, SUN cohort, prospective studies
velopment Fund, regional government of Navarra, and University of Navarra.
2
Supplemental Table 1 is available from the “Online Supporting Material”
INTRODUCTION link in the online posting of the article and from the same link in the online
table of contents at http://ajcn.nutrition.org.
Obesity is a worldwide problem with increasing prevalence. In *To whom correspondence should be addressed. E-mail: [email protected].
2014, more than half a billion adults were obese, and 39% were Received March 18, 2016. Accepted for publication September 8, 2016.
overweight. In the Americas, w30% of women and 25% of men First published online October 12, 2016; doi: 10.3945/ajcn.116.135004.

Am J Clin Nutr 2016;104:1433–40. Printed in USA. Ó 2016 American Society for Nutrition 1433
1434 MENDONÇA ET AL.

and adults (7, 8). A study of children aged 3–8 y in a low-income consumption were measured in 9 categories (ranging from never or
community in Brazil showed that ultraprocessed food con- almost never to .6 servings/d), and the FFQ included a typical
sumption was an important predictor of the increase in total portion size for each item. Daily food consumption was estimated by
cholesterol and LDL cholesterol (9). On the other hand, a study multiplying the portion size by the consumption frequency for each
with data from the 2008–2012 United Kingdom National Diet food item.
and Nutrition Survey showed no association between ultra- The foods were classified according to NOVA (2) based on the
processed food consumption and body weight. However, diets extent and purpose of applied food processing. There are 4
with smaller quantities of ultraprocessed foods have better nu- groups in this classification scheme. The first group includes
tritional quality (10). foods that are fresh or processed in ways that did not add sub-
Studies have shown an association between specific types of stances such as salt, sugar, oils, or fats and infrequently contain
ultraprocessed foods, such as soft drinks, and being overweight or additives. Processes used are aimed to extend life, allow storage
obese. However, no prospective cohort study to our knowledge for long use, and facilitate or diversify preparation (freezing,
has been performed to evaluate the association between all drying, and pasteurization). Examples are fruits and vegetables,
ultraprocessed foods as a group and the incidence of overweight grains (cereals), flours, nuts and seeds, fresh and pasteurized
and obesity in adults. Therefore, in this study (NCT02669602), milk, natural yogurt with no added sugar or artificial sweeteners,
we evaluated the association of ultraprocessed food consumption meat and fish, tea, coffee, drinking water, spices, and herbs. The
with the incidence of overweight and obesity in a Mediterranean second group contains processed culinary ingredients. These are
cohort with a prolonged follow-up. substances obtained from the first group or from nature and may
contain additives to preserve the original properties. Examples
are salt, sugar, honey, vegetable oils, butter, lard, and vinegar.
METHODS The third group is processed food made with the addition of
substances such as salt, sugar, or oil and the use of processes
Study population
such as smoking, curing, or fermentation. Examples are canned
The SUN (University of Navarra Follow-Up) Project is a dy- or bottled vegetables and legumes, fruits in syrup, canned fish,
namic and multipurpose prospective cohort study with perma- cheeses, freshly made bread, and salted or sugared nuts and
nently open recruitment conducted in Spain among university seeds. The fourth group is ultraprocessed food and drink prod-
graduates since December 1999. The participants are followed up ucts that are made predominantly or entirely from industrial
biennially with the use of questionnaires distributed by post or substances and contain little or no whole foods. These products
electronic mail. Details of its design have been published else- are ready to eat, drink, or heat. Examples include carbonated
where (11, 12). drinks, sausages, biscuits (cookies), candy (confectionery), fruit
Up to March 2012, the SUN data set included 21,291 par- yogurts, instant packaged soups and noodles, sweet or savory
ticipants who had answered the baseline questionnaire. In this study, packaged snacks, and sugared milk and fruit drinks (2). It is this
we excluded those classified as overweight or obese [BMI (in kg/m2) fourth group that is the main subject of this study. Supplemental
$25] at baseline (n = 6340), individuals who reported total energy Table 1 describes the classification of FFQ foods according to
intake values outside of predefined limits [low: ,3347 kJ/d NOVA. The frequency of ultraprocessed food consumption was
or ,800 kcal/d in men and ,2092 kJ/d or ,500 kcal/d in women; estimated with the use of the sum of the food items from the
high: .16,736 kJ/d or .4000 kcal/d in men and .14,644 kJ/d fourth group in the FFQ (total of 33 items). The sample was di-
or .3500 kcal/d in women (n = 1713)] (13), women who were vided into quartiles according to total consumption (servings/d).
pregnant at baseline or became pregnant during the follow-up
period (n = 2739), and individuals who reported a previously di- Outcome assessment
agnosed chronic disease at baseline (e.g., diabetes, cancer, car-
diovascular disease) (n = 618). In addition, we excluded participants Self-reported weight and height were validated with a previous
with a weight change .10 kg in the 5 y preceding entry into the study (16). The outcome used the incidence of overweight and
study to reduce potential sources of confounding by other causes of obesity (BMI $25) during follow-up and was defined as the first
weight changes (n = 260). Among the remaining participants, 1106 time participants reached a BMI of 25 during follow-up.
subjects were lost to follow-up, and 64 participants had missing
values in $1 variable of interest. After these exclusions, a total of Assessment of other variables
8451 participants were included in the final analyses (Figure 1). The baseline questionnaire also included questions relating to
The retention rate of the study was w89%. the following variables: sex, age, marital status, educational status,
This study was conducted according to Declaration of Helsinki smoking status, physical activity, television watching, siesta sleep,
guidelines, and all procedures involving human subjects were diet and dietary habits, and snacking between main meals. Physical
approved by the University of Navarra institutional review board. activity was evaluated with the use of a validated 17-item ques-
Voluntary completion of the baseline self-administrated ques- tionnaire (17). Total energy, macronutrient, fiber, and alcohol intake
tionnaire was considered to imply informed consent. and fruit, vegetable, fast-food, fried-food, processed meat, non-
processed meat, and sugar-sweetened beverage consumption was
assessed with the use of the FFQ (15). Nutrient intake scores were
Exposure assessment: ultraprocessed food consumption computed with the use of a computer program developed spe-
Dietary exposures were assessed at baseline through a self- cifically by our dietitians for this purpose. Adherence to the
administered 136-item semiquantitative food-frequency questionnaire Mediterranean dietary pattern was evaluated with the use of a well-
(FFQ) that was previously validated in Spain (14, 15). Frequencies of known score (18).
 ULTRAPROCESSED FOODS AND OVERWEIGHT AND OBESITY RISK 1435

FIGURE 1 Flowchart of participants.

Statistical analyses and following a special diet. Total energy intake was not included
Differences in baseline characteristics of participants according as a covariate because it may plausibly mediate the association of
to ultraprocessed food consumption quartiles were evaluated with ultraprocessed foods and overweight and obesity. We evaluated
ANOVA and adjusted for sex and age. To evaluate the relation the interaction between exposure, sex, and BMI with the use of
a likelihood ratio test that compared the fully adjusted Cox re-
between ultraprocessed food consumption at baseline and the
gression model and the same model with interaction product
subsequent risk of the development of overweight and obesity
terms (3 df). Nelson-Aalen cumulative hazards estimates were
during follow-up, we used Cox proportional hazards models, and to
plotted for overweight and obesity incidence according to
estimate HRs and 95% CIs we used the lowest quartile as the
ultraprocessed food consumption quartiles at baseline. We used
reference category. inverse probability weighting to adjust the Nelson-Aalen curves
The follow-up period was defined as the interval between the for baseline potential confounders.
date of recruitment and date of the return of the follow-up To test the proportional hazard assumption, we calculated a Cox
questionnaire in which the participant was classified as overweight regression with the exposure as a continuous time-varying covariate
or obese for the first time (for incident cases). The date of death or to check that the HR did not vary over time, obtaining a non-
of the last questionnaire was used for noncases. significant result, suggesting that the proportionality assumption was
Tests for linear trends were conducted by assigning medians of met. We also checked the proportionality of hazards model with the
ultraprocessed food consumption to each category and treating use of a Grambsch-Therneau test of the scaled Schoenfeld residuals
this variable as a continuous variable in the respective Cox re- from a Cox model on the 3 dummy variables of the upper ultra-
gression model. We fitted a first model without any adjustment processed quartiles (19). The P value of the global test was 0.72.
(crude), a second model adjusted for age and sex, and a third To determine the contribution of each food item to the between-
multivariable-adjusted model adjusted for age, sex, marital sta- person variance in ultraprocessed food consumption (13), we
tus, educational status, baseline BMI, physical activity, television constructed a series of nested least-squares linear regression
watching, siesta sleep, smoking status, snacking between meals, models after stepwise-selection regression analyses. The additional
1436 MENDONÇA ET AL.

contribution of a given food item was reflected in the change in All analyses were performed with Stata version 12.1 (Stata-
the cumulative R2. Corp LP). P , 0.05 was considered significant.
Sensitivity analyses were conducted by repeating the multivariable-
adjusted Cox regression models with the following changes: 1) ad-
ditional adjustment excluding fruit and vegetable consumption, 2) RESULTS
exclusion of those participants under the 5th percentile and over the
A total of 2967 (35.1%) men and 5484 (64.9%) women were
95th percentile of total energy intake, 3) additional adjustment for total
included in this analysis, and the mean 6 SD age of the par-
energy intake, 4) additional adjustment for family history of obesity, ticipants was 37.6 6 11.0 y. The main baseline characteristics of
5) additional adjustment for weight gain .3 kg in the 5 y before participants according to quartiles of total ultraprocessed food
entering the cohort, 6) exclusion of participants who were early consumption are presented in Table 1. Participants in the fourth
incident cases of overweight (those who became overweight after quartile of ultraprocessed food consumption had the highest
only 2 y of follow-up), and 7) inclusion of subjects with prevalent BMI, were more likely to be current smokers, watched more
chronic diseases. television, and had the highest total energy and fat intake and the

TABLE 1
Baseline characteristics of participants according to their consumption of ultraprocessed foods1
Quartile

Characteristics 1 2 3 4 P

n 2118 2108 2116 2109

Total ultraprocessed consumption, servings/d 1.5 6 0.9 2.7 6 0.9 3.8 6 0.9 6.1 6 0.9 ,0.001
Marital status, %
Single 47.8 44.8 45.8 48.6 0.29
Married 45.8 49.7 49.1 46.6 0.74
Educational status, %
Graduated 79.3 77.4 76.9 76.7 0.09
Master or doctoral 16.4 19.2 19.4 18.4 0.19
Baseline weight, kg 61.9 6 6.4 62.1 6 6.3 62.6 6 6.3 62.6 6 6.4 ,0.001
Baseline BMI, kg/m2 21.9 6 1.7 21.9 6 1.6 22.0 6 1.6 22.0 6 1.7 0.004
Family history of obesity, % 20.4 20.7 20.2 21.5 0.55
Smoking status, %
Current 19.9 19.8 22.8 24.1 ,0.001
Former 26.4 25.8 25.9 24.5 0.25
Physical activity, MET-h/wk 23.3 6 24.1 22.8 6 23.5 22.4 6 23.6 22.7 6 23.9 0.41
Television watching, h/d 1.5 6 1.2 1.5 6 1.2 1.5 6 1.2 1.6 6 1.2 ,0.001
Sleeping siesta, h/d 0.3 6 0.8 0.3 6 0.7 0.3 6 0.7 0.3 6 0.7 0.12
Total energy intake, kcal/d 1967 6 542 2264 6 527 2464 6 529 2796 6 535 ,0.001
Macronutrients, % energy
Carbohydrate intake 43.8 6 7.4 43.8 6 7.2 43.7 6 7.2 43.8 6 7.3 0.77
Protein intake 18.9 6 3.2 18.2 6 3.1 17.5 6 3.1 17.0 6 3.1 ,0.001
Fat intake 35.2 6 6.5 36.2 6 6.3 36.9 6 6.4 37.5 6 6.4 ,0.001
SFAs 11.6 6 3.2 12.4 6 3.1 12.8 6 3.1 13.0 6 3.1 ,0.001
MUFAs 15.6 6 3.8 15.6 6 3.7 15.8 6 3.7 15.9 6 3.7 0.004
PUFAs 4.9 6 1.6 5.0 6 1.5 5.3 6 1.5 5.4 6 1.5 ,0.001
Total dietary fiber intake, g/kcal 31.2 6 10.4 28.5 6 10.1 27.2 6 10.1 25.5 6 10.3 ,0.001
Fruit consumption, g/d 358 6 284 347 6 277 338 6 278 336 6 281 0.01
Vegetable consumption, g/d 543 6 341 507 6 331 516 6 333 517 6 337 0.06
Fast-food consumption,2 g/d 14.1 6 18.8 19.8 6 18.3 23.7 6 18.4 29.2 6 18.6 ,0.001
Processed meat consumption,3 servings/d 0.4 6 0.5 0.6 6 0.4 0.8 6 0.4 1.0 6 0.4 ,0.001
Nonprocessed meat consumption, servings/d 0.7 6 0.4 0.8 6 0.4 0.8 6 0.4 0.9 6 0.4 ,0.001
Fried food consumption, servings/wk 2.8 6 4.4 3.5 6 4.3 3.9 6 4.3 4.6 6 4.3 ,0.001
Sugar-sweetened beverage consumption, mL/d 20.6 6 108.9 35.9 6 106.0 58.4 6 106.4 119.2 6 107.7 ,0.001
Alcohol intake, g/d 5.0 6 8.4 5.8 6 8.1 6.1 6 8.2 6.5 6 8.3 ,0.001
Between-meal snacking, % 22.5 28.0 31.5 39.0 ,0.001
Mediterranean diet adherence score, %
0–2 16.2 18.9 20.5 21.1 ,0.001
3–6 70.0 69.5 70.5 70.0 0.95
7–9 13.8 11.6 9.0 8.9 ,0.001
Special diet at baseline, % 8.0 5.0 5.0 5.8 0.01
1
All values are means 6 SDs unless otherwise indicated. P values were adjusted for sex and age with the use of
ANOVA. MET, metabolic equivalent; SUN, University of Navarra Follow-Up.
2
Sum of hamburgers, sausages, and pizza.
3
Sum of sausages, hamburgers, and ham.
 ULTRAPROCESSED FOODS AND OVERWEIGHT AND OBESITY RISK 1437
TABLE 2 processed food intake (Figure 2). No significant interaction
Main sources of ultraprocessed foods1 between sex or BMI and ultraprocessed food consumption was
Foods Change in R2 Cumulative R2 observed. Results in the sensitivity analyses described previously
were not substantially changed in any of the scenarios (Table 4).
Processed meat2 — 0.1901
Cookies3 0.1224 0.3125
Sugar-sweetened beverages 0.0974 0.4098
Pastries4 0.0716 0.4814 DISCUSSION
Breakfast cereals 0.0619 0.5433
Chocolate 0.0451 0.5884 In this prospective cohort study with initially healthy, middle-
Fruit drinks in bottles 0.0346 0.6229 aged, Spanish adults, higher ultraprocessed food consumption
Margarine 0.0262 0.6491 was associated with an increased risk of overweight and obesity
1 incidence for a mean of w9 y of follow-up, even after adjusting
Cumulative R2 values were determined with the use of nested regres-
sion analyses after a stepwise selection. for potential confounding factors.
2
Includes ham, sausages, chorizo, salami, mortadella, and hamburgers. Recent ecological and cross-sectional studies on ultraprocessed
3
Includes biscuits and chocolate cookies. food consumption and the risk of overweight and obesity have
4
Includes muffins, doughnuts, croissants or other pastries, and shown an association between increased ultraprocessed food
confectionery. consumption and obesity among adolescents and adults (5, 7, 8,
20). To our knowledge, this is the first prospective epidemiologic
lowest protein and total fiber than those in the first quartile. study to assess this association among adults.
Moreover, on average, they consumed more fast food, fried The food and drinks classified as ultraprocessed foods are
foods, processed and other meats, and sugar-sweetened beverages. mostly manufactured by large and powerful transnational com-
In contrast, they had the lowest intakes of vegetables and were less panies. These companies focus their marketing on the individual’s
likely to follow special diets. In addition, they had the highest lack of time compared with the expediency, convenience, and
prevalence of snacking between main meals. In addition, the accessibility of these products (21, 22). However, ultraprocessed
adherence to the Mediterranean diet score varied according to foods tend to be nutritionally unbalanced and have a high energy
quartiles of ultraprocessed food (i.e., the higher the con- content and amounts of fat, added or free sugar, sodium, and
sumption of ultraprocessed foods, the less adherence to the chemical additives in addition to being poor in micronutrients
Mediterranean diet). and fiber (3, 21, 23, 24). This nutritional composition is
The contributions of different food groups to ultraprocessed caused by several industrial processes, such as the removal of
food intake are shown in Table 2. Processed meat, biscuits and water, which increases the shelf life and reduces trans-
cookies, sugar-sweetened beverages, and candies were among portation costs but increases the energy density per portion. In
the major contributors to ultraprocessed food consumption addition, to confer greater stability, ultraprocessed foods have
variability. a greater amount of trans and saturated fats and synthetic
During the follow-up period (median: 8.9 y; person-years: additives (24). Ultraprocessed foods are also designed to be
66,625), we observed 1939 incident cases of overweight and irresistible and to favor consumption (because of the use of
obesity. In the multivariate models, when we assessed the risk of salt, sugar, and fat but also because of marketing and other
overweight and obesity according to quartiles of ultraprocessed characteristics such as portion size and convenience), and the
foods consumption (Table 3), participants with higher ultra- processing also affects food behavior and the capacity for self-
processed food intake (highest quartile) presented a 26% rela- control (2, 3). This new classification proposal of foods ac-
tively higher risk of developing overweight or obesity than those cording to the degree of processing (2) therefore seems to be
in the lowest quartile (HR: 1.26; 95% CI: 1.10, 1.45). The esti- a good way of assessing their intake and relation to health
mates showed a statistically significant linear trend (P = 0.001). outcomes because it can identify the consumption of a group
The Nelson-Aalen curves exhibited a higher incidence of over- of foods and drinks with low nutritional quality rather than
weight and obesity with increasing baseline quartiles of ultra- just a certain food or nutrient.

TABLE 3
Cox proportional HRs and 95% CIs for incident overweight and obesity according to baseline consumption of
ultraprocessed foods1
Quartile

1 2 3 4 P-trend

Incident cases 440 466 512 521

Person-years 16,889 16,790 16,522 16,423
Crude 1.00 (reference) 1.17 (1.03, 1.34) 1.40 (1.23, 1.59) 1.49 (1.31, 1.70) ,0.001
Age- and sex-adjusted 1.00 (reference) 1.12 (0.98, 1.28) 1.27 (1.11, 1.45) 1.29 (1.13, 1.47) ,0.001
Multivariable-adjusted2 1.00 (reference) 1.15 (1.01, 1.32) 1.24 (1.09, 1.43) 1.26 (1.10, 1.45) 0.001
1
All values are HRs; 95% CIs in parentheses unless otherwise indicated.
2
Adjusted for sex, age, marital status, educational status, physical activity, television watching, siesta sleep, smoking
status, snacking between meals, following a special diet at baseline, baseline BMI, and consumption of fruit and vegetables.
1438 MENDONÇA ET AL.

FIGURE 2 Nelson-Aalen curves showing the outcome of new-onset overweight depending on ultraprocessed food consumption at baseline. Adjusted for
sex, age, marital status, educational status, physical activity, television watching, siesta sleep, smoking status, snacking between meals, following a special diet
at baseline, baseline BMI, and consumption of fruit and vegetables with the use of inverse probability weighting. Q, quartile.

The sales of ultraprocessed foods increased between 2000 and food consumption and a diet with a high energy density and intake
2013. Despite the United States being the largest buyer, sales of of sugars, sodium, and total and saturated fats and an inverse as-
these products in Spain grew 18.5%, followed by Canada, sociation with fiber intake (3, 21). Moreover, a study from Brazil
Germany, and Mexico (5). In a longitudinal study of food and that examined data from a national household food budget survey
beverage consumer packaged goods purchased by US house- showed that higher ultraprocessed food consumption was associ-
holds, the 2000–2012 Nielsen Homescan Panel showed that 61% ated with the lowest intakes of vitamin B-12, vitamin D, niacin,
of the energy in purchases by US households in 2012 was derived iron, selenium, and magnesium (26).
from ultraprocessed foods (25). The 2008–2012 United Kingdom We believe that ultraprocessed food consumption may increase
National Diet and Nutrition Survey, a large national cross-sectional the risk of overweight and obesity by increasing the total intake of
study of diet, showed that diets with a higher intake of minimally calories, added and free sugars, and fats and providing an in-
processed foods and lower intake of ultraprocessed foods are as- adequate relation of nutrients potentially involved in the genesis
sociated with a more healthy food profile, but this was not asso- of the accumulation of body fat (1, 3, 21–24, 26). A cross-
ciated with body weight (10). Studies from Canada and Brazil, sectional study from the United States with data from NHANES
conducted with data from national household food budget surveys, 2009–2010 showed that ultraprocessed food represented 57.9%
have also demonstrated a direct association between ultraprocessed of energy intake and that w90% of this amount was derived

TABLE 4
Sensitivity analyses of HRs (95% CIs) for incident overweight and obesity according to quartiles of consumption of ultraprocessed foods1
Quartile

Cases/person-years, n 1 2 3 4 P-trend

Overall 1939/66,625 1.00 (reference) 1.15 (1.01, 1.32) 1.24 (1.09, 1.43) 1.26 (1.10, 1.45) 0.001
Excluding adjustment for fruit and vegetable 1939/66,625 1.00 (reference) 1.15 (1.01, 1.32) 1.25 (1.09, 1.43) 1.26 (1.10, 1.35) 0.001
consumption
Further adjusted for energy total intake 1939/66,625 1.00 (reference) 1.15 (1.01, 1.32) 1.25 (1.09, 1.44) 1.27 (1.09, 1.49) 0.003
Further adjusted for family history of obesity 1939/66,625 1.00 (reference) 1.16 (1.01, 1.33) 1.25 (1.10, 1.44) 1.27 (1.11, 1.45) 0.001
Further adjusted for weight gain .3 kg in the 1939/66,625 1.00 (reference) 1.15 (1.00, 1.32) 1.24 (1.09, 1.42) 1.26 (1.10, 1.44) 0.001
5 y before entering the cohort
Inclusion of prevalent cancer, diabetes, or 2073/70,617 1.00 (reference) 1.14 (1.00, 1.29) 1.20 (1.06, 1.37) 1.24 (1.09, 1.41) 0.002
cardiovascular disease
Exclusion of early incident cases of overweight 1293/65,030 1.00 (reference) 1.17 (0.99, 1.38) 1.14 (0.97, 1.35) 1.32 (1.12, 1.56) 0.002
(until 2 y of follow-up)
Energy limits between 5th and 95th percentiles 1828/63,401 1.00 (reference) 1.21 (1.06, 1.39) 1.21 (1.05, 1.39) 1.28 (1.12, 1.47) 0.002
1
All values are HRs (95% CIs) unless otherwise indicated. Analyses were adjusted for sex, age, marital status, educational status, physical activity,
television watching, siesta sleep, smoking status, snacking between meals, following a special diet at baseline, baseline BMI, and consumption of fruit and
vegetables.
 ULTRAPROCESSED FOODS AND OVERWEIGHT AND OBESITY RISK 1439
from added sugars (23). The participants in our cohort with risk between ultraprocessed food consumption and excess weight
a higher daily consumption of ultraprocessed foods had a higher may be even greater in the general population. Conversely, the
consumption of total calories and fat, a lower consumption of participants have a higher educational status and are capable of
proteins and fibers, and a lower adherence to the Mediterranean providing better quality self-reported data, thus reducing the
dietary pattern. potential for misclassification bias. Nevertheless, we must be
The Nurses’ Health Study showed a strong association be- cautious with regard to the generalization of the results because
tween the percentage of calories from saturated and trans fats they do not depend on the statistical representativeness of the
and weight gain (27). Results from the Nurses’ Health Study II sample but on the biological mechanisms underlying the ob-
also showed that an increase in dietary energy density was a risk served association.
factor for obesity (28). A meta-analysis of cohort studies and The main strengths of this study are its prospective design, the
randomized clinical trials has provided evidence of the relation use of validated methods (15, 16), the relatively large sample size,
between the intake of sugar and the development of overweight and the long follow-up period. In addition, this is one of the first
and obesity (29). On the other hand, the consumption of natural longitudinal studies to our knowledge to evaluate the relation
foods such as fruits, vegetables, fish, and fiber—but not red between ultraprocessed food intake and overweight and obesity.
meat—is inversely associated with the incidence of overweight Further studies should be performed to investigate whether
and obesity (4, 30). there is a relation between increased ultraprocessed food intake
Our findings stress the importance of promoting and enhancing and major causes of death, such as cardiovascular diseases
healthy and sustainable food standards, which requires structural and cancer.
and behavioral changes. Food systems should improve the supply The results suggest that increased ultraprocessed food con-
of healthy foods and foster healthier food choices (31). This may sumption is associated with a greater risk of overweight and
be accomplished by promoting and enhancing eating habits based obesity. Strategies for reducing the consumption of this group
on foods and cuisine, such as the Mediterranean dietary pattern. of foods, such as the maintenance of a traditional food culture
Several studies have demonstrated its association with the re- and strengthening of the Mediterranean diet, should be encouraged
duction of weight gain because it consists predominantly of fruits, as preventive approaches for obesity. However, further longi-
vegetables, legumes, nuts, olive oil, and fish (17, 32). Therefore, tudinal studies in different contexts are necessary to confirm our
the Mediterranean diet is based on the consumption of fresh findings.
foods; multiple studies have demonstrated its preventive effect
The authors’ responsibilities were as follows—RdDM, ACSL, and MB-R:
against chronic diseases (32–35). analyzed the data and drafted the manuscript; MB-R: obtained the funding
Some limitations may be observed in this study. The assess- and designed the study; AMP, AG, and CdlF-A: helped to design the study
ment of ultraprocessed food consumption was performed with the and interpreted the data; MAM-G: helped to design the study and was the
use of an FFQ that was not specifically designed to collect data founder of the SUN project; and all authors: read and approved the final
regarding this new classification of foods. Therefore, there is the manuscript. None of the authors reported a conflict of interest related to
potential for some degree of misclassification of ultraprocessed the study.
food consumption inherent in our methodology. However, our
FFQ was previously validated and represents the main foods
ingested by the studied population (15), including ultraprocessed REFERENCES
foods. Hamburgers are classified as an ultraprocessed food 1. WHO. Global status report on noncommunicable diseases 2014. Ge-
according to NOVA (2). Hamburgers are not ground beef but are neva (Switzerland): WHO; 2014.
2. Monteiro CA, Cannon G, Levy R, Moubarac JC, Jaime P, Martins AP,
a dish made of industrial breads, sauces, reconstituted meat,
Canella D, Louzada M, Parra D, Ricardo C, et al. Food classification.
processed cheese, bacon, etc. Therefore, we included hamburgers Public health. J World Public Health Nutr 2016;7:28–38.
in the ultraprocessed food group. Some hamburgers can be homemade, 3. Moubarac JC, Martins AP, Claro RM, Levy RB, Cannon G, Monteiro
and doubts may arise about their inclusion in this group. Un- C. Consumption of ultra-processed foods and likely impact on human
fortunately, our study could not differentiate between fast-food and health. Evidence from Canada. Public Health Nutr 2013;16:2240–8.
4. Mozaffarian D, Hao T, Rimm EB, Willett WC, Hu FB. Changes in diet
homemade hamburgers. Even when we removed hamburgers from and lifestyle and long-term weight gain in women and men. N Engl J
the ultraprocessed foods, the results were only slightly attenuated Med 2011;364:2392–404.
(HR: 1.23; 95% CI: 1.07, 1.41; P-trend = 0.003). 5. Pan American Health Organization. Ultra-processed food and drink
Self-reported data were used to evaluate the BMI and to products in Latin America: trends, impact on obesity, policy implica-
classify the overweight and obesity incident cases. This may tions. Washington (DC): Pan American Health Organization; 2013.
6. Tavares LF, Fonseca SC, Garcia Rosa ML, Yokoo EM. Relationship
not be precise enough, especially because we cannot accu- between ultra-processed foods and metabolic syndrome in adolescents
rately specify the date of onset of the overweight or obese status. from a Brazilian family doctor program. Public Health Nutr 2012;15:
Nevertheless, this method was previously validated (16). In 82–7.
addition, the participants were health professionals, and this fact 7. Louzada MLC, Baraldi LG, Steele EM, Martins AP, Canella DS,
Moubarac JC, Levy RB, Cannon G, Afshin A, Imamura F, et al.
is likely to have increased the validity of their self-reported Consumption of ultra-processed foods and obesity in Brazilian ado-
anthropometric data. lescents and adults. Prev Med 2015;81:9–15.
Potential confounding may still be possible. Nevertheless, we 8. Canella DS, Levy RB, Martins APB, Claro RM, Moubarac JC, Baraldi
adjusted for several potential confounding factors related to LG, Cannon G, Monteiro CA. Ultra-processed food products and
lifestyle. Caution is needed in extrapolating the results to the obesity in Brazilian households (2008–2009). PLoS One 2014;9:
e92752.
general population because the SUN participants are volunteer 9. Rauber F, Campagnolo PD, Hoffman DJ, Vitolo MR. Consumption of
university graduates. This group could be more aware of health, ultra-processed food products and its effects on children’s lipid pro-
thus causing a selection bias that implies that the magnitude of the files: a longitudinal study. Nutr Metab Cardiovasc Dis 2015;25:116–22.
1440 MENDONÇA ET AL.
10. Adams J, White M. Characterisation of UK diets according to degree 24. Ludwig DS. Technology, diet, and the burden of chronic disease. JAMA
of food processing and associations with socio-demographics and 2011;305:1352–3.
obesity: cross-sectional analysis of UK National Diet and Nutrition 25. Poti JM, Mendez MA, Ng SW, Popkin BM. Is the degree of food
Survey (2008–12). Int J Behav Nutr Phys Act 2015;12:160. processing and convenience linked with the nutritional quality of
11. Seguı́-Gómez M, de la Fuente C, Vázquez Z, de Irala J, Martı́nez- foods purchased by US households? Am J Clin Nutr 2015;101:
González MA. Cohort profile: the ‘Seguimiento Universidad de Navarra’ 1251–62.
(SUN) study. Int J Epidemiol 2006;35:1417–22. 26. Louzada MLC, Martins APB, Canella DS, Baraldi LG, Levy RB, Claro
12. Martı́nez-González MA, Sanchez-Villegas A, de Irala J, Marti A, Martı́nez RM, Moubarac JC, Cannon G, Monteiro CAImpact of ultra-processed
JA. Mediterranean diet and stroke: objectives and design of the SUN foods on micronutrient content in the Brazilian diet. Rev Saúde Pública
project. Seguimiento Universidad de Navarra. Nutr Neurosci 2002;5:65–73. 2015;49:45.
13. Willett W. Nutritional epidemiology. 3rd ed. New York: Oxford Uni- 27. Field A, Willett W, Lissner L, Colditz GADietary fat and weight gain
versity Press; 2013. among women in the Nurses’ Health Study. Obesity 2007;15:967e76.
14. de la Fuente-Arrillaga C, Ruiz ZV, Bes-Rastrollo M, Sampson L, 28. Bes-Rastrollo M, van Dam RM, Martı́nez-Gonzalez MA, Li TY,
Martinez-González MA. Reproducibility of an FFQ validated in Spain. Sampson LL, Hu FB. A prospective study of dietary energy density and
Public Health Nutr 2010;13:1364–72. weight gain in women. Am J Clin Nutr 2008;88:769–77.
15. Martı́n-Moreno JM, Boyle P, Gorgojo L, Maisonneuve P, Fernandez- 29. Te Morenga L, Mallard S, Mann J. Dietary sugars and body weight:
Rodriguez JC, Salvini S, Willett WC. Development and validation of systematic review and meta-analyses of randomized controlled trials
a food frequency questionnaire in Spain. Int J Epidemiol 1993;22:512–9. and cohort studies. BMJ 2012;346:e7492.
16. Bes-Rastrollo M, Pérez VR, Sánchez-Villegas A, Alonso A, Martinez- 30. Zazpe I, Bes-Rastrollo M, Canela MR, Sánchez-Villegas A, Serrano-
Gonzalez MA. Validation of the weight and body mass index self-declared by Martı́nez M, Martı́nez-González MA. A brief assessment of eating
participants in a cohort of university graduates. Rev Esp Obes 2005;3:352–8.
habits and weight gain in a Mediterranean cohort. Br J Nutr 2011;105:
17. Martı́nez-González MA, López-Fontana C, Varo JJ, Sánchez-Villegas
765–75.
A, Martinez JA. Validation of the Spanish version of the physical ac-
31. Anand SS, Hawkes C, Souza RJ, Mente A, Dehghan M, Nugent R,
tivity questionnaire used in the Nurses’ Health Study and Health
Zulyniak MA, Weis T, Bernstein AM, Krauss RM, et al. Food con-
Professionals’ Follow-up Study. Public Health Nutr 2005;8:920–7.
sumption and its impact on cardiovascular disease: importance of so-
18. Trichopoulou A, Costacou T, Bamia C, Trichopoulos D. Adherence to
a Mediterranean diet and survival in a Greek population. N Engl J Med lutions focused on the globalized food system. J Am Coll Cardiol 2015;
2003;348:2599–608. 66:1590–614.
19. Royston P, Lambert PC. Flexible parametric survival analysis using 32. Beunza JJ, Toledo A, Hu FB, Bes-Rastrollo M, Serrano-Martı́nez M,
Stata: beyond the Cox model. College Station (TX): Stata Press; 2011. Sánchez-Villegas A, Martı́nez JA, Martı́nez-González MA. Adherence
20. Juul F, Hemmingsson E. Trends in consumption of ultra-processed to the Mediterranean diet, long-term weight change, and incident
foods and obesity in Sweden between 1960 and 2010. Public Health overweight or obesity: the Seguimiento Universidad de Navarra (SUN)
Nutr 2015;18:3096–107. cohort. Am J Clin Nutr 2010;92:1484–93.
21. Monteiro CA, Levy RB, Claro RM, de Castro IR, Cannon G. In- 33. Martı́nez-González MA, Salas-Salvado J, Estruch R, Corella D, Fitó
creasing consumption of ultra-processed foods and likely impact on M, Ros E. Benefits of the Mediterranean diet: insights form the
human health: evidence from Brazil. Public Health Nutr 2011;14:5–13. PREDIMED Study. Prog Cardiovasc Dis 2015;58:50–60.
22. Moodie R, Stuckler D, Monteiro C, Sheron N, Neal B, Thamarangsi T, 34. Martı́nez-González MA, Estruch R, Corella D, Ros E, Salas-Salvadó J.
Lincoln P, Casswell S. Profits and pandemics: prevention of harmful Prevention of diabetes with Mediterranean diets. Ann Intern Med 2014;
effects of tobacco, alcohol, and ultra-processed food and drink in- 161:157–8.
dustries. Lancet 2013;381:670–9. 35. Toledo E, Salas-Salvadó J, Donat-Vargas C, Buil-Cosiales P, Estruch R,
23. Martı́nez Steele E, Baraldi LG, Louzada MLC, Moubarac JC, Ros E, Corella D, Fitó M, Hu FB, Arós F, et al. Mediterranean diet and
Mozaffarian D, Monteiro CA. Ultra-processed foods and added invasive breast cancer risk among women at high cardiovascular risk in
sugars in the US diet: evidence from a nationally representative the PREDIMED Trial: a randomized clinical trial. JAMA Intern Med
cross-sectional study. BMJ Open 2016;6:e009892. 2015;175:1752–60.