Ultra-processed foods and excessive free sugar intake in the UK: A nationally representative cross-sectional study

Abstract

Objectives
To describe dietary sources of free sugars in different age groups of the UK population considering food groups classified according to the NOVA system and to estimate the proportion of excessive free sugars that could potentially be avoided by reducing consumption of their main sources.

Design and setting
Cross-sectional data from the UK National Diet and Nutrition Survey (2008–2014) were analysed. Food items collected using a 4-day food diary were classified according to the NOVA system.

Participants
9364 individuals aged 1.5 years and above.

Main outcome measures
Average dietary content of free sugars and proportion of individuals consuming more than 10% of total energy from free sugars.

Data analysis
Poisson regression was used to estimate the associations between each of the NOVA food group and intake of free sugars. We estimated the per cent reduction in prevalence of excessive free sugar intake from eliminating ultra-processed foods and table sugar. Analyses were stratified by age group and adjusted for age, sex, ethnicity, survey year, region and equivalised household income (sterling pounds).

Results
Ultra-processed foods account for 56.8% of total energy intake and 64.7% of total free sugars in the UK diet. Free sugars represent 12.4% of total energy intake, and 61.3% of the sample exceeded the recommended limit of 10% energy from free sugars. This percentage was higher among children (74.9%) and adolescents (82.9%). Prevalence of excessive free sugar intake increased linearly across quintiles of ultra-processed food consumption for all age groups, except among the elderly. Eliminating ultra-processed foods could potentially reduce the prevalence of excessive free sugar intake by 47%.

Conclusion
Our findings suggest that actions to reduce the ultra-processed food consumption generally rich in free sugars could lead to substantial public health benefits.

Full text

1
RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
Ultra-processed foods and excessive free
sugar intake in the UK: a nationally
representative cross-sectional study
Fernanda Rauber ,1,2 Maria Laura da Costa Louzada,2,3
Euridice Martinez Steele,1,2 Leandro F M de Rezende,2,4 Christopher Millett,2,5
Carlos A Monteiro,1,2 Renata B Levy2,6
To cite: RauberF,
LouzadaMLdC, Martinez
SteeleE, etal. Ultra-processed
foods and excessive free
sugar intake in the UK: a
nationally representative cross-
sectional study. BMJ Open
2019;9:e027546. doi:10.1136/
bmjopen-2018-027546
►Prepublication history and
additional material for this
paper are available online. To
view these les, please visit
the journal online (http:// dx. doi.
org/ 10. 1136/ bmjopen- 2018-
027546).
Received 27 October 2018
Revised 06 September 2019
Accepted 12 September 2019
For numbered afliations see
end of article.
Correspondence to
Dr Fernanda Rauber;
rauber. fernanda@ gmail. com
Original research
© Author(s) (or their
employer(s)) 2019. Re-use
permitted under CC BY-NC. No
commercial re-use. See rights
and permissions. Published by
BMJ.
ABSTRACT
Objectives To describe dietary sources of free sugars in
different age groups of the UK population considering food
groups classied according to the NOVA system and to
estimate the proportion of excessive free sugars that could
potentially be avoided by reducing consumption of their
main sources.
Design and setting Cross-sectional data from the UK
National Diet and Nutrition Survey (2008–2014) were
analysed. Food items collected using a 4-day food diary
were classied according to the NOVA system.
Participants 9364 individuals aged 1.5 years and above.
Main outcome measures Average dietary content of free
sugars and proportion of individuals consuming more than
10% of total energy from free sugars.
Data analysis Poisson regression was used to estimate
the associations between each of the NOVA food group
and intake of free sugars. We estimated the per cent
reduction in prevalence of excessive free sugar intake
from eliminating ultra-processed foods and table sugar.
Analyses were stratied by age group and adjusted for
age, sex, ethnicity, survey year, region and equivalised
household income (sterling pounds).
Results Ultra-processed foods account for 56.8% of total
energy intake and 64.7% of total free sugars in the UK
diet. Free sugars represent 12.4% of total energy intake,
and 61.3% of the sample exceeded the recommended
limit of 10% energy from free sugars. This percentage
was higher among children (74.9%) and adolescents
(82.9%). Prevalence of excessive free sugar intake
increased linearly across quintiles of ultra-processed food
consumption for all age groups, except among the elderly.
Eliminating ultra-processed foods could potentially reduce
the prevalence of excessive free sugar intake by 47%.
Conclusion Our ndings suggest that actions to reduce
the ultra-processed food consumption generally rich in
free sugars could lead to substantial public health benets.
INTRODUCTION
Excessive consumption of free sugar is asso-
ciated with obesity, type 2 diabetes, dental
caries and several other health outcomes.1–4
To address this associated health burden, the
WHO5 recommends that free sugars should
be reduced to less than 10% of total energy
intake and also suggests a level below 5% to
obtain additional health benefits, such as
reduction of dental caries. Achievement of
this ambitious target will require bold and
systematic efforts to reduce sugar across a
variety of food products in most settings.
As defined by the NOVA food classification
system, ultra-processed foods are industrial
formulations of many ingredients, mostly
of exclusive industrial use, that result from
a sequence of industrial processes (hence
ultra-processed).6 In some high-income coun-
tries, including the UK, ultra-processed foods
account for more than half of total dietary
energy intake.7–9 Importantly, national
dietary surveys conducted in high-income
and middle-income countries8–12 have shown
a strong and positive association between
consumption of ultra-processed foods and
excessive dietary added (or free) sugar intake.
Free sugars include sugars added to foods by
the manufacturer, cook and consumer, plus
sugars naturally present in honey, syrups and
fruit juices,5 while added sugars captures all
free sugars, but exclude naturally occurring
sugars in fruit juices.
Strengths and limitations of this study
►Use of a large and nationally representative sample
of the UK population, increasing generalisability.
►Use of data on free sugars rather than total sugars
or sugar-sweetened beverages, which correspond
to the guidelines relevant area of prioritisation.
►Use of NOVA system that has been recognised as a
valid tool for public health and nutrition research and
policy by international organisations.
►Dietary data obtained by food diaries are subject to
potential error and bias.
►UK national dietary survey collects limited informa-
tion indicative of food processing (eg, place of meals
and product brands), which may lead to misclassi-
cation of food items.
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from

2RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
Free sugar intake in the UK is high, ranging from 11%
to 15% of total energy intake.13 To address this, the UK
has implemented a number of measures including a
sugar-sweetened beverage levy in 2018. However, action
on sugar-sweetened beverages alone is unlikely to reduce
population level sugar intake to WHO-recommended
levels. In a more recent publication, the voluntary sugar
reduction programme continues being endorsed by the
government, but other measures such as restriction of
advertising and in-store promotions of some sugary foods
are also being considered as strategies to reduce child-
hood obesity.14 A better understanding of the key sources
of sugar intake in the UK diet is required to inform policy
development. This study aims to describe the dietary
sources of free sugars in different age groups of the UK
population taking into account food groups classified
according to the NOVA classification system and estimate
the proportion of excessive free sugars that could be
potentially avoided by reducing the consumption of their
main dietary sources.
METHODS
Data source and collection
We used data from the National Diet and Nutrition Survey
Rolling Programme (NDNS) years 1–6 (2008/2009–
2009/2010, 2010/2011–2011/2012, 2012/2013–
2013/2014) combined, which is a cross-sectional survey of
people aged 1.5 years or older. The survey was designed
to be representative of the UK population and provides
comprehensive information on food intake. Details of the
rationale, design and methods of the survey have been
described elsewhere.15 Briefly, the sample was drawn from
households randomly selected from the UK Postcode
Address File, a list of all UK addresses. One adult (aged
19 years and older) and one child (aged 1.5–18 years), if
available, were randomly selected from each household.
Only a child was selected from some households to be
part of a ‘child boost’ to ensure approximately equal
numbers of children and adults. Participants (or in the
case of children ≤11 years, their parent/carer) completed
a 4-day food diary and participated in an interview that
included data on socio-demographic status.
Participants were asked to report all foods and drinks
consumed both within and outside the home. Portion
sizes were estimated using household measures or
weights from packaging. Once completed, diaries were
checked by interviewers with respondents and missing
details added to improve completeness. Diary days were
randomly selected to ensure balanced representation of
all days of the week. All individuals who completed 3 or
4 days of dietary recording were eligible for inclusion in
the study, giving a sample size of 9374 (4738 adults and
4636 children) participants for years 1 to 6 (2008/2009 to
2013/2014) combined.
The food intake data from completed records were
coded and edited using the software DINO (Diet In
Nutrients Out) and food and nutrient intakes estimated
using nutrient composition data from the Department
of Health’s Nutrient Databank, updated for each survey
year.16 17 Free sugars are defined as sugars added to foods
by the manufacturer, cook or consumer, plus sugars
naturally present in honey, syrups, fruit juices and fruit
concentrates.5 Intakes in the UK NDNS years 1–6 were
expressed as non-milk extrinsic sugars (NMES). The
term NMES captures all sugars defined by the term free
sugars while also including half of the sugars present in
dried, stewed or canned fruit. The NMES values could be
slightly higher in some cases than the free sugar values,
mostly in the non-ultra-processed food group since the
term free sugar does not include sugars contributed by
dried and processed fruits. Based on the assumption that
those definitions are sufficiently similar for assessment
and monitoring purposes,1 3 this study used the term free
sugars.
Computerised raw data files and documentation from
this survey were obtained under license from the UK Data
Archive (http://www. esds. ac. uk).
Food classication according to processing
We classified all recorded food items according to NOVA,
a food classification system based on the nature, extent
and purpose of the industrial food processing.6 This
classification includes four groups: (1) unprocessed or
minimally processed foods (eg, fresh, dry or frozen fruits
or vegetables; grains, flours and pasta; pasteurised or
power plain milk, plain yoghurt, fresh or frozen meat);
(2) processed culinary ingredients (eg, table sugar, oils,
butter and salt); (3) processed foods (eg, vegetables in
brine, cheese, simple breads, fruits in syrup, canned fish)
and (4) ultra-processed foods (eg, soft drinks, sweet or
savoury packaged snacks, confectionery; packaged breads
and buns; reconstituted meat products and preprepared
frozen or shelf-stable dishes) (see online supplementary
table S1). The detailed description of NOVA classification
can be found elsewhere.6 18
All foods in NDNS are coded as food number and
grouped into subsidiary food groups (n=155). When
possible, subsidiary food groups were directly classified
according to NOVA (see online supplementary table S2).
When foods within a subsidiary food group pertained
to different NOVA groups (n=52), it was the food codes
instead of the group, which were individually classified. By
doing so, we were able to classify each underlying ingre-
dient of homemade dishes in its corresponding NOVA
group. Subsidiary food groups as classified by NOVA are
described in the online supplementary table S2.
Although the NDNS database was provided with most
food items systematically disaggregated into their indi-
vidual components, about 4% of composite food codes
were still mixed dishes compiled from two or more
single-ingredient food code.19 The method we adopted
to disaggregate food codes has been described previ-
ously.19 Using the core sample of years 1 to 4 (2008/2009
to 2011/2012) (n=4125), we estimated that composite
food codes represented only 3% of total calories. In
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from

3
RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
this case, dishes were categorised according to the main
constituent ingredient. Dishes in which a main constit-
uent ingredient was not clearly identified (eg, chicken
and vegetable soup) were classified as a specific subgroup
of freshly prepared dishes based on one or more unpro-
cessed or minimally processed food (group 1). Non-ca-
loric supplements were not included in the analyses.
Covariates
Covariates included were age (years), sex, ethnicity
(White, mixed ethnic group, Black or Black British, Asian
or Asian British and other race), region (England North,
England Central/Midlands, England South (including
London), Scotland, Wales and Northern Ireland), survey
year (years 1–6) and equivalised household income
(equivalised for different household sizes and composi-
tion using the McClements equivalence scale15). Due to
the significant proportion of missing values for the equiv-
alised household income (12.8%), we applied multiple
imputation by chained equation method based on age,
sex, ethnicity, excessive free sugar intake and ultra-pro-
cessed food consumption. Multiple imputation was
performed 20 times, and the Monte Carlo error analysis
showed good statistical reproducibility of the results.20
We used the average of estimates from each imputed data
set. Sensitivity analysis was conducted comparing findings
from imputed data and complete case analysis.
Data analysis
For each survey day and age group (1.5–10 years, 11–18
years, 19–64 years and ≥64 years), we defined extreme
total energy intake outliers as values below the 1st and
above the 99th percentiles21 (see online supplementary
figure S1). Based on these criteria, we excluded 10 individ-
uals who had all days of food diary classified as outliers. In
total, 9364 (4729 adults and 4635 children) participants
were eligible for inclusion in the analyses and more than
91% completed the 4-day food diary. We used the mean
of all available days of food diary for each individual.
Food items were sorted into mutually exclusive food
groups according to NOVA classification. We combined
the group of unprocessed or minimally processed foods
with the group of processed culinary ingredients, as foods
belonging to these two groups are usually mixed together
in culinary preparations and, therefore, consumed
together. Thus, we performed the analyses consid-
ering three groups of foods: unprocessed or minimally
processed foods and processed culinary ingredients (indi-
viduals are able to determine the amount of table sugars
they add), processed foods (sugar added by the food
industry) and ultra-processed foods (sugar added by the
food industry).
First, we estimated the distribution of total energy and
free sugar intake according to the food groups. Then, we
calculated the mean free sugar intake of the overall diet
and the prevalence of excessive intake of free sugars. We
used the WHO recommendations5 to assess the excessive
intake of free sugars (≥10% of total energy). Analyses
using the UK recommendations to further limit free
sugar intake to less than 5% of total energy intake are
presented in a supplementary table (online supplemen-
tary table S3). Analyses were carried out for the entire
population and also stratified by age group.
Next, the prevalence of excessive intake of free sugars
(≥10% of total energy) was compared across quintiles
of the energy share provided by each of the three food
groups. Poisson regression was used to estimate prev-
alence ratios (PRs) and 95% CIs for the associations
between each of the three NOVA food group quintiles
and prevalence of individuals consuming more than 10%
of total energy from free sugars. Tests of linear trend were
performed to evaluate the quintiles as a single contin-
uous variable. All analyses were stratified by age group.
Multiple regression models were also performed to adjust
for age, sex, ethnicity, region, survey year and equivalised
household income (sterling pounds). Analyses using the
entire population are presented in a supplementary table
(online supplementary table S4). We also evaluated the
extent to which the association between the exposure
(dietary contribution of NOVA food groups) and the
dietary content in free sugars changed according to the
survey year, by including a multiplicative interaction term
(survey year×dietary contribution of NOVA food groups)
in the fully adjusted models.
Finally, we estimated the proportion of excessive free
sugar intake that could be potentially avoided if exposure
to the risk factors was eliminated (theoretical minimum
risk exposure level scenarios).22 23 The counterfactual
scenarios were defined considering the main dietary
sources of free sugars. The first counterfactual scenario
assumed no consumption of ultra-processed food (poten-
tially hidden sugars), while in the second scenario table
sugar consumption was set to zero. Table sugar included
honey, molasses, maple syrup (100%) and sugar added
to coffee/juice and homemade dishes (potentially sugar
that can be measured by the consumer). Examples of
homemade dishes include biscuits, fruit pies, buns, cakes
and pastries, cereal-based milk puddings and sponge
pudding (see online supplementary table S2).
In both scenarios, we first calculated the prevalence of
excessive free sugar intake in the UK population (Ppopu-
lation). We then estimated the predicted prevalence of
excessive free sugar intake that would be expected had
the consumption of each of these main sources of free
sugars being zero (Pnon-exposed). Lastly, we calculated the
proportion of excessive free sugar intake that could be
potentially avoided in each scenario using the following
formula: (Ppopulation − Pnon-exposed)/Ppopulation. Prevalences were
adjusted for sex, age, ethnicity, region, survey year and
household income. To test more feasible scenarios, we
also estimated the per cent reduction in prevalence of
excessive free sugar intake from reducing the consump-
tion of ultra-processed foods and table sugar by 50% (see
online supplementary figure S2).
NDNS study weights were used in all analyses to account
for sampling and non-response error. All statistical
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from

4RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
analyses were carried out using Stata Statistical Soft-
ware V.14. The p values reported were two-tailed, and a
threshold of <0.05 was considered for statistically signifi-
cant associations.
Patient and public involvement
Patients and/or public were not involved in the design or
conduct of this study.
RESULTS
Ultra-processed foods account for 56.8% of total
energy intake and 64.7% of total free sugars in the UK
diet. Unprocessed or minimally processed foods and
processed culinary ingredients represented an additional
34.3% of total energy intake and 23.8% of free sugars,
and processed foods the remaining 8.8% of total energy
intake and 11.5% of free sugars. Ultra-processed foods
accounted for a higher percentage of total energy intake
among children (63.5%) and adolescents (68%). The
average UK daily intake of free sugars was 12.4% (SE 0.1)
of total energy intake and 61.3% of British exceeded the
recommended limit of 10% energy from free sugars. This
proportion was even higher among children (74.9%) and
adolescents (82.9%) (table 1).
No significant interaction was observed between the
exposure and the survey year for the total energy intake
from free sugars (unprocessed or minimally processed
foods+processed culinary ingredients: p=0.254; processed
foods: p=0.538; ultra-processed foods: p=0.137) nor for
the prevalence of excessive intake of free sugars (unpro-
cessed or minimally processed foods+processed culinary
ingredients: p=0.609; processed foods: p=0.262; ultra-pro-
cessed foods: p=0.258). Even so, we included variable
survey year (1–6) in the adjusted model.
Indicators of the dietary content in free sugars
according to quintiles of the dietary contribution of
NOVA food groups stratified by age groups are shown in
tables 2–5 (1.5–10 years, 11–18 years, 19–64 years and ≥64
years, respectively). The dietary contents of free sugars
increased linearly across quintiles of ultra-processed food
consumption for children (from 10.4% in the lowest
quintile to 15.3% in the highest quintile), adolescents
(from 12.7% to 17.4%, respectively) and adults (from
9.6% to 15.2%, respectively), whereas the increase for
elderly was not significant (from 10.6% to 11.7%, respec-
tively). The prevalence of excessive free sugar intake
also increased linearly across quintiles of ultra-processed
food consumption for children, adolescents and adults.
Children in the highest quintiles of ultra-processed food
consumption had a prevalence of excessive free sugar
intake 60% higher (PRadj 1.6; 95% CI 1.3 to 1.9) than
those in the lowest quintile group. The same trend was
observed for adolescents (PRadj 1.6; 95% CI 1.2 to 1.9)
and adults (PRadj 1.7; 95% CI 1.5 to 1. 9). Although no
linear trend was found between quintiles of ultra-pro-
cessed food consumption and excessive free sugar intake
among elderly (p>0.05), the fourth quintile group had
a prevalence of excessive free sugar intake 35% higher
(PRadj 1.3; 95% CI 1.1 to 1.7) than those in the lowest
quintile group.
Opposite trends were observed for the group of unpro-
cessed or minimally processed foods and processed culi-
nary ingredients, where the prevalence of excessive free
sugar intake decreased from the first to the last quintile
of these food groups in all age groups. The prevalence of
excessive free sugar intake also decreased from the first
to the last quintile of processed foods, but only in adoles-
cents and adults.
Sensitivity analysis performed by considering complete
cases only indicated that the results of the multiple impu-
tations did not differ significantly from the complete case
analysis (data not shown).
In our counterfactual scenarios, we calculated the
percentage of excessive free sugar intake avoided if the
consumption of ultra-processed foods and table sugar
was zero (figure 1). We estimated that about 47% of the
prevalence of excessive free sugar intake in the UK popu-
lation could be potentially avoided if the consumption of
ultra-processed foods was eliminated. Eliminating table
sugar could potentially avoid 9.4% of the prevalence of
excessive free sugar intake. This greater reduction in the
percentage of excessive free sugar intake due to elimina-
tion of ultra-processed foods, relative to table sugar, was
observed in all age groups, except in the elderly group
where both scenarios had similar impacts on total free
sugar intake. For the more feasible scenario, we found a
similar trend where a greater reduction in the percentage
of excessive free sugar intake due to a 50% reduction of
ultra-processed foods, relative to table sugar, was observed
in all age groups, except in the elderly group (see online
supplementary figure S2).
DISCUSSION
In this large, nationally representative sample of the UK
population, higher consumption of ultra-processed food
was associated with greater dietary content of free sugars
in children, adolescents and adults. Using theoretical
minimum risk exposure level scenarios, we also showed
that by eliminating ultra-processed food consumption,
the prevalence of excessive free sugar intake (10% or
more of total energy intake) could be potentially reduced
from 60% to 31%. In children and adolescents, the poten-
tial reduction could be from 74% to 45% and from 83%
to 53%, respectively.
Our findings confirm an excessive consumption of free
sugars in the UK diet13 and show that ultra-processed
foods contributed to nearly 65% of all free sugars in all
age groups and nearly 80% in children and adolescents.
Unprocessed or minimally processed foods (mostly fresh
juice) and processed culinary ingredients (mostly table
sugar) contributed between 19% and 27% of the dietary
content of free sugars, while processed foods provided
the lowest contribution in all age groups.
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from

5
RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
Table 1 Dietary contribution of NOVA food groups and indicators of the dietary content in free sugars according to age groups in the UK population aged 1.5 years or
over (2008–2014)
Age
groups
Dietary contribution (% of total energy intake) % of total energy intake from free sugars
Individuals with ≥10% of
total energy intake from
free sugars
Unprocessed or
minimally processed
foods+processed
culinary ingredients
Processed
foods
Ultra-
processed
foods
Unprocessed or
minimally processed
foods+processed
culinary ingredients
Processed
foods
Ultra-
processed
foods Total Overall diet
Mean SE Mean SE Mean SE Mean SE Mean SE Mean SE Mean SE % 95% CI
1.5–10
years
31.96 0.33 4.51 0.10 63.53 0.34 18.82 0.45 5.15 0.22 76.03 0.49 14.00 0.14 74.94 72.78 to 76.99
11–18
years
27.25 0.37 4.75 0.16 68.00 0.40 18.63 0.55 2.48 0.19 78.89 0.57 15.78 0.19 82.91 80.72 to 84.90
19–64
years
34.75 0.32 10.37 0.19 54.89 0.35 24.68 0.50 12.96 0.38 62.36 0.56 11.93 0.14 56.59 54.47 to 58.68
≥65
years
38.57 0.49 8.45 0.29 52.98 0.52 26.77 0.96 15.38 0.69 57.86 1.01 11.36 0.23 56.83 52.98 to 60.59
Total 34.35 0.22 8.83 0.13 56.82 0.24 23.78 0.36 11.46 0.27 64.75 0.40 12.44 0.10 61.27 59.76 to 62.76
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from

6RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
Table 2 Indicators of the dietary content in free sugars according to quintiles of the dietary contribution of NOVA food groups in the UK population aged 1.5–10 years
(2008–2014)
Dietary contribution (% of total energy intake)
% of total energy intake from
free sugars Individuals with ≥10% of total energy intake from free sugars
Quintile Mean Min Max Mean SE % PR* PRadj† 95% CI
Unprocessed or minimally processed foods+processed culinary ingredients
First 15.36 0.00 20.92 15.80 0.33 82.99 1.00 1.00 –
Second 24.86 20.93 28.41 14.60 0.30 79.62 0.96 0.95 0.89 to 1.02
Third 31.57 28.46 34.96 14.37 0.28 81.68 0.98 0.99 0.93 to 1.06
Fourth 39.30 34.98 43.86 13.66 0.36 73.40 0.88 0.91 0.84 to 0.99
Fifth 52.46 43.97 79.93 11.13‡ 0.26 53.87 0.65‡ 0.69‡ 0.61 to 0.78
Processed foods
First 0.41 0.00 1.33 13.93 0.29 72.58 1.00 1.00 –
Second 2.56 1.34 3.79 14.82 0.30 80.23 1.11 1.11 1.03 to 1.19
Third 5.18 3.79 6.82 13.77 0.25 73.85 1.02 1.04 0.95 to 1.13
Fourth 8.96 6.83 11.95 13.37 0.31 73.23 1.01 1.02 0.93 to 1.12
Fifth 16.05 12.04 41.71 13.16 0.52 69.20 0.95 0.99 0.86 to 1.14
Ultra-processed foods
First 36.38 15.11 43.67 10.35 0.38 46.41 1.00 1.00 –
Second 49.00 43.72 53.03 12.37 0.30 66.78 1.44 1.39 1.15 to 1.70
Third 57.17 53.06 60.95 13.84 0.37 74.22 1.60 1.50 1.24 to 1.81
Fourth 65.58 60.96 70.14 14.48 0.26 80.95 1.74 1.62 1.35 to 1.95
Fifth 78.05 70.15 100 15.32‡ 0.25 81.41 1.75‡ 1.62‡ 1.35 to 1.95
*PR=prevalence ratios estimated using Poisson regression.
†PRadj=prevalence ratios adjusted for sex, age, race/ethnicity (White, mixed ethnic group, Black or Black British, Asian or Asian British and other race), region, survey year and household
income.
‡Signicant linear trend across all quintiles (p≤0.01).
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from

7
RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
Table 3 Indicators of the dietary content in free sugars according to quintiles of the dietary contribution of NOVA food groups in the UK population aged 11–18 years
(2008–2014)
Dietary contribution (% of total energy intake)
% of total energy intake from
free sugars Individuals with ≥10% of total energy intake from free sugars
Quintile Mean Min Max Mean SE % PR* PRadj† 95% CI
Unprocessed or minimally processed foods+processed culinary ingredients
First 14.43 0.00 20.89 17.28 0.36 88.89 1.00 1.00 –
Second 24.61 20.92 28.43 15.87 0.35 84.3 0.95 0.95 0.89 to 1.01
Third 31.46 28.44 34.93 15.5 0.37 81.82 0.92 0.92 0.86 to 0.99
Fourth 39.24 34.98 43.84 13.96 0.43 78.15 0.88 0.89 0.82 to 0.96
Fifth 52.96 43.88 79.86 13.60‡ 0.8 66.92 0.75‡ 0.77‡ 0.66 to 0.88
Processed foods
First 0.29 0.00 1.33 17.18 0.41 85.11 1.00 1.00 –
Second 2.56 1.34 3.79 15.81 0.35 81.74 0.96 0.96 0.9 to 1.03
Third 5.16 3.8 6.81 15.62 0.35 86.87 1.02 1.02 0.96 to 1.09
Fourth 8.94 6.82 11.95 14.52 0.43 79.4 0.93 0.93 0.86 to 1.01
Fifth 17.53 12.05 41.62 13.68‡ 0.57 74.89 0.88§ 0.87§ 0.78 to 0.99
Ultra-processed foods
First 35.29 18.4 42.94 12.72 1.39 56.18 1.00 1.00 –
Second 49.35 43.7 53.03 13.65 0.56 75.73 1.35 1.34 1.03 to 1.74
Third 56.91 53.08 60.96 14.19 0.4 79.24 1.41 1.4 1.09 to 1.8
Fourth 65.63 60.96 70.13 14.99 0.32 80.76 1.44 1.42 1.11 to 1.82
Fifth 79.05 70.14 100 17.37‡ 0.29 89.04 1.58‡ 1.56‡ 1.23 to 1.99
*PR=prevalence ratios estimated using Poisson regression.
†PRadj=prevalence ratios adjusted for sex, age, race/ethnicity (White, mixed ethnic group, Black or Black British, Asian or Asian British and other race), region, survey year and household
income.
‡Signicant linear trend across all quintiles (p≤0.01).
§Signicant linear trend across all quintiles (p≤0.05).
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from

8RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
Table 4 Indicators of the dietary content in free sugars according to quintiles of the dietary contribution of NOVA food groups
in the UK population aged 19–64 years (2008–2014)
Dietary contribution (% of total energy
intake)
% of total energy
intake from free
sugars
Individuals with ≥10% of total energy intake from
free sugars
Quintile Mean Min Max Mean SE % PR* PRadj† 95% CI
Unprocessed or minimally processed foods+processed culinary ingredients
First 15.06 0.00 20.92 15.11 0.36 35.87 1.00 – –
Second 24.93 20.95 28.41 12.87 0.31 31.12 0.85 0.87 0.79 to 0.96
Third 31.65 28.43 34.96 11.97 0.31 30.87 0.79 0.85 0.77 to 0.94
Fourth 38.95 34.97 43.88 11.01 0.28 28.45 0.66 0.72 0.64 to 0.8
Fifth 54.24 43.93 91.9 9.89‡ 0.25 25.28 0.57‡ 0.62‡ 0.55 to 0.71
Processed foods
First 0.28 0.00 1.32 13.09 0.5 59.14 1.00 1.00 –
Second 2.6 1.34 3.79 12.82 0.41 60.65 1.03 1.04 0.92 to 1.19
Third 5.35 3.79 6.82 12.17 0.3 61.42 1.04 1.04 0.92 to 1.18
Fourth 9.36 6.82 12.03 11.62 0.26 55.92 0.95 0.98 0.87 to 1.11
Fifth 19.8 12.04 65.22 11.27‡ 0.22 52.47 0.89‡ 0.92‡ 0.82 to 1.03
Ultra-processed foods
First 34.45 1.82 43.67 9.62 0.27 39.42 1.00 1.00 –
Second 48.7 43.69 53.04 11.11 0.25 53.34 1.35 1.3 1.13 to 1.5
Third 57.08 53.06 60.96 11.83 0.29 56.84 1.44 1.37 1.19 to 1.57
Fourth 65.34 60.96 70.14 13.09 0.32 66.31 1.68 1.57 1.37 to 1.79
Fifth 78.04 70.15 100 15.21‡ 0.38 74.3 1.88‡ 1.67‡ 1.46 to 1.92
*PR=prevalence ratios estimated using Poisson regression.
†PRadj=prevalence ratios adjusted for sex, age, race/ethnicity (White, mixed ethnic group, Black or Black British, Asian or Asian British and
other race), region, survey year and household income.
‡Signicant linear trend across all quintiles (p≤0.01).
Our findings are similar to previous studies conducted
in high-income and middle-income countries that have
shown strong associations between the intake of ultra-pro-
cessed foods and the dietary content of free sugars.8–11 A
previous study conducted in Chile similarly showed that
the association between ultra-processed food consump-
tion and the dietary content of added sugars is more
pronounced among children and adolescents.12 In our
study, there was no linear association between ultra-pro-
cessed food consumption and dietary content of free
sugars among the elderly. Although the prevalence of
excessive free sugar intake was higher in the fourth
with regard to the first quintile of ultra-processed food
consumption, the prevalence in the highest quintile
group was not different from the first. A possible expla-
nation for this finding could be changes in the composi-
tion of different types of ultra-processed across quintiles
in the elderly. Actually, while in the overall population,
ultra-processed sweetened products such as soft/fruit
drinks, confectionery, milk-based drinks and biscuits
monotonically increased across quintiles (from 18% to
23% of the total calories from ultra-processed foods), in
the elderly a drop in consumption was observed between
the fourth and fifth quintiles (from 18% to 15%) (data
not shown).
There is strong evidence that the high consump-
tion of free sugars contributes to excess obesity, type 2
diabetes, dyslipidaemia, hypertension and coronary heart
disease.2–4 Consequently, most dietary recommendations
now advise limiting free sugar intake, but more focused
efforts are needed to put this recommendation into prac-
tice. Changing personal behaviour and choice alone is
not an effective or realistic option as our findings confirm
that the majority of free sugar is added to food before
it is marketed and sold. Voluntary agreements between
industry and government have been shown repeatedly
to be ineffective in improving public health.24 This is
confirmed by recent UK experience where the early
stages of the government’s sugar reduction programme,
which challenged the food industry to voluntarily cut
sugar in some products, have produced only slow progress
towards proposed targets.25 Thus, more drastic measures
that change the availability, price and marketing of these
products are necessary.
The analyses presented here suggest that actions to
reduce the consumption of ultra-processed foods often
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from

9
RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
Table 5 Indicators of the dietary content in free sugars according to quintiles of the dietary contribution of NOVA food groups
in the UK population aged 65 years or over (2008–2014)
Dietary contribution (% of total
energy intake)
% of total energy intake from free
sugars
Individuals with ≥10% of total energy intake from
free sugars
Quintile Mean Min Max Mean SE % PR* PRadj† 95% CI
Unprocessed or minimally processed foods+processed culinary ingredients
First 16.63 6.34 20.82 11.67 0.87 56.16 1.00 1.00 –
Second 25.04 20.95 28.36 12.83 0.61 67.39 1.2 1.19 0.9 to 1.57
Third 32.06 28.44 34.9 11.98 0.48 64.37 1.15 1.15 0.87 to 1.52
Fourth 39.3 34.98 43.85 10.93 0.44 53.96 0.96 0.97 0.73 to 1.28
Fifth 52.26 43.89 78.36 10.7 0.42 50.94 0.91‡ 0.91‡ 0.69 to 1.21
Processed foods
First 0.38 0.00 1.32 0.72 43.52 1.00 1.00 –
Second 2.42 1.34 3.78 9.7 0.56 64.3 1.48 1.49 1.14 to 1.96
Third 5.23 3.79 6.81 12.16 0.45 65 1.49 1.52 1.17 to 1.98
Fourth 9.27 6.82 12.02 11.1 0.47 54.46 1.25 1.27 0.96 to 1.67
Fifth 19.1 12.04 50.86 11.23 0.46 53.62 1.23 1.29 0.97 to 1.69
Ultra-processed foods
First 35.98 7.79 43.69 10.63 0.49 47.63 1.00 1.00 –
Second 48.67 43.74 53.02 11.3 0.48 58.67 1.23 1.2 0.97 to 1.47
Third 56.97 53.05 60.91 11.61 0.45 59.89 1.26 1.21 0.98 to 1.5
Fourth 64.99 61.01 70.08 12.01 0.54 65.53 1.38 1.35 1.09 to 1.66
Fifth 75.66 70.17 92.3 11.67 0.7 53.75 1.13 1.06 0.81 to 1.4
*PR=prevalence ratios estimated using Poisson regression.
†PRadj=prevalence ratios adjusted for sex, age, race/ethnicity (White, mixed ethnic group, Black or Black British, Asian or Asian British and
other race), region, survey year and household income.
‡Signicant linear trend across all quintiles (p≤0.05).
Figure 1 Percentage of excessive free sugar intake that would be avoided under two counterfactual scenarios regarding the
consumption of the main dietary sources of free sugar. UK population aged 1.5 years or over (2008–2014).
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from

10 RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
rich in free sugars could lead to larger public health
benefits. Policies concerning the use of fiscal measures
to reduce intake of free sugars and improve diet quality
should consider extending beyond artificially sweetened
beverages to include the main driver of excessive free
sugar intake, including dairy drinks, cakes, biscuits and
confectionery.13
To our knowledge, this is the first study to examine
the association between consumption of ultra-processed
foods, as defined per NOVA,6 and dietary content of free
sugar in different age groups of the UK population. The
use of NOVA is a key strength of the study as it classified
foods by their level of processing level using standardised
and objective criteria. NOVA has been recognised as a valid
tool for public health and nutrition research and policy
by the Food and Agricultural Organisation of the United
Nations26 and the Pan American Health Organisation.27
In addition, we used data from the NDNS—a large and
nationally representative sample of the UK population,
applying weighting to reduce any sampling and non-re-
sponse bias. Unlike household budget data, food diaries
employed in the NDNS take food wastage into account,
include food eaten out of home and do not assume that
all individuals within a household consume the same diet.
Importantly, the dietary data also allowed for the disag-
gregation of dishes into their constituents and classifi-
cation of the underlying ingredients, which enabled the
calculation of more precise estimates of intakes of each
NOVA group and reduced misclassification.
Potential limitations should be considered. The dietary
data we used were self-reported and may be subject
to misclassification. A constant limitation of dietary
assessment methods is under-reporting of some foods
(particularly unhealthy foods), though food diaries are
recognised to be one of the most comprehensive methods
for assessing dietary intake. Possible under-reporting of
unhealthy foods may lead to an underestimation of the
dietary contribution of ultra-processed foods and the
overall intake of free sugars but may less likely affect the
association between these variables. Nevertheless, accu-
rate and valid NDNS data were achieved through optimal
methods for collecting dietary intake,28 which helped to
minimise missing information. NDNS collects limited
information indicative of food processing (eg, place of
meals and product brands), which may lead to misclassi-
fication of food items. This bias is more likely for a small
number of specific food items such as pizza where there
is insufficient information for classification purposes (see
online supplementary table S2). In those cases, the most
frequently consumed alternative (culinary preparation or
manufactured product) was chosen. Finally, our theoret-
ical minimum risk exposure models estimate the poten-
tial impact of eliminating each of the main sources of free
sugars on excessive free sugar intake, ignoring substitu-
tions that may occur in the consumption of other foods.
Although our findings suggest that greater reduction in
excessive free sugar intake could be achieved by elimi-
nating ultra-processed food consumption, guidance to
the public about reducing the consumption of table sugar
remains an important component of any public health
guidance.
Conclusion
Almost half of excessive intake of free sugars in the UK
can be attributed to ultra-processed foods. Policies to
reduce sugar consumption should focus on minimising
consumption of ultra-processed foods and replacing them
with unprocessed or minimally processed food alterna-
tives. The study adds to a growing body of evidence that
ultra-processed foods are a major contributor to growth
of diet-related non-communicable diseases globally.
Author afliations
1Departamento de Nutrição, Universidade de Sao Paulo, Sao Paulo, Brazil
2Núcleo de Pesquisas Epidemiológicas em Nutrição e Saúde, Universidade de Sao
Paulo, Sao Paulo, Brazil
3Departamento de Políticas Públicas e Saúde Coletiva, Universidade Federal de Sao
Paulo, Sao Paulo, Brazil
4Escola Paulista de Medicina, Departamento de Medicina Preventiva, Universidade
Federal de São Paulo, Sao Paulo, Brazil
5Public Health Policy Evaluation Unit, School of Public Health, Imperial College
London, London, UK
6Departamento de Medicina Preventiva, Faculdade de Medicina, Universidade de
Sao Paulo, Sao Paulo, Brazil
Contributors CAM, EMS, FR, MLdCL and RBL designed the research. FR and RBL
undertook data management and analysis. CAM, CM, EMS, FR, LFMdR, MLdCL and
RBL interpreted the data. FR wrote the rst draft of the manuscript. All authors read,
edited and approved the nal manuscript.
Funding This work was supported by the Fundação de Amparo à Pesquisa do
Estado de São Paulo (FAPESP), grant numbers 2015/14900-9, 2016/14302-7
(FR is a beneciary of a postdoctoral fellowship) and 2014/25614-4 (LFMR is a
beneciary of a doctoral fellowship).
Disclaimer FAPESP had no role in the design, analysis or writing of this
manuscript.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval All relevant research ethics and governance committees approved
the survey.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Computerized raw data les and documentation
from this survey were obtained under license from the U.K. Data Archive (http://
www. esds. ac. uk). Details of how food item classication was accomplished are
further explained in previously published papers (Rauber et al, Nutrients. 2018 - see
Supplementary Table S1 http://www. mdpi. com/ 2072- 6643/ 10/ 5/ 587/ s1).
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, appropriate credit is given, any changes made indicated, and the use
is non-commercial. See:http:// creativecommons. org/ licenses/ by- nc/ 4. 0/.
ORCID iD
FernandaRauber http:// orcid. org/ 0000- 0001- 9693- 7954
REFERENCES
1 Scientic Advisory Committee on Nutrition. SACN’s sugars and
health recommendations: why 5%. London, UK: Scientic Advisory
Committee on Nutrition, Department of Health, 2015.
2 Te Morenga LA, Howatson AJ, Jones RM, etal. Dietary sugars
and cardiometabolic risk: systematic review and meta-analyses of
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from

11
RauberF, etal. BMJ Open 2019;9:e027546. doi:10.1136/bmjopen-2018-027546
Open access
randomized controlled trials of the effects on blood pressure and
lipids. Am J Clin Nutr 2014;100:65–79.
3 Scientic Advisory Committee on Nutrition. Carbohydrates and
health report. London, UK: Scientic Advisory Committee on
Nutrition, Department of Health, 2015.
4 Te Morenga L, Mallard S, Mann J. Dietary sugars and body weight:
systematic review and meta-analyses of randomised controlled trials
and cohort studies. BMJ 2012;346:e7492.
5 World Health Organization. Sugars intake for adults and children.
Geneva, Switzerland: World Health Organization, 2015.
6 Monteiro CA, Cannon G, Moubarac J-C, etal. The un decade of
nutrition, the nova food classication and the trouble with ultra-
processing. Public Health Nutr 2018;21:5–17.
7 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.
8 Moubarac J-C, Batal M, Louzada ML, etal. Consumption of
ultra-processed foods predicts diet quality in Canada. Appetite
2017;108:512–20.
9 Rauber F, da Costa Louzada ML, Steele EM, etal. Ultra-Processed
food consumption and chronic non-communicable Diseases-
Related dietary nutrient prole in the UK (2008⁻2014). Nutrients
2018;10:E587.
10 Martínez Steele E, Baraldi LG, Louzada MLdaC, 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.
11 Louzada MLdaC, Ricardo CZ, Steele EM, etal. The share of ultra-
processed foods determines the overall nutritional quality of diets in
Brazil. Public Health Nutr 2018;21:94–102.
12 Cediel G, Reyes M, da Costa Louzada ML, etal. Ultra-processed
foods and added sugars in the Chilean diet (2010). Public Health Nutr
2018;21:125–33.
13 Public Health England. National diet and nutrition survey results from
years 7 and 8 (combined) of the rolling programme (2014/2015 to
2015/2016). London, UK: Public Health England, 2018.
14 Department of Health and Social Care: Global Public Health
Directorate:Obesity, Food and Nutrition. Childhood obesity: a plan
for action, chapter 2. London, UK: Department of Health and Social
Care, 2018.
15 Public Health England. National diet and nutrition survey results from
years 1, 2, 3 and 4 (combined) of the rolling programme (2008/2009–
2011/2012). London, UK: Public Health England, 2014.
16 Fitt E, Cole D, Ziauddeen N, etal. DINO (Diet In Nutrients Out)
- an integrated dietary assessment system. Public Health Nutr
2015;18:234–41.
17 Public Health England. McCance and Widdowson’s the composition
of foods integrated dataset 2015. London, UK: Public Health
England, 2015.
18 Monteiro CA, Cannon G, Levy RB, etal. The StAR shines bright.
World Nutrition 2016;7:28–38.
19 Fitt E, Mak TN, Stephen AM, etal. Disaggregating composite food
codes in the UK national diet and nutrition survey food composition
databank. Eur J Clin Nutr 2010;64(Suppl 3):S32–6.
20 White IR, Royston P, Wood AM. Multiple imputation using chained
equations: issues and guidance for practice. Stat Med
2011;30:377–99.
21 Nielsen SJ, Adair L. An alternative to dietary data exclusions. J Am
Diet Assoc 2007;107:792–9.
22 Steenland K, Armstrong B. An overview of methods for calculating
the burden of disease due to specic risk factors. Epidemiology
2006;17:512–9.
23 Rezende LFMde, Eluf-Neto J. Population attributable fraction:
planning of diseases prevention actions in Brazil. Rev Saude Publica
2016;50:S0034-89102016000100601.
24 Moodie R, Stuckler D, Monteiro C, etal. Prots and pandemics:
prevention of harmful effects of tobacco, alcohol, and ultra-
processed food and drink industries. Lancet
2013;381:670–9.
25 Public Health England. First measure of industry progress to cut
sugar unveiled [press release]. London, UK: Public Health England,
2018.
26 Food and Agriculture Organization of the United Nations. Guidelines
on the collection of information on food processing through food
consumption surveys. Rome, Italy: Food and Agriculture Organization
of the United Nations, 2015.
27 Pan American Health Organization. Ultra-processed food and
drink products in Latin America: trends, impact on obesity, policy
implications. Washington, DC: Pan American Health Organization,
2015.
28 Public Health England. Dietary data collection and editing. in
national diet and nutrition survey. results from years 1–4 (combined)
of the rolling programme (2008/2009–2011/2012. London, UK:
Public Health England, 2014Available online:. https://www. gov. uk/
government/ statistics/ national- diet- and-nutrition-survey-results-
from-years-1-to-4-combined-of-the-rolling-programme-for-2008-
and-2009-to-2011-and-2012. (accessed on 15 January 2018).
Sao Paulo. Protected by copyright. on October 29, 2019 at UNIFESP - Universidade Federal dehttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2018-027546 on 28 October 2019. Downloaded from