Available via license: CC BY-NC 4.0
Content may be subject to copyright.
1
BaraldiLG, etal. BMJ Open 2018;8:e020574. doi:10.1136/bmjopen-2017-020574
Open Access
Consumption of ultra-processed foods
and associated sociodemographic factors
in the USA between 2007 and 2012:
evidence from a nationally
representative cross-sectional study
Larissa Galastri Baraldi,1,2 Euridice Martinez Steele,1,2 Daniela Silva Canella,2,3
Carlos Augusto Monteiro1,2
To cite: BaraldiLG, Martinez
SteeleE, CanellaDS, etal.
Consumption of ultra-
processed foods and associated
sociodemographic factors
in the USA between 2007
and 2012: evidence from a
nationally representative cross-
sectional study. BMJ Open
2018;8:e020574. doi:10.1136/
bmjopen-2017-020574
►Prepublication history for
this paper is available online.
To view these les, please visit
the journal online (http:// dx. doi.
org/10.1136/bmjopen-2017-
020574).
Received 14 November 2017
Revised 5 January 2018
Accepted 31 January 2018
1Department of Nutrition, School
of Public Health, University of
São Paulo, São Paulo, Brazil
2Center for Epidemiological
Studies in Health and Nutrition,
University of São Paulo, São
Paulo, Brazil
3Department of Applied
Nutrition, Institute of Nutrition,
Rio de Janeiro State University,
Rio de Janeiro, Brazil
Correspondence to
DrCarlos AugustoMonteiro;
carlosam@ usp. br
Research
ABSTRACT
Objectives To compare ultra-processed food consumption
across sociodemographic groups and over time (2007–
2008, 2009–2010, 2011–2012) in the USA.
Design Cross-sectional study.
Setting National Health and Nutrition Examination Survey
(NHANES) 2007–2012.
Participants All individuals aged ≥2 years with at least
one 24-hour dietary recall were included (n=23 847).
Main outcome measures Average dietary contribution of
ultra-processed foods (expressed as a percentage of the
total caloric value of the diet), obtained after classifying all
food items according to extent and purpose of industrial
food processing using NOVA classication.
Data analysis Linear regression was used to evaluate the
association between sociodemographic characteristics or
NHANES cycles and dietary contribution of ultra-processed
foods.
Results Almost 60% of calories consumed in the
period 2007–2012 came from ultra-processed foods.
Consumption of ultra-processed foods decreased with age
and income level, was higher for non-Hispanic whites or
non-Hispanic blacks than for other race/ethnicity groups
and lower for people with college than for lower levels of
education, all differences being statistically signicant.
Overall contribution of ultra-processed foods increased
signicantly between NHANES cycles (nearly 1% point
per cycle), the same being observed among males,
adolescents and high school education-level individuals.
Conclusions Ultra-processed food consumption in the
USA in the period 2007–2012 was overall high, greater
among non-Hispanic whites or non-Hispanic blacks, less
educated, younger, lower-income strata and increased
across time.
INTRODUCTION
Ultra-processed food and drink products
are packaged formulations resulting from
several sequences of industrial processes
(hence ‘ultra-processed’). They are manu-
factured mostly or entirely from substances
derived from foods and several additives
used to imitate sensory properties of foods
or to disguise unpalatable aspects of the final
product. They typically contain little or no
intact foods, and are ready to drink, eat or
heat up.1 2
Worldwide, the relationship between
consumption of specific ultra-processed
foods such as soft drinks and diet-related
chronic non-communicable diseases is
Strengths and limitations of this study
►Use of a large, nationally representative sample of
the US population, increasing generalisability.
►Unlike most articles which have focused on specic
food items such as soft drinks or fast food, our study
evaluates the impact of a comprehensive group of
products whose consumption is increasing rapidly
in most countries.
►Dietary data recallbias may lead to underestimation
of ultra-processed food consumption, especially
if some individuals tend to under-report these
types of food items. Should this under-reporting
have increased with time in response to a growing
awareness of health effects of ultra-processed
foods, this could result in a greater underestimation
of ultra-processed foods in later years.
►Information indicative of food processing is not
consistently determined for all food items in National
Health and Nutrition Examination Survey, which could
lead to modest overestimation or underestimation of
the consumption of ultra-processed foods.
►Social desirability bias may lead to underestimation
of ultra-processed food consumption. Should
this under-reporting have increased with time in
response to a growing awareness of the health
effects of ultra-processed foods, this could result
in a greater underestimation of ultra-processed
food consumption in later years. Differential social
desirability bias across socioeconomic groups
(should this exist) could lead to both underestimation
or overestimation of the studied associations.
group.bmj.com on March 10, 2018 - Published by http://bmjopen.bmj.com/Downloaded from
2BaraldiLG, etal. BMJ Open 2018;8:e020574. doi:10.1136/bmjopen-2017-020574
Open Access
well documented.3 4 Cross-sectional studies have also
shown an association between ultra-processed food
intake and outcomes such as obesity and metabolic
syndrome.5–7 Increased risks of obesity, hypertension and
dyslipidaemia among higher consumers of ultra-processed
foods have also been reported by cohort studies.8 9 There-
upon, the rise in obesity and chronic diseases observed in
the USA in the last decades10 11 may have been triggered
by the increase in ultra-processed food and drink product
availability12 13 and its negative impact on dietary quality.
Indeed, studies have shown that in the USA, ultra-pro-
cessed foods are the major dietary contributors of added
sugars with the mean added sugars content almost tripling
between the first and last quintiles of ultra-processed food
consumption1 14 An increase in dietary contribution of
ultra-processed foods has also been associated with both
a decrease in protein, fibre, vitamins A, C, D and E, zinc,
potassium, phosphorus, magnesium and calcium densi-
ties, and with an increase of carbohydrate, added sugar
and saturated fat contents.14
Although the USA has been recognised to be one of
the first countries to suffer a meaningful rise in ultra-pro-
cessed food and drink product market availability,12 13
there is a lack of evidence regarding recent consumption
changes and also differences among sociodemographic
strata. Thus, the aim of this study was to compare ultra-pro-
cessed food consumption across different sociodemo-
graphic groups of the US population and to describe the
recent evolution of ultra-processed food consumption in
this population.
METHODS
Study population and data collection
This study included data from three cycles of National
Health and Nutrition Examination Survey (NHANES):
2007–2008, 2009–2010 and 2011–2012. NHANES is a
continuous, nationally representative, cross-sectional
survey of non-institutionalised, civilian US residents.15
NHANES sample was obtained using a complex, multi-
stage, probability sampling design.
In the first step of the survey, all participants receive
an interviewer at home and complete questionnaires
about health history, housing characteristics and family
demographic background. Within 1 or 2 weeks, phys-
ical examinations and laboratory studies are performed
at a mobile examination centre (MEC). During these
examinations, participants complete a dietary interview
component that includes one 24-hour dietary recall.
A follow-up dietary recall is carried out 3–10 days after
the MEC examination with all examinees by telephone.
Both dietary recalls are administered by trained inter-
viewers using US Department of Agriculture Automated
Multiple-Pass Method. Portion sizes were estimated by
NHANES as further explained in NHANES Manual.16
Shortly, each MEC dietary interview room contains a
standard set of measuring guides that are used to help
the respondent report the volume and dimensions of the
food items consumed. On completion of the in-person
interview, participants are given measuring cups, spoons,
a ruler and a food model booklet to use for reporting
food amounts during the telephone interview.
For children under 9 years of age, the interview was
conducted with a proxy; for children between 6 and 8
years of age, in the presence of the child. Children aged
9–11 years provided their own data assisted by an adult
household member (assistant). The preferred proxy/
assistant was the most knowledgeable person about the
child’s consumption on the day before the interview. If
the child had more than one caregiver, several individuals
could contribute to the intake data.
Of the 10 149 people screened in NHANES 2007–2008,
9255 (91.2%) participated in the dietary interview in the
MEC and 7838 (77.2%) answered the follow-up dietary
recall. Similarly, in the second NHANES cycle (2009–
2010) of 10 537 people screened, 9754 (92.6%) and 8406
(79.8%) completed the first and the follow-up dietary
recall, respectively. In 2011–2012, of the 9756 individuals
screened, 8519 (87.3%) responded to the first dietary
recall and 7605 (78.0%) completed the second dietary
recall.
In our study, all individuals aged ≥2 years— who
completed at least one dietary recall—were considered
eligible. Pregnant women and breastfeeding mothers
were excluded, resulting in a final study sample of 23 847
individuals. For adjusted analyses, we evaluated those
individuals who had complete information on sociode-
mographic variables. Since 4307 participants had missing
values on family income and/or educational attainment,
multivariable-adjusted analysis included 19 540 individ-
uals. These individuals with missing values had similar
sociodemographic characteristics (gender, age, race/
ethnicity) and similar mean consumption of ultra-pro-
cessed foods to the full sample of interviewed participants
aged ≥2 years.
Sociodemographic variables
The sociodemographic variables considered in this study
were: gender (male and female); age group (2–9 years,
10–19 years, 20–39 years, 40–59 years, 60+ years); race/
ethnicity (American non-Hispanic white, American
non-Hispanic black, Mexican-American, other Hispanic
and other race, including Asians and multiracial).
Educational attainment of respondents for participants
aged more than 24 years and of household reference
person otherwise (less than high school, high school
and college). There were 113 households (1.51% of the
sample) where the reference person was younger than
25 years old. In these households, the education achieve-
ment of the interviewee (in all cases aged between 18 and
24 years) was adopted.
Ratio of family income to poverty categorised based on
Supplemental Nutrition Assistance Program eligibility
(low, 0.00–1.30; medium, >1.30–3.50 and high, >3.50 and
above).15 Income-to-poverty ratio is defined as ‘the
ratio of family or unrelated individual income to their
group.bmj.com on March 10, 2018 - Published by http://bmjopen.bmj.com/Downloaded from
3
BaraldiLG, etal. BMJ Open 2018;8:e020574. doi:10.1136/bmjopen-2017-020574
Open Access
appropriate poverty threshold. Ratios below 1.00 indi-
cate that the income for the respective family or unre-
lated individual is below the official definition of poverty,
while a ratio of 1.00 or greater indicates income above the
poverty level’.17
Food classication according to processing
All recorded food items from the three NHANES cycles
were classified according to NOVA (a name, not an
acronym), a food classification based on the extent and
purpose of industrial food processing.2 NOVA includes
four groups: ‘unprocessed or minimally processed foods’
(such as fresh, dry or frozen fruits or vegetables; pack-
aged grains and pulses; grits, flakes or flours made from
corn, wheat or cassava; pasta, fresh or dry, made from
flours and water; eggs; fresh or frozen meat and fish and
fresh or pasteurised milk); ‘processed culinary ingredi-
ents’ (including sugar, oils, fats, salt and other substances
extracted from foods and used in kitchens to season and
cook unprocessed or minimally processed foods and to
make culinary preparations), ‘processed foods’ (including
canned foods, sugar-coated dry fruits, salted meat prod-
ucts, cheeses and freshly made unpackaged breads, and
other ready-to-consume products manufactured with the
addition of salt or sugar or other substances of culinary
use to unprocessed or minimally processed foods) and
‘ultra-processed foods’.
Ultra-processed foods, of particular interest in this study,
include soft drinks, sweet or savoury packaged snacks,
confectionery and industrialised desserts, mass-produced
packaged breads and buns, poultry and fish nuggets and
other reconstituted meat products, instant noodles and
soups, and many other ready-to-consume formulations
of several ingredients. Besides salt, sugar, oils and fats,
these ingredients include food substances not commonly
used in culinary preparations, such as modified starches,
hydrogenated oils, protein isolates and classes of addi-
tives whose purpose is to imitate sensorial qualities of
unprocessed or minimally processed foods and their culi-
nary preparations, or to disguise undesirable qualities
of the final product. These additives include colourants,
flavourings, non-sugar sweeteners, emulsifiers, humec-
tants, sequestrants, and firming, bulking, de-foaming,
anticaking and glazing agents. Unprocessed or minimally
processed foods represent a small proportion of or are
even absent from the list of ingredients of ultra-processed
products. The rationale underlying the classification, a
detailed definition of each NOVA food group and exam-
ples of food items classified in each group has been previ-
ously published.1
For all food items (Food Codes) judged to be a hand-
made recipe, the classification was applied to the under-
lying ingredients (Standard Reference Codes or SR
Codes) obtained from the US Department of Agriculture
(USDA) Food and Nutrient Database for Dietary Studies
(FNDDS 4.0, 5.0 and 6.6). For example, for cakes, cookies
or pies underlying SR Codes were used unless underlying
SR Codes or ingredients were unlikely to be used in home
recipe (ie, ‘cellulose (alpha-cellulose, powdered cellu-
lose and poly-cellulose)’ or ‘oil, industrial, soy (partially
hydrogenated), multiuse for non-dairy butter flavour’
or ‘shortening, industrial, soybean (hydrogenated) and
cottonseed’, or ‘whey, sweet, dried’).
Other examples are ‘salsa, red, cooked, not home-
made’ and ‘salsa, red, cooked, homemade’. Food Code
‘salsa, red, cooked, not homemade’ was classified as
ultra-processed based on information from a similar
product included in the Food Code website (‘Red Gold
Salsa, Mild’: tomato concentrate (water, tomato paste),
diced tomatoes, jalapeno peppers, green chiles, yellow
chiles, vinegar, salt, dried onion, dried garlic, cilantro,
natural flavour). On the other hand, for Food Code
‘salsa, red, cooked, homemade’ it was each underlying SR
Code (salt, table; garlic, raw; onions, raw; tomatoes, red,
ripe, canned, packed in tomato juice; peppers, hot chiles,
sun dried; water, tap, drinking; vegetable oil) instead, that
was classified according to NOVA, as further explained in
a previously published paper.1
Assessing energy content
For this study, Food code energy values were used as
provided by NHANES.
For handmade recipes, the underlying ingredient
(SR Code) energy values were calculated using variables
from both FNDDS (4.0, 5.0 or 6.6) and USDA National
Nutrient Database for SR, Release 23, 24 or 25 (SR23,
SR24 or SR25) depending on the NHANES cycle.
Data analysis
Dietary intake was assessed using means of both recall
days when available and 1 day otherwise.
Food items were sorted into mutually exclusive food
subgroups within unprocessed or minimally processed
foods (n=11 subgroups), processed culinary ingredi-
ents (n=4), processed foods (n=5) and ultra-processed
foods (n=11), as shown in table 1. The dietary intake was
described according to the average absolute (total daily
energy intake) and relative (expressed as a percentage of
the total caloric value of the diet) consumption, provided
by each NOVA food group and subgroups within these
groups.
Crude and multivariate-adjusted linear regressions
were used to evaluate the association between socio-
demographic characteristics and dietary contribution
of ultra-processed foods. Tests of linear trend were
performed in order to assess the effect of age, education
and income as single continuous variables. Linear regres-
sion was also used to assess how the mean dietary share
of ultra-processed foods varied across the three studied
cycles, both overall and according to age group, gender,
education, income and ethnicity. In order to estimate the
evolution of ultra-processed food consumption, Wald tests
were performed to compare 2007–2008 and 2011–2012
values. Interaction tests were used to explore differential
effects of time across sociodemographic variables. Tests of
linear trend across the whole period were also performed.
group.bmj.com on March 10, 2018 - Published by http://bmjopen.bmj.com/Downloaded from
4BaraldiLG, etal. BMJ Open 2018;8:e020574. doi:10.1136/bmjopen-2017-020574
Open Access
Table 1 Distribution of daily energy intake according to NOVA food groups
Food groups
Mean energy intake
(kcal) SE
Mean percentage of
total energy intake SE
Unprocessed or minimally processed foods 537.3 6.4 27.4 0.3
Cereal 99.4 2.8 5.0 0.1
Meat 94.2 2.3 4.6 0.1
Milk 88.6 1.6 4.5 0.1
Poultry 67.3 1.8 3.4 0.1
Fruits 58.2 1.3 3.2 0.1
Roots and tubers 34.6 0.9 1.8 0.0
Eggs 29.8 0.7 1.5 0.0
Legumes 17.8 0.8 0.9 0.0
Fish 15.4 0.9 0.8 0.0
Vegetables 13.9 0.4 0.8 0.0
Other* 18.1 0.7 0.9 0.0
Processed culinary ingredients 85.2 1.5 4.1 0.1
Plant oils 32.1 0.9 1.5 0.9
Sugar 26.3 0.7 1.3 0.7
Animal fats 25.1 0.8 1.2 0.8
Other† 1.6 0.2 0.1 0.2
Processed foods 215.3 4.8 10.0 0.2
Cheese 79.4 1.7 3.7 0.1
Ham and other salted, smoked or canned meat
or sh 28.1 0.7 1.4 0.0
Vegetables and other plant foods preserved in
brine 14.9 0.5 0.7 0.0
Fruits in syrup, jams, marmalades 9.1 0.4 0.5 0.0
Other‡ 83.8 2.7 3.7 0.1
Ultra-processed foods 1205.4 7.8 58.5 0.3
Breads 197.6 2.6 9.9 0.1
Frozen/shelf-stable dishes§ 185.7 3.9 8.6 0.1
Confectionery 128.0 2.6 6.3 0.1
Fruit and milk drinks 114.3 1.8 5.8 0.1
Cakes, cookies and pies 123.6 2.2 5.7 0.1
Soft drinks 96.0 3.2 4.6 0.2
Saltysnacks 85.4 1.9 4.1 0.1
Breakfast cereals 56.8 1.1 3.0 0.1
Sauces, dressings and gravies 51.7 1.2 2.5 0.1
Sausages, hamburgers, reconstituted meat
products
51.0 1.3 2.4 0.1
Other¶ 115.3 1.8 5.6 0.0
All foods 2043.2 9.6 100.0
US population aged≥2 years(NHANES 2007–2012).
*Including nuts and seeds (unsalted); yeast; dried fruits (without added sugars) and vegetables; non-presweetened, non-whitened,
non-avoured coffee and tea; coconut water and meat; homemade soup and sauces; ours; tapioca.
†Including starches; coconut and milk cream; unsweetened baking chocolate, cocoa powder and gelatin powder; vinegar; baking
powder and baking soda.
‡Including salted or sugared nuts and seeds; peanut, sesame, cashew and almond butter or spread; beer and wine.
§Including pizzas and ready-to-eat dishes.
¶Including soy products such as meatless patties and sh sticks; baby food and baby formula; dips, spreads, mustard and catsup;
margarine; sugar substitutes, sweeteners and all syrups (excluding 100% maple syrup); distilled alcoholic drinks.
NHANES, National Health and Nutrition Examination Survey.
group.bmj.com on March 10, 2018 - Published by http://bmjopen.bmj.com/Downloaded from
5
BaraldiLG, etal. BMJ Open 2018;8:e020574. doi:10.1136/bmjopen-2017-020574
Open Access
NHANES sample weights were used in all analyses to
account for differential probabilities of selection for the
individual domains, non-response to survey instruments
and differences between the final sample and the total
US population. The Taylor series linearisation variance
approximation procedure was used for variance estima-
tion, in order to account for the complex sample design
and the sample weights.15
Statistical hypotheses were tested using a two-tailed
P<0.05 level of significance. Data were analysed using
Stata/SE statistical software package version 14.1.
RESULTS
Distribution of total energy intake according to NOVA food
groups
The average US daily energy intake in 2007–2012 was
2042.5 kcal, and most calories (58.5%) came from
ultra-processed foods. Unprocessed or minimally
processed foods contributed 27.5% of total energy,
processed foods an additional 10.0% and processed culi-
nary ingredients the remaining 4.0% (table 1).
Within ultra-processed foods, most calories came from
breads and frozen/shelf-stable meals (9.9% and 8.6% of
total daily intake, respectively), followed by confectionery
(6.1%), fruit and milk drinks (5.8%), cakes, cookies and
pies (5.7%), soft drinks (4.6%), salty snacks (4.1%) and
breakfast cereals (3.0%) (table 1).
All together food of animal origin accounted for more
than a half of calories of unprocessed or minimally
processed foods (approximately 15% of total calories),
with milk corresponding to 4.7% of total calories, meat
4.5%, poultry 3.4% and eggs 1.5%. Among processed
foods, most calories came from cheese (3.7% of daily calo-
ries) and ham or other salted, smoked or canned meat
or fish (1.4%). Plant oils (1.5%) and table sugar (2.3%)
were the highest calorie contributors among processed
culinary ingredients (table 1).
Association of sociodemographic characteristics with the
dietary contribution of ultra-processed foods
Except for gender, all other sociodemographic charac-
teristics were associated with the dietary contribution of
ultra-processed foods. Both the crude and adjusted contri-
bution decreased with age, education and income. The
non-adjusted contribution was highest among non-His-
panic blacks (62.1% of total energy intake), followed
by non-Hispanic whites (59.2%), Mexican-American
(57.7%), other Hispanic (53.5%) and other (49.9%).
Adjusted estimates indicated similar consumption among
non-Hispanic black and non-Hispanic white but did not
change substantially the consumption of the other ethnic
groups (table 2).
Time changes in the dietary contribution of ultra-processed
foods
As shown in table 3, the overall dietary contribution of
ultra-processed foods linearly increased across the three
cycles (P<0.05): from 57.6% in 2007–2008 to 59.7% in
2011–2012 (nearly +1% per cycle). The same statistically
significant linear time trend was found for men, adoles-
cents and respondents with high school education. These
linear trends remained unchanged after adjusting for
sociodemographic variables (data not shown). Further-
more, when comparing the last cycle against the first one,
an increase in the consumption of ultra-processed foods
was observed for all sociodemographic strata, ranging
from a minimum of +1.4% (in high income/poverty family
ratio) and a maximum of +3.6% (in other race). Signif-
icant interaction terms were found for gender-year and
age-year (P<0.05).
DISCUSSION
This study of recent and representative data of US popula-
tion shows that ultra-processed foods represented nearly
60% of dietary calories over a 6-year period. Small differ-
ences in consumption within the population reflect how
ultra-processed foods have permeated and reached all
social strata, modifying eating behaviours by displacing
handmade meals and turning ultra-processed food
consumption into an important eating pattern. Individ-
uals with a college education consumed the least ultra-pro-
cessed foods whereas adolescents and American black
and white ethnic groups were the highest consumers.
Ultra-processed food consumption was inversely associ-
ated with both age and income levels and did not vary
according to sex.
There was an overall time trend increase in ultra-pro-
cessed consumption between 2007–2008 and 2011–2012,
and more specifically among men, adolescents and
individuals with high school education. Among middle
income-to-poverty ratio individuals and non-Hispanic
white Americans, an increase in ultra-processed food
consumption was observed between the first and last
cycles.
Population-based studies evaluating ultra-processed
food consumption according to sociodemographic vari-
ables carried out in different countries obtained similar
results. In the UK (2008–2009), the average ultra-pro-
cessed food consumption was 53% of total energy intake
and decreased with age (from 58.2% among 18–29
years to 50.6% among 70+ years).18 In Canada (2004),
the average consumption of ultra-processed foods was
47.7% remaining high in all socioeconomic groups,
especially among children and adolescents (55.1%) and
less educated individuals (51.7%).19 In Chile (2010) and
Mexico (2012), ultra-processed food consumption was
much lower than in USA—28.6% and 29.8%, respec-
tively—and also decreased with age.20 21 In Chile, ultra-pro-
cessed food consumption increased with family income
from 25.8% to 30.1% of total energy intake.20 In Mexico,
ultra-processed food consumption increased with socio-
economic status (SES).21 In a cohort study carried out in
France (including participants recruited between 2009
and 2014), ultra-processed foods contributed 35.9% of
group.bmj.com on March 10, 2018 - Published by http://bmjopen.bmj.com/Downloaded from
6BaraldiLG, etal. BMJ Open 2018;8:e020574. doi:10.1136/bmjopen-2017-020574
Open Access
total energy intake and consumption was higher among
younger aged and lower educated individuals.22
While our study observed an increase in ultra-processed
food consumption between 2007 and 2012, an investiga-
tion using the alternative index of diet quality showed an
increase in US adult diet quality between 1999 and 2012.23
This stands in apparent contradiction if we take into
account results from studies on NHANES food intake and
on national US household food purchases. Indeed, these
studies have demonstrated that the dietary contribution
of ultra-processed foods is inversely associated with the
dietary content of protein, fibre, and most micronutri-
ents and directly associated with carbohydrate, saturated
fat, total sugar, added sugar and sodium contents.14 24
Still, it must be noted that more than 50% of the improve-
ment in AHEI-2010 overall score was attributed to a
reduction in trans fat intake,25 which is not inconsis-
tent with an increasing consumption of ultra-processed
foods, especially in view of the reduction in trans fat use
in industrial products. The limits of ultra-processed food
reformulation have been discussed elsewhere.26
The negative gradient between age group and
ultra-processed food consumption is consistent with
the significant positive association observed between
age and dietary quality in US adults from 1999 to
2010.23 This negative gradient may be shaped by differ-
ential age-dependent food preferences.27 Indeed,
children and adolescents may have a higher demand
of ultra-processed foods because of a greater promo-
tion through marketing and advertising among these
age groups,28 and also availability in school food
environment. The influence of the school food envi-
ronment, including school vending machines and
school stores/canteens/snack bars, on children and
adolescents’ dietary intake is well known.29 Indeed, in
recent years, the healthiness of diets among children
Table 2 Dietary contribution of ultra-processed foods according to sociodemographic variables
Variables
% of energy intake from ultra-
processed foods
Crude mean (95% CI) Adjusted mean† (95% CI)
Gender
Male 58.4(57.6to59.1) 58.3(57.6to59.0)
Female 58.6(57.9to59.4) 58.8(58.1to59.5)
Age group
2–9 65.2(64.3to66.1) 65.9(65.0to66.8)
10–19 66.7(65.9to67.5) 66.8(65.9to67.7)
20–39 59.0(58.0to60.1) 59.5(58.7to60.3)
40–59 55.2(54.0to56.3) 55.2(54.1to56.4)
≥60 53.5(52.5to54.4)* 52.8(51.9to53.7)*
Education‡
Less than high school 59.0(57.9to60.0) 59.5(58.4to60.6)
High school 59.9(59.2to60.6) 59.7(59.1to60.3)
College or higher 55.4(54.2to56.8)* 55.9(54.6to57.2)*
Family income-to-poverty ratio§
≤1.30 61.1(60.0to62.2) 59.6(58.6to60.7)
1.31–1.50 58.9(57.9to59.9) 58.7(57.8to59.7)
>1.50 56.8(55.9to57.8)* 57.7(56.9to58.6)*
Ethnicity
Non-Hispanic white 59.2(58.4to60.1) 60.2(59.4to60.9)
Non-Hispanic black 62.1(61.1to63.3) 60.6(59.7to61.5)
Mexican-American 57.7(56.7to58.8) 54.8(53.2to56.3)
Other Hispanic 53.3(51.5to55.1) 52.0(50.3to53.7)
Other 49.9(47.7to52.2) 49.6(47.3to51.8)*
US population aged≥2 years(NHANES 2007–2012).
*P for linear trend<0.05.
†Adjusted for all the other variables in the table.
‡For individuals under 25 years, the education of the household head was considered.
§Income-to-poverty ratios represent the ratio of family or unrelated individual income to their appropriate poverty threshold. Categories based
on SNAP eligibility.38
NHANES,National Health and Nutrition Examination Survey; SNAP, Supplemental Nutrition Assistance Program.
group.bmj.com on March 10, 2018 - Published by http://bmjopen.bmj.com/Downloaded from
7
BaraldiLG, etal. BMJ Open 2018;8:e020574. doi:10.1136/bmjopen-2017-020574
Open Access
and adolescents are the ones to have declined the
most. Snacks, pizzas, pastries, sweetened fruit juices
and ready-made Mexican dishes (mixed dishes with
corn or flour tortillas and corn-based dishes), most
of them ultra-processed foods, were the ones to
present the greatest caloric intake increase in the last
20 years among American children aged 2–6 years.30
In addition, in the last decades among adolescents,
a total of 17% of all calories came from fast food.24
All these previous findings are consistent with our
findings of higher ultra-processed food consumption
among younger age groups and positive time trend in
ultra-processed food intake among adolescents.
Although some studies have reported that non-His-
panic black Americans have less favourable dietary
patterns than non-Hispanic white Americans,31 in our
study both groups presented the highest consumption of
ultra-processed foods. A previous study also showed that
in most NHANES survey cycles between 1999 and 2010,
non-Hispanic whites had a significantly lower dietary
quality than Mexican-Americans adults.23 A possible
explanation for differences between non-Hispanic white
and non-Hispanic black Americans and the remaining
ethnicities may be greater promotion and advertising
of ultra-processed foods to these race/ethnicities.32
Differences may also arise from the fact that house-
holds with foreign-born reference persons tend to cook
more dinners at home than those with US-born refer-
ence person32 perhaps in an attempt to maintain their
culinary traditions or out of simple routine and not
having been fully acculturated to the fast food culture
yet. However, it is known that throughout the process
of acculturation (at least among Mexicans), migrants
progressively incorporate negative eating habits and
with the passing of the generations end up adhering to
the North American diet.33
Table 3 Time changes in the dietary contribution of ultra-processed foods (% or total energy intake) according to
sociodemographic variables
Variables 2007–2008 2009–2010 2011–2012
Pvalue for linear
trend
Gender
Male 57.3 58.8 59.7 0.0368
Female 57.9 58.9 59.6 0.1834
Age group
2–9 63.4 63.9 65.4 0.4518
10–19 64.9 67.0 68.3 0.0128
20–39 58.0 59.2 59.9 0.3529
40–59 54.2 55.0 56.3 0.3821
≥60 52.1 54.2 54.0 0.1800
Education
Less than high school 58.0 58.6 60.4 0.1632
High school 58.4 60.2 61.1 0.0122
College 54.4 55.9 56.0 0.4667
Family income-to-poverty ratio†
≤1.30 60.6 60.9 62.6 0.1910
1.31–3.50 57.7 59.5 60.1 0.0380
>3.50 56.1 57.4 57.5 0.2310
Ethnicity
Non-Hispanic white 57.9 59.4 60.4* 0.0749
Non-Hispanic black 61.6 61.1 63.6 0.1512
Mexican-American 56.4 58.4 58.3 0.0501
Other Hispanic 51.4 55.3 53.1 0.2563
Other Race 47.4 50.5 51.0 0.4002
Total 57.6 58.9 59.7 0.038
US population aged≥2 years(NHANES 2007–2012).
*P<0.05 Wald test for difference in consumption between the rst (2007–2008) and the last (2011–2012) cycles.
†Income-to-poverty ratios represent the ratio of family or unrelated individual income to their appropriate poverty threshold. Categories based
on SNAP eligibility.38
NHANES,National Health and Nutrition Examination Survey; SNAP, Supplemental Nutrition Assistance Program.
group.bmj.com on March 10, 2018 - Published by http://bmjopen.bmj.com/Downloaded from
8BaraldiLG, etal. BMJ Open 2018;8:e020574. doi:10.1136/bmjopen-2017-020574
Open Access
The small inverse association between education and
ultra-processed food consumption is consistent with
previous findings of direct association between education
and dietary quality of adults in all cycles between 1999
and 2010.23 Increased knowledge about nutrition and
concern with health may lead individuals to value nutri-
tion over taste and convenience when choosing foods
among more highly educated individuals.27
Also for income, the small inverse association with
ultra-processed food consumption is consistent with
previous findings of direct association between income
and dietary quality (AHEI-2010) of adults from 1999 to
2010.23 On the other hand, the fact that only a small nega-
tive gradient was found between SES and ultra-processed
food consumption, counters the stereotype that lower-in-
come individuals are higher consumers of ultra-processed
foods because of price. As previously highlighted in others
studies eating ‘junk food’ is not necessarily cheaper than
eating ‘real food’.34–36 Our results are also consistent with
those from a recent US study showing little evidence of
a gradient in adult fast-food consumption in regards to
income/wealth.27 If consumer’s food choices should
depend on price, income, household manager’s time
and preferences,37 our study shows that the cumulative
effect of all four did not lead to an important differen-
tial consumption of ultra-processed foods across income
strata.
SES disparities in health are thought to be partially
caused by an SES gradient in nutrition23 31 Our study
suggests that a gradient in ultra-processed food consump-
tion is probably not the main or only underlying reason
for these health inequities. Indeed, ultra-processed food
consumption was high overall, despite being slightly
higher among less educated, younger, lower income and
American non-Hispanic black and non-Hispanic white
strata. Healthier eating should be encouraged among
all sociodemographic strata especially by promoting
healthier food environments (in schools among others)
and regulating marketing.28 29
The probabilistic nature of the studied sample and
the national representativeness of the American popula-
tion are strengths of this study. Study limitations should
be considered. Recall bias may lead to underestimation
of ultra-processed food consumption, especially if some
individuals tend to under-report these types of food
items. Should this under-reporting have increased with
time in response to a growing awareness of health effects
of ultra-processed foods, this could result in a greater
underestimation of ultra-processed foods in later years.
Still, recall bias should be minimised through the use of
the five-step validated USDA’s Automated Multiple-Pass
Method. As NHANES was not specifically designed to
classify food items according to degree of processing,
misclassification errors may lead to underestimation or
overestimation of ultra-processed food consumption.
Cases of classification uncertainty were solved using a
conservative approach, opting for the lesser degree of
processing or assuming a homemade recipe, which could
have led to underestimation of ultra-processed food
consumption.
Social desirability bias may lead to underestimation of
ultra-processed food consumption. Should this under-re-
porting have increased with time in response to a growing
awareness of the health effects of ultra-processed foods,
this could result in a greater underestimation of ultra-pro-
cessed food consumption in later years. Differential social
desirability bias across socioeconomic groups (should this
exist) could lead to both underestimation or overestima-
tion of the studied associations.
Even though data from two 24-hour dietary recalls may
not represent the usual diet of individuals, these data can
be useful to estimate group means as was done in this
study. Though analyses were controlled for several socio-
demographic variables, residual confounding by variables
such as region or urban/rural area is always possible.
CONCLUSION
In this study, we show that ultra-processed food consump-
tion in the USA in the period 2007–2012 was overall high,
greater among less educated, younger, lower income and
American non-Hispanic black and non-Hispanic white
strata, and increased across time. Healthier eating should
be promoted among all sociodemographic groups, espe-
cially among children and adolescents, which have been
shown to be the highest consumers of ultra-processed
foods in several countries.
Contributors CAM and LGB designed the research. LGB and EMS took care of
data management. LGB, EMS and DSC analysed the data. LGB, EMS and CAM
wrote the paper. CAM and LGB had primary responsibility for the nal content. All
authors revised critically the nal version to be published, had full access to all of
the data (including statistical reports and tables), read and ofcially approved the
nal version. All authors absolutely agreed to be accountable for all aspects of the
study.
Funding This work was supported by Conselho Nacional de Desenvolvimento
Cientíco e Tecnológico, Edital MCTI/CNPq/Universal (Processo CNPq number
443477/2014-0) and from Fundação de Amparo à Pesquisa do Estado de São Paulo
(Processo FAPESP number 2015/14900-9).
Competing interests None declared.
Patient consent Obtained.
Ethics approval Secondary publicly available data from NHAES were used in this
study. NHANES obtained Ethics Committee/Institutional Review Board approval by
NCHS Research Ethics Review Board under Continuation of Protocol #2005-06 and
Protocol #2011-17 for 2007–2008/2009–2010 cycles and for 2011–2012 cycle,
respectively.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement This study is based on open data of the American
population that is available by Centers for Disease Control and Prevention in their
website: https://www. cdc. gov/ nchs/ nhanes/ index. htm.
Open Access This is an Open Access article distributed in accordance with the
Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which
permits others to distribute, remix, adapt, build upon this work non-commercially,
and license their derivative works on different terms, provided the original work is
properly cited and the use is non-commercial. See: http:// creativecommons. org/
licenses/ by- nc/ 4. 0/
© Article author(s) (or their employer(s) unless otherwise stated in the text of the
article) 2018. All rights reserved. No commercial use is permitted unless otherwise
expressly granted.
group.bmj.com on March 10, 2018 - Published by http://bmjopen.bmj.com/Downloaded from
9
BaraldiLG, etal. BMJ Open 2018;8:e020574. doi:10.1136/bmjopen-2017-020574
Open Access
REFERENCES
1. Martínez Steele E, Baraldi LG, Louzada ML, et al. Ultra-processed
foods and added sugars in the US diet: evidence from a nationally
representative cross-sectional study. BMJ Open 2016;6:e009892.
2. Monteiro CA, Cannon G, Moubarac JC, et al. The UN Decade of
Nutrition, the NOVA food classication and the trouble with ultra-
processing. Public Health Nutr 2018.21.
3. Micha R, Khatibzadeh S, Shi P, et al. Global, regional and national
consumption of major food groups in 1990 and 2010: a systematic
analysis including 266 country-specic nutrition surveys worldwide.
BMJ Open 2015;5:e008705.
4. Singh GM, Micha R, Khatibzadeh S, et al. Global, regional, and
national consumption of sugar-sweetened beverages, fruit juices,
and milk: a systematic assessment of beverage intake in 187
Countries. PLoS One 2015;10:e0124845.
5. Louzada ML, Baraldi LG, Steele EM, et al. Consumption of ultra-
processed foods and obesity in Brazilian adolescents and adults.
Prev Med 2015;81:9–15.
6. Canella DS, Levy RB, Martins AP, et al. Ultra-processed food
products and obesity in Brazilian households (2008-2009). PLoS One
2014;9:e92752.
7. Tavares LF, Fonseca SC, Garcia Rosa ML, et al. Relationship
between ultra-processed foods and metabolic syndrome in
adolescents from a Brazilian Family Doctor Program. Public Health
Nutr 2012;15:1–6.
8. Mendonça RD, Pimenta AM, Gea A, et al. Ultraprocessed food
consumption and risk of overweight and obesity: the University
of Navarra Follow-Up (SUN) cohort study. Am J Clin Nutr
2016;104:1433–40.
9. Rauber F, Campagnolo PD, Hoffman DJ, et al. Consumption of ultra-
processed food products and its effects on children's lipid proles: a
longitudinal study. Nutr Metab Cardiovasc Dis 2015;25:116–22.
10. Flegal KM, Carroll MD, Kit BK, et al. Prevalence of obesity and trends
in the distribution of body mass index among us adults, 1999-2010.
JAMA 2012;307:491–7.
11. Murray CJ, Atkinson C, Bhalla K, et al. The state of US health,
1990-2010: burden of diseases, injuries, and risk factors. JAMA
2013;310:591–608.
12. Hawkes C. Uneven dietary development: linking the policies and
processes of globalization with the nutrition transition, obesity and
diet-related chronic diseases. Global Health 2006;2:4.
13. Stuckler D, McKee M, Ebrahim S, et al. Manufacturing epidemics:
the role of global producers in increased consumption of unhealthy
commodities including processed foods, alcohol, and tobacco. PLoS
Med 2012;9:e1001235.
14. Martínez Steele E, Popkin BM, Swinburn B, et al. The share of ultra-
processed foods and the overall nutritional quality of diets in the
US: evidence from a nationally representative cross-sectional study.
Popul Health Metr 2017;15:6.
15. Johnson CL, Paulose-Ram R, Cl O, et al. National Health and
Nutrition Examination Survey: Analytic guidelines, 1999–2010.
National Center for Health Statistics 2013.
16. Centers for Disease Control and Prevention (CDC). NC for HS
(NCHS. National Health and Nutrition Examination Survey. Measuring
Guides for the Dietary Recall Interview. 2015. https://www. cdc. gov/
nchs/ nhanes/ measuring_ guides_ dri/ measuringguides. htm
17. U.S. Census Bureau, Population Division,Fertility & Family Statistics
Branch. Current Population Survey: denitions and explanations.
2004. http://www. census. gov/ population/ www/ cps/ cpsdef. html
(accessed 18 Aug 2017).
18. Adams J, White M. Characterisation of UK diets according to degree
of food processing and associations with socio-demographics and
obesity: cross-sectional analysis of UK National Diet and Nutrition
Survey (2008-12). Int J Behav Nutr Phys Act 2015;12:160.
19. Moubarac JC, Batal M, Louzada ML, et al. Consumption of
ultra-processed foods predicts diet quality in Canada. Appetite
2017;108:512–20.
20. Cediel G, Reyes M, da Costa Louzada ML, et al. Ultra-processed
foods and added sugars in the Chilean diet (2010). Public Health Nutr
2017.9.
21. Marrón-Ponce JA, Sánchez-Pimienta TG, Louzada M, et al.
Energy contribution of NOVA food groups and sociodemographic
determinants of ultra-processed food consumption in the Mexican
population. Public Health Nutr 2018.21.
22. Julia C, Martinez L, Allès B, et al. Contribution of ultra-processed
foods in the diet of adults from the French NutriNet-Santé study.
Public Health Nutr 2018.21.
23. Wang DD, Leung CW, Li Y, et al. Trends in dietary quality among
adults in the United States, 1999 through 2010. JAMA Intern Med
2014;174:1587–95.
24. Poti JM, Popkin BM. Trends in energy intake among US children
by eating location and food source, 1977-2006. J Am Diet Assoc
2011;111:1156–64.
25. Wang DD, Li Y, Chiuve SE, et al. Improvements In US diet helped
reduce disease burden and lower premature Deaths, 1999-2012;
Overall Diet Remains Poor. Health Aff 2015;34:1916–22.
26. Reformulation SG. fortication and functionalization: big food
corporations’ nutritional engineering and marketing strategies. J
Peasant Stud 2016;43:17–37.
27. Zagorsky JL, Smith PK. The association between socioeconomic
status and adult fast-food consumption in the U.S. Econ Hum Biol
2017;27:12–25.
28. Boyland EJ, Nolan S, Kelly B, et al. Advertising as a cue to consume:
a systematic review and meta-analysis of the effects of acute
exposure to unhealthy food and nonalcoholic beverage advertising
on intake in children and adults. Am J Clin Nutr 2016;103:519–33.
29. Kubik MY, Wall M, Shen L, et al. State but not district nutrition
policies are associated with less junk food in vending machines
and school stores in US public schools. J Am Diet Assoc
2010;110:1043–8.
30. Ford CN, Slining MM, Popkin BM. Trends in dietary intake among
US 2- to 6-year-old children, 1989-2008. J Acad Nutr Diet
2013;113:35–42.
31. Kirkpatrick SI, Dodd KW, Reedy J, et al. Income and race/ethnicity
are associated with adherence to food-based dietary guidance
among US adults and children. J Acad Nutr Diet 2012;112:624–35.
32. Virudachalam S, Long JA, Harhay MO, et al. Prevalence and patterns
of cooking dinner at home in the USA: National Health and Nutrition
Examination Survey (NHANES) 2007-2008. Public Health Nutr
2014;17:1022–30.
33. Batis C, Hernandez-Barrera L, Barquera S, et al. Food acculturation
drives dietary differences among Mexicans, Mexican Americans, and
Non-Hispanic Whites. J Nutr 2011;141:1898–906.
34. Bittman M. Is junk food really cheaper? The New York Times 2011.
http://www. nytimes. com/ 2011/ 09/ 25/ opinion/ sunday/ is- junk- food-
really- cheaper. html (accessed 6 Oct 2017).
35. Rao M, Afshin A, Singh G, et al. Do healthier foods and diet patterns
cost more than less healthy options? A systematic review and meta-
analysis. BMJ Open 2013;3:e004277.
36. Kern DM, Auchincloss AH, Robinson LF, et al. Healthy and unhealthy
food prices across neighborhoods and their association with
neighborhood socioeconomic status and proportion black/hispanic.
J Urban Health 2017;94:494–505.
37. Becker GS. A theory of the allocation of time. Econ J
1965;75:493–517.
38. U.S. Department of Health & Human Services. Poverty guidelines,
research, and measurement. Washington, DC: U.S. Department of
Health & Human Services, 2012.
group.bmj.com on March 10, 2018 - Published by http://bmjopen.bmj.com/Downloaded from
study
nationally representative cross-sectional
USA between 2007 and 2012: evidence from a
associated sociodemographic factors in the
Consumption of ultra-processed foods and
and Carlos Augusto Monteiro
Larissa Galastri Baraldi, Euridice Martinez Steele, Daniela Silva Canella
doi: 10.1136/bmjopen-2017-020574
2018 8: BMJ Open
http://bmjopen.bmj.com/content/8/3/e020574
Updated information and services can be found at:
These include:
References http://bmjopen.bmj.com/content/8/3/e020574#ref-list-1
This article cites 29 articles, 7 of which you can access for free at:
Open Access
http://creativecommons.org/licenses/by-nc/4.0/non-commercial. See:
provided the original work is properly cited and the use is
non-commercially, and license their derivative works on different terms,
permits others to distribute, remix, adapt, build upon this work
Commons Attribution Non Commercial (CC BY-NC 4.0) license, which
This is an Open Access article distributed in accordance with the Creative
service
Email alerting box at the top right corner of the online article.
Receive free email alerts when new articles cite this article. Sign up in the
Notes
http://group.bmj.com/group/rights-licensing/permissions
To request permissions go to:
http://journals.bmj.com/cgi/reprintform
To order reprints go to:
http://group.bmj.com/subscribe/
To subscribe to BMJ go to:
group.bmj.com on March 10, 2018 - Published by http://bmjopen.bmj.com/Downloaded from