Healthy and sustainable

diets in the 21st century

Healthy and sustainable
fr om a n i n d i v i d ua l a n d s o c i e ta l p er s p ec t i v e

diets in the 21st century

F R O M A N IND IVID UAL AN D SOC I E TAL PE R S PEC TI VE
Marjolein C. Harbers

Marjolein C. Harbers
 Healthy and sustainable diets in the 21st
century from an individual and societal
perspective

Gezonde en duurzame voedingspatronen in de 21ste eeuw vanuit

individueel en maatschappelijk perspectief
(met een samenvatting in het Nederlands)

Proefschrift

COLOFON
Healthy and sustainable diets in the 21st century from an individual and societal ter verkrijging van de graad van doctor aan de
perspective Universiteit Utrecht
op gezag van de
rector magnificus, prof.dr. H.R.B.M. Kummeling,
Printed by Ipskamp Printing | proefschriften.net ingevolge het besluit van het college voor promoties
Layout and design: Sara Terwisscha van Scheltinga, persoonlijkproefschrift.nl in het openbaar te verdedigen op
Cover design: Sara Terwisscha van Scheltinga, persoonlijkproefschrift.nl
dinsdag 31 mei 2022 des middags te 2.15 uur

ISBN: 978-94-6421-688-2

door
The research described in this thesis was part of the Supreme Nudge project which was
supported by a grant of the Dutch Heart Foundation and the Netherlands Organization
for Health Research and Development (CVON2016-04).
Marjoleine Catharine Harbers
Financial support by the Dutch Heart Foundation and Julius Center for the publication
geboren op 2 februari 1994
of this thesis is gratefully acknowledged. te Wageningen

Copyright 2022 © Marjolein C. Harbers

The Netherlands. All rights reserved. No parts of this thesis may be reproduced, stored
in a retrieval system or transmitted in any form or by any means without permission
of the author.
Promotoren:
Prof. dr. Y.T. van der Schouw
TABLE OF CONTENTS
Prof. dr. J.W.J. Beulens

Copromotor: Chapter 1 General Introduction 6

Dr. F. Rutters

Part I Healthy and sustainable diets from an individual perspective

Chapter 2 Adherence to the Dutch dietary guidelines and 15‑year incidence of heart 16
failure the EPIC‑NL cohort
Financial support by the Dutch Heart Foundation for the publication of this thesis is gratefully
acknowledged. Chapter 3 Adherence to the DASH diet and risk of heart failure in the EPIC-NL 42
cohort

Chapter 4 Adherence to the EAT-Lancet Healthy Reference Diet in relation to 58

coronary heart disease, all-cause mortality risk and environmental
impact: results from the EPIC-NL cohort

Chapter 5 The association between adherence to the EAT-Lancet Healthy Reference 86

Diet and coronary heart disease and stroke: the EPIC-CVD cohort study

Part II Healthy and sustainable diets from a societal perspective

Chapter 6 The effects of nudges on purchases, food choice, and energy intake 116
or content of purchases in real-life food purchasing environments: a
systematic review and evidence synthesis

Chapter 7 Residential exposure to fast-food restaurants and its association with 168
diet quality, overweight and obesity in the Netherlands: a cross-sectional
analysis in the EPIC-NL cohort

Chapter 8 Determinants of food choice and perceptions of supermarket-based 194

nudging interventions among adults with low socioeconomic position: the
SUPREME NUDGE project

Chapter 9 General Discussion 220

Appendices Summary in English 239

Summary in Dutch 242
List of publications 245
Dankwoord 247
About the author 251
1
1.1 HEALTHY AND SUSTAINABLE DIETS FROM AN
INDIVIDUAL PERSPECTIVE
Cardiovascular disease (CVD) is one of the leading causes of disease burden and mortality
globally (1). From 1990 to 2019 the number of CVD cases almost doubled from 271 million
to 523 million, and the number of deaths due to CVD increased from 12.1 million to
18.6 million (2). Consequently, CVD is a major public health problem associated with a
significant burden of disease on an individual level as well as a high economic burden on
a societal level (3). Despite national and global efforts to curb the rise in CVD, set targets
– such as a 25% decrease in mortality from non-communicable diseases, including CVD,
by the year 2025 as put forward by the World Health Organization – are not likely to be
met (4, 5). Therefore, reducing the burden of CVD remains a significant public health
challenge (6).

The occurrence of CVD shows a socioeconomic gradient, with more disadvantaged

individuals experiencing a higher burden of CVD compared to more affluent individuals
(7, 8). The pathways trough which socioeconomic position (SEP) may contribute to
socioeconomic disparities in CVD constitute a complex interplay of factors on the
psychosocial level (e.g., stress, health literacy), behavioural level (e.g., unhealthy diets)
and environmental level (e.g., access to healthy food). Diet in particular is one of the
major behavioural modifiable risk factors for CVD, accounting for 49% of total CVD
deaths and 48-67% of the socioeconomic inequalities in CVD occurrence (9, 10).

Meta-analyses of randomized controlled trials (RCTs) show that adherence to healthy

dietary patterns such as the Dietary Approaches to Stop Hypertension (DASH) diet and
Mediteranean diet positively affect key CVD risk factors such as blood pressure and
weight status compared to control diets (11-13). The PREDIMED study – a large-scale RCT
investigating the effects of a dietician-supervised Mediterranean diet on CVD incidence
in a high risk population with a follow-up of nearly 5 years – showed that the intervention
groups had significantly lower risk of developing CVD as compared to the control group
(14). However, despite the potential for prevention of CVD through dietary modification,

General Introduction adherence to dietary recommendations remains low (15), especially among populations
with low SEP (16, 17).

Most of the evidence on the role of diet in CVD focusses on coronary heart disease and
stroke, while much less research has been devoted to heart failure (18). Heart failure is
a chronic condition pertaining to the heart’s inability to pump around adequate supplies
of blood. It usually presents at older age and is characterized by a poor prognosis (19,
20) limited treatment options (21), and a severe impact on quality of life (22). Therefore,
chapter 1 general introduc tion
primary prevention of heart failure seems to be an attractive strategy. Especially given
the ageing population and the expected increase in HF incidence (23), it is of importance
to further elucidate the role of diet in onset of heart failure.
At the same time, there is an increasing recognition that population diets should not
only be healthy but also sustainable (24). At this moment, food production takes a high
toll on our environment, by taking up more than 43% of habitable land, causing two
thirds of freshwater withdrawals, 26% of global greenhouse gas emissions, 32% of global
terrestrial acidification, and 78% of eutrophication (25, 26). Therefore, The EAT-Lancet
Commission on Healthy Diets From Sustainable Food Systems proposed the Healthy
Reference Diet, which incorporates both health and sustainability in universal dietary
recommendations (24). However, empirical research examining the effects of the Healthy
Reference Diet on cardiovascular health, mortality, and sustainability indicators using
population-based data is scarce.
1.2 HEALTHY AND SUSTAINABLE DIETS FROM A
SOCIETAL PERSPECTIVE
Concluding from the previous, there is an urgent need to equitably shift towards
healthy and sustainable diets on a population level. However, behaviour change on an
individual level is difficult to sustain. A large network meta-analysis comparing multiple
dietary patterns to a control diet found that reduced weight loss and blood pressure
at six month follow-up largely disappeared at 12 month follow-up (13). This raises an
important question on how the potential for prevention through dietary modification
can be unlocked on a population level.
For several decades, public health efforts to improve diet quality have been mainly
focused on achieving behaviour change through information-based intervention
strategies (e.g., mass information campaigns, dietary counselling). However, such
information-based intervention strategies preferentially improve diets among those
of higher SEP, and thereby contribute to widening of socioeconomic inequalities (27).
This is mainly due to the fact that such interventions heavily rely on individual agency,
for which the resources (e.g., time, income, social networks) are more scarce among
disadvantages groups (28, 29). Alternatively, interventions that target the overarching
conditions that shape our daily lives, i.e. our environment, create healthier conditions for
all and may therefore be an equitable strategy for improving population diets (30, 31).
The food environment is such an overarching condition which can be leveraged as an
entry-point for improving population diets. Nudging is an example of an intervention
8 9
strategy that uses subtle non-monetary changes in the choice architecture that make
the healthy choice the more easy choice, for example by placing healthier products
at eye level in supermarkets (32, 33). At a more macro level, there are indications that
the availability of (un)healthy food outlets around our home, school, or workplace
also affect the food choices we make (34-37). However, evidence on the role of the
food environment – both on micro-level (e.g., in the supermarket) and on macro-level 1
(e.g., the presence of food outlets in our physical environment) remains inconclusive.
Furthermore, it is not yet clear how adults with low SEP perceive interventions such as
nudging strategies.
1.3 CONTENT OF THIS THESIS
Viewing from an individual perspective, the first part of this thesis aims to explore the
association between various healthy and/or sustainable dietary patterns in relation
to cardiovascular outcomes, mortality, and sustainability indicators. From an societal
perspective, the second part of this thesis aims to explore how the food environment
affects dietary behaviours and health outcomes, and how adults with low SEP perceive
interventions in the food environment. This research was part of the larger Supreme
Nudge project which aims to develop, implement, and evaluate the effects of an
intervention package consisting out of pricing and nudging strategies – combined with
a physical activity app – in a real-life supermarket setting on dietary behaviour and
cardiometabolic health among people with lower SEP.
In Chapter 2 and 3, we explore the associations between healthy dietary patterns (e.g.,
Dutch Healthy Diet 2015 index and the DASH diet, respectively) and risk of heart failure
using data from the European Prospective Investigation into Cancer and Nutrition
(EPIC-NL) study. In Chapter 4 and 5, we explore the association between the Healthy
Reference Diet – a healthy and sustainable dietary pattern – in relation to cardiovascular,
mortality, and sustainability outcomes using data from the EPIC-NL and EPIC-CVD
cohort, respectively. Chapter 6 provides a systematic literature review examining
the effectiveness of nudges in promoting healthy dietary choices within real-life food
purchasing environments. Additionally, we explore whether there was evidence for a
moderating role of SEP. In Chapter 7, we explore the association between exposure
to fast-food restaurants in the neighborhood environment in relation to diet quality,
overweight an obesity, using data from the EPIC-NL study. In Chapter 8 we present
findings of a qualitative study in which we explored how the determinants of food choice
among adults with low SEP shape the perceptions regarding supermarket-based nudging
strategies. In Chapter 9, I will summarize the main findings of this thesis and put them
in a broader perspective.
chapter 1 general introduc tion
REFERENCES
1. Global burden of 369 diseases and injuries in 204 countries and territories, 1990-
2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet.
2020;396(10258):1204-22.
2. Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. Global
Burden of Cardiovascular Diseases and Risk Factors, 1990-2019: Update From the GBD
2019 Study. J Am Coll Cardiol. 2020;76(25):2982-3021.
3. Leal J, Luengo-Fernández R, Gray A, Petersen S, Rayner M. Economic burden of
cardiovascular diseases in the enlarged European Union. Eur Heart J. 2006;27(13):1610-9.
4. World Health Organization. Global action plan for the prevention and control of
noncommunicable diseases 2013-2020. Geneva, Switzerland: World Health Organization; 2013.
5. Sacco RL, Roth GA, Reddy KS, Arnett DK, Bonita R, Gaziano TA, et al. The Heart of 25 by 25:
Achieving the Goal of Reducing Global and Regional Premature Deaths From Cardiovascular
Diseases and Stroke. Circulation. 2016;133(23):e674-e90.
6. Dugani S, Gaziano TA. 25 by 25: Achieving Global Reduction in Cardiovascular Mortality.
Curr Cardiol Rep. 2016;18(1):10-.
7. Schultz WM, Kelli HM, Lisko JC, Varghese T, Shen J, Sandesara P, et al. Socioeconomic
Status and Cardiovascular Outcomes: Challenges and Interventions. Circulation.
2018;137(20):2166-78.
8. Abdalla SM, Yu S, Galea S. Trends in Cardiovascular Disease Prevalence by Income Level
in the United States. JAMA Network Open. 2020;3(9):e2018150-e.
9. Meier T, Gräfe K, Senn F, Sur P, Stangl GI, Dawczynski C, et al. Cardiovascular mortality
attributable to dietary risk factors in 51 countries in the WHO European Region from 1990
to 2016: a systematic analysis of the Global Burden of Disease Study. Eur J Epidemiol.
2019;34(1):37-55.
10. Méjean C, Droomers M, van der Schouw YT, Sluijs I, Czernichow S, Grobbee DE, et al. The
contribution of diet and lifestyle to socioeconomic inequalities in cardiovascular morbidity
and mortality. Int J Cardiol. 2013;168(6):5190-5.
11. Filippou CD, Thomopoulos CG, Kouremeti MM, Sotiropoulou LI, Nihoyannopoulos PI,
Tousoulis DM, et al. Mediterranean diet and blood pressure reduction in adults with and
without hypertension: A systematic review and meta-analysis of randomized controlled
trials. Clin Nutr. 2021;40(5):3191-200.
12. Filippou CD, Tsioufis CP, Thomopoulos CG, Mihas CC, Dimitriadis KS, Sotiropoulou LI, et
al. Dietary Approaches to Stop Hypertension (DASH) Diet and Blood Pressure Reduction
in Adults with and without Hypertension: A Systematic Review and Meta-Analysis of
Randomized Controlled Trials. Adv Nutr. 2020;11(5):1150-60.
13. Ge L, Sadeghirad B, Ball GDC, da Costa BR, Hitchcock CL, Svendrovski A, et al. Comparison
of dietary macronutrient patterns of 14 popular named dietary programmes for weight
and cardiovascular risk factor reduction in adults: systematic review and network meta-
analysis of randomised trials. Bmj. 2020;369:m696.
14. Estruch R, Ros E, Salas-Salvadó J, Covas MI, Corella D, Arós F, et al. Primary Prevention of
Cardiovascular Disease with a Mediterranean Diet Supplemented with Extra-Virgin Olive
Oil or Nuts. N Engl J Med. 2018;378(25):e34.
10
15. Imamura F, Micha R, Khatibzadeh S, Fahimi S, Shi P, Powles J, et al. Dietary quality among
men and women in 187 countries in 1990 and 2010: A systematic assessment. The Lancet
Global Health. 2015;3(3):e132-e42.
16. Dijkstra SC, Neter JE, Brouwer IA, Huisman M, Visser M. Adherence to dietary guidelines
for fruit, vegetables and fish among older Dutch adults; the role of education, income and
job prestige. J Nutr Health Aging. 2014;18(2):115-21.
17. Giskes K, Avendano M, Brug J, Kunst AE. A systematic review of studies on socioeconomic
inequalities in dietary intakes associated with weight gain and overweight/obesity 1
conducted among European adults. Obes Rev. 2010;11(6):413-29.
18. Rich MW, Hauptman PJ. Nutrition in Heart Failure: More Than Drugs and Devices. J Card
Fail. 2015;21(12):943-4.
19. Magnussen C, Niiranen TJ, Ojeda FM, Gianfagna F, Blankenberg S, Vartiainen E,
et al. Sex-Specific Epidemiology of Heart Failure Risk and Mortality in Europe:
Results From the BiomarCaRE Consortium. JACC Heart Fail. 2019;7(3):204-13.
20. Jones NR, Roalfe AK, Adoki I, Hobbs FDR, Taylor CJ. Survival of patients with chronic
heart failure in the community: a systematic review and meta-analysis. Eur J Heart Fail.
2019;21(11):1306-25.
21. The Lancet. Heart failure: the need for improved treatment and care. The Lancet.
2018;392(10146):451.
22. Moradi M, Daneshi F, Behzadmehr R, Rafiemanesh H, Bouya S, Raeisi M. Quality of life
of chronic heart failure patients: a systematic review and meta-analysis. Heart Failure
Reviews. 2020;25(6):993-1006.
23. Groenewegen A, Rutten FH, Mosterd A, Hoes AW. Epidemiology of heart failure. Eur J Heart
Fail. 2020;22(8):1342-56.
24. Willett W, Rockström J, Loken B, Springmann M, Lang T, Vermeulen S, et al. Food in the
Anthropocene: the EAT-Lancet Commission on healthy diets from sustainable food
systems. Lancet. 2019;393(10170):447-92.
25. Poore J, Nemecek T. Reducing food’s environmental impacts through producers and
consumers. Science. 2018;360(6392):987-92.
26. Crist E, Mora C, Engelman R. The interaction of human population, food production, and
biodiversity protection. Science. 2017;356(6335):260-4.
27. McGill R, Anwar E, Orton L, Bromley H, Lloyd-Williams F, O’Flaherty M, et al. Are
interventions to promote healthy eating equally effective for all? Systematic review of
socioeconomic inequalities in impact. BMC Public Health. 2015;15(1):457.
28. Mechanic D. Disadvantage, inequality, and social policy. Health Aff (Millwood).
2002;21(2):48-59.
29. White M, Adams J, Heywood P. How and why do interventions that increase health overall
widen inequalities within populations. In: Babones SJ, editor. Social Inequality and Public
Health. Bristol: Policy Press; 2009. p. 65-80.
30. Capewell S, Graham H. Will cardiovascular disease prevention widen health inequalities?
PLoS Med. 2010;7(8):e1000320-e.
31. van Trier TJ, Mohammadnia N, Snaterse M, Peters RJG, Jørstad HT, Bax WA, et al. An appeal
to our government for nationwide policies in the prevention of cardiovascular disease.
Neth Heart J. 2021:1-5.
32. Thaler RH, Sunstein CR. Nudge: Improving decisions about health, wealth, and happiness.
London: Penguin Books; 2009.
11
chapter 1 general introduc tion

33. Hollands GJ, Bignardi G, Johnston M, Kelly MP, Ogilvie D, Petticrew M, et al. The TIPPME
intervention typology for changing environments to change behaviour. Nature Human
Behaviour. 2017;1(8):0140.
34. Gonçalves VSS, Figueiredo A, Silva SA, Silva SU, Ronca DB, Dutra ES, et al. The food
environment in schools and their immediate vicinities associated with excess weight in
adolescence: A systematic review and meta-analysis. Health Place. 2021;71:102664.
35. Bivoltsis A, Trapp G, Knuiman M, Hooper P, Ambrosini GL. The influence of the local food
environment on diet following residential relocation: longitudinal results from RESIDential 1
Environments (RESIDE). Public Health Nutrition. 2020;23(12):2132-44.
36. Black C, Moon G, Baird J. Dietary inequalities: what is the evidence for the effect of the
neighbourhood food environment? Health Place. 2014;27:229-42.
37. Caspi CE, Sorensen G, Subramanian SV, Kawachi I. The local food environment and diet: a
systematic review. Health Place. 2012;18(5):1172-87.

12 13
part I Healthy and sustainable diets from
an individual perspective
2
ABSTRACT

Purpose
A healthy diet may contribute to the primary prevention of heart failure (HF), but
evidence is still inconclusive. We aimed to study the association between adherence to
the Dutch dietary guidelines and incidence of HF.

Methods
We studied 37,468 participants aged 20-70 years and free of HF at baseline from the
EPIC-NL cohort. At baseline (1993-1997) data were collected on demographics, lifestyle
and presence of chronic diseases. Dietary intake was assessed using a 178-item validated
food frequency questionnaire. Dietary intake data were used to calculate scores on
the Dutch Healthy Diet 2015 Index (DHD15-index) measuring adherence to the Dutch
dietary guidelines. The DHD15-index is based on the average daily intake of 14 food
groups resulting in a total score ranging between 0 and 140, with higher scores indicating
better adherence. HF morbidity and mortality during follow-up were ascertained
through linkage with national registries. Cox proportional hazards analysis was used to
estimate hazard ratios (HRs) and 95% confidence intervals (95% CIs) for the association
between DHD15 adherence and HF risk, adjusting for sociodemographic and lifestyle
characteristics.

Results
The average score on the DHD15-index was 71 (SD=15). During a median follow-up of
15.2 years (IQR: 14.1 – 16.5), 674 HF events occurred. After adjustment for demographic
and lifestyle characteristics, higher scores on the DHD15-index were associated with
lower risk of HF (HRQ4vsQ1: 0.73; 95% CI: 0.58 – 0.93; P-trend: 0.001).

Conclusion
Adherence to the Dutch dietary guidelines In a large Dutch population of middle-aged adults, higher adherence to the Dutch dietary
guidelines was associated with lower risk of HF.
and 15‑year incidence of heart failure in
the EPIC‑NL cohort

This chapter was published as:

Harbers MC, de Kroon AM, Boer JMA, Asselbergs FW, Geleijnse JM, Verschuren WMM, et al.
Adherence to the Dutch dietary guidelines and 15-year incidence of heart failure in the EPIC-
NL cohort. Eur J Nutr. 2020;59(8):3405-13
chapter 2 adherence to the dutch dietary guidelines and risk of heart failure

INTRODUCTION
Heart failure (HF) is a chronic condition of the heart in which the heart is not sufficiently
able to contract or relax, resulting in an inadequate supply of oxygen and nutrients to
the rest of the body (1). As a consequence, patients with HF experience symptoms such
as shortness of breath and fatigue. HF is a growing public health concern worldwide,
and typically presents at older age. Given the ageing population, the lifetime risk of HF is
high, ranging between 20% and 45% (2). Consequently, it is expected that the prevalence
of HF will increase substantially in the future (3).
Coronary heart disease, hypertension, diabetes mellitus, obesity and smoking are
responsible for 52% of incident cases of HF in the US (4). Given the fact that an unhealthy
diet is an important risk factor for coronary heart disease, hypertension, diabetes
mellitus and obesity, it could be hypothesized that diet may be an important modifiable
risk factor for the primary prevention of HF as well. Potential mechanisms by which
healthy diets may be protective for HF development include decreased oxidative stress
and inflammation, and increased antioxidant defense and nitric oxide bioavailability (5).
Previously, adherence to the DASH diet has been associated with 22-37% lower risk of
HF (6,7) and improved left ventricular function (8), although another study remained
inconclusive (9). Similarly, high adherence to the Mediterranean diet has been associated
with 21-31% lower risk of HF (10,11), although after adjustment for lifestyle characteristics
this was not the case in a German cohort (12). The heterogeneity in study findings may
be due to differences in length of follow-up, ascertainment of HF, and age of study
participants. As the evidence-base for the protective association of healthy dietary
patterns with HF risk remains inconclusive, more prospective observational studies
are warranted.
METHODS
Study population
EPIC-NL is the Dutch contribution to The European Prospective Investigation into Cancer
and Nutrition- and consists of 40 011 participants from the MORGEN and PROSPECT
cohorts (14). For the MORGEN cohort (n=22,654), men and women from the general
population aged 20 to 59 years were randomly sampled from three Dutch towns in the
Netherlands (Amsterdam, Doetinchem, and Maastricht) between 1993 and 1997. For
the Prospect cohort (n=17,357) women aged 49 to 70 years, participating in the breast 2
cancer screening programme between 1993 and 1995, and living in Utrecht or its vicinity
were recruited. Participants provided informed consent prior to study inclusion. Both
studies complied with the Declaration of Helsinki and were approved by the medical
ethics committee of the Netherlands Organization for Applied Scientific Research (TNO)
(MORGEN), and the institutional review board of the University Medical Centre Utrecht
(Prospect).
For the present study, participants were excluded who withheld permission for linkage
with disease registries (n=1,756); had prevalent HF (n=22); did not participate in the food
frequency questionnaire (n=179); had implausible energy intake (i.e., those in the lowest
and highest 0.5% of the ratio of energy intake over basal metabolic rate; n=356); had
missing data on the covariates in model 1 and model 2 (n=229); withdrew permission for
use of the data for analyses (n=1). Consequently, the population for analysis comprised
37,468 participants (Figure 1).
• EPIC-NL cohort: Prospect (n=17,357) and
40,011 Morgen (n=22,654)
• No permission for linkage with disease
38,255 registries (n=1,756; 4.4%)

• Prevalent heart failure at baseline (n=22;

38,233 0.1%)
The Health Council of the Netherlands released an updated version of the Dutch dietary
• Missing food frequency questionnaire data
guidelines in 2015, in which they compiled all the evidence on nutrients, foods, and 38,054 (n=179; 0.4%)

dietary patterns in relation to the ten most important chronic diseases and related • Individuals with implausible energy intakea
37,698 (n=356; 0.9%)
risk factors – including cardiovascular disease and risk factors such as systolic blood
• Individuals with missing data on confounders
pressure, LDL-cholesterol, and body weight. Therefore, higher adherence to the Dutch 37,469 (n=229; 0.6%)

dietary guidelines potentially reduces the risk of HF. The DHD15-index was developed to • Individuals who withdrew informed consent
37,468 (n=1; 0.0%)
measure adherence to the Dutch dietary guidelines (13). Hence, the aim of the present
study is to study the association between adherence to the DHD15-index and the • Population for analysis (93.6% of 40,011)
37,468

incidence of HF in a Dutch population.

Figure 1. Flowchart of participant exclusions. aImplausible energy intake was defined as in-
dividuals within the lower and upper 0.5% ratio of energy intake over basal metabolic rate

18 19
chapter 2 adherence to the dutch dietary guidelines and risk of heart failure
Dietary intake assessment
Dietary intake was assessed using a self-administered validated food frequency
questionnaire (FFQ) that included questions on the habitual frequency of consumption
of 178 food items during the year preceding enrolment. Consumption frequency could
be indicated in times per day, per week, per month or per year, or as never. Consumed
amounts could be indicated using specified units (e.g. glass or slice). For selected foods,
photographs with portion sizes assisted in portion size estimation (15). Reproducibility
and validity were previously assessed in 121 men and women (15,16). The median validity
of 16 food groups with 12 monthly 24-hour recalls as reference, was 0.61 for men and
0.53 for women. Correlation coefficients for specific food groups ranged from 0.21 for
cooked vegetables to 0.78 for sugar and sweet products among men and from 0.31 for
vegetables and 0.87 for alcoholic beverages among women. Average food intake (g/d)
was calculated by multiplying the consumption frequency with the consumed amounts,
and nutrient intakes were calculated using the Dutch food composition table of 1996 (17).
Calculation of Dutch Healthy Diet 2015-index score
Average food intake was used to construct scores on the DHD15-index (13). The DHD15-
Index represents adherence to the Dutch food-based dietary guidelines released by
the Health Council of the Netherlands in 2015. The DHD15-index has previously been
validated by comparing mean scores as assessed by a food frequency questionnaire
and two 24h recalls as a reference, which showed a correlation coefficient of 0.56 (13).
The calculation of the DHD15-Index in the EPIC-NL study population has been described
previously (18). In short, the index consists of 15 food groups which are assigned a
proportional score between 0 and 10. In the present study, we included only fourteen
components since no data was available on type of coffee (filtered vs. unfiltered)
consumed. Consequently, scores on the DHD15-index could range between 0 (no
adherence) to 140 (complete adherence) for each participant. The scoring system of
the DHD15-Index is shown in Supplementary Table 1.
Ascertainment of heart failure
Hospitalization for and death from HF were used to define HF incidence. Hospitalization
for HF was determined based on both primary and secondary hospital discharge
diagnoses which were obtained from the Hospital Discharge Register (ICD9-code: 428).
A primary hospital diagnosis was defined as the underlying disease for hospitalization.
A secondary hospital diagnosis was defined as a comorbidity of the primary hospital
admission. The Hospital Discharge Register was linked to the EPIC-NL cohort on the
basis of birth date, sex, postal code, and general practitioner by a validated probabilistic
method (19). Vitality information was obtained through the municipal registry and causes
of death were obtained from the Cause of Death Register at Statistics Netherlands
20
(ICD10-code: I50). Death from HF was based on both primary and secondary causes of
death. A primary cause of death was defined as the underlying disease that led to death.
A secondary cause of death was defined as either a complication of the primary cause,
or another disease which might have contributed to the death. Follow-up was complete
until December 31st 2010.
Assessment of covariates
A self-administered general questionnaire provided information on age, sex, educational
level, smoking status, and physical activity. Educational level was categorized into low 2
(lower vocational training or primary school), middle (secondary school and intermediate
vocational training), or high (higher vocational training or university), and smoking status
was categorized into current, former or never (14). Physical activity was assessed using
the validated EPIC questionnaire and categorized according to the Cambridge Physical
Activity Index (CPAI) into active, moderately active, moderately inactive and inactive
(20,21) . In the first year of the MORGEN study (1993, 14.2% of the study population),
physical activity was not assessed with the EPIC questionnaire. The missing physical
activity data were imputed using single imputation SPSS Missing Values Analysis
procedure as described previously (22).
A physical examination at baseline provided information on blood pressure, weight,
height, and waist and hip circumference. During this examination systolic and diastolic
blood pressure measurements were performed twice in supine position, from which
the mean was taken. For the MORGEN-EPIC these measurements were performed on
the left arm using a random zero sphygmomanometer, and for the PROSPECT-EPIC on
the right arm using a Boso Oscillomat (Bosch & Son, Jungingen, Germany). Participants
were categorized as being hypertensive in case systolic blood pressure ≥140 mm Hg
or diastolic blood pressure ≥ 90 mm Hg, when antihypertensive medications were
used (self-reported) or if hypertension had been diagnosed by a physician. BMI was
calculated as height divided by weight squared and waist-hip ratio was calculated as
waist circumference divided by hip circumference.
Blood samples were drawn during the physical examination. Serum total cholesterol
was measured using enzymatic methods and LDL- and HDL-cholesterol were measured
using a homogeneous assay with enzymatic endpoint. The ratio between total/HDL
cholesterol ratio was computed (14). Presence of hyperlipidaemia was based on self-
reported diagnosis and / or use of medication.
Prevalent cases of type 2 diabetes were identified through linkage with the National
medical registry (1990–1997) and by self-report using the general baseline questionnaire.
21
chapter 2 adherence to the dutch dietary guidelines and risk of heart failure
Cases detected by either of these methods have been verified by consulting the general
practitioners (14). Only verified cases were used in the analyses.
Statistical analysis
Participant characteristics at baseline were shown by quartiles of DHD15-index scores.
The characteristics were expressed as means with standard deviations for normally
distributed variables, medians and interquartile range for skewed variables, or as counts
and percentages for categorical variables.
Follow-up time was calculated from the date of enrolment into the study to the date
of HF diagnosis, date of death or date of censoring. Cox proportional hazards models
were used to estimate hazard ratios (HRs) and 95% confidence intervals (95% CIs) for
the association between DHD15-index quartiles and incidence of HF, using the lowest
quartile as reference. Additionally, a test for linear trend across DHD15-index quartiles
was performed by assigning the median value for each quartile and modelling this
variable as a continuous variable. The proportional hazard assumption was fulfilled
according to inspection of log-log plots and by including time-dependent covariates in
the Cox models (p<0.05).
The first model was adjusted for age and sex. The second model, which is considered
to be the main model, was additionally adjusted for smoking status, physical activity,
total energy intake, and educational level. Additionally, we conducted an explorative
analysis where BMI, waist-hip ratio, diabetes mellitus type 2, systolic blood pressure,
hyperlipidaemia, hypertension, and total/HDL cholesterol ratio were added to model
2, both simultaneously and individually. We did this in a separate model since these
factors may be mediators in the association of DHD15-index and HF risk, rather than
confounders. Only participants with complete information on these additional covariates
were included in this explorative analysis (n=35,709), and a flowchart of participant
exclusions is shown in Supplementary Figure 1. All analyses were stratified by cohort
by including cohort in the strata statement in the Cox model, which allowed the hazard
of HF to vary across cohorts.
Several sensitivity analyses were carried out. First, to minimize the possibility of reverse
causation, the analyses were repeated after exclusion of the first two years of follow-
up. Second, the analyses were repeated using the DHD15-index excluding the sodium
component since the FFQ may not be a suitable method for estimating sodium intake
(23). Third, since participants with prevalent myocardial infarction (MI), type 2 diabetes,
hypertension and stroke at baseline may receive lifestyle advice to change dietary
behaviours, we conducted sensitivity analyses excluding these groups of participants.
22
Fourth, given that MIs frequently precede HF, the analyses were repeated after exclusion
of persons who developed MI during follow-up, to test whether this association is
independent of MI.
Statistical analyses were performed using IBM SPSS Statistics 24 (IBM Analytics, United
States of America, New York). A p-value below 0.05 was considered to be statistically
significant.
RESULTS 2
The average score on the DHD15-index was 71 (SD=15), with a minimum score of 16 and
a maximum score of 130. Participants with higher scores on the DHD15-index were more
likely to be older, highly educated, and more physically active. Moreover, they were less
likely to be male and to be a current smoker, and had lower BMI compared to participants
with low scores on the DHD index (Table 1).
During a median follow up of 15.2 (IQR: 14.1; 16.5) years, 674 HF events occurred.
Comparing extreme quartiles, the highest adherence to the DHD15-index was associated
with a lower incidence of HF (HR: 0.73; 95% CI: 0.58 – 0.93; P-trend = 0.001) after
adjustment for sociodemographic and lifestyle characteristics (Table 2).
The characteristics of the study population after exclusion of participants with missing
data on BMI, waist-hip ratio, systolic blood pressure, hyperlipidaemia and total/HDL
cholesterol ratio are shown in Supplementary Table 2. No substantial differences were
observed between the characteristics of the main population and subpopulation after
participant exclusions. The association between DHD15 adherence and HF incidence
after adjustment for sociodemographic and lifestyle characteristics was similar in the
subpopulation as compared to the main population (HR: 0.74; 95% CI: 0.57 – 0.95; P-trend:
0.003). When adjusting for potential intermediates simultaneously, the association
between DHD15-index and incidence of HF attenuated to non-significant (HR: 0.93;
95% CI: 0.72 – 1.21; P-trend: 0.27). When these potential intermediates were added
individually to the main model, adjustment for BMI led to the strongest attenuation
(HR: 0.84; 95%CI: 0.65-1.08; P-trend: 0.06) (Supplementary Table 3).
Exclusion of cases in the first two years of follow-up or excluding the sodium component
from the DHD15-index had no substantial impact on the results (Supplementary Table
4). Excluding participants with prevalent MI (n=504), type 2 diabetes (n=570) or stroke
(n=443) at baseline also had little effect on the observed hazard-ratio’s. In comparison,
exclusion of participants with hypertension (n=14,044) resulted in slight attenuation of
23
chapter 2 adherence to the dutch dietary guidelines and risk of heart failure
the association. Exclusion of participants who experienced MI at follow-up (n=53) did
not alter the results.
DISCUSSION
In the present study among Dutch middle-aged adults, we observed that high adherence
to the DHD15-index was associated with 27% lower risk of HF. These findings suggest
that adherence to the Dutch dietary guidelines may contribute to the prevention of HF.
Before we interpret our findings, some limitations of our study need to be addressed.
The first limitation concerns the reliability of the calculated sodium intake in our study.
Food frequency questionnaires are generally no suitable method for estimating sodium
intake (23). About 80% of sodium intake in the Netherlands comes from (processed)
foods (24). However, sodium content may differ among products brands, which is not
captured by the FFQ since no questions are included on specific food brands consumed.
Moreover, the amount of sodium added by consumers was not captured by the FFQ.
This may have resulted in misclassification and bias of the HRs towards the null. Indeed,
exclusion of sodium from the DHD15-index showed little change in the observed
association, suggesting that sodium was captured inadequately. Additionally, the coffee
component in the DHD15-index which represents the dietary recommendation to replace
unfiltered coffee by filtered coffee, was not included in the calculation of DHD15-scores
because no information was available on type of coffee consumed (filtered/unfiltered).
Furthermore, dietary intake was self-reported using a validated FFQ. The validity of
vegetable and fish consumption is of concern (15), and this may have contributed to
diluted associations. More specifically, higher intakes of these food groups have been
associated with lower risk of heart failure (25,26), which could imply that measurement
error in these components may have attenuated the inverse risk observed in the
present study. Consequently, the true risk associated with a DHD15 compliant dietary
pattern may have been underestimated. Moreover, diet was assessed only once and
may have changed during follow-up, resulting in non-differential misclassification that
may have weakened the observed associations. Also, the number of HF events in our
study population is likely an underestimation as HF may be unrecorded in patients who
have other CVD (1). Additionally, the validity of HF cases retrieved from the Hospital
Discharge Registry may be of concern, as a previous study in the Netherlands showed
low sensitivity (43%) but adequate positive predictive value (80%) when HF incidence
from the Hospital Discharge Registry was compared with HF incidence obtained from
a golden standard cardiovascular disease registry, potentially contributing to a further
underestimation of HF cases in the present study (27). Moreover, we only included
cases of HF resulting from hospitalization or death. Therefore, the generalizability of
24
the present findings to less severe cases of HF treated in primary care is uncertain. Also,
it is well-known that individuals participating in cohort studies are generally more health-
conscious, and therefore may not fully reflect the general population. In addition, we
did not have information on HF subtypes, such as based on preserved versus reversed
ejection fraction. Therefore, it is not yet clear whether the association observed in the
present study is different across subtypes of HF.
Strengths of our study include its prospective design, broad age range of the study
population and extensive follow-up period. Moreover, we had detailed information on 2
dietary intake. Finally, our associations were robust against exclusion of cases in the first
two years of follow-up, minimizing the possibility that associations can be explained by
reverse causation. Additionally, exclusion of patients participants with MI at baseline
or follow-up, prevalent type 2 diabetes, or stroke at baseline showed similar findings.
Exclusion of participants with prevalent hypertension attenuated the association,
indicating that hypertension may be an intermediate in the diet-heart failure pathway.
Data on the association between adherence to a healthy dietary pattern – and to the
DHD15-index in specific – with HF from large prospective observational studies are
scarce. However, the association between the DHD-index reflecting the dietary guidelines
of 2006 and incidence of CVD previously yielded a null association in the EPIC-NL cohort
(28). This may be due to the fact that HF was included in a composite end-point of
incident CVD that may differ in disease pathology and effects of dietary modification
from other CVD. Moreover, substantial differences exist between both indices in dietary
components included, with the DHD15-having its primary focus on food groups whereas
the DHD-index primarily includes individual nutrients and only a limited number of food
groups (vegetables, fruit and fish) (13,29).
The inverse association observed in the present study is further substantiated by
other prospective cohort studies investigating the association between healthy diet
index scores, such as the DASH score and the Mediterranean diet score, and HF risk.
First, adherence to the DASH diet was associated with 22 and 37% lower HF risk in two
Swedish cohorts with evidence of a dose-response relationship, supporting our finding
and suggesting that healthy eating patterns could protect against HF (6,7). In contrast,
a US cohort did not observe an association of the DASH score with HF risk (9). Second,
in the same two Swedish cohorts higher adherence to the modified Mediterranean diet
score was also associated with 21 and 31% lower risk of HF (10,11). In contrast to the
present study, these inverse associations remained after inclusion of BMI, history of
hypertension, hypercholesterolemia, and diabetes in the multivariable adjusted model.
In an analysis in the EPIC-Potsdam study (Germany), high adherence to the traditional
25
chapter 2 adherence to the dutch dietary guidelines and risk of heart failure

Mediterranean diet was associated with 41% lower risk of HF after adjustment for age, direction for the Mediterranean diet score and in a beneficial direction for the DHD15-
sex, and total energy intake (12). Adjustment for lifestyle characteristics resulted in index and DASH score.
slight attenuation of the association. Furthermore, inclusion of BMI, waist circumference,
diabetes, hypertension and hyperlipidaemia in this analysis resulted in further In conclusion, the present study showed that higher adherence to the Dutch Healthy
attenuation of the association to non-significance. Guidelines 2015 was associated with lower incidence of HF. Future research is warranted
to study the association between healthy dietary patterns and HF risk in study
One of the potential explanations for these previous contrasting findings may be due to populations including less severe cases of HF and different subtypes of HF.
differences in HF ascertainment. The US cohort installed a centralized events committee
which used all available outpatient and inpatient data to establish whether participants 2
had a confirmed HF diagnosis by a treating physician, experienced HF symptoms and
were on medical therapy. Similarly, in the EPIC-Potsdam study, HF was ascertained on the
basis of self-reported diagnosis, death due to HF, hospital information system data, and
validation of medication use typical for HF. Additionally, HF diagnosis was validated by
the treating physician. In the Swedish cohort and in EPIC-NL however, only cases resulting
from hospitalization or death were included. Consequently, the Swedish cohorts and
EPIC-NL may represent patients with more severe disease as compared to patients in
the US and German cohorts. As heart failure is a heterogeneous syndrome with multiple
subtypes, pathophysiology and the role of risk-factors – such as diet – may differ. For
example, a study conducted by Ahmad et al. (30) showed that different HF phenotypes
as established by cluster analysis, responded differently to therapeutic intervention
and had distinct outcomes on selected end-points including all-cause hospitalization
and mortality. Therefore, it may be important to consider the subtypes of HF when
studying the association between healthy dietary patterns and incident HF. Moreover,
differences in background diet and the age of participants included may also explain
these contrasting findings.

The effect size observed in the present study compares well to the inverse associations
reported for other dietary scores, previously. However, some notable differences exist
among the DHD15-index, the DASH score, and the Mediterranean diet score with regard
to their composition and scoring. Compared to the DASH score and Mediterranean diet
score, the DHD15-index is the most extensive diet score with more dietary components
included. For example, the DASH score does not include fish or alcohol, which have been
suggested as relevant food components for HF risk (31,32). Similarly, the Mediterranean
diet score does not include a tea of sugar-sweetened beverage component (33).
Additionally, the DHD15-index is based on absolute dietary cut-offs instead of scores
being assigned relative to other participants in the study population, which may be more
meaningful in terms of impact on disease risk. Finally, scoring of included components
differs among the diet scores, with high intakes of dairy contributing in a detrimental

26 27
 Table 1. Baseline characteristics across quartiles of DHD15-index scores in the EPIC-NL cohort a (n=37,468)

28
Quartiles of DHD15-index score
chapter
2

Q1 Q2 Q3 Q4
(16 – 61) (61 – 71) (71 – 81) (81 – 130)
N (%) 9,367 (25.0) 9,367 (25.0) 9,367 (25.0) 9,367 (25.0)
Age, years 49 (36, 55) 52 (42, 57) 53 (45, 59) 53 (46, 59)
Cohort
Morgen, n (%) 6,497 (69.4) 5,225 (55.8) 4,656 (49.7) 4,605 (49.2)
Prospect, n (%) 2,870 (30.6) 4,142 (44.2) 4,711 (50.3) 4,762 (50.8)
Sex
Female, n (%) 6,070 (64.8) 6,973 (74.4) 7,476 (79.8) 7,457 (79.6)
Educational level
Low, n (%) 6,179 (66.0) 5,866 (62.6) 5,498 (58.7) 4,218 (45.0)
Moderate, n (%) 2,143 (22.9) 2,004 (21.4) 1,918(20.5) 2,064 (22.0)
High, n (%) 1,045 (11.2) 1,497 (16.0) 1,951 (20.8) 3,085 (32.9)
Smoking status
Never, n (%) 3,014 (32.2) 3,534 (37.7) 3,863 (41.2) 3,892 (41.6)
Former, n (%) 2,301 (24.6) 2,837 (30.3) 3,131 (33.4) 3,528 (37.7)
Current, n (%) 4,052 (43.3) 2,996 (32.0) 2,373 (25.3) 1,947 (20.8)
Physical activity
Inactive, n (%) 956 (10.2) 715 (7.6) 618 (6.6) 523 (5.6)
Moderate Inactive, n (%) 2,346 (25.0) 2,405 (25.7) 2,408 (25.7) 2,165 (23.1)
Moderate active, n (%) 2,274 (24.3) 2,383 (25.4) 2,468 (26.3) 2,602 (27.8)
Active, n (%) 3,791 (40.5) 3,864 (41.3) 3,873 (41.3) 4,077 (43.5)

Table 1. Baseline characteristics across quartiles of DHD15-index scores in the EPIC-NL cohort a (n=37,468) (continued)

Quartiles of DHD15-index score

Q1 Q2 Q3 Q4
(16 – 61) (61 – 71) (71 – 81) (81 – 130)
DHD15 food groups
Vegetables, g/d 89 (65, 118) 98 (75, 128) 107 (82, 138) 124 (95, 161)
Fruit, g/d 115 (52, 185) 141 (96, 251) 190 (121, 274) 242 (147, 328)
Whole grains, g/d 14 (2, 72) 47 (6, 106) 73 (19, 126) 99 (64, 135)
Legumes, g/d 4 (1, 10) 6 (2, 12) 7 (3, 14) 10 (5, 17)
Nuts, g/d 2 (0, 5) 3 (1, 7) 3 (1, 7) 6 (2, 14)
Dairy, g/d 330 (156, 644) 391 (224, 605) 405 (254, 586) 399 (277, 536)
Fish, g/d 4 (2, 11) 6 (2, 12) 7 (3, 14) 9 (4, 15)
Tea, ml/d 71 (3, 250) 179 (36, 375) 250 (125, 450) 375 (250, 450)
Soft fats and oils, g/d 10 (4, 20) 11 (5, 19) 11 (5, 18) 11 (6, 18)
Solid fats, g/d 12 (5, 23) 9 (4, 18) 7 (3, 15) 6 (2, 13)
Red Meat, g/d 83 (61, 102) 68 (42, 87) 54 (33, 78) 39 (21, 59)
Processed meat, g/d 37 (21, 60) 26 (14, 39) 18 (9, 31) 12 (4, 20)
Sugar-sweetened beverages, ml/d 155 (73, 267) 119 (48, 196) 99 (40, 161) 76 (28, 138)
Alcohol, g/d 5 (1, 21) 4 (1, 15) 4 (1, 14) 6 (1, 14)
Sodium, mg/d 2668 (2098, 3333) 2337 (1864, 2868) 2157 (1763, 2613) 2039 (1700, 2428)
BMI, kg/m2 25.6 (23.2, 28.6) 25.4 (23.1, 28.1) 25.2 (23.1, 27.8) 24.5 (22.5, 26.9)
Hip-waist ratio 0.84 ± 0.09 0.83 ± 0.09 0.82 ± 0.09 0.81 ± 0.08
Type 2 diabetes, n (%) 167 (1.8) 162 (1.7) 145 (1.5) 96 (1.0)
29
adherence to the dutch dietary guidelines and risk of heart failure

2
chapter 2 adherence to the dutch dietary guidelines and risk of heart failure

Model 1 was adjusted for age and sex; b Model 2 was additionally adjusted for educational level, energy intake, physical activity, and smoking status.
Values are displayed as medians (P25, P75) or as means ± SD. Missing values, N (%): BMI 21 (0.1%), hip-waist ratio 68 (0.2%), systolic blood pressure
REFERENCES

<0.001
0.001
P trend

0.11
3.6 (2.9, 4.5)
3,306 (35.3)
(81 – 130) 1. Ponikowski P, Anker SD, AlHabib KF, Cowie MR, Force TL, Hu S, et al. Heart failure: preventing

804 (8.6)
126 ± 19
77 ± 10
Q4

disease and death worldwide. ESC Heart Fail. 2014;1(1):4-25.

2. Huffman MD, Berry JD, Ning H, Dyer AR, Garside DB, Cai X, et al. Lifetime risk for heart

0.66 (0.52 – 0.83)

0.73 (0.58 – 0.93)
0.89 (0.71 – 1.11)
failure among white and black Americans: cardiovascular lifetime risk pooling project. J

155 / 9367
Am Coll Cardiol. 2013;61(14):1510-7.

138,893
81-130
Q4

88
3. Heidenreich PA, Albert NM, Allen LA, Bluemke DA, Butler J, Fonarow GC, et al. Forecasting
the impact of heart failure in the United States: a policy statement from the American
Heart Association. Circ Heart Fail. 2013;6(3):606-19.
2
Table 1. Baseline characteristics across quartiles of DHD15-index scores in the EPIC-NL cohort a (n=37,468) (continued)

3.7 (3.0, 4.7)

3,631 (38.8)

795 (8.5)
(71 – 81)
Quartiles of DHD15-index score

127 ± 19

4. Benjamin EJ, Virani SS, Callaway CW, Chamberlain AM, Chang AR, Cheng S, et al. Heart
78 ± 11
Q3

Disease and Stroke Statistics-2018 Update: A Report From the American Heart Association.

Table 2. HRs (95% CI) for the association between quartiles of DHD15-index scores and incident HF (n = 37,468)
Circulation. 2018;137(12):e67-e492.

0.85 (0.68 – 1.07)

0.79 (0.63 – 0.99)
0.98 (0.79 – 1.23)
5. Kerley CP. Dietary patterns and components to prevent and treat heart failure: a

172 / 9367
comprehensive review of human studies. Nutr Res Rev. 2018:1-27.

139,020
71-81
82 (0.2%), diastolic blood pressure 64 (0.2%), hyperlipidaemia 77 (0.2), cholesterol ratio 1,556 (4.2%).

Q3

76
6. Levitan EB, Wolk A, Mittleman MA. Consistency with the DASH diet and incidence of heart
failure. Arch Intern Med. 2009;169(9):851-7.
3.9 (3.1, 4.9)
3,631 (38.8)

7. Levitan EB, Wolk A, Mittleman MA. Relation of consistency with the dietary approaches
794 (8.5)
(61 – 71)
127 ± 19
78 ± 11

to stop hypertension diet and incidence of heart failure in men aged 45 to 79 years. Am J
Q2

Cardiol. 2009;104(10):1416-20.
8. Nguyen HT, Bertoni AG, Nettleton JA, Bluemke DA, Levitan EB, Burke GL. DASH Eating

1.04 (0.84 – 1.29)

1.09 (0.88 – 1.35)
1.20 (0.97 – 1.49)
Pattern Is Associated with Favorable Left Ventricular Function in the Multi-Ethnic Study
199 / 9367
138,494

of Atherosclerosis. J Am Coll Nutr. 2012;31(6):401-7.

61-71
Q2

66

9. Del Gobbo LC, Kalantarian S, Imamura F, Lemaitre R, Siscovick DS, Psaty BM, et al.
Contribution of Major Lifestyle Risk Factors for Incident Heart Failure in Older Adults The
4.0 (3.2, 5.1)

Cardiovascular Health Study. Jacc-Heart Fail. 2015;3(7):520-8.

3,476 (37.1)

812 (8.7)
(16 – 61)
126 ± 19
78 ± 11

10. Tektonidis TG, Akesson A, Gigante B, Wolk A, Larsson SC. A Mediterranean diet and risk
Q1

of myocardial infarction, heart failure and stroke: A population-based cohort study.

Atherosclerosis. 2015;243(1):93-8.
11. Tektonidis TG, Akesson A, Gigante B, Wolk A, Larsson SC. Adherence to a Mediterranean
148 / 9367

1.00 (ref)
1.00 (ref)
1.00 (ref)
139,516

diet is associated with reduced risk of heart failure in men. Eur J Heart Fail. 2016;18(3):253-9.
16-61
Q1

53

12. Wirth J, di Giuseppe R, Boeing H, Weikert C. A Mediterranean-style diet, its components

and the risk of heart failure: a prospective population-based study in a non-Mediterranean
country. Eur J Clin Nutr. 2016;70(9):1015-21.
13. Looman M, Feskens EJM, de Rijk M, Meijboom S, Biesbroek S, Temme EHM, et al.
Development and evaluation of the Dutch Healthy Diet index 2015. Public Health Nutr.
Diastolic blood pressure, mmHg
Systolic blood pressure, mmHg

2017;20(13):2289-99.
Total/HDL cholesterol ratio

14. Beulens JWJ, Monninkhof EM, Verschuren WMM, van der Schouw YT, Smit J, Ocke MC, et
al. Cohort Profile: The EPIC-NL study. Int J Epidemiol. 2010;39(5):1170-8.
Hyperlipidaemia, n (%)

DHD15-index, range
Hypertension, n (%)

15. Ocke MC, Bueno-de-Mesquita HB, Goddijn HE, Jansen A, Pols MA, van Staveren WA, et al.
Cases / at risk (n)
Quartile median

The Dutch EPIC food frequency questionnaire. I. Description of the questionnaire, and
Person-years

relative validity and reproducibility for food groups. Int J Epidemiol. 1997;26 Suppl 1:S37-48.
Univariable

b
a

16. Ocke MC, Bueno-de-Mesquita HB, Pols MA, Smit HA, van Staveren WA, Kromhout D. The
Model 1
Model 2

Dutch EPIC food frequency questionnaire. II. Relative validity and reproducibility for
nutrients. Int J Epidemiol. 1997;26 Suppl 1:S49-58.
a

30 31
chapter 2 adherence to the dutch dietary guidelines and risk of heart failure

17. Dutch Food Composition Table (NEVO - Nederlandse Voedingsstoffenbetand - in Dutch).

Supplementary Table 1. Components of the DHD15-index and corresponding dietary recommendations and their threshold (minimum score) and

This component comprises two sub-components of which each sub-component has a maximum score of 5 points; b Maximum of 40 g cheese can be
included; c Maximum of 4 g lean fish can be included; d The coffee component was not included in the calculation of the DHD15-score as no data was
No consumption of butter, hard
Zeist: NEVO Foundation; 1996.

OR ratio of liquid cooking fats

Maximum score (10 points)

margarines and cooking fats

grains to refined grains ≥ 11

filtered coffee OR no coffee

products OR ratio of whole
No consumption of whole grain No consumption of refined
18. Biesbroek S, Verschuren WMM, van der Schouw YT, Sluijs I, Boer JMA, Temme EHM.

to solid cooking fats ≥ 13

Women: ≤10 g ethanol/d

Identification of data-driven Dutch dietary patterns that benefit the environment and

Men: ≤10 g ethanol/d

Consumption of only

available on type of coffee (filtered/unfiltered); e Sodium only originated from foods; intake from added salt was not captured by the FFQ.
are healthy. Climatic Change. 2018;147(3-4):571-83.
19. Herings RM, Bakker A, Stricker BH, Nap G. Pharmaco-morbidity linkage: a feasibility study

≤ 1.9 g sodium/d
comparing morbidity in two pharmacy based exposure cohorts. J Epidemiol Community

consumption
300–450 g/d
Health. 1992;46(2):136-40.

≥ 200 g/d
≥ 200 g/d

≥450 g/d
≥ 90 g/d

≤45 g/d
20. Wareham NJ, Jakes RW, Rennie KL, Schuit J, Mitchell J, Hennings S, et al. Validity and

≥10 g/d
≥15 g/d

≥15 g/d

0 g/d
0 g/d
repeatability of a simple index derived from the short physical activity questionnaire
used in the European Prospective Investigation into Cancer and Nutrition (EPIC) study.
Public Health Nutr. 2003;6(4):407-13.

margarines, liquid cooking fats

Any consumption of unfiltered

products OR ratio whole grain

of liquid cooking fats to solid

21. Pols MA, Peeters PHM, Ocke MC, Slimani N, BuenoDeMesquita HB, Collette HJA. Estimation

Minimum score (0 points)

and vegetable oils OR ratio

of reproducibility and relative validity of the questions included in the EPIC physical activity

Women: ≥20 g ethanol/d

No consumption of soft
questionnaire. Int J Epidemiol. 1997;26:S181-S9.

Men: ≥30 g ethanol/d

to refined grain ≤ 0.7
22. Joosten MM, Grobbee DE, van der A DL, Verschuren WMM, Hendriks HFJ, Beulens JWJ.

0 g/d OR ≥750 g/d

cooking fats ≤ 0.6

≥ 3.8 g sodium/d
Combined effect of alcohol consumption and lifestyle behaviors on risk of type 2 diabetes.
Am J Clin Nutr. 2010;91(6):1777-83.

≥ 250 g/d
≥ 100 g/d
23. McLean RM. Measuring population sodium intake: a review of methods. Nutrients.

≥ 50 g/d
coffee
2014;6(11):4651-62.

0 g/d
0 g/d
0 g/d

0 g/d
0 g/d

0 g/d
0 g/d
24. Van Rossum C, Buurma-Rethans E, Fransen H, Verkaik-Kloosterman J, Hendriksen M.
Salt consumption of children and adults in the Netherlands: Results from the Dutch

Eat a few portions of dairy produce daily, including

National Food Consumption Survey 2007-2010. National Institute for Public Health and

Eat one serving of fish weekly, preferably oily fish.

b. Replace refined cereal products by whole-grain

limited to one Dutch unit (10 gram ethanol) daily

Limit consumption of sweetened beverages and
fats by soft margarines, liquid cooking fats, and
a. Eat at least 90 g of wholegrain products daily
the Environment 2012.

Replace butter, hard margarines, and cooking

If alcohol is consumed at all, intake should be

Eat at least 15 grams of unsalted nuts a day.

Drink three cups of black or green tea a day

25. Mozaffarian D, Bryson CL, Lemaitre RN, Burke GL, Siscovick DS. Fish intake and risk of

Replace unfiltered coffee by filtered coffee.

Limit consumption of table salt to 6 g daily

incident heart failure. J Am Coll Cardiol. 2005;45(12):2015-21.

Limit consumption of processed meat.

26. Rautiainen S, Levitan EB, Mittleman MA, Wolk A. Fruit and vegetable intake and rate of heart

Eat at least 200 g of vegetables daily

CHAPTER 2 – SUPPLEMENTARY FILES
failure: a population-based prospective cohort of women. Eur J Heart Fail. 2015;17(1):20-6.

Limit consumption of red meat.

27. Merry AHH, Boer JMA, Schouten LJ, Feskens EJM, Verschuren WMM, Gorgels APM, et al.

Eat at least 200 g of fruit daily

Dietary recommendation
Validity of coronary heart diseases and heart failure based on hospital discharge and
mortality data in the Netherlands using the cardiovascular registry Maastricht cohort
study. Eur J Epidemiol. 2009;24(5):237-47.

Eat legumes weekly.

28. Struijk EA, May AM, Wezenbeek NLW, Fransen HP, Soedamah-Muthu SS, Geelen A, et al.

milk or yoghurt.
Adherence to dietary guidelines and cardiovascular disease risk in the EPIC-NL cohort. Int

vegetable oils
J Cardiol. 2014;176(2):354-9.

fruit juices.
products
29. van Lee L, Geelen A, van Huysduynen EJCH, de Vries JHM, van’t Veer P, Feskens EJM. The
Dutch Healthy Diet index (DHD-index): an instrument to measure adherence to the Dutch

cut-off (maximum score) values.

Guidelines for a Healthy Diet. Nutr J. 2012;11.
30. Ahmad T, Pencina MJ, Schulte PJ, O’Brien E, Whellan DJ, Pina IL, et al. Clinical implications

beverages and fruit juices

a
3. Whole grain products
of chronic heart failure phenotypes defined by cluster analysis. J Am Coll Cardiol.

13. Sugar-sweetened
DHD15-component
2014;64(17):1765-74.

12. Processed meat

31. Wang C, Xiong B, Huang J. The Role of Omega-3 Polyunsaturated Fatty Acids in Heart

9. Fats and oils

Failure: A Meta-Analysis of Randomised Controlled Trials. Nutrients. 2016;9(1).

1. Vegetables

11. Red meat

15. Sodiume
4. Legumes

14. Alcohol
10. Coffeed
32. Larsson SC, Wallin A, Wolk A. Alcohol consumption and risk of heart failure: Meta-analysis

b
6. Dairy
5. Nuts
2. Fruit

7. Fishc
of 13 prospective studies. Clin Nutr. 2018;37(4):1247-51.

8. Tea
33. Mahtani KR, Heneghan C, Onakpoya I, Tierney S, Aronson JK, Roberts N, et al. Reduced
Salt Intake for Heart Failure A Systematic Review. Jama Intern Med. 2018;178(12):1693-700.

a
32 33
34
chapter
2

missing data on potential intermediates

Supplementary Figure 1. Flowchart of additional participant exclusions for participants with

Supplementary Table 2. Baseline characteristics across quartiles of DHD15-index scores after exclusion of participants with missing data on BMI,
waist-hip ratio, systolic blood pressure, hyperlipidaemia, and total/HDL cholesterol ratio (n=35,709) a

Quartiles of DHD15-index score

Q1 Q2 Q1 Q4
(16 – 61) (61 – 71) (71 – 81) (81 – 130)
N (%) 8927 (25.0) 8927 (25.0) 8928 (25.0) 8927 (25.0)
Age in years 48 (36, 54) 51 (42, 57) 52 (44, 59) 53 (45, 59)
Cohort
Morgen, n (%) 6385 (71.5) 5189 (58.1) 4628 (51.8) 4543 (50.9)
Prospect, n (%) 2542 (28.5) 3738 (41.9) 4300 (48.2) 4384 (49.1)
Sex
Female, n (%) 5682 (63.6) 6553 (73.4) 7045 (78.9) 7037 (78.8)
Education
Low, n (%) 5836 (65.4) 5526 (61.9) 5142 (57.6) 3978 (44.6)
Middle, n (%) 2079 (23.3) 1951 (21.9) 1880 (21.1) 1967 (22.0)
High, n (%) 1012 (11.3) 1450 (16.2) 1906 (21.3) 2982 (33.4)
Smoking status
Never, n (%) 2871 (32.2) 3345 (37.5) 3668 (41.1) 3706 (41.5)
Former, n (%) 2170 (24.3) 2701 (30.3) 2972 (33.3) 3347 (37.5)
Current, n (%) 3886 (43.5) 2881 (32.3) 2288 (25.6) 1874 (21.0)
Physical activity
Inactive, n (%) 890 (10.0) 660 (7.4) 573 (6.4) 488 (5.5)
Moderate Inactive, n (%) 2233 (25.0) 2278 (25.5) 2282 (25.6) 2061 (23.1)
Moderate active, n (%) 2179 (24.4) 2286 (25.6) 2369 (26.5) 2476 (27.7)
Active, n (%) 3625 (40.6) 3703 (41.5) 3704 (41.5) 3902 (43.7)
35
adherence to the dutch dietary guidelines and risk of heart failure

2
36
Supplementary Table 2. Baseline characteristics across quartiles of DHD15-index scores after exclusion of participants with missing data on BMI,
waist-hip ratio, systolic blood pressure, hyperlipidaemia, and total/HDL cholesterol ratio (n=35,709) a (continued)
chapter
2

Quartiles of DHD15-index score

Q1 Q2 Q1 Q4
(16 – 61) (61 – 71) (71 – 81) (81 – 130)
DHD15 food groups
Vegetables, g/d 89 (65, 118) 98 (75, 128) 107 (82, 137) 124 (95, 161)
Fruit, g/d 113 (51, 182) 140 (95, 250) 190 (120, 273) 242 (145, 326)
Whole grains, g/d 14 (2, 72) 47 (6, 106) 73 (19, 127) 99 (64, 136)
Legumes, g/d 4 (1, 10) 6 (2, 12) 7 (3, 13) 10 (5, 17)
Nuts, g/d 2 (0, 5) 3 (1, 7) 3 (1, 7) 6 (2, 14)
Dairy, g/d 327 (156, 641) 390 (225, 606) 404 (253, 585) 399 (276, 536)
Fish, g/d 4 (2, 11) 6 (2, 12) 7 (3, 14) 9 (4, 15)
Tea, ml/d 71 (3, 250) 179 (36, 375) 250 (125, 450) 375 (250, 450)
Soft fats and oils, g/d 10 (4, 20) 11 (5, 19) 11 (5, 18) 11 (6, 18)
Solid fats, g/d 12 (5, 23) 9 (4, 18) 7 (3, 15) 6 (2, 13)
Red Meat, g/d 84 (61, 102) 69 (42, 87) 55 (33, 78) 39 (21, 60)
Processed meat, g/d 37 (22, 60) 26 (14, 40) 18 (9, 31) 12 (4, 20)
Sugar-sweetened beverages, ml/d 155 (73, 269) 120 (49, 198) 99 (40, 163) 76 (29, 138)
Alcohol, g/d 6 (1, 21) 5 (1, 16) 5 (1, 14) 6 (1, 14)
Sodium, mg/d 2687 (2113, 3345) 2349 (1885, 2884) 2165 (1768, 2626) 2045 (1707, 2437)
BMIa, kg/m2 25.6 (23.1, 28.5) 25.3 (23.1, 28.0) 25.1 (23.0, 27.7) 24.4 (22.5, 26.9)
Hip-waist ratio 0.84 ± 0.09 0.83 ± 0.09 0.82 ± 0.09 0.81 ± 0.08
Type 2 diabetes, n (%) 148 (1.7) 140 (1.6) 132 (1.5) 84 (0.9)

Supplementary Table 2. Baseline characteristics across quartiles of DHD15-index scores after exclusion of participants with missing data on BMI,
waist-hip ratio, systolic blood pressure, hyperlipidaemia, and total/HDL cholesterol ratio (n=35,709) a (continued)

Quartiles of DHD15-index score

Q1 Q2 Q1 Q4
(16 – 61) (61 – 71) (71 – 81) (81 – 130)
Systolic blood pressure, mmHg 125 ± 18 127 ± 19 127 ± 19 125 ± 19
Diastolic blood pressure, mmHg 78 ± 11 78 ± 11 78 ± 11 77 ± 10
Hypertension, n (%) 3244 (36.3) 3391 (38.0) 3385 (37.9) 3091 (34.6)
Total/HDL cholesterol ratio 4.0 (3.2, 5.0) 3.9 (3.1, 4.9) 3.7 (3.0, 4.7) 3.6 (2.9, 4.5)
Hyperlipidaemia, n (%) 767 (8.6) 745 (8.3) 744 (8.3) 763 (8.5)
a
Values are displayed as medians (P25, P75) or as means ± SD.
37
adherence to the dutch dietary guidelines and risk of heart failure

2
 Supplementary Table 3. HRs (95% CI) for the association between quartiles of DHD15-index scores and incident HF with both simultaneous and

38
individual adjustment for potential intermediates in a subpopulation of EPIC-NL (n=35,709) a
chapter

Q1 Q2 Q1 Q4 P trend
2

DHD15-index, range 16 – 61 61 – 71 71 – 81 81 – 130

Quartile median 53 66 76 88
Cases / at risk (n) 130 / 8927 165 / 8927 135 / 8928 134 / 8927
Person-years 133,338 132,419 132,843 132,578
b
Model 2 1.00 (ref) 1.05 (0.83 – 1.32) 0.78 (0.61 – 1.00) 0.74 (0.57 – 0.95) 0.003
Model 2b + all intermediatesc 1.00 (ref) 1.16 (0.92 – 1.46) 0.89 (0.69 – 1.14) 0.93 (0.72 – 1.21) 0.27
b
Model 2 + BMI 1.00 (ref) 1.10 (0.88 – 1.40) 0.85 (0.66 – 1.09) 0.84 (0.65 – 1.08) 0.06
Model 2b + Waist-hip ratio 1.00 (ref) 1.07 (0.85 – 1.36) 0.82 (0.64 – 1.05) 0.79 (0.61 – 1.02) 0.02
b
Model 2 + Type 2 diabetes 1.00 (ref) 1.09 (0.86 – 1.37) 0.81 (0.63 – 1.04) 0.79 (0.61 – 1.03) 0.02
b
Model 2 + Systolic blood pressure 1.00 (ref) 1.08 (0.85 – 1.36) 0.81 (0.63 – 1.04) 0.79 (0.61 – 1.02) 0.02
Model 2b + Hyperlipidaemia 1.00 (ref) 1.05 (0.83 – 1.32) 0.78 (0.61 – 1.00) 0.73 (0.57 – 0.95) 0.003
Model 2b + Hypertension 1.00 (ref) 1.08 (0.85 – 1.36) 0.82 (0.64 – 1.05) 0.79 (0.61 – 1.03) 0.02
Model 2b + Total/HDL cholesterol ratio 1.00 (ref) 1.05 (0.83 – 1.32) 0.80 (0.62 – 1.02) 0.76 (0.59 – 0.99) 0.009
a
Analyses were conducted after exclusion of participants with missing data on potential intermediates (Supplementary Figure 1); b Model 2 was adjusted
for age, sex, educational level, energy intake, physical activity and smoking status ; c Additionally adjusted for all potential intermediates simultaneously,
including BMI, hip-waist ratio, type 2 diabetes, systolic blood pressure, hypertension, hyperlipidaemia, total/HDL cholesterol ratio.

Supplementary Table 4. Sensitivity analyses excluding various subpopulations and using DHD15-index scores without the sodium component as
end-point

Q1 Q2 Q1 Q4 P trend
First two years of follow-up excluded
DHD15-index, range 16-61 61-71 71-81 81-130
Quartile median 53 66 76 88
Cases / at risk (n) 143/9321 188/9293 160/9310 147/9307
Person-years 139,457 138,410 138,956 138,820
Univariable 1.00 (ref) 1.17 (0.94, 1.46) 0.94 (0.75, 1.18) 0.87 (0.69, 1.09) 0.07
Model 1a 1.00 (ref) 1.02 (0.82, 1.27) 0.76 (0.60, 0.95) 0.64 (0.51, 0.81) <0.001
Model 2b 1.00 (ref) 1.07 (0.85, 1.33) 0.82 (0.65, 1.03) 0.72 (0.56, 0.92) 0.001
DHD15-index without sodium component
DHD15-index, range 14-54 54-64 64-74 74-120
Quartile median 47 59 68 80
Cases / at risk (n) 172/9,367 178/9,367 175/9,367 149/9,367
Person-years 139,252 138,815 138,915 138,940
Univariable 1.00 (ref) 0.95 (0.77-1.17) 0.91 (0.73-1.12) 0.78 (0.63-0.97) 0.11
Model 1a 1.00 (ref) 0.85 (0.69-1.05) 0.75 (0.60-0.92) 0.60 (0.48-0.75) <0.001
Model 2b 1.00 (ref) 0.92 (0.74-1.13) 0.84 (0.68-1.04) 0.71 (0.56-0.89) 0.001
Participants with prevalent MI at baseline excluded
DHD15-index, range 16-61 61-71 71-81 81-130
Quartile median 53 66 76 88
Cases / at risk (n) 130/9,231 184/9,233 160/9,228 145/9,257
Person-years 137,775 136,737 137,095 137,382
Univariable 1.00 (ref) 1.27 (1.01-1.59) 1.05 (0.83-1.32) 0.94 (0.74-1.20) 0.29
39
adherence to the dutch dietary guidelines and risk of heart failure

a
Model 1 1.00 (ref) 1.10 (0.88-1.38) 0.84 (0.67-1.06) 0.70 (0.55-0.89) <0.001
b
Model 2 1.00 (ref) 1.14 (0.91-1.43) 0.90 (0.71-1.14) 0.77 (0.60-0.99) 0.007
40
Supplementary Table 4. Sensitivity analyses excluding various subpopulations and using DHD15-index scores without the sodium component as
end-point (continued)
chapter
2

Q1 Q2 Q1 Q4 P trend
Participants with prevalent type 2 diabetes at baseline excluded
DHD15-index, range 16-61 61-71 71-81 81-130
Quartile median 53 66 76 88
Cases / at risk (n) 125/9200 180/9205 159/9222 144/9271
Person-years 137,320 136,348 136,980 137,592
Univariable 1.00 (ref) 1.29 (1.02 – 1.62) 1.08 (0.85 – 1.37) 0.97 (0.76 – 1.23) 0.40
a
Model 1 1.00 (ref) 1.11 (0.89 – 1.40) 0.87 (0.68 – 1.10) 0.72 (0.56 – 0.91) 0.001
Model 2b 1.00 (ref) 1.17 (0.93 – 1.47) 0.93 (0.73 – 1.19) 0.79 (0.62 – 1.02) 0.02
Participants with prevalent hypertension at baseline excluded
DHD15-index, range 18-61 61-71 71-81 81-130
Quartile median 53 66 76 88
Cases / at risk (n) 43/5891 58/5736 60/5736 58/6061
Person-years 88,566 86,036 85,862 90,482
Univariable 1.00 (ref) 1.20 (0.81 – 1.79) 1.18 (0.80 – 1.76) 1.08 (0.72 – 1.61) 0.82
Model 1a 1.00 (ref) 1.06 (0.71 – 1.58) 0.95 (0.64 – 1.41) 0.79 (0.53 – 1.19) 0.17
b
Model 2 1.00 (ref) 1.09 (0.73 – 1.63) 1.01 (0.67 – 1.52) 0.87 (0.57 – 1.33) 0.41

Supplementary Table 4. Sensitivity analyses excluding various subpopulations and using DHD15-index scores without the sodium component as
end-point (continued)

Q1 Q2 Q1 Q4 P trend
Participants with prevalent stroke at baseline excluded
DHD15-index, range 16-61 61-71 71-81 81-130
Quartile median 53 66 76 88
Cases / at risk (n) 141/9201 190/9219 160/9231 152/9246
Person-years 137,126 136,361 137,142 137,193
Univariable 1.00 (ref) 1.21 (0.97 – 1.50) 0.96 (0.76 – 1.21) 0.91 (0.72 – 1.15) 0.16
a
Model 1 1.00 (ref) 1.05 (0.84 – 1.31) 0.77 (0.61 – 0.97) 0.67 (0.53 – 0.85) <0.001
b
Model 2 1.00 (ref) 1.10 (0.88 – 1.37) 0.82 (0.65 – 1.04) 0.75 (0.59 – 0.95) 0.002
Participants with MI during follow-up excluded
DHD15-index, range 16 – 61 61 – 71 71 – 81 81 – 130
Quartile median 53 66 76 88
Cases / at risk (n) 139/9358 181/9349 158/9353 143/9355
Person-years 139,409 138,273 138,873 138,752
Univariable 1.00 (ref) 1.15 (0.92 – 1.44) 0.95 (0.75 – 1.19) 0.86 (0.68 – 1.09) 0.07
a
Model 1 1.00 (ref) 1.00 (0.80 – 1.25) 0.76 (0.61 – 0.96) 0.64 (0.50 – 0.81) <0.001
b
Model 2 1.00 (ref) 1.05 (0.84 – 1.31) 0.82 (0.65 – 1.04) 0.71 (0.55 – 0.91) 0.001
a
Model 1 was adjusted for age and sex; b Model 2 was additionally adjusted for educational level, energy intake, physical activity, and smoking status.
41
adherence to the dutch dietary guidelines and risk of heart failure

2
3
ABSTRACT

Background
The Dietary Approaches to Stop Hypertension (DASH) diet is an effective non-
pharmacologic approach for lowering blood pressure, but evidence of its effect on heart
failure (HF) remains scarce and inconsistent.

Objectives
We aimed to study the association between adherence to the DASH diet and incident HF.

Methods
We studied 37,671 participants from the EPIC-NL cohort, free from cardiovascular diseases
at baseline (1993-1997). Data on demographics, lifestyle, disease end-points and diet were
collected at baseline. Dietary intake was assessed using a food-frequency questionnaire.
DASH diet adherence was calculated based on intake rankings of eight food and nutrient
components, resulting in a score ranging 8-40 points. HF events were ascertained through
linkage to nation-wide registries. Cox proportional hazards models, adjusted for potential
confounders, were used to determine hazard ratios (HR) and 95% confidence intervals
(CI)s for the association between tertiles of adherence to DASH diet and HF incidence.
Further adjustments for body mass index (BMI), hypertension, cholesterol ratio and type
2 diabetes (T2D) were made to assess their potential mediating role.

Results
During a median follow-up of 15.3 years, 650 cases of HF occurred. Higher adherence
to DASH was associated with 22% lower HF risk in confounder-adjusted models (HRT3vsT1
0.78; 95%CI 0.64-0.95). Individual mediator adjustments did not materially change the
association, however, when these were included simultaneously an attenuation was
observed (HRT3vsT1 0.86; 95%CI 0.70-1.05).

Conclusions
Adherence to the DASH diet and risk of Higher adherence to the DASH diet was associated with lower risk of HF, which could
partly be attributed to BMI, hypertension, cholesterol ratio and T2D.
heart failure in the EPIC-NL cohort
Stefanie Costa, Marjolein C. Harbers, Jolanda M.A. Boer, W.M. Monique Verschuren,
Yvonne T. van der Schouw

Manuscript submitted to Nutrition, Metabolism, and Cardiovascular Diseases

chapter 3 adherence to the dash diet and risk of heart failure
INTRODUCTION
Heart failure (HF) is a condition in which the heart is unable to fill up and/or pump
blood around the body to fulfil its needs for oxygen and nutrients (1,2). HF prevalence
in the United States is expected to increase from 2.4% in 2012 to 3% in 2030, resulting in
more than 8 million American adults living with this condition (3). Despite the advances
that have been made with regard to therapy and management, HF continues to have a
substantial adverse impact on the overall quality of life of the patients living with this
condition (4,5) and along with that, HF related mortality rate remains high, with only 35%
of the individuals surviving 5 years after the initial diagnosis (6). Therefore, there is an
urgent need to explore effective heart failure prevention strategies (7,8).
High blood pressure is known to be one of the major risk factors for incident HF
(9,10,11,12) and its pharmacological treatment has been associated with a lower risk of
HF (13,14). Additionally, evidence has shown that other non-pharmacologic strategies
such as, lifestyle and dietary changes, can also have a favorable effect on blood pressure
reduction and control (15,16,17,18). The Dietary Approaches to Stop Hypertension
(DASH) diet, in specific, has been recommended for blood pressure lowering in adults
with hypertension (19). Previously, a meta-analysis of RCTs showed that the DASH
diet significantly reduced systolic blood pressure by 6.74 mm Hg (95% CI -4.23, -5.23;
I =78.1%) and diastolic blood pressure by 3.54 mmHg (95% CI -4.29, -2.79; I =56.7%) (20).
2 2
Consequently, the DASH diet may also represent a promising heart failure prevention
strategy.
Few studies examined the association between adherence to the DASH diet and heart
failure incidence, reporting both inverse (21,22) and null associations (23,24). In two
cohorts of Swedish men and women, a higher adherence to the DASH diet was associated
with lower risk of developing HF (21,22); however, in two other studies from the U.S. no
association was found (23,24). Therefore, the objective of the present study was to assess
the relationship between adherence to the DASH diet and the risk of HF.
METHODS
Study population
The EPIC-NL study resulted from the merging of two cohort studies ongoing in the
Netherlands, the Prospect cohort and the MORGEN cohort, which are both part of
the EPIC (European Prospective Investigation Into Cancer and Nutrition). The Prospect
cohort comprises of 17,357 women recruited in Utrecht and its surroundings through the
Dutch breast screening program with a range of ages 49 to 70 years old. The MORGEN
44
study recruited 22,654 men and women aged 20 to 64 in three different locations in the
Netherlands (Amsterdam, Doetinchem and Maastricht). The recruitment of individuals
for both cohorts was carried out simultaneously between 1993 and 1997 and resulted
in a combined total of 40,011 people (25,26,27). EPIC-NL was approved by the Medical
Ethical Committee of TNO Nutrition and Food Research (MORGEN-EPIC) and by the
institutional review boards of the University Medical Center Utrecht (Prospect-EPIC) and
was conducted in accordance with the guidelines of the Declaration of Helsinki. A written
informed consent was obtained from all the individuals before inclusion in the study.
Since enrolment, one participant (n=1) withdrew permission for the use of his/her data.
We additionally excluded participants who did not fill in the dietary questionnaire (n=218);
participants with implausible energy intakes (those in the 0.5% lowest and 0.5% highest 3
score of the ratio of reported energy intake/estimated energy requirement; n=400);
participants with missing information on prevalent cardiovascular diseases before study
inclusion (n=1185) and participants with prevalent cardiovascular diseases at baseline
(n=536). After these exclusions, 37,671 individuals remained for the analysis (Figure 1).
Dietary intake measurement and DASH-score construction
Dietary intake was assessed using a validated self-administered food-frequency
questionnaire (FFQ) completed at baseline. The FFQ contained questions regarding
the frequency of consumption of 178 food items during the year preceding enrolment.
Frequency of food consumption could be indicated in times per day, per week, per
month, per year, or as never. As for the amounts consumed, these could be indicated
using specified units (e.g., number of glasses or slices). Portion size was estimated with
the aid of photographs (26,28,29). The average daily food intake (g/d) was calculated by
multiplying the consumption frequency with the amounts consumed. Nutrient intakes
were calculated based on the Dutch food composition table of 1996 (30).
The level of adherence to the DASH diet was calculated according to the scoring proposed
by Fung et al. (2008) (31) based on intake (in g/day) of eight food and nutrient components:
(1) fruits; (2) vegetables; (3) nuts and legumes; (4) low-fat dairy products; (5) whole grains;
(6) red and processed meats; (7) sodium intake and (8) sweetened beverages. For fruit,
vegetables, nuts and legumes, low-fat dairy products, and whole grains, high intakes
are regarded as desirable. Therefore, a score of 5 is given to individuals in the highest
quintile, and a score of 1 to those who are in the lowest quintile. For red and processed
meat, sodium and sugar-sweetened beverages, low intakes are considered desirable.
For these food groups, the reverse scoring is used. Individuals in intermediate quintiles
obtain intermediate scores. Then the subscores for each food and nutrient component
are summed per participant to a score ranging from 8 (low adherence) to 40 (high
45
chapter 3 adherence to the dash diet and risk of heart failure
adherence). Considering that men and women have different energy requirements,
sex-specific quintiles were used for each dietary component (31).
Figure 1. Flowchart of participant exclusions. * Individuals in the 0.5% lowest and 0.5% highest
score of the ratio of reported energy intake/estimated energy requirement.
Ascertainment of heart failure
Incident HF as the endpoint was defined as first hospital admission or death from HF
as a primary or secondary diagnosis, after study inclusion. Data on hospitalization for
HF during follow-up was obtained through linkage to the Dutch Center for Health Care
Information (32). Information on vital status was obtained through the Municipal Registry
and on causes of death through the Cause of Death Register at Statistics Netherlands.
Hospital discharge for and death from HF were identified according to the International
Classification of Diseases, Ninth or Tenth Revision, Clinical modification as ICD-9: 428 and
ICD-10: I50, respectively. Follow-up was complete until 31 December 2010.
st
Assessment of covariates and health indicators
A general questionnaire, self-administered at baseline, provided information on socio-
demographic factors, lifestyle, and health indicators, including age, sex, educational
level, physical activity level and smoking habits. Education was categorized into low
(primary school up to completed advanced elementary education), middle (intermediate
vocational education up to completed higher general secondary education) and high
(high vocational education and university). Physical activity was assessed using a set
of questions common to all EPIC cohorts and categorized according to the validated
46
Cambridge Physical Activity Index into inactive, moderately inactive, moderately active
and active (26,33). Smoking habits were categorized as current smoker and non-smoker.
A physical examination was performed for each study participant, including measurement
of blood pressure, weight and height. Systolic and diastolic blood pressure were obtained
from two sequential measurements on the left arm, taken in supine position. In Prospect,
blood pressure measurements were performed 5-15 min after the start of the study visit
and repeated after 10 min using a Boso Oscillomat (Bosch & Son, Germany). In MORGEN,
blood pressure (systolic Korotkoff phase-I, diastolic Korotkoff phase-V) was measured at
the end of the study visit using a random-zero Sphygmomanometer (26). Comparability
of the measurement procedures for both cohorts has been reported in more detail
(34). Hypertension was considered present if at least one of the following criteria were 3
met: systolic blood pressure >140mm Hg, diastolic blood pressure >90mm Hg, use of
antihypertensive medication (self-reported) or physician-diagnosed hypertension (self-
reported). Body weight was measured to the nearest 0.5 kg, in light indoor clothing and
without shoes, using calibrated scales (Seca, Atlanta, GA, USA). Height was measured
to the nearest 0.5 cm. BMI was calculated as the weight divided by height squared
(kg/m2). During the physical examination non-fasting blood samples were drawn and
stored in liquid nitrogen until analysis. Total cholesterol was measured using enzymatic
methods and low-density lipoprotein and high-density lipoprotein (HDL) cholesterol
were measured using a standard homogeneous assay with enzymatic endpoint (26,27).
The cholesterol ratio was calculated as total cholesterol divided by HDL cholesterol level
(25). Alcohol consumption as assessed with the FFQ and categorized into ≤11g ethanol/d,
11-25g ethanol/d, 25-50g ethanol/d and >50g ethanol/d.
Prevalent type-2 diabetes (T2D) cases were assessed at baseline through linkage with
the National Medical Registry (1990-1997) and/or by self-report upon completion of the
baseline questionnaire. Only cases verified by general practitioner’s medical records
or pharmacy records were used for analyses (26,35). Information on occurrence of
myocardial infarction during follow-up was obtained through linkage to medical registries
(ICD-9: 410 for hospital discharges and ICD-10: I21-I22 for death due to myocardial
infarction) (26).
Data analysis
Baseline characteristics of the study population were summarized as mean ± standard
deviation (SD) or median [25th and 75th percentiles] for continuous variables and
percentages for categorical variables, presented across tertiles of DASH-score adherence.
47
chapter 3 adherence to the dash diet and risk of heart failure
Because there were missing data on outcome (HF – 1.3%) and other covariates (smoking
habits – 0.3%; educational level – 0.5%; BMI – 0.1%; cholesterol ratio – 3.7%; myocardial
infarction – 1.3%), Multiple Imputation by Chained Equations (MICE package in R) was
used (36,37). We simulated 10 complete datasets.
Length of follow-up was calculated as the period between the date of study entry and
the date of the first-ever outcome occurrence, loss to follow-up, death, or end of follow-
up (31-12-2010), whichever came first. Cox proportional hazard models (38,39) were
used to determine the hazard ratios (HR) and the 95% confidence intervals (CIs) for the
associations between tertiles of adherence to the DASH diet and the incidence of HF.
The first tertile was taken as the reference category. A test for linear trend across the
tertiles of the DASH-score was performed by assigning the median DASH-score in each
tertile and modelling this variable as a continuous variable. The association was also
modelled continuously per SD increase in the DASH-score Model 1 was adjusted for sex
and age. Model 2 was further adjusted for educational level, physical activity, energy
intake, smoking status, and alcohol intake. All analyses were stratified for cohort. The
proportional hazards assumption was tested by calculating the Schoenfeld residuals
and visually by checking log-minus-log plots, with no significant violations detected.
The potential mediating role of BMI, hypertension, total/HDL cholesterol ratio and
prevalent T2D in the association between level of adherence to the DASH diet and the
risk of HF was explored by including the variables into the multivariable model, both
simultaneously and individually. Sensitivity analyses were performed to investigate
whether pre-clinical HF had an influence on the results. The multivariable model analyses
were repeated after excluding the HF cases detected during the first 2 years of follow-up.
Furthermore, since myocardial infarction is a risk factor for HF, additional analyses were
conducted after excluding the HF cases that occurred following a myocardial infarction.
All statistical analyses were performed within each of the 10 imputed dataset, the results
were averaged using Rubin’s rule (40) and the 95% CIs were calculated accounting for
the uncertainty of the imputed estimates (36). All statistical analyses were performed
using the statistical software program R Version 3.5.1 - © 2009-2018 (37).
RESULTS
At baseline, the study population was predominantly female (74.7%), the median age
was 51.4 (interquartile range = 41.9; 57.6) years and the level of adherence to the DASH
diet was modest, with an average score of 24 (standard deviation = 5). Participants in
the highest tertile of the DASH score were more likely to be older, highly educated,
48
more physically active, have a lower BMI and less likely to be smokers compared to the
participants in the lowest tertile (Table 1).
During a median follow-up of 15.3 years, 650 cases of HF occurred. In fully adjusted
models, a higher level of adherence to the DASH diet was associated with a lower risk
of developing HF when comparing extreme tertiles (HR T3vsT1: 0.78; 95% CI 0.64-0.95).
Similarly, when analyzing the DASH-score as a continuous variable per standard
deviation increase an inverse association with HF was observed (HR SD: 0.87; 95% CI
0.79-0.94) (Table 2).
Adding the potential mediators BMI, hypertension, total/HDL cholesterol ratio and
T2D to model 2 individually did not materially alter the results. When these potential 3
mediators were added simultaneously, the association slightly attenuated and was no
longer statistically significant (HRT3vsT1: 0.86; 95% CI 0.70-1.05) (Table 3). Furthermore,
associations remained similar when excluding the 32 HF cases occurring during the first
two years of follow-up (Supplementary Table 1), as well as when excluding 79 HF cases
that occurred following a myocardial infarction (Supplementary Table 2).
DISCUSSION
In this large prospective cohort study among Dutch adults, high adherence to the DASH
diet was associated with a 22% lower risk of developing HF. Hypertension, BMI, total/
HDL cholesterol ratio and T2D may partially mediate the association.
The main strengths of the present study include its prospective study design, extensive
follow-up and the detailed information on dietary intake. Furthermore, the analyses
performed after exclusion of HF cases in the first two years of follow-up and HF cases
following a myocardial infarction minimized the possibility of the results being explained
by pre-clinical disease and further corroborated the robustness of our results.
Nevertheless, this study also has limitations that should be addressed before interpreting
the results. The FFQ was reasonably valid for ranking individuals according to food group
intake; however, lower validity was observed for the intake assessment of vegetables (28).
Vegetable intake is one of DASH components for which a higher intake is awarded with
more points, given the fact that a higher consumption of vegetables has been associated
with a lower risk of HF (31,41). Consequently, random measurement error in vegetable
intake could have resulted in a dilution of the inverse association found in our study.
Another limitation inherent to the FFQ concerns its reliability regarding the measurement
of sodium intake. In countries with western style diets, such as the Netherlands, it is
49
chapter 3 adherence to the dash diet and risk of heart failure

estimated that 75% to 80% of the dietary sodium is derived from processed foods.
However, products from different brands contain different levels of sodium and the
FFQ does not capture those differences (42). Furthermore, added salt used in food
preparation was not captured by the FFQ (28). Misclassification of sodium intake may
have occurred with regard to the added salt component, as cases may have used more
added salt when compared to the non-cases, resulting in an underestimation of the
observed effect. Moreover, dietary intake was measured only once at baseline, and the
same is true for the socio-demographic and lifestyle factors. Finally, HF assessment was
based on hospital discharge diagnoses and cause of death registries only, therefore we
probably ascertained only severe cases for whom treatment by the general practitioner
does not suffice, and we lack information on HF subtype. Given that hypertension has
been strongly associated with HFpEF (43), and DASH may work specifically through
its effects on blood pressure, it could be hypothesized that the association between
DASH and risk of HF is more pronounced in this subpopulation. Additionally, women
are more likely to develop HFpEF than HFrEF (44, 45). The predominantly female study
population and the fact that most HF cases were identified in women, are suggestive for
the assumption that the majority of our endpoints were most likely HFpEF.
HF was adjudicated by a centralized committee that used all outpatient and inpatient
medical records, possibly capturing more HF cases, including the less severe that were
taken care of by general practitioners. Moreover, HF has been consistently associated
with older age; thus, the differences in age of the participants between these studies
might also explain the contrast in the number of registered HF cases (3). All studies used
the same method proposed by Fung el al. (31) to generate their DASH-score. However,
DASH-scores are based on population-specific quintiles rather than on absolute food
intake cut-offs. It is likely that the background diets of the populations in each study
differ. Therefore, the hypothesis that a high DASH-score in our study population relates
to different intake amounts in study populations from other countries, cannot be
discarded.
3
In conclusion, greater adherence to the DASH diet was inversely associated with the risk
of developing HF in this population of Dutch adults. This association should be further
investigated in future studies preferably including larger populations accounting for both
severe and less-severe HF cases and with precise HF phenotyping, in order to facilitate
the analysis of possible differential associations across HF subtypes.

The inverse association observed in our study is in line with the findings of two Swedish
cohort studies on DASH diet adherence and incident HF in middle aged and older persons
(21,22). In the male cohort a 22% lower HF risk for individuals in the highest quartile of
adherence to the DASH diet was observed (21) versus a 37% lower risk in the females
(22). The multivariable models in both Swedish studies were adjusted for the potential
mediators in our study. In our study, including BMI, hypertension, total/HDL cholesterol
ratio and T2D simultaneously in the multivariable model led to an attenuation of the
risk reduction.

Nevertheless, the protective effect of a higher adherence to DASH diet on HF was not
observed in two other studies from the U.S. (23,24). First, in a cohort of older adults,
no association was found when investigating the role of the DASH diet (24). Second, in
a more age diverse cohort free of cardiovascular diseases, a trend towards a lower HF
incidence with higher DASH-score was observed, but associations were not statistically
significant (23).

A potential explanation for the inconsistency of study results might be due to differences
in HF ascertainment. In EPIC-NL as well as in both Swedish studies cases were ascertained
based on hospital discharge diagnoses and causes of death registries. Therefore, it is
likely that less severe cases not admitted into the hospital remained undocumented
and did not get accounted for (46,47,48). In the U.S. studies, on the other hand, incident

50 51
chapter 3 adherence to the dash diet and risk of heart failure

Table 1. Baseline characteristics of the EPIC-NL study participants by tertiles of the DASH-score Table 2. Hazards Ratios and 95% Confidence Intervals for the association between tertiles of
(N= 37,671) adherence to the DASH diet and HF incidence among the EPIC-NL participants (n=37,671).

Tertiles of the DASH Score Tertiles DASH-score

T1 (8-22) T2 (23-26) T3 (27-29) T1 (8-22) T2 (23-26) T3 (27-29) Continuous per P-trend
SD increase
n (individuals) 14,288 11,596 11,787
n (individuals) 14,288 11,596 11,787 37,671
DASH-score 19 ± 3 25 ± 1 30 ± 2
HF cases (n) 220 221 209 650
Sex (% female) 74.3 (10,615) 74.5 (8637) 75.5 (8905)
Model 1 Ref (1.00) 0.91 (0.75; 1.10) 0.69 (0.57; 0.84) 0.82 (0.76; 0.89) 0.0001
Age (years) 49.1 [37.0, 55.0] 51.7 [42.9, 57.9] 53.8 [48.0, 59.7]
Model 2 Ref (1.00) 0.97 (0.80; 1.18) 0.78 (0.64; 0.95) 0.87 (0.79; 0.94) 0.01
BMI (Kg/m2) 25.2 [22.9, 28.1] 25.3 [23.0, 27.9] 24.9 [22.8, 27.4]
Values are hazard ratio (95% confidence interval). DASH = Dietary Approaches to Stop
Current smokers (%) 38.7 (5,508) 29.4 (3,395) 21.7 (2,549) Hypertension; EPIC-NL = European Prospective Investigation Into Cancer and Nutrition-
Netherlands; HF = heart failure. Model 1: adjusted for age and sex; Model 2: adjusted for Model
High education (%) * 13.5 (1,923) 20.4 (2359) 28.4 (3,338) 1 and for educational level, physical activity level, smoking status and alcohol intake.
Physically Active (%) † 39.2 (5,580) 42.0 (4,860) 44.6 (5,262) 3
Hypertension (%) ‡ 36.3 (5,185) 37.6 (4,358) 37.5 (4,417)
Cholesterol Ratio § 3.9 [3.1, 4.9] 3.8 [3.1, 4.8] 3.7 [3.0, 4.7]
Dietary Intake Table 3. Hazards Ratios and 95% Confidence Intervals for the association between tertiles
of adherence to the DASH diet and HF incidence with adjustments for potential mediators
Energy (Kcal/d) 2143 ± 635 2032 ± 612 1954 ± 550 (n=37,671).
Alcohol (g/d) 4.5 [0.6, 16.1] 5.3 [0.8, 16.1] 5.3 [0.8, 14.8] Tertiles DASH-score
DASH-score components T1 (8-22) T2 (23-26) T3 (27-29) P-trend
Fruit (g/d) 118.9 [60.4, 181.1] 178.5 [117.1, 256.0] 250.1 [170.6, 353.7] n (individuals) 14,288 11,596 11,787
Vegetables (g/d) 110.4 [85.5, 138.9] 131.9 [104.1, 164.8] 157.2 [126.3, 195.8] HF cases (n) 220 221 209
Nuts (g/d) 1.9 [0.6, 7.1] 3.3 [0.8, 7.1] 3.7 [1.4, 10.7] Model 2 Ref (1.00) 0.97 (0.80; 1.18) 0.78 (0.64; 0.95) 0.01
Low Fat Dairy (g/d) 120.2 [53.8, 231.2] 220.3 [ 111.2, 357.8] 314.6 [187.7, 463.2] Model 2 + all mediators * Ref (1.00) 1.00 (0.83; 1.21) 0.86 (0.70; 1.05) 0.11
Whole Grains (g/d) 15.0 [1.9, 71.5] 67.3 [14.0, 118.3] 100.9 [64.9, 139.0] Model 2 + BMI Ref (1.00) 0.99 (0.82; 1.21) 0.82 (0.67; 1.00) 0.04
Meat (g/d) 107.8 [77.9, 138.8] 86.2 [53.9, 114.0] 58.8 [33.8, 89.8] Model 2 + Hypertension Ref (1.00) 0.98 (0.81; 1.19) 0.81 (0.66; 0.99) 0.03
Sodium (mg/d) 2593.1 ± 895.6 2363.6 ± 819.1 2187.7 ± 704.0 Model 2 + Cholesterol Ratio † Ref (1.00) 0.99 (0.81; 1.20) 0.79 (0.65; 0.97) 0.02
Sweetened drinks (g/d) 65.9 [23.1, 135.0] 27.7 [6.6, 69.2] 12.0 [0.6, 35.7] Model 2 + T2D Ref (1.00) 0.97 (0.80; 1.17) 0.79 (0.65; 0.97) 0.02
Values are percentage (n), mean ± SD or median [interquartile range]. BMI = body mass index; Values are hazard ratio (95% confidence interval). BMI = body mass index; DASH = Dietary
DASH = Dietary Approaches to Stop Hypertension; EPIC-NL = European Prospective Investigation Approaches to Stop Hypertension; HF = heart failure; T2D = type 2 diabetes. Model 2: adjusted
Into Cancer and Nutrition-Netherlands; * Higher vocational education and university; † for age, sex, educational level, physical activity level, smoking status and alcohol intake; * Model
Physically active according to the Cambridge Physical Activity Index; ‡ Present if at least one of 2 additionally adjusted for all potential mediator variables simultaneously: BMI, hypertension,
the following criteria were met: systolic blood pressure >140mm Hg, diastolic blood pressure total/high-density lipoprotein cholesterol ratio and T2D. † Cholesterol ratio calculated as total
>90mm Hg, antihypertensive medication (self-reported) or physician-diagnosed (self-reported); cholesterol (mmol/l) / high-density lipoprotein cholesterol (mmol/l).
§ Cholesterol Ratio calculated as Total cholesterol (mmol/l) / High-density lipoprotein cholesterol
(mmol/l).

52 53
chapter 3 adherence to the dash diet and risk of heart failure
REFERENCES
1. Orso F, Fabbri G, Maggioni A Pietro. Epidemiology of Heart Failure. In: Bauersachs J, Butler
J, Sandner P, eds. Heart Failure. Cham, Switzerland: Springer International Publishing;
2017: 15-33.
2. Heart Failure - Also known as Congestive heart failure. The National Heart, Lung, and
Blood Institute (NHLBI). Available at: https://www.nhlbi.nih.gov/health-topics/heart-failure.
Accessed November 18, 2019.
3. Heidenreich PA, Albert NM, Allen LA, et al. Forecasting the Impact of Heart Failure in the
United States. Circ Hear Fail 2013; 6(3):606-619.
4. Gallagher AM, Lucas R, Cowie MR. Assessing health-related quality of life in heart failure
patients attending an outpatient clinic: a pragmatic approach. ESC Hear Fail 2019; 6(1):3-9.
5. de Leon CFM, Grady KL, Eaton C, et al. Quality of Life in a Diverse Population of Patients
With Heart Failure. J Cardiopulm Rehabil Prev 2009;29(3):171-178.
6. Bleumink G, Knetsch A, Sturkenboom M, et al. Quantifying the heart failure epidemic:
prevalence, incidence rate, lifetime risk and prognosis of heart failure - The Rotterdam
Study. Eur Heart J 2004; 25(18):1614-1619.
7. Schocken DD, Benjamin EJ, Fonarow GC, et al. Prevention of Heart Failure. Circulation
2008; 117(19):2544-2565.
8. Dunlay SM, Weston SA, Jacobsen SJ, et al. Risk Factors for Heart Failure: A Population-Based
Case-Control Study. Am J Med 2009; 122(11):1023-1028.
9. Kalogeropoulos A, Georgiopoulou V, Kritchevsky SB, et al. Epidemiology of Incident Heart
Failure in a Contemporary Elderly Cohort. Arch Intern Med 2009; 169(7):708.
10. Shah S, Henry A, Roselli C, et al. Genome-wide association and Mendelian randomisation
analysis provide insights into the pathogenesis of heart failure. Nat Commun 2020;
11(1):163.
11. Slivnick J, Lampert BC. Hypertension and Heart Failure. Heart Fail Clin 2019; 15(4):531-541.
12. Yancy CW, Jessup M, Bozkurt B, et al. 2013 ACCF/AHA Guideline for the Management of
Heart Failure. Circulation 2013; 128(16).
13. Moser M, Hebert PR. Prevention of disease progression, left ventricular hypertrophy and
congestive heart failure in hypertension treatment trials. J Am Coll Cardiol 1996; 27(5):1214-1218.
14. Messerli FH, Rimoldi SF, Bangalore S. The Transition From Hypertension to Heart Failure.
Am Coll Cardiol HF 2017; 5(8):543-551.
15. Appel LJ, Moore TJ, Obarzanek E, et al. A Clinical Trial of the Effects of Dietary Patterns on
Blood Pressure. N Engl J Med 1997; 336(16):1117-1124.
16. Appel L, Champagne C, Harsha D, et al. Effects of Comprehensive Lifestyle Modification
on Blood Pressure Control. JAMA 2003; 289(16).
17. Moore LL, Singer MR, Bradlee ML, et al. Intake of Fruits, Vegetables, and Dairy Products in
Early Childhood and Subsequent Blood Pressure Change. Epidemiology 2005; 16(1):4-11.
18. Samadian F, Dalili N, Jamalian A. Lifestyle Modifications to Prevent and Control
Hypertension. Iran J Kidney Dis 2016 Sep; 10(5):237-263.
19. Arnett DK, Blumenthal RS, Albert MA, et al. 2019 ACC/AHA Guideline on the Primary
Prevention of Cardiovascular Disease. J Am Coll Cardiol 2019; 74(10):e177-e232.
54
20. Saneei P, Salehi-Abargouei A, Esmaillzadeh A, et al. Influence of Dietary Approaches to Stop
Hypertension (DASH) diet on blood pressure: A systematic review and meta-analysis on
randomized controlled trials. Nutr Metab Cardiovasc Dis 2014; 24(12):1253-1261.
21. Levitan EB, Wolk A, Mittleman MA. Relation of Consistency With the Dietary Approaches
to Stop Hypertension Diet and Incidence of Heart Failure in Men Aged 45 to 79 Years. Am
J Cardiol 2009; 104(10):1416-1420.
22. Levitan EB, Wolk A, Mittleman MA. Consistency with the DASH diet and incidence of heart
failure. Arch Intern Med 2009; 169(9):851-857.
23. Campos CL, Wood A, Burke GL, et al. Dietary Approaches to Stop Hypertension Diet
Concordance and Incident Heart Failure: The Multi-Ethnic Study of Atherosclerosis. Am J
Prev Med 2019; 56(6):819-826.
24. Del Gobbo LC, Kalantarian S, Imamura F, et al. Contribution of Major Lifestyle Risk Factors
for Incident Heart Failure in Older Adults. JACC Hear Fail 2015; 3(7):520-528.
25. Beulens JWJ, Monninkhof EM, Verschuren WMM, et al. Cohort Profile: The EPIC-NL study.
Int J Epidemiol 2010; 39(5):1170-1178. 3
26. Beulens JWJ, Monninkhof EM, Verschuren WMM, et al. Cohort Profile: The EPIC-NL study.
Int J Epidemiol 2010; 39(5):1170-1178.
27. Dutch Contribution to European Prospective Investigation into Cancer and Nutrition-
EpicNL. Available at: https://www.epicnl.eu/Home/EPICNL. Accessed November 23, 2019.
28. Ocké MC, Bueno-de-mesquita HBAS, Goddijn HE, et al. The Dutch EPIC Food Frequency
Questionnaire. I. Description of the Questionnaire, and Relative Validity and Reproducibility
for Food Groups. Int J Epidemiol 1997; 26:37-48.
29. Ocke M. The Dutch EPIC food frequency questionnaire. II. Relative validity and
reproducibility for nutrients. Int J Epidemiol 1997; 26(90001):49S - 58.
30. Dutch Food Composition Table (1996) (NEVO - Nederlandse Voedingsstoffenbestand - in
Dutch) NEVO Foundation, Zeist. Print.
31. Fung TT, Chiuve SE, McCullough ML, et al. Adherence to a DASH-style diet and risk of
coronary heart disease and stroke in women. Arch Intern Med 2008; 168(7):713-720.
32. Herings RM, Bakker A, Stricker BH, et al. Pharmaco-morbidity linkage: a feasibility study
comparing morbidity in two pharmacy based exposure cohorts. J Epidemiol Community
Heal 1992; 46(2):136-140.
33. Wareham NJ, Jakes RW, Rennie KL, et al. Validity and repeatability of a simple index
derived from the short physical activity questionnaire used in the European Prospective
Investigation into Cancer and Nutrition (EPIC) study. Public Health Nutr 2003; 6(4):407-413.
34. Schulze MB, Kroke A, Saracci R, et al. The effect of differences in measurement procedure
on the comparability of blood pressure estimates in multi-centre studies. Blood Press
Monit 2002; 7(2):95-104.
35. Sluijs I, van der A DL, Beulens JWJ, et al. Ascertainment and verification of diabetes in the
EPIC-NL study. Neth J Med 2010; 68(1):333-339.
36. Buuren S van, Groothuis-Oudshoorn K. mice : Multivariate Imputation by Chained
Equations in R. J Stat Softw 2011; 45(3).
37. R Core Team. R: A language and environment for statistical computing. R Foundation for
Statistical Computing - version 3.5.1. Published 2018. Available at https://www.r-project.org/.
38. Therneau T, Grambsch P. The Cox Model. In: Modeling Survival Data: Extending the Cox
Model. New York: Springer; 2000: 39-77.
55
chapter 3 adherence to the dash diet and risk of heart failure

39. Therneau T. A Package for Survival Analysis in S - version 2.38. Published 2015. Available
at https://cran.r-project.org/package=survival. CHAPTER 3 – SUPPLEMENTARY FILES
40. Rubin, DB. Multiple Imputation for Nonresponse in Surveys. New York: Wiley, 1987. Print.
41. Rautiainen S, Levitan EB, Mittleman MA, et al. Fruit and vegetable intake and rate of heart Supplementary Table 1. HR and 95% CIs for the association between adherence to the DASH
diet and HF incidence after removal of HF cases occurring during the first 2 years of follow-up
failure: a population-based prospective cohort of women. Eur J Heart Fail 2015; 17(1):20-26. (n = 32).
42. McLean R. Measuring Population Sodium Intake: A Review of Methods. Nutrients 2014;
Tertiles DASH Score
6(11):4651-4662.
T1 (8 - 22) T2 (23 -26) T3 (27 - 39) P-trend
43. Paulus WJ, Tschöpe C. A novel paradigm for heart failure with preserved ejection
fraction: comorbidities drive myocardial dysfunction and remodeling through coronary n (individuals) 14,277 11,582 11,780
microvascular endothelial inflammation. J Am Coll Cardiol. 2013;62(4):263-71. n (HF cases) 209 207 202
44. Dunlay SM, Roger VL, Redfield MM. Epidemiology of heart failure with preserved ejection Model 2 Reference 0.96 (0.79; 1.17) 0.79 (0.64; 0.97) 0.01
fraction. Nat Rev Cardiol. 2017;14(10):591-602.
DASH = Dietary Approaches to Stop Hypertension; HF = heart failure. Model 2: The multivariable
45. Ponikowski P, Voors AA, Anker SD, Bueno H, Cleland JGF, Coats AJS, et al. 2016 ESC model was adjusted for age, sex, educational level, physical activity level, smoking status, alcohol
Guidelines for the diagnosis and treatment of acute and chronic heart failure: The Task
Force for the diagnosis and treatment of acute and chronic heart failure of the European
intake.
3
Society of Cardiology (ESC)Developed with the special contribution of the Heart Failure
Association (HFA) of the ESC. Eur Heart J. 2016;37(27):2129-200. Supplementary Table 2. HR and 95% CIs for the association between adherence to the DASH
diet and HF incidence after excluding people who experienced a MI before HF (n = 79).
46. Merry AHH, Boer JMA, Schouten LJ, et al. Validity of coronary heart diseases and heart
failure based on hospital discharge and mortality data in the Netherlands using the Tertiles DASH Score
cardiovascular registry Maastricht cohort study. Eur J Epidemiol 2009; 24(5):237-247. T1 (8 - 22) T2 (23 -26) T3 (27 - 39) P-trend
47. Khand AU, Shaw M, Gemmel I, et al. Do discharge codes underestimate hospitalisation n (individuals) 14,257 11,568 11,767
due to heart failure? Validation study of hospital discharge coding for heart failure. Eur J
n (HF cases) 190 193 189
Heart Fail 2005; 7(5):792-797.
Model 2 Reference 0.98 (0.80; 1.20) 0.80 (0.65; 0.99) 0.03
48. Ingelsson E, Ärnlöv J, Sundström J, et al. The validity of a diagnosis of heart failure in a
hospital discharge register. Eur J Heart Fail 2005; 7(5):787-791. DASH = Dietary Approaches to Stop Hypertension; HF = heart failure. Model 2: The multivariable
model was adjusted for age, sex, educational level, physical activity level, smoking status, alcohol
intake.

56 57
4
ABSTRACT

Background
The EAT-Lancet Healthy Reference Diet (HRD) was suggested to co-benefit public and
planetary health. We aimed to construct a diet-score measuring adherence to the HRD,
and to explore its association with coronary heart disease (CHD), all-cause mortality,
and environmental impact.

Methods
We included 37,349 particpants from the population-based EPIC-NL study. Data from
food frequency questionnaires were used to calculate scores on the HRD (0-140), which
was categorized in quartiles. Data on incident CHD and all-cause mortality were retrieved
through linkage with national registries. Data on environmental impact indicators were
obtained from life cycle assessments.

Results
High adherence to the HRD was associated with 15% lower risk of CHD (HRQ4vsQ1=0.85;
95% CI: 0.75, 0.96), and 17% lower risk of all-cause mortality (HRQ4vsQ1=0.83; 95% CI 0.77,
0.90). High adherence to the HRD was associated with lower environmental impact
from greenhouse gas emissions (βQ4vsQ1= -0.10 kg CO2-eq; 95% CI: -0.13, -0.07), land use
(βQ4vsQ1= -0.11 m2; 95% CI: -0.12, -0.09), freshwater eutrophication (βQ4vsQ1= -0.000002
kg P-eq; 95% CI: -0.000004, -0.000001), marine eutrophication (β Q4vsQ1= -0.00035 kg
N-eq; 95% CI: -0.00042, -0.00029), and terrestrial acidification (βQ4vsQ1 = -0.004 kg SO2-
eq; 95% CI: -0.004, -0.003), but with higher environmental impact from blue water use
(βQ4vsQ1=0.044 m3; 95% CI: 0.043, 0.045).

Adherence to the EAT-Lancet Healthy Conclusion

Reference Diet in relation to coronary High adherence to the HRD was associated with lower risk of CHD, all-cause mortality,

heart disease, all-cause mortality risk and and with lower environmental impact, although attention is needed for the associated

environmental impact: results from the

increase in blue water use.

EPIC-NL cohort

Chiara Colizzi, Marjolein C Harbers, Reina E Vellinga, WM Monique Verschuren, Jolanda

MA Boer, Elisabeth HM Temme, Yvonne T van der Schouw

Manuscript submitted to Circulation

chapter 4 the healthy reference diet in rel ation to health outcomes and environmental impac t
INTRODUCTION
Diet has a profound impact on human health as well as the environment (1). According
to the Global Burden of Disease Study 2017, 11 million deaths and 255 million DALYs can
be attributed to high sodium intake, and low intake of whole grains and fruit across the
world (2). Unhealthy diets are considered one of the main risk factors for the development
of cardiovascular diseases (3). At the same time, current dietary practices are likely to
exhaust our planet in the light of the expected growth of the world population (1). Food
production practices account for up to 30% of global greenhouse-gas emissions (GHGE)
and 70% of freshwater use(1), most of which is intended for meat and dairy production
(4-6). For these reasons, shifting towards healthy and sustainable diets could co-benefit
public and planetary health.
The EAT-Lancet Commission on Healthy Diets From Sustainable Food Systems is the
first large-scale and coordinated scientific collaboration to provide dietary guidelines on
healthy diets within the food production boundaries for the world population (7). The
commission proposed the Healthy Reference Diet (HRD), that was constructed based
on scientifically established targets for healthy diets and fitting within a safe operating
space of food systems, for which the Planetary Boundaries framework was used. The
diet includes high consumption of fruits and vegetables, whole grains, legumes, nuts,
and unsaturated oils; low to moderate consumption of dairy, starchy vegetables, poultry
and fish; and no or low consumption of saturated fats, red meat, and all sweeteners.
As such, the HRD generally emphasizes the intake of plant-based foods and suggests to
limit the intake of animal-sourced foods and starchy vegetables.
There is still limited evidence directly linking the HRD to cardiovascular outcomes and
mortality. The EAT-Lancet report projected that 19.0-23.6% of premature adult deaths
could potentially be avoided by adopting the HRD, while remaining within acceptable
environmental boundaries (7). However, these projections were based on theoretical
models. To date, only one study empirically assessed the association between the HRD
and the risk of coronary heart disease (CHD) and all-cause mortality, showing that better
adherence to the HRD was associated with 28% lower risk for CHD, but not with risk of
stroke or all-cause mortality (8). Potentially, this may relate to the dichotomous scoring
system that was applied, which consequently did not allow for large variation in HRD-
scores. Thus, evidence on the potential cardiovascular benefits of the HRD coming from
prospective cohort studies using a refined diet-score to measure adherence is currently
lacking. Additionally, the environmental impact of the HRD has not been previously
assessed empirically. Insight into the cardiovascular and planetary consequences of
adhering to the HRD would help to identify win-win or win-lose aspects of the HRD.
60
Therefore, the present study aimed to construct a refined HRD-score allowing for wide
variation in adherence to the HRD. Second, we aimed to estimate the association of
adherence to the HRD with CHD and all-cause mortality risk in a population-based cohort
study. Third, we aimed to estimate the associated environmental impact of the HRD
using a wide range of environmental indicators relating to the planetary boundaries in
the same population-based cohort study.
METHODS
Study population
We used data from the Dutch contribution to the European Prospective Investigation
into Cancer and Nutrition (EPIC-NL) (9). The EPIC study was designed to assess the
associations between diet, lifestyle, dietary intake, and the incidence of cancer and
other chronic conditions. The EPIC-NL cohort combines the MORGEN cohort (n = 22,654)
and the Prospect cohort (n = 17,357), resulting in a total of 40,011 participants. The 4
MORGEN cohort included both men and women, aged 20-64 years, from three Dutch
cities (Amsterdam, Doetinchem, and Maastricht), recruited between 1993 and 1997.
The Prospect cohort included women participating in a breast screening program, aged
49-70 years, recruited between 1993 and 1995 from Utrecht and its vicinity. At baseline,
participants completed a general questionnaire and a validated semi-quantitative food
frequency questionnaire (FFQ). During a physical examination a non-fasting blood
sample was taken, aliquoted and stored for future research. The EPIC-NL study was
conducted according to the guidelines in the Declaration of Helsinki and all procedures
involving the participants were approved by the institutional review board of the
University Medical Center Utrecht (Prospect-EPIC) and the medical ethical committee
of TNO Nutrition and Food Research (MORGEN-EPIC). All participants provided written
informed consent.
For the current study, we excluded participants who withheld permission for linkage
with national disease registries (n=1,666), those who withdrew informed consent during
follow-up (n=1), participants with prevalent CHD at baseline (n=366), participants with
missing dietary intake data (n=218), and particpants with implausible energy intake
(defined as those in the lowest and highest 0.5% of the ratio of energy intake over basal
metabolic rate) (n=400), leaving 37,360 persons for analysis (Supplementary Figure 1).
Calculation of the HRD adherence score
The FFQ included questions on the consumption of 178 food items in the year prior
to enrolment (9, 10). For some food items, questions were accompanied by images of
the food in different portion sizes, to assist in portion size estimation. Frequency of
61
chapter 4 the healthy reference diet in rel ation to health outcomes and environmental impac t
consumption was estimated in times per day, week, month, year or never. Average food
intake (g/d) was calculated by multiplying the consumption frequency with the consumed
amounts and nutrient intakes were calculated using the Dutch food composition table
of 1996 (11).
To asses adherence to the HRD, a Healthy Reference Diet score (HRD-score) was
constructed. To calculate the adherence scores, the dietary recommendations from
the EAT-Lancet report were recalculated on the basis of 2000 kcal/day for women, in
line with the recommended energy intake proposed by the Dutch dietary guidelines
(Supplementary Table 1 and 2). Participants were assigned proportional scores ranging
from 0-10 for each of the 14 dietary recommendations in the HRD (as proposed by EAT-
Lancet), that were then summed, resulting in a score ranging between 0 (no adherence)
and 140 (complete adherence). Each food group in the HRD-score was categorized into
one of the following scoring components adapted from Looman et al. (12): adequacy,
moderation, optimum or ratio. The allocation of scoring components to the dietary
recommendations in the HRD was informed by literature investigating the associations
of those food groups with chronic disease (13-35). Adequacy components are used to
score foods generally considered healthy and for which a high intake is recommended. In
the HRD-score, foods assigned to this component were whole grains, vegetables, fruits,
legumes, and soy foods. Participants received 10 points for meeting the recommended
intake for these food groups, 0 points for no consumption, and a proportional score
for intakes between zero and the recommended level. Moderation components were
used to score foods that could increase the risk of chronic diseases. The moderation
component was used to score beef, lamb, pork, and sweeteners. For these foods, 0
points were assigned if the intake was above the reference intake, 10 points were
assigned for an intake equal to or lower than the reference intake, and intermediate
intakes were assigned a proportional scoring. Optimum components comprise foods
which are nutritious yet potentially detrimental if eaten in large quantities on a daily
basis. The optimum component was used to score the following food groups: potatoes,
dairy, chicken, eggs, fish, and nuts. For these foods, participants with intakes within the
required optimum intake range would receive 10 points, while those with intakes lower
or higher than the optimum would be scored proportionally and symmetrically from 0
to 10 and from 10 to 0. Finally, a ratio component was used to describe the added fats
food group. For the added fats, no consumption of unsaturated fats or an unsaturated
to saturated fats ratio lower than 0.6 was assigned 0 points, while no consumption of
saturated fats or an unsaturated to saturated fats ratio higher than 13 was assigned 10
points. Ratios in between were scored proportionally. Cut-offs and threshold values for
the ratio component were derived from the 15th percentile and 85th percentile of the
intake distribution of the Dutch reference population, as described in Looman et al. (12).
62
Finally, the HRD-score was adjusted for energy intake (HRDea-score) using the energy-
adjusted nutrient residual model to remove the variance in dietary intake related to
total energy intake (36).
CHD and all-cause mortality ascertainment
CHD events included both fatal and non-fatal cases of CHD. Morbidity data were
obtained from the Dutch Center for Health Care Information, which holds a standardized
computerized registry of hospital discharge diagnoses. The hospital discharge diagnosis
database was linked to the cohort based on information of birthdate, sex, postal code,
and general practitioner with a validated probabilistic method (37). Hospitalization for
CHD was based on the principal diagnoses (ICD 10: I20-I25).
Information on vital status was obtained through linkage with the Dutch municipal
registry. All-cause mortality was defined as death from any cause after study inclusion.
For deceased participants, information on the causes of death was ascertained through 4
linkage with the Causes of Death Registry of Statistics Netherlands. Death from CHD was
based on both primary and secondary causes of death. A primary cause of death was
defined as death due to a CHD event, while a secondary cause of death was defined as
death due to complications of the primary cause, or another disease which could have
led to death. All participants were followed until CHD event, death, emigration, or end
of follow-up, whichever came first. Follow-up was complete until December 31st, 2010.
Environmental impact assessment
The ‘planetary boundaries’ within the planetary boundaries framework provide the
safe operating space for the Earth’s biophysical subsystems and or processes, and also
underlie the EAT-Lancet’s commission’s environmental impact assessments (38). Within
the planetary boundaries framework, the main environmental systems and processes
that are affected by food production are climate change, biodiversity loss, land system
change, freshwater use, and nitrogen and phosphorus flows (7). Within this framework,
the state of these systems is further defined by so-called control variables. As the main
environmental systems are interlinked and interdependent, most control variables
relate to multiple environmental systems. For example, greenhouse gas emissions
(GHGE) are an indicator of biodiversity loss and climate change; land use is an indicator
of biodiversity loss and land system change; blue water use (e.g, irrigation water) is
an indicator of biodiversity loss and freshwater use; eutrophication (e.g., through
application of fertilizer) is an indicator of nitrogen and phosphorus cycles, biodiversity
loss and climate change, and terrestrial acidification is an indicator of biodiversity loss (7,
38, 39). Therefore, the assessment of a wide range of environmental indicators provides
a holistic assessment of the environmental impact of the HRD. In the present study,
63
chapter 4 the healthy reference diet in rel ation to health outcomes and environmental impac t
we evaluate the effects of the HRD on GHGE (kg CO2-eq per day), land use (m2 per
year), blue water use (m3 per day), freshwater eutrophication (kg P-eq per day), marine
eutrophication (kg N-eq per day), and terrestrial acidification (kg SO2-eq day).
The associated environmental impact of the 178 foods and beverages were assessed
using the most recent Life Cycle Assessments (LCA) data from the Dutch LCA Food
database (40). This database is established by the National Institute for Public Health
and the Environment (RIVM) and contains information on the environmental impact
for approximately 250 Dutch foods and beverages. A full description of the data and
assumptions can be found elsewhere (41). In short, the LCAs had an attributional
approach and hierarchical perspective. System boundaries were from cradle till plate,
including primary production, processing, primary packaging, distribution, retail,
supermarket, storage, preparation by the consumer (e.g., cooking), and incineration
of packaging waste. Transport between all phases, except from retail to the consumer
was included. Economic allocation was applied for all food items, except for milk, where
physical allocation was used. In order to estimate daily environmental impact, LCA data
from the Dutch LCA Food database, referred to as primary data, was linked via NEVO-
codes to FFQ items. Extrapolations were carried out in case no primary LCA data were
available.
Ascertainment of covariates
Details on data collection on covariates are described elsewhere (9). In short, for age,
sex, educational level, smoking status and history, physical activity, and medication use
data from the baseline general questionnaire were used. Education was categorized into
low (lower vocational training and primary school), moderate (secondary school and
intermediate vocational training), and high educational level (higher vocational training
and university). Smoking status was categorized into never smoker, former smoker, or
current smoker. Alcohol intake was assessed from the FFQ, and measured in grams/day.
Physical activity was categorized into inactive, moderately inactive, moderately active,
and active, according to the Cambridge Physical Activity Index (CPAI) (42). Total energy
intake was also derived from the FFQ, and expressed in kilocalories/day.
The baseline physical examination provided data on body weight and height, blood
pressure and cholesterol levels (10). BMI was calculated as height divided by weight
squared, and participants were categorized as normal weight for a BMI ≤ 24.9 kg/m , 2
overweight for a BMI between 25 and 29.9 kg/m2, and obese for a BMI ≥ 30 kg/m2.
Both systolic and diastolic blood pressure were measured twice in supine position,
from which the mean was taken. Blood pressure measurements were performed on
the left arm, using a Boso Oscillomat in the MORGEN-EPIC cohort, and a random zero
64
Sphygmomanometer in the Prospect-EPIC cohort (9). Hypertension was defined as use of
hypertensive medication, and/or systolic blood pressure >140 and/or diastolic pressure
>90.9 Serum total cholesterol (mmol/l) was measured using enzymatic methods.9
Statistical analysis
All baseline characteristics are reported by quartiles of the HRDea-score. Normally
distributed continuous variables are presented as means with standard deviations.
Continuous variables with a skewed distribution are presented as median with
interquartile range (IQR). Categorical variables are presented as counts and percentages.
A Cox proportional hazard model was used to obtain hazard ratios (HR) and 95%
confidence intervals (CI) for the association between quartiles of the HRDea-score
and CHD risk and all-cause mortality. The lowest quartile was used as reference. The
underlying time variable was age from study entry to either diagnosis, death, or end of
follow-up (31-12-2010), whichever came first. The proportional hazards assumption was
checked using the Schoenfeld test, with no violations observed. 4
For CHD and all-cause mortality outcomes, the analyses present first the unadjusted
model with crude estimates. Model 1 was adjusted for age and sex and model 2 was
additionally adjusted for educational level, smoking, alcohol consumption, physical
activity, and energy intake. A sensitivity analysis was conducted, where we added BMI,
total cholesterol, and hypertension to the multivariable-adjusted model (model 2), as
these factors may be potential mediators in the association between the HRDea-score
and CHD. All mediators were first added individually and then simultaneously.
All foods in the FFQ, expressed in grams/day, had an estimated environmental impact
calculated with LCA. We used linear regression models to estimate the association
between HRDea-score and each environmental indicator. In this linear regression the
exposure was the HRDea-score and the outcome was the environmental indicator,
calculated as the sum of the associated environmental impact of the food groups
included in the HRD. The lowest quartile was used as reference. The analyses present
first the crude estimates, and then in model 1 estimates were adjusted for age, sex,
and energy intake. No other variables were included, based on current literature on the
environmental impact of diets. The p-value for trend across quartiles was estimated
by modelling the median value of each quartile as a continuous variable. Statistical
significance was set at a two-tailed P < 0.05. All statistical analyses were carried out
using STATA 13.SE (StataCorp LP, College Station, TX, USA). Reporting was guided by
the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE)
recommendations for nutritional epidemiology (43).
65
chapter 4 the healthy reference diet in rel ation to health outcomes and environmental impac t
RESULTS
Table 1 shows the baseline characteristics of the study population across quartiles of
the HRDea-score. The average score was 73 (SD=10), and ranged between 32 and 116.
Participants most adherent to the HRD were more likely to be female, have a normal BMI,
be highly educated, have never smoked, and consume less calories per day compared
to the least adherent.
HRDea-score and CHD
During a median follow-up of 15.1 years, a total of 2,543 CHD events occurred. High
adherence to the HRD was associated with a lower risk of CHD (HRQ4vsQ1 0.85; 95% CI
0.75 – 0.96) in fully adjusted models (Table 2). Adding each potential mediator separately
to the multivariable-adjusted model only slightly attenuated the results (Supplementary
Table 3). The cumulative effect of all mediators also did not substantially alter the results
(HRQ4vsQ1 0.88; 95% CI 0.77 – 0.99).
HRDea-score and all-cause mortality
During a median follow-up of 15.3 years, 5648 people died from all causes. High
adherence to the HRD was associated with a lower risk of all-cause mortality (HRQ4vsQ1
0.83; 95% CI 0.77 – 0.90) in fully adjusted models (Table 3).
HRDea-score and environmental impact
Table 4 shows the baseline means of GHG emissions, land use, blue water use, freshwater
eutrophication, marine eutrophication, and terrestrial acidification across quartiles of
the HRDea-score. Participants most adherent to the HRD were more likely to consume
diets that were associated with less GHGE, land use, freshwater eutrophication, marine
eutrophication and terrestrial acidification compared to the least adherent. Yet, diets of
those most adherent to the HRD have higher blue water use compared to diets of those
least adhering to the HRD.
In multivariable adjusted models, high adherence to the HRD was associated with lower
GHGE (β= -0.10 kg CO2-eq; 95%CI: -0.13, -0.07), less land use (β= -0.11 m2 per year; 95%
CI: -0.12, -0.09), less freshwater eutrophication (β= -0.000002 kg P-eq; 95%CI: -0.000004,
-0.000001), less marine eutrophication (β= -0.00035 kg N-eq; 95%CI: -0.00042, -0.00029)
and less terrestrial acidification (β = -0.004 kg SO2-eq; 95% CI: -0.004, -0.003) and with
higher blue water use (β=0.044; 95% CI; 95%CI: 0.043, 0.045) when comparing extreme
quartiles (Table 5). These beta-coefficients correspond to 1.7% lower GHGE, 3.2% less
land use, 0.5% less freshwater eutrophication, 3.5% less marine eutrophication, 6.3%
66
less terrestrial acidification, but with 31.4% higher blue water use, when comparing
extreme quartiles.
DISCUSSION
In the present study among 37,349 Dutch adults, we found that higher adherence to
the HRD as proposed by the EAT-Lancet Commission was associated with 15% lower
risk of CHD, and with 17% lower risk of all-cause mortality. Higher adherence to the diet
was also associated with 1.7% lower GHGE, 3.2% less land use, 0.5% less freshwater
eutrophication, 3.5% less marine eutrophication, 6.3% less terrestrial acidification, but
with 31.4% higher blue water use.
Before we can interpret our results, we need to address the limitations of the present
study. Even though overall the FFQ was considered adequate to assess food intake
of the EPIC-NL population, the validity of vegetable and fish intakes was found to be 4
quite poor (10). This would suggest possible measurement error in the present study
for vegetable and fish intake as well. As misclassification of these food groups is likely
to be random given the prospective design of the present study, and considering that
these food groups are generally associated with lower risk for CHD, misclassification
of the intake of these foods could indicate an attenuation of the inverse association
with CHD and mortality risk. Similarly, there may be underestimation of the effects on
environmental impact indicators. Moreover, dietary assessment was conducted only at
baseline and dietary intake might have changed during follow-up. However, a previous
study in EPIC-NL showed dietary changes between baseline and 20 years follow-up to
be relatively modest (44). Finally, the current study used the Dutch LCA Database to
calculate environmental indicators. It should be noted that, although the LCA database
is a comprehensive source of LCA indicators, there is also some uncertainty in the data
since they are modelled and not actually measured (41). Furthermore, LCA estimates for
the Netherlands will likely not be fully generalizable to other contexts.
The main strength of this study is the use of a prospective design, based on a large
population cohort, and a long follow-up period. Moreover, we used a proportional
scoring from 0 to 10 for each component of the HRD-score, which is likely to capture
the variability in dietary intake. Additionally, the current study created a refined diet
score which could be used or adapted by other studies who wish to study the HRD in
other settings. Another strength is the linkage with national registries to ascertain health
outcomes which is considered a valid method to reach near-complete follow-up and to
reduce possible outcome misclassification (45). Finally, the present study included a wide
67
chapter 4 the healthy reference diet in rel ation to health outcomes and environmental impac t

range of environmental indicators, which appeals to the need for an integrated analysis
of the core environmental impact dimensions of food systems (7).
The EAT-Lancet report leaves some space for definition of the HRD, so that
recommendations can be tailored to different populations. Thus, for the construction
of the HRD-score, several choices were made in assigning foods to each scoring
component, such as the inclusion of dairy and starchy vegetables in the optimum
component. Depending on the population and cultural context, some might prefer
assigning these food groups to an adherence or moderation component. Additionally,
intake recommendations in grams per day from the EAT-Lancet report were energy-
adjusted for women, to account for their generally lower energy requirements. Since
these choices were mostly based on the baseline characteristics of this study population,
they might not be entirely appropriate when replicating this study in a different setting.
Findings from the present study are largely in line with the study from Knuppel et al.
which used a more simple score to reflect HRD-adherence, and found similar inverse
associations for CHD risk and all-cause mortality, although the latter did not reach
statistical significance (8). Even though the HRD-score is unique to this study, other
studies investigating dietary indices focusing on plant-based diets, show inverse
associations with either CHD or all-cause mortality (46, 47). Differences in the magnitude
of risk reductions between the present study and available literature are likely related to
the scoring methods, the baseline characteristics of the populations, and/or to residual
confounding.
The Dutch Healthy Diet index 2015 (DHD15-index) – a diet-score reflecting adherence
to the Dutch national dietary guidelines – has in EPIC-NL previously been related to
environmental sustainability. In line with our findings, these studies also observed
adherence to the DHD15 to be associated with lower GHGE and less land use, but
with higher use of blue water (48, 49). Indeed, several plant-based foods – which are
emphasized in both the DHD15 and the HRD - do have a relatively high blue water use
per kg product, such as several fruits and nuts (50). Plant-based foods with a relatively
high blue water use are often imported into The Netherlands from areas with a high
water scarcity, such as citrus fruits from Spain or almonds from the USA. In order to
reduce the blue water footprint of the HRD diet in a Dutch context, choosing locally-
grown and seasonal fruits and vegetables may be recommended (50). Thus, is plausible
that with small changes in the choice for type of fruits, and when choosing for seasonal
and locally grown fruits and vegetables, the high blue water use associated with high
HRD-adherence could be diminished.
4
CONCLUSION
This study provides evidence from a prospective cohort study in The Netherlands for
an inverse association of adherence to the HRD with CHD and all-cause mortality. This
research also found that increasing adherence to the HRD could lower some aspects of
the environmental impact of diets, but attention is needed for the associated increase
in blue water use.

With regard to the environmental impact of the HRD, the indicators used in this study
are largely in line with the planetary boundaries framework (38, 39), which is also
applied by the EAT-Lancet Commission to model the environmental effects of the HRD.
Although there was a significant increase in blue water use, the observed percentage
reductions for GHGE, land use, freshwater eutrophication, marine eutrophication and
terrestrial acidification in fully adjusted models seem modest. These findings are in line
with findings from the EAT-Lancet Commission, showing that dietary changes alone are
not sufficient to stay within most planetary boundaries, except for GHGE, for which a
reduction of 49% was observed when comparing current diets with the HRD (7). The
discrepancy in GHGE reductions between the EAT-Lancet report and the current study
could be due to the fact that in the EPIC-NL population the maximum HRD-score reached
was only 116, while complete adherence would yield 140 points. Thus, observed diets
may still be suboptimal, and further improvements towards the HRD may have larger
effects on environmental impact indicators. Also, other measures such as improved
production practices and less food waste and loss are needed as well (7).

68 69
 Table 1. Baseline characteristics of the EPIC-NL cohort by quartiles of the HRDea-score (n=37,360)1

70
Quartiles of HRDea-scores (range)
chapter
4

Q1 (32-66) Q2 (67-73) Q3 (74-79) Q4 (80-116)

N=9340 N=9340 N=9340 N=9340
Sex
Male 3773 (40.4) 2753 (29.5) 1799 (19.3) 1086 (11.6)
Female 5567 (59.6) 6587 (70.5) 7541 (80.7) 8254 (88.4)
Age 48.5 (36.8, 55.5) 51.0 (41.4, 57.3) 52.5 (44.9, 58.6) 52.7 (46.9, 58.9)
BMI
Normal weight 4114 (44.7) 4391 (47.7) 4359 (47.4) 4848 (52.7)
Overweight 3731 (40.5) 3559 (38.6) 3640 (39.6) 3319 (36.1)
Obesity 1357 (14.7) 1261 (13.7) 1195 (13.0) 1035 (11.2)
Educational level
Low 5912 (63.7) 5612 (60.4) 5375 (57.8) 4639 (49.9)
Moderate 2122 (22.9) 2036 (21.9) 1951 (21.0) 1976 (21.3)
High 1250 (13.5) 1645 (17.7) 1966 (21.2) 2681 (28.8)
Smoking
Never 3152 (33.9) 3487 (37.5) 3751 (40.3) 3852 (41.3)
Former 2466 (26.5) 2860 (30.7) 3151 (33.9) 3397 (36.5)
Current 3687 (39.6) 2958 (31.8) 2403 (25.8) 2067 (22.2)
Physical activity
Inactive 889 (9.5) 728 (7.8) 666 (7.1) 524 (5.6)
Moderately inactive 2259 (24.2) 2371 (25.4) 2348 (25.1) 2294 (24.6)
Moderately active 2260 (24.2) 2412 (25.8) 2442 (26.1) 2588 (27.7)
Active 3932 (42.1) 3829 (41.0) 3884 (41.6) 3934 (42.1)

Table 1. Baseline characteristics of the EPIC-NL cohort by quartiles of the HRDea-score (n=37,360)1 (continued)

Quartiles of HRDea-scores (range)

Q1 (32-66) Q2 (67-73) Q3 (74-79) Q4 (80-116)
N=9340 N=9340 N=9340 N=9340
Alcohol consumption, g/day 5.3 (0.7, 17.3) 4.9 (0.7, 15.6) 4.8 (0.7, 14.8) 5.0 (0.7, 14.8)
Energy intake, kcal/day 2269 (1872, 2738) 2020 (1704, 2424) 1860 (1568, 2199) 1736 (1473, 2046)
Food consumption, g/day
Whole grains 20.7 (2.4, 85.2) 49.1 (6.4, 107.4) 72.1 (15.0, 127.4) 96.5 (46.8, 134.1)
Vegetables 89.5 (66.7, 117.0) 99.2 (74.7, 129.4) 105.9 (81.9, 136.9) 125.8 (95.4, 166.2)
Fruit 105.1 (49.8, 180.7) 136.9 (90.7, 250.2) 190.6 (122.4, 278.7) 241.4 (158.4, 323.5)
Potatoes and cassava 142.9 (105.1, 183.9) 106.8 (67.9, 156.9) 76.9 (49.6, 111.4) 60.1 (38.7, 81.0)
Dairy foods2 538.6 (258.5, 722.6) 417.5 (232.4, 613.2) 381.0 (232.5, 543.5) 323.6 (207.8, 427.2)
Legumes 3 23.7 (14.7, 35.9) 26.9 (17.2, 40.2) 29.0 (18.8, 42.0) 33.8 (23.2, 46.4)
Soy 0.0 (0.0, 0.0) 0.0 (0.0, 0.0) 0.0 (0.0, 0.1) 0.0 (0.0, 4.8)
Beef, lamb, and pork 103.7 (70.8, 136.3) 93.5 (59.3, 123.9) 82.8 (51.9, 111.4) 64.0 (34.5, 97.5)
Chicken 10.3 (4.7, 17.3) 9.6 (4.3, 16.3) 9.4 (4.1, 15.9) 7.9 (2.9, 14.6)
Eggs 21.4 (10.5, 28.6) 14.3 (7.6, 21.4) 13.8 (7.1, 17.6) 10.5 (5.8, 15.7)
Fish 7.2 (2.8, 14.0) 7.4 (3.3, 14.0) 8.0 (3.3, 15.2) 8.4 (3.3, 15.9)
Nuts 4.3 (1.4, 11.7) 4.1 (1.4, 10.7) 3.8 (1.4, 9.7) 4.7 (1.7, 11.4)
Unsaturated fats 11.3 (4.9, 20.8) 10.9 (5.1, 19.6) 10.1 (5.0, 17.8) 9.9 (5.1, 16.9)
Saturated fats 33.5 (21.4, 49.2) 28.5 (17.6, 42.5) 24.7 (14.4, 37.8) 22.3 (12.3, 35.0)
Added sugars 193.7 (117.4, 305.9) 178.4 (108.9, 271.4) 170.1 (101.3, 251.0) 159.8 (93.3, 231.7)
1
Estimates are presented as counts n and percentages (%) or as medians (p25, p75). 2Including whole milk, derivate equivalents and cheese. 3Including
beans, lentils, and peas.
71
the healthy reference diet in rel ation to health outcomes and environmental impac t

4
 Table 2. Hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between quartiles of the HRDea-score and incident of CHD (n=37,360).

72
Quartiles of HRDea-scores (range)
chapter
4

Q1 (32-66) Q2 (67-73) Q3 (74-79) Q4 (80-116) P-trend

Cases, n 700 684 606 553
Persons-years 135165 135188 135810 135745
Unadjusted Model 1.00 [ref] 0.98 (0.88 – 1.09) 0.86 (0.77 - 0.96) 0.79 (0.70 – 0.88) < 0.001
a
Model 1 1.00 [ref] 0.91 (0.82 – 1.01) 0.79 (0.71 - 0.88) 0.74 (0.66 – 0.83) < 0.001
Model 2b 1.00 [ref] 0.96 (0.86 – 1.07) 0.86 (0.77 – 0.96) 0.85 (0.75 – 0.96) 0.003
a b
Adjusted for age and sex. Adjusted for age, sex, educational level, smoking status, alcohol consumption, physical activity, and energy intake. Data
was missing for smoking status n=129 (0.0034%) and educational level n=195 (0.0052%).

Table 3. Hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between quartiles of the HRDea-score and all-cause mortality
(n=37,360).

Quartiles of HRDea-scores (range)

Q1 (32-66) Q2 (67-73) Q3 (74-79) Q4 (80-116) P-trend
Cases, n 1327 1470 1516 1335
Persons-years 139528 139337 139473 139236
Unadjusted Model 1.00 [ref] 1.12 (1.04 – 1.21) 1.18 (1.10 - 1.27) 1.05 (0.98 – 1.14) 0.079
a
Model 1 1.00 [ref] 0.92 (0.86 – 0.99) 0.86 (0.80 - 0.93) 0.74 (0.68 – 0.80) < 0.001
Model 2b 1.00 [ref] 0.97 (0.90 – 1.05) 0.94 (0.87 – 1.02) 0.83 (0.77 – 0.90) < 0.001
a b
Adjusted for age and sex. Adjusted for age, sex, educational level, smoking status, alcohol consumption, physical activity, and energy intake. Data
was missing for smoking status n=129 (0.0034%) and educational level n=195 (0.0052%).

Table 4. Descriptive statistics of the environmental impact indicators by quartiles of the HRDea-score (n=37,360)1

Quartiles of HRDea-score (range)

Q1 (32-66) Q2 (67-73) Q3 (74-79) Q4 (80-116)
Greenhouse gases (Kg CO2-eq) 6.07 (1.72) 5.61 (1.51) 5.32 (1.41) 5.03 (1.32)
2
Land use (m per year) 3.47 (0.95) 3.19 (0.85) 3.00 (0.79) 2.82 (0.74)
Blue water use (m3 per day) 0.14 (0.05) 0.15 (0.05) 0.16 (0.05) 0.17 (0.05)
Freshwater eutrophication (Kg P-eq) 0.00043 (0.00012) 0.00039 (0.00011) 0.00037 (0.00010) 0.00035 (0.00009)
Marine eutrophication (Kg N-eq) 0.010 (0.003) 0.010 (0.003) 0.009 (0.003) 0.009 (0.003)
Terrestrial acidification (Kg SO2-eq) 0.064 (0.021) 0.059 (0.018) 0.055 (0.017) 0.051 (0.017)
1
All values are presented as means (SD).
73
the healthy reference diet in rel ation to health outcomes and environmental impac t

4
 chapter 4 the healthy reference diet in rel ation to health outcomes and environmental impac t

Table 5. Regression coefficients and 95% confidence intervals (CI) for the association between quartiles of the HRDea-score and environmental

P-trend

< 0.001
< 0.001

< 0.001
< 0.001

< 0.001
< 0.001

< 0.001

< 0.001

< 0.001
< 0.001

< 0.001
< 0.001
REFERENCES
1. Tilman D, Clark M. Global diets link environmental sustainability and human health. Nature.
2014;515:518-22.

-0.0019 (-0.0020, -0.0018)

(-0.000004, -0.000001)
(-0.000077, -0.000071)

-0.004 (-0.004, -0.003)

-0.012 (-0.013, -0.012)
(-0.00042, -0.00029)
2. Afshin A, Sur PJ, Fay KA, Cornaby L, Ferrara G, Salama JS, et al. Health effects of dietary

0.044 (0.043, 0.045)

0.031 (0.030, 0.032)
-0.66 (-0.68, -0.63)
-0.10 (-0.13, -0.07)
-1.04 (-1.08, -0.99)

-0.11 (-0.12, -0.09)

risks in 195 countries, 1990–2017: a systematic analysis for the Global Burden of Disease
Q4 (80-114)

-0.000002
-0.000074

-0.00035
Study 2017. The Lancet. 2019;393:1958-72.
3. Arnett DK, Blumenthal RS, Albert MA, Buroker AB, Goldberger ZD, Hahn EJ, et al. 2019
ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease: A Report of the
American College of Cardiology/American Heart Association Task Force on Clinical Practice
Guidelines. Circulation. 2019;140(11):e596-e646.
4. de Vries M, de Boer IJM. Comparing environmental impacts for livestock products: A review
of life cycle assessments. Livestock Science. 2010;128(1):1-11.

-0.0013 (-0.0014, -0.0012)

(-0.000004, -0.000001)
(-0.000058, -0.000053)

-0.008 (-0.009, -0.008)

-0.002 (-0.002, -0.001)
5. Tukker A, Goldbohm RA, de Koning A, Verheijden M, Kleijn R, Wolf O, et al. Environmental

(-0.00021, -0.00008)
HRDea-score (range)

-0.07 (-0.08, -0.05)

-0.47 (-0.49, -0.45)

0.029 (0.027, 0.03)

-0.75 (-0.79, -0.71)

0.018 (0.017, 0.02)

-0.05 (-0.08, -0.3)

impacts of changes to healthier diets in Europe. Ecological Economics. 2011;70:1776-88.

Q3 (74-79)

-0.000003
-0.000056

-0.00014
6. Hedenus F, Wirsenius S, Johansson DJA. The importance of reduced meat and dairy
consumption for meeting stringent climate change targets. Climatic Change. 2014;124:79-91.
4
7. Willett W, Rockström J, Loken B, Springmann M, Lang T, Vermeulen S, et al. Food in the
Anthropocene: the EAT-Lancet Commission on healthy diets from sustainable food
systems. Lancet. 2019;393(10170):447-92.
8. Knuppel A, Papier K, Key TJ, Travis RC. EAT-Lancet score and major health outcomes: the
EPIC-Oxford study. Lancet. 2019;394(10194):213-4.
-0.0008 (-0.0008, -0.0007)
(-0.000004, -0.000001)
(-0.000036, -0.000031)

-0.005 (-0.006, -0.005)

-0.001 (-0.001, -0.001)
9. Beulens JWJ, Monninkhof EM, Monique Verschuren WM, van der Schouw YT, Smit J,
(-0.00015, -0.00002)
0.017 (0.016, 0.018)
-0.04 (-0.06, -0.03)
-0.28 (-0.31, -0.26)

0.01 (0.009, 0.012)

-0.05 (-0.07, -0.02)
-0.46 (-0.5, -0.42)

Ocke MC, et al. Cohort profile: The EPIC-NL study. International Journal of Epidemiology.
Q2 (68-73)

-0.000034

-0.000002

-0.00009

2010;39:1170-8.
10. Ocké MC, Bueno-De-Mesquita HB, Goddijn HE, Jansen A, Pols MA, Van Staveren WA, et al.
The Dutch EPIC Food Frequency Questionnaire. I. Description of the questionnaire, and
relative validity and reproducibility for food groups. International Journal of Epidemiology.
1997;26.
11. NEVO-table. Nederlands Voedingsstoffen-tabel (NEVO-tabel) 2011 (Dutch Food
Composition Table 2011) version 3 [Internet]. RIVM/Dutch Nutrition Centre. 1996.
Q1 (29-67)

0 [ref]
0 [ref]

0 [ref]
0 [ref]

0 [ref]
0 [ref]

0 [ref]

0 [ref]

0 [ref]
0 [ref]

0 [ref]
0 [ref]

12. Looman M, Feskens EJM, De Rijk M, Meijboom S, Biesbroek S, Temme EHM, et al.
Development and evaluation of the Dutch Healthy Diet index 2015. Public Health Nutrition.
Adjusted for age, sex, and energy intake.

2017;20:2289-99.
Freshwater eutrophication (Kg P-eq)

13. Wang X, Ouyang Y, Liu J, Zhu M, Zhao G, Bao W, et al. Fruit and vegetable consumption
Terrestrial acidification (Kg SO2-eq)

and mortality from all causes, cardiovascular disease, and cancer: Systematic review and
Marine eutrophication (Kg N-eq)

dose-response meta-analysis of prospective cohort studies. BMJ (Online). 2014;349.

14. Aune D, Giovannucci E, Boffetta P, Fadnes LT, Keum NN, Norat T, et al. Fruit and vegetable
intake and the risk of cardiovascular disease, total cancer and all-cause mortality-A
Blue water use (m3/d)
indicators (n=37,360).

systematic review and dose-response meta-analysis of prospective studies. International

Unadjusted model

Unadjusted model

Unadjusted model

Unadjusted model

Unadjusted model

Unadjusted model
GHGE (kg CO2-eq)

Journal of Epidemiology. 2017;46:1029-56.

Land Use (m /y)

15. Yip CSC, Chan W, Fielding R. The Associations of Fruit and Vegetable Intakes with Burden
2

of Diseases: A Systematic Review of Meta-Analyses. Journal of the Academy of Nutrition

a

Model 1a

Model 1a

Model 1a
Model 1

Model 1

Model 1

and Dietetics. 2019;119:464-81.

a

74 75
chapter 4 the healthy reference diet in rel ation to health outcomes and environmental impac t
16. Micha R, Shulkin ML, Peñalvo JL, Khatibzadeh S, Singh GM, Rao M, et al. Etiologic effects
and optimal intakes of foods and nutrients for risk of cardiovascular diseases and diabetes:
Systematic reviews and meta-analyses from the nutrition and chronic diseases expert
group (NutriCoDE). PLoS ONE. 2017;12.
17. Lea B, Eric B, Walter C, John PF. Potato intake and incidence of hypertension: Results from
three prospective US cohort studies. BMJ (Online). 2016;353.
18. Bertoia ML, Mukamal KJ, Cahill LE, Hou T, Ludwig DS, Mozaffarian D, et al. Changes in Intake
of Fruits and Vegetables and Weight Change in United States Men and Women Followed for
Up to 24 Years: Analysis from Three Prospective Cohort Studies. PLoS Medicine. 2015;12.
19. Guo J, Astrup A, Lovegrove JA, Gijsbers L, Givens DI, Soedamah-Muthu SS. Milk and
dairy consumption and risk of cardiovascular diseases and all-cause mortality: dose–
response meta-analysis of prospective cohort studies. European Journal of Epidemiology.
2017;32:269-87.
20. Aune D, Navarro Rosenblatt DA, Chan DSM, Vieira AR, Vieira R, Greenwood DC, et al. Dairy
products, calcium, and prostate cancer risk: A systematic review and meta-analysis of
cohort studies. American Journal of Clinical Nutrition. 2015;101:87-117.
21. Chen M, Sun Q, Giovannucci E, Mozaffarian D, Manson JE, Willett WC, et al. Dairy
consumption and risk of type 2 diabetes: 3 cohorts of US adults and an updated meta-
analysis. BMC Medicine. 2014;12:215.
22. Song M, Fung TT, Hu FB, Willett WC, Longo VD, Chan AT, et al. Association of animal and
plant protein intake with all-cause and cause-specific mortality. JAMA Internal Medicine.
2016;176:1453-63.
23. Pan A, Sun Q, Bernstein AM, Schulze MB, Manson JAE, Stampfer MJ, et al. Red meat
consumption and mortality: Results from 2 prospective cohort studies. Archives of Internal
Medicine. 2012;172:555-63.
24. Del Gobbo LC, Imamura F, Aslibekyan S, Marklund M, Virtanen JK, Wennberg M, et al. ω-3
Polyunsaturated fatty acid biomarkers and coronary heart disease: Pooling project of 19
cohort studies. JAMA Internal Medicine. 2016;176:1155-66.
25. Rong Y, Chen L, Zhu T, Song Y, Yu M, Shan Z, et al. Egg consumption and risk of coronary
heart disease and stroke: Dose-response meta-analysis of prospective cohort studies.
BMJ (Online). 2013;346.
26. Viguiliouk E, Glenn AJ, Nishi SK, Chiavaroli L, Seider M, Khan T, et al. Associations between
Dietary Pulses Alone or with Other Legumes and Cardiometabolic Disease Outcomes: An
Umbrella Review and Updated Systematic Review and Meta-analysis of Prospective Cohort
Studies. Advances in Nutrition. 2019;10(Supplement_4):S308-S19.
27. Nöthlings U, Schulze MB, Weikert C, Boeing H, Van Der Schouw YT, Bamia C, et al. Intake of
vegetables, legumes, and fruit, and risk for all-cause, cardiovascular, and cancer mortality
in a European diabetic population. Journal of Nutrition. 2008;138:775-81.
28. Grosso G, Marventano S, Yang J, Micek A, Pajak A, Scalfi L, et al. A comprehensive meta-
analysis on evidence of Mediterranean diet and cardiovascular disease: Are individual
components equal? Critical Reviews in Food Science and Nutrition. 2017;57:3218-32.
29. Nagata C, Wada K, Tamura T, Konishi K, Goto Y, Koda S, et al. Dietary soy and natto intake
and cardiovascular disease mortality in Japanese adults: The Takayama study. American
Journal of Clinical Nutrition. 2017;105:426-31.
30. Talaei M, Koh WP, van Dam RM, Yuan JM, Pan A. Dietary soy intake is not associated with
risk of cardiovascular disease mortality in Singapore Chinese adults. Journal of Nutrition.
2014;144:921-8.
76
31. Grosso G, Micek A, Godos J, Pajak A, Sciacca S, Galvano F, et al. Dietary Flavonoid and Lignan
Intake and Mortality in Prospective Cohort Studies: Systematic Review and Dose-Response
Meta-Analysis. American Journal of Epidemiology. 2017;185:1304-16.
32. Aune D, Keum NN, Giovannucci E, Fadnes LT, Boffetta P, Greenwood DC, et al. Nut
consumption and risk of cardiovascular disease, total cancer, all-cause and cause-specific
mortality: A systematic review and dose-response meta-analysis of prospective studies.
BMC Medicine. 2016;14:1-14.
33. Morenga LT, Mallard S, Mann J. Dietary sugars and body weight: Systematic review and
meta-analyses of randomised controlled trials and cohort studies. BMJ (Online). 2013;345.
34. Malik VS, Popkin BM, Bray GA, Després JP, Willett WC, Hu FB. Sugar-sweetened beverages
and risk of metabolic syndrome and type 2 diabetes: A meta-analysis. Diabetes Care.
2010;33:2477-83.
35. Yang Q, Zhang Z, Gregg EW, Flanders WD, Merritt R, Hu FB. Added sugar intake and
cardiovascular diseases mortality among us adults. JAMA Internal Medicine. 2014;174:516-24.
36. Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake in epidemiologic studies:
American Society for Nutrition; 1997.
37. Herings RMC, Bakker A, Stricker BHC, Nap G. Pharmaco-morbidity linkage: a feasibility
study comparing morbidity in two pharmacy based exposure cohorts. Journal of 4
Epidemiology and Community Health. 1992;46:136-40.
38. Rockström J, Steffen W, Noone K, Persson Å, Chapin FS, Lambin EF, et al. A safe operating
space for humanity. Nature. 2009;461(7263):472-5.
39. Steffen W, Richardson K, Rockström J, Cornell SE, Fetzer I, Bennett EM, et al.
Planetary boundaries: Guiding human development on a changing planet. Science.
2015;347(6223):1259855.
40. Milieubelasting van voedingsmiddelen [Internet]. 2021 [cited 06-04-2021]. Available from:
https://www.rivm.nl/voedsel-en-voeding/duurzaam-voedsel/database-milieubelasting-
v o e d i n g s m i d d e l e n # :~ : t e x t = M i l i e u % 2 D i n d i c a t o r e n , - D e % 2 0 l e v e n s c y c l u s % 2 0
milieubelasting&text=Vermesting%20zoet%20water%20(kg%20P,Waterverbruik%20(m3)
41. Vellinga RE, van de Kamp M, Toxopeus IB, van Rossum CTM, de Valk E, Biesbroek S, et al.
Greenhouse Gas Emissions and blue water use of dutch diets and its association with
health. Sustainability (Switzerland). 2019;11:6027.
42. Wareham NJ, Jakes RW, Rennie KL, Schuit J, Mitchell J, Hennings S, et al. Validity and
repeatability of a simple index derived from the short physical activity questionnaire
used in the European Prospective Investigation into Cancer and Nutrition (EPIC) study.
Public Health Nutrition. 2003;6:407-13.
43. Lachat C, Hawwash D, Ocké MC, Berg C, Forsum E, Hörnell A, et al. Strengthening the
Reporting of Observational Studies in Epidemiology-Nutritional Epidemiology (STROBE-
nut): An Extension of the STROBE Statement. PLoS Med. 2016;13(6):e1002036.
44. Biesbroek S, Verschuren WM, Boer JM, van der Schouw YT, Sluijs I, Temme EH. Are our
diets getting healthier and more sustainable? Insights from the European Prospective
Investigation into Cancer and Nutrition - Netherlands (EPIC-NL) cohort. Public Health Nutr.
2019;22(16):2931-40.
45. Merry AHH, Boer JMA, Schouten LJ, Feskens EJM, Verschuren WMM, Gorgels APM, et al.
Validity of coronary heart diseases and heart failure based on hospital discharge and
mortality data in the Netherlands using the cardiovascular registry Maastricht cohort
study. European Journal of Epidemiology. 2009;24:237-47.
77
chapter 4 the healthy reference diet in rel ation to health outcomes and environmental impac t

46. Satija A, Bhupathiraju SN, Spiegelman D, Chiuve SE, Manson JAE, Willett W, et al. Healthful
and Unhealthful Plant-Based Diets and the Risk of Coronary Heart Disease in U.S. Adults. CHAPTER 4 – SUPPLEMENTARY FILES
Journal of the American College of Cardiology. 2017;70:411-22.
47. Martínez-González MA, Sánchez-Tainta A, Corella D, Salas-Salvadó J, Ros E, Arós F, et al. A
provegetarian food pattern and reduction in total mortality in the Prevención con Dieta • Baseline EPIC-NL cohort (PROSPECT n= 17,357; MORGEN n=22,654)
Mediterránea (PREDIMED) study: American Society for Nutrition; 2014. 40,011
48. Biesbroek S, Verschuren WMM, Boer JMA, Van De Kamp ME, Van Der Schouw YT,
Geelen A, et al. Does a better adherence to dietary guidelines reduce mortality risk and • Individual who withdrew consent (n=1)
40,010
environmental impact in the Dutch sub-cohort of the European Prospective Investigation
into Cancer and Nutrition? British Journal of Nutrition. 2017;118:69-80.
49. Biesbroek S, Monique Verschuren WM, van der Schouw YT, Sluijs I, Boer JMA, Temme EHM. • Individuals with incomplete FFQ (n=218)
39,792
Identification of data-driven Dutch dietary patterns that benefit the environment and are
healthy. Climatic Change. 2018;147:571-83.
• Individuals with extreme energy intake (n=400)
50. Hollander A, Vellinga R, De V, Ido T, De K. The Global Blue Water Use for the Dutch Diet 39,392
and Associated Environmental Impact on Water Scarcity. World Journal of Food Science
and Technology. 2021;5:10-8.
• Prevalent cases of CHD (n=366)
39,026

4
• Individuals who withdrew permission for linkage with disease registries (n=1,666)
37,360

• Final study population

37,360

Supplementary Figure 1. Flowchart of the study population

78 79
 Supplementary Table 1. Construction of HRD-score, intake recommendations (g/day) and scoring system for men based on 2500 kcal/day

80
Food group Component HRD recommendation Minimum score Proportional Maximum points Proportional
chapter

type1 (g/day) (0 points) score (10 points) score

4

Rice, wheat, corn, and other2 A 464 (or 60% of total 0 g/d 0-464 g/d ≥464 g/d
energy)
All vegetables 3 A 300 0 g/d 0-300 g/d ≥300 g/d
All fruit4 A 200 0 g/d 0-200 g/d ≥200 g/d
Potatoes and cassava O 50 0 g/d 0-50 g/d 50-100 g/d 100-150 g/d
Whole milk or derivative O 250 0 g/d 0-250 g/d 250 – 500 g/d 500-750 g/d
equivalents (e.g., cheese)
Dry beans, lentils, and peas A 50 0 g/d 0-50 g/d ≥50 g/d
Soy foods A 25 0 g/d 0-25 g/d ≥25 g/d
Beef, lamb and pork M 14 ≥14 g/d 14 -0 g/d 0 g/d
Chicken and other poultry O 29 0 g/d 0-29 g/d 29-58 g/d 58-88 g/d
Eggs O 13 0 g/d 0-13 g/d 13-25 g/d 25-38 g/d
Fish O 28 0 g/d 0-28 g/d 28-100 g/d 100-128 g/d
Nuts O 50 0 g/d 0-50 g/d 50-100 g/d 100-150 g/d
All sweeteners M 31 ≥31 g/d 31-0 g/d 0 g/d

Supplementary Table 1. Construction of HRD-score, intake recommendations (g/day) and scoring system for men based on 2500 kcal/day (continued)

Food group Component HRD recommendation Minimum score Proportional Maximum points Proportional
type1 (g/day) (0 points) score (10 points) score
Added fats R
Palm oil 6.8 No consumption No consumption of
of unsaturated saturated fats OR
unsaturated oils 40 fats OR ratio of ratio of unsaturated to
unsaturated to saturated fats of ≥135
Dairy fats included in milk 0 saturated fats of
Lard and tallow 5 ≤0·65
1
A = adequacy component; O = optimum component; M = moderation component; R = ratio component.
2
Reference diet refers to dry, raw weight. Recommendations for whole grains were converted, as described by Dooren et al.1
3
Including dark green vegetables, red and orange vegetables, other vegetables.
4
Excluding fruit juice.
5
Cut-offs and threshold values were derived from the 15th percentile and 85th percentile of the intake distribution of the Dutch reference population,
as described in Looman et al. 2

References
1. van Dooren C, Mensink F, Eversteijn K, et al. Development and Evaluation of the Eetmaatje Measuring Cup for Rice and Pasta as an Intervention to
Reduce Food Waste. Front Nutr 2019;6:197.
2. Looman M, Feskens EJ, de Rijk M, et al. Development and evaluation of the Dutch Healthy Diet index 2015. Public Health Nutr 2017;20:2289–99
81
the healthy reference diet in rel ation to health outcomes and environmental impac t

4
 Supplementary Table 2. Construction of HRD-score, intake recommendations (g/day) and scoring system for women based on 2000 kcal/day

82
Food group Component HRD recommendation Minimum score Proportional Maximum points Proportional
chapter

type1 (g/day) (0 points) score (10 points) score

4

Rice, wheat, corn, and other2 A 372 (or 60% of total 0 g/d 0-372 g/d ≥372 g/d
energy)
All vegetables 3 A 240 0 g/d 0-240 g/d ≥240 g/d
All fruit4 A 160 0 g/d 0-160 g/d ≥160 g/d
Potatoes and cassava O 40 0 g/d 0-40 g/d 40-80 g/d 80-120 g/d
Whole milk or derivative O 200 0 g/d 0-200 g/d 200 – 400 g/d 400-600 g/d
equivalents (e.g., cheese)
Dry beans, lentils, and peas A 40 0 g/d 0-40 g/d ≥40 g/d
Soy foods A 20 0 g/d 0-20 g/d ≥20 g/d
Beef, lamb and pork M 12 ≥12 g/d 12-0 g/d 0 g/d
Chicken and other poultry O 23 0 g/d 0-23 g/d 23-46 g/d 46-69 g/d
Eggs O 10 0 g/d 0-10 g/d 10-20 g/d 20-30 g/d
Fish O 22 0 g/d 0-22 g/d 22-80 g/d 80-102 g/d
Nuts O 40 0 g/d 0-40 g/d 40-80 g/d 80-120 g/d
All sweeteners M 25 ≥25 g/d 25-0 g/d 0 g/d

Supplementary Table 2. Construction of HRD-score, intake recommendations (g/day) and scoring system for women based on 2000 kcal/day (continued)

Food group Component HRD recommendation Minimum score Proportional Maximum points Proportional
type1 (g/day) (0 points) score (10 points) score
Added fats R
Palm oil 5 No consumption of No consumption of
unsaturated fats OR unsaturated fats OR
unsaturated oils 32 ratio of unsaturated ratio of unsaturated
to saturated fats of to saturated fats of
dairy fats included in milk 0 ≤0·55 ≤0·55
lard and tallow 4
1
A = adequacy component; O = optimum component; M = moderation component; R = ratio component.
2
Reference diet refers to dry, raw weight. Recommendations for whole grains were converted, as described by Dooren et al.1
3
Including dark green vegetables, red and orange vegetables, other vegetables.
4
Excluding fruit juice.
5
Cut-offs and threshold values were derived from the 15th percentile and 85th percentile of the intake distribution of the Dutch reference population,
as described in Looman et al. 2

References
1. van Dooren C, Mensink F, Eversteijn K, et al. Development and Evaluation of the Eetmaatje Measuring Cup for Rice and Pasta as an Intervention to
Reduce Food Waste. Front Nutr 2019;6:197.
2. Looman M, Feskens EJ, de Rijk M, et al. Development and evaluation of the Dutch Healthy Diet index 2015. Public Health Nutr 2017;20:2289–99
83
the healthy reference diet in rel ation to health outcomes and environmental impac t

4
 Supplementary Table 3. Hazard ratios (HR) and 95% confidence intervals (CI) for the association between quartiles of the HRDea-score and CHD, with

84
adjustment for mediating factors (n=37,360).
chapter

HRDea-score (range)
4

Q1 (32-66) Q2 (67-73) Q3 (74-79) Q4 (80-116) P-trend

Cases, n 700 684 606 553
HR (95% CI)
Model 2 + BMI 1.00 [ref] 0.98 (0.88 – 1.10) 0.88 (0.79 – 0.99) 0.88 (0.78 – 1.00) 0.019
Model 2 + cholesterol 1.00 [ref] 0.96 (0.86 – 1.07) 0.85 (0.76 – 0.96) 0.85 (0.75 – 0.96) 0.003
Model 2 + hypertension 1.00 [ref] 0.96 (0.87 -1.07) 0.86 (0.77 – 0.97) 0.87 (0.77 – 0.98) 0.006
Model 2 + all possible mediators 1.00 [ref] 0.98 (0.87 – 1.09) 0.87 (0.78 – 0.98) 0.88 (0.77 – 0.99) 0.014

85
the healthy reference diet in rel ation to health outcomes and environmental impac t

4
5
ABSTRACT

Background
There is little and inconsistent evidence regarding the association between adherence
to the EAT-Lancet Healthy Reference Diet (HRD) and risk of cardiovascular disease.
Therefore, we aimed to study the association between adherence to the HRD with risk
of CHD and stroke.

Methods
We used data from 401,187 men and women in 8 countries participating in the European
Investigation into Cancer and Nutrition (EPIC) study aged between 25-70 years at
baseline. Dietary intake data was assessed with food frequency questionnaires and
used to create a score reflecting adherence to the HRD. Participants were assigned
a proportional score ranging between 0-10 points for each dietary recommendation
in the HRD, totaling to a score between 0 (no adherence) and 140 points (complete
adherence). Incident CHD and stroke were defined as any first fatal or non-fatal event.
Cox Proportional Hazard analysis was used to study the association between adherence
to the HRD-score and CHD incidence and stroke, both categorically and continuously.
Models were adjusted for age, sex, educational level, smoking, alcohol consumption,
and physical activity, energy intake, and stratified by EPIC study center.

Results
The HRD-score ranged from 10 to 120 and the mean was 63 (SD=14) and. After a median
follow-up time of 12.5 years, 12,753 cases of CHD and 7,126 cases of stroke occurred. High
adherence to the HRD was associated with a lower risk of incident CHD (HRQ5vsQ1: 0.94,
95%CI = 0.88 – 1.00; HR10-point increment
: 0.98, 95%CI = 0.97 – 1.00) and stroke (HRQ5vsQ1: 0.76,

The association between adherence to the 95%CI = 0.70 – 0.84; HR10-point increment: 0.93, 95%CI = 0.91 – 0.95) in fully adjusted models.

EAT-Lancet Healthy Reference Diet and Conclusion

coronary heart disease and stroke: the Adherence to the EAT-Lancet HRD was associated with lower risk of CHD and stroke.

EPIC-CVD cohort study

Marjolein C. Harbers, Joline W.J. Beulens, Femke Rutters, Anne M. May, Timothy J. Key,
Inge Huybrechts, Paolo Vineis, Jessica Laine, Keren Papier, Anika Knuppel, Elisabete
Weiderpass, Krasimira Aleksandrova, Christina C. Dahm, Yvonne T. van der Schouw

Manuscript in preparation for submission

chapter 5 the healthy reference diet in rel ation to c ardiovascul ar disease
INTRODUCTION
Our current food production processes put high pressure on environmental resources,
with agriculture taking up more than 43% of habitable land and causing about two thirds
of freshwater withdrawals for irrigation purposes (1). Moreover, agriculture causes 26%
of global greenhouse gas emissions, 32% of global terrestrial acidification, and 78% of
eutrophication, resulting in profound adverse effects on biodiversity and resilience of
ecosystems (2). At the same time, the United Nations estimates that the world population
will continue to grow from 7.7 billion people in 2019 to 9.7 billion people in 2050 (3).
Providing the expanding world population with access to adequate and healthy diets is
therefore becoming an increasing global concern.
In response to this challenge, the EAT-Lancet Commission on Healthy Diets From
Sustainable Food Systems proposed the Healthy Reference Diet (HRD) (4). The HRD
is a universal diet emphasizing intake of plant-based foods and suggesting a limited
intake of animal-sourced foods and starchy vegetables. The EAT-Lancet Commission has
estimated that global adoption of the HRD would prevent 10.8 to 11.6 million deaths per
year, equaling 19-24% of total deaths among adults. At the same time, adherence to the
HRD would aid in keeping within the earth’s food production boundaries with regard to
environmental resources, although the Commission estimates that other measures (e.g.,
reducing food waste, improving production practices) are needed as well (4).
The health and environmental effects have been modelled by the EAT-Lancet Commission,
yet empirical evidence from population-based cohort studies is still scarce. Both coronary
heart disease (CHD) and stroke are important drivers of mortality, accounting for 5.3
million and 5.5 million annual deaths, respectively, and have high attributable risks for
dietary factors (5, 6). Only two previous studies explored the association of the HRD with
risk of CHD and stroke, showing inconsistent findings (7, 8). Moreover, these studies used
a diet-score which could range between 0 and 14. Given the importance of variation in
the exposure measure for establishing diet-disease associations, development of a more
elaborate diet-score measuring adherence to the HRD is warranted (9). Therefore, we
aimed to study the association between adherence to an elaborate HRD-score with risk
of CHD and stroke in a pan-European study with heterogeneous diets (10).
88
METHODS
Study population
We used data from the European Investigation into Cancer and Nutrition (EPIC) study:
an ongoing prospective cohort study of 521,323 men and women aged between 25-70
years at baseline between 1992 and 2000 from 23 centers across 10 European countries.
For the present analysis we used data from Italy, Spain, the United Kingdom (UK), the
Netherlands, Germany, Denmark, Sweden and Norway, and excluded participants
from Greece due to an unresolved data protection regulation issue (n=28,561) and
participants from France due inconsistencies in the outcome definition with the other
EPIC-centers (n=74,523). We further excluded participants with missing diet assessment
data (n=6,310), and those with implausible energy intakes defined as being in the top
and bottom 1% of the distribution of the ratio of energy intake over estimated energy
requirement (n=8,196). Additionally excluding participants with prevalent CHD or stroke
at baseline (n=2,546) resulted in an analytical sample of 401,187 participants.
Most centers recruited participants from the general population, with a few exceptions.
First, participants from the Utrecht (Netherlands) and Florence (Italy) centers were 5
recruited through a population-based breast cancer screening program. Second,
participants from some of the Spanish and Italian centers were recruited from local
blood donor associations. In Oxford (UK), half of the cohort was recruited among (lacto-
ovo) vegetarian and vegan subjects, thereby representing a generally ‘health-conscious’
cohort. The cohorts in Norway, Utrecht (Netherlands), and Naples (Italy) only included
women. Detailed information on the rationale and design of EPIC has been described
previously (10, 11). The EPIC study was approved by the Ethical Review Boards of the
International Agency for Research on Cancer (IARC) and the Institutional Review Board
of each participating EPIC center.
Diet assessment
At baseline, habitual dietary intake over the past 12 months was assessed using
country-specific validated dietary questionnaires. Most centers used a validated (semi-)
quantitative food frequency questionnaire (FFQ), although a combination of dietary
assessment methods was used in Malmo (Sweden) and the UK. Nutrient and food intakes
were derived through the standardized EPIC Nutrient Database (12).
Diet intake assessments were used to calculate adherence to the HRD. Participants
were assigned a proportional score ranging between 0-10 points for each dietary
recommendation in the HRD, totaling to a score between 0 (no adherence) and 140
points (complete adherence). The methodology of the scoring was informed by the
89
chapter 5 the healthy reference diet in rel ation to c ardiovascul ar disease
Dutch Healthy Diet 2015 index distinguishing adequacy, moderation, optimum and
ratio components (Supplemental File 1) (13). The dietary recommendations for the
food groups in the HRD are based on intakes of 2500 kcal per day for both men and
women. As energy requirements for women are lower, we re-calculated the food group
recommendations to 2000 kcal/day for women (except for the ratio-component and fiber
intake, since we deemed these to be energy-independent).
The scoring approach is shown in Supplemental File 2. Adequacy components were
used to score food groups generally considered healthy and for which a high intake is
generally recommended, including 1) rice, wheat, corn and other grains 2) vegetables,
3) fruit, 4) dry beans, lentils and peas, 5) soy foods and 6) peanuts and 7) tree nuts. For
these food groups, participants were assigned 10 points if they had an intake equaling
or larger than the HRD reference level, 0 points for no consumption, and a proportional
score for intermediate intakes. For the recommendation on whole grains (including rice,
wheat, corn and other grains) two sub-components were created. As information on type
of cereal (e.g., wholegrain) was not available, fiber intake was used as an indicator of
wholegrain consumption. This sub-component was scored as an adequacy component,
with participants being assigned with 5 points if they had an intake equaling or larger
than 30g of fiber per day, 0 points for no consumption, and a proportional score for
intermediate intakes. The second subcomponent reflected the recommendation to limit
consumption of (dry) grains to 232g/day, equaling to 464g in converted wet weight,
for which participants were assigned 5 points or 0 points, for meeting or not meeting
the recommendation, respectively. The two separate recommendations on tree nuts
and peanuts were merged since they are consumed in a similar way and distinguishing
between the two from a nutritional point of view was not deemed necessary.
Consequently, the HRD reference level for the peanuts and tree nuts was 50g.
Moderation components were used to score foods generally considered to increase risk
of non-communicable diseases, including 1) beef and lamb, 2) pork and 3) all sweeteners.
For these foods, participants were assigned 10 points for no consumption, 0 points if
they had an intake equaling or larger than the HRD reference level, and a proportional
score for intermediate intakes. The two separate recommendations on beef, and lamb
and pork were merged together for reasons outlined above. Consequently, the HRD
reference level for beef, lamb and pork was 14g.
Optimum components were used to score foods which are nutritious yet potentially
detrimental if eaten in large quantities, including 1) potatoes and cassava, 2) whole
milk or derivative equivalents, 3) chicken and other poultry, 4) eggs, and 5) fish. For
these foods, the optimum range was defined as the range between the HRD reference
90
level and the upper range value provided in the HRD recommendations. Participants
were assigned with 10 points if their intake fell within the optimum range, and with a
proportionally increasing or decreasing score for intakes left and right of the optimum,
respectively.
Finally, a ratio component was used to score the added fats food group. For the added
fats, no consumption of unsaturated fats or an unsaturated to saturated fats ratio lower
than 0.6 was assigned 0 points, while no consumption of saturated fats or an unsaturated
to saturated fats ratio higher than 13 was assigned 10 points. Ratios in between were
scored proportionally.
In the original scoring approach, no or low consumption of selected animal-sourced
foods (whole milk or derivative equivalents, chicken and other poultry, eggs, and fish)
were scored as optimum components and assigned zero points or an proportionally
increasing score, respectively. In order to accommodate vegetarian and vegan diets,
we also constructed an alternative score in which no or low intakes of animal-sourced
foods were assigned 10 points (HRDvv-score).
5
Outcome ascertainment
Incident CHD was defined as any first fatal or non-fatal CHD event, which was a composite
myocardial infarction (International Classification of Diseases, 10th Revision (ICD-10)
codes I21, I22), angina (ICD-10 code I20) and other types of acute or chronic ischemic
heart diseases (ICD-10 codes I23, I24, I25). Incident stroke was defined as any first fatal or
non-fatal stroke event, which was a composite of hemorrhagic stroke (I60-I61), ischemic
stroke (I63), unclassified stroke (I64), and other acute cerebrovascular events (I62, I65-
69, F01). Fatal outcome events were generally ascertained through linkage with death
registries. Non-fatal outcome events were ascertained through a variety of methods
across centers, including follow-up questionnaires or linkage with morbidity/hospital
registries.
Assessment of other covariates
Data on socio-demographic (age, sex), lifestyle and other factors were collected at
baseline through validated questionnaires (10). Educational level was categorized into
primary school, technical/professional school, secondary school, and longer education
(including university degree). Alcohol intake was a continuous variable, measured in
grams per day. Physical activity was assessed through the Cambridge Physical Activity
index, which captures occupational physical activity and other physical exercise (e.g.,
cycling, walking), and was used to categorize participants as inactive, moderately inactive,
91
chapter 5 the healthy reference diet in rel ation to c ardiovascul ar disease
moderately active, and active. Smoking status was categorized into never, former, and
current smoking.
Body mass index (BMI) was calculated from measured height and weight, although for
participants from Norway and for some participants from France and the UK these
data were self-reported. Waist circumference was measured either at the narrowest
circumference of the torso or at the midpoint between the lower ribs and the iliac crest.
Systolic and diastolic blood pressures were measured in millimeters of mercury by
trained personnel (data not available for France, Norway, and the Asturias and Navarra
centers in Spain).
Statistical analysis
Baseline characteristics were presented across quintiles of HRD-adherence. Normally
distributed continuous variables were presented as means with standard deviations.
Categorical variables were presented as counts and percentages. For some of the
categorical covariates there was missing data (educational level n=15,317; physical
activity n=9,340; for smoking status n=5,044). These missings were imputed using the
SAS Procedure PROC MI, using conditionally specified imputation models (FCS), using
20 imputation sets and 5 burn-it iterations. The SAS Procedure MIANALYZE was used to
combine results from the Cox regression analyses across the imputed datasets using
Rubin’s rule.
Cox Proportional Hazard analysis was used to study the association between quintiles of
adherence to the HRD-score and CHD incidence and stroke, with the lowest quartile as
reference. Additionally, associations were examined linearly with HRD-scores modelled
per 10-point increment. Confounder adjustments were determined a priori and were
informed by literature. In model 1 we adjusted for age, sex, educational level, smoking,
alcohol consumption, and physical activity. In model 2 we additionally adjusted for
energy intake. All models were stratified by EPIC study center. To examine whether
associations were consistent across various subgroups, we conducted separate analyses
for levels of age, sex, educational level and BMI. The proportional hazards assumption
was tested using the Schoenfeld test, indicating no violation of the assumption.
We carried out a number of sensitivity analyses. First, we additionally adjusted for
cardiovascular risk-factors, including BMI, waist circumference, and blood pressure.
Second, we explored the association between adherence to the HRDvv-score with risk
of CHD and stroke. Third, we excluded cases occurring in the first 2 years of follow-
up, as subclinical disease may have affected eating habits and have induced reverse
causation. Fourth, we adjusted for food groups that were not included in the HRD
92
recommendations, but are possibly associated with risk of CHD and stroke, including
consumption of coffee and tea, snacks, sauces and condiments, and vegetarian products.
Last, we excluded food group recommendations individually in order to explore the food
groups driving the association.
RESULTS
Baseline characteristics
The mean HRD-score was 63 (SD=14) and ranged from 10 to 120. Participants generally
scored most points on the recommendations for grains (rice, wheat and corn), fruit,
and added fats, and the least points on the recommendations for meat, added sugar,
legumes, soy and nuts. Participants in Spain, UK, and Italy had the highest average
HRD-scores and in Sweden, the Netherlands, and Germany the lowest (Supplemental
File 3). Participants were on average 51 years old (SD=10), 33.5% was male, 31.3% had
a low educational level, and 19.7% was physically inactive. On average, participants
had a BMI of 25.6 kg/m2 (SD=4.2) and an energy intake of 2057 (SD=623) kcal per day.
Participants in the highest HRD-adherence quintile were more likely to be female and 5
alcohol abstainers, less likely to be current smokers, and have higher energy intake
(Table 1).
Association of HRD-adherence with incident CHD
After a median follow-up time of 12.5 years 12,753 cases of CHD occurred. High
adherence to the HRD was associated with a borderline significantly lower risk of
incident CHD after adjustment for demographic and lifestyle characteristics (HRQ5vsQ1:
0.94, 95%CI = 0.88 – 1.00; HR10-point increment: 0.98, 95%CI = 0.97 – 1.00) (Table 2). Further
adjustment for cardiovascular risk factors slightly strengthened the association (HRQ5vsQ1:
0.92, 95%CI = 0.85 – 0.99; HR10-point increment: 0.98, 95%CI = 0.96 – 0.99) (Supplemental File
4). The association of HRD-adherence with incident CHD was more pronounced in
the younger age strata (HR<55years,Q5vsQ1: 0.88, 95%CI = 0.78 – 0.99; HR<55-64years,Q5vsQ1: 0.90,
95%CI = 0.81 – 1.00; HR>64,Q5vsQ1: 1.14, 95%CI = 1.00 – 1.31) but comparable in strata of sex,
educational level, and BMI (Supplemental File 5). When exploring the association of the
HRDvv-score with CHD risk, the association strengthened (HRQ5vsQ1: 0.87, 95%CI = 0.82 –
0.93; HR10-point increment: 0.96, 95%CI = 0.94 – 0.97) which was mainly driven by the adjusted
scoring of eggs, fish, and chicken (Supplemental File 6). Neither exclusion of cases in the
first two years of follow-up (data not shown) nor adjusting the models for food groups
not included in the HRD (Supplemental File 7) altered our findings. No individual HRD
recommendation seemed to be driving the association (Supplemental File 8).
93
chapter 5 the healthy reference diet in rel ation to c ardiovascul ar disease
Association of HRD-adherence with incident stroke
After a median follow-up time of 12.5 years 7,126 cases of stroke occurred. High
adherence to the HRD was significantly associated with a lower risk of incident stroke
after adjustment for demographic and lifestyle characteristics (HRQ5vsQ1: 0.76, 95%CI = 0.70
– 0.84; HR10-point increment: 0.93, 95%CI = 0.91 – 0.95) (Table 2). Further adjustment for
cardiovascular risk factors resulted in a slight attenuation of the association (HRQ5vsQ1:
0.77, 95%CI = 0.69 – 0.86; HR10-point increment: 0.93, 95%CI = 0.91 – 0.95) (Supplemental File
4). The association of HRD-adherence with incident stroke was not materially different
in strata of age, sex, educational level, and BMI (Supplemental File 9). When exploring
the association of the HRDvv-score with stroke risk , HRs attenuated (HRQ5vsQ1: 0.82,
95%CI = 0.75 – 0.90; HR10-point increment: 0.94, 95%CI = 0.92 – 0.97), which was mainly due to
the altered scoring of the dairy component (Supplemental File 6). Exclusion of cases
in the first two years of follow-up did not alter findings (data not shown). Adjusting the
models for food groups not included in the HRD (Supplemental File 7) did not alter
findings. No individual HRD recommendation seemed to be driving the associations
(Supplemental File 8).
DISCUSSION
In the present study, we aimed to explore the association between adherence to the
HRD with risk of CHD and stroke in a large pan-European prospective cohort study.
We observed that high adherence to the HRD was associated with a 6% lower risk of
CHD and a 24% lower risk of stroke compared with low adherence. Additionally, we
observed differences in adherence between countries, with Spain, the UK, and Italy
being most adherent to the HRD, and Sweden, the Netherlands, and Germany being the
least adherent. With regard to individual recommendations, scores on meat (e.g., lamb,
beef, and pork), added sugar, soy, nuts and legumes were particularly low. Generally, the
maximum HRD-score was 120, indicating that in observed diets across Europe there is
still room for improvement with regard to reaching HRD recommendations.
The present study should be interpreted in the light of its strengths and limitations.
First, dietary data were based on self-report and were only assessed at baseline, which
may have caused misclassification of exposure status. Second, due to the multicenter
design, non-fatal CHD and stroke events were ascertained through a variety of methods
(e.g., linkage with disease registries, follow-up questionnaires), which may have led to
misclassification of outcome status. Third, although we adjusted for a wide range of
demographic, lifestyle and cardiometabolic risk-factors, we cannot rule out residual
confounding bias. Strengths of the current study include the large pan-European
population with a substantial number of cases, the prospective design, the variability
94
of diets captured, and the availability of validated FFQ data. Furthermore, our scoring
approach complies with various preferable features of a priori dietary indices, including
the use of various index dimensions (e.g., adequacy, moderation, optimum, and ratio
components), the use of metric measures opposed to ordinal or dichotomous metrics,
and the use of normative cut-off values (9) .
The present study is the first to report on an elaborate diet-score measuring adherence
to the HRD in relation to cardiovascular outcomes. Previously, a simpler diet-score (0-14
points) measuring adherence to the HRD was not associated with stroke in a Danish
population. However, when the threshold for the fiber recommendation was increased
similar to that of the present study, a 25% lower risk of stroke was observed (8). In
contrast, the same diet-score measuring HRD-adherence was not associated with risk
of stroke in the EPIC-Oxford population, although it did associate with a 28% lower
risk of CHD (7). It must be noted that EPIC-Oxford has a relatively high proportion of
vegetarian/vegan participants which may have resulted in a distinct nutritional status
of study participants, potentially limiting comparability of study findings. Still, using the
HRDvv-score, associations with CHD became more inverse and the association with
stroke attenuated, which is in line with the results of EPIC-Oxford. Apparently, the choices 5
for operationalizing the recommendations with regard to animal-sourced protein impact
the associations found with disease risk, which should be explored further in future
research.
In our main analysis, we observed that the association of HRD-adherence with stroke
was stronger as compared to the association with CHD. This may be explained by the fact
that the HRD may have a differential impact on the underlying risk-factors for CHD and
stroke. Risk profiles for stroke and CHD are comparable, but do show some differences
in magnitude with low-density lipoprotein cholesterol being of particular importance
for CHD, and hypertension being of greater influence for stroke (14, 15). However, taking
into account the fact that associations with CHD and stroke became more similar when
analyzing the HRDvv-score, it remains uncertain whether the apparent difference is
risk estimates is due to operationalization of the score or represents a meaningful
pathophysiological difference.
Overall, our findings are further substantiated by previous studies investigating
individual foods included in the HRD with risk of CHD and stroke. For example, high
consumption of fruit, vegetables, fish, dairy (e.g., cheese, fermented dairy) and limited
consumption of red and processed meat and sugar-sweetened beverages have all been
related to both lower risk of stroke and CHD (16-21). Intakes of whole grains, nuts, and
legumes have been more consistently associated with CHD rather than stroke (16, 17,
95
chapter 5 the healthy reference diet in rel ation to c ardiovascul ar disease

19), while intakes of poultry (18-20) and eggs (16-19) show fairly neutral associations

58,549 (73.0)

43,291 (54.0)
80,237 (20.0)

26,661 (33.2)

28,210 (35.6)

35,814 (45.2)
14,442 (18.0)
21,688 (27.0)

17,973 (22.4)
21,187 (26.4)

14,031 (17.5)
13,660 (17.0)
21,738 (27.1)

11,161 (14.1)

2101 ± 589
49.1 ± 10.9
with cardiovascular risk.

(76-120)

11 ± 15
Q5
From a public health perspective, our stratified analyses showed that adoption of the
Healthy Reference Diet is not likely to be of particular benefit for subgroups based on
sex, educational level, or BMI. However, we observed that risk of CHD was particularly

20,334 (25.3)
80,238 (20.0)

26,596 (33.2)

23,558 (29.7)
16,248 (20.3)
16,352 (20.4)

39,670 (50.0)
38,629 (48.1)
24,103 (30.0)
23,179 (28.9)

15,724 (19.6)

12,784 (16.1)
57,059 (71.1)

9,711 (12.1)
2077 ± 610
51.4 ± 10.2
lower among younger age groups, which may suggest that early adoption of a HRD-

(67-76)

12 ± 16
Q4
compliant pattern is of importance for CHD risk. Further, the HRD represents a global diet
that should be translated to national food-based dietary guidelines. A recent modelling
study showed that adherence to HRD recommendations resulted in greater reductions
in mortality as compared to adherence to current food-based dietary guidelines, mainly

24,435 (30.5)

25,826 (32.2)
80,237 (20.0)

22,799 (28.7)
25,006 (31.2)

35,814 (44.6)

13,004 (16.4)
18,916 (23.6)

16,831 (21.0)
21,641 (27.0)
14,099 (17.6)
55,231(68.8)

41,416 (52.1)

8,661 (10.8)
2057 ± 621
driven by more ambitious recommendations for whole grains, nuts and seeds, legumes,

52.0 ± 9.7
(59-67)

12 ± 17
Q3
vegetables, and processed meat (22). At the same time, none of the 85 countries included
in the analysis adhered to all food-based dietary guidelines, highlighting an important
concern about the feasibility of transitioning to HRD compliant dietary patterns. Last,
we did not consider the nutritional adequacy of the HRD, which is a prerequisite for

28,084 (35.0)
80,238 (20.0)

23,472 (29.5)
23,678 (29.5)
33,330 (41.5)

13,270 (16.5)
52,154 (65.0)

13,081 (16.4)
41,765 (52.4)
22,109 (27.6)
25,170 (31.4)
22,145 (27.6)

17,328 (21.6)

8,647 (10.8)
widespread dissemination of the diet. Two previous studies aimed to translate the

2036 ± 630
51.9 ± 9.6
5

(51-59)

12 ± 18
generic HRD recommendations to country-specific contexts and found country-specific

Q2
HRD diets generally to be nutritionally adequate, except for vitamin D, iodine and calcium
(23, 24). Additionally, we did not consider the affordability of the diet, about which

Table 1. Baseline characteristics across quintiles of HRD-score (N=401,187)1

concerns have been raised for low-income countries (25).

80,237 (20.0)

26,628 (33.2)

39,358 (49.3)
43,918 (54.7)
36,319 (45.3)

21,425 (26.7)

26,149 (32.8)
31,217 (38.9)

21,184 (26.4)
16,747 (20.9)
24,928 (31.1)
14,331 (17.9)

13,629 (17.1)

9,632 (12.0)
50.8 ± 10.1

2015 ± 661
(10-51)

12 ± 20
To conclude, the present study showed that adherence to the EAT-Lancet HRD was

Q1
associated with lower risk of CHD and stroke. Future research should further explore
the role of animal-sourced foods, operationalization of the diet, nutritional adequacy
and affordability of the diet.

Alcohol consumption in current drinkers (g/day)

Not current alcohol drinker, n (%)
Number of participants, n (%)

Vocational or university
Educational level, n (%)2
Physical activity, n (%)2
2
Age at recruitment (y)

Energy intake (kcal/d)

Moderately inactive
Smoking status, n (%)

Moderately active

None or primary
Secondary
Sex, n (%)

Inactive
Current
Former
Female

Active
Never
96 Male 97
98
Table 1. Baseline characteristics across quintiles of HRD-score (N=401,187)1 (continued)
chapter

Q1 Q2 Q3 Q4 Q5
5

(10-51) (51-59) (59-67) (67-76) (76-120)

Body mass index, kg/m2 25.5 ± 4.1 25.5 ± 4.1 25.5 ± 4.1 25.6 ± 4.2 25.9 ± 4.4
Waist circumference, cm 87 ± 13 86 ± 13 85 ± 13 84 ± 12 84 ± 13
Systolic blood pressure, mm/Hg 132 ± 20 133 ± 20 133 ± 20 132 ± 20 131 ± 20
Diastolic blood pressure, mm/Hg 82 ± 11 82 ± 11 82 ± 11 82 ± 11 81 ± 11
Coffee and tea, g/day 625 (386, 929) 625 (350, 986) 625 (300, 1000) 558 (190, 986) 352 (105, 950)
Sauces and condiments, g/day 10 (5, 21) 13 (6, 25) 15 (7, 28) 17 (7, 32) 17 (7, 37)
Snacks, g/day 0 (0, 0) 0 (0, 0) 0 (0, 0) 0 (0, 0.1) 0 (0, 1.1)
Vegetarian products, g/day 0 (0, 0) 0 (0, 0) 0 (0, 0) 0 (0, 0.1) 0 (0, 0.33)
HRD-score 44 ± 6 55 ± 2 63 ± 2 71 ± 3 84 ± 6
1 2
Continuous variables are displayed as means ± SDs or as medians (P25, P75). Data was missing on smoking status (n=5,044), physical activity (n=9,340),
educational level (n=15,317), waist circumference (n=61,571), and systolic and diastolic blood pressure (n=126,523).

Table 2. Hazard ratio’s and 95% confidence intervals for quintiles of HRD adherence with incident CHD and stroke.

Q1 Q2 Q3 Q4 Q5 Per 10-point increment

CHD
Incident cases, n 2851 2643 2586 2414 2259
Person-years 931,572 929,140 933,908 948,251 990,456
Crude model 1.00 0.85 (0.80 - 0.89) 0.76 (0.72 - 0.80) 0.66 (0.62 - 0.70) 0.61 (0.57 - 0.65) 0.88 (0.86 - 0.89)
a
Model 1 1.00 0.97 (0.92 - 1.03) 0.96 (0.91 - 1.02) 0.91 (0.86 - 0.97) 0.94 (0.88 - 1.00) 0.98 (0.97 - 1.00)
b
Model 2 1.00 0.97 (0.92 - 1.03) 0.96 (0.91 - 1.02) 0.91 (0.86 - 0.97) 0.94 (0.88 - 1.00) 0.98 (0.97 - 1.00)
Stroke
Incident cases, n 1828 1619 1429 1274 976
Person-years 969,789 967,399 968,471 975,610 1,022,044
Crude model 1.00 0.86 (0.81 - 0.92) 0.75 (0.69 - 0.80) 0.68 (0.63 - 0.73) 0.55 (0.50 - 0.61) 0.86 (0.84 - 0.88)
a
Model 1 1.00 0.94 (0.88 - 1.00) 0.87 (0.81 - 0.93) 0.85 (0.79 - 0.92) 0.76 (0.70 - 0.84) 0.93 (0.91 - 0.95)
Model 2b 1.00 0.94 (0.88 - 1.00) 0.87 (0.81 - 0.93) 0.85 (0.79 - 0.92) 0.76 (0.70 - 0.84) 0.93 (0.91 - 0.95)
a b
Model 1 was adjusted for age, sex, educational level, smoking, alcohol consumption, and physical activity; Model 2 was additionally adjusted for
energy intake. Models were stratified by EPIC center.
99
the healthy reference diet in rel ation to c ardiovascul ar disease

5
chapter 5 the healthy reference diet in rel ation to c ardiovascul ar disease

17. Iacoviello L, Bonaccio M, Cairella G, Catani MV, Costanzo S, D’Elia L, et al. Diet and primary
REFERENCES prevention of stroke: Systematic review and dietary recommendations by the ad hoc
Working Group of the Italian Society of Human Nutrition. Nutr Metab Cardiovasc Dis.
1. Poore J, Nemecek T. Reducing food’s environmental impacts through producers and 2018;28(4):309-34.
consumers. Science. 2018;360(6392):987-92. 18. Key TJ, Appleby PN, Bradbury KE, Sweeting M, Wood A, Johansson I, et al. Consumption
2. Crist E, Mora C, Engelman R. The interaction of human population, food production, and of Meat, Fish, Dairy Products, and Eggs and Risk of Ischemic Heart Disease. Circulation.
biodiversity protection. Science. 2017;356(6335):260. 2019;139(25):2835-45.
3. World Population Prospects 2019: Highlights. United Nations, Department of Economic 19. Tong TYN, Appleby PN, Key TJ, Dahm CC, Overvad K, Olsen A, et al. The associations of
and Social Affairs PD; 2019. major foods and fibre with risks of ischaemic and haemorrhagic stroke: a prospective study
4. Willett W, Rockström J, Loken B, Springmann M, Lang T, Vermeulen S, et al. Food in the of 418 329 participants in the EPIC cohort across nine European countries. European Heart
Anthropocene: the EAT-Lancet Commission on healthy diets from sustainable food Journal. 2020;41(28):2632-40.
systems. Lancet. 2019;393(10170):447-92. 20. Mozaffarian D. Dietary and Policy Priorities for Cardiovascular Disease, Diabetes, and
5. Global, regional, and national burden of stroke, 1990-2016: a systematic analysis for the Obesity: A Comprehensive Review. Circulation. 2016;133(2):187-225.
Global Burden of Disease Study 2016. Lancet Neurol. 2019;18(5):439-58. 21. Alexander DD, Bylsma LC, Vargas AJ, Cohen SS, Doucette A, Mohamed M, et al. Dairy
6. Dai H, Much AA, Maor E, Asher E, Younis A, Xu Y, et al. Global, regional, and national burden consumption and CVD: a systematic review and meta-analysis. British Journal of Nutrition.
of ischaemic heart disease and its attributable risk factors, 1990–2017: results from the 2016;115(4):737-50.
Global Burden of Disease Study 2017. European Heart Journal - Quality of Care and Clinical 22. Springmann M, Spajic L, Clark MA, Poore J, Herforth A, Webb P, et al. The healthiness and
Outcomes. 2020. sustainability of national and global food based dietary guidelines: modelling study. Bmj.
7. Knuppel A, Papier K, Key TJ, Travis RC. EAT-Lancet score and major health outcomes: the 2020;370:m2322.
EPIC-Oxford study. Lancet. 2019;394(10194):213-4. 23. Lassen AD, Christensen LM, Trolle E. Development of a Danish Adapted Healthy Plant-
Based Diet Based on the EAT-Lancet Reference Diet. Nutrients. 2020;12(3).
8. Ibsen D, Christiansen A, Olsen A, Tjønneland A, Overvad K, Wolk A, et al. Adherence to the
Eat-Lancet Diet and Risk of Stroke and Stroke Subtypes – A Danish Cohort Study. Current 24. Tucci M, Martini D, Del Bo C, Marino M, Battezzati A, Bertoli S, et al. An Italian-Mediterranean 5
Developments in Nutrition. 2021;5(Supplement_2):414-. Dietary Pattern Developed Based on the EAT-Lancet Reference Diet (EAT-IT): A Nutritional
9. Burggraf C, Teuber R, Brosig S, Meier T. Review of a priori dietary quality indices in relation Evaluation. Foods. 2021;10(3).
to their construction criteria. Nutr Rev. 2018;76(10):747-64. 25. Hirvonen K, Bai Y, Headey D, Masters WA. Affordability of the EAT-Lancet reference diet:
10. Riboli E, Hunt KJ, Slimani N, Ferrari P, Norat T, Fahey M, et al. European Prospective a global analysis. Lancet Glob Health. 2020;8(1):e59-e66.
Investigation into Cancer and Nutrition (EPIC): study populations and data collection.
Public Health Nutr. 2002;5(6b):1113-24.
11. Riboli E, Kaaks R. The EPIC Project: rationale and study design. European Prospective
Investigation into Cancer and Nutrition. Int J Epidemiol. 1997;26 Suppl 1:S6-14.
12. Slimani N, Deharveng G, Unwin I, Southgate DA, Vignat J, Skeie G, et al. The EPIC nutrient
database project (ENDB): a first attempt to standardize nutrient databases across the 10
European countries participating in the EPIC study. Eur J Clin Nutr. 2007;61(9):1037-56.
13. Looman M, Feskens EJ, de Rijk M, Meijboom S, Biesbroek S, Temme EH, et al. Development
and evaluation of the Dutch Healthy Diet index 2015. Public Health Nutr. 2017;20(13):2289-99.
14. Leening MJG, Cook NR, Franco OH, Manson JE, Lakshminarayan K, LaMonte MJ, et al.
Comparison of Cardiovascular Risk Factors for Coronary Heart Disease and Stroke Type
in Women. J Am Heart Assoc. 2018;7(19):e007514.
15. Stoekenbroek RM, Boekholdt SM, Luben R, Hovingh GK, Zwinderman AH, Wareham NJ,
et al. Heterogeneous impact of classic atherosclerotic risk factors on different arterial
territories: the EPIC-Norfolk prospective population study. Eur Heart J. 2016;37(11):880-9.
16. Bechthold A, Boeing H, Schwedhelm C, Hoffmann G, Knüppel S, Iqbal K, et al. Food groups
and risk of coronary heart disease, stroke and heart failure: A systematic review and dose-
response meta-analysis of prospective studies. Crit Rev Food Sci Nutr. 2019;59(7):1071-90.

100 101
102
10
Points

10
Points
chapter
5

Adequacy component

Optimum component
Intake (g/day)

Intake (g/day)
10
10

Points Points
B.

Ratio component
component
Moderation
CHAPTER 5 – SUPPLEMENTARY FILES

Intake (g/day)

Ratio

Supplemental File 1. Graphical illustration of scoring components applied to HRD-score

Supplemental File 2. Scoring approach for the Healthy Reference Diet1

Scoring Points (0-10)

component
0 5 10 5 0
2
1. Rice, wheat, corn etc.
a. Limit consumption ~ > 464 g/d ≤ 464 g/d

b. Fiber intake A 0 g/d ≥ 30 g/d

2. Potatoes & cassava O 0 g/d 50 ≤ g/day ≤ 100 ≥ 150 g/d

3. All vegetables A 0 g/d ≥ 300 g/d

4. All fruit A 0 g/d ≥ 200 g/d

5. Dairy O 0 g/d 250 ≤ g/day ≤ 500 ≥ 750 g/d

6. Beef, lamb, pork 3 M ≥ 28 g/d < 14 g/d

7. Chicken, poultry O 0 g/d 29 ≤ g/day ≤ 58 ≥ 87 g/d

8. Eggs O 0 g/d 13 ≤ g/day ≤ 25 ≥ 38 g/d

9. Fish O 0 g/d 28 ≤ g/day ≤ 100 ≥ 128 g/d

10. Beans, lentils, peas A 0 g/d ≥ 50 g/d

11. Soy foods A 0 g/d ≥ 25 g/d

12. Peanuts/tree nuts 3 A 0 g/d ≥ 50 g/d

13. Added fats 3 R 13 0.6

14. Added sugar M ≥ 31 g/d < 0 g/d

1
Values represent intakes for men (based on 2500 kcal/d); intake values for women were re-calculated to 2000 kcal/d (e.g., conversion factor of 0.8).
2
The EAT-Lancet Commission proposed 232g of dry rice, wheat, and corn so HRD reference level was converted to wet weight with a conversion factor
of 2.
103
the healthy reference diet in rel ation to c ardiovascul ar disease

3
HRD reference levels collapsed for a) beef and lamb and b) pork, and a) tree nuts and b) peanuts.
4
Expressed as a ratio of unsaturated to saturated fat.
5
 Supplemental File 3. Sub-scores on food components included in HRD-score across countries

104
Italy Spain UK NL Germany Sweden Denmark Norway Full cohort
chapter
5

Food group Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD
Rice, wheat, corn etc. 7.2 2.0 8.7 1.2 8.4 1.5 8.6 1.1 8.5 1.0 8.0 1.3 8.9 1.1 8.4 0.9 8.3 1.4
Potatoes & cassava 5.3 3.1 6.7 3.8 5.9 4.2 5.5 4.3 6.6 3.9 3.8 4.3 3.7 4.3 3.7 4.5 5.2 4.2
All vegetables 6.3 2.7 7.6 2.7 8.3 2.2 5.1 2.0 4.7 2.1 4.5 2.9 6.2 2.8 5.6 2.6 6.2 2.9
All fruit 9.4 1.7 8.5 2.9 8.3 2.7 7.7 2.9 6.5 2.9 7.2 3.1 6.9 3.3 7.1 3.2 7.7 3.0
Dairy 6.9 3.3 7.3 3.4 6.4 3.9 5.9 4.0 6.8 3.2 6.5 3.8 5.9 3.9 6.9 3.0 6.5 3.7
Beef, lamb, pork 0.4 1.8 0.7 2.3 4.2 4.8 0.5 2.1 0.5 2.0 0.5 2.0 0.1 1.0 0.5 2.1 1.2 3.1
Chicken, poultry 6.8 3.2 6.6 3.7 4.2 3.8 4.1 3.1 4.1 3.0 3.3 3.3 6.0 3.1 4.9 3.3 4.9 3.6
Eggs 6.9 3.4 4.9 4.2 5.3 3.2 6.4 3.6 6.4 3.6 3.8 3.5 5.7 4.1 7.0 3.5 5.7 3.7
Fish 7.6 3.0 7.9 3.5 6.4 4.3 4.0 3.0 6.2 3.5 6.9 3.4 8.9 2.4 5.7 4.6 6.8 3.8
Beans, lentils, peas 2.4 2.9 7.8 3.0 4.4 3.4 2.2 2.1 1.0 1.3 0.8 1.8 0.2 0.5 0.0 0.0 2.4 3.3
Soy foods 0.0 0.0 0.0 0.5 2.6 3.8 0.9 2.0 0.1 0.2 0.0 0.4 0.0 0.0 0.0 0.0 0.6 2.1
Peanuts/tree nuts 0.2 0.5 0.9 2.0 1.3 2.0 1.9 2.3 0.8 1.3 0.2 0.7 0.4 0.9 0.5 0.9 0.8 1.6
Added fats 7.6 3.1 9.7 1.6 6.3 4.4 2.4 2.1 3.7 4.2 8.8 3.0 7.9 3.9 9.0 2.8 6.9 4.1
Added sugar 0.2 1.2 1.2 2.7 0.0 0.4 0.0 0.5 0.1 0.6 0.1 0.8 0.1 0.9 0.1 1.0 0.2 1.2
Total HRD-score 67 12 78 11 72 12 55 11 56 11 54 12 61 11 59 11 63 14

Supplemental File 4. Additional adjustment for cardiovascular risk-factors

Q1 Q2 Q3 Q4 Q5 Per 10-point increment

CHD
Incident cases, n 2851 2643 2586 2414 2259
Person-years 931,572 929,140 933,908 948,251 990,456
Crude model 1.00 0.85 (0.80 - 0.89) 0.76 (0.72 - 0.80) 0.66 (0.62 - 0.70) 0.61 (0.57 - 0.65) 0.88 (0.86 - 0.89)
a
Model 1 1.00 0.97 (0.92 - 1.03) 0.96 (0.91 - 1.02) 0.91 (0.86 - 0.97) 0.94 (0.88 - 1.00) 0.98 (0.97 - 1.00)
b
Model 2 1.00 0.97 (0.92 - 1.03) 0.96 (0.91 - 1.02) 0.91 (0.86 - 0.97) 0.94 (0.88 - 1.00) 0.98 (0.97 - 1.00)
Model 3c 1.00 0.98 (0.92 – 1.04) 0.96 (0.91 – 1.02) 0.92 (0.86 – 0.98) 0.92 (0.85 – 0.99) 0.98 (0.96 – 0.99)
Stroke
Incident cases, n 1828 1619 1429 1274 976
Person-years 969,789 967,399 968,471 975,610 1,022,044
Crude model 1.00 0.86 (0.81 - 0.92) 0.75 (0.69 - 0.80) 0.68 (0.63 - 0.73) 0.55 (0.50 - 0.61) 0.86 (0.84 - 0.88)
Model 1a 1.00 0.94 (0.88 - 1.00) 0.87 (0.81 - 0.93) 0.85 (0.79 - 0.92) 0.76 (0.70 - 0.84) 0.93 (0.91 - 0.95)
b
Model 2 1.00 0.94 (0.88 - 1.00) 0.87 (0.81 - 0.93) 0.85 (0.79 - 0.92) 0.76 (0.70 - 0.84) 0.93 (0.91 - 0.95)
c
Model 3 1.00 0.93 (0.86 – 1.00) 0.90 (0.83 – 0.97) 0.85 (0.78 – 0.93) 0.77 (0.69 – 0.86) 0.93 (0.91 – 0.95)
a
Model 1 was adjusted for age, sex, educational level, smoking, alcohol consumption, and physical activity; b Model 2 was additionally adjusted for energy
intake; c Model 3 was additionally adjusted for BMI, waist circumference, systolic and diastolic blood pressure. Models were stratified by EPIC center.
105
the healthy reference diet in rel ation to c ardiovascul ar disease

5
 Supplemental File 5. Hazard ratio’s and 95% confidence intervals for HRD adherence with incident CHD across levels of age, sex, educational level

106
and BMI1
chapter

Cases/N Q1 Q2 Q3 Q4 Q5
5

Age group
<55 4,124/256,592 1.00 0.93 (0.84 – 1.02) 0.97 (0.88 – 1.07) 0.91 (0.82 – 1.02) 0.88 (0.78 – 0.99)
55-64 5,737/119,878 1.00 1.02 (0.95 – 1.11) 0.95 (0.87 – 1.03) 0.88 (0.80 – 0.96) 0.90 (0.81 – 1.00)
>64 2,892/24,717 1.00 1.03 (0.90 – 1.17) 1.06 (0.94 – 1.21) 1.03 (0.91 – 1.17) 1.14 (1.00 – 1.31)
Sex
Male 7,527/134,276 1.00 0.98 (0.91 – 1.05) 0.97 (0.90 – 1.04) 0.91 (0.84 – 0.99) 0.97 (0.89 – 1.06)
Female 5,226/266,911 1.00 1.00 (0.91 – 1.09) 0.99 (0.91 – 1.09) 0.96 (0.87 – 1.05) 0.96 (0.87 – 1.07)
Educational level
Low 5,287/124,188 1.00 1.02 (0.93 – 1.10) 1.03 (0.94 – 1.12) 0.96 (0.88 – 1.05) 1.00 (0.90 – 1.12)
Middle 1,344/63,659 1.00 1.11 (0.94 – 1.31) 1.05 (0.88 – 1.26) 1.17 (0.97 – 1.40) 1.11 (0.90 – 1.36)
High 5,089/198,023 1.00 0.90 (0.83 – 0.97) 0.88 (0.81 – 0.96) 0.81 (0.74 – 0.88) 0.84 (0.76 – 0.93)
BMI
<25 4,190/197,543 1.00 1.01 (0.92 – 1.11) 0.95 (0.86 – 1.04) 0.92 (0.83 – 1.02) 0.90 (0.80 – 1.01)
25-30 6,038/148,686 1.00 0.98 (0.90 – 1.06) 0.97 (0.90 – 1.06) 0.89 (0.81 – 0.97) 0.99 (0.90 – 1.09)
>30 2,525/54,958 1.00 0.88 (0.78 – 1.00) 0.94 (0.83 – 1.07) 0.95 (0.83 – 1.08) 0.89 (0.76 – 1.03)
1
Models were adjusted for age, sex, educational level, smoking, alcohol consumption, and physical activity and energy intake, and stratified by EPIC
center.
levels of age, sex, BMI, and educational level.

107
the healthy reference diet in rel ation to c ardiovascul ar disease

Supplemental File 5 (continued). Association of quintiles of HRD-scores with CHD risk across

5
 Supplemental File 6. Hazard ratio’s and 95% confidence intervals for quintiles of HRD adherence with incident CHD and stroke using HRDvv-score1

108
Q1 Q2 Q3 Q4 Q5 Per 10-point increment
chapter
5

CHD
HRD-score 1.00 (ref) 0.97 (0.92 - 1.03) 0.96 (0.91 - 1.02) 0.91 (0.86 - 0.97) 0.94 (0.88 - 1.00) 0.98 (0.97 - 1.00)
HRDvv-score2 1.00 (ref) 0.96 (0.91 – 1.01) 0.95 (0.90 – 1.00) 0.95 (0.90 – 1.00) 0.87 (0.82 – 0.93) 0.96 (0.94 – 0.97)
Stroke
HRD-score 1.00 (ref) 0.94 (0.88 - 1.00) 0.87 (0.81 - 0.93) 0.85 (0.79 - 0.92) 0.76 (0.70 - 0.84) 0.93 (0.91 - 0.95)
HRDvv-score2 1.00 (ref) 0.94 (0.88 – 1.01) 0.89 (0.83 – 0.96) 0.85 (0.79 – 0.92) 0.82 (0.75 – 0.90) 0.94 (0.92 – 0.97)
1
Models were adjusted for age, sex, educational level, smoking, alcohol consumption, and physical activity and energy intake, and stratified by EPIC
center. 2No or low intakes of animal-sourced foods (e.g., those initially scored as optimum components, including whole milk or derivative equivalents,
chicken and other poultry, eggs, and fish) were assigned with 10 points.

Supplemental File 7. Additional confounder adjustment1

Q1 Q2 Q3 Q4 Q5 Per 10-point increment

CHD
Model 21 1.00 0.97 (0.92 - 1.03)
Model 2 + coffee and tea 1.00 0.97 (0.92 – 1.03)
Model 2 + sauces and condiments 1.00 0.97 (0.92 – 1.03)
Model 2 + snacks 1.00 0.97 (0.92 – 1.03)
Model 2 + vegetarian products 1.00 0.97 (0.92 – 1.03)
Stroke
Model 21 1.00 0.94 (0.88 - 1.00)
Model 2 + coffee and tea 1.00 0.94 (0.88 – 1.00)
Model 2 + sauces and condiments 1.00 0.94 (0.88 – 1.00)
Model 2 + snacks 1.00 0.94 (0.88 – 1.00)
Model 2 + vegetarian products 1.00 0.94 (0.88 – 1.00)
1
Models were adjusted for age, sex, educational level, smoking, alcohol consumption, and physical activity, and energy intake, and stratified by EPIC
center.
109
0.96 (0.91 - 1.02) 0.91 (0.86 - 0.97) 0.94 (0.88 - 1.00) 0.98 (0.97 - 1.00)
0.96 (0.91 – 1.02) 0.91 (0.86 – 0.97) 0.94 (0.88 – 1.00) 0.98 (0.97 – 1.00)
0.96 (0.91 – 1.02) 0.92 (0.86 – 0.97) 0.94 (0.88 – 1.01) 0.98 (0.97 – 1.00)
0.96 (0.91 – 1.02) 0.91 (0.86 – 0.97) 0.94 (0.88 – 1.01) 0.98 (0.97 – 1.00)
0.96 (0.91 – 1.02) 0.92 (0.86 – 0.97) 0.95 (0.89 – 1.02) 0.98 (0.97 – 1.00)
0.87 (0.81 - 0.93) 0.85 (0.79 - 0.92) 0.76 (0.70 - 0.84) 0.93 (0.91 - 0.95)
0.87 (0.81 – 0.93) 0.85 (0.79 – 0.92) 0.76 (0.70 – 0.84) 0.93 (0.91 – 0.95)
0.87 (0.81 – 0.93) 0.85 (0.79 – 0.92) 0.77 (0.70 – 0.84) 0.93 (0.91 – 0.95)
0.87 (0.81 – 0.93) 0.85 (0.79 – 0.92) 0.77 (0.70 – 0.84) 0.93 (0.91 – 0.95)
0.87 (0.81 – 0.93) 0.85 (0.79 – 0.92) 0.76 (0.69 – 0.84) 0.93 (0.91 – 0.95)
the healthy reference diet in rel ation to c ardiovascul ar disease

5
 1

110
None

9. Fish
8. Eggs
4. Fruit
chapter
5

11. Soy foods

3. Vegetables

13. Added fats

14. Added sugars
6. Beef, lamb and pork
2. Potatoes and cassava

12. Peanuts and tree nuts

7. Chicken and other poultry
1. Rice, wheat, corn and other
recommendations individually1

10. Dry beans, lentils and peas

Excluded HRD recommendation

5. Whole milk or derivative equivalents

activity, and energy intake and stratified by EPIC center.

HR (95% CI)
Risk of CHD

0.98 (0.97 - 1.00)

0.99 (0.97 – 1.01)

0.99 (0.98 – 1.01)

0.97 (0.95 – 0.98)

0.97 (0.95 – 0.98)
0.98 (0.97 – 1.00)
0.98 (0.97 – 1.00)
0.98 (0.97 – 1.00)

0.99 (0.97 – 1.00)

0.99 (0.97 – 1.00)
0.99 (0.97 – 1.00)

0.97 (0.96 – 0.99)

0.98 (0.96 – 0.99)

0.98 (0.96 – 0.99)
0.98 (0.96 – 0.99)
HR (95% CI)
Risk of stroke

0.93 (0.91 - 0.95)

0.93 (0.91 – 0.95)

0.93 (0.91 – 0.95)
0.93 (0.91 – 0.95)
0.93 (0.91 – 0.95)
0.93 (0.91 – 0.95)
0.93 (0.91 – 0.95)
0.93 (0.91 – 0.95)
0.93 (0.91 – 0.95)
0.93 (0.91 – 0.95)
0.93 (0.91 – 0.95)
0.93 (0.91 – 0.95)

0.92 (0.91 – 0.94)

0.92 (0.90 – 0.94)
0.94 (0.91 – 0.96)
per 10-point increment per 10-point increment
Supplemental File 8. Hazard ratio’s and 95% confidence intervals for the association between

Models were adjusted for age, sex, educational level, smoking, alcohol consumption, physical
HRD-scores (continuous) and incident CHD and stroke excluding one out of the fourteen

Supplemental File 9. Hazard ratio’s and 95% confidence intervals for HRD adherence with incident stroke across levels of age, sex, educational level
and BMI1

Cases/N Q1 Q2 Q3 Q4 Q5
Age group
<55 2,056 /256,592 1.00 1.00 (0.88 – 1.14) 0.87 (0.75 – 1.00) 0.81 (0.70 – 0.95) 0.81 (0.68 – 0.97)
55-64 3,595 /119,878 1.00 0.88 (0.80 – 0.97) 0.82 (0.74 – 0.90) 0.83 (0.75 – 0.93) 0.72 (0.63 – 0.83)
>64 1,475/24,717 1.00 1.11 (0.94 – 1.31) 1.08 (0.92 – 1.28) 1.00 (0.84 – 1.19) 0.85 (0.69 – 1.04)
Sex
Male 3,512/134,276 1.00 0.88 (0.80 – 0.97) 0.82 (0.74 – 0.91) 0.78 (0.69 – 0.87) 0.77 (0.67 – 0.88)
Female 3,614 /266,911 1.00 1.02 (0.92 – 1.13) 0.95 (0.85 – 1.05) 0.95 (0.85 – 1.06) 0.81 (0.71 – 0.92)
Educational level
Low 3,123/124,188 1.00 0.97 (0.88 – 1.08) 0.93 (0.83 – 1.03) 0.85 (0.76 – 0.97) 0.78 (0.68 – 0.91)
Middle 723/63,659 1.00 0.99 (0.80 – 1.23) 0.94 (0.75 – 1.18) 0.93 (0.73 – 1.18) 0.87 (0.65 – 1.16)
High 2,921/198,023 1.00 0.88 (0.79 – 0.98) 0.79 (0.71 – 0.88) 0.81 (0.72 – 0.91) 0.71 (0.61 – 0.82)
BMI
<25 2,717/197,543 1.00 0.92 (0.83 – 1.03) 0.90 (0.80 – 1.01) 0.86 (0.76 – 0.98) 0.82 (0.71 – 0.96)
25-30 3,105/148,686 1.00 0.97 (0.88 – 1.08) 0.81 (0.72 – 0.90) 0.87 (0.77 – 0.97) 0.77 (0.66 -0.88)
>30 1,304/54,958 1.00 0.88 (0.74 – 1.04) 0.95 (0.80 – 1.13) 0.78 (0.64 – 0.94) 0.67 (0.54 – 0.83)
1
Models were adjusted for age, sex, educational level, smoking, alcohol consumption, and physical activity and energy intake, and stratified by EPIC
center.
111
the healthy reference diet in rel ation to c ardiovascul ar disease

5
chapter 5 the healthy reference diet in rel ation to c ardiovascul ar disease

Supplemental File 9 (continued). Association of quintiles of HRD-scores with stroke risk across
levels of age, sex, BMI, and educational level.

112 113
part II Healthy and sustainable diets from
a societal perspective
6
ABSTRACT
Background
Adults with a low socioeconomic position (SEP) are more likely to engage in unhealthy
diets as compared to adults with high SEP. However, individual-level educational
interventions aiming to improve food choices have shown limited effectiveness in adults
with low SEP. Environmental-level interventions such as nudging strategies however,
may be more likely to benefit low SEP groups. We aimed to review the evidence for the
effectiveness of nudges as classified according to interventions in proximal physical
micro-environments typology (TIPPME) to promote healthy purchases, food choice,
or affecting energy intake or content of purchases, within real-life food purchasing
environments. Second, we aimed to investigate the potentially moderating role of SEP.

Methods
We systematically searched PubMed, EMBASE, and PsycINFO until 31 January 2018.
Studies were considered eligible for inclusion when they i) complied with TIPPME
intervention definitions; ii) studied actual purchases, food choice, or energy intake or
content of purchases, ii) and were situated in real-life food purchasing environments.
Risk of bias was assessed using a quality assessment tool and evidence was synthesized
using harvest plots.

Results
From the 9210 references identified, 75 studies were included. Studies were generally
of weak to moderate quality. The most frequently studied nudges were information
(56%), mixed (24%), and position nudges (13%). Harvest plots showed modest tendencies
towards beneficial effects on outcomes for information and position nudges. Less

The effects of nudges on purchases, food evidence was available for other TIPPME nudging interventions for which the harvest
plots did not show compelling patterns. Only six studies evaluated the effects of nudges
choice, and energy intake or content of across levels of SEP (e.g., educational level, food security status, job type). Although

purchases in real-life food purchasing there were some indications that nudges were more effective in low SEP groups, the

environments: a systematic review and

evidence synthesis
This chapter was published as:
Harbers MC, Beulens JWJ, Rutters F, de Boer F, Gillebaart M, Sluijs I, et al. The effects of
nudges on purchases, food choice, and energy intake or content of purchases in real-life food
purchasing environments: a systematic review and evidence synthesis. Nutr J. 2020;19(1):103.
limited amount of evidence and different proxies of SEP used warrant caution in the
interpretation of findings.
Conclusions
Information and position nudges may contribute to improving population dietary
behaviours. Evidence investigating the moderating role of SEP was limited, although
some studies reported greater effects in low SEP subgroups. We conclude that more
high-quality studies obtaining detailed data on participant’s SEP are needed.
Registration
This systematic review is registered in the PROSPERO database (CRD42018086983).
chapter 6 effec ts of nudges in real - life purchasing environments
INTRODUCTION
An unhealthy diet is one of the major risk factors for non-communicable diseases (NCDs),
such as type 2 diabetes and cardiovascular disease (1). Adults with a low socioeconomic
position (SEP) in particular are at high risk for NCDs, as they are more likely to engage
in unhealthy diets as compared to adults with high SEP (2). Despite this, individual-level
educational interventions that aim to improve healthy food choices have shown to have
limited effectiveness in adults with low SEP and may increase health inequalities (3).
This may partly be attributed to the fact that these interventions often necessitate
access to various resources (e.g., knowledge, skills, social networks) which may be more
limited in low SEP groups (4, 5). Alternatively, environmental-level interventions are more
likely to benefit adults with low SEP and reduce health inequalities (3), because they
rely to a lesser extent on an individual’s access to resources but rather create healthy
opportunities for all.
The rationale underlying such environmental-level interventions is rooted in dual process
models of human behaviour, which conceptualize the regulation of human behaviour
into two main cognitive processes: 1) an unconscious, fast, and automatic cognitive
process, and 2) a conscious, slow, and more effortful cognitive process (6). Whereas
individual-level educational interventions tap into the conscious and effortful processes
– by for example providing nutrition knowledge to target populations – environmental
interventions make use of environmental cues or heuristics that subconsciously guide
food-decision making (7), thus requiring limited amounts of cognitive resources.
Nudging has been proposed as a promising environmental intervention strategy for
modifying food choices. The term ‘nudge’ was originally coined by Thaler and Sunstein
in 2008 and defined as: ‘Any aspect of the choice architecture that alters people’s
behaviour in a predictable way, without forbidding any options or significantly changing
their economic incentives’ (p.6) (8). Nudging became popular as it opposed the reigning
idea that humans are rational actors who constantly seek opportunities that maximize
their utility. Instead, it acknowledges that people’s ability to make rational decisions is
limited by cognitive boundaries, biases and habits, leading people to make choices not
compatible with their long-term goals (9). Nudges make use of the same principles that
cause flawed decision-making, to steer people towards choices that serve them in their
own interest. When applied to modifying diets, this means that nudges make healthy
choices more easy, by for example making them more salient, without constraining
choice for unhealthy alternatives (9).
118
So far, numerous nudging studies have been performed describing a wide range of
interventions, for example placing healthier foods at convenient and visible locations
in supermarkets (e.g., position nudge) or making healthy foods salient through the use
of signage (e.g., information nudge). To establish more conceptual clarity regarding
nudging interventions and to facilitate evidence synthesis, the typology of interventions
in proximal physical micro-environments (TIPPME) was introduced, distinguishing six
distinct nudging interventions types: availability, position, functionality, presentation,
size, and information (10).
The multiple systematic reviews and meta-analyses on the effectiveness of TIPPME
nudging interventions in modifying food choices or consumption (11-13) mainly focused
on availability and position nudges (9, 10) or specific foods (11), and studies were primarily
conducted in laboratory settings. Only one of these systematic review addressed the
question whether the effects of nudging interventions are moderated by SEP, for which
indications were found (13). Therefore, insights are lacking on the effectiveness of
other TIPPME intervention types in real-life food purchasing environments, and the
moderating role of SEP.
In the present systematic review, our first aim is to review the evidence for the
effectiveness of nudges as classified according to the TIPPME typology in promoting
healthy purchases, food choice, or affecting energy intake or content of purchases within 6
real-life food purchasing environments among adult populations. Second, we aimed to
investigate the potentially moderating role of SEP.
METHODS
The protocol for the present systematic review was registered in the PROSPERO database
(registration number: CRD42018086983). A systematic literature search was conducted
in accordance with the guidelines in the Reporting Items for Systematic Reviews and
Meta-Analysis (PRISMA) statement (www.prisma-statement.org) (Additional File 1).
Data sources and searches
In order to maximize the yield of our search, we adopted an elaborate search strategy
including general nudging terms (e.g. nudging and choice architecture) as well as more
specific nudging terms (e.g. signage) according the TIPPME typology (Table 1). Types of
nudges considered in other categorizations were evaluated on their applicability to the
current review (14, 15). As a result, the search strategy was further extended by adding
the default nudge, which we defined as follows: ‘to provide a standard food option for
which no active choice needs to be made’. For the search queries, search terms for the
119
chapter 6 effec ts of nudges in real - life purchasing environments
(type of) nudging intervention, outcome, and setting were combined using Boolean
operators and were limited to title and abstract. The search strategies for each of the
databases can be found in Additional file 2. We systematically searched the databases
PubMed, EMBASE, and PsycINFO until 31 January 2018. Additionally, references included
in existing reviews were included for screening (11, 12, 16).
Table 1. Overview of nudging interventions in TIPPME as defined by Hollands et al. (10)
Intervention type Definition
Availability To add or remove (some or all) products or objects to increase,
decrease, or alter their range, variety, or number
Position To alter the position, proximity, or accessibility of products or objects
Functionality To alter the functionality or design of products or objects to change
how they work, or guide or constrain how people use or physically
interact with them
Presentation To alter visual, tactile, auditory or olfactory properties of products,
objects or stimuli
Size To alter size or shape of products or objects
Information Add, remove, or change words, symbols, numbers or pictures that
convey information about the product or object or its use
Study selection
Titles, abstracts, and full-text articles retrieved from database searches were
screened for eligibility in duplicate by a team of five researchers (MH, FdB, IS, JWJB,
FR). Studies were included if they: 1) involved a manipulation of the food purchasing
environment, in such a way that the availability, position, functionality, presentation,
size, and/or information of products (e.g. foods), related objects (e.g. shelfs), or the
wider environment (e.g. supermarket) was altered; 2) examined the effects on actual
food purchases, energy intake or energy content of purchases, or food choice; 3) were
situated in a food purchasing environment where people purchase food or meals on a
regular basis; 4) were conducted among adult populations; 5) were originally published
articles and were written in English language.
Studies were excluded if they: 1) did not report the effects of the nudges separately from
other non-nudge interventions, such as pricing interventions; 2) studied the effects of
nudges on behavioural intent; 3) were performed in settings in which people do not
purchase food or meals on a regular basis (e.g. sit-down restaurants); 4) changed the
intrinsic characteristics of foods (e.g. dietary composition); 5) examined the effects of
mandatory legislation.
120
Inconsistencies in eligibility judgements were resolved by discussion among two
reviewers (MH and IS) and if consensus could not be reached, inconsistencies were
resolved by discussion with a third reviewer (JWJB, FR, or FdB). After this process was
completed, titles, abstracts, and full-text articles retrieved from the reference lists of
existing reviews were screened for eligibility by MH. A 10% subsample of the studies
retrieved from the reference lists was checked by a second reviewer (IS), which revealed
no inconsistencies in eligibility judgements.
Quality assessment
Risk of bias was assessed using the Quality Assessment Tool for Quantitative Studies (17),
as this tool was specifically designed to critically appraise public health interventions and
encompassed a wide range of research designs, including non-randomized designs. This
tool evaluates the risk of bias with regard to selection of study participants, study design,
confounding variables, blinding, data collection methods, and withdrawals and drop-
outs. Each domain can be attributed a weak, moderate or strong quality score. Articles
were considered of i) strong quality if no domains were rated as weak; ii) moderate
quality if only one domain was rated as weak; 3) weak if at least two domains were rated
as weak. Quality assessment was conducted in duplicate by a team of five researchers
(MH, FdB, IS, JWJB, FR). Inconsistencies were resolved by discussion with a third reviewer.
Data extraction 6
Data extraction was performed by one researcher (MH) using a predefined data
extraction form, and conducted in duplicate for a subsample of the included studies
(n=8), which showed high levels of agreement. Data was extracted on the type of nudge
(including nudge description), country, study design, intervention duration, SEP, setting,
study outcomes, outcome assessment, and main findings.
Data synthesis
For the tabulation of study characteristics and main findings, nudges were classified using
the TIPPME intervention typology (MH & FdB) into either one of the following intervention
types: availability, position, functionality, presentation, size or information. On the
basis of the quality assessment, study design was categorized into before-after studies
(both within- and between-subjects), controlled trials, or randomized controlled trials.
Intervention duration was defined as the duration for which the nudge was implemented
and categorized according to the following categories: ≤ 1 week; > 1 week & ≤ 1 month;
1 < month(s) ≤ 6; 6 < months ≤ 12 and > 1 year. Study size could pertain to amount
of purchases and/or transactions, number of customers, or number of stores. Study
outcomes could pertain to purchases, energy intake or energy content of purchases or
food choice. Outcome assessment was categorized as either one or a combination of
121
chapter 6 effec ts of nudges in real - life purchasing environments
the following: point-of-sale system, observer-reported, computer-generated response,
digital photographic method, food weighing, hand counts, questionnaires, dietary recall,
and records of inventory movement. Lastly, we report SEP characteristics for each study
based on descriptive characteristics for proxies of SEP reported in the baseline table or
in-text (e.g., educational level, job type).
Besides the tabulation of study characteristics and main findings, we visualized the
main findings and study characteristics of studies within each of the TIPPME categories
in harvest plots (18). The harvest plot groups studies according to their intervention
effect (positive/negative or no effect) in a matrix, and allows to further incorporate
relevant study information by varying characteristics of the matrix, including bar length,
width, and colour, and by adding rows to the matrix. As such, harvest plots provide
a qualitative summary to the reader by enabling them to visually appraise the most
prominent patterns in the matrix, and judge study characteristics and study quality.
For the present review, the matrix comprises three columns representing the
intervention effect (increase, no change, or decrease) and three rows comprising the
types of outcomes (purchases, energy intake or energy content of purchases or food
choice). Studies were plotted in the matrix based on the direction of the association that
was reported for each outcome (e.g. if a nudge is associated with higher purchases, this
study was plotted in the ‘increase’ column). Each study was plotted in the matrix using
bars, with a study reference number below the bar corresponding to the tabulation of the
study characteristics and main findings in Table 2. If studies assessed multiple outcomes,
studies appear in the matrix for each outcome denoted by an additional letter (e.g. 1a,
1b). The bars were further modified to represent several relevant study characteristics.
More specifically, high bars represent RCTs and controlled trials and low bars represent
before-after study designs; narrow bars indicate shorter study duration and increasing
width indicates longer study duration; red bars indicate unhealthy foods, blue bars
indicate healthy foods, and white bars indicate calorie intake or content of purchases.
Lastly, settings as retrieved from the data extraction were categorized into cafeterias
(denoted by letter C) and supermarkets and small food stores (denoted by letter S).
We were not able to visuzalize nine studies in harvest plots, due to outcomes that were
difficult to categorize on relative healthiness (e.g., targeted foods for which insufficient
information was available to determine this); the absence of formal statistical analysis
or the use of a factorial design. These studies can be found in Additional file 3.
122
RESULTS
From the 9210 references identified from the database searches and reference list
screening, 224 were eligible for full-text review, and 68 references were included in
the narrative synthesis of findings. The 68 references comprised 75 studies (Figure 1).
Descriptive characteristics of included studies
Of the 75 retrieved studies, 42 studies were categorized as studying information nudges,
ten studies were categorized as studying position nudges, 18 studies were categorized
as studying mixed nudging interventions, two studies were categorized as studying
size nudges, two studies were categorized as studying a functionality nudge, and one
study was categorized as studying a presentation nudge. No studies were categorized
as studying default or availability nudges. Given the vast amount of information nudges
identified, we further categorized these groups of interventions into the following
categories: information nudges using symbols (n=15); information nudges providing
nutrition information (n=13); and information nudges using signage (n=14). Studies most
often employed a before-after design (n=56), followed by a controlled trial design (n=32)
and randomized controlled trial design (n=12). Only 19% of studies had an intervention
duration longer than 6 months, and studies were most often situated in cafeterias (55%),
followed by supermarkets (25%) and small food stores (16%).
6
Effects of nudging by TIPPME category
Information nudges using symbols
The harvest plot for information nudges using symbols is shown in Figure 2 and study
characteristics and main findings are presented in Table 2. Eight studies received a
moderate quality rating, four received a weak quality rating, and three received a strong
quality rating. Studies examining information nudges via symbols generally highlighted
healthy or unhealthy foods using symbols such as star-ratings and promotional logos.
The effects of information nudges using symbols were most often studied in association
to purchasing outcomes. Overall, in mainly cafeteria settings, identifying healthy food
items through the use of symbols generally did not affect purchases of those items
(1a, 2c, 4, 5, 11, 13a, 13b, 14a, 14b, 15b, 15c, 15d, 15e), caloric content of purchases or
caloric intake (2d, 6b, 7b, 8), or healthier food choice (7a). Contrary, some other studies
conducted in supermarket and cafeteria settings showed increased purchases of healthy
foods and decreased purchases of unhealthy foods (1b, 2a, 2b, 3a, 6a, 10, 12, 13c, 15a)
and decreased energy intake or content of purchases (3b, 9). Concluding, the effects of
highlighting healthy and unhealthy foods through the use of symbols in supermarket,
123
chapter 6 effec ts of nudges in real - life purchasing environments
small food store, and cafeteria settings were heterogeneous but showed a modest
tendency towards no effects on studied outcomes.
Information nudges providing nutrition information
The harvest plot of information nudges providing nutrition information is shown in
Figure 3 and study characteristics and main findings are presented in Table 2. Three
studies could not be visualized in the harvest plots and are presented in Additional file
3. Seven studies received a moderate quality rating, five studies received a weak quality
rating, and one study received a strong quality rating. Studies examining information
nudges providing nutrition information usually did so by providing nutritional labels at
the point-of-choice. The effects of nutrition information nudges were most often studied
in relation to purchases as the outcome as well as energy intake or energy content of
purchases. Some studies provide evidence that the provision of nutrition information
in food purchasing environments increases purchases of or choice for healthy items
(1a, 7a, 7b, 8a, 10a), decreases purchases of unhealthy items (1b, 10b), and similarly,
decreases energy intake or energy content of purchases (1c, 2a, 2b, 3, 4), although one
study observed increased energy intake (5). Contrary, other studies found no effects
on purchases of healthy or unhealthy items (7c, 9a, 9b), or on energy intake or content
of purchases (6) or food choice (8b). Concluding, the effects of providing nutrition
information in supermarket, small food store and cafeteria settings were heterogeneous
but showed a modest tendency towards beneficial effects on studied outcomes.
Information nudges using signage
The harvest plot of information nudges using signage is shown in Figure 4 and study
characteristics and main findings are presented in Table 2. Two studies could not be
visualized in the harvest plots and are presented in Additional file 3. Eight studies
received a moderate quality rating, three studies received a weak quality rating, and
three studies received a strong quality rating. Studies examining information nudges
using signage generally displayed posters with health prompts, social norms, or health
primes. The effects of signage nudges were generally evaluated on purchasing outcomes
and studies were primarily conducted within cafeteria settings. Signage was associated
with increased purchases of healthy items in several studies (2b, 2c, 3, 5a, 6, 7a, 7b, 7c,
9a), increased choice for healthy food (4) and with decreased purchases of unhealthy
items (1a). Contrary, also no change in purchases of healthy or unhealthy (1b, 2a, 2d,
5b, 8a, 8b, 9b) items were observed. Concluding, effects for information nudges using
signage in supermarket, small food store, and cafeteria settings were heterogeneous
but showed a modest tendency towards beneficial effects on studied outcomes.
124
Position nudges
The harvest plot for position nudges is shown in Figure 5 and study characteristics and
main findings are presented in Table 2. Eight studies received a moderate quality rating
and two received a weak quality rating. Studies examining position nudges generally
manipulated proximity to healthy and unhealthy foods (e.g. decreasing proximity to
healthy foods and increasing proximity to unhealthy foods). The effects of position nudges
were most often studied in relation to purchasing outcomes. Overall, it can be concluded
that in small food stores and cafeterias, increasing or decreasing the accessibility or
visibility of healthy and unhealthy foods, respectively, showed increased purchases of
healthy foods and decreased choice for unhealthy foods (1a, 2a, 3, 5, 6, 9). However,
other studies conducted in larger purchasing contexts such as supermarkets showed no
effects on healthy food purchases (8, 10a). Moreover, purchases of relocated unhealthy
snacks (e.g. snacks that were relocated to more distant locations as a consequence of
making healthy foods more accessible) (1b, 10b), energy intake (2b), or food choice (4)
were not affected in both small and larger purchasing contexts. Lastly, one study showed
counterintuitive findings, with increased and decreased purchases of unhealthy and
healthy items, respectively, when healthy items had been made more accessible (7a, 7b).
Concluding, the effects of altering the proximity of healthy and unhealthy foods showed
a modest tendency towards beneficial effects on outcomes in primarily smaller food
purchasing environments, but not in larger food purchasing environments.
6
Mixed nudging interventions
Several studies were identified that studied a combination of TIPPME intervention
categories, which we phrased ‘mixed nudging interventions’. The harvest plot for mixed
nudging intervention is shown in Figure 6 and study characteristics and main findings
are presented in Table 2. Four studies could not be visualized in the harvest plots and
are presented in Additional file 3.
Eight studies received a moderate quality rating, eight studies received a weak quality
rating, and two studies received a strong quality rating. The effects of mixed intervention
nudges were most often studied in relation to purchasing outcomes in cafeteria or
supermarket settings. Moreover, studies were often characterized by high quality study
designs (e.g. RCTs and controlled trials). As for the effects of mixed nudging interventions
on the outcomes studied, mixed nudging interventions generally did not affect purchases
of healthy items (1a, 2b, 2d, 3c, 3d, 3e, 4a, 5b, 5e, 9, 11, 13a, 14) or unhealthy items
(1b, 2a, 2c, 4b, 13b), or energy intake or -content of purchases (4c, 6). Contrary, some
studies observed increased purchases of healthier items (3a, 3b, 5a, 5c, 5f, 7a, 10a,
12a), decreased purchases of unhealthy items (7b, 10b, 12b), and decreased calorie
content of purchases (8). Also some counterintuitive findings were observed, with mixed
125
chapter 6 effec ts of nudges in real - life purchasing environments
nudging interventions being associated with increased purchases of unhealthy items
(2e) and decreased purchases of healthy items (2f, 5d). Concluding, the effects mixed
nudging interventions in supermarket, small food store, and cafeteria settings were
heterogeneous but showed a modest tendency towards no changes in studied outcomes.
Availability, size, functionality, and presentation nudges
Two studies were categorized as size nudges (74, 75). In these studies, increasing the
portion size of an entrée (74) and decreasing the portion size of sausages (75), was
associated with increased energy intake and decreased meat purchases, respectively.
Two studies described the effects of a functionality nudge (76). In these studies, arrows
on supermarket floors indicating the location of fresh fruits and vegetables were
associated with increased fruit and vegetable purchasing. One study was categorized
as a presentation nudge, during which participants were provided with a healthy or
unhealthy sample and subsequent purchases in a supermarket were monitored (77). The
study showed that the consumption of a healthy sample was associated with increased
subsequent healthy purchases.
Evidence for differential effects across SEP
Six studies evaluated the effects of nudges across levels of SEP, for which several
indicators were used including educational level, food security, job type, and income.
In subgroup analyses, there were modest indications that nudges – including signage,
mixed nudging interventions, and position nudges – were significantly more effective
among people with a lower educational level (44), in people with food insecurity (63), or
in people on a food assistance program (59), respectively. Similarly, in two other mixed
nudging intervention studies which used traffic-light labelling and accessibility changes,
the extent to which red and green-labelled purchases were affected by the intervention
differed in magnitude across job type in subgroup analyses (71) and the effect of the
intervention on red-labelled purchases was significantly modified by job type, but not for
overall purchases (67). However, no evident pattern in purchasing differences across job
types could be discerned, as job types could not be clearly classified by SEP. In another
study which examined the effect of an information nudge providing nutrition information
on calorie intake, no significant effect modification by income or educational level was
observed (37).
126
DISCUSSION
Main findings
In the present review, we aimed to assess the evidence for the effectiveness of nudges as
classified according to TIPPME in promoting healthy purchases, food choice, or affecting
energy content of purchases or intake within real-life food purchasing environments.
Additionally, we aimed to investigate whether the effects of nudges are moderated by
SEP. We observed that the evidence to date predominately focused on the effectiveness
of information nudges (56%) and position nudges (13%), while less evidence is available
on the effectiveness of other types of TIPPME nudging interventions. We also observed
that studies often investigated short-term outcomes, with 81% of studies having an
intervention duration shorter than 6 months. Also, the studies often relied on non-
randomized designs and were most often conducted in cafeteria or supermarket
settings.
The harvest plots showed that for information and position nudges modest tendencies
towards beneficial effects on studied outcomes were present. Finally, we found
indications that the effects of nudges may be moderated by SEP, showing larger effects
among low SEP individuals. However, evidence was limited in quantity and the use of
different measures of SEP hampered comparison of the evidence. Overall, studies were
generally considered of moderate or weak quality, raising concerns about potential bias 6
and warranting caution in the interpretation of the results.
Findings from the present review are in line with previous literature. Similar to the
present study, a scoping review conducted by Hollands et al. concluded that most
studies focused on information nudges (78). The effectiveness of information nudges
is however debated, as they deviate from the original definition of nudging, by relying
partly on cognitive processing. One previous meta-analysis of field studies by Cadario
and Chandon explored the effectiveness of nudges, using their own categorization
of cognitive nudges, affective nudges and behavioural nudges. They concluded that
cognitive nudges were least effective in affecting selection and consumption outcomes
(79), observing a small effect size of d = 0.12, supporting the argument that information
nudges are ‘sub-optimal’. In the present review, we observed that information nudges
– largely overlapping with the definition of cognitive nudges by Cadario and Chandon
– positively affected outcomes, but we could not compare the magnitude of effects to
other TIPPME nudges given the inability to meta-analyse findings. Further evidence
that information nudges work, even though considered ‘sub-optimal’ in terms of how
they operate on a psychological level, comes from two recent systematic reviews and
meta-analyses of nutritional package and/or point-of-purchase labelling in primarily
127
chapter 6 effec ts of nudges in real - life purchasing environments
supermarkets, cafeterias, and restaurants, showing statistically significant average
decreases of 6.6% and 7.8% in energy intake, respectively (80, 81), although for the
latter review the quality of evidence was rated as low.
We also observed a tendency towards healthier purchasing in smaller food purchasing
contexts for position nudges. Although evidence is tentative and qualitative in nature,
this finding is in line with multiple systematic reviews that examined the effects of
position nudges on consumption and selection; choice, sales or servings; or on sales
and consumption in primarily laboratory settings (13), school settings (11), and a range
of micro-environments including cafeteria and laboratory settings (12), respectively.
However, all reviews highlight that effects are generally small in magnitude and that the
quality of evidence is considered to be low.
Finally, we observed that the effects of nudges may differ by SEP, with limited studies
observing somewhat stronger effects in low SEP populations. Only one other systematic
review and meta-analysis that examined the effectiveness of availability and proximity
nudges systematically assessed whether the effects of these interventions were
potentially modified by SEP, and found that effect sizes for position nudges were larger
among studies conducted among populations with low deprivation status, as compared
to studies conducted among populations from both high and low deprivation status (13).
For availability nudges, insufficient data was available to assess whether intervention
effects were modified by SEP. An important reason for why evidence is limited in the
present review, may be due to the fact that it is challenging to obtain detailed information
on SEP in studies conducted in real-life food purchasing environments, as there is often
less active engagement with the research population. For example, studies often monitor
purchases following a nudging intervention, without consent or active participation of
customers.
Strengths and limitations
Some limitations of the present review need to be addressed. First, given the substantial
heterogeneity in study characteristics and incomplete study reporting, it was not possible
to quantify the effects of the TIPPME intervention types using conventional meta-
analyses techniques. An important reason for the heterogeneous study characteristics
and study findings may relate to the focus on real-life purchasing contexts which are
naturally less controlled environments as compared to laboratory settings. Additionally
it may be due to our studied outcomes which were heterogeneous in terms of the types
of foods that were targeted with nudging strategies. However, the use of harvest plots
offers a visually appealing way to summarize the study information and study findings.
This approach is preferable over a narrative analysis of study findings, as information is
128
more easily digested by the reader and also less prone to bias, as studies are plotted in
a systematic way (82). Second, very few studies assessed dietary intake as outcome of
nudging interventions. Alternatively, energy content of purchases was often calculated as
a proxy of energy intake. Therefore, the majority of evidence is based on the evaluation
food purchases. As nudging is often suggested as a potentially important strategy in
battling the obesity epidemic, it is crucial to evaluate its effects on more proximal health
parameters, such as dietary intake, as well. Third, we adopted a broad search strategy,
including general nudging terms (e.g., nudging and choice architecture) as well as more
specific nudging terms (e.g., signage) according the TIPPME typology. As a result of this
search strategy, studies were included that did not clearly indicate to test a nudge, but
did comply with nudging definitions laid out by the TIPPME typology. As these studies
provided little theoretical background of the intervention under study, there was often
limited information available to categorize the study according to TIPPME. For example,
studies we categorized as information nudges based on the TIPPME definition, may partly
rely on cognitive processing, and therefore may not satisfy the criteria for nudging.
Finally, the majority of studies received a moderate to weak quality rating. Major quality
issues related to the study design, which was often not randomized, which consequently
raised concerns about potential for confounding. Concerns about the quality of nudging
studies have also been highlighted in previous reviews (11-13).
Strengths of the current review include that it used an extensive search strategy, not 6
only using ‘nudging’ and ‘choice architecture’ as search terms, but adding specific
nudging intervention types as search terms as well. Indeed, a previous systematic
review investigating the effectiveness of nudging strategies only included studies if
they were specified as such by the original authors, resulting in only thirteen eligible
publications (83). Additionally, the present review builds upon the TIPPME typology
which was the result of an extensive scoping review, and therefore provides a useful
conceptual framework for structuring the evidence base. However, we acknowledge that
categorizations remain broad and may be susceptible to different interpretations, and
further enhancement of conceptual clarity is needed.
Implications for improved methods
Given the limitations of the evidence base addressed in this review, we provide several
suggestions for improved methods. First, given the level of heterogeneity in study
characteristics there is an urgent need for harmonization of methods in nudging studies
to facilitate evidence accumulation. It is therefore important to establish common
measures to asses SEP, such as composite measures combining both income, education,
and job status (84). Additionally, adherence to reporting standards such as Journal Article
Reporting Standards (JARS) as laid out by the American Psychological Association would
129
chapter 6 effec ts of nudges in real - life purchasing environments

improve study reporting and therefore enhance evidence synthesis. Moreover, the field compare. More high-quality studies focusing on non-information nudges and examining
of psychological and behavioural science has been scrutinized for its inability to replicate long-term effectiveness in real-life food purchasing environments and obtaining detailed
some of its findings (85). For example, a recent pre-registered study found no association data on participant’s SEP are needed.
between plate size and food consumption, which contrasted with earlier findings (86).
Therefore, efforts such as pre-registration of study protocols which allow replication are
warranted to further advance the field of (nutrition) nudging (87).

Implications for future research and practice

From the present evidence, we highlight the following knowledge gaps present in
nudging literature. First, future studies should focus on studying the effectiveness of
non-information nudges (e.g. availability, position, functionality, or sizing nudges) in
real-world settings. Second, given the limited available data on potential moderators
of nudging effectiveness in real-world settings, the use of loyalty cards containing
customer’s personal information would be a valuable contribution to the existing
literature, allowing to examine the role of potential moderators such as age, sex, and SEP.
Third, nudging studies often only targeted limited food categories, which does not justify
complex food environments in which multiple other food choices are made. Moreover,
it is difficult to make inferences about what changes in purchases of a selected number
of foods actually constitutes in term of an individual’s health. Therefore, future nudging
studies that use loyalty cards, could nudge a wider array of food products and estimate
changes in overall dietary quality on an individual level. Fourth, as the included literature 6
in the present study mainly studied short-term effects, future studies should consider
including a longer follow-up, as this long-term effectiveness is crucial to assess potential
public health impact. Lastly, the present review highlights the viability of conducting
nudging interventions in real-life purchasing contexts. Consequently, local policy makers
or owners of local food stores could be encouraged to implement nudging interventions
at local level. From a policy perspective, it is also of importance to consider the ethical
aspects of nudging, which have been outlined previously (88).

CONCLUSION
This systematic review was the first to examine the effectiveness of nudging interventions
on purchases, energy intake or content of purchases, and food choice in real-life food
purchasing environments, using an elaborate search strategy drawing upon the TIPPME
framework. We showed that evidence mainly focuses information and position nudges,
while less evidence is available on the effectiveness of other TIPPME intervention
types. We qualitatively demonstrated that information and position nudges might be
effective in improving outcomes, especially purchasing outcomes, and that SEP may be
a moderator for the effectiveness of nudges. However, evidence is limited and difficult to

130 131
chapter 6 effec ts of nudges in real - life purchasing environments

Figure 2. Harvest plot for information nudges using symbols

Figure 1. PRISMA flowchart of study inclusion

132 133
134
6 chapter

Figure 3. Harvest plot for information nudges providing nutrition information

135
effec ts of nudges in real - life purchasing environments

Figure 4. Harvest plot for information nudges using signage

6
136
6 chapter

Figure 5. Harvest plot for position nudges

137
effec ts of nudges in real - life purchasing environments

Figure 6. Harvest plot for mixed nudging interventions

6
Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Information nudges (symbols)

Cawley et 2015 USA Supermarket Pre-post 168 > 1 year Supermarket N/A Purchases Point of sale 1a. Purchases Moderate
al. [19] items were supermarkets healthy items system of healthy items
assigned with (any stars); were not affected
stars indicating Purchases of 1b. Purchases of
their relative unhealthy items unhealthy items
healthiness (no stars) decreased*

Dubbert et 1984 USA Labels Pre-post 6,970 > 1 week & ≤ 1 Cafeteria N/A Purchases of Point of sale 2a. Increased Weak
al. [20] indicating low- customers month vegetables; system and vegetable
calorie choices Purchases of observer purchases*
were placed salad; reported 2b. Increased
besides serving Purchases of salad purchases*
location entrees; 2c. Entrée
Caloric content purchases not
of meal affected
purchased 2d. Caloric
content of meals
purchased not
affected

Elbel et al. 2013 USA Unhealthy items Pre-post 3,680 > 1 week & ≤ 1 Small food Store Purchases of Point of sale 3a. Probability Strong
[21] were assigned a purchases month store catered to healthy items; system of purchasing
tag stating ‘less low-income, Caloric content healthy items
healthy’ minority and of purchases increased*
immigrant 3b. Caloric
population content of items
purchased
decreased*

Eldridge et 1997 USA Menu boards Pre-post 7 cafeterias 6 < months ≤ 12 Cafeteria N/A Purchases of all Point of sale 4. Purchases of Moderate
al. [22] indicated targeted items system targeted items
healthy items were not affected
with a green
check-mark

Freedman 2011 USA Healthy Pre-post 1 small food 1 < month(s) ≤ 6 Small food N/A Purchases of all Point of sale 5. Purchases of Moderate
et al. [23] foods were store store targeted items system targeted items
identified with were not affected
a promotional
logo on shelf-
tags

Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Hobin et al. 2017 Canada Supermarket CT 44 6 < months ≤ 12 Supermarket Supermarkets Purchases of Point of sale 6a. Average Strong
[24] items were intervention were located healthy items; system mean star rating
assigned with supermarkets; in area where Caloric content per product
stars indicating 82 control 13.1% had no of purchases purchased
their relative supermarkets secondary increased, so
healthiness school healthy purchases
diploma increased*
6b. Caloric
content of
purchases was
not affected

Hoefkens 2011 Belgium Healthy meal Pre-post 224 > 1 week & ≤ 1 Cafeteria N/A Meal choice Questionnaire 7a. Meal choice Weak
et al. [25] suggestions customers month (0-3 stars) and was not affected;
were assigned energy intake 7b. Energy intake
with stars (0-3 was not affected
stars)

Johnson et 1990 USA Labels Pre-post 413 > 1 week & ≤ 1 Cafeteria N/A Caloric content Observer 8. Caloric content Weak
al. [26] indicating low- customers month of purchases reported of purchases was
calorie choices not affected
were placed
besides serving
location

Lassen et 2014 Norway Healthy choices CT 270 6 < months ≤ 12 Cafeteria 59% employed Energy density Digital 9. Energy density Strong
al. [27] were labelled customers as office and of consumed photographic decreased*
with the Keyhole administrative foods method
symbol personnel or
as technical
staff

Levin et al. 1996 USA Low-fat entrees CT 2 cafeterias 6 < months ≤ 12 Cafeteria N/A Purchases of Point of sale 10. Purchases of Moderate
[28] were labelled targeted items system targeted items
with a heart- increased*
shaped symbol

Sproul et 2003 USA Healthy Pre-post 1 cafeteria 1 < month(s) ≤ 6 Cafeteria N/A Purchases of Point of sale 11. Purchases of Moderate
al. [29] entrees were targeted entrees system targeted entrees
labelled with were not affected
a promotional
logo, which
additionally
provided
nutritional
information
6
Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Sutherland 2010 USA Supermarket Pre-post 168 > 1 year Supermarket N/A Purchases of Point of sale 12. Purchases Moderate
et al. [30] items were supermarkets star-labelled system of star-labelled
assigned with items items increased*
stars indicating
their relative
healthiness

Vyth et al. 2011 The Healthy RCT 13 > 1 week & ≤ 1 Cafeteria N/A Purchases Point of sale 13a. Purchases Moderate
[31] Netherlands sandwiches, intervention month of healthy system of healthy
soups, and cafeterias; sandwiches; sandwiches were
fresh fruit were 12 control Purchases of not affected
identified with cafeterias healthy soups; 13b. Purchases
a promotional Purchases of of healthy soups
logo fresh fruit were not affected
13c. Fruit
purchases
increased*

Mazza et 2017 USA Emoticons Pre-post 1 cafeteria 1 < month(s) ≤ 6 Cafeteria N/A Purchases Point-of-sale 14a. Purchases Moderate
al. [32] highlighted of healthy system of healthy chips
healthy items beverages were not affected
Purchases of 14b. Purchases of
healthy chips healthy beverages
were not affected

Steenhuis 2004 The In the labelling RCT 17 cafeterias 1 < month(s) ≤ 6 Worksite 2% low Purchases of Point of sale 15a. Purchases of Weak
et al. [33] Netherlands program, low-fat were cafeteria educational low-fat items system and low-fat desserts
products were randomly level (milk, butter, questionnaire increased*
identified with assigned to cheese, meat, 15b. Purchases
a promotional either of 4 desserts). of milk were not
logo. conditions affected
(including 15c. Purchases of
control and butter were not
labelling affected
program) 15d. Purchases of
cheese were not
affected
15e. Purchases
of meat were not
affected

Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Information nudges (nutrition information)

Cioffi et al. 2015 USA Nutrition labels Pre-post 20 small food 6 < months ≤ 12 Small food N/A Purchases of low Point of sale 1a. Purchases of Moderate
[34] were added to a stores store calorie foods; system low calorie foods
selection of pre- Purchases of increased*
packaged meals high calorie 1b. Purchases of
and snacks foods; high calorie foods
Caloric content decreased
of purchases 1c. Caloric
content of items
purchased
decreased*

Hammond 2015 Canada Calorie labels Pre-post 159 ≤ 1 week University N/A Caloric content Questionnaire 2a. Caloric Weak
et al. [35] were added customers cafeteria of purchases; content of
to all cafeteria Calories purchases
menu boards consumed decreased*
and food 2b. Calorie intake
stations decreased*

Milich et al. 1976 USA Foods were Pre-post 450 ≤ 1 week Hospital N/A Caloric content Observer 3. Caloric content Weak
[36] labelled with customers cafeteria of purchases reported of purchases
their caloric decreased;
value (p=0.06)

Vanderlee 2014 Canada Energy, sodium CT 497 1 < month(s) ≤ 6 Hospital 14% low Calorie intake Questionnaire 4. Caloric intake Weak
et al. [37] and fat content customers at cafeteria educational decreased*
were displayed intervention level (high
on digital menu site; 506 school or
boards, as well customers at less)
as a health logo control site 15% low
for healthier income
items (<$CAN 40
000)

Aron et al. 1995 UK Foods were CT 65 ≤ 1 week University N/A Calorie intake Questionnaire 5. Caloric intake Weak
[38] provided with intervention cafeteria increased*
nutrition labels customers;
35 control
customers
6
Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Chu et al. 2009 USA Simplified Pre-post 1 cafeteria > 1 week & ≤ 1 University N/A Caloric content Point of sale 6. Caloric content Moderate
[39] nutrition labels month cafeteria of purchases system of purchases was
were posted not affected
at the point of
selection for
entrée dishes

Webb et al. 2011 USA Calorie CT 1 1 < month(s) ≤ 6 Hospital 13% low Purchases of Point of sale 7a. Purchases Moderate
[40] information experimental cafeteria educational healthy side system of sides dishes
was posted on cafeteria; level (< dishes; increased*
menu boards or 1 control eighth grade, Purchases of 7b. Purchases of
was provided cafeteria some high healthy snacks; snacks increased*
only on posters school and Purchases of 7c. Purchases of
placed away high school healthy entrees entrees were not
from the point graduate) affected
of decision.

Chen et al. 2017 Taiwan Entrees and Pre-post 276 6 < months ≤ 12 Worksite N/A Choice for green- Questionnaire 8a. Choice for Moderate
[41] side dishes customers cafeteria labelled food; green entrée
were labeled for first Attempt to avoid increased*;
with traffic-light survey; 205 red-labelled food 8b. Attempt to
labels customers avoid red-
for second coloured items
survey was not affected.
Sonnenberg 2013 USA Food and Pre-post 389 1 < month(s) ≤ 6 Hospital N/A Purchases of Point of sale 9a. Healthy Strong
et al. [42] beverages customers cafeteria green items system (green) item
were labelled Purchases of red purchases were
red, yellow, items not affected
or green on 9b. Unhealthy
either the menu (red) item
board, shelf, or purchases were
directly on the not affected
packaging.

Whitt et al. 2017 USA Items were Pre-post 1 small food 1 < month(s) ≤ 6 Small food N/A Purchases of Point of sale 10a. Purchases of Moderate
[43] labelled green store store green items system healthy (green)
(healthy), yellow Purchases of red items increased*
(neutral) or red items 10b. Purchases of
(unhealthy). unhealthy (red)
items decreased*

Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Information nudges (signage)

Allan et al. 2015 UK Signs visually RCT >20,000 1 < month(s) ≤ 6 Small food N/A Purchases of Point of sale 1a. Purchases Moderate
[44] arranged snacks purchases store high-calorie system of high
and drinks from snacks calorie snacks
least caloric to Purchases high- decreased*;
most caloric, calorie drinks 1b. Purchases
with arrows of high calorie
indicating their drinks were not
location in store affected.

Buscher 2001 Canada Signs with Pre-post 2,280 > 1 week & ≤ 1 University N/A Purchases of Point of sale 2a. Vegetable Moderate
et al. [45], promotional students month cafeteria vegetable basket system and basket purchases
study 1 prompts were potentially Purchases of hand-counts were not affected
located at exposed pretzels 2b. Pretzel
the cafeteria to the Purchases of purchases
entrance and intervention yoghurt increased*
in front of the Purchases of 2c. Yoghurt
targeted foods fruit basket purchases
increased*
2d. Fruit basket
purchases were
not affected

Buscher 2001 Canada Signs with Pre-post 2,280 > 1 week & ≤ 1 University N/A Purchases of Point of sale 3. Yoghurt Moderate
et al. [45], promotional students month cafeteria yoghurt system purchases
study 2 prompts were potentially increased*
located at exposed
the cafeteria to the
entrance and intervention
in front of
the targeted
yoghurt

Montuclard 2017 USA A water sign Pre-post 357 students 1 < month(s) ≤ 6 University N/A Choice for water Questionnaire 4. Choice for Moderate
et al. [46] was taped to pre- cafeteria water increased*
the cafeterias intervention
soda dispensers survey; 301
and coffee students
dispensers post-
intervention
survey
6
Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Ogawa et 2011 Japan Health and CT 1 intervention 1 < month(s) ≤ 6 Supermarket N/A Purchases of Point of sale 5a. Vegetable Moderate
al. [47] nutrition supermarket; fruits system purchases
information 1 control Purchases of increased*
related to supermarket vegetables 5b. Fruit
consumption purchases were
of fruits and not affected.
vegetables was
displayed on
posters near
fruit/vegetable
display and/
or checkout
counter

Policastro 2017 USA Water Pre-post 2,393 ≤ 1 week University N/A Choice for water Point of sale 6. Water Moderate
et al. [48] consumption students cafeteria system purchases
was promoted covering increased*
through signage 6,730
promoting transactions
swapping soda
for water
Scourboutakos 2017 Canada Posters Pre-post 368 to 510 1 < month(s) ≤ 6 University N/A Purchases of Observer 7a. Purchases of Weak
et al. [49] promoted students cafeteria water reported water increased*
water and fruit per data Purchases of 7b. Purchases of
and vegetable collection day fruits fruit increased*
consumption Purchases of 7c. Purchases
vegetables of vegetables
increased*

Mazza et 2018 USA A health Pre-post 1 cafeteria > 1 week & ≤ 1 Hospital N/A Purchases of Point of sale 8a. Purchases Moderate
al. [32] message month cafeteria healthy chips system of healthy chips
stating the % of Purchases were not affected
daily calories of healthy 8b. Purchases of
contained in beverages healthy beverages
beverages, were not affected
and required
exercise to burn
calories of chips

Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Mazza et 2018 USA A health Pre-post 1 cafeteria > 1 week & ≤ 1 Hospital N/A Purchases of Point of sale 9a. Purchases Moderate
al. [32] message month cafeteria healthy chips system of healthy chips
stating the % of Purchases increased*
daily calories of healthy 9b. Purchases of
contained in beverages healthy beverages
chips, and were not affected
required
exercise to
burn calories of
beverages

Payne et 2015 USA Messages CT 396,017 > 1 week & ≤ 1 Supermarket Supermarkets Purchases Point-of-sale 10. Fruit and Strong
al. [50], on grocery individual month were located of fruits and system vegetable
study 1 carts stated person in area with 7% vegetables purchases
the number transactions unemployment increased*
of fruits and and
vegetable items 24% only
customers of high school
that particular education
store normally
purchased

Payne et 2015 USA Messages Pre-post 575,689 > 1 week & ≤ 1 Supermarket Supermarkets Purchases Point-of-sale 11. Fruit and Strong
al. [50], on grocery individual month were located of fruits and system vegetable
study 2 carts stated person in area with 7% vegetables purchases
the number transactions unemployment increased*
of F&V items and
customers of 24% only
that particular high school
store normally education
purchased

Salmon et 2015 The The presence CT 127 N/A Supermarket 10% low Purchases of Collection of 12. Purchases of Strong
al. [51] Netherlands of a banner customers educational low-fat cheese receipts low-fat cheese
with was level were not affected
manipulated (primary
(absent/present, school or
which stated lower levels
that a particular of high
low-fat cheese school)
was the most
sold brand of
cheese in the
supermarket.
6
Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Position nudges

Kroese et 2016 The Unhealthy CT 2 intervention ≤ 1 week Small food N/A Purchases of Point-of-sale 1a. Healthy Moderate
al. [52] Netherlands snacks at the stores; 1 store healthy snacks; system snack purchases
check-out control store Purchases increased*
counter were of unhealthy 1b. Unhealthy
replaced by snacks snack purchases
healthy snacks were not affected

Meiselman 1994 UK Candy was Pre-post 43 students ≤ 1 week University N/A Choice for candy Questionnaire 2a. Candy Weak
et al. [53], repositioned cafeteria Energy intake and food selection
study 1 from four cash- weighing decreased*
points to one 2b. Energy intake
distant cash- was not affected
point

Meiselman 1994 UK Potato Pre-post 60 students > 1 week & ≤ 1 University N/A Choice for potato Questionnaire 3. Potato chips Weak
et al. [53], chips were month cafeteria chips selection
study 2 repositioned decreased*
from meal line
to distant snack
bar

Meyers et 1980 USA High calorie Pre-post 4,412 food ≤ 1 week Hospital N/A Choice for high Observer- 4. High calorie Moderate
al. [54] desserts were choices were cafeteria calorie desserts reported dessert choice
placed in the observed was not affected
rear position on
buffet line

Rozin et 2011 USA Salad bar Pre-post 1 cafeteria 1 < month(s) ≤ 6 Hospital N/A Purchases Food 5. Salad bar Moderate
al. [55], ingredients cafeteria of salad bar weighing purchases
study 3 were placed on ingredients increased*
edge position
of salad bar vs.
middle position

Van Gestel 2017 The Unhealthy Pre-post 1 small food > 1 week & ≤ 1 Small food N/A Purchases of Point of sale 6. Healthy snack Moderate
et al. [56] Netherlands snacks at the store month store healthy snacks system purchases
check-out increased*
counter were
replaced by
healthy snacks

Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Chapman 2012 UK Confectionery Pre-post 1 cafeteria ≤ 1 week University N/A Purchases of Point of sale 7a. Healthy Moderate
et al. [57], was removed cafeteria fruit; system fruit purchases
study 1 from check-out Purchases of decreased*
counters and confectionery 7b. Unhealthy
replaced by fruit confectionary
purchases
increased*

De Wijk et 2016 The Wholegrain CT 2 1 < month(s) ≤ 6 Supermarket N/A Purchases of Point of sale 8. Healthy bread Moderate
al. [58] Netherlands bread was supermarkets wholegrain system purchases were
placed near bread not affected
entrance vs.
away from
entrance

Thorndike 2017 USA Stores improved RCT 3 intervention 1 < month(s) ≤ 6 Small food Store was Purchases Point of sale 9. Fruit and Moderate
et al. [59] visibility of fruits stores; 3 store located in of fruits and system (WIC vegetable
and vegetables control stores low-income vegetables voucher purchases
through new urban redemption) increased*
supplies community
(e.g. baskets,
shelving)

Winkler et 2016 Denmark Sugar CT 4 intervention > 1 week & ≤ 1 Supermarket N/A Purchases of Point of sale 10a. Healthy Moderate
al. [60] confectionery stores; 2 month healthy snacks; system snack purchases
at one checkout control stores Purchases were generally
counter was of sugar not affected
replaced by confectionary 10b. Unhealthy
healthy snacks purchases were
not affected

Mixed nudges

Gittelsohn 2013 USA Environmental RCT 98 > 1 year Supermarket Years of Healthy food Questionnaire 1a. Healthy food Weak
et al. [61] changes participants schooling; purchasing score purchasing score
included from 10.9y Unhealthy food was not affected
demonstrations intervention (intervention purchasing score 1b. Unhealthy
of healthier condition; 47 particpants) food purchasing
cooking participants and 9.3y score was not
methods, from control (control affected
taste-tests, and condition participants)
display of point-
of-purchase
materials (e.g.
posters and
shelf labels)
6
Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Dorresteijn 2013 The Environmental Pre-post 1 cafeteria > 1 week & ≤ 1 Hospital N/A Purchases of Point of sale 2a. Normal soup Weak
et al. [62] Netherlands changes month cafeteria normal soup system and purchases were
included signage Purchases of hand-counts not affected
promoting healthier soup 2b. Healthier soup
low-sodium Purchases purchases were
soup and low- of normal not affected
fat croissants. croissants 2c. Normal crois-
Also, margarine Purchases sant purchases
was made less of healthier were not affected
accessible croissants 2d. Healthier
whereas butter Purchases of croissant pur-
was made more butter chases were not
accessible. Purchases of affected
margarine 2e. Butter pur-
chases increased*
2f. Margarine
purchases de-
creased*
Gamburzew 2016 France Environmental CT 6,625 1 < month(s) ≤ 6 Supermarket N/A Purchases of Point of sale 3a. Purchases of Weak
et al. [63] changes customers targeted foods system fruits and vegeta-
included shelf (fruits and bles increased*
labels indicating vegetables; 3b. Purchases
healthy foods, starches; meat/ of starches
signage fish/eggs; mixed increased*
explaining the dishes and 3c. Purchases of
labelling system, sandwiches; meat/fish/eggs
placement dairy products). were not affected
strategies, and 3d. Purchases of
a taste-testing mixed dishes/
booth. sandwiches were
not affected
3e. Purchases of
dairy were not
affected

Gittelsohn 2010 USA Environmental CT 64 6 < months ≤ 12 Supermarket For inter- Healthy food Questionnaires 4a. Healthy food Strong
et al. [64] changes intervention vention and purchasing score and dietary purchasing score
included participants; comparison Unhealthy food recall was not affected
posters, shelf 53 control group, re- purchasing score 4b. Unhealthy
labels, cooking participants spectively: Calorie intake food purchasing
demonstrations Years of score was not
and taste tests. schooling, affected
12.5y and 4c. Calorie intake
12.4y; was not affected
Percentage
unemployed,
35.9% and
18.8%.

Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Foster et 2014 USA Environmental RCT 4 intervention 6 < months ≤ 12 Supermarket Supermarkets Purchases of tar- Point of sale 5a. Purchases of Moderate
al. [65] changes includ- supermarkets; located in geted foods (milk, system some targeted
ed placement 4 control low-income, cereal, frozen products within
strategies supermarkets high-minority meals, in aisle-bev- the milk category
(multiple neighborhoods erages, checkout increased*
facings, prime cooler beverages, 5b. Purchases of
placement),sig- water) cereals were not
nage, shelf-tags, affected
cross-promo- 5c. Purchases of
tion of healthy some targeted
foods, and taste products within
testing. the frozen
meal category
increased*
5d. Purchases of
some targeted
products within
the in-aisle bev-
erages category
decreased to a
lesser extent in
the intervention
stores as com-
pared to control*
5e. Purchases of
check-out cooler
beverages were
not affected
5f. Purchases of
water increased*

Lawman et 2015 USA Availability of Pre-post 8,671 6 < months ≤ 12 Small food N/A Mean energy Observer 6. The Moderate
al. [66] healthy foods customers store purchased reported intervention
was increased at baseline; did not affect
and promot- 5,949 mean energy
ed through customers at purchased.
banners, shelf follow-up
labels, and rec-
ipes. A subset
of stores was
provided addi-
tional business
trainings and
mini-grants for
storing their
inventory of
healthy foods
(high-intensity
intervention).
6
Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Levy et al. 2012 USA Foods were Pre-post 4,642 1 < month(s) ≤ 6 Hospital 28% employed Purchases of Point of sale 7a. Green item Strong
[67] labelled green, customers cafeteria as service green items system purchases
orange or red. workers, Purchases of red increased∞
Additionally, administrative items 7b. Red item
a choice support, purchases
architecture technicians decreased∞
intervention
was added
which increased
visibility and
accessibility of
green-labelled
foods and
beverages while
decreasing the
same for certain
red-labelled
items.

Lowe et al. 2010 USA Environmental Pre-post 49 customers 1 < month(s) ≤ 6 Hospital N/A Caloric content Point of sale 8. Caloric content Moderate
[68] changes cafeteria of purchases system of purchases
included decreased*
increased
availability of
foods lower in
energy density.
Additionally, a
labelling system
was introduced
which color-
coded food
items.

Cardenas 2015 Peru Fruit was repo- Pre-post 150 > 1 week & ≤ 1 University N/A Fruit purchases Hand-counts 9. Fruit purchases Moderate
et al. [69] sitioned from a customers month cafeteria were not affected
distant position
to a more acces-
sible location
near the point
of purchase.
Additionally,
signage high-
lighted health
benefits of fruit
consumption.

Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Thorndike 2012 USA Foods were Pre-post 1 cafeteria 1 < month(s) ≤ 6 Hospital N/A Purchases of Point of sale 10a. Purchases Moderate
et al. [70] labelled cafeteria green items system of green items
red, yellow Purchases of red increased∞.
and green. items 10b. Purchases
Additionally, of red items
healthy foods decreased∞
were located
to convenient
positions and
unhealthy op-
tions were made
less convenient.

Steenhuis 2004 The In the food RCT 17 cafeterias 1 < month(s) ≤ 6 Worksite 2% low Purchases of Point of sale 11. Purchases Weak
et al. [33] Netherlands supply program, were random- cafeteria educational low-fat items system and of low-fat items
the availability ly assigned level questionnaire were not affected
of low-fat items to either of
increased and 4 conditions
was made (including
salient with control and
signage. food supply
program)

Thorndike 2014 USA Items were Pre-post 1 cafeteria > 1 year Hospital 29% low Purchases of Point of sale 12a. Purchases Weak
et al. [71] labelled green, cafeteria educated green items system of green-labelled
yellow or red. jobs (service Purchases of red items increased*
Additionally, workers, items 12b. Purchases of
items were administrative red-labelled items
rearranged to support, decreased*
make some of technicians)
the green items
more visible and
some red items
less visible.

Seward et 2016 USA The full inter- CT 4 1 < month(s) ≤ 6 University N/A Purchases of Observer 13a. Purchases Moderate
al. [72] vention included experimental cafeteria green items reported of green items
traffic-light cafeterias; Purchases of red were not affected
labels, accessi- 2 control items in neither full or
bility changes cafeterias minimal interven-
and tray stickers tion cafeterias.
visualizing 13b. Purchases
recommended of red items were
portions of food not affected in
types. The mini- neither full or
mal intervention minimal interven-
only included tion cafeterias.
accessibility
changes.
6
Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Lee-Kwan 2015 USA During phase CT 3 intervention 1 < month(s) ≤ 6 Small food Carry-outs Purchases of Point of sale 14. Purchases Moderate
et al. [73] 1, menus stores; 4 store were based healthy items system of healthy items
were revised control stores in low-in- were not affected.
to emphasize come neigh-
healthy foods bourhoods
with labels.
Consecutively,
during phase
2, (additional)
healthy sides
and beverages
were introduced
and promoted.

Availability, size, functionality, and presentation nudges

Diliberti et 2004 USA During baseline CT 180 > 1 week & ≤ 1 Cafeteria N/A Energy intake Food Energy intake Moderate
al. [74] conditions, the customers month weighing from the
portion size of pasta entree,
the entree was accompaniments,
the standard and entire meal
100% portion; in increased*
the experimen-
tal condition
the size was
increased to
150%.
Vandenbroele 2018 Belgium Different sizes CT 161 > 1 week & ≤ 1 Supermarket N/A Meat purchases Point-of-sale The introduction Moderate
et al. [75] of sausages customers month system of smaller portion
were available: a who bought size alternatives
150 g por- targeted was associated
tion(default); a product with less meat
125 g in-be- being purchased*
tween portion;
or a small, 100 g
portion.

Table 2. Study characteristics and main findings of included studies categorized by TIPPME intervention type (continued)

Author Year Country Nudge Study Study size Intervention Setting SEP Study Outcome Main finding(s) Quality
description design duration outcome(s) assessment assessment

Payne et 2016 USA Large green CT 1 intervention > 1 week & ≤ 1 Supermarket N/A Purchases Point-of-sale Green arrows Strong
al., study 1 arrows were store; 1 month of fruit & system on floors were
[76] placed on the control store vegetables associated with
floor directing increased fruit
attention to the and vegetable
store’s produce purchases*
section.

Payne et 2016 USA Large green CT 1 intervention > 1 week & ≤ 1 Supermarket N/A Purchases Point-of-sale Green arrows Strong
al., study 2 arrows were store; 1 month of fruit & system on floors were
[76] placed on the control store vegetables associated with
floor directing increased fruit
attention to the and vegetable
store’s produce purchases*
section.

Tal et al. 2015 USA Samples (no, CT 120 N/A Supermarket N/A Purchases fruit & Observer- Receiving an Weak
[77] apple, or cookie) customers vegetables reported apple sample was
were offered to associated with
participants at increased subse-
the entrance of quent purchases
the store of fruits and veg-
etables vs. cookie
or no sample
(p=0.06).

*Asterix indicates statistical significance (p <0.05); ∞ Main findings are aggregated across two consecutive intervention phases for which data on statistical significance was
not available.
6
chapter 6 effec ts of nudges in real - life purchasing environments
REFERENCES
1. GBD 2017 Diet Collaborators. Health effects of dietary risks in 195 countries, 1990-
2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet.
2019;393(10184):1958-72. doi: 10.1016/s0140-6736(19)30041-8.
2. Darmon N, Drewnowski A. Does social class predict diet quality? The American Journal of
Clinical Nutrition. 2008;87(5):1107-17. doi: 10.1093/ajcn/87.5.1107.
3. McGill R, Anwar E, Orton L, Bromley H, Lloyd-Williams F, O’Flaherty M, et al. Are
interventions to promote healthy eating equally effective for all? Systematic review of
socioeconomic inequalities in impact. BMC Public Health. 2015;15:457. doi: 10.1186/s12889-
015-1781-7.
4. Mechanic D. Disadvantage, inequality, and social policy. Health Aff (Millwood).
2002;21(2):48-59. doi: 10.1377/hlthaff.21.2.48.
5. White M, Adams J, Heywood P. How and why do interventions that increase health overall
widen inequalities within populations. In: Babones (Ed.). Social inequality and public health.
Bristol: Policy Press; 2009. p.64-81.
6. Houlihan S. Dual-process models of health-related behaviour and cognition: a review of
theory. Public Health. 2018;156:52-9. doi: 10.1016/j.puhe.2017.11.002.
7. Marteau TM, Hollands GJ, Fletcher PC. Changing human behavior to prevent disease: the
importance of targeting automatic processes. Science. 2012;337(6101):1492-5. doi: 10.1126/
science.1226918.
8. Thaler RH, Sunstein CS. Nudge: Improving decisions about health, wealth, and happiness.
Yale University Press; 2008.
9. Hansen PG, Skov LR, Skov KL. Making Healthy Choices Easier: Regulation versus
Nudging. Annual Review of Public Health. 2016;37:237-51. doi: 10.1146/annurev-
publhealth-032315-021537.
10. Hollands GJ, Bignardi G, Johnston M, Kelly MP, Ogilvie D, Petticrew M, et al. The TIPPME
intervention typology for changing environments to change behaviour. Nature Human
Behaviour. 2017;1(8). doi: 10.1038/s41562-017-0140.
11. Broers VJV, De Breucker C, Van den Broucke S, Luminet O. A systematic review and meta-
analysis of the effectiveness of nudging to increase fruit and vegetable choice. Eur J Public
Health. 2017;27(5):912-20. doi: 10.1093/eurpub/ckx085.
12. Bucher T, Collins C, Rollo ME, McCaffrey TA, De Vlieger N, Van der Bend D, et al. Nudging
consumers towards healthier choices: a systematic review of positional influences on food
choice. British Journal of Nutrition. 2016;115(12):2252-63. doi: 10.1017/s0007114516001653.
13. Hollands GJ, Carter P, Anwer S, King SE, Jebb SA, Ogilvie D, et al. Altering the availability
or proximity of food, alcohol, and tobacco products to change their selection and
consumption. Cochrane Database of Systematic Reviews. 2019(8). doi: 10.1002/14651858.
CD012573.pub2.
14. Dolan P, Hallsworth M, Halpern D, King D, Metcalfe R, Vlaev I. Influencing behaviour:
The mindspace way. Journal of Economic Psychology. 2012;33(1):264-77. doi: 10.1016/j.
joep.2011.10.009.
15. Marchiori DR, Adriaanse MA, De Ridder DTD. Unresolved questions in nudging research:
Putting the psychology back in nudging. Social and Personality Psychology Compass.
2017;11(1):e12297. doi: 10.1111/spc3.12297.
154
16. Arno A, Thomas S. The efficacy of nudge theory strategies in influencing adult dietary
behaviour: a systematic review and meta-analysis. BMC Public Health. 2016;16:676. doi:
10.1186/s12889-016-3272-x.
17. Thomas BH, Ciliska D, Dobbins M, Micucci S. A process for systematically reviewing
the literature: providing the research evidence for public health nursing interventions.
Worldviews on Evidence-Based Nursing. 2004;1(3):176-84. doi: 10.1111/j.1524-
475X.2004.04006.x.
18. Ogilvie D, Fayter D, Petticrew M, Sowden A, Thomas S, Whitehead M, et al. The harvest
plot: a method for synthesising evidence about the differential effects of interventions.
BMC Medical Research Methodology. 2008;8:8. doi: 10.1186/1471-2288-8-8.
19. Cawley J, Sweeney MJ, Sobal J, Just DR, Kaiser HM, Schulze WD, et al. The impact of a
supermarket nutrition rating system on purchases of nutritious and less nutritious foods.
Public Health Nutrition. 2015;18(1):8-14. doi: 10.1017/s1368980014001529.
20. Dubbert PM, Johnson WG, Schlundt DG, Montague NW. The influence of caloric information
on cafeteria food choices. Journal of Applied Behavior Analysis. 1984;17(1):85-92.
21. Elbel B, Taksler GB, Mijanovich T, Abrams CB, Dixon LB. Promotion of healthy eating
through public policy: a controlled experiment. American Journal of Preventive Medicine.
2013;45(1):49-55. doi: 10.1016/j.amepre.2013.02.023.
22. Eldridge AL, Snyder MP, Faus NG, Kotz K. Development and evaluation of a labeling program
for low-fat foods in a discount department store foodservice area. Journal of Nutrition
Education. 1997;29(3):159-61. doi: http://dx.doi.org/10.1016/S0022-3182%2897%2970181-X.
23. Freedman MR, Connors, R. Point-of-purchase nutrition information influences food-
purchasing behaviors of college students: a pilot study. Journal of the American Dietetic
Association. 2011;111(5 Suppl):S42-6. doi: 10.1016/j.jada.2011.03.008.
24. Hobin E, Bollinger B, Sacco J, Liebman E, Vanderlee L, Zuo F, et al. Consumers’ Response 6
to an On-Shelf Nutrition Labelling System in Supermarkets: Evidence to Inform Policy and
Practice. Milbank Q. 2017;95(3):494-534. doi: 10.1111/1468-0009.12277.
25. Hoefkens C, Lachat C, Kolsteren P, Van Camp J, Verbeke W. Posting point-of-purchase
nutrition information in university canteens does not influence meal choice and nutrient
intake. The American Journal of Clinical Nutrition. 2011;94(2):562-70. doi: 10.3945/
ajcn.111.013417.
26. Johnson WG, Corrigan SA, Schlundt DG, Dubbert PM. Dietary restraint and eating behavior
in the natural environment. Addictive Behaviors. 1990;15(3):285-90.
27. Lassen AD, Beck A, Leedo E, Andersen EW, Christensen T, Mejborn H, et al. Effectiveness of
offering healthy labelled meals in improving the nutritional quality of lunch meals eaten in
a worksite canteen. Appetite. 2014;75:128-34. doi: 10.1016/j.appet.2013.12.005.
28. Levin S. Pilot study of a cafeteria program relying primarily on symbols to promote healthy
choices. Journal of Nutrition Education. 1996;28(5):282-5. doi: http://dx.doi.org/10.1016/
S0022-3182%2896%2970102-4.
29. Sproul AD, Canter DD, Schmidt JB. Does point-of-purchase nutrition labeling influence meal
selections? A test in an Army cafeteria. Military Medicine. 2003;168(7):556-60.
30. Sutherland LA, Kaley LA, Fischer L. Guiding stars: the effect of a nutrition navigation
program on consumer purchases at the supermarket. The American Journal of Clinical
Nutrition. 2010;91(4):1090s-4s. doi: 10.3945/ajcn.2010.28450C.
155
chapter 6 effec ts of nudges in real - life purchasing environments
31. Vyth EL, Steenhuis IH, Heymans MW, Roodenburg AJ, Brug J, Seidell JC. Influence of
placement of a nutrition logo on cafeteria menu items on lunchtime food Choices at Dutch
work sites. Journal of the American Dietetic Association. 2011;111(1):131-6. doi: 10.1016/j.
jada.2010.10.003.
32. Mazza MC, Dynan L, Siegel RM, Tucker AL. Nudging Healthier Choices in a Hospital
Cafeteria: Results From a Field Study. Health Promotion Practice. 2018;19(6):925-34. doi:
10.1177/1524839917740119.
33. Steenhuis I, Van Assema P, Van Breukelen G, Glanz K, Kok G, De Vries H. The impact
of educational and environmental interventions in Dutch worksite cafeterias. Health
Promotion International. 2004;19(3):335-43. doi: 10.1093/heapro/dah307.
34. Cioffi CE, Levitsky DA, Pacanowski CR, Bertz F. A nudge in a healthy direction. The effect
of nutrition labels on food purchasing behaviors in university dining facilities. Appetite.
2015;92:7-14. doi: 10.1016/j.appet.2015.04.053.
35. Hammond D, Lillico HG, Vanderlee L, White CM, Reid JL. The impact of nutrition labeling on
menus: a naturalistic cohort study. American Journal of Health Behavior. 2015;39(4):540-8.
doi: 10.5993/ajhb.39.4.10.
36. Milich R, Anderson J, Mills M. Effects of visual presentation of caloric values on food buying
by normal and obese persons. Perceptual and Motor Skills. 1976;42(1):155-62. doi: 10.2466/
pms.1976.42.1.155.
37. Vanderlee L, Hammond D. Does nutrition information on menus impact food choice?
Comparisons across two hospital cafeterias. Public Health Nutrition. 2014;17(6):1393-402.
doi: 10.1017/s136898001300164x.
38. Aron JI, Evans RE, Mela DJ. Paradoxical effect of a nutrition labelling scheme in a student
cafeteria. Nutrition Research. 1995;15(9):1251-61. doi: 10.1016/0271-5317(95)02001-C.
39. Chu YH, Frongillo EA, Jones SJ, Kaye GL. Improving patrons’ meal selections through the use
of point-of-selection nutrition labels. American Journal of Public Health. 2009;99(11):2001-
5. doi: 10.2105/ajph.2008.153205.
40. Webb KL, Solomon LS, Sanders J, Akiyama C, Crawford PB. Menu labeling responsive to
consumer concerns and shows promise for changing patron purchases. Journal of Hunger
and Environmental Nutrition. 2011;6(2):166-78. doi: 10.1080/19320248.2011.576210.
41. Chen HJ, Weng SH, Cheng YY, Lord AYZ, Lin HH, Pan WH. The application of traffic-light food
labelling in a worksite canteen intervention in Taiwan. Public Health. 2017;150:17-25. doi:
10.1016/j.puhe.2017.04.005.
42. Sonnenberg L, Gelsomin E, Levy DE, Riis J, Barraclough S, Thorndike AN. A traffic light
food labeling intervention increases consumer awareness of health and healthy
choices at the point-of-purchase. Preventive Medicine. 2013;57(4):253-7. doi: 10.1016/j.
ypmed.2013.07.001.
43. Whitt OR, Jilcott Pitts SB, Rafferty AP, Payne CR, Ng SW. The effects of traffic light labelling
versus cartoon labelling on food and beverage purchases in a children’s hospital setting.
Pediatric Obesity. 2018;13(4):265-8. doi: 10.1111/ijpo.12232.
44. Allan JL, Johnston M, Campbell N. Snack purchasing is healthier when the cognitive demands
of choice are reduced: A randomized controlled trial. Health Psychology. 2015;34(7):750-5.
doi: 10.1037/hea0000173.
45. Buscher LA, Martin KA, Crocker S. Point-of-purchase messages framed in terms of cost,
convenience, taste, and energy improve healthful snack selection in a college foodservice
setting. Journal of the American Dietetic Association. 2001;101(8):909-13. doi: 10.1016/
s0002-8223(01)00223-1.
156
46. Montuclard AL, Park-Mroch J, O’Shea AMJ, Wansink B, Irvin J, Laroche HH. College Cafeteria
Signage Increases Water Intake but Water Position on the Soda Dispenser Encourages
More Soda Consumption. Journal of Nutrition Education and Behavior. 2017;49(9):764-71.
e1. doi: 10.1016/j.jneb.2017.05.361.
47. Ogawa Y, Tanabe N, Honda A, Azuma T, Seki N, Suzuki T, et al. Point-of-purchase health
information encourages customers to purchase vegetables: objective analysis by using a
point-of-sales system. Environmental Health and Preventive Medicine. 2011;16(4):239-46.
doi: 10.1007/s12199-010-0192-8.
48. Policastro P, Palm T, Schwartz J, Chapman G. Targeted Calorie Message Promotes
Healthy Beverage Consumption Better than Charity Incentive. Obesity (Silver Spring).
2017;25(8):1428-34. doi: 10.1002/oby.21885.
49. Scourboutakos MJ, Mah CL, Murphy SA, Mazza FN, Barrett N, McFadden B, et al. Testing a
Beverage and Fruit/Vegetable Education Intervention in a University Dining Hall. Journal
of Nutrition Education and Behavior. 2017;49(6):457-65.e1. doi: 10.1016/j.jneb.2017.02.003.
50. Payne CR, Niculescu M, Just DR, Kelly MP. Shopper marketing nutrition interventions: Social
norms on grocery carts increase produce spending without increasing shopper budgets.
Preventive Medicine Reports. 2015;2:287-91. doi: 10.1016/j.pmedr.2015.04.007.
51. Salmon SJ, De Vet E, Adriaanse MA, Fennis BM, Veltkamp M, De Ridder DTD. Social proof in
the supermarket: Promoting healthy choices under low self-control conditions. Food Quality
and Preference. 2015;45:113-20. doi: http://dx.doi.org/10.1016/j.foodqual.2015.06.004.
52. Kroese FM, Marchiori DR, de Ridder DT. Nudging healthy food choices: a field experiment at the
train station. Journal of Public Health (Oxf). 2016;38(2):e133-7. doi: 10.1093/pubmed/fdv096.
53. Meiselman HL, Hedderley D, Staddon SL, Pierson BJ, Symonds CR. Effect of effort on meal
selection and meal acceptability in a student cafeteria. Appetite. 1994;23(1):43-55. doi:
10.1006/appe.1994.1033.
6
54. Meyers AW, Stunkard AJ, Coll M. Food accessibility and food choice. A test of Schachter’s
externality hypothesis. Archives of General Psychiatry. 1980;37(10):1133-5.
55. Rozin P, Scott S, Dingley M, Urbanek JK, Jiang H, Kaltenbach M. Nudge to nobesity I:
Minor changes in accessibility decrease food intake. Judgment and Decision Making.
2011;6(4):323-32.
56. Van Gestel LC, Kroese FM, De Ridder DTD. Nudging at the checkout counter - A longitudinal
study of the effect of a food repositioning nudge on healthy food choice. Psychology &
Health. 2017:1-10. doi: 10.1080/08870446.2017.1416116.
57. Chapman K, Ogden J. Nudging customers towards healthier choices. An intervention in
the university canteen. Journal of Food Research. 2012;1(2):13-21.
58. De Wijk RA, Maaskant AJ, Polet IA, Holthuysen NT, van Kleef E, Vingerhoeds MH. An In-
Store Experiment on the Effect of Accessibility on Sales of Wholegrain and White Bread in
Supermarkets. PloS one. 2016;11(3):e0151915. doi: 10.1371/journal.pone.0151915.
59. Thorndike AN, Bright OM, Dimond MA, Fishman R, Levy DE. Choice architecture to promote
fruit and vegetable purchases by families participating in the Special Supplemental
Program for Women, Infants, and Children (WIC): randomized corner store pilot study.
Public Health Nutrition. 2017;20(7):1297-305. doi: 10.1017/s1368980016003074.
60. Winkler LL, Christensen U, Glumer C, Bloch P, Mikkelsen BE, Wansink B, et al. Substituting sugar
confectionery with fruit and healthy snacks at checkout - a win-win strategy for consumers
and food stores? a study on consumer attitudes and sales effects of a healthy supermarket
intervention. BMC Public Health. 2016;16(1):1184. doi: 10.1186/s12889-016-3849-4.
157
chapter 6 effec ts of nudges in real - life purchasing environments
61. Gittelsohn J, Kim EM, He S, Pardilla M. A food store-based environmental intervention
is associated with reduced BMI and improved psychosocial factors and food-related
behaviors on the Navajo nation. The Journal of Nutrition. 2013;143(9):1494-500. doi:
10.3945/jn.112.165266.
62. Dorresteijn JA, van der Graaf Y, Zheng K, Spiering W, Visseren FL. The daily 10 kcal
expenditure deficit: a before-and-after study on low-cost interventions in the work
environment. BMJ Open. 2013;3(1). doi: 10.1136/bmjopen-2012-002125.
63. Gamburzew AD, N.; Gazan, R.; Dubois, C.; Maillot, M.; Tome, D.; Raffin, S.; Darmon, N.
In-store marketing of inexpensive foods with good nutritional quality in disadvantaged
neighborhoods: increased awareness, understanding, and purchasing. The International
Journal of Behavioral Nutrition and Physical Activity. 2016;13(1):104. doi: 10.1186/s12966-
016-0427-1.
64. Gittelsohn J, Vijayadeva V, Davison N, Ramirez V, Cheung LW, Murphy S, et al. A food store
intervention trial improves caregiver psychosocial factors and children’s dietary intake in
Hawaii. Obesity (Silver Spring). 2010;18 Suppl 1:S84-90. doi: 10.1038/oby.2009.436.
65. Foster GD, Karpyn A, Wojtanowski AC, Davis E, Weiss S, Brensinger C, et al. Placement
and promotion strategies to increase sales of healthier products in supermarkets in low-
income, ethnically diverse neighborhoods: a randomized controlled trial. The American
Journal of Clinical Nutrition. 2014;99(6):1359-68. doi: 10.3945/ajcn.113.075572.
66. Lawman HG, Vander Veur S, Mallya G, McCoy TA, Wojtanowski A, Colby L, et al. Changes
in quantity, spending, and nutritional characteristics of adult, adolescent and child
urban corner store purchases after an environmental intervention. Preventive Medicine.
2015;74:81-5. doi: 10.1016/j.ypmed.2014.12.003.
67. Levy DE, Riis J, Sonnenberg LM, Barraclough SJ, Thorndike AN. Food choices of minority and
low-income employees: a cafeteria intervention. American Journal of Preventive Medicine.
2012;43(3):240-8. doi: 10.1016/j.amepre.2012.05.004.
68. Lowe MR, Tappe KA, Butryn ML, Annunziato RA, Coletta MC, Ochner CN, et al. An
intervention study targeting energy and nutrient intake in worksite cafeterias. Eating
Behaviors. 2010;11(3):144-51. doi: 10.1016/j.eatbeh.2010.01.002.
69. Cardenas MK, Benziger CP, Pillay TD, Miranda JJ. The effect of changes in visibility and
price on fruit purchasing at a university cafeteria in Lima, Peru. Public Health Nutrition.
2015;18(15):2742-9. doi: 10.1017/s1368980014002730.
70. Thorndike AN, Sonnenberg L, Riis J, Barraclough S, Levy DE. A 2-phase labeling and choice
architecture intervention to improve healthy food and beverage choices. American Journal
of Public Health. 2012;102(3):527-33. doi: 10.2105/ajph.2011.300391.
71. Thorndike AN, Riis J, Sonnenberg LM, Levy DE. Traffic-light labels and choice architecture:
promoting healthy food choices. American Journal of Preventive Medicine. 2014;46(2):143-
9. doi: 10.1016/j.amepre.2013.10.002.
72. Seward MW, Block JP, Chatterjee A. A Traffic-Light Label Intervention and Dietary Choices in
College Cafeterias. American Journal of Public Health. 2016;106(10):1808-14. doi: 10.2105/
ajph.2016.303301.
73. Lee-Kwan SH, Bleich SN, Kim H, Colantuoni E, Gittelsohn J. Environmental Intervention
in Carryout Restaurants Increases Sales of Healthy Menu Items in a Low-Income
Urban Setting. American Journal of Health Promotion. 2015;29(6):357-64. doi: 10.4278/
ajhp.130805-QUAN-408.
158
74. Diliberti N, Bordi PL, Conklin MT, Roe LS, Rolls BJ. Increased portion size leads to increased
energy intake in a restaurant meal. Obesity Research. 2004;12(3):562-8. doi: 10.1038/
oby.2004.64.
75. Vandenbroele J, Slabbinck H, Van Kerckhove A, Vermeir I. Curbing portion size effects by
adding smaller portions at the point of purchase. Food Quality and Preference. 2018;64:82-
7. doi: http://dx.doi.org/10.1016/j.foodqual.2017.10.015.
76. Payne CR, Niculescu M, Just DR, Kelly MP. This Way to Produce: Strategic Use of Arrows on
Grocery Floors Facilitate Produce Spending Without Increasing Shopper Budgets. Journal
of Nutrition Education and Behavior. 2016;48(7):512-3.e1. doi: 10.1016/j.jneb.2016.05.001.
77. Tal A, Wansink B. An apple a day brings more apples your way: Healthy samples prime
healthier choices. Psychology & Marketing. 2015;32(5):575-84. doi: http://dx.doi.
org/10.1002/mar.20801.
78. Hollands GJ, Shemilt I, Marteau TM, Jebb SA, Kelly MP, Nakamura R, et al. Altering micro-
environments to change population health behaviour: towards an evidence base for choice
architecture interventions. BMC Public Health. 2013;13(1):1218. doi: 10.1186/1471-2458-
13-1218.
79. Cadario R, Chandon P. Which healthy eating nudges work best? A meta-analysis of field
experiments. Marketing Science. 2020. doi: 10.1287/mksc.2018.1128.
80. Shangguan S, Afshin A, Shulkin M, Ma W, Marsden D, Smith J, et al. A Meta-Analysis of Food
Labeling Effects on Consumer Diet Behaviors and Industry Practices. American Journal of
Preventive Medicine. 2019;56(2):300-14. doi: 10.1016/j.amepre.2018.09.024.
81. Crockett RA, King SE, Marteau TM, Prevost AT, Bignardi G, Roberts NW, et al. Nutritional
labelling for healthier food or non-alcoholic drink purchasing and consumption. Cochrane
Database of Systematic Reviews. 2018;2(2):Cd009315. doi: 10.1002/14651858.CD009315.pub2.
82. Harrison S, Jones HE, Martin RM, Lewis SJ, Higgins JPT. The albatross plot: A novel graphical 6
tool for presenting results of diversely reported studies in a systematic review. Research
Synthesis Methods. 2017;8(3):281-9. doi: 10.1002/jrsm.1239.
83. Wilson AL, Buckley E, Buckley JD, Bogomolova S. Nudging healthier food and beverage
choices through salience and priming. Evidence from a systematic review. Food Quality
and Preference. 2016;51:47-64. doi: 10.1016/j.foodqual.2016.02.009.
84. Shavers VL. Measurement of socioeconomic status in health disparities research. Journal
of the National Medical Association. 2007;99(9):1013-23.
85. Maxwell SE, Lau MY, Howard GS. Is psychology suffering from a replication crisis? What
does “failure to replicate” really mean? American Psychologist. 2015;70(6):487.
86. Kosīte D, König LM, De-loyde K, Lee I, Pechey E, Clarke N, et al. Plate size and food
consumption: a pre-registered experimental study in a general population sample.
International Journal of Behavioral Nutrition and Physical Activity. 2019;16(1):75. doi:
10.1186/s12966-019-0826-1.
87. Simons DJ, Holcombe AO, Spellman BA. An Introduction to Registered Replication
Reports at Perspectives on Psychological Science. Perspectives on Psychological Science.
2014;9(5):552-5. doi: 10.1177/1745691614543974.
88. Schmidt AT, Engelen B. The ethics of nudging: An overview. Philosophy Compass.
2020;15(4):e12658.
159
160
CHAPTER 6 – SUPPLEMENTARY FILES
chapter

Additional File 1. PRISMA Checklist

6

Section/topic # Checklist item Reported

on page #
TITLE
Title 1 Identify the report as a systematic review, meta-analysis, or both. Title page
ABSTRACT
Structured summary 2 Provide a structured summary including, as applicable: background; objectives; data sources; p. 5-6
study eligibility criteria, participants, and interventions; study appraisal and synthesis
methods; results; limitations; conclusions and implications of key findings; systematic review
registration number.
INTRODUCTION
Rationale 3 Describe the rationale for the review in the context of what is already known. p. 7-8
Objectives 4 Provide an explicit statement of questions being addressed with reference to participants, p. 8
interventions, comparisons, outcomes, and study design (PICOS).
METHODS
Protocol and registration 5 Indicate if a review protocol exists, if and where it can be accessed (e.g., Web address), and, if p. 9
available, provide registration information including registration number.
Eligibility criteria 6 Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., p. 10-11
years considered, language, publication status) used as criteria for eligibility, giving rationale.
Information sources 7 Describe all information sources (e.g., databases with dates of coverage, contact with study p. 9
authors to identify additional studies) in the search and date last searched.
Search 8 Present full electronic search strategy for at least one database, including any limits used, such Additional
that it could be repeated. File 2
Study selection 9 State the process for selecting studies (i.e., screening, eligibility, included in systematic review, p. 10-11
and, if applicable, included in the meta-analysis).
Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in p. 12
duplicate) and any processes for obtaining and confirming data from investigators.

Additional File 1. PRISMA Checklist (continued)

Section/topic # Checklist item Reported

on page #
Data items 11 List and define all variables for which data were sought (e.g., PICOS, funding sources) and any p. 12-13
assumptions and simplifications made.
Risk of bias in individual studies 12 Describe methods used for assessing risk of bias of individual studies (including specification p. 11
of whether this was done at the study or outcome level), and how this information is to be used
in any data synthesis.
Summary measures 13 State the principal summary measures (e.g., risk ratio, difference in means). p. 12-13
Synthesis of results 14 Describe the methods of handling data and combining results of studies, if done, including p. 12-13
measures of consistency (e.g., I2) for each meta-analysis.
Risk of bias across studies 15 Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., N/A1
publication bias, selective reporting within studies).
Additional analyses 16 Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, meta- N/A 2
regression), if done, indicating which were pre-specified.
RESULTS
Study selection 17 Give numbers of studies screened, assessed for eligibility, and included in the review, with p. 14 +
reasons for exclusions at each stage, ideally with a flow diagram. Fig 13
Study characteristics 18 For each study, present characteristics for which data were extracted (e.g., study size, PICOS, Table 2
follow-up period) and provide the citations.
Risk of bias within studies 19 Present data on risk of bias of each study and, if available, any outcome level assessment (see Table 2
item 12).
Results of individual studies 20 For all outcomes considered (benefits or harms), present, for each study: (a) simple summary Table 24
data for each intervention group (b) effect estimates and confidence intervals, ideally with a
forest plot.
Synthesis of results 21 Present results of each meta-analysis done, including confidence intervals and measures of N/A 5
consistency.
Risk of bias across studies 22 Present results of any assessment of risk of bias across studies (see Item 15). N/A1
Additional analysis 23 Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, meta- N/A 2
regression [see Item 16]).
161
effec ts of nudges in real - life purchasing environments

6
chapter 6 effec ts of nudges in real - life purchasing environments

effect estimates and confidence intervals. 5 Given our inability to perform a meta-analysis, we have adopted a novel approach in which we visualized
Given our inability to perform a meta-analysis, we could not assess the risk of bias across studies, but only at the level of individual studies. 2 In
the present review, we did not perform any additional analyses. 3 Studies were often excluded due failure to meet multiple inclusion criteria. We did
not apply a hierarchy to these reasons for exclusion, and therefore we do not provide data for the N for each reason for exclusion. 4 In line with the
qualitative synthesis of results presented in the harvest plots, our main findings have also been described qualitatively. Therefore, we do not provide
Additional file 2 Search strategy for bibliographic databases
on page #
Reported

Title page
p. 20-22

p. 23-24
PubMed

p. 22
(nudg*[Title/Abstract] OR choice architect*[Title/Abstract] OR environmental
inter vention*[ Title/Abstract] OR environmental change*[ Title/Abstract] OR

Describe sources of funding for the systematic review and other support (e.g., supply of data);
environmental cue*[Title/Abstract] OR behavioral economic*[Title/Abstract]
consider their relevance to key groups (e.g., healthcare providers, users, and policy makers).
Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level (e.g.,
OR behavioural economic*[ Title/Abstract] OR heuristic*[ Title/Abstract] OR
Summarize the main findings including the strength of evidence for each main outcome;

Provide a general interpretation of the results in the context of other evidence, and availability[Title/Abstract] OR assortment[Title/Abstract] OR variety[Title/Abstract]
OR position[Title/Abstract] OR placement[Title/Abstract] OR proximity[Title/Abstract]
OR visibility[Title/Abstract] OR salience[Title/Abstract] OR accessibility[Title/Abstract]
OR convenience[Title/Abstract] OR trolley[Title/Abstract] OR cart[Title/Abstract] OR
tray[Title/Abstract] OR presentation[Title/Abstract] OR packaging[Title/Abstract] OR
portion size[Title/Abstract] OR unit size[Title/Abstract] OR package size[Title/Abstract]
OR nutrition information[Title/Abstract] OR health information[Title/Abstract] OR
incomplete retrieval of identified research, reporting bias).

label*[Title/Abstract] OR prompt*[Title/Abstract] OR social norm*[Title/Abstract] OR

reminder*[Title/Abstract] OR feedback[Title/Abstract] OR default[Title/Abstract]) AND
(diet[Title/Abstract] OR dietary[Title/Abstract] OR food*[Title/Abstract] OR intake[Title/
our findings in harvest plots. These plots are presented in figure 2 up to and including 6.

Abstract] OR purchas*[Title/Abstract] OR meal*[Title/Abstract] OR calori*[Title/

Abstract] OR consum*[Title/Abstract] OR spend*[Title/Abstract] OR choice*[Title/
role of funders for the systematic review.

Abstract] OR sale*[Title/Abstract]) AND (store[Title/Abstract] OR shop[Title/Abstract]

OR supermarket[Title/Abstract] OR market[Title/Abstract] OR retail*[Title/Abstract]
6
implications for future research.

OR lunchroom[Title/Abstract] OR cafeteria[Title/Abstract] OR canteen[Title/Abstract]

OR kiosk[Title/Abstract] OR point of purchase[Title/Abstract] OR fast-food outlet[Title/
Abstract] OR fast-food restaurant[Title/Abstract] OR food-service operat*[Title/
Abstract]) NOT (animals[mh] NOT humans[mh])
Checklist item

Filters: Journal Article; English; Dutch

Additional File 1. PRISMA Checklist (continued)

Embase
(‘nudg*’:ab,ti OR ‘choice architect*’:ab,ti OR ‘environmental intervention*’:ab,ti
OR ‘environmental change*’:ab,ti OR ‘environmental cue*’:ab,ti OR ‘behavioral
25
24

27
26
#

economic*’:ab,ti OR ‘behavioural economic*’:ab,ti OR ‘heuristic*’:ab,ti OR ‘availability’:ab,ti

OR ‘assortment’:ab,ti OR ‘variety’:ab,ti OR ‘position’:ab,ti OR ‘placement’:ab,ti OR
‘proximity’:ab,ti OR ‘visibility’:ab,ti OR ‘salience’:ab,ti OR ‘accessibility’:ab,ti OR
‘convenience’:ab,ti OR ‘trolley’:ab,ti OR ‘cart’:ab,ti OR ‘tray’:ab,ti OR ‘presentation’:ab,ti
Summary of evidence

OR ‘packaging’:ab,ti OR ‘portion size’:ab,ti OR ‘unit size’:ab,ti OR ‘package size’:ab,ti

OR ‘nutrition information’:ab,ti OR ‘health information’:ab,ti OR ‘label*’:ab,ti OR
Section/topic

DISCUSSION

Conclusions
Limitations

‘prompt*’:ab,ti OR ‘social norm*’:ab,ti OR ‘reminder*’:ab,ti OR ‘feedback’:ab,ti OR

FUNDING
Funding

‘default’:ab,ti) AND (‘diet’:ab,ti OR ‘dietary’:ab,ti OR ‘food*’:ab,ti OR ‘intake’:ab,ti OR

‘purchas*’:ab,ti OR ‘meal*’:ab,ti OR ‘calori*’:ab,ti OR ‘consum*’:ab,ti OR ‘spend*’:ab,ti OR
1

162 163
 164
6chapter

PsycINFO

food-service operat*).ab,ti.
NOT ([animals]/lim NOT [humans]/lim) AND ‘article’/it

nutrition information OR health information OR label* OR prompt* OR social norm* OR

OR heuristic* OR availability OR assortment OR variety OR position OR placement OR
(nudg* OR choice architect* OR environmental intervention* OR environmental

(store OR shop OR supermarket OR market OR retail* OR lunchroom OR cafeteria OR

purchas* OR meal* OR calori* OR consum* OR spend* OR choice* OR sale*).ab,ti. AND
reminder* OR feedback OR default).ab,ti. AND (diet OR dietary OR food* OR intake OR
‘market’:ab,ti OR ‘retail*’:ab,ti OR ‘lunchroom’:ab,ti OR ‘cafeteria’:ab,ti OR ‘canteen’:ab,ti

canteen OR kiosk OR point of purchase OR fast-food outlet OR fast-food restaurant OR

change* OR environmental cue* OR behavioral economic* OR behavioural economic*
OR ‘kiosk’:ab,ti OR ‘point of purchase’:ab,ti OR ‘fast-food outlet’:ab,ti OR ‘fast-food

OR tray OR presentation OR packaging OR portion size OR unit size OR package size OR

proximity OR visibility OR salience OR accessibility OR convenience OR trolley OR cart
restaurant’:ab,ti OR ‘food-service operat*’:ab,ti) AND [embase]/lim NOT [medline]/lim
‘choice*’:ab,ti OR ‘sale*’:ab,ti) AND (‘store’:ab,ti OR ‘shop’:ab,ti OR ‘supermarket’:ab,ti OR

Additional file 3. Studies not appropriate for visualisation in harvest plots

Author Year Type of Country Nudge Study size Study Intervention Setting Study Outcome Main finding(s) Quality Reason for
nudge description design duration outcome(s) assessment assessment not being
included in
harvest plot

Roy et al. (1) 2016 Nutrition USA Caloric 1 food outlet Pre-post 1 < month(s) Small food Purchases Point of sale Calorie labelling was Moderate Outcome
information information was ≤6 store of menu system associated with increased difficult to
displayed on items purchases of one relatively categorize
menus and on low-calorie menu option* (healthy vs.
laminated menus and decreased purchases unhealthy)
at the counter. of the highest calorie
menu-option*

Sacks et 2009 Nutrition UK Products were Supermarkets Pre-post > 1 week & ≤ 1 Supermarket Purchases Point of sale Labelling was associated Moderate Outcome
al. (2) information labelled with across the UK month of targeted system with increased sales of difficult to
traffic-light labels products ready meals*; but these categorize
were not more healthy. The (healthy vs.
sales of sandwiches were unhealthy)
unaffected.

Cinciripini 1984 Nutrition USA The caloric value 5,542 Pre-post 1 < month(s) University Purchases Observer Caloric feedback Weak Outcome
et al. (3) information of menu items observations ≤ 6 (both cafeteria of cafeteria reported was associated with difficult to
was displayed interventions) items decreased purchases of categorize
on large red meat, poultry/fish, (healthy vs.
tripods (caloric carbohydrates, and dairy; unhealthy)
feedback). and increased purchases
Subsequently, of fruit and vegetables,
healthy low-fat dairy and salad.
foods were Labelling was associated
identified with with increased purchases
a green triangle of high-fat desserts and
(labelling). red meat, and decreased
purchases of salads, dairy
and carbohydrates.

Papies et 2010 Signage The A poster 156 customers CT N/A Butcher Selection Observer- The healthy recipe Weak Factorial
al. (4) Netherlands announcing a of grilled reported decreased selection of design
recipe for a ‘good chicken grilled chicken among
and slim figure’ restrained but not
was manipulated unrestrained eaters*
(poster present/
absent) in a
butcher while
participants
were exposed to
attractive food
cues (grilled
chicken samples).
6
Additional file 3. Studies not appropriate for visualisation in harvest plots (continued)

Author Year Type of Country Nudge Study size Study Intervention Setting Study Outcome Main finding(s) Quality Reason for
nudge description design duration outcome(s) assessment assessment not being
included in
harvest plot

Wagner et 1988 Signage USA Messages with 1 intervention CT > 1 week & ≤ 1 Fast food Salad bar Point of sale Salad purchases increased Weak No formal
al. (5) ‘health’ prompts store; 1 month restaurant purchases system statistical
were placed at control store analysis
several locations
in the restaurant

Sigurdsson 2014 Mixed Norway Confectionary 1 intervention Pre-post > 1 week & ≤ 1 Supermarket Purchases Point of sale Purchases of healthy Moderate No formal
et al. (6) nudging at the checkout store; 1 month of healthy system targeted check-out foods statistical
intervention was replaced by control store check-out increased; purchases analysis
healthy foods. foods of relocated unhealthy
Additionally, Purchases products decreased
advertisement of
was added, relocated
reinforcing the unhealthy
healthiness foods
of the healthy
check-out foods.

Wisdom et 2010 Mixed USA The menus 638 CT N/A Fast food Total Questionnaire Provision of calorie Weak Factorial
al. (7) nudging varied in a 2 participants restaurant calorie information* and calorie design
intervention (daily calorie intake recommendations* was
recommendation associated with decreased
offered or not) calorie intake; grouping
× 2 (calorie healthy menu options
information together also decreased
for menu items total calorie intake*
shown or not) × 3
(convenience of
healthy options)
design

Rushakoff 2017 Mixed USA Environmental n=233 at Pre-post > 1 year Small food Purchases Questionnaire Increased purchasing Weak No formal
et al. (8) nudging changes included baseline; store and and consumption were statistical
intervention inventory n=211 at reported reported for some healthy analysis
changes, follow-up intake of items (no statistical
installation of foods testing).
point-of-choice
materials (shelf
strips and
cookbook), taste
tests, and display
improvements.

Additional file 3. Studies not appropriate for visualisation in harvest plots (continued)

Author Year Type of Country Nudge Study size Study Intervention Setting Study Outcome Main finding(s) Quality Reason for
nudge description design duration outcome(s) assessment assessment not being
included in
harvest plot

Van Kleef 2012 Mixed The The shelf 291 purchases RCT > 1 week & ≤ 1 Hospital Sales of Hand-counts Predominantly healthy Weak Factorial
et al., study nudging Netherlands arrangement month cafeteria healthy snack assortment design
2 (9) intervention of healthy and structure was associated
snacks (healthy unhealthy with increased purchases
snacks on top snacks of healthy snacks*; shelf
vs. bottom arrangement did not
shelves) and impact on healthy snack
the assortment purchases.
structure (75%
healthy snacks
vs. 25% healthy
snacks) was
manipulated.

*p<0.05

References
1. Roy R, Beattie-Bowers J, Ang SM et al. (2016) The Effect of Energy Labelling on Menus and a Social Marketing Campaign on Food-Purchasing Behaviours of University Students.
BMC Public Health 16, 727. 2. Sacks G, Rayner M, Swinburn B (2009) Impact of front-of-pack ‘traffic-light’ nutrition labelling on consumer food purchases in the UK. Health
Promot Int 24, 344-352. 3. Cinciripini PM (1984) Changing food selections in a public cafeteria: An applied behavior analysis. Behav Modif 8, 520-539. 4. Papies EK, Hamstra P
(2010) Goal priming and eating behavior: enhancing self-regulation by environmental cues. Health Psychol 29, 384-388. 5. Wagner JL, Winett RA (1988) Prompting one low-fat,
high-fiber selection in a fast-food restaurant. J Appl Behav Anal 21, 179-185. 6. Sigurdsson V, Larsen NM, Gunnarsson D (2014) Healthy food products at the point of purchase:
An in-store experimental analysis. J Appl Behav Anal 47, 151-154. 7. Wisdom J, Downs JS, Loewenstein G (2010) Promoting Healthy Choices: Information versus Convenience.
American Economic Journal: Applied Economics 2, 164-178. 8. Rushakoff JA, Zoughbie DE, Bui N et al. (2017) Evaluation of Healthy2Go: A country store transformation project
to improve the food environment and consumer choices in Appalachian Kentucky. Prev Med Rep 7, 187-192.
9. Van Kleef E, Otten K, van Trijp HC (2012) Healthy snacks at the checkout counter: a lab and field study on the impact of shelf arrangement and assortment structure on
consumer choices. BMC Public Health 12, 1072.
6
7
ABSTRACT

Background
Unhealthy food environments may contribute to unhealthy diets and risk of overweight
and obesity through increased consumption of fast-food. Therefore, we aimed to study
the association of relative exposure to fast-food restaurants (FFR) with overall diet quality
and risk of overweight and obesity in a sample of older adults.

Methods
We analyzed cross-sectional data of the EPIC-NL cohort (n=8,231). Data on relative FFR
exposure was obtained through linkage of home address in 2015 with a retail outlet
database. We calculated relative exposure to FFR by dividing the densities of FFR in
street-network buffers of 400, 1000, and 1500m around the home of residence by the
density of all food retailers in the corresponding buffer. We calculated scores on the
Dutch Healthy Diet 2015 (DHD15) index using data from a validated food-frequency
questionnaire. BMI was categorized into normal weight (BMI < 25), overweight (25 ≤ BMI
< 30), and obesity (BMI ≥ 30). We used multivariable linear regression (DHD15-index) and
multinomial logistic regression (weight status), using quartiles of relative FFR exposure
as independent variable, adjusting for lifestyle and environmental characteristics.

Results
Relative FFR exposure was not significantly associated with DHD15-index scores in the
400, 1000, and 1500m buffers (βQ4vsQ1= -0.21 [95%CI: -1.12; 0.70]; βQ4vsQ1= -0.12 [95%CI:
-1.10; 0.87]; βQ4vsQ1= 0.37 [95%CI: -0.67; 1.42], respectively). Relative FFR exposure was
also not related to overweight in consecutive buffers (ORQ4vsQ1=1.10 [95%CI: 0.97; 1.25];

Residential exposure to fast-food ORQ4vsQ1=0.97 [95%CI: 0.84; 1.11]; ORQ4vsQ1= 1.04 [95%CI: 0.90 – 1.20]); estimates for obesity
were similar to those of overweight.
restaurants and its association with diet
quality, overweight and obesity in the Conclusion

Netherlands: a cross-sectional analysis in

A high proportion of FFR around the home of residence was not associated with diet
quality or overweight and obesity in this large Dutch cohort of older adults. We conclude
the EPIC-NL cohort that although the food environment may be a determinant of food choice, this may
not directly translate into effects on diet quality and weight status. Methodological
improvements are warranted to provide more conclusive evidence.
This chapter was published as:
Harbers MC, Beulens JWJ, Boer JM, Karssenberg D, Mackenbach JD, Rutters F, et al. Residential
exposure to fast-food restaurants and its association with diet quality, overweight and obesity
in the Netherlands: a cross-sectional analysis in the EPIC-NL cohort. Nutr J. 2021;20(1):56.
chapter 7 fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y
INTRODUCTION
An unhealthy diet and overweight are important modifiable risk factors for a wide
range of chronic diseases. In 2017, globally 11 million deaths and 255 million disability
adjusted life-years were attributable to poor diets (1). The prevalence of unhealthy
diets and overweight has risen substantially over the past decades. In 2014, more than
half of the European population was overweight (2), and worldwide, a trend towards
increased healthy food consumption between 1990 and 2010 was off-set by an even
larger increase in unhealthy food consumption (3). Causes for these trends are complex
and multifactorial, and relate to factors on both the individual, socio-economic, and
environmental level.
The community food environment has gained recognition as an important environmental
determinant of diet quality and weight status and can be defined as the number, type
and location of food outlets in a certain geographical area (4). In recent decades, food
environments have changed towards increased availability of food outlets offering
relatively less healthy foods. For example, in the Netherlands, the number of out-of-home
eating locations such as food delivery outlets, restaurants, and fast-food restaurants
(FFR) increased by 120%, 35%, and 6%, respectively between 2004 and 2018 while the
number of local shops (e.g., greengrocer, butcher, bakery) decreased (5). Similar changes
in the food environment have also been observed in the UK (6) and US (7).
Out-of-home eating locations such as FFR generally serve foods which are energy-dense
and nutrient-poor, and often serve these in large portions promoting overconsumption
(8). Previously, fast-food consumption has been linked to increased risk of overweight
and obesity (9) and poor diet quality (10-12). In the past decade, the idea has gained
traction that exposure to FFR may influence diet quality and weight status through
increased consumption of fast-food. Despite its plausibility, the evidence base remains
inconclusive, with various studies across the UK, Denmark, the Netherlands and
Australia, associating FFR exposure to poor dietary outcomes such as increased fast-
food (13-15) or meat consumption (16) and higher risk of obesity (14, 16), while others
report null associations for FFR exposure with diet (11, 17, 18) or weight status (19, 20).
These previous studies mainly investigated the association of FFR exposure with dietary
outcomes or weight status in middle-aged populations, while less is known about these
associations in older populations. In high-income countries, approximately 49% of the
total disease burden is attributable to disorders in people aged 60 years old and older,
with diet-related disorders such as cardiovascular disease being the leading contributor
(21). Primary and secondary prevention strategies in older age groups relating to the
170
control of modifiable risk-factors, such as diet, may not only contribute to lower mortality
and morbidity, but may also improve quality of life and physical function and foster
personal independence (22). As approximately 1 out of 10 older adults consumes fast-
food at least 1-2 times a week (23), it is worthwhile to investigate whether FFR exposure
affects diet quality or weight status in this specific population.
Additionally, previous studies mainly investigated absolute measures using counts or
densities (11, 13-15, 17, 18, 20, 24-26) rather than relative measures of FFR exposure, in
which exposure is often defined as the proportion of FFR relative to other food outlets
or restaurants (16, 19, 27, 28). As such, relative measures of FFR capture the availability
of other competing food outlets that may co-locate with FFR, and therefore reflect the
opportunities for (un)healthy food choice. Consequently, a relative measure may better
predict dietary outcomes as compared to absolute measures of FFR availability (29).
The objective of the current study is to study the association between relative exposure
to FFR in relation to overall diet quality, overweight and obesity in a large population of
older adults. We hypothesize that high relative FFR is associated with lower diet quality
and higher risk of overweight or obesity, through providing relatively more opportunities
for unhealthy food purchases.
METHODS
Study population
EPIC-NL is the Dutch contribution to the European Prospective Investigation into Cancer 7
and Nutrition (EPIC), and consists of the MORGEN and Prospect cohorts for which
recruitment took place between 1993 and 1997. The design and rationale EPIC-NL have
been described previously (30). Briefly, the MORGEN cohort includes 22,654 men and
women aged 20-69 years who were randomly sampled from three towns across the
Netherlands (Maastricht, Doetinchem and Amsterdam). Prospect includes 17,357 women
aged 49-70 from Utrecht and vicinity who participated in the national Dutch breast
cancer screening program. The study was conducted in accordance with the Declaration
of Helsinki and the institutional review boards of the University Medical Center Utrecht
and Medical Ethics Committee of TNO Nutrition and Food Research approved the study.
Written consent was obtained from all participants.
In 2015, participants who responded to a follow-up questionnaire on electromagnetic
radiation in 2011, who were still alive and residing in the Netherlands in 2015, and who
provided informed consent (n=13,421 from Prospect and MORGEN Amsterdam and
Maastricht only; the Doetinchem cohort was not invited) were invited to fill out a food
171
chapter 7 fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y
frequency questionnaire (FFQ) for which the response rate was 62.7% (n=8,409). From
this population, we excluded those with extreme energy intake defined as those within
the lower and upper 0.5% of the ratio between energy intake and basal metabolic rate
(n=84). We further excluded participants who did not return the general questionnaire at
follow-up (n=94). Consequently, the population for analysis comprised 8,231 participants.
Exposure assessment – relative FFR exposure
Data on relative FFR exposure was obtained through linkage of the 6-digit postal code
from home address in 2015 with a commercial retail database from Locatus. Locatus
is a commercial company that collects data (e.g., location, type) on food outlets in the
Netherlands based on systematic field audits, which are conducted regularly (e.g.,
once per year for busy shopping areas and once per two to three years for less busy
shopping areas). Additionally, field audits are complemented by surveys and telephone
interviews with retailers, assuring up-to-date data. A recent validation study showed
good to excellent agreement between the Locatus data and research field audits (31).
Although the participants included in this study were originally recruited in Amsterdam,
Utrecht and Maastricht (1993-1997), approximately 40% of study participants included
in the present study still lived in the recruitment areas, whereas the remaining study
participants had moved elsewhere (Additional File 1).
From the Locatus database, we extracted data on food retailers operating in 2015. FFR
were defined as traditional fast-food restaurants, grill-rooms and take-away outlets.
Using PCRaster Python (pcraster.eu), we calculated a relative measure of FFR exposure
in a 400, 1000 and 1500 street-network buffer (e.g., the distance someone can cover
using the street-network) around the home of residence. This was done by dividing the
densities of FFR by the density of all food retailers in the corresponding buffer.
Outcome assessment – diet quality and BMI
In 2015, dietary intake was assessed with a validated semi-quantitative food-frequency
questionnaire, the FFQ-NL 1.0 (32). The FFQ assessed habitual consumption of 160
food items in the preceding year, through questions on consumption frequency and
consumed amounts. Estimated food group intake was validated against an average of 2.7
telephone-based 24-hour recalls. Spearman correlation coefficients between estimates
of the FFQ and estimates from the 24-hour recalls were 0.66 for fruits, 0.61 for bread
and bread products, 0.38 for meat, 0.29 for vegetables, 0.27 for fish, 0.20 for nuts, seeds
and snacks, and 0.13 for legumes. Average daily energy intake was estimated using the
Dutch food composition table from 2011 (33).
172
In order to assess diet quality, we calculated scores on the Dutch Healthy Diet 2015 index
(DHD15-index) (34). Studying overall diet quality in contrast to isolated food groups or
nutrients is nowadays preferred in nutritional epidemiology, as overall dietary patterns
capture synergistic properties of individual foods and are more likely to affect health/
weight status as compared to consumption of individual foods (35). The DHD15-index
reflects adherence to the Dutch dietary guidelines as issued by the Health Council in
2015. The index consists of fifteen food groups for which participants are allocated points
based on absolute intakes of the respective food groups, resulting in a score ranging from
0-150 (Additional file 2). For the present analysis three out of the fifteen original food
groups were excluded in the calculation of the DHD15-score. First, the coffee component
was dropped from the score as the FFQ did not differentiate between types of coffee
(filtered vs. unfiltered). Secondly, we excluded the sodium component from the DHD15-
score, as sodium intake was not reliably captured with the FFQ. Third, we excluded the
alcohol component from the score since we deemed it to be inappropriate in the context
of our research question as alcohol is usually not sold in FFR. Moreover, we did not have
data on type of cereal product (wholegrain vs. refined) except for bread. Therefore, the
scoring of the wholegrain component was based on bread only, with an intake equal
to or above 90 grams receiving the maximum score of 10 points, and a proportionate
decrease in points with decreased intake to the point where participants are assigned
with 0 points when consuming no wholegrain bread. Taken together, the DHD15-score
could range between 0 and 120 with higher scores indicating better adherence to the
dietary guidelines and thus better diet quality.
Lastly, we used Body Mass Index (BMI) as a measure of overweight and obesity. Weight 7
and height were self-reported in the follow-up questionnaire in 2015. BMI was calculated
by dividing the weight by the height squared. Participants were categorized as normal
weight (BMI < 25 kg/m2), overweight (25 ≤ BMI <30), or obese (BMI ≥ 30).
Covariate assessment
At baseline, participants completed a general questionnaire providing data on age, sex
and educational level. Given the high proportion of older women in EPIC-NL, individual
educational level may not be representative of women’s socioeconomic position when
for example their partner is more highly educated. Therefore, we included the highest
attained household educational level of the participant or the partner as a covariate in
our analyses. Household educational level was categorized into low (primary education
or intermediate vocational education), moderate (higher secondary education), and
high (higher vocational education or university). At follow-up, participants provided
information and smoking status, which was was categorized into never, former and
current.
173
chapter 7 fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y
Data on neighborhood level socioeconomic position (SEP) in 2014 and 2016 – based
on neighborhood income, educational level and job status – was obtained from the
Netherlands Institute for Social Research and linked to the address information. This
provided a continuous summary measure of neighborhood SEP for each participant
based on their address, with higher scores indicating higher neighborhood SEP. The
score ranges from approximately -8 to 3. The summary measure of neighborhood
level SEP in 2014 and 2016 was averaged to approximate neighborhood level SEP in
2015. Data on urbanicity of the neighborhood in 2015 was obtained from Statistics
Netherlands, providing a categorical variable for each participant indicating very high
level of urbanisation (≥ 2,500 addresses per km²), high level of urbanisation (1,500 – 2,500
addresses per km²), moderate level of urbanisation (1,000 – 1,500 addresses per km²), low
level of urbanisation (500 – 1,000 addresses per km²), and very low level of urbanisation
(< 500 addresses per km²). Data were linked to the EPIC-NL database through 4-digit
postal codes and GWB-codes (Gemeente Wijk Buurt codes, or City Neighborhood Area
codes) for the neighborhood level SEP and level of urbanisation, respectively.
Statistical analyses
Baseline characteristics are displayed as means with standard deviations (SDs) for
normally distributed variables, medians and interquartile range (IQR) for non-normally
distributed variables, and frequencies and percentages for categorical variables, across
quartiles of relative FFR exposure.
We performed multiple imputation on missing data (n=422 for smoking; n=193 for
level of urbanisation; n=92 for BMI; n=22 for household educational level, n=16 for
neighborhood SEP), using age, sex, cohort, physical activity, smoking status, educational
level, neighborhood SEP, level of urbanisation, DHD-15 score, and BMI as predictor
variables and using 20 imputation sets.
Given the fact that recruitment took place across three cities in the Netherlands, we
tested for a multilevel-structure by including a random intercept and random slope for
recruitment area, with a variance component covariance pattern. The model without
the random slope showed better model fit based on lowest AIC. The Wald statistic for
the random intercept for recruitment area was non-significant (p=0.32), indicating that
accounting for clustering of participants was not necessary. We performed multivariable
linear regression, with quartiles of relative FFR exposure as the independent variable and
DHD15-scores as the dependent variable in order to allow for categorical comparisons
between participants with varying relative FFR exposure. Models were adjusted for
confounding variables based on previous literature. Model 1 was adjusted for age at
follow-up, sex, and cohort. Model 2 was additionally adjusted for smoking, household
174
educational level, and energy intake. Model 3 was additionally adjusted for neighborhood
level SEP and level of urbanisation. Model 4 was additionally adjusted for the total count
of food outlets in the corresponding buffer. We checked the assumptions of linearity
and homoscedasticity by plotting the standardized residuals against standardized
predicted values. The plots indicated that there was no evidence of non-linearity or
heteroscedasticity. We checked the assumption of multicollinearity by examining the
correlation coefficients among predictor variables, and the corresponding variance
inflation factors and tolerance values, which showed no indication of multicollinearity.
We performed multinomial logistic regression for weight status (overweight and
obesity vs. normal weight) as dependent variable, with the lowest quartile of relative
FFR exposure as the reference category. Model structure was similar as to the linear
regression analysis.
We checked effect modification by household educational level, neighborhood SEP, and
level of urbanisation in both the linear and multinomial logistic regression analysis by
including an interaction-term between the continuous variable of relative FFR exposure
and the potential effect-modifier in fully adjusted models. We considered a p-value of <
0.20 to be indicative of possible effect modification. In sensitivity analyses, we excluded
participants who lived at their address for < 1 year and examined the associations in
strata of age. For the analyses on weight status, we also conducted two additional
sensitivity analyses: one using four instead of three BMI categories as outcome variable
(adding an underweight category, defined as BMI <18.5), and another excluding energy
intake as confounder from the model since it might well be an intermediate in the relative 7
FFR exposure – weight status pathway.
Statistical analyses were performed using IBM SPSS Statistics 24 (IBM Analytics, United
States of America, New York). A p-value of <0.05 was considered to be statistically
significant.
RESULTS
Baseline characteristics
The baseline characteristics of the study population across quartiles of relative FFR
exposure in the 400m buffer are presented in Table 1. Baseline characteristics across
quartiles of FFR proportion for the 1000m and 1500m buffer are presented in Additional
file 3 and Additional file 4, respectively. The median proportion of FFR in the 400m
buffer ranged from 0 (IQR: 0 – 0) in the first quartile to 0.29 (IQR: 0.22 – 0.34) in the fourth
quartile; ranges for the 1000 and 1500m buffers were somewhat smaller. On average,
175
chapter 7 fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y
participants were 70 years old (SD = 10) and 80% was female. Moreover, 31.4% had a low
educational level, 6.7% were current smokers, 36.9% was overweight and 12.6% of the
population had obesity. The average score on the DHD15-index was 73 (SD=17).
Association between relative FFR exposure and diet quality
The regression coefficients from the multiple linear regression models for the association
between quartiles of FFR proportion and scores on the DHD15-index are presented
in Table 2. Relative FFR exposure was not significantly associated with DHD15-index
scores when comparing extreme quartiles in the fully adjusted model in the 400m buffer
(β = -0.21, 95% CI: -1.12; 0.70), 1000m buffer (β = -0.12, 95% CI: -1.10; 0.87), or 1500m
buffer (β = 0.37, 95% CI: -0.67; 1.42). Relative FFR exposure was associated with DHD15-
index scores when comparing quartile 2 with quartile 1 in model 1 (β = 1.56, 95% CI:
0.46; 2.66), model 2 (β = 1.46, 95% CI: 0.41; 2.52), and model 3 (β = 1.37, 95% CI: 0.29;
2.45). However, this association attenuated to non-significance after adjustment for
the total presence of food retailers in model 4 (β = 1.10, 95% CI: -0.07; 2.28). Excluding
participants who had lived at their current address for <1 year did not substantially alter
findings (data not shown), and associations were similar in strata of age (data not shown).
Furthermore, there was no significant effect modification by household educational level,
neighborhood SEP, or level of urbanisation (all p-values > 0.20).
Association between relative FFR exposure, overweight and obesity
The odds ratio’s from the multinomial logistic regression for the association between
quartiles of FFR proportion and overweight and obesity are presented in Table 3. Relative
FFR exposure was not significantly associated with overweight in the 400m buffer (1.10;
95% CI: 0.97; 1.25), 1000m buffer (OR = 0.97; 95% CI: 0.84; 1.11) and 1500m (OR = 1.04;
95% CI: 0.90; 1.20) in fully adjusted models when comparing extreme quartiles. Similarly,
relative FFR exposure was not significantly associated with obesity in the 400m buffer
(OR=1.08, 95% CI: 0.90; 1.30), 1000m buffer (OR=0.98, 95% CI: 0.81; 1.20), or 1500m
buffer (OR=1.06, 95% CI: 0.85; 1.31). Excluding participants who had lived at their current
address for <1 year did not substantially alter findings (data not shown), and associations
were similar in strata of age (data not shown). There was no significant effect modification
by educational level, neighborhood SEP, or level of urbanisation (all p-values > 0.20).
Using four categories of weight status as outcome variable and excluding energy intake
as confounder from the model did not materially alter the results (data not shown).
176
DISCUSSION
In the present study, we aimed to study the association of relative FFR exposure with
diet quality, overweight and obesity. We did not confirm our hypothesis that a relatively
unhealthy residential food environment – characterized by a high proportion of FFR
relative to all food outlets – was associated with lower diet quality and higher risk of
overweight or obesity.
Although the evidence base on food environment research is large, substantial
heterogeneity exists in the way FFR exposure (e.g., relative vs. absolute measures)
and diet outcomes (e.g., fast-food purchasing, fast-food consumption, diet quality)
are operationalized. The only other study that examined relative FFR exposure with a
diet quality score was conducted in a sample of younger adults from the US. It found
that relative FFR exposure expressed as the ratio of FFR to total restaurants was over
time associated with lower scores on a diet quality index using instrumental variable
analysis, but was not in ordinary least squares regression models (28). This suggests
that residual confounding attenuates associations between relative FFR exposure and
diet quality in non-causal models. Since our results were based on ordinary regression
models, this may explain the null findings observed in the present study. Three other
studies investigated the association of absolute FFR exposure with diet quality scores,
showing null findings as well (11, 17, 18). However, comparison of studies using different
measures of FFR exposure (e.g., relative vs. absolute) should be done with caution, as
these measures represent different aspects of the FFR environment (36).
7
As for associations of relative FFR exposure with weight status, evidence seems somewhat
more conclusive. Previously, relative FFR exposure expressed as the ratio of FFR to all
food outlets has been associated with higher BMI and higher odds of obesity (in two out
of four buffers studied) in Australia (19) and the UK (16), respectively. Similarly, relative
FFR exposure expressed as the ratio of FFR to total restaurants was associated with
higher BMI and increased odds of obesity in Canada (27). These findings contrast with
the null findings observed in the present study, and may be attributed to the relatively
high age of our study population (on average ~70 years). Indeed, there is evidence that
suggests that consumption of take-away meals declines with age (23), which may be
due to the fact that older adults have less income (37), experience reduced mobility (26)
(e.g., go outside less, so are less susceptible to the tempting food environment) or did
not acquire the habit of obtaining fast-food in their youth as fast-food restaurants were
less proliferate (38). Moreover, other methodological differences (e.g., different model
adjustment and buffer sizes) or different geographical contexts could account for the
contrasting findings.
177
chapter 7 fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y

Besides abovementioned explanations for the contrasting findings of the present study
CONCLUSION
in light of previous evidence, we cannot not preclude the possibility that null associations
with diet quality and weight status are true null findings. For example, a recent study In the present study among predominantly older women we did not find any evidence
by Hobbs et al. challenged the hypothesis that FFR exposure is associated with diet of an association of relative FFR exposure around the home with diet quality or weight
quality and BMI, and alternatively suggested that area-level deprivation rather than FFR status. Although the food environment may be a determinant of food choice, this may
exposure may account for associations that have previously been observed with BMI not directly translate into effects on diet quality and weight status. Methodological
and diet quality through residual confounding (17). Indeed, using a structural equation improvements in accurately assessing FFR exposure and outcomes, and incorporating
modelling approach, Hobbs et al. showed that area-level deprivation was associated behavioural moderators, are warranted to provide more conclusive evidence.
with higher absolute FFR exposure, higher BMI and lower diet quality, whereas FFR
exposure was not associated with diet quality of BMI. Although this study used absolute
FFR exposure rather than relative FFR exposure, it may explain why we were unable to
confirm our initial hypothesis.

The present study should be interpreted in light of several limitations. First, this study
employed a cross-sectional design, which does not allow to draw conclusions on the
temporality of the association. Moreover, weight and height were self-reported which
may have resulted in non-differential outcome misclassification and bias towards the
null. Moreover, we only measured exposure to FFR around the home, which is likely to
provide an underestimation of an individual’s true exposure to FFR (14). However, given
the older population in the present study, we might expect that activity patterns will
generally be centered around the home to a greater extent as compared to younger
populations (39). Also, it has been hypothesized that individuals may differ in the
extent to which they are susceptible to the food environment (18, 25). However, we
lacked data on behavioral constructs that may moderate the association between the 7
food environment, diet and weight status. Additionally, participants in cohort studies
generally represent a more health-conscious population. Given the fact that we only
included participants that took part in the data-collection at follow-up (18-22 years after
inclusion in the cohort), we cannot exclude the potential of healthy volunteer bias. As a
result, our study population may have been generally health-conscious and not likely to
obtain FFR, leading to our null associations. Last, although we adjusted for a wide range
of potential confounders, we cannot exclude possibility of residual confounding. For
example, we used educational level as a proxy of SEP and did not have data on income,
which may be an important confounder as well.

Strengths of the present study include the use of a validated FFQ which allowed us
to compute a comprehensive measure of diet quality, opposed to short screeners
used often in food environment research (40). Lastly, the study population included
participants from various areas in the Netherlands, providing a geographically diverse
sample and assuring variation in relative FFR exposure.

178 179
 Table 1. Participant characteristics across quartiles of FFR proportion in the 400m buffera.

180
Q1 Q2 Q3 Q4
7 chapter

N, (%) 2871 (34.9) 1240 (15.1) 2063 (25.1) 2057 (25.0)

Median FFR proportion 0.00 (0.00 – 0.00) 0.07 (0.06 - 0.09) 0.14 (0.13 - 0.17) 0.29 (0.22 - 0.34)
Age, y 70 ± 10 70 ± 11 71 ± 10 70 ± 10
Sex, n (%)
Male 539 (18.8) 239 (19.3) 437 (21.2) 404 (19.6)
Female 2332 (81.2) 1001 (80.7) 1626 (78.8) 1653 (80.4)
Household educational level, n (%)
Low 901 (31.5) 408 (33.0) 653 (31.7) 622 (30.3)
Moderate 730 (25.5) 283 (22.9) 482 (23.4) 515 (25.1)
High 1229 (43.0) 547 (44.2) 923 (44.8) 916 (44.6)
Smoking, n (%)
Current 192 (7.0) 86 (7.3) 145 (7.4) 125 (6.4)
Former 1290 (47.4) 534 (45.6) 983 (50.2) 961 (49.2)
Never 1242 (45.6) 552 (47.1) 831 (42.4) 868 (44.4)
BMI, kg/m2 25.5 ± 4.2 25.5± 4.4 25.5± 4.3 25.6 ± 4.1
Weight status
Normal weight, n (%) 1434 (50.4) 612 (49.9) 1035 (50.8) 990 (48.7)
Overweight, n (%) 1049 (36.9) 447 (36.5) 759 (37.2) 780 (38.4)
Obesity, n (%) 360 (12.7) 167 (13.6) 244 (12.0) 262 (12.9)
Kcal/d 1,891 ± 634 1,901 ± 661 1,894 ± 627 1,888 ± 647

Table 1. Participant characteristics across quartiles of FFR proportion in the 400m buffera. (continued)

Q1 Q2 Q3 Q4
DHD-15 food groups, g/day
Vegetables 117 (69 - 171) 124 (76 - 181) 118 (68 - 171) 119 (69 - 174)
Fruit 165 (77 - 234) 187 (89 - 241) 156 (70 - 237) 149 (71 - 235)
Wholegrain bread 71 (22 - 107) 70 (21 - 106) 70 (19 - 106) 70 (22 - 106)
Legumes 6 (0 - 16) 6 (0 - 17) 9 (0 - 17) 6 (0 - 16)
Nuts 6 (1 - 19) 7 (1 - 21) 6 (1 - 20) 7 (1 - 20)
Dairy 266 (141 - 402) 268 (149 - 399) 263 (145 - 406) 254 (136 - 392)
Fish 14 (7 - 29) 15 (7 - 36) 14 (5 - 29) 14 (7 - 36)
Tea 340 (121 - 510) 340 (146 - 510) 340 (121 - 510) 340 (97 - 680)
Butter and solid fats 0 (0 - 5) 0 (0 - 6) 0 (0 - 6) 0 (0 - 6)
Oils and diet margarines 12 (3 - 32) 10 (3 - 31) 11 (3 - 30) 11 (3 - 30)
Red meat 40 (19 - 74) 39 (19 - 74) 39 (17 - 76) 41 (18 - 77)
Processed meat 25 (10 - 45) 23 (8 - 41) 23 (8 - 42) 22 (9 - 41)
Sweetened beverages and fruit juices 75 (11 - 176) 58 (6 - 182) 71 (7 - 176) 63 (7 - 175)
Neighbourhood socioeconomic statusb 0.4 (-0.3; 0.9) 0.3 (-0.6; 1.1) 0.4 (-0.6; 1.1) 0.4 (-0.8; 1.1)
Level of urbanisationc, n (%)
Very low level of urbanisation 593 (21.0) 456 (39.2) 839 (41.3) 546 (27.1)
Low level or urbanisation 924 (32.7) 273 (23.5) 609 (30.0) 723 (35.8)
Moderate level of urbanisation 683 (24.2) 184 (15.8) 226 (11.1) 408 (20.2)
High level of urbanisation 260 (9.2) 187 (16.1) 211 (10.4) 196 (9.7)
Very high level or urbanisation 366 (13.0) 63 (5.4) 146 (7.2) 145 (7.2)
a b
Continuous variables are presented as means (standard deviation) or as medians (p25 – p75). Higher scores represent higher neighbourhood
socioeconomic status. cVery low level of urbanisation ≤ 500 addresses/km2; low level of urbanisation = 500-1000 addresses/ km2; moderate level of
181
fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y

urbanisation = 1000-1500 addresses/ km2; high level of urbanisation =1500-2000 addresses/ km2; very high level or urbanisation ≥ 2000 addresses/
km2. The following variables had missing data: smoking status (n=422); level of urbanisation (n=193); BMI (n=92); household educational level (n=22);
neighbourhood socioeconomic status (n=16).
7
 Table 2. Regression coefficients (95%CI) for the association between quartiles of FFR proportion and DHD15-index scores.

182
Q1 Q2 Q3 Q4
7 chapter

β (95% CI) β (95% CI) β (95% CI) β (95% CI)

400m buffer
N, (%) 2871 (34.9) 1240 (15.1) 2063 (25.1) 2057 (25.0)
Median FFR (IQR) 0.00 (0.00 – 0.00) 0.07 (0.06 - 0.09) 0.14 (0.13 - 0.17) 0.29 (0.22 - 0.34)
a
Model 1 ref 1.56 (0.46; 2.66)* 0.43 (-0.51; 1.36) 0.10 (-0.83; 1.04)
Model 2b ref 1.46 (0.41; 2.52)* 0.28 (-0.62; 1.18) -0.05 (-0.94; 0.85)
Model 3c ref 1.37 (0.29; 2.45)* 0.05 (-0.87; 0.96) -0.15 (-1.05; 0.75)
Model 4 d ref 1.10 (-0.07; 2.28) -0.22 (-1.24; 0.81) -0.21 (-1.12; 0.70)
1000m buffer
N, (%) 2057 (25.0) 2057 (25.0) 2044 (24.8) 2073 (25.2)
Median FFR (IQR) 0.00 (0.00 - 0.09) 0.13 (0.11 - 0.13) 0.16 (0.15 - 0.17) 0.21 (0.20 - 0.25)
Model 1a ref -0.22 (-1.22; 0.79) 0.38 (-0.63; 1.39) -0.08 (-1.09; 0.93)
Model 2b ref -0.21 (-1.18; 0.76) 0.10 (-0.87; 1.07) -0.15 (-1.12; 0.82)
c
Model 3 ref -0.48 (-1.45; 0.49) -0.27 (-1.25; 0.71) -0.34 (-1.31; 0.63)
d
Model 4 ref -0.82 (-1.82; 0.18) -0.16 (-1.14; 0.82) -0.12 (-1.10; 0.87)
1500m buffer
N, (%) 2057 (25.0) 2060 (25.0) 2128 (25.9) 1986 (24.1)
Median FFR (IQR) 0.10 (0.00 - 0.10) 0.13 (0.12 - 0.13) 0.17 (0.16 - 0.19) 0.22 (0.20 - 0.24)
Model 1a ref -0.38 (-1.39; 0.63) -0.56 (-1.55; 0.44) 0.16 (-0.85; 1.18)
b
Model 2 ref -0.50 (-1.47; 0.47) -0.53 (-1.49; 0.43) -0.07 (-1.05; 0.91)
Model 3c ref -0.70 (-1.68; 0.27) -0.66 (-1.63; 0.30) -0.18 (-1.16; 0.81)
d
Model 4 ref -0.75 (-1.72; 0.23) -0.27 (-1.27; 0.73) 0.37 (-0.67; 1.42)
a b
Model 1 is adjusted for age, sex, and cohort. Model 2 is additionally adjusted for smoking, energy intake, and household educational level. cModel 3 is
additionally adjusted for neighbourhood level socioeconomic status and level of urbanisation. dModel 4 is additionally adjusted for the total presence
of food retailers in the corresponding buffer. *Indicates statistical significance.

Table 3. Odds ratios (95%CI) for the association between quartiles of FFR proportion and risk of overweight and obesity.

400m buffer 1000m buffer 1400m buffer

Odds of being Odds of being Odds of being Odds of being Odds of being Odds of being
overweight obese overweight obese overweight obese
Model 1a
Q1 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Q2 0.99 (0.86; 1.15) 1.08 (0.88; 1.33) 0.89 (0.78; 1.02) 0.88 (0.73; 1.07) 1.08 (0.95; 1.24) 1.04 (0.86; 1.27)
Q3 0.97 (0.86; 1.10) 0.92 (0.77; 1.11) 0.91 (0.80; 1.04) 0.86 (0.71; 1.04) 1.06 (0.92; 1.21) 1.04 (0.86; 1.26)
Q4 1.07 (0.95; 1.21) 1.05 (0.88; 1.26) 0.97 (0.85; 1.11) 0.99 (0.82; 1.20) 1.07 (0.93; 1.22) 1.06 (0.87; 1.29)
Model 2b
Q1 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Q2 1.00 (0.86; 1.17) 1.10 (0.89; 1.36) 0.89 (0.77; 1.02) 0.88 (0.72; 1.07) 1.08 (0.94; 1.24) 1.03 (0.85; 1.26)
Q3 0.99 (0.87; 1.12) 0.94 (0.78; 1.13) 0.92 (0.81; 1.06) 0.88 (0.72; 1.07) 1.05 (0.91; 1.20) 1.03 (0.85; 1.25)
Q4 1.09 (0.96; 1.23) 1.07 (0.89; 1.29) 0.96 (0.84; 1.10) 0.97 (0.80; 1.18) 1.06 (0.93; 1.22) 1.05 (0.86; 1.29)
c
Model 3
Q1 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Q2 1.01 (0.87; 1.17) 1.12 (0.90; 1.39) 0.90 (0.78; 1.03) 0.89 (0.73; 1.09) 1.08 (0.94; 1.24) 1.02 (0.84; 1.25)
Q3 1.00 (0.88; 1.13) 0.96 (0.80; 1.16) 0.95 (0.83; 1.09) 0.91 (0.74; 1.11) 1.04 (0.91; 1.20) 1.01 (0.83; 1.23)
Q4 1.09 (0.96; 1.23) 1.06 (0.88; 1.27) 0.98 (0.86; 1.12) 0.99 (0.81; 1.20) 1.08 (0.94; 1.24) 1.07 (0.87; 1.30)
d
Model 4
Q1 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Q2 1.08 (0.91; 1.27) 1.19 (0.94; 1.51) 0.92 (0.79; 1.06) 0.89 (0.73; 1.10) 1.08 (0.94; 1.24) 1.02 (0.84; 1.25)
Q3 1.06 (0.92; 1.23) 1.02 (0.83; 1.26) 0.95 (0.83; 1.09) 0.91 (0.74; 1.11) 1.02 (0.88; 1.17) 1.00 (0.82; 1.22)
Q4 1.10 (0.97; 1.25) 1.08 (0.90; 1.30) 0.97 (0.84; 1.11) 0.98 (0.81; 1.20) 1.04 (0.90; 1.20) 1.06 (0.85; 1.31)
a b
Model 1 is adjusted for age, sex, and cohort. Model 2 is additionally adjusted for smoking, energy intake, and household educational level. cModel 3 is
183
fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y

additionally adjusted for neighbourhood level socioeconomic status and level of urbanisation. dModel 4 is additionally adjusted for the total presence
of food retailers in the corresponding buffer
7
chapter 7 fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y
REFERENCES
1. GBD 2017 Diet Collaborators. Health effects of dietary risks in 195 countries, 1990-
2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet.
2019;393(10184):1958-72.
2. Marques A, Peralta M, Naia A, Loureiro N, de Matos MG. Prevalence of adult overweight
and obesity in 20 European countries, 2014. Eur J Public Health. 2018;28(2):295-300.
3. Imamura F, Micha R, Khatibzadeh S, Fahimi S, Shi P, Powles J, et al. Dietary quality among
men and women in 187 countries in 1990 and 2010: a systematic assessment. Lancet Glob
Health. 2015;3(3):e132-42.
4. Glanz K, Sallis JF, Saelens BE, Frank LD. Healthy nutrition environments: concepts and
measures. Am J Health Promot. 2005;19(5):330-3, ii.
5. Pinho MGM, Mackenbach JD, den Braver NR, Beulens JJW, Brug J, Lakerveld J. Recent
changes in the Dutch foodscape: socioeconomic and urban-rural differences. Int J Behav
Nutr Phys Act. 2020;17(1):43.
6. Maguire ER, Burgoine T, Monsivais P. Area deprivation and the food environment over time:
A repeated cross-sectional study on takeaway outlet density and supermarket presence
in Norfolk, UK, 1990-2008. Health Place. 2015;33:142-7.
7. James P, Seward MW, James O’Malley A, Subramanian SV, Block JP. Changes in the food
environment over time: examining 40 years of data in the Framingham Heart Study.
International Journal of Behavioral Nutrition and Physical Activity. 2017;14(1):84.
8. Jaworowska A, Blackham T, Davies IG, Stevenson L. Nutritional challenges and health
implications of takeaway and fast food. Nutrition Reviews. 2013;71(5):310-8.
9. Rosenheck R. Fast food consumption and increased caloric intake: a systematic review of
a trajectory towards weight gain and obesity risk. Obes Rev. 2008;9(6):535-47.
10. Barnes TL, French SA, Mitchell NR, Wolfson J. Fast-food consumption, diet quality and body
weight: cross-sectional and prospective associations in a community sample of working
adults. Public Health Nutr. 2016;19(5):885-92.
11. Moore LV, Diez Roux AV, Nettleton JA, Jacobs DR, Franco M. Fast-food consumption, diet
quality, and neighborhood exposure to fast food: the multi-ethnic study of atherosclerosis.
Am J Epidemiol. 2009;170(1):29-36.
12. Schroder H, Fito M, Covas MI. Association of fast food consumption with energy intake,
diet quality, body mass index and the risk of obesity in a representative Mediterranean
population. Br J Nutr. 2007;98(6):1274-80.
13. Bernsdorf KA, Lau CJ, Andreasen AH, Toft U, Lykke M, Glümer C. Accessibility of fast food
outlets is associated with fast food intake. A study in the Capital Region of Denmark. Health
& Place. 2017;48:102-10.
14. Burgoine T, Forouhi NG, Griffin SJ, Wareham NJ, Monsivais P. Associations between
exposure to takeaway food outlets, takeaway food consumption, and body weight in
Cambridgeshire, UK: population based, cross sectional study. BMJ : British Medical Journal.
2014;348:g1464.
15. Burgoine T, Forouhi NG, Griffin SJ, Brage S, Wareham NJ, Monsivais P. Does neighborhood
fast-food outlet exposure amplify inequalities in diet and obesity? A cross-sectional study.
Am J Clin Nutr. 2016;103(6):1540-7.
184
16. Burgoine T, Sarkar C, Webster CJ, Monsivais P. Examining the interaction of fast-food outlet
exposure and income on diet and obesity: evidence from 51,361 UK Biobank participants.
Int J Behav Nutr Phys Act. 2018;15(1):71-.
17. Hobbs M, Green M, Roberts K, Griffiths C, McKenna J. Reconsidering the relationship
between fast-food outlets, area-level deprivation, diet quality and body mass index: an
exploratory structural equation modelling approach. J Epidemiol Community Health.
2019;73(9):861-6.
18. Mackenbach JD, Lakerveld J, Generaal E, Gibson-Smith D, Penninx B, Beulens JWJ. Local
fast-food environment, diet and blood pressure: the moderating role of mastery. Eur J
Nutr. 2019;58(8):3129-34.
19. Feng X, Astell-Burt T, Badland H, Mavoa S, Giles-Corti B. Modest ratios of fast food
outlets to supermarkets and green grocers are associated with higher body mass index:
Longitudinal analysis of a sample of 15,229 Australians aged 45 years and older in the
Australian National Liveability Study. Health Place. 2018;49:101-10.
20. Lamb KE, Thornton LE, Olstad DL, Cerin E, Ball K. Associations between major chain fast-
food outlet availability and change in body mass index: a longitudinal observational study
of women from Victoria, Australia. BMJ Open. 2017;7(10):e016594.
21. Prince MJ, Wu F, Guo Y, Gutierrez Robledo LM, O’Donnell M, Sullivan R, et al. The burden
of disease in older people and implications for health policy and practice. Lancet.
2015;385(9967):549-62.
22. Fleg Jerome L, Forman Daniel E, Berra K, Bittner V, Blumenthal James A, Chen Michael A,
et al. Secondary Prevention of Atherosclerotic Cardiovascular Disease in Older Adults.
Circulation. 2013;128(22):2422-46.
23. Adams J, Goffe L, Brown T, Lake AA, Summerbell C, White M, et al. Frequency and socio-
demographic correlates of eating meals out and take-away meals at home: cross-sectional
analysis of the UK national diet and nutrition survey, waves 1–4 (2008–12). International
Journal of Behavioral Nutrition and Physical Activity. 2015;12(1):51.
24. Boone-Heinonen J, Gordon-Larsen P, Kiefe CI, Shikany JM, Lewis CE, Popkin BM. Fast food
restaurants and food stores: longitudinal associations with diet in young to middle-aged 7
adults: the CARDIA study. Arch Intern Med. 2011;171(13):1162-70.
25. Paquet C, Daniel M, Knauper B, Gauvin L, Kestens Y, Dube L. Interactive effects of reward
sensitivity and residential fast-food restaurant exposure on fast-food consumption. Am
J Clin Nutr. 2010;91(3):771-6.
26. Hobbs M, Griffiths C, Green MA, Jordan H, Saunders J, Christensen A, et al. Fast-food
outlet availability and obesity: Considering variation by age and methodological diversity
in 22,889 Yorkshire Health Study participants. Spatial and Spatio-temporal Epidemiology.
2019;28:43-53.
27. Polsky JY, Moineddin R, Dunn JR, Glazier RH, Booth GL. Absolute and relative densities
of fast-food versus other restaurants in relation to weight status: Does restaurant mix
matter? Preventive Medicine. 2016;82:28-34.
28. Rummo PE, Guilkey DK, Ng SW, Meyer KA, Popkin BM, Reis JP, et al. Understanding bias
in relationships between the food environment and diet quality: the Coronary Artery
Risk Development in Young Adults (CARDIA) study. J Epidemiol Community Health.
2017;71(12):1185-90.
29. Cummins S, Clary C, Shareck M. Enduring challenges in estimating the effect of the food
environment on obesity. Am J Clin Nutr. 2017;106(2):445-6.
185
chapter 7 fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y

30. Beulens JWJ, Monninkhof EM, Verschuren WMM, Schouw YTvd, Smit J, Ocke MC, et al.
Cohort Profile: The EPIC-NL study. International Journal of Epidemiology. 2009;39(5):1170-8. CHAPTER 7 – SUPPLEMENTARY FILES
31. Canalia C, Pinho MGM, Lakerveld J, Mackenbach JD. Field Validation of Commercially
Available Food Retailer Data in the Netherlands. Int J Environ Res Public Health. 2020;17(6). Additional file 1. Distribution of home locations in 2015 in the study population.
32. Sluik D, Geelen A, de Vries JH, Eussen SJ, Brants HA, Meijboom S, et al. A national FFQ for
the Netherlands (the FFQ-NL 1.0): validation of a comprehensive FFQ for adults. Br J Nutr.
2016;116(5):913-23.
33. Dutch Food Composition Table (NEVO) [Internet]. RIVM/Dutch Nutrition Centre. 2011.
34. Looman M, Feskens EJ, de Rijk M, Meijboom S, Biesbroek S, Temme EH, et al. Development
and evaluation of the Dutch Healthy Diet index 2015. Public Health Nutr. 2017;20(13):2289-99.
35. Hu FB. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin
Lipidol. 2002;13(1):3-9.
36. Pinho MGM, Mackenbach JD, Oppert JM, Charreire H, Bardos H, Rutter H, et al. Exploring
absolute and relative measures of exposure to food environments in relation to dietary
patterns among European adults. Public Health Nutr. 2019;22(6):1037-47.
37. OECD. Incomes of older people. 2017.
38. Janssen HG, Davies IG, Richardson LD, Stevenson L. Determinants of takeaway and fast
food consumption: a narrative review. Nutr Res Rev. 2018;31(1):16-34.
39. Al Snih S, Peek KM, Sawyer P, Markides KS, Allman RM, Ottenbacher KJ. Life-space mobility
in Mexican Americans aged 75 and older. J Am Geriatr Soc. 2012;60(3):532-7.
40. Kirkpatrick SI, Reedy J, Butler EN, Dodd KW, Subar AF, Thompson FE, et al. Dietary
assessment in food environment research: a systematic review. Am J Prev Med.
2014;46(1):94-102.

186 187
 Additional file 2. Components of the DHD15-index and corresponding dietary recommendations and their threshold (minimum score) and cut-off
(maximum score) values.

188
DHD15-component Dietary recommendations Minimum score (0 points) Maximum score (10 points)
7 chapter

1. Vegetables Eat at least 200 g of vegetables daily 0 g/d ≥ 200 g/d

2. Fruit Eat at least 200 g of fruit daily 0 g/d ≥ 200 g/d
a
3. Whole grain products a. Eat at least 90 g of wholegrain products daily 0 g/d ≥ 90 g/d
b. Replace refined cereal products by whole- No consumption of whole grain No consumption of refined
grain products products OR ratio whole grain to products OR ratio of whole
refined grain ≤ 0.7 grains to refined grains ≥ 11
4. Legumes Eat legumes weekly. 0 g/d ≥10 g/d
5. Nuts Eat at least 15 grams of unsalted nuts a day. 0 g/d ≥15 g/d
b
6. Dairy Eat a few portions of dairy produce daily, 0 g/d OR ≥750 g/d 300–450 g/d
including milk or yoghurt.
7. Fishc Eat one serving of fish weekly, preferably oily 0 g/d ≥15 g/d
fish.
8. Tea Drink three cups of black or green tea a day 0 g/d ≥450 g/d
9. Fats and oils Replace butter, hard margarines, and cooking No consumption of soft No consumption of butter, hard
fats by soft margarines, liquid cooking fats, and margarines, liquid cooking fats margarines and cooking fats
vegetable oils and vegetable oils OR ratio of OR ratio of liquid cooking fats
liquid cooking fats to solid cooking to solid cooking fats ≥ 13
fats ≤ 0.6
10. Coffeed Replace unfiltered coffee by filtered coffee. Any consumption of unfiltered Consumption of only
coffee filtered coffee OR no coffee
consumption
11. Red meat Limit consumption of red meat. ≥ 100 g/d ≤45 g/d
12. Processed meat Limit consumption of processed meat. ≥ 50 g/d 0 g/d
13. Sugar-sweetened Limit consumption of sweetened beverages and ≥ 250 g/d 0 g/d
beverages and fruit juices fruit juices.
14. Alcohol If alcohol is consumed at all, intake should be Women: ≥20 g ethanol/d Women: ≤10 g ethanol/d
limited to one Dutch unit (10 gram ethanol) daily Men: ≥30 g ethanol/d Men: ≤10 g ethanol/d
15. Sodiume Limit consumption of table salt to 6 g daily ≥ 3.8 g sodium/d ≤ 1.9 g sodium/d
a
The wholegrain component comprises two sub-components of which each sub-component has a maximum score of 5 points; bFor the dairy component,
a maximum of 40 g cheese can be included. cFor the fish component, a maximum of 4 g lean fish can be included. dThe coffee component was not
included in the calculation of the DHD15-score as no data was available on type of coffee (filtered/unfiltered). eSodium only originated from foods;
intake from added salt was not captured by the FFQ.

Additional file 3. Participant characteristics across quartiles of FFR proportion in the 1000m buffera.

Q1 Q2 Q3 Q4
N, (%) 2057 (25.0) 2057 (25.0) 2044 (24.8) 2073 (25.2)
Median FFR proportion 0.00 (0.00 - 0.09) 0.13 (0.11 - 0.13) 0.16 (0.15 - 0.17) 0.21 (0.20 - 0.25)
Age, y 70 ± 10 70 ± 10 69 ± 10 70 ± 10
Sex, n (%)
Male 398 (19.3) 392 (19.1) 430 (21.0) 399 (19.2)
Female 1659 (80.7) 1665 (80.9) 1614 (79.0) 1674 (80.8)
Household educational level, n (%)
Low 654 (31.9) 671 (32.7) 594 (29.1) 665 (32.2)
Moderate 514 (25.0) 495 (24.1) 492 (24.1) 509 (24.7)
High 884 (43.1) 886 (43.2) 955 (46.8) 890 (43.1)
Smoking, n (%)
Current 152 (7.8) 155 (8.0) 130 (6.7) 111 (5.6)
Former 923 (47.4) 922 (47.4) 955 (48.9) 968 (49.2)
Never 871 (44.8) 868 (44.6) 866 (44.4) 888 (45.1)
BMI, kg/m2 25.6 ± 4.2 25.5 ± 4.3 25.4 ± 4.1 25.6 ± 4.2
Weight status
Normal weight, n (%) 987 (48.5) 1040 (51.3) 1035 (51.2) 1009 (49.1)
Overweight, n (%) 781 (38.3) 735 (36.3) 745 (36.9) 774 (37.7)
Obesity, n (%) 269 (13.2) 251 (12.4) 241 (11.9) 272 (13.2)
Kcal/d 1,901 ± 646 1,885 ± 639 1,904 ± 639 1,881 ± 634
189
fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y

7
 Additional file 3. Participant characteristics across quartiles of FFR proportion in the 1000m buffera. (continued)

190
Q1 Q2 Q3 Q4
chapter

DHD-15 food groups, g/day

7

Vegetables 120 (73 - 171) 116 (67 - 172) 121 (71 - 177) 117 (69 - 172)
Fruit 176 (81 - 237) 163 (77 - 236) 151 (71 - 234) 154 (74 - 236)
Wholegrain bread 70 (24 - 106) 70 (18 - 106) 70 (22 - 109) 70 (23 - 106)
Legumes 6 (0 - 17) 6 (0 - 17) 6 (0 - 17) 6 (0 - 17)
Nuts 6 (1 - 21) 6 (1 - 19) 7 (1 - 20) 7 (1 - 20)
Dairy 266 (142 - 399) 262 (139 - 397) 258 (143 - 403) 263 (143 - 402)
Fish 17 (7 - 36) 14 (7 - 29) 14 (7 - 29) 14 (7 - 36)
Tea 340 (121 - 510) 340 (121 - 510) 340 (121 - 680) 340 (146 - 680)
Butter and solid fats 0 (0 - 5) 0 (0 - 6) 0 (0 - 6) 0 (0 - 6)
Oils and diet margarines 11 (3 - 31) 11 (3 - 31) 11 (3 - 31) 12 (3 - 30)
Red meat 42 (19 - 75) 40 (18 - 76) 39 (17 - 74) 40 (18 - 75)
Processed meat 25 (10 - 44) 23 (8 - 42) 22 (9 - 42) 23 (9 - 41)
Sweetened beverages and fruit juices 68 (9 - 179) 57 (6 - 175) 64 (7 - 175) 75 (9 - 182)
Alcohol 9 (1 - 22) 9 (0 - 22) 10 (1 - 22) 8 (1 - 20)
b
Neighbourhood socioeconomic status 0.2 (-0.5; 0.9) 0.2 (-0.7; 1.1) 0.4 (-0.3; 1.1) 0.4 (-0.4; 1.1)
Level of urbanizationc, n (%)
Very low level of urbanisation 506 (25.7) 753 (37.0) 706 (35.4) 469 (23.0)
Low level or urbanisation 607 (30.9) 580 (28.5) 648 (32.5) 694 (34.0)
Moderate level of urbanisation 375 (19.1) 305 (15.0) 318 (15.9) 503 (24.6)
High level of urbanisation 242 (12.3) 256 (12.6) 200 (10.0) 156 (7.6)
Very high level or urbanisation 236 (12.0) 141 (6.9) 124 (6.2) 219 (10.7)
a
Continuous variables are presented as means (standard deviation) or as medians (p25 – p75). bHigher scores represent higher neighbourhood
socioeconomic status. cVery low level of urbanisation ≤ 500 addresses/km2; low level of urbanisation = 500-1000 addresses/ km2; moderate level of
urbanisation = 1000-1500 addresses/ km2; high level of urbanisation =1500-2000 addresses/ km2; very high level or urbanisation ≥ 2000 addresses/
km2. The following variables had missing data: smoking status (n=422); level of urbanisation (n=193); BMI (n=92); household educational level (n=22);
neighbourhood socioeconomic status (n=16).

Additional file 4. Participant characteristics across quartiles of FFR proportion in the 1500m buffera.

Q1 Q2 Q3 Q4
N, (%) 2057 (25.0) 2060 (25.0) 2128 (25.9) 1986 (24.1)
Median FFR proportion 0.10 (0.00 - 0.10) 0.13 (0.12 - 0.13) 0.17 (0.16 - 0.19) 0.22 (0.20 - 0.24)
Age, y 70 ± 10 71 ± 10 69 ± 10 70 ± 10
Sex, n (%)
Male 367 (17.8) 445 (21.6) 442 (20.8) 365 (18.4)
Female 1690 (82.2) 1615 (78.4) 1686 (79.2) 1621 (81.6)
Household educational level, n (%)
Low 639 (31.2) 681 (33.1) 666 (31.4) 598 (30.2)
Moderate 507 (24.7) 481 (23.4) 534 (25.2) 488 (24.6)
High 905 (44.1) 895 (43.5) 921 (43.4) 894 (45.2)
Smoking, n (%)
Current 158 (8.1) 136 (7.0) 149 (7.3) 105 (5.6)
Former 921 (47.0) 948 (49.0) 965 (47.6) 934 (49.5)
Never 879 (44.9) 851 (44.0) 915 (45.1) 848 (44.9)
BMI, kg/m2 25.4 ± 4.3 25.7 ± 4.2 25.6 ± 4.3 25.5 ± 4.2
Weight status
Normal weight, n (%) 1042 (51.3) 994 (49.0) 1055 (50.0) 980 (49.8)
Overweight, n (%) 734 (36.1) 780 (38.4) 788 (37.3) 733 (37.3)
Obesity, n (%) 255 (12.6) 256 (12.6) 269 (12.7) 253 (12.9)
Kcal/d 1,876 ± 637 1,913 ± 654 1,897 ± 632 1,884 ± 634
191
fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y

7
 Additional file 4. Participant characteristics across quartiles of FFR proportion in the 1500m buffera. (continued)

192
Q1 Q2 Q3 Q4
DHD-15 food groups, g/day
7 chapter

Vegetables 119 (72 - 173) 119 (66 - 173) 118 (71 - 172) 120 (71 - 173)
Fruit 165 (77 - 236) 163 (77 - 236) 155 (71 - 236) 161 (79 - 235)
Wholegrain bread 70 (21 - 106) 70 (18 - 106) 70 (19 - 106) 71 (29 - 106)
Legumes 6 (0 - 17) 6 (0 - 17) 9 (0 - 17) 6 (0 - 17)
Nuts 6 (1 - 20) 6 (1 - 20) 6 (1 - 19) 7 (2 - 21)
Dairy 269 (138 - 400) 258 (144 - 397) 255 (142 - 401) 268 (145 - 401)
Fish 18 (7 - 36) 14 (7 - 29) 14 (5 - 32) 14 (7 - 29)
Tea 340 (121 - 510) 340 (146 - 583) 340 (109 - 510) 340 (121 - 680)
Butter and solid fats 0 (0 - 5) 0 (0 - 7) 0 (0 - 5) 0 (0 - 6)
Oils and diet margarines 11 (3 - 31) 12 (3 - 32) 11 (3 - 28) 11 (3 - 32)
Red meat 39 (17 - 73) 41 (19 - 79) 40 (18 - 75) 40 (18 - 74)
Processed meat 23 (9 - 43) 24 (9 - 44) 23 (9 - 43) 23 (9 - 40)
Sweetened beverages and fruit juices 71 (13 - 176) 63 (6 - 175) 66 (7 - 180) 69 (9 - 175)
Alcohol 9 (1 - 22) 10 (0 - 22) 9 (0 - 21) 8 (1 - 20)
b
Neighbourhood socioeconomic status 0.2 (-0.4; 0.9) 0.2 (-1.0; 1.1) 0.4 (-0.4; 1.0) 0.5 (-0.2; 1.1)
Level of urbanizationc, n (%)
Very low level of urbanisation 566 (28.7) 796 (39.1) 674 (32.2) 398 (20.5)
Low level or urbanisation 620 (31.5) 590 (29.0) 635 (30.3) 684 (35.3)
Moderate level of urbanisation 311 (15.8) 294 (14.5) 411 (19.6) 485 (25.0)
High level of urbanisation 241 (12.2) 220 (10.8) 236 (11.3) 157 (8.1)
Very high level or urbanisation 232 (11.8) 134 (6.6) 140 (6.7) 214 (11.0)
a
Continuous variables are presented as means (standard deviation) or as medians (p25 – p75). bHigher scores represent higher neighbourhood
socioeconomic status. cVery low level of urbanisation ≤ 500 addresses/km2; low level of urbanisation = 500-1000 addresses/ km2; moderate level of
urbanisation = 1000-1500 addresses/ km2; high level of urbanisation =1500-2000 addresses/ km2; very high level or urbanisation ≥ 2000 addresses/
km2. The following variables had missing data: smoking status (n=422); level of urbanisation (n=193); BMI (n=92); household educational level (n=22);
neighbourhood socioeconomic status (n=16).

193
fast - food restaurant exposure in rel ation to diet qualit y , overweight and obesit y

7
8
ABSTRACT
Nudging has received ample attention in scientific literature as an environmental
strategy to promote healthy diets, and may be effective for reaching populations with
low socioeconomic position (SEP). Therefore, the objective of this study was to investigate
how the determinants of food choice shape the perceptions regarding supermarket-
based nudging strategies among adults with low SEP. We conducted semi-structured
interviews among fifteen adults with low SEP using a pre-defined topic list and visual
examples of nudges. Interviews were recorded and transcribed verbatim and content
analysis was used to analyse the data. The results show that food costs, convenience,
healthiness, taste, and habits were frequently mentioned as determinants of food
choice. However, the relative importance of these determinants seemed to context-
dependent. Interviewees generally had a positive attitude towards nudges, especially
when they were aligned with product preferences, information needs, and beliefs about
the food environment. Still, some interviewees also expressed distrust towards nudging
strategies, suspecting ulterior motives. We conclude that nudging strategies should
target foods which align with product preferences and information needs. However,
the suspicion of ulterior motives highlights an important concern which should be
considered when implementing supermarket-based nudging strategies.

Determinants of food choice and

perceptions of supermarket-based
nudging interventions among adults with
low socioeconomic position: the SUPREME
NUDGE project

This chapter was published as:

Harbers MC, Middel CNH, Stuber JM, Beulens JWJ, Rutters F, van der Schouw YT. Determinants
of Food Choice and Perceptions of Supermarket-Based Nudging Interventions among Adults
with Low Socioeconomic Position: The SUPREME NUDGE Project. Int J Environ Res Public Health.
2021;18(11).
chapter 8 determinants of food choice and perceptions of supermarket - based nudging interventions
INTRODUCTION
Obesity presents a global threat to public health: in 2016, more than 1.9 billion adults
suffered from overweight and obesity worldwide (1), predisposing them to a range of
non-communicable diseases including type 2 diabetes and cardiovascular disease. The
burden of overweight and obesity shows a strong social gradient. In the Netherlands,
those with the lowest educational attainment – a key indicator for socioeconomic
position (SEP) – are 1.6 times more likely to be overweight and 2.7 times more likely to
be obese, compared to people with the highest educational attainment (2) and similar
inequalities have been observed in other European countries (3). Ad-dressing social
inequalities is an important step in improving general public health.
One of the major drivers of overweight and obesity is an unhealthy diet, characterized
by high intakes of nutrient-poor and energy-dense foods (4), stressing the need for
improvement of diet quality, especially among populations with low SEP. However, not all
interventions that aim to improve healthy food choices are equally effective across levels
of SEP. Individual-based interventions (e.g., mainly focusing on information provision)
preferentially improve healthy eating outcomes in individuals with high SEP (5), whereas
interventions focused on creating healthier food-environments seem to affect all SEP-
levels equally (5, 6), and could be instrumental in reducing the social inequalities in the
burden of overweight and obesity.
Nudging in particular has become an increasingly popular strategy in creating healthier
food environments. Nudges make healthy choices easier or more intuitive, without
constraining choice for unhealthy alternatives or using financial incentives, for example
by placing healthy foods at eye level. Although a wide range of nudging strategies have
been applied in various contexts, such as supermarkets and cafeterias (7-11), they often
follow a one-size-fits-all approach. The challenge is that customers often constitute a
heterogeneous population (12), with potentially differing responses to nudges based on
their determinants of food choice (e.g., food values, taste preferences, cooking habits).
In order to tackle this issue, recent literature has explored the potential of choice and
delivery personalization, which in fact uses this het-erogeneity data in order to determine
which nudge and which outcome (e.g., food product) will be most effective for targeting a
certain group of individuals (13). Thus, to maximize the potential effectiveness of nudges
in populations with low SEP, we need to understand the determinants of food choice in
this group in order to improve the design of nudges so that they fit with their habitual
food behaviour practices.
196
Previous research has reported food costs, familiarity, habits, role models, and food
outlet availability as important determinants of food choice among populations with
low SEP (14-18), but evidence is relatively scarce. Further, although there is evidence to
suggest that people perceive nudging strategies to be acceptable (19), little is known
about perceptions regarding supermarket-based nudging strategies among adults with
low SEP and how determinants of food choice – such as food costs and habits – may
shape these perceptions.
The SUPREME NUDGE project aims to improve lifestyle behaviours and lower
cardiometabolic risk among adults with low SEP through implementing nudging and
pricing strategies in a real-life supermarket set-ting (20). For the design of the nudges
in this project, we aimed to explore how the determinants of food choice shape the
perceptions regarding supermarket-based nudging strategies among adults with low
SEP in order to better inform the design of the nudges.
METHODS
Participant recruitment
Participants were recruited in two low-SEP neighborhoods – defined as having a SEP-
score below the national average – in the city of Utrecht during two rounds of data
collection. During the first round (October and November 2018), participants (n=14)
were recruited in a shopping centre in the city district of Overvecht. Of the fourteen
participants that were recruited, two participants were not eligible for inclusion in the
study based on their educational level and therefore are omitted from data analysis. A
preliminary analysis of the data was conducted in order to grasp whether data saturation
had been reached. As some new topics emerged from the last interviews which prompted
further exploration, additional participants (n=3) were recruited in a community centre 8
in the city district of Rivierenwijk. Formal assessment of thematic saturation using the
methodology described by Guest et al. (21) showed that by using a base size of 4 we
reached the 0% new information threshold at 12+1 interviews, indicating that during the
13th interview, no additional relevant themes were identified.
During the first round of data collection, the researcher approached shoppers and briefly
explained the purpose of the interview. If shoppers were interested in participating in
the interview, the researcher immediately invited the participant to a nearby restaurant
where the interview took place. Reasons for refusing to participate were mainly related
to time-concerns (e.g., people were busy doing their groceries). During the second
round of data col-lection, the researcher approached potential participants during
an informal coffee morning in a community centre among mostly elderly community
197
chapter 8 determinants of food choice and perceptions of supermarket - based nudging interventions

members and volunteers. If they expressed interest in participating in the interview,
the researcher immediately invited them to a more quiet table in the community centre
where the interview took place. During the second round of data-collection, none of the
approached potential interviewees refused to participate. Table 1 provides an overview
of participants characteristics.
Table 1. Participant characteristics
level, the interview was shortened. All participant were reimbursed with a gift voucher
worth of 20€, including the participants (n=2) who did not have a lower educational level.
We conducted semi-structured interviews that lasted on average 25 minutes following a
pre-defined topic list (Supplementary File 1). The topic list was used in a flexible manner
by the researcher, based on the issues that were mentioned by the participants. The
first part of the interview was focused on the everyday life of participants, in order to

Interviewee # Age Sex Round of data Educational level get a feeling of the relevant wider contextual factors that shape dietary choices (e.g.,
collection employment, living situation) and in order for the interviewee to feel at ease. In the
Interviewee 1 33 male 1 intermediate vocational education second part of the interview, the participants were asked how they valued food, and
Interviewee 2 51 female 1 intermediate vocational education what they perceived to determine their food choices. First, the participants were given
Interviewee 3 46 female 1 intermediate vocational education the opportunity to answer freely and in their own words, and if deemed appropriate,
Interviewee 4 68 female 1 advanced elementary education the researcher probed by suggesting other potential determinants of food choice. In the
Interviewee 5 69 female 1 lower vocational education third part of the interview, we asked under which circumstances it was easy or difficult
Interviewee 6 54 male 1 lower vocational education to eat healthy for the participant. If appropriate, the researcher referred back to what
Interviewee 7 70 male 1 intermediate vocational education was mentioned in the first and second part of the interview in order to probe further.
Interviewee 8 71 female 1 lower vocational education
Interviewee 9 79 male 1 advanced elementary education In the last part of the interview, the researcher showed various photos of nudges
Interviewee 10 65 female 1 intermediate vocational education that could be applied in supermarkets to promote healthier food choices, and briefly
Interviewee 11 58 female 1 intermediate vocational education explained them (Supplementary File 2). Instead of using the term of nudging, we
Interviewee 12 79 female 1 intermediate vocational education introduced the photos as strategies that supermarkets could implement in order to

Interviewee 13 73 female 2 advanced elementary education assist customers in making more healthy food choices. Thereafter, we explored relative

Interviewee 14 57 male 2 lower vocational education judgement of the nudges by asking participants to choose the photo that appealed to

Interviewee 15 50 female 2 advanced elementary education

them the most, and we also asked them to explain why. Vice versa, we asked participants
to choose the photo which appealed to them the least, and explain why this was the case.

Interview procedure
First, the purpose of the interview was once again explained and in-formed consent was
obtained. Prior to the start of the interview, we asked the participants to fill out a short
questionnaire on demographic characteristics, including age, sex, and highest attained
educational level. Answer possibilities for educational level were based on the Dutch
education system, and lower educational level was defined as primary education up to
completing intermediate vocational education. This short questionnaire was used as a
screener for the researcher, in order to verify that the participants had lower educational
level which was used a proxy for SEP. As such, the recruitment of low-SEP participants was
achieved in two ways: first, by recruiting in are-as which had a low neighborhood SEP-
score, and second, by verifying the highest educational level attained by the participant.
If the highest attained educational level was not within the definition of low educational
The study was conducted in accordance with the Declaration of Helsinki and was 8
approved by the ethics committee of the University Medical Center Utrecht. Reporting
of this qualitative study follows the guidelines set out by the Consolidated criteria for
reporting qualitative research (COREQ) checklist (22).
Research team
The first author (female PhD candidate in the field of epidemiology) conducted twelve
of the fifteen interviews. The third author (female PhD candidate in the field of lifestyle
epidemiology) conducted three of the fifteen interviews. The interviewers piloted the
topic guide together in order for it to be used in a consistent manner. Both interviewers
had no prior relationship with either of the interviewees. Interviewees were informed on
the scientific background of the interviewers and personal motivations for conducting the
interviews (e.g., learning how to help people make more healthy choices in supermarkets)

198 199
chapter 8 determinants of food choice and perceptions of supermarket - based nudging interventions
were made explicit. The second author is a male PhD candidate conducting research in
the field of systems innovation and transition theory using qualitative research methods,
and assisted the first author in performing data analysis and data interpretation. The
other members of the research team are (assistant) professors in the field of (lifestyle)
epidemiology, and were involved in designing the study, data interpretation, and writing
of the manuscript.
Data-analysis
All interviews were audio-recorded and transcribed verbatim by a re-search assistant
and a professional audio transcription company, and all personal identifying information
was removed from the transcripts. Tran-scripts were imported in Atlas.ti (ATLAS.ti
Scientific Development GmbH, Berlin, Germany). The process of coding was informed
by the Qualitative Analysis Guide of Leuven (QUAGOL) (23), and was based on content
analysis using a combination of inductive and deductive coding. First, the transcripts
were read and re-read in order for the first-author to become familiar with the data.
This phase included underlining relevant passages and noting emerging themes in
the margin of the transcript. In the second phase, the key elements of the transcripts
were summarized in ‘narrative interview reports’, which aimed to holistically capture
the experience of the participant. In the third phase, key elements that emerged from
the transcripts and narrative interview reports appearing to be relevant for answering
the research questions were translated into more abstract concepts. These emerging
concepts were used to construct the initial codebook (e.g., inductive coding). For the
first re-search question, this inductive approach was complemented with a deductive
approach, by also including relevant concepts identified from literature (e.g., deductive
coding) (24, 25). For the second research question, the code-book only included concepts
that emerged from the narrative interview re-ports, thus following a primarily inductive
coding approach, as research and theory concerning perceptions, needs, and preferences
on nudging interventions is scarce. After coding the first five interviews, the codes and
quotations were carefully examined and compared by the first and second author,
and minor revisions to the codebook were made. The following five interviews were
coded using the updated codebook, and the first five interviews were again analysed
in order to adjust the coding following the revisions to the codebook. This iterative
process of comparing codes and quotations among the first and second author was
repeated until all interviews were analysed. Codes were structured into intrapersonal,
interpersonal, socioeconomic and environmental determinants of food choice, following
socio-ecological models of health (24). In order to provide an indication of the relative
importance of the identified determinants of food choice, the narrative report of findings
is accompanied by a bar graph indicating the number of interviewees mentioning the
200
specific determinant (interview frequency), and the frequency of coding this determinant
in the interview reports (quotation frequency).
RESULTS
Determinants of food choice
Intrapersonal determinants of food choice
Physiological factors
We identified three physiological determinants of food choice: ill health, hunger and
satiety, and taste preferences (Figure 1). Eight interviewees shared various experiences
about ill health, including coping with cancer, diabetes, heart failure, and physical
disability. Interviewees said that these conditions influenced their food choices in
various ways, for example through affecting taste perceptions, requiring dietary
adjustments, affecting product choice, or triggering stress-related unhealthy snacking.
Some interviewees also expressed that their own or a relative’s health motivated them
to eat healthy:
“Despite being this unhealthy already I still want to reach the age of 100. And then I
think, well maybe, despite all these illnesses, [making healthy dietary choices] helps me
in achieving that.” (R5)
The role of hunger and satiety was mentioned by five interviewees. Interviewees talked
about hunger with respect to meal-timing, for example a preference for skipping
breakfast due to not feeling hungry in the morning. For two other interviewees this
theme related to food choice, as they perceived some foods as more satiating than 8
others. The role of taste preferences was mentioned by eight interviewees. Taste
preferences were often phrased as if they were part of one’s identity, and they often
expressed very definite likes and dislikes for various foods:
“I am more of a bread eater” (R3) and “I am not really a fruit type of person” (R2)
Attitudes, beliefs, and perceptions
Attitudes towards food varied and could co-occur. Five interviewees had a functional
attitude, mainly describing food as a source of energy and nutrients, which are necessary
for the body to function. Four interviewees held a hedonic attitude, mainly describing
how much they enjoyed food. Five interviewees had a complicated relationship with food
201
chapter 8 determinants of food choice and perceptions of supermarket - based nudging interventions

by describing that food choice required elaborate contemplation, which was prompted
by their health condition or other goals, such as weight loss:

“Since I have heart failure, I have to carefully watch my potassium intake […]. So that
is really difficult for me. Because I would love to make a big pan of tomato soup. But I
cannot do that anymore” (R5)

Interviewees held various beliefs regarding properties of foods, which influenced their
food choices. Four interviewees expressed that food should be of ‘good quality’, referring
to the shop where they bought their food (e.g., fish at a local fish shop), the physical
appearance of food, or preferences for certain brands. Seven interviewees expressed to
prefer fresh foods (e.g., mostly fresh fruits and vegetables), because they perceived this
to have better taste, structure, and to be healthier. For some interviewees, these factors
outweigh the perceived higher cost or longer preparation time. Finally, five interviewees
expressed a general distrust towards the food industry, which was perceived to ‘trick’
consumers or ‘mess’ with the foods.

Figure 1. Interview and quotation frequency of perceived determinants of food choice

“Vegetarian food. Who says that is healthy? Maybe they add all kinds of chemical stuff.
There is no other way!” (R13)

Habits
Nearly all interviewees framed food choices as habits they were ‘just used to’ or that
‘worked out’ for them. Interviewees usually reported a traditional Dutch dietary pattern,
characterized by potatoes, vegetables, fruits, meats, fish, bread, dairy, and some simple
international dishes, such as fried rice and pastas. Some interviewees reported a
distinction in food habits be-tween weekdays, weekends, and holidays. During the latter
two, interviewees allowed themselves to make more unhealthy food choices. Skipping 8
breakfast was a recurring food habit, mentioned by four interviewees. Despite the ‘static’
notion of habitual food choices, six interviewees had an open attitude towards trying
out new foods. Motivations to try new foods included perceived health benefits (e.g.,
buying spice mixes without salt because of salt restriction), expectations of a good taste,
or as means of dietary variation.

Motivation and values

Food choices were influenced by several commonly valued characteristics: perceived
food costs, convenience, healthiness, and taste. Ani-mal-friendliness was also mentioned
by one interviewee. These values were often weighted against each other when
interviewees faced certain food choices. As such, their relative importance was not
static, but rather con-text-specific.

202 203
chapter 8 determinants of food choice and perceptions of supermarket - based nudging interventions
Food costs were mentioned by ten interviewees. Interviewees who valued (low) food
costs highly, perceived it to be a waste of money to spend too much money on food,
over other ‘better purposes’. One interviewee also ex-pressed to find it important to
teach his children that healthy eating could be affordable. Participants who did not
value food costs, generally found other food values (e.g., animal friendliness, taste) to
be more important.
Convenience was mentioned in ten interviews. Interviewees generally found it important
that meals had a shorter preparation time and required a limited number of ingredients.
Three interviewees mentioned that convenience was especially important when they
had to cook for themselves; when cooking for others they were prepared to put more
effort into cooking. Interviewees generally appreciated convenience in terms of products
sold in supermarkets that aided in preparing convenient and healthy meals (e.g., meal
boxes or pre-cut vegetables).
The healthiness of foods was mentioned by thirteen interviewees. Only one interviewee
expressed to not value the healthiness of foods, which was regarded as not tasty.
Other interviewees generally stated to value the healthiness of foods, which they often
substantiated by explaining how they found it important to eat (fresh) vegetables,
vitamins, drink water, and to minimize intake of pork, additives, fat, sugar-sweetened
beverage sand sugar. However, the competing value of taste (mentioned by thirteen
interviewees) seemed to be a boundary condition for choosing healthy foods, as
participants indicated that foods should not be ‘bland’, but should be ‘enjoyable’.
Knowledge and skills
Interviewees generally seemed familiar with what a healthy diet constitutes, which they
described as eating plenty of vegetables, vitamins, fruit, fish, and minimizing intake
of sugar, alcohol, and sugar-sweetened beverages in twelve of the interviews. Four
interviewees also expressed concerns about additives and artificial sweeteners present
in food, particularly in packaged foods and ready-made meals, which they considered
detrimental to health.
Eight interviewees mentioned various media sources from which they obtained
information about nutrition, including television shows, the inter-net, and doctors.
For example, one interviewee described that she had received medication for her
cholesterol, which prompted her to search online to search for dietary regimens:
“And then I read that the combination of cinnamon and honey reduces cholesterol levels.
So every morning I add that to my cereal.” (R5)
204
Three interviewees perceived nutrition information to be confusing, due to conflicting
statements and a lack of clarity regarding which sources are credible.
“And I have heard that butter is better than margarine. So yeah, there are just so many
perspectives.” (R6)
Five interviewees expressed that they had sought professional help for their dietary
behaviours from dieticians or health care institutions, or indicated to have participated
in a weight loss program or group therapy for binge-eating. Moreover, five interviewees
indicated they were adhering to a diet, such as a low-carb diet or a salt-restricted diet.
Furthermore, six participants described various strategies or skills, which facilitated
them in making healthy dietary choices. Some strategies were mainly driven by health
values, such as making sure that in-home food availability was healthy, restricting
snacking after a certain time, or only allowing oneself to fill up their plate once. Other
strategies, however, were not only driven by health motives, but also incorporated other
values such as convenience and foods costs. These included: buying foods on offer in
large quantities, and/or making meals ahead that one would keep in the freezer, and
restricting (un-healthy and unplanned) food purchases by using a grocery list.
Interpersonal determinants of food choice
Partners, family members, neighbours and friends were mentioned as actors influencing
food choice in either direct or indirect ways. Partners and direct family members often
exerted a direct influence on food choice as they provided input on dinner choices, which
was mentioned by nine interviewees. Parents of interviewees exerted more of an indirect
influence, as interviewees often perceived certain food habits (e.g., choosing for fresh
vegetables) to originate from their upbringing. The composition of the household also
influenced food choice, as interviewees were generally more likely to put more effort 8
in preparing meals when they lived or had dinner with others, which was mentioned by
three interviewees. Similarly, eating was also often associated with social activities by
nine interviewees, such as gatherings with friends and family. Often these occasions
were associated with choosing more unhealthy foods.
Socio-economic determinants of food choice
Two interviewees explicitly mentioned that they had limited income, which caused them
to carefully watch what they spend on food purchases:
“I am price conscious. I have to be, because I only have state pension (AOW)” (R4)
205
chapter 8 determinants of food choice and perceptions of supermarket - based nudging interventions
Two other interviewees mentioned that they were brought up with limited financial
possibilities, which also affected their perceptions on the affordability of food and food
behaviours at present:
“Look, I come from a big family where you could not do just whatever you wanted. And
that is also what I want to teach my children. It does not have to cost a lot of money to
live healthily.” (R1)
“I have never had a big appetite […] Back in the days, my parents made dinner, and
that was split up. So you could never have a second plate. […] That was in a time with
limited possibilities” (R9)
Strikingly, as the first quote illustrates, a limited income was not necessarily perceived
as a barrier to eating healthy, and the portion restrictions imposed by growing up in
a family with limited income actually may have been a facilitator to portion control.
Employment also influenced food choices and was mentioned by five interviewees,
and primarily exerted influence through the (limited) amount of time interviewees
perceived to have available for meal preparation and food shopping. For two other
interviewees, (anticipated) employment provided motivation to eat healthy given a
physically demanding job, or because employment was considered to give structure to
the interviewee’s daily routine:
“A job would really give some structure to my day. Right now there is just no structure
to my day. [..] You just don’t take the time to sit down and eat. Then it’s just quick, I will
pop by the supermarket and have a croissant.” (R14)
Environmental determinants of food choice
Twelve interviewees mentioned that food products that were on offer in supermarkets
often guided their decisions on their purchases. These foods were sometimes bought
in large quantities with which interviewees prepared meals in advance. Moreover,
interviewees shared various price comparisons for fruits and vegetables available in
the supermarket, including the price of organic vs. regular vegetables, fresh vegetables
vs. canned vegetables, freshly squeezed orange juice vs. pre-packaged orange juice,
or comparing prices amongst food outlets (e.g., discount supermarket vs. regular
supermarket). One interviewee also expressed to dislike the seemingly increasing prices
of fresh fruits and vegetables.
“But if I see that snacks are cheap en that fresh food is only getting more expensive…
And if you only receive social assistance benefits. That is just undoable” (R14)
206
Moreover, eight interviewees described the food environment – and particularly the
supermarket environment – as tempting, inducing unhealthy food purchases. One
interviewee explicitly mentioned that he was aware of some ‘smart’ marketing strategies
employed by supermarkets, including secondary placement of food items. In this light,
six participants described various strategies which facilitated healthy food purchasing
decisions in the supermarket, including making a grocery list prior to going to the
supermarket, or shopping for groceries on a weekly basis instead of daily, and thereby
minimizing the temptation to unnecessarily buy unhealthy foods.
Perceptions of nudging
The closing parts of each interview explored interviewees’ perceptions regarding
a number of nudging strategies, by means of visual examples (Supplemental File 2).
Generally, interviewees had positive perceptions to-wards the nudges shown during the
interviews. More specifically, interviewees seemed to perceive nudges more positively
when these were aligned with their perceived determinants of food choice, e.g., with
their product preferences, information needs, and beliefs about the food environment.
Product preferences
Generally, nudges were perceived more positively if they nudged foods which aligned with
interviewees’ product preferences in terms of taste, healthiness, costs, and preparation
time. Similarly, product characteristics that seemingly made a nudge less appealing were
perceived non-freshness, expected poor taste, high costs, and complicated or lengthy
preparation. For example, an interviewee indicated to dislike a nudge applied to nuts,
because she did not like the taste of nuts. This may suggest that a nudge is less likely
to affect a consumer if the nudged product itself does not fit their or their family’s food
preferences
8
“It’s just as with this: do I find it tasty? Do I like it? Wat does my partner like, that’s also
something you need to consider. So, it quickly becomes for me that if there’s something
I like, I take it, and otherwise I skip it completely.” (R2)
“But this appeals to me, those recipes. But then it has to be somewhat easy, or saltless.”
(R5)
Knowledge and information
Generally, nudges were perceived more positively if they anticipated on some kind
of information need, for example regarding calorie content and other nutritional
information. At times this need was linked to ill health and dietary restrictions (e.g., a
sodium-restricted diet), which prompted a need for individualized dietary information.
207
chapter 8 determinants of food choice and perceptions of supermarket - based nudging interventions
As an exception, one interviewee indicated that a specific type of information (product
popularity) was of very low interest to them. Finally, four other interviewees expressed
a certain extent of distrust towards nudges presenting such information, or indicating
‘healthy’ products, suspecting ulterior motives of the supermarket or food industry. As
such, these types of nudges will need to be designed in a way which strengthens their
credibility, or risk not being trusted by part of the target group.
“The first thing that comes to my mind is: who says this and who gets to decide on this.
You get what I mean? At a first glance, this all seems very nice, but then it gets fiddled
with by Coca Cola, and they make something different out of it so it does not contain
sugar, but then it does contain something else.” (R3)
Food environment
As previously described, some of the interviewees held rather negative perceptions
towards the food environment in supermarkets, which they perceived as tempting and
counter-productive in making healthy food choices. In light of this, several interviewees
expressed that nudges would actually provide them with the freedom of autonomous
choice between healthy and unhealthy foods, illustrating intervention acceptability:
“They should put this on more products. So if you are in front of the crisps shelf, you can
think, well, I can take Lays because that is easy. But that then you have an alternative
next to it, from which you can see, well, it does actually provide less calories and it is
just as tasty.” (R2)
Additionally, it appeared that interviewees had different strategies for navigating
the tempting supermarket environment: some adhered to a grocery list or described
themselves as habitual shoppers while others let in-store food availability guide their
food purchases. This shopping strategy also influenced interviewee’s perception of
nudges: when interviewees described themselves as habitual shoppers they had less
positive perceptions towards nudging strategies as compared to interviewees who let
in-store food availability guide their food purchases.
DISCUSSION
The present study aimed to explore how the determinants of food choice shape the
perceptions regarding supermarket-based nudging strategies among adults with low
SEP. Overall, interviewees mentioned a wide range of factors on individual level, socio-
economic level and environmental level to influence their food choices. Food costs,
convenience, healthiness, taste, and habits were frequently mentioned and their relative
208
importance seemed to be context-specific. Moreover, interviewees generally had
positive perceptions towards nudges, especially when they aligned with their product
preferences, information needs, and beliefs about the food environment. Still, a small
number of interviewees also expressed distrust towards nudging strategies, suspecting
ulterior motives of supermarkets or food industry, irrespective of their determinants
of food choice.
Determinants of food choice and nudging
In the present study, we confirm earlier findings on the importance of determinants
such as food costs (14-18, 26), convenience (15), healthiness (15, 26), and taste (15,
26). Additionally, this study provides preliminary evidence that nudges that align with
interviewee’s determinants of food choice, are perceived more positively. This was most
notably observed for product preferences: if nudges targeted foods which interviewees
liked (e.g., in terms of taste, convenience), then nudges were also perceived more
positively. This may imply that the type of food which is nudged may be instrumental
for nudge effectiveness in populations with low SEP. For example, following international
dietary recommendations, the intake of red and processed meat should be reduced
(27). From the present study it may be deduced, that in or-der to nudge people from
meat to another substitute, this alternative should align with values such as such as
taste and convenience. As a result, nudging a healthy meat substitute (e.g., similar taste
and preparation method) may be a better option than for example tofu (e.g., notably
different taste, potentially different preparation method). We also noted that nudges
were perceived more positively when they tapped into certain information needs, which
may imply that information nudges (28) may be of particular interest to populations
with low SEP.
However, a number of interviewees also voiced concerns about the legitimacy of 8
supermarket-driven health-promoting initiatives, suspecting ulterior motives. Distrust
of information on health behaviours has previously been highlighted as a barrier for
attaining behaviour change in community-based interventions among populations with
low SEP (29), but has not been documented in the specific case of supermarket-based
nudging strategies. As opponents of nudging often frame nudging as infringement on
autonomous choice (30), one way to address the underlying feeling of distrust is for
supermarkets to be transparent about applying nudging strategies using positive framing
and to explain why, how, and with whom (e.g., govern-mental agencies) they participate
in such initiatives. Especially with regard to the application of information nudges, this
would imply that the information provided is perceived to come from a credible source.
Still, it must be noted that interviewees did not voice ethical concerns about nudging
209
chapter 8 determinants of food choice and perceptions of supermarket - based nudging interventions
per se, and some even perceived nudges to actually provide them with autonomy to
make in-store food choices.
Beliefs about the food environment also influenced perceptions of nudging strategies. We
observed that interviewees often viewed the super-market environment in a relatively
negative way, describing it as a tempting environment for which the interviewees had
developed strategies to withhold themselves from engaging in alluring unhealthy
purchases (e.g., restricting supermarket visits). As such, interviewees perceived nudging
to be a welcome strategy for making the supermarket food environment more healthy.
This contrasts heavily with the perception among retailers of customer demand for
healthy foods to be low and customer interest in health to be limited, which is often cited
as a barrier for retailer’s engagement in health promoting interventions (31). This could
indicate a mismatch between consumer needs, and retailers’ perceptions of them, and
present a potential argument to convince retailers to invest more into health-promoting
initiatives.
The role of habits
The role of habit as an important determinant of food choice in adults with low SEP
warrants some extra consideration in the context of nudging. Nearly all interviewees
indicated that their food habits and/or shopping habits were largely habitual. Habits
have been defined as behaviours that have become established through a history of
systematic repetition and reinforcement, and therefore have become automatized (32).
Since habits and nudges share the same proposed working mechanism of automated
unconscious cognitive processes, the question arises to what extent nudges are able
to overrule hard-wired automated habitual food choices. Emerging literature suggests
that strong habits or a priori preferences indeed may form boundary conditions for
nudge effectiveness regarding food choice (33, 34). For example, a study conducted
by Venema et al. showed that the effects of a portion size nudge aimed at decreasing
sugar consumption in tea was less effective in participants with strong habits concerning
the addition of sugar to tea beverages, compared to participants with less strong
habits (34). Similarly, in a study aiming to nudge participants to a smaller soda drink,
it was found that a priori preferences for soda drinks (e.g., liking of those drinks or
intentions to reduce consumption of soda drinks), were stronger predictors of the chosen
drinks than the nudge itself (33). Although these former studies were conducted in
laboratory settings, these findings may suggest that also in re-al-life environments
such as supermarkets, a priori preferences of customers for specific foods are likely to
compete with nudges. Examples of strong preferences that were noted in the present
study include the preference for fresh vegetables as opposed to canned vegetables,
preference for certain brands, or preference for foods that are convenient. A practical
210
implication that may follow from this, is making sure that the nudged foods are not
subject to strong a priori preferences, but represent foods for which adults with low
SEP are likely to be fairly indifferent.
Needs concerning other public health interventions
The needs expressed by interviewees also underline the importance of other public
health interventions other than nudging strategies that aim to promote healthy food
choices. Especially in the case of ill health there was a need to obtain objective and
individualized dietary information (e.g., information on salt content prompted by heart
failure induced salt restriction), which may be fulfilled by the use of mobile apps that
can be used in the supermarket environment. An example of such an app is the ‘Kies
ik Gezond?’ (do I make a healthy choice?) application from the Dutch Nutrition Center,
which allows users to scan barcodes in the supermarket and subsequently informs
the user whether the scanned food item is contained in the Wheel of Five (the Dutch
dietary recommendations), and also provides healthy alter-native suggestions for
unhealthy foods. The use of these apps could be promoted, especially given the fact
that some interviewees perceived nutrition information to be conflicting, and even
obtained very erroneous nutrition in-formation from sources, such as the internet.
Lastly, the interviews high-lighted that price of foods often guided food choice within
the supermarket setting, indicating that pricing strategies are likely to be supported by
populations with low SEP. Concluding, the needs expressed by the interviewees suggest
that nudging is not likely to be a silver bullet for achieving healthy diets (35), but that in
fact we need a multitude of interventions (e.g., personalized nutrition advice, pricing
strategies) to match the varying needs and preferences of adults with low SEP.
Limitations and strengths
The present study should be interpreted in light of some limitations. First, we only 8
used educational level as an indicator of SEP, as we deemed it stigmatizing to ask
interviewees about other proxies of SEP such as income. As it is well-known that SEP is a
multidimensional construct, the use of educational level only is likely to be a simplification
of reality. Moreover, given the difficulties associated with recruiting individuals with low
SEP (36), we adhered to a broader definition of lower educational than for example the
definition used by the Dutch Central Bureau of Statistics for pragmatic rea-sons. Third,
our sample was characterized by a relatively high age, which may limit the generalizability
of our findings to younger populations.
The study also had several strengths. The analysis and presentation of the data were
thoroughly reviewed by and discussed with a second researcher, which strengthens the
internal validity of the study. Although not systematically recorded, the interviewers
211
chapter 8 determinants of food choice and perceptions of supermarket - based nudging interventions
judged the interviewees to constitute a representative sample of the population with
regard to weight status, which may be of importance when exploring the topic of food
choice. Lastly, although the sample size was relatively small, saturation was reached,
which indicates that all major determinants for the interviewed group are likely to have
been identified.
CONCLUSION
Food costs, convenience, healthiness, taste, and habits were perceived to be
important determinants of food choice among adults with low SEP. Interviewees
especially appreciated nudges when they targeted foods that aligned with their
product preferences, information needs, and beliefs about the food environment.
These underlying preferences can be taken into account in the design of nudges, in
particular for choosing the type of food to be nudged. Additionally, given the variation
in determinants of food choice and their context-specific relevance, applying a wide
variation of nudging strategies may also be viable since they may appeal to a wider
audience. Last, some interviewees also voiced concerns about the legitimacy of nudging
strategies suspecting ulterior motives, which may undermine nudge effectiveness and
should be considered when implementing supermarket-based nudging strategies.
212
REFERENCES
1. NCD Risk Factor Collaboration. Worldwide trends in body-mass index, underweight,
overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based
measurement studies in 128·9 million children, adolescents, and adults. Lancet 2017, 390,
2627-2642, doi:10.1016/s0140-6736(17)32129-3.
2. Leefstijlmonitor. Centraal Bureau voor de Statistiek. [cited 13-07-2018]. Available from:
https://www.volksgezondheidenzorg.info/onderwerp/overgewicht/cijfers-context/huidige-
situatie#node-overgewicht-naar-opleiding.
3. Hoffmann, K.; De Gelder, R.; Hu, Y.; Bopp, M.; Vitrai, J.; Lahelma, E.; Menvielle, G.; Santana,
P.; Regidor, E.; Ekholm, O.; et al. Trends in educational inequalities in obesity in 15 European
countries between 1990 and 2010. Int J Behav Nutr Phy 2017, 14, 63, doi:10.1186/s12966-
017-0517-8.
4. Swinburn, B.A.; Caterson, I.; Seidell, J.C.; James, W.P. Diet, nutrition and the prevention
of excess weight gain and obesity. Public Health Nutr 2004, 7, 123-146, doi:10.1079/
phn2003585.
5. McGill, R.; Anwar, E.; Orton, L.; Bromley, H.; Lloyd-Williams, F.; O’Flaherty, M.; Taylor-
Robinson, D.; Guzman-Castillo, M.; Gillespie, D.; Moreira, P.; et al. Are interventions to
promote healthy eating equally effective for all? Systematic review of socioeconomic
inequalities in impact. BMC Public Health 2015, 15, 457, doi:10.1186/s12889-015-1781-7.
6. Hoenink, J.C.; Mackenbach, J.D.; Waterlander, W.; Lakerveld, J.; van der Laan, N.; Beulens,
J.W.J. The effects of nudging and pricing on healthy food purchasing behavior in a virtual
supermarket setting: a randomized experiment. The international journal of behavioral
nutrition and physical activity 2020, 17, 98-98, doi:10.1186/s12966-020-01005-7.
7. Bucher, T.; Collins, C.; Rollo, M.E.; McCaffrey, T.A.; De Vlieger, N.; Van der Bend, D.; Truby,
H.; Perez-Cueto, F.J. Nudging consumers towards healthier choices: a systematic review
of positional influences on food choice. British Journal of Nutrition 2016, 115, 2252-2263,
doi:10.1017/s0007114516001653.
8. Arno, A.; Thomas, S. The efficacy of nudge theory strategies in influencing adult dietary
behaviour: a systematic review and meta-analysis. BMC Public Health 2016, 16, 676,
doi:10.1186/s12889-016-3272-x.
9. Hollands, G.J.; Carter, P.; Anwer, S.; King, S.E.; Jebb, S.A.; Ogilvie, D.; Shemilt, I.; Higgins, J.P.T.; 8
Marteau, T.M. Altering the availability or proximity of food, alcohol, and tobacco products
to change their selection and consumption. Cochrane Database of Systematic Reviews
2019, doi:10.1002/14651858.CD012573.pub2.
10. Broers, V.J.V.; De Breucker, C.; Van den Broucke, S.; Luminet, O. A systematic review and
meta-analysis of the effectiveness of nudging to increase fruit and vegetable choice. Eur
J Public Health 2017, 27, 912-920, doi:10.1093/eurpub/ckx085.
11. Harbers, M.C.; Beulens, J.W.J.; Rutters, F.; de Boer, F.; Gillebaart, M.; Sluijs, I.; van der
Schouw, Y.T. The effects of nudges on purchases, food choice, and energy intake or content
of purchases in real-life food purchasing environments: a systematic review and evidence
synthesis. Nutrition Journal 2020, 19, 103, doi:10.1186/s12937-020-00623-y.
12. Sunstein, C.R. Impersonal default rules vs. active choices vs. personalized default rules:
A triptych. Active Choices vs. Personalized Default Rules: A Triptych (May 19, 2013) 2013.
13. Mills, S. Personalized nudging. Behavioural Public Policy 2020, 1-10, doi:10.1017/bpp.2020.7.
213
chapter 8 determinants of food choice and perceptions of supermarket - based nudging interventions

14. Turrell, G.; Kavanagh, A.M. Socio-economic pathways to diet: modelling the association
between socio-economic position and food purchasing behaviour. Public Health Nutr 2006,
9, 375-383, doi:10.1079/phn2006850.
15. Konttinen, H.; Sarlio-Lähteenkorva, S.; Silventoinen, K.; Männistö, S.; Haukkala, A. Socio-
economic disparities in the consumption of vegetables, fruit and energy-dense foods:
the role of motive priorities. Public Health Nutrition 2013, 16, 873-882, doi:10.1017/
S1368980012003540.
16. Evans, A.; Banks, K.; Jennings, R.; Nehme, E.; Nemec, C.; Sharma, S.; Hussaini, A.; Yaroch,
A. Increasing access to healthful foods: a qualitative study with residents of low-income
communities. Int J Behav Nutr Phys Act 2015, 12 Suppl 1, S5, doi:10.1186/1479-5868-12-S1-S5.
17. Bukman, A.J.; Teuscher, D.; Feskens, E.J.; van Baak, M.A.; Meershoek, A.; Renes, R.J.
Perceptions on healthy eating, physical activity and lifestyle advice: opportunities for
adapting lifestyle interventions to individuals with low socioeconomic status. BMC Public
Health 2014, 14, 1036, doi:10.1186/1471-2458-14-1036.
18. van der Velde, L.A.; Schuilenburg, L.A.; Thrivikraman, J.K.; Numans, M.E.; Kiefte-de Jong, J.C.
Needs and perceptions regarding healthy eating among people at risk of food insecurity:
a qualitative analysis. International Journal for Equity in Health 2019, 18, 184, doi:10.1186/
s12939-019-1077-0.
19. Reynolds, J.P.; Archer, S.; Pilling, M.; Kenny, M.; Hollands, G.J.; Marteau, T.M. Public
acceptability of nudging and taxing to reduce consumption of alcohol, tobacco, and food:
A population-based survey experiment. Soc Sci Med 2019, 236, 112395, doi:10.1016/j.
socscimed.2019.112395.
20. Stuber, J.M.; Mackenbach, J.D.; de Boer, F.E.; de Bruijn, G.J.; Gillebaart, M.; Harbers,
M.C.; Hoenink, J.C.; Klein, M.C.A.; Middel, C.N.H.; van der Schouw, Y.T.; et al. Reducing
cardiometabolic risk in adults with a low socioeconomic position: protocol of the Supreme
Nudge parallel cluster-randomised controlled supermarket trial. Nutr J 2020, 19, 46,
doi:10.1186/s12937-020-00562-8.
21. Guest, G.; Namey, E.; Chen, M. A simple method to assess and report thematic saturation
in qualitative research. PLoS One 2020, 15, e0232076-e0232076, doi:10.1371/journal.
pone.0232076.
22. Tong, A.; Sainsbury, P.; Craig, J. Consolidated criteria for reporting qualitative research
28. Hollands, G.J.; Bignardi, G.; Johnston, M.; Kelly, M.P.; Ogilvie, D.; Petticrew, M.; Prestwich,
A.; Shemilt, I.; Sutton, S.; Marteau, T.M. The TIPPME intervention typology for changing
environments to change behaviour. Nature Human Behaviour 2017, 1, 0140, doi:10.1038/
s41562-017-0140.
29. Everson-Hock, E.S.; Johnson, M.; Jones, R.; Woods, H.B.; Goyder, E.; Payne, N.; Chilcott, J.
Community-based dietary and physical activity interventions in low socioeconomic groups
in the UK: A mixed methods systematic review. Preventive Medicine 2013, 56, 265-272,
doi:https://doi.org/10.1016/j.ypmed.2013.02.023.
30. Schmidt, A.T.; Engelen, B. The ethics of nudging: An overview. Philosophy Compass 2020,
15, e12658, doi:https://doi.org/10.1111/phc3.12658.
31. Middel, C.N.H.; Schuitmaker-Warnaar, T.J.; Mackenbach, J.D.; Broerse, J.E.W. Systematic
review: a systems innovation perspective on barriers and facilitators for the
implementation of healthy food-store interventions. Int J Behav Nutr Phy 2019, 16, 108,
doi:10.1186/s12966-019-0867-5.
32. Orbell, S.; Verplanken, B. The automatic component of habit in health behavior: habit as
cue-contingent automaticity. Health Psychol 2010, 29, 374-383, doi:10.1037/a0019596.
33. Venema, T.A.G.; Kroese, F.M.; De Vet, E.; De Ridder, D.T.D. The One that I Want: Strong
personal preferences render the center-stage nudge redundant. Food Quality and
Preference 2019, 78, 103744, doi:https://doi.org/10.1016/j.foodqual.2019.103744.
34. Venema, T.A.G.; Kroese, F.M.; Verplanken, B.; de Ridder, D.T.D. The (bitter) sweet taste of
nudge effectiveness: The role of habits in a portion size nudge, a proof of concept study.
Appetite 2020, 151, 104699, doi:10.1016/j.appet.2020.104699.
35. Waterlander, W.E. Are Front-of-Pack Nutrition Labels the Silver Bullet for Achieving Healthier
Population Diets? Am J Public Health 2019, 109, 1067-1068, doi:10.2105/ajph.2019.305169.
36. Stuber, J.M.; Middel, C.N.H.; Mackenbach, J.D.; Beulens, J.W.J.; Lakerveld, J. Successfully
Recruiting Adults with a Low Socioeconomic Position into Community-Based Lifestyle
Programs: A Qualitative Study on Expert Opinions. Int J Environ Res Public Health 2020,
17, doi:10.3390/ijerph17082764.

8
(COREQ): a 32-item checklist for interviews and focus groups. International Journal for
Quality in Health Care 2007, 19, 349-357, doi:10.1093/intqhc/mzm042.
23. Dierckx de Casterlé, B.; Gastmans, C.; Bryon, E.; Denier, Y. QUAGOL: A guide for qualitative
data analysis. International Journal of Nursing Studies 2012, 49, 360-371, doi:https://doi.
org/10.1016/j.ijnurstu.2011.09.012.
24. Dahlgren, G.; Whitehead, M. Policies and strategies to promote social equity in health.
Background document to WHO - Strategy paper for Europe; 1991; Volume 14.
25. Contento, I. Overview of determinants of food choice and dietary change: Implications for
nutrition education. Nutrition Education: Linking research, theory and practice 2011, 26-42.
26. Kamphuis, C.B.; de Bekker-Grob, E.W.; van Lenthe, F.J. Factors affecting food choices of
older adults from high and low socioeconomic groups: a discrete choice experiment. Am
J Clin Nutr 2015, 101, 768-774, doi:10.3945/ajcn.114.096776.
27. Qian, F.; Riddle, M.C.; Wylie-Rosett, J.; Hu, F.B. Red and Processed Meats and Health Risks:
How Strong Is the Evidence? Diabetes Care 2020, 43, 265-271, doi:10.2337/dci19-0063.

214 215
216
CHAPTER 8 – SUPPLEMENTARY FILES
chapter

Supplementary File 1. Topic list

8

Topic Interviewing questions

1. Account of daily life GENERAL
- How would you rate your life at this moment in time?
- Which factors in your life contribute to this? (e.g. physical/mental health, financial, work, social life)

COMMUNITY
- Can you describe your living situation?
- Do you have close relations with family / friends? What do they mean to you?
- How often do you meet up with them?
- When do you meet up with them?
- What do you do when you meet up with them? Why?

ACTIVITIES [pick day before interview as guidance]

Daily schedule
- What time did you get up? Why at that time?
- What did you do once you got out of bed?
- Did you have breakfast? Why (not)? What did you have for breakfast? With whom?
- At what time did you leave home?
- When leaving home, did you bring some food with you? Why (not)?
- Did you have lunch? At what time? What did you have? Where do you get your lunch?
- At what time did you get home?
- Did you have dinner? Why (not)? What did you have for dinner? With whom? Why?

Food acquisition
- Did you acquire any food that day?
- Where do you acquire food (e.g. supermarket, market)? Why there? Do you know the employees there?
- When do you normally acquire food? Weekly, daily?
- Do you enjoy acquiring food? Why (not)?
- How do you acquire food? Do you use of a grocery list? Do you consider promotions?
- Can you describe how a shopping trip looks like?

Work
- How did you get to work? Why do you choose that mode of transportation?
- Can you describe the type of work you do?
- Can you describe your working pattern? Is it constant or irregular?
- Do you enjoy your work? Why (not)?

Leisure time
- Do you have any spare time? When do you have spare time?
- What do you like to do in your spare time? With whom?
- Why do you enjoy that?

Supplementary File 1. Topic list (continued)

Topic Interviewing questions

2. Account of dietary - How do you value a healthy diet? Is it important to you?
behaviours - What does a healthy diet mean to you?
- Do you consider yourself to have a healthy diet?
- Why do you eat what you eat?
• Health effects
• Taste
• Price
• Social network
• Habit
• Emotions
• Cultural influences
• Special occasions
• Stress
3. Enablers and barriers - What makes it easier for you to eat healthily? Why?
Probe, factors related to:
• Individual
• Lifestyle
• Community
• Local economy
• Activities
• Built environment
• Natural environment

- What makes it difficult for you to eat healthily? Why?

Probe, factors related to:
• Individual
• Lifestyle
• Community
• Local economy
• Activities
• Built environment
• Natural environment
217
determinants of food choice and perceptions of supermarket - based nudging interventions

8
218
Supplementary File 1. Topic list (continued)
chapter

Topic Interviewing questions

8

4. Needs for the The first part should really anticipate on previous answers of the interviewee. Anticipate on previously mentioned aspects of
promotion of healthy daily routine; factors that influence eating behaviours (e.g. price, taste, etc.); and barriers that make it difficult to eat healthily.
dietary behaviour • You previously indicated … What could help you to …?

For the second part, we present the interviewee with graphical cards of possible interventions in the supermarket that promote
a healthy lifestyle (e.g. including visualized nudges and pricing interventions).
• Pick one card which appeals to you to the most. Can you describe which elements appeal to you? And why do they
appeal to you?
• Pick one card which doesn’t appeal to you at all. Can you describe which elements you don’t like about it? What
would be the conditions that would make it more appealing to you?

Supplementary File 2. Visual examples nudges

219
determinants of food choice and perceptions of supermarket - based nudging interventions

8
9
In this thesis I highlighted the challenges of healthy and sustainable dietary patterns in
the 21st century. In the first part of this thesis, the focus was on the perspective of the
individual and I explored the association between various healthy and/or sustainable
dietary patterns with cardiovascular outcomes, mortality, and sustainability indicators.
In the second part of this thesis, focusing on the societal perspective, I explored how the
food environment affects dietary behaviour and health outcomes, and how adults with
low SEP perceive interventions in the food environment. In this chapter, I will provide a
summary of the main findings presented in this thesis, and put the findings in a broader
perspective by reflecting on them. Also, I will reflect on the research methodology and
provide recommendations for future research and policy.

9.1. HEALTHY AND SUSTAINABLE DIETS FROM AN

INDIVIDUAL PERSPECTIVE

9.1.1 Diet and heart failure

Although dietary modification – such as sodium and fluid restriction – is well established
in treatment of heart failure (1, 2), the potential of healthy diets for the primary
prevention of heart failure has received far less attention (3). In Chapter 1 and 2, we
therefore examined the association of adherence to the Dutch dietary guidelines and the
DASH diet with risk of heart failure. Using data from approximately 37,500 Dutch adults
participating in the EPIC-NL study, we observed that high adherence to these dietary
patterns was associated with a 22 to 27% lower risk of heart failure, which is well in line
with evidence from previous observational and experimental studies (4).

Still, a particular knowledge gap persists with regard to the effects of healthy diets on
heart failure subtypes. Heart failure with reduced ejection fraction (HFrEF) and heart
failure with preserved ejection fraction (HFpEF) have a distinct pathophysiology (5)
and different underlying risk-factors (6-8). Coronary heart disease is an important risk-
factor for HFrEF, whereas more systemic factors such as hypertension, type 2 diabetes,
and obesity are important risk-factors for HFpEF. Therefore, it could be hypothesized

General Discussion that healthy diets may have differential effects on HF subtypes, by influencing different
underlying risk-factors. For example, adherence to the DASH may be particularly
efficacious for the prevention of HFpEF through its effects on hypertension (9).
Preliminary evidence substantiating this hypothesis comes from the Women’s Health
Initiative, a prospective cohort study including an ethnically diverse population of
postmenopausal women from the US. In an analysis covering 118,057 women, calibrated
dietary sodium – one of the eight components included in the DASH diet – was indeed
more strongly associated with risk of HFpEF (HR: 2.87; 95% CI: 2.29 – 3.60) than with
chapter 9 general discussion
HFrEF (HR: 1.71; 95% CI: 1.28 – 2.30) (10). Additionally, adherence to the DASH diet has
been cross-sectionally associated with improved diastolic function in men and women
free of cardiovascular disease (11). In contrast, in a US cohort including 16,068 ethnically
diverse adults free from coronary heart disease at baseline, associations of data-driven
dietary patterns with risk of HFrEF and HFpEF were not materially different (12). However,
the number of cases was relatively small (n=157 for HFrEF and n=133 for HFpEF; 73 cases
with intermediate or undocumented ejection fraction), and data-derived dietary patterns
may not represent diets that are optimal for cardiovascular health.
Thus, further research examining the association of healthy diets with heart failure
subtypes is warranted. However, accurate diagnosis of heart failure subtype is challenging,
as the signs and symptoms are often non-specific and need to be corroborated with
echocardiographic measures which often miss (early-stage) HFpEF diagnoses (13, 14).
Such echocardiographic data are often not available in epidemiological cohort studies,
which frequently rely on disease and mortality registries for heart failure ascertainment
and thus depend on availability of such detailed coding in these registers (15). In the
registers used for follow-up of EPIC-NL, only generic ICD codes were available. Therefore,
the application of machine learning techniques to predict heart failure subtype from
routine clinical data may facilitate research on diet and heart failure subtypes. A previous
algorithm developed in the Swedish Heart Failure Registry showed good accuracy in
predicting HFpEF and HFrEF (16). Additionally, the incorporation of multidimensional
measurements resulting from cardiac imaging with other biological and clinical data
may aid improved taxonomical classifications in the clinical setting (13, 17).
9.1.2 Incorporating sustainability in food-based dietary guidelines
In Chapter 3, we studied adherence to the EAT-Lancet Healthy Reference Diet in
association with coronary heart disease (CHD), all-cause mortality and environmental
indicators. High adherence to the Healthy Reference Diet was associated with 15% lower
risk of CHD and 17% lower risk of all-cause mortality. Additionally, high adherence to
the Healthy Reference Diet was associated with lower environmental impact from
greenhouse gas emissions, land use, freshwater eutrophication, marine eutrophication,
and terrestrial acidification, but with higher environmental impact from blue water use.
In Chapter 4, we examined adherence to the Healthy Reference Diet in association with
CHD and stroke using data from the European-wide EPIC cohort. Similarly, adherence
to the Healthy Reference Diet was associated with 6% lower risk of CHD and also with
24% lower risk of stroke. Our findings further contribute to the evidence base that
predominantly plant-based dietary patterns are associated with cardiovascular health
benefits (18-20) and lower environmental impact (21).
222
Until now, food-based dietary guidelines have been mainly formulated from a public
health perspective (22). However, given the urgency of the climate crisis, there is
increasing recognition that sustainability is also an important issue to be taken
into account (23). Additionally, following the definition of the Food and Agriculture
Organization, sustainable diets should not only be healthy and have little environmental
impact, but also be nutritionally adequate, safe, culturally acceptable, and economically
affordable (24). Therefore, the question of what should be recommended in food-based
dietary guidelines is a complex multidimensional challenge as the various objectives
(e.g., sustainability, nutritional adequacy, cultural acceptability, and affordability) may
not always converge. An illustrative example of such a trade-off between health and
sustainability is the recommendation to consume (fatty) fish regularly (25). Fatty fish
consumption has been associated with lower risk of cardiovascular disease and other
health conditions (26, 27). At the same time, a global increase in fish demand has
negatively affected marine biodiversity (28).
So how can this multidimensional challenge be addressed? It is evident that with classical
epidemiological methods – as applied in the first part of this thesis – we can only
address the bidirectional questions relating to diet–health associations. However, they
cannot handle additional parameters (e.g., sustainability, nutritional adequacy, cultural
acceptability), nor identify synergies and trade-offs between the various objectives.
In this regard, mathematical optimization techniques – originating from operations
research – have gained popularity as they allow to handle the multidimensional
complexity of the problem (29). Diet optimization models provide ‘optimal’ diets (e.g., a
set of foods) that minimize or maximize an objective function while satisfying predefined
constraints (29). For example, a study by Perignon et al. modelled a sustainable and
culturally acceptable diet by minimizing deviation from the observed diet (objective
function), while imposing constraints on macro- and micronutrients (e.g., diets must
contain ≥2.4 mg of vitamin B12) and on the level of greenhouse gas emissions (e.g.,
modelling incremental reductions of 10%). In this study it was found that 30% reduction
in greenhouse gas emissions could be achieved without major shifts from observed 9
diets, while at the same time maintaining nutritional adequacy (30). However, achieving
>30% reductions in greenhouse gas emissions could only be reached at the cost of either
cultural acceptability or nutritional adequacy. For example, more stringent constraints
on greenhouse gas emissions required lower intakes of meat, fish and eggs and tended
to decrease the mean adequacy ratio – a composite measure of nutritional adequacy.
As such, diet optimization studies elegantly integrate the various dimensions at stake,
and highlight synergies and trade-offs between these dimensions (31).
223
chapter 9 general discussion
Previously, Brink et al. applied diet optimization models to derive healthy and sustainable
food-based dietary guidelines for various age and sex groups in the Netherlands (32).
In this study, diets were optimized on cultural acceptability, i.e. minimizing deviation
from current Dutch diets as measured in the Dutch National Food Consumption
Survey. The current Dutch food-based dietary guidelines, subgroup-specific nutrient
requirements, and environmental considerations were as imposed constraints. This
resulted in several optimized diets for the different age and sex groups. Interesting to
note is that recommended intakes in the optimized diets diverged for some food groups.
For example, the recommended intake for vegetables ranged between 200 g/day for
males in de age category 19-30 years old to 700 g/day for females in the age category
31-50 years old. In light of effective communication of dietary guidelines, it may be of
concern whether diverging recommendations cause confusion amongst the public. On
the other hand, providing target-group specific guidelines may be a valid approach given
distinct nutritional requirements along the life course. Additionally, the recommended
intakes from the model represent significant challenges for behavioural change, despite
the fact that cultural acceptability was taken into account in the optimization model. To
illustrate: the recommended intake of red meat from the diet optimization model by
Brink et al. for male adults ranged between 26 g/day (31-50 years old) and 70 g/day (≥ 70
years old). In contrast, current consumption of red meat among male adults according
to the Dutch national food consumption survey is 92 g/day (33), indicating that red meat
intake should be reduced with no less than 24-72%.
Despite the versatility of diet optimization models to account for the various dimensions of
healthy and sustainable diets, there also some limitations that need to be taken into account.
Most notably, the validity of diet optimization models depends on the decisions that are
made with regard to specification of the objective function and constraints (34). For example,
modeling cultural acceptability as minimal deviation of current diets is a model assumption
which could be questioned. Additionally, the availability of reliable datasets on sustainability
indicators other than the commonly used greenhouse gas emissions and land use (e.g.,
biodiversity) on both national and international level is of importance to comprehensively
model diets with regard to environmental sustainability (35).
To conclude, providing dietary recommendations on sustainable dietary patterns is a
balancing act as multiple dimensions need to be taken into account. Diet optimization
models may offer insights into synergies and trade-offs between the various
dimensions, and shed light on recommended intakes for key food groups. Despite
scientific challenges in defining healthy and sustainable diets, the overall direction
seems clear: Western-style diets should shift towards more plant-based diets, which
– despite not fully eliminating animal-sourced foods from diets – still constitutes a big
224
challenge. Some countries (e.g., Sweden, Brazil) have already integrated sustainability
in their food-based dietary guidelines on a rather generic level, by emphasizing the
importance of consuming predominantly plant-based foods and highlighting the high
environmental impact from meat (36). However, the guidelines lack specificity as the
guidelines themselves remain primarily health-oriented. In some other cases (e.g.,
USA), efforts to integrate sustainability in the food-based dietary guidelines have failed
due to absence of governmental endorsement. At the same time, establishing food-
based dietary guidelines incorporating sustainability are not enough, and should be
backed up by congruent policy promoting healthy and sustainable food choices. A
recent benchmarking study investigating policy measures aimed at improving the food
environment in the Netherlands showed that such measures (e.g., taxing unhealthy
foods, restricting unhealthy food marketing) are largely absent, highlighting significant
room for improvement (37). Therefore, in the next part of this discussion, I will reflect
on population-based approaches that may aid in promoting healthy and sustainable
dietary choices.
9.1.3 Methodological considerations in the field of nutritional epidemiology
The first part of thesis relied on analyses of diet-health associations in observational
studies, which are characterized by various challenges. The first challenge in the field
of nutritional epidemiology is the reliance on observational study designs, which are
generally prone to information, selection, confounding and reverse causation bias.
The golden standard for inferring causality – the randomized controlled trial (RCT)
– is often infeasible, invalid, or not generalizable (38, 39). Even if an RCT is possible,
short follow-up durations on mostly intermediate disease markers, inability of blinding
participants, high dropout rates, non-compliance because of the difficulties associated
with changing dietary behaviour, and determination of a ‘control’ group, may all result
in similar biases as compared to observational designs (40). New approaches such as
target trial emulation may improve causal inference from observational research (41-43).
In target trial emulation, researchers describe the hypothetical target trial that would
answer the causal question of interest. Subsequently, the trial is emulated based on 9
observational data. For example, random assignment of treatment and control groups
may be emulated by using repeated measures of dietary intake in a observational cohort
study. Participants having stable diets at t1 (diet X) and t2 (diet X), may be assigned to
the ‘control diet’. Participants that switch diets in an a priori defined direction between
t1 (diet X) and t2 (diet Y), may be assigned to the treatment group. As diet change may
be driven by several factors that also influence the outcome of interest, adjustment for
those variables serves to mimic randomization. Trial emulation is no panacea for causal
inference, e.g., unmeasured confounding bias still remains a problem. However, it can
serve as an intermediate step between observational and RCT research by identifying
225
chapter 9 general discussion
promising dietary interventions (43). Additionally, the availability of repeated food
frequency questionnaires in cohort studies is more likely to be an exception than the
rule, as they are relatively burdensome to undertake, which may hamper uptake of trial
emulation designs.
A second challenge in the field of nutritional epidemiology is the valid and reliable
assessment of dietary intake (44, 45). In this thesis, we primarily used validated food
frequency questionnaires (FFQs), which are suitable for assessing habitual intake and
ranking participants on intake. However, just like any other dietary assessment method,
FFQs are prone to recall bias and systematic over- and underreporting, which may affect
risk estimates in epidemiological studies (46, 47). Therefore, innovations in dietary
assessment methods that circumvent reliance on memory, and improve the validity
and reliability of dietary data would advance the field. Image-based dietary assessment
methods, which asses dietary intake through photos or videos and are complemented
with artificial intelligence technology to allow for dietary intake calculations, are promising
avenues of research (48). Still, challenges with regard to accurate volume estimation and
application in real-life settings opposed to lab settings must be addressed, before such
methods can be readily applied in large-scale epidemiological studies (49, 50).
Lastly, the application of metabolomics may further advance the field of nutritional
epidemiology. First, metabolomics data may corroborate self-reported dietary intake
data, which have inherent limitations as outline previously. Although metabolomics data
have the advantage of being objective measurements, they are often not food-specific
and/or (partly) reflect metabolism rather than intake. Therefore they cannot be used
to quantify dietary intake, but can be useful as measures of dietary compliance (51).
Secondly, the field of metabolomics may contribute to further understanding the effects
of diets on disease risk (51). Although it has been hypothesized that enhanced glycemic
control, improved lipid profile, reduced blood pressure, decreased inflammation, and
improved composition of the gut microbiome play a role in lowering CVD risk (52,
53), the exact mechanism underlying such associations largely remains a ‘black box’.
Understanding the biological pathways by which diet may affect CVD risk is important as
biological plausibility is one of the Bradford-Hill criteria for evaluating whether observed
associations may be causal (54). Additionally, inclusion of omics technology may shed
light on intra-person variability in responses to dietary patterns, allowing for precision
nutrition approaches (55).
In conclusion, the field of nutritional epidemiology has provided important insights
on diet-health associations, which have a pivotal role in informing food-based dietary
guidelines. Still, significant scientific challenges remain, including improvement of causal
226
inference based on observational data, improving dietary assessment methods, and
leveraging data from the metabolome for better understanding the relation between
diet and health.
9.2. HEALTHY AND SUSTAINABLE DIETS FROM A
SOCIETAL PERSPECTIVE
9.2.1 Complexity: a challenge in food environment research
In Chapter 6, we systematically reviewed the literature on the effectiveness of nudges
in promoting healthy dietary choices within real-life food purchasing environments.
Additionally, we explored whether there was evidence for a moderating role of SEP. We
observed that the majority of the evidence-base focused on the effects of information
and position nudges, while other types of nudges (e.g., availability, presentation, size)
were less often studied. Information and position nudges showed modest tendencies
towards beneficial effects, although findings remain qualitative due to the substantial
heterogeneity in underlying evidence encountered, prohibiting meta-analysis of findings.
Additionally, only a few studies examined the effects of nudges across levels of SEP.
Although there were some indications that effects were more pronounced among those
with low SEP, the scarcity of evidence and the use of different proxies of SEP made it
difficult to draw conclusions. In Chapter 7, we studied relative exposure to fast-food
restaurants (FFR) in relation to diet quality, overweight, and obesity using data from the
EPIC-NL study. We observed that high exposure to FFR (e.g., proportion of FFR to total
food outlets of 30%) in a radius of 400m, 1000m, or 1400m around the home was not
associated with scores on the Dutch Health Diet 2015 index or weight status.
These findings add to the mixed bag of evidence in food environment research. From
the totality of evidence on nudging strategies it seems that nudging may have a small
effect on food choice (56-62), but it remains difficult to synthesize and quantify their
effects in real-life settings. Additionally, it remains unclear how (some types of) nudges 9
affect health inequalities (63). At the same time, methodological limitations hamper
our ability to validly assess the association between neighborhood environments and
health outcomes. Then, how should we move forward in food environment research?
A common challenge in chapter 6 and 7 was the (in)ability to capture the real-world
complexity of how our food environment affects dietary choices. With regard to
Chapter 6, the nudges we studied can be regarded as ‘complex interventions’ (64).
Given their application in real-life settings, contextual factors such as type of food
store, characteristics of its visitors and other competing influences such as pricing
227
chapter 9 general discussion
may have interacted with the intervention, producing context-sensitive findings which Table 1. Comparison of evidence-based medicine paradigm and the complex system paradigm
[Adapted from Greenhalgh (76)]
cannot be readily integrated in meta-analyses (64, 65). For example, the qualitative
findings in Chapter 8 highlighted that people may distrust supermarket-engagement Evidence-based medicine
paradigm
in health-promoting initiatives, suspecting ulterior motives. We also showed that when
Perspective on Singular, independent of the
nudges aligned with product preferences, information needs and beliefs about the food scientific truth observer, ascertainable through inquiry and perspective taken
empirical inquiry
environment, interventions were perceived more positively. These findings indicate that
Goal of research Establishing the truth; finding
attitudes to interventions may vary, and may therefore influence the effects we observe. more or less generalizable
solutions to well-defined
Similarly, the pathways through which neighborhood food environments may influence problems
dietary choices in Chapter 7 may be equally complex and dynamic, with factors on
Assumed model of Linear, cause-and-
various levels (e.g., individual, behavioural, environmental) interacting in non-linear ways causality effect causality (perhaps
incorporating mediators and
to produce population diets (66). In response to this complexity that is inherent to food moderators)
environment research and public health research in general, a paradigm shift has been Typical format of “What is the effect size of the
research question intervention on the predefined
advocated, moving from the evidence-based medicine paradigm towards a complex outcome, and is it statistically
systems paradigm (Table 1) (64, 65, 67, 68). Viewed from a complex systems perspective, significant?”
one might argue that there is no singular effect of food environments on outcomes,
Mode of Attempt to represent science in
and that such effects very much depend on other elements of the system. As such, the representation one authoritative voice
emphasis within the complex systems paradigm is on the interacting factors that give
Good research is Methodological ‘rigor’, i.e.
rise to a certain outcome, rather than viewing those factors as having fixed effect-sizes. characterized by strict application of structured
and standardized design,
conventional approaches to
9.2.1 Capturing complexity in food environment research methodology generalizability and validity
Adopting the complex systems paradigm has a great influence on our research approach. Purpose of Disjunctive: simplification and
theorizing abstraction; breaking problems
In the first place, the paradigm shift has implications for the research questions that we down into analyzable parts
pose. More specifically, instead of asking ‘How does A influence B?’, we are interested
Approach to data Research should continue until
in ‘How do A, B, C, D relate and give rise to Y?’ (Table 1). As such, there is a much data collection is complete
more prominent role for ‘extraneous’ factors that traditionally are cancelled out in
epidemiological research through randomization or statistical adjustments. In the second
Analytic focus Dualisms: influence of X on Y
place, this paradigm shift requires different methodology. An interesting approach in
this regard is the use of agent-based models, as they may provide novel insights into
the conditions that contribute to the effectiveness of population-based interventions
(69). Agent-based models are simulation frameworks in which agents (e.g., individuals,
food outlets) make decisions and pursue goals (e.g., health, economic gains) according to
decisions heuristics (70). They can account for the real-world complexity of food-choice
decision making by incorporating various characteristics of agents such as age, sex,
SEP, environments and social networks (71). Using empirical data from epidemiological
studies to inform parameter estimates, interactions between the various agents can
be modelled to show emerging patterns in outcomes of interest, for example policy
effectiveness (69).
228
Complex systems paradigm
Multiple, influenced by mode of
Exploring tensions; generating
insights and wisdom, exposing
multiple perspectives, viewing
complex systems as moving targets
Emergent causality: multiple
interacting influences account for
a particular outcome but none can
be said to have a fixed ‘effect size’
“What combination of influences
has generated this phenomenon?
What happens to the system and
its actors if intervene in a particular
way?”
Attempt to illustrate the plurality of
voices inherent in the research and
phenomena under study
Strong theory, flexible methods,
pragmatic adaptation to emerging
circumstances, contribution to
generative learning and theoretical
transferability
Conjunctive: drawing parts of the
problem together to produce a rich,
nuanced picture of what is going on
and why
Data will never be complete of
perfect, decisions often need to
be made despite incomplete or
contested data
Dualities: inter-relationships and
dynamic tension between A, B, C,
and other emergent aspects
9
229
chapter 9 general discussion
Although studies have used agent-based modelling to evaluate the effectiveness of
pricing strategies (72, 73) and food access (74), no studies have yet applied agent-
based models to the domain of healthy food nudging. Interestingly, a previous study
applied agent-based modeling to studying the effectiveness of a social norm nudge on
promoting recycling behaviour, and found that the amount of waste present on the
streets moderated the effectiveness of the social norm (75), illustrating the applicability
of agent-based modelling in the nudging domain.
Another interesting approach focused on unravelling the complex dynamic nature of
neighborhood food environments is causal loop diagramming, which visualizes the
complex interplay of factors that give rise to a certain outcome within a system (77).
For example, Sawyer et al. created a causal loop diagram by systematically mapping
quantitative and qualitative evidence on the determinants – and associations between
these determinants – of dietary intake and food environments in low income groups
(66). Several causal loop diagrams were identified visualizing the 60 determinants (e.g.
‘elements’), their connections, and existing feedback loops. As such, these systems maps
may contribute to understanding how systems produce certain outcomes (e.g. unhealthy
diets), and provide insight into the promising entry points for intervention (66).
At last, the complexity perspective also has implications for the way we address evidence
synthesis in food environment research. The clearly defined ‘PICO’ question that is
central to most systematic reviews does not seem to be compatible with the complex
systems paradigm. Therefore, it has been suggested to reframe questions in evidence
synthesis not it terms of ‘what works’, but rather in terms of ‘what happens’ (67). This
means that systematic reviews of complex interventions should not rely on ‘frequentist’
hypothesis-testing (does it work: yes or no?), but rather answer a broader question:
“What has happened previously when this intervention been implemented across a range
of contexts, populations and subpopulations, and how have those effects come about?”
(Petticrew, 2015, p.2) (67). In answering this question, the use and integration of different
types of evidence is needed, as they may act like the pieces of a puzzle, and together
contribute to a better understanding of the intervention and system as a whole (67).
To conclude, research methodology stemming from the evidence-based medicine
paradigm is not well-suited to address questions relating to the effects of the food
environment on food choice and health outcomes. Rather, we need research methods
that allow to take into account the complexity of the food environment. Agent-based
models and causal loop diagramming are two approaches that aid in doing so.
Additionally, the central focus of systematic reviews in the food environment research
domain should shift from focusing on ‘what works’ to ‘what happens’.
230
9.3 RECOMMENDATIONS FOR FUTURE RESEARCH
Based on the findings from this thesis, I will provide several main recommendations
for future research. First, more research is needed to explore the association between
healthy dietary patterns and risk of heart failure subtype in observational cohort studies.
In this regard, the use of innovative methods in nutritional epidemiology, including
target trial emulation and the application of metabolomics, may aid in improving causal
inference from these types of studies and deepen our mechanistic understanding,
respectively. Second, more research is needed to explore how sustainability can be
integrated in food-based dietary guidelines. It is especially important to further elucidate
the role of animal-sourced protein and what would constitute optimal intake, as these
recommendations may have implications for sustainability, nutritional adequacy, and
cultural acceptability. Diet optimization methods may be appropriate methods to do so.
Also – in line with the focus on health inequalities – it is of importance to further explore
the affordability of healthy and sustainable diets. Third, food environment research
should move away from ‘linear thinking’ and embrace the complexity of lived reality.
Therefore, new research approaches such as the application of agent-based models and
causal loop diagramming, may be prioritized above observational research methodology
in studying research questions relating to the food environment.
9.4 CONSIDERATIONS FROM A POLICY VIEWPOINT
The challenges that are faced in food environment research should not be an excuse to
refrain from implementing policy measures tackling the unhealthy food environment.
In fact, framing the causes of poor diets as complex – while essential from a scientific
perspective – may act as a discursive strategy justifying inaction by governments
and food industry (78). The undisputed amount of evidence showing that current
diets are associated with a significant disease burden and environmental impact,
and acknowledging that the drivers of poor diets operate dynamically within complex
systems, justifies policy action at various levels. Therefore, a broad pallet of measures, 9
including nudging strategies, taxing unhealthy foods and subsidizing healthy foods,
imposing restrictions on marketing of unhealthy and highly palatable foods, food
reformulation, and expanding legal possibilities that facilitate regulation of food outlet
availability are likely to bring about positive changes in the food system (37, 79).
It is promising that at the moment this thesis is written the Dutch government announced
intentions to introduce a tax on sugar-sweetened beverages and make binding
agreements with food industry regarding food reformulation (80). At the same time,
these population-based approaches should be complemented with long-term cross-
231
chapter 9 general discussion
domain policy measures aimed at tackling the structural societal causes of health
inequalities (e.g., poverty) (79). Such policy-measures may require a bigger role for local
governments, as they can best identify which causes of health inequalities are relevant
on local level and identify appropriate responses. Targeted lifestyle-interventions may
be used as complementary approaches, but do risk being only effective for a selective
part of the population, namely those with higher SEP (79).
9.5 GENERAL CONCLUSION
On an individual level, consuming dietary patterns rich in plant-based foods and
including moderate amounts of animal-sourced foods may benefit both cardiovascular
and planetary health. The food environment is an important determinant of food
choice, but it remains difficult to quantify its effects. The complex systems perspective
acknowledges that causes for poor diets are multifactorial and dynamic, stressing the
need for policy action on multiple levels. As such, the food environment can be leveraged
as entry-point for promoting healthy and sustainable food choices at a societal level.
232
REFERENCES
1. McDonagh TA, Metra M, Adamo M, Gardner RS, Baumbach A, Böhm M, et al. 2021 ESC
Guidelines for the diagnosis and treatment of acute and chronic heart failure. Eur Heart
J. 2021;42(36):3599-726.
2. Bozkurt B, Hershberger RE, Butler J, Grady KL, Heidenreich PA, Isler ML, et al. 2021 ACC/
AHA Key Data Elements and Definitions for Heart Failure: A Report of the American
College of Cardiology/American Heart Association Task Force on Clinical Data Standards
(Writing Committee to Develop Clinical Data Standards for Heart Failure). Circulation:
Cardiovascular Quality and Outcomes. 2021;14(4):e000102.
3. Wang DD. Dietary Patterns and Precision Prevention of Heart Failure. J Am Coll Cardiol.
2019;73(16):2046-8.
4. Kerley CP. Dietary patterns and components to prevent and treat heart failure: a
comprehensive review of human studies. Nutr Res Rev. 2019;32(1):1-27.
5. Kemp CD, Conte JV. The pathophysiology of heart failure. Cardiovascular Pathology.
2012;21(5):365-71.
6. Lee MP, Glynn RJ, Schneeweiss S, Lin KJ, Patorno E, Barberio J, et al. Risk Factors for Heart
Failure with Preserved or Reduced Ejection Fraction Among Medicare Beneficiaries:
Application of Competing Risks Analysis and Gradient Boosted Model. Clin Epidemiol.
2020;12:607-16.
7. Ho JE, Lyass A, Lee DS, Vasan RS, Kannel WB, Larson MG, et al. Predictors of New-Onset
Heart Failure. Circulation: Heart Failure. 2013;6(2):279-86.
8. Gong FF, Jelinek MV, Castro JM, Coller JM, McGrady M, Boffa U, et al. Risk factors for incident
heart failure with preserved or reduced ejection fraction, and valvular heart failure, in a
community-based cohort. Open Heart. 2018;5(2):e000782.
9. Filippou CD, Tsioufis CP, Thomopoulos CG, Mihas CC, Dimitriadis KS, Sotiropoulou LI, et
al. Dietary Approaches to Stop Hypertension (DASH) Diet and Blood Pressure Reduction
in Adults with and without Hypertension: A Systematic Review and Meta-Analysis of
Randomized Controlled Trials. Adv Nutr. 2020;11(5):1150-60.
10. Gu X, Tinker LF, Field AE, Roberts MB, Allison M, Haring B, et al. Abstract P170: Associations
of Calibrated Dietary Sodium and Potassium Intakes With the Risk of Heart Failure and
its Subtypes in Postmenopausal Women: The Women’s Health Initiative. Circulation.
2018;137(suppl_1):AP170-AP.
11. Nguyen HT, Bertoni AG, Nettleton JA, Bluemke DA, Levitan EB, Burke GL. DASH eating
pattern is associated with favorable left ventricular function in the multi-ethnic study of 9
atherosclerosis. J Am Coll Nutr. 2012;31(6):401-7.
12. Lara KM, Levitan EB, Gutierrez OM, Shikany JM, Safford MM, Judd SE, et al. Dietary Patterns
and Incident Heart Failure in U.S. Adults Without Known Coronary Disease. J Am Coll
Cardiol. 2019;73(16):2036-45.
13. Omar AMS, Bansal M, Sengupta PP. Advances in Echocardiographic Imaging in Heart Failure
With Reduced and Preserved Ejection Fraction. Circulation Research. 2016;119(2):357-74.
14. Huis In ‘t Veld AE, de Man FS, van Rossum AC, Handoko ML. How to diagnose heart
failure with preserved ejection fraction: the value of invasive stress testing. Neth Heart
J. 2016;24(4):244-51.
233
chapter 9 general discussion
15. Brouwers FP, Hillege HL, van Gilst WH, van Veldhuisen DJ. Comparing New Onset Heart
Failure with Reduced Ejection Fraction and New Onset Heart Failure with Preserved
Ejection Fraction: An Epidemiologic Perspective. Curr Heart Fail Rep. 2012;9(4):363-8.
16. Uijl A, Lund LH, Vaartjes I, Brugts JJ, Linssen GC, Asselbergs FW, et al. A registry-based
algorithm to predict ejection fraction in patients with heart failure. ESC Heart Fail.
2020;7(5):2388-97.
17. Banerjee A, Chen S, Fatemifar G, Zeina M, Lumbers RT, Mielke J, et al. Machine learning for
subtype definition and risk prediction in heart failure, acute coronary syndromes and atrial
fibrillation: systematic review of validity and clinical utility. BMC Medicine. 2021;19(1):85.
18. Gan ZH, Cheong HC, Tu YK, Kuo PH. Association between Plant-Based Dietary Patterns
and Risk of Cardiovascular Disease: A Systematic Review and Meta-Analysis of Prospective
Cohort Studies. Nutrients. 2021;13(11).
19. Quek J, Lim G, Lim WH, Ng CH, So WZ, Toh J, et al. The Association of Plant-Based Diet With
Cardiovascular Disease and Mortality: A Meta-Analysis and Systematic Review of Prospect
Cohort Studies. Front Cardiovasc Med. 2021;8:756810.
20. Gibbs J, Gaskin E, Ji C, Miller MA, Cappuccio FP. The effect of plant-based dietary patterns
on blood pressure: a systematic review and meta-analysis of controlled intervention trials.
J Hypertens. 2021;39(1):23-37.
21. Aleksandrowicz L, Green R, Joy EJM, Smith P, Haines A. The Impacts of Dietary Change on
Greenhouse Gas Emissions, Land Use, Water Use, and Health: A Systematic Review. PLOS
ONE. 2016;11(11):e0165797.
22. Herforth A, Arimond M, Álvarez-Sánchez C, Coates J, Christianson K, Muehlhoff E. A Global
Review of Food-Based Dietary Guidelines. Advances in Nutrition. 2019;10(4):590-605.
23. Mazac R, Renwick K, Seed B, Black JL. An Approach for Integrating and Analyzing
Sustainability in Food-Based Dietary Guidelines. Frontiers in Sustainable Food Systems.
2021;5(84).
24. Burlingame B, Dernini S. Sustainable diets and biodiversity directions and solutions for
policy, research and action: FAO Headquarters, Rome; 2012.
25. Kromhout D, Spaaij CJK, de Goede J, Weggemans RM, for the Committee Dutch Dietary
G. The 2015 Dutch food-based dietary guidelines. European Journal of Clinical Nutrition.
2016;70(8):869-78.
26. Mohan D, Mente A, Dehghan M, Rangarajan S, O’Donnell M, Hu W, et al. Associations of Fish
Consumption With Risk of Cardiovascular Disease and Mortality Among Individuals With or
Without Vascular Disease From 58 Countries. JAMA Internal Medicine. 2021;181(5):631-49.
27. Jayedi A, Shab-Bidar S. Fish Consumption and the Risk of Chronic Disease: An Umbrella
Review of Meta-Analyses of Prospective Cohort Studies. Advances in Nutrition.
2020;11(5):1123-33.
28. Thurstan RH, Roberts CM. The past and future of fish consumption: Can supplies meet
healthy eating recommendations? Marine Pollution Bulletin. 2014;89(1):5-11.
29. Gazan R, Brouzes CMC, Vieux F, Maillot M, Lluch A, Darmon N. Mathematical Optimization
to Explore Tomorrow’s Sustainable Diets: A Narrative Review. Advances in Nutrition.
2018;9(5):602-16.
30. Perignon M, Masset G, Ferrari G, Barré T, Vieux F, Maillot M, et al. How low can dietary
greenhouse gas emissions be reduced without impairing nutritional adequacy, affordability
and acceptability of the diet? A modelling study to guide sustainable food choices. Public
Health Nutr. 2016;19(14):2662-74.
234
31. Wilson N, Cleghorn CL, Cobiac LJ, Mizdrak A, Nghiem N. Achieving Healthy and Sustainable
Diets: A Review of the Results of Recent Mathematical Optimization Studies. Adv Nutr.
2019;10(Suppl_4):S389-s403.
32. Brink E, van Rossum C, Postma-Smeets A, Stafleu A, Wolvers D, van Dooren C, et
al. Development of healthy and sustainable food-based dietary guidelines for the
Netherlands. Public Health Nutrition. 2019;22(13):2419-35.
33. Van Rossum C, Buurma-Rethans E, Dinnissen C, Beukers M, Brants H, Ocké M. The diet
of the Dutch: Results of the Dutch National Food Consumption Survey 2012-2016. 2020.
34. van Dooren C. A Review of the Use of Linear Programming to Optimize Diets, Nutritiously,
Economically and Environmentally. Front Nutr. 2018;5(48).
35. Dooren C, Aiking H, Vellinga P. In search of indicators to assess the environmental impact
of diets. The International Journal of Life Cycle Assessment. 2018;23.
36. Fischer CG, Garnett T. Plates, pyramids, and planets: developments in national healthy
and sustainable dietary guidelines: a state of play assessment: Food and Agriculture
Organization of the United Nations; 2016.
37. Djojosoeparto S, Kamphuis C, Vandevijvere S, Poelman M. The Healthy Food Environment
Policy Index (Food-EPI): the Netherlands. An assessment of national governmental policies
affecting the food environment in the Netherlands and policy actions for creating a healthy
food environment. Utrecht: Utrecht University; 2020.
38. Satija A, Stampfer MJ, Rimm EB, Willett W, Hu FB. Perspective: Are Large, Simple Trials the
Solution for Nutrition Research? Advances in Nutrition. 2018;9(4):378-87.
39. Hébert JR, Frongillo EA, Adams SA, Turner-McGrievy GM, Hurley TG, Miller DR, et al.
Perspective: Randomized Controlled Trials Are Not a Panacea for Diet-Related Research.
Adv Nutr. 2016;7(3):423-32.
40. Satija A, Yu E, Willett WC, Hu FB. Understanding nutritional epidemiology and its role in
policy. Adv Nutr. 2015;6(1):5-18.
41. Chiu YH, Chavarro JE, Dickerman BA, Manson JE, Mukamal KJ, Rexrode KM, et al. Estimating
the effect of nutritional interventions using observational data: the American Heart
Association’s 2020 Dietary Goals and mortality. Am J Clin Nutr. 2021;114(2):690-703.
42. Hernán MA. Methods of Public Health Research - Strengthening Causal Inference from
Observational Data. N Engl J Med. 2021;385(15):1345-8.
43. Tobias DK, Lajous M. What would the trial be? Emulating randomized dietary intervention
trials to estimate causal effects with observational data. Am J Clin Nutr. 2021;114(2):416-7.
44. Ioannidis JP. Implausible results in human nutrition research. Bmj. 2013;347:f6698.
45. Archer E, Pavela G, Lavie CJ. The Inadmissibility of What We Eat in America and NHANES 9
Dietary Data in Nutrition and Obesity Research and the Scientific Formulation of National
Dietary Guidelines. Mayo Clin Proc. 2015;90(7):911-26.
46. Shim J-S, Oh K, Kim HC. Dietary assessment methods in epidemiologic studies. Epidemiol
Health. 2014;36:e2014009-e.
47. Bennett DA, Landry D, Little J, Minelli C. Systematic review of statistical approaches
to quantify, or correct for, measurement error in a continuous exposure in nutritional
epidemiology. BMC Medical Research Methodology. 2017;17(1):146.
48. Skinner A, Toumpakari Z, Stone C, Johnson L. Future Directions for Integrative Objective
Assessment of Eating Using Wearable Sensing Technology. Front Nutr. 2020;7:80-.
49. Tay W, Kaur B, Quek R, Lim J, Henry CJ. Current Developments in Digital Quantitative
Volume Estimation for the Optimisation of Dietary Assessment. Nutrients. 2020;12(4).
235
chapter 9 general discussion
50. Lo FPW, Sun Y, Qiu J, Lo B. Image-Based Food Classification and Volume Estimation
for Dietary Assessment: A Review. IEEE Journal of Biomedical and Health Informatics.
2020;24(7):1926-39.
51. Brennan L, Hu FB, Sun Q. Metabolomics Meets Nutritional Epidemiology: Harnessing the
Potential in Metabolomics Data. Metabolites. 2021;11(10):709.
52. Satija A, Hu FB. Plant-based diets and cardiovascular health. Trends Cardiovasc Med.
2018;28(7):437-41.
53. Szabo Z, Koczka V, Marosvolgyi T, Szabo E, Frank E, Polyak E, et al. Possible Biochemical
Processes Underlying the Positive Health Effects of Plant-Based Diets-A Narrative Review.
Nutrients. 2021;13(8).
54. Fedak KM, Bernal A, Capshaw ZA, Gross S. Applying the Bradford Hill criteria in the 21st
century: how data integration has changed causal inference in molecular epidemiology.
Emerg Themes Epidemiol. 2015;12:14-.
55. Berry SE, Valdes AM, Drew DA, Asnicar F, Mazidi M, Wolf J, et al. Human postprandial
responses to food and potential for precision nutrition. Nature Medicine. 2020;26(6):964-73.
56. Arno A, Thomas S. The efficacy of nudge theory strategies in influencing adult dietary
behaviour: a systematic review and meta-analysis. BMC Public Health. 2016;16:676.
57. Crockett RA, King SE, Marteau TM, Prevost AT, Bignardi G, Roberts NW, et al. Nutritional
labelling for healthier food or non-alcoholic drink purchasing and consumption. Cochrane
Database Syst Rev. 2018;2(2):Cd009315.
58. Cadario R, Chandon P. Which Healthy Eating Nudges Work Best? A Meta-Analysis of Field
Experiments. Marketing Science. 2020;39(3):465-86.
59. Shangguan S, Afshin A, Shulkin M, Ma W, Marsden D, Smith J, et al. A Meta-Analysis of Food
Labeling Effects on Consumer Diet Behaviors and Industry Practices. American Journal of
Preventive Medicine. 2019;56(2):300-14.
60. An R, Shi Y, Shen J, Bullard T, Liu G, Yang Q, et al. Effect of front-of-package nutrition labeling
on food purchases: a systematic review. Public Health. 2021;191:59-67.
61. Broers VJV, De Breucker C, Van den Broucke S, Luminet O. A systematic review and meta-
analysis of the effectiveness of nudging to increase fruit and vegetable choice. Eur J Public
Health. 2017;27(5):912-20.
62. Wilson AL, Buckley E, Buckley JD, Bogomolova S. Nudging healthier food and beverage
choices through salience and priming. Evidence from a systematic review. Food Quality
and Preference. 2016;51:47-64.
63. Schüz B, Meyerhof H, Hilz LK, Mata J. Equity Effects of Dietary Nudging Field Experiments:
Systematic Review. Front Public Health. 2021;9:668998.
64. Petticrew M, Knai C, Thomas J, Rehfuess EA, Noyes J, Gerhardus A, et al. Implications of
a complexity perspective for systematic reviews and guideline development in health
decision making. BMJ Global Health. 2019;4(Suppl 1):e000899.
65. Ogilvie D, Bauman A, Foley L, Guell C, Humphreys D, Panter J. Making sense of the evidence
in population health intervention research: building a dry stone wall. BMJ Glob Health.
2020;5(12).
66. Sawyer ADM, van Lenthe F, Kamphuis CBM, Terragni L, Roos G, Poelman MP, et al. Dynamics
of the complex food environment underlying dietary intake in low-income groups: a
systems map of associations extracted from a systematic umbrella literature review.
International Journal of Behavioral Nutrition and Physical Activity. 2021;18(1):96.
236
67. Petticrew M. Time to rethink the systematic review catechism? Moving from ‘what works’
to ‘what happens’. Syst Rev. 2015;4:36-.
68. Greenhalgh T, Papoutsi C. Studying complexity in health services research: desperately
seeking an overdue paradigm shift. BMC Medicine. 2018;16(1):95.
69. Langellier BA, Bilal U, Montes F, Meisel JD, Cardoso LO, Hammond RA. Complex Systems
Approaches to Diet: A Systematic Review. Am J Prev Med. 2019;57(2):273-81.
70. Railsback SF, Grimm V. Agent-based and individual-based modeling: a practical
introduction: Princeton university press; 2019.
71. Xue H, Slivka L, Igusa T, Huang TT, Wang Y. Applications of systems modelling in obesity
research. Obes Rev. 2018;19(9):1293-308.
72. Li Y, Zhang D, Thapa JR, Madondo K, Yi S, Fisher E, et al. Assessing the role of access and
price on the consumption of fruits and vegetables across New York City using agent-based
modeling. Preventive Medicine. 2018;106:73-8.
73. Zhang D, Giabbanelli PJ, Arah OA, Zimmerman FJ. Impact of Different Policies on Unhealthy
Dietary Behaviors in an Urban Adult Population: An Agent-Based Simulation Model.
American Journal of Public Health. 2014;104(7):1217-22.
74. Widener MJ, Metcalf SS, Bar-Yam Y. Agent-based modeling of policies to improve urban
food access for low-income populations. Applied Geography. 2013;40:1-10.
75. Ceschi A, Sartori R, Dickert S, Scalco A, Tur EM, Tommasi F, et al. Testing a norm-based
policy for waste management: An agent-based modeling simulation on nudging recycling
behavior. J Environ Manage. 2021;294:112938.
76. Greenhalgh T. Will COVID-19 be evidence-based medicine’s nemesis? PLOS Medicine.
2020;17(6):e1003266.
77. Bala BK, Arshad FM, Noh KM. Causal loop diagrams. System dynamics: Springer; 2017. p.
37-51.
78. Savona N, Thompson C, Smith D, Cummins S. ‘Complexity’ as a rhetorical smokescreen for
UK public health inaction on diet. Critical Public Health. 2021;31(5):510-20.
79. Een eerlijke kans op gezond leven [Internet]. Den Haag: Raad voor Volksgezondheid en
Samenleving; 2021 [Accessed 17-12-2021]. Available from: https://www.rijksoverheid.nl/
documenten/rapporten/2021/04/29/een-eerlijke-kans-op-gezond-leven.
80. Omzien naar elkaar, vooruitkijken naar de toekomst: Coalitieakkoord 2021-2025 [Internet].
[Accessed 17-12-2021]. Available from: https://www.kabinetsformatie2021.nl/documenten/
publicaties/2021/12/15/coalitieakkoord-omzien-naar-elkaar-vooruitkijken-naar-de-
toekomst.
9
237
A
SUMMARY
This thesis focused on healthy and sustainable diets from an individual and societal
perspective. The first part of this thesis aimed to explore the association between various
healthy and/or sustainable dietary patterns and cardiovascular outcomes, mortality as
well as sustainability indicators. The second part of this thesis aimed to explore how the
food environment affects dietary behaviours and health outcomes, and how adults with
low socioeconomic position (SEP) perceive food environment interventions.

The first chapters of this thesis focused on the individual perspective of healthy and
sustainable diets. In Chapter 2, we investigated the association between adherence to
the Dutch dietary guidelines and incidence of heart failure among 37,468 participants
from the EPIC-NL study. High adherence to the Dutch dietary guidelines was associated
with a 27% lower risk of heart failure (HRQ4vsQ1: 0.73; 95% CI: 0.58 – 0.93), after adjustment
for sociodemographic and lifestyle characteristics. As healthy diets may specifically
prevent heart failure through effects on blood pressure, we additionally investigated
whether adherence to the DASH diet – an effective non-pharmacologic approach for
lowering blood pressure – would be associated with risk of heart failure in Chapter 3.
Using data from 37,671 participants of the EPIC-NL study, we found that high adherence
to the DASH diet was associated with 22% lower heart failure risk after adjustment
for sociodemographic and lifestyle characteristics (HRT3vsT1: 0.78; 95% CI: 0.64 – 0.95),
corroborating our previous findings. In Chapter 4 and 5, we focused on the Healthy
Reference Diet: a set of dietary recommendations which have been proposed to co-
benefit human and planetary health. In Chapter 4, we studied the association between
adherence to the Healthy Reference Diet and risk of coronary heart disease, all-cause
mortality, and environmental impact using data from 37,349 particpants in the EPIC-NL
study. High adherence to the Healthy Reference Diet was associated with 15% lower risk
of coronary heart disease (HRQ4vsQ1: 0.85; 95% CI: 0.75 – 0.96), and 17% lower risk of all-
cause mortality (HRQ4vsQ1: 0.83; 95% CI: 0.77 – 0.90) after adjustment for sociodemographic
and lifestyle characteristics. Additionally, high adherence to the Healthy Reference Diet
was associated with 1.7% lower impact from greenhouse gas emissions, 3.2% less land

Appendices use, 0.5% less freshwater eutrophication, 3.5% less marine eutrophication, 6.3% less
terrestrial acidification, but with 31.4% more blue water use. In Chapter 5, we studied
adherence to the Healthy Reference Diet in the pan-European EPIC cohort, comprised of
401,187 participants from eight participating countries. We observed that high adherence
to the Healthy Reference diet was associated with a lower risk of incident coronary heart
disease (HRQ5vsQ1: 0.94; 95% CI: 0.88 – 1.00) and stroke (HRQ5vsQ1: 0.76; 95% CI: 0.70 – 0.84)
after adjustment for demographic and lifestyle characteristics.
appendices summary

The subsequent chapters of this thesis focused on the societal perspective of healthy determinant of food choice, but it remains difficult to quantify its effects. The complex
and sustainable diets, zooming in on the role of the food environment. In Chapter 6, systems perspective acknowledges that causes for poor diets are multifactorial and
we systematically reviewed current evidence on the effectiveness of nudges – small dynamic, stressing the need for policy action on multiple levels. As such, the food
adjustments in the choice architecture of for example a supermarket – in promoting environment can be leveraged as entry-point for promoting healthy and sustainable
healthy dietary choices within real-life food purchasing environments. Additionally, we food choices at a societal level.
explored evidence for a moderating role of SEP. We observed that the majority of the
evidence-base focused on the effects of information and position nudges, while other
types of nudges (e.g., availability, presentation, size) were less often studied. Information
and position nudges showed modest tendencies towards beneficial effects, although
findings were not meta-analyzed due to the substantial heterogeneity in underlying
evidence encountered. Additionally, only a few studies examined the effects of
nudges across levels of SEP. Although there were some indications that effects were
more pronounced among those with low SEP, the scarcity of evidence and the use of
different proxies of SEP made it difficult to draw firm conclusions. In Chapter 7, we
studied the association between relative exposure to fast-food restaurants and diet
quality, overweight, and obesity, using cross-sectional data from 8,231 participants
in the EPIC-NL study. We did not observe an association between high exposure to
fast-food restaurants around the home of residence and diet quality or weight status.
In Chapter 8, we aimed to explore how the determinants of food choice shape the
perceptions regarding supermarket-based nudging strategies among adults with low
SEP. In a qualitative study including 15 interviewees with lower socioeconomic position,
we observed that interviewees evaluated supermarket-based nudging strategies
more positively when they aligned with their product preferences, information needs
and beliefs about the food environment. At the same time, we observed that some
interviewees distrusted supermarket engagement in health-promoting initiatives,
suspecting ulterior motives.

In Chapter 9, findings from this thesis were put in a broader perspective. I discussed
how future studies could further examine the association between healthy diets with
heart failure subtypes, given the distinct differences in underlying risk-factors and
pathophysiology. Additionally, I reflected on the analytical challenge of incorporating
sustainability in food-based-dietary guidelines, and on the limitations and opportunities
that are faced in the field of nutritional epidemiology. Finally, I reflected on the complexity A
of food environment research, and on policy measures that are needed to prevent
cardiovascular disease and other non-communicable diseases.

From this thesis, I conclude that on an individual level, consuming dietary patterns rich
in plant-based foods and including moderate amounts of animal-sourced foods may
benefit both cardiovascular and planetary health. The food environment is an important

240 241
appendices
SAMENVATTING
In dit proefschrift stond het bestuderen van gezonde en duurzame voedingspatronen
vanuit individueel en maatschappelijk perspectief centraal. Het eerste deel van dit
proefschrift had als doel om de relatie te onderzoeken tussen diverse gezonde en/
of duurzame voedingspatronen en cardiovasculaire uitkomsten, sterfte en milieu-
indicatoren. Het tweede deel van dit proefschrift had als doel om te exploreren hoe de
voedingsomgeving ons eetgedrag beïnvloedt, en hoe mensen met een lagere sociaal-
economische positie de voedingsomgeving ervaren.
In de eerste hoofdstukken van dit proefschrift keken we naar het individuele perspectief
van gezonde en duurzame voedingspatronen. In hoofdstuk 2 onderzochten we de
relatie tussen het naleven van de Richtlijnen Goede Voeding en het optreden van
hartfalen bij 37,468 deelnemers van de EPIC-NL studie. We vonden dat goede naleving
van de Richtlijnen Goede Voeding geassocieerd was met een 27% lager risico op
hartfalen (HRQ4vsQ1: 0.73; 95% CI: 0.58 – 0.93) na correctie voor sociaal-demografische-
en leefstijlkarakteristieken. Aangezien een gezond voedingspatroon hartfalen mogelijk
voorkomt door het effect op bloeddruk, hebben we daarnaast onderzocht of goede
naleving van het DASH voedingspatroon – een effectieve non-farmacologische manier
om bloeddruk te verlagen – geassocieerd is met het risico op hartfalen in hoofdstuk
3. Met de gegevens van 37,671 deelnemers van de EPIC-NL studie vonden we dat
het goed naleven van het DASH dieet geassocieerd was met een 22% lager risico of
hartfalen (HRT3vsT1: 0.78; 95% CI: 0.64 – 0.95) na correctie voor sociaal-demografische-
en leefstijlkarakteristieken, wat in overeenstemming is met de resultaten uit hoofdstuk
2. In hoofdstuk 4 en 5 onderzochten we het Healthy Reference Diet: een set van
voedingsrichtlijnen die opgesteld zijn met het oog op zowel het bevorderen van de
volksgezondheid als het beschermen van het klimaat. In hoofdstuk 4, onderzochten
we de relatie tussen de naleving van het Healthy Reference Diet en het optreden
van coronaire hartziekte, sterfte, en milieuimpact met behulp van data van 37,349
deelnemers van de EPIC-NL studie. Het goed naleven van het Healthy Reference Diet
was geassocieerd met een 15% lager risico op het optreden van coronaire hartziekten
(HRQ4vsQ1: 0.85; 95% CI: 0.75 – 0.96), en een 17% lager risico op sterfte (HRQ4vsQ1: 0.83; 95%
CI: 0.77 – 0.90) na correctie voor sociaal-demografische- en leefstijlkarakteristieken.
Daarnaast was het goed naleven van het Healthy Reference Diet geassocieerd met 1.7%
lagere broeikasuitstoot, 3.2% minder landgebruik, 0.5% minder eutrofiëring van zoet
water, 3.5% minder eutrofiëring van oppervlaktewater, 6.3% minder bodemverzuring,
maar met een toename van 31.4% in het gebruik van irrigatiewater. In hoofdstuk 5
onderzochten we het Healthy Reference Diet in het pan-Europese EPIC cohort, waarin
we gebruik maakten van de gegevens van 401,187 deelnemers uit acht deelnemende
242
samenvat ting
landen. Daar zagen we dat het goed naleven van het Healthy Reference Diet geassocieerd
was met een lager risico op het optreden van coronaire hartziekte (HRQ5vsQ1: 0.94; 95%
CI: 0.88 – 1.00) en beroerte (HRQ5vsQ1: 0.76; 95% CI: 0.70 – 0.84) na correctie voor sociaal-
demografische- en leefstijlkarakteristieken.
De daaropvolgende hoofdstukken van dit proefschrift zoomden in op het
maatschappelijke perspectief van gezonde en duurzame voedingspatronen. In
hoofdstuk 6 hebben we een systematisch literatuuronderzoek gedaan naar de
effectiviteit van nudges – kleine duwtjes richting de gezonde keuzes in bijvoorbeeld
een supermarkt – in het stimuleren van gezonde voedingskeuzes in een ‘real-life’ setting.
Daarnaast hebben we onderzocht of de effectiviteit van nudges verschillend was tussen
groepen die verschillen in sociaal-economische positie. We vonden dat een groot deel
van het bewijs zich voornamelijk toespitste op het bestuderen van information en
position nudges, terwijl ander typen nudges (bijvoorbeeld availability, presentation, en
size nudges) minder vaak bestudeerd werden. Information en position nudges lieten een
voorzichtige tendens naar positieve effecten zien, alhoewel we niet in staat waren de
bevindingen te meta-analyseren door de heterogeniteit in het onderliggende bewijs.
Daarnaast hebben we maar een klein aantal studies geïdentificeerd die de effecten
van nudges uitsplitsten naar sociaal-economische positie. Alhoewel er een indicatie
was dat de effecten van nudges sterker waren onder groepen met een lagere sociaal-
economische positie, kunnen we door de beperkte omvang van studies en het gebruik
van verschillende indicatoren voor sociaal-economische positie, geen sterke conclusies
hieraan verbinden. In hoofdstuk 7 hebben we de relatie tussen relatieve blootstelling
aan fast-food restaurants enerzijds, en dieetkwaliteit en het risico op overgewicht
en obesitas anderzijds, bestudeerd met behulp van cross-sectionele data van 8,231
deelnemers van de EPIC-NL studie. In deze studie vonden we geen verband tussen
een hoge relatieve blootstelling van fast-food restaurants in de thuisomgeving en
dieetkwaliteit of risico op overgewicht/obesitas. In hoofdstuk 8, hebben we onderzocht
welke determinanten van voedingskeuzes percepties rondom het toepassen van nudges
in de supermarkt beïnvloeden bij mensen met een lagere sociaal-economische positie.
In een kwalitatieve studie met 15 participanten vonden we dat geïnterviewden nudges in
de supermarkt positiever beoordeelden wanneer deze aansloten op productvoorkeuren,
informatiebehoeften, en overtuigingen over de voedselomgeving van de deelnemers. A
Tegelijkertijd zagen we dat sommige deelnemers wantrouwen koesteren tegenover
dergelijke initiatieven in de supermarkt, vanwege vermoedens van verborgen motieven
van de supermarkt of de voedingsindustrie.
In hoofdstuk 9 zijn de belangrijkste bevindingen uit dit proefschrift in breder perspectief
geplaatst. Ik heb besproken hoe toekomstige studies zich verder kunnen toeleggen op
243
appendices
het bestuderen van gezonde voedingspatronen in relatie tot de verschillende subtypen
van hartfalen, gezien de verschillen in onderliggende risicofactoren en pathofysiologie.
Daarnaast heb ik gereflecteerd op de analytische uitdaging van het meenemen van
het duurzaamheidaspect in het opstellen van toekomstige voedingsrichtlijnen, en
de limitaties en kansen in het veld van de voedingsepidemiologie. Tenslotte heb ik
gereflecteerd op de complexiteit van het doen van voedingsomgeving onderzoek, en
op de beleidsmaatregelen die nodig zijn om cardiovasculaire ziekte en andere chronische
ziekten te voorkomen.
Op basis van dit proefschrift concludeer ik dat op individueel niveau, het consumeren van
voedingspatronen rijk aan plantaardige voedingsmiddelen en met kleine hoeveelheden
dierlijke voedingsmiddelen zowel de volksgezondheid als het milieu ten goede kan komen.
De voedingsomgeving is een belangrijke determinant van voedingskeuzes, maar het is
moeilijk om de effecten hiervan op uitkomsten te kwantificeren. Het systeemdenken
paradigma onderkent dat de oorzaken van ongezonde voedingspatronen multifactorieel
en dynamisch zijn, wat het belang van beleid op meerde niveaus onderstreept. Zodoende
kan de voedingsomgeving als ingangspunt worden gebruikt om gezonde en duurzame
voedingskeuzes op maatschappelijk niveau te stimuleren.
244
list of public ations
LIST OF PUBLICATIONS
Manuscripts published and included in this doctoral thesis
Harbers MC, de Kroon AM, Boer JMA, Asselbergs FW, Geleijnse JM, Verschuren WMM et
al. Adherence to the Dutch dietary guidelines and 15-year incidence of heart failure in
the EPIC-NL cohort. European Journal of Nutrition. 2020 Dec;59(8):3405-3413. https://
doi.org/10.1007/s00394-019-02170-7
Harbers MC, Beulens JWJ, Rutters F, de Boer F, Gillebaart M, Sluijs I et al. The effects of
nudges on purchases, food choice, and energy intake or content of purchases in real-life
food purchasing environments: a systematic review and evidence synthesis. Nutrition
Journal. 2020 Sep 17;19(1). 103. https://doi.org/10.1186/s12937-020-00623-y
Harbers MC, Beulens JWJ, Boer JM, Karssenberg D, Mackenbach JD, Rutters F et al.
Residential exposure to fast-food restaurants and its association with diet quality,
overweight and obesity in the Netherlands: a cross-sectional analysis in the EPIC-NL
cohort. Nutrition Journal. 2021 Jun 16;20(1):1-10. 56. https://doi.org/10.1186/s12937-021-
00713-5
Harbers MC, Middel CNH, Stuber JM, Beulens JWJ, Rutters F, van der Schouw YT.
Determinants of food choice and perceptions of supermarket-based nudging
interventions among adults with low socioeconomic position: The supreme nudge
project. International Journal of Environmental Research and Public Health. 2021 Jun
7;18(11):1-14. 6175. https://doi.org/10.3390/ijerph18116175
Manuscripts published but not included in this doctoral thesis
Lakerveld J, Mackenbach JD, de Boer F, Brandhorst B, Broerse JEW, de Bruijn G-J et al.
Improving cardiometabolic health through nudging dietary behaviours and physical
activity in low SES adults: design of the Supreme Nudge project. BMC Public Health. 2018
Jul 20;18(1). 899. https://doi.org/10.1186/s12889-018-5839-1
Pertiwi K, Wanders AJ, Harbers MC, Küpers LK, Soedamah-Muthu SS, de Goede J et al. A
Plasma and Dietary Linoleic Acid and 3-Year Risk of Type 2 Diabetes After Myocardial
Infarction: A Prospective Analysis in the Alpha Omega Cohort. Diabetes Care. 2020
Feb;43(2):358-365. https://doi.org/10.2337/dc19-1483
Stuber JM, Mackenbach JD, De Boer FE, De Bruijn GJ, Gillebaart M, Harbers MC et al.
Reducing cardiometabolic risk in adults with a low socioeconomic position: Protocol of
245
appendices
the Supreme Nudge parallel cluster-randomised controlled supermarket trial. Nutrition
Journal. 2020 May 19;19(1):1-19. 46. https://doi.org/10.1186/s12937-020-00562-8
Hoenink JC, Beulens JWJ, Harbers MC, Boer JMA, Dijkstra SC, Nicolaou M et al. To what
extent do dietary costs explain socio-economic differences in dietary behavior? Nutrition
Journal. 2020 Aug 24;19(1). 88. https://doi.org/10.1186/s12937-020-00608-x
Pinho MGM, Lakerveld J, Harbers MC, Sluijs I, Vermeulen R, Huss A et al. Ultra-processed
food consumption patterns among older adults in the Netherlands and the role of the
food environment. European Journal of Nutrition. 2021 Aug;60(5):2567-2580. https://
doi.org/10.1007/s00394-020-02436-5
Manuscripts in preparation or submitted
Harbers MC, Beulens JWJ, Rutters F, May AM, Key TJ, Huybrechts I, Paolo Vineis, Laine
J, Papier K, Knuppel A, Weiderpass E, Aleksandrova K, Dahm CC, EPIC-cohorts and IARC
representatives, van der Schouw YT. The association between adherence to the EAT-
Lancet Healthy Reference Diet and coronary heart disease and stroke: the EPIC-CVD
cohort study. In preparation.
Colizzi C, Harbers MC, Vellinga RE, Verschuren WMM, Boer JMA, Temme EHM, van der
Schouw YT. Adherence to the EAT-Lancet Healthy Reference Diet in relation to risk of
Cardiovascular Events and Environmental Impact: Results from the EPIC-NL Cohort.
Submitted to Circulation.
Costa S, Harbers MC, Boer JMA, Verschuren WMM, van der Schouw YT. Adherence to
the DASH diet and risk of heart failure in the EPIC-NL cohort. Submitted to Nutrition,
Metabolism, and Cardiovascular Diseases.
246
dankwoord
DANKWOORD
Na vier jaar hard werken zit het er dan eindelijk op: mijn proefschrift is klaar, en nu
mag ik het meest gelezen (en leukste?) stukje van het proefschrift gaan schrijven – het
dankwoord! Ik kijk terug op vier leuke en leerzame jaren, waarvoor ik een paar personen
in het bijzonder wil bedanken.
Allereerst gaat mijn dank uit naar mijn promotieteam: prof. dr. Yvonne T. van der
Schouw, prof. dr. Joline W.J. Beulens, en dr. Femke Rutters. Ik wil jullie bedanken voor
de fijne samenwerking in de afgelopen jaren, en voor de vrijheid en het vertrouwen
dat jullie mij hebben gegeven om mijzelf te ontwikkelen tot zelfstandig onderzoeker.
Vaak als ik mensen vertelde over de omvang van mijn team – zeker toen er nog een
vierde co-promotor in het spel was – werd er vaak gereageerd met de aanname dat
al die verschillende meningen binnen zo’n team vast wel heel lastig werken zou zijn.
Niets bleek minder waar: jullie leken elkaar altijd perfect aan te vullen, waren altijd
(zonder uitzondering) snel om te reageren op mijn mailtjes en teksten, en de neuzen
stonden vrijwel altijd dezelfde kant op (met wat nodige wetenschappelijke discussie
daargelaten). Dat was enorm prettig werken, en na onze overleggen had ik altijd een
keurig lijstje van acties hoe ik weer verder kon. Beste Yvonne, ongeveer halverwege in
mijn promotietraject werd jij naast promotor ook mijn dagelijks begeleider en gingen
we intensiever samenwerken. Ik vond het ontzettend fijn hoe ik altijd met mijn vragen
bij je terecht kon, en heb veel geleerd van de rustige en gestructureerde manier hoe jij
dingen aanvliegt. Beste Joline, ik heb een enorme bewondering voor hoe jij – ondanks je
drukke agenda – zo betrokken bent bij (je vele) promovendi, en altijd in staat bent snel
en gedetailleerd feedback te geven. Wat fijn was het ook om te kunnen sparren over
voedingsepi-onderzoek – toch wel een beetje de roots die jij, Yvonne en ik delen! Beste
Femke, gelukkig bestond ons team niet alleen maar uit voedingsmiepen, en daardoor
wist jij vaak net die juiste kritische vraag te stellen die mij uitdaagde om dingen beter
uit te leggen. Ook jij veel dank voor je snelle en gedetailleerde opmerkingen op mijn
manuscripten.
Daarnaast wil ik ook graag dr. Ivonne Sluijs bedanken. De eerste 2 jaar van mijn
promotietraject ben jij betrokken geweest als co-promotor, en heb je mij onder jouw A
hoede genomen en wegwijs gemaakt in het Julius Centrum. Ik vond het altijd enorm fijn
dat ik voor alles bij je terecht kon. Dankjewel voor je blijvende interesse in het vergaan
van mijn promotietraject.
Mijn dank gaat ook uit naar de leden van de leescommissie: prof. dr. M.L. Bots, prof.
dr. J.M. Geleijnse, prof. dr. F.J. van Lenthe, prof. dr. D.T.D. de Ridder en prof. dr. F.L.J.
247
appendices
Visseren. Dank voor het lezen en beoordelen van dit manuscript. Ook wil ik alle co-
auteurs bedanken voor hun bijdrage aan de manuscripten in dit proefschrift.
Marianne en Sabita – jullie wil ik bedanken voor de kansen die jullie mij hebben
gegeven tijdens mijn Master Nutrition and Health in Wageningen. Tijdens het schrijven
van mijn scriptie over de relatie tussen linolzuur en diabetes onder jullie supervisie,
werd mijn interesse in het doen van onderzoek verder aangewakkerd. Dat bleef jullie
hoogstwaarschijnlijk niet onopgemerkt, want jullie stuurden mij naar het EDEG congres
in Dubrovnik en naar de WEON in Antwerpen – kansen waar ik jullie enorm dankbaar
voor ben. Dank voor het zelfvertrouwen wat jullie in mij deden groeien.
Om in het Wageningse te blijven… Dear Nena, what started out as two students
complaining about errors in SAS (#weloveSAS), grew into a friendship which I cherish
with all my heart. You are the most thoughtful person I know, and your cheerleading
along this PhD journey motivated me to keep going. Hopefully we can re-unite soon (I’m
opting for Zurich) and have a little celebration of us being PhDone. Lieve Mies, nog een
Wageningse langeafstandsvriendin – wat leuk dat we contact zijn blijven houden ondanks
het feit dat je naar Spanje bent verhuisd. Ik hoop dat we elkaar snel weer kunnen zien
om bij te kletsen. Joke, jou wil ik bedanken voor het leuke contact wat we zijn blijven
houden na mijn tutor-tijd – daar kijk ik met heel veel plezier op terug!
Lieve mede-promovendi van Supreme Nudge: Josine, Jody, Cédric, Anne, en Femke. Wat
was het een feestje om met jullie samen te werken op het Supreme Nudge project. Ik
heb de overleggen altijd erg gezellig gevonden, en hoop dat we elkaar nog tegen blijven
komen in de toekomst. Cédric, jou wil ik in het bijzonder bedanken voor je hulp bij de
analyse van de kwalitatieve studie. Het analyseren van de transcripten voelde voor mij
wat onwennig (kan ik dat wel zeggen? Is toch n=1?), maar door het sparren met jou is
er toch een mooie studie komen te liggen. Josine, wat heb ik een enorme bewondering
voor hoe jij als hoofdverantwoordelijke promovenda het Supreme Nudge project in
uitdagende omstandigheden hebt geleid. Ook dank voor je gezelligheid, betrokkenheid
en interesse in het verlopen van mijn onderzoek. In het kader van Supreme Nudge
wil ik ook graag Joreintje en Jeroen bedanken voor het organiseren van alle leuke en
inspirerende consortiummeetings; ik kwam hier altijd vol energie en inspiratie van terug.
Chiara, Stefanie, Marleen, Niek, en Manon: het begeleiden van studenten vond ik
misschien wel één van mijn leukste taken tijdens mijn promotietijd. Dank voor jullie
bijdrage aan mijn onderzoek, ik vond het heel leuk om met jullie samen te werken!
248
dankwoord
De collega’s uit het Stratenum (kamer 6.119) en van Geuns (kamer 5.16) wil ik bedanken
voor alle gezellige lunches en kantoorpraat – het voelt als eeuwen geleden, maar het
zorgde voor een fijne en warme start van mijn promotietijd. Dank aan alle cardiovasculaire
epi collega’s voor de inhoudelijke sessies tijdens de Geoffrey Rose meetings. In het
bijzonder wil ik Wout bedanken voor alle hulp in het regelen van handtekeningen,
organiseren van overleggen, en andere administratieve zaken. Fijn dat je dit uit handen
kon nemen Wout! Lieve Anouk, wij begonnen ongeveer tegelijkertijd aan onze promotie
en er was direct een klik. Je bent een geboren onderzoeker, maar bovenal een hele fijne
en lieve collega. Ik heb erg genoten van al onze lunchwandelingen, en wat fijn dat we
ook tijdens het thuiswerken regelmatig afspreken voor een wandeling met koffie of iced
tea to go! Ik ben er heel erg blij mee dat jij mijn paranimf wil zijn tijdens de grote dag.
Lieve Marte en Marloes: wat begon als het zijn van AID-zusjes tijdens de introdagen in
Wageningen is uitgegroeid tot een hele hechte vriendschap. Wat heb ik enorm genoten
van onze talloze borrels, diners, en lunches – ik hoop dat er nog velen mogen volgen.
Marloes, wat ben ik blij dat jij aan mijn zijde staat als paranimf. Ondanks dat we ons hele
leven elkaar al kennen, zijn we pas besties geworden tijdens onze studie in Wageningen.
Wellicht zijn het onze gemeenschappelijke roots, maar wat heerlijk hoe we vaak hetzelfde
over dingen denken en hetzelfde leuk vinden. Dank dat ik stoom bij je mocht afblazen
– ons gezamenlijk geklaag verzachtte de PhD-dipjes wel degelijk – en dank voor al je
gezelligheid. Ik ben enorm trots op jou als kersverse moeder van lieve kleine Elise, en
geniet ervan dat we nu lekker dichtbij elkaar wonen en met onze lieve teckeltjes het bos
in kunnen.
Lieve oma Hennie, helaas heeft u alleen het begin van mijn promoveren kunnen
meemaken. Wat dat nu allemaal precies inhield was u een raadsel, maar zolang ik
er maar plezier in had dan was het voor u goed. Wat was het fijn thuis komen aan
de Flessestraat op zaterdagochtend met een kop koffie bij de kachel. Dank voor alle
warmte en liefde die ik van u heb mogen ontvangen. Lieve oma Vink, uw trots over uw
(“geleerde”) kleinkinderen steekt u niet onder stoelen of banken. Wat ben ik trots op
een oma die nog zo hip en bij de tijd is. Dank voor uw belangstelling in het onderzoek
van mij en Martijn.
A
Lieve schoonfamilie – lieve Gemma, Berrie, Jasper, Ciske, Willem, Benthe, Julian en
Liz. Dankjulliewel voor jullie interesse in mijn onderzoek, maar vooral voor alle fijne
middagen en avonden in de Achterhoek. Ik vind het ontzettend gezellig dat we elkaar
nu we naar Bennekom zijn verhuisd vaker zien. Ook Saar is daar volgens mij heel blij
mee – op naar nog meer gezelligheid in de toekomst, en ik ben benieuwd hoeveel onze
hondenroedel nog gaat uitdijen.
249
appendices
Lieve kleine zus, lieve Lotte – wat houd ik ontzettend veel van jou! Je bent inmiddels
echt niet meer zo klein – lees inmiddels bijna afgestudeerd – en wat ben ik toch enorm
trots op je. Dank voor alle gezelligheid tijdens de weekenden bij papa en mama thuis, en
voor de fijne lunches in Wageningen tijdens al het thuiswerken. Ook dank voor je hulp
met mijn discussie – na talloze WhatsApp videogesprekken hoop ik dat het stukje over
linear programming goed op papier staat. Aan jouw uitleg zal het niet hebben gelegen.
Lieve papa en mama – dank voor de liefdevolle, warme en stabiele basis die jullie mij
hebben gegeven. Jullie hebben mij altijd aangemoedigd om mijn best te doen, maar
bovenal dicht bij mezelf te blijven. Dank dat jullie altijd voor mij klaar staan – ik houd
ontzettend veel van jullie. Lieve pap, er zijn al heel veel grapjes gemaakt over wat ik hier
zou opschrijven over jou, maar ik houd het serieus. Dank voor je relativering, humor, en
je eindeloze optimisme en energie – you are one of a kind. Lieve mam, wat vind ik het
heerlijk om met jou te kletsen over alles en nog wat. Dankjewel dat ik altijd met alles bij
je terecht kan – jouw zorgzaamheid en liefde zijn eindeloos. Love you!
Tenslotte, aan mijn allerliefste Martijn – mijn rots in de branding. What a ride it has been!
Samen promoveren in lockdown tijd: als dat geen ultieme relatietest is dan weet ik het
ook niet meer… Wat begon in een klein appartementje aan de Maliebaan, werd een
volledige casa in Bennekom, waar we ook nog eens vergezeld werden door de liefste (en
meest eigenwijze) teckel op aarde. Dankjewel liefste voor je onvoorwaardelijke support
en liefde de afgelopen jaren. Ik ben enorm trots op jou, en ik kan niet wachten op wat
de toekomst ons brengt. Ik hou van je!
250
about the author
ABOUT THE AUTHOR
Marjolein Harbers was born on the 2nd of February 1994 in Wageningen, the Netherlands.
She attended pre-university education at the Hendrik Pierson College in Zetten, after
which she pursued a bachelor degree in Health and Society at Wageningen University.
Given her growing interest in nutrition, she then obtained her MSc in Nutrition and
Health from Wageningen University, specializing in Epidemiology and Public Health. In
September 2017 Marjolein started as a PhD candidate on the Supreme Nudge project
at the Department of Epidemiology at the Julius Center for Health Sciences and Primary
Care in the University Medical Center Utrecht. She worked under the supervision of prof.
dr. Yvonne van der Schouw, prof. dr. Joline Beulens, and dr. Femke
Rutters. From January 2022 onwards, Marjolein continues
to work as a postdoctoral researcher at the Julius Center,
focusing on prevention of cardiometabolic diseases.
251