Adherence to the healthy Nordic diet is associated with weight change during 7 years of follow-up

Abstract

Studies indicate that the healthy Nordic diet may improve heart health, but its relation to weight change is less clear. We studied the association between the adherence to the healthy Nordic diet and long-term changes in weight, BMI and waist circumference. Furthermore, the agreement between self-reported and measured body anthropometrics was examined. The population-based DIetary, Lifestyle and Genetic Determinants of Obesity and Metabolic syndrome Study in 2007 included 5024 Finns aged 25–75 years. The follow-up was conducted in 2014 ( n 3735). One-third of the participants were invited to a health examination. The rest were sent measuring tape and written instructions along with questionnaires. The Baltic Sea Diet Score (BSDS) was used to measure adherence to the healthy Nordic diet. Association of the baseline BSDS and changes in BSDS during the follow-up with changes in body anthropometrics were examined using linear regression analysis. The agreement between self-reported and nurse-measured anthropometrics was determined with Bland–Altman analysis. Intra-class correlation coefficients between self-reported and nurse-measured anthropometrics exceeded 0·95. The baseline BSDS associated with lower weight ( β =−0·056, P =0·043) and BMI ( β =−0·021, P =0·031) over the follow-up. This association was especially evident among those who had increased their BSDS. In conclusion, both high initial and improved adherence to the healthy Nordic diet may promote long-term weight maintenance. The self-reported/measured anthropometrics were shown to have high agreement with nurse-measured values which adds the credibility of our results.

Full text

Adherence to the healthy Nordic diet is associated with weight change
during 7 years of follow-up
Noora Kanerva
1,2
*, Kennet Harald
1
, Satu Männistö
1
, Niina E. Kaartinen
1
, Mirkka Maukonen
1
,
Ari Haukkala
3
and Pekka Jousilahti
1
1
Department of Public Health Solutions, National Institute for Health and Welfare, PO Box 30, 27100 Helsinki, Finland
2
Department of Public Health, University of Helsinki, PO Box 20, 00014 Helsinki, Finland
3
Department of Social Research, University of Helsinki, PO Box 54, 00014 Helsinki, Finland
(Submitted 4 December 2017 –Final revision received 30 March 2018 –Accepted 15 April 2018)
Abstract
Studies indicate that the healthy Nordic diet may improve heart health, but its relation to weight change is less clear. We studied the association
between the adherence to the healthy Nordic diet and long-term changes in weight, BMI and waist circumference. Furthermore, the
agreement between self-reported and measured body anthropometrics was examined. The population-based DIetary, Lifestyle and Genetic
Determinants of Obesity and Metabolic syndrome Study in 2007 included 5024 Finns aged 25–75 years. The follow-up was conducted in 2014
(n3735). One-third of the participants were invited to a health examination. The rest were sent measuring tape and written instructions along
with questionnaires. The Baltic Sea Diet Score (BSDS) was used to measure adherence to the healthy Nordic diet. Association of the baseline
BSDS and changes in BSDS during the follow-up with changes in body anthropometrics were examined using linear regression analysis. The
agreement between self-reported and nurse-measured anthropometrics was determined with Bland–Altman analysis. Intra-class correlation
coefficients between self-reported and nurse-measured anthropometrics exceeded 0·95. The baseline BSDS associated with lower weight
(β=−0·056, P=0·043) and BMI (β=−0·021, P=0·031) over the follow-up. This association was especially evident among those who had
increased their BSDS. In conclusion, both high initial and improved adherence to the healthy Nordic diet may promote long-term weight
maintenance. The self-reported/measured anthropometrics were shown to have high agreement with nurse-measured values which adds the
credibility of our results.
Key words: Baltic Sea Diet Score: Dietary score: Nordic diet: Validation: Weight change
Overweight (BMI ≥25 kg/m
2
) and obesity (BMI ≥30 kg/m
2
) are
global health problems that concern over 50 % of western
populations
(1)
. Excess adipose tissue with its array of metabolic
disturbances increases the risk of chronic diseases, such as the
metabolic syndrome, type 2 diabetes, CVD and cancer
(2)
.
Therefore, even a small weight loss may lead to health benefits.
Effective permanent weight loss and weight maintenance
depend largely on favourable changes in diet and physical
activity. Furthermore, instead of focusing on single macro-
nutrients (such as fat or sugar) one should focus on the heal-
thiness of the whole diet
(3,4)
. Of the healthy dietary patterns, the
Mediterranean diet has been the most extensively studied
regarding obesity and weight gain. Studies have reported
approximately 30 % lower risk of overweight or obesity and
10 % lower weight gain in participants with high adherence to
the Mediterranean diet compared with those with low adher-
ence
(5,6)
. A meta-analysis of randomised controlled trials
showed that Mediterranean diet results in 1·75 kg larger weight
loss compared with control diets
(7)
. Despite these proven health
effects of the Mediterranean diet, other populations may have
difficulties in adopting this dietary pattern due to differences in
food culture and resources.
In Nordic countries (Denmark, Finland, Iceland, Norway and
Sweden), the concept of the healthy Nordic diet is increasingly
studied
(8)
. The rationale behind the Nordic diet is to emphasis
use of healthy foods that can be produced locally and are
therefore easily available, affordable and culturally acceptable.
The diet is rich in foods grown in Nordic countries, such as
apples, and berries, roots and cabbages, rye, oats and barley,
low-fat milk products, rapeseed oil and fish (e.g. salmon and
Baltic herring). Furthermore, it is described to be low in red
meat, processed meat products and alcohol. Many studies have
indicated that this kind of diet may improve heart health
(9–13)
.
However, its effects on weight loss and maintenance are less
clear. In two randomised controlled trials, the healthy Nordic
diet lowered weight significantly more compared with the
Abbreviations: BSDS, Baltic Sea Diet Score; DILGOM, DIetary, Lifestyle and Genetic Determinants of Obesity and Metabolic syndrome; ICC, intra-class
correlation; WC, waist circumference.
*Corresponding author: N. Kanerva, email noora.kanerva@helsinki.fi
British Journal of Nutrition (2018), 120, 101–110 doi:10.1017/S0007114518001344
© The Authors 2018
https://doi.org/10.1017/S0007114518001344
Downloaded from https://www.cambridge.org/core. YBP Library Services, on 20 Aug 2018 at 13:17:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

control diet among obese individuals
(11,12,14)
. Weight loss in the
Nordic diet group was achieved by decreased energy intake,
which was achieved possibly due to diet’s large fibre con-
tent
(11,12)
. In cross-sectional epidemiological studies, adherence
to the healthy Nordic diet was associated with lower likelihood
of excess waist circumference (WC)
(15)
, but no association
between adherence to the diet and changes in body weight
or WC was observed in a cohort followed over 12 years
(16)
.
Furthermore, a case–control study reported no association
between healthy Nordic diet, body weight and WC among
participants from five European countries
(17)
. These long-
itudinal observations were based largely on self-reported data
which may have attenuated the results.
The main aim was to examine the association between the
healthy Nordic diet and 7-year changes in anthropometric
measures (weight, BMI and WC) among the general population.
Our study complements the previous studies which have
mainly focused on short-term results on high-risk groups.
Furthermore, we examined not onlythebaselineadherencetothe
diet, but also the changes occurred in the adherence during
thefollow-up.Lastlyasasecondaryaim,assomeparticipants
provided only self-reported anthropometrics at the follow-up,
we examined the agreement between self-reported and
nurse-measured anthropometrics among those participants
from who both measurements were available in order to interpret
our results.
Methods
Participants
The DIetary, Lifestyle and Genetic Determinants of Obesity and
Metabolic syndrome (DILGOM) 2007
(18)
study was an expan-
sion in the framework of the National FINRISK Studies
(19)
–the
main health monitoring system on Finnish adults every 5 years
since 1972 (online Supplementary Fig. S1). In brief, the FINRISK
2007 is a population-based, random sample of 10 000 Finns
aged 25–75 years from five large geographical regions of
Finland. The FINRISK 2007 participants (n6258, participation
rate 63 %) who went through clinical examination between
January and March were invited to the DILGOM baseline study
with more specific measurements regarding obesity and the
metabolic syndrome between April and June 2007. Of the
invited, 5024 (2325 men and 2699 women; participation rate
80 %) participated in the examination.
The follow-up of DILGOM study was conducted 7 years after
baseline between April and June 2014 similar than the baseline
study (online Supplementary Fig. S1). After excluding those
who died during the follow-up, who had moved outside
Finland, and those whose contact information was not available
(n434), 4581 participants were invited to the follow-up. Of the
invited, 3735 individuals participated (participation rate 82 %).
The follow-up was carried out in two groups:
(1) a health questionnaire and clinical examination for partici-
pants who lived in the capital area (Helsinki, Vantaa and
Espoo) and southwestern Finland, (n1312, participation rate
74 %); and
(2) questionnaires and measuring tape for those who lived in
North Karelia, Northern Savo and Ostrobothnia (n2423,
participation rate 87 %).
Of the 1312 participants who attended the clinical examina-
tion in 2014, both measured and self-reported height were
obtained from 557 men and 719 women, whereas measured
and self-reported weight were obtained from 520 men and
672 women. Furthermore, to validate self-measured WC, a
randomly selected subset that was stratified according to
participants’age, sex and living area was drawn from the
Group 1. Within this subset, both nurse- and self-measured WC
values were obtained from 140 men and 141 women.
The DILGOM and FINRISK studies have been conducted
according to the guidelines laid down in the Declaration of
Helsinki, and all procedures involving human subjects were
approved by the Ethics Committee of the Hospital District of
Helsinki and Uusimaa. Written informed consent was obtained
from all participants.
Questionnaires and measurements
Participants’age and sex were originally derived from Popula-
tion Information System. Along with the invitation letters, both
Groups 1 and 2 received standardised questionnaires via mail.
The main health questionnaire included questions on socio-
economic position, smoking, physical activity, medical history,
psychosocial factors and sleep. Leisure-time physical activity
was assessed by asking the participants to define their activity
outside work using four categories: inactive (mainly reading,
watching television or other light activities) moderately active
(walking, cycling, gardening or other activity at least 4 h/week),
active (brisk running, walking cross-country skiing, swimming
or other physically demanding activities at least 3 h/week), and
highly active (competition sports aiming and physically
demanding exercising several times in week). Furthermore, the
questionnaire queried on participants’weight, and height as
‘What is Your current weight? (kg)’and ‘What is Your current
height? (cm)’. Self-reported BMI (kg/m
2
) was calculated based
on answers to these questions. Those who were not invited to
the health examination (Group 2) received also measuring tape
and written instructions to measure their WC. The instructions
are described below.
In addition to the main health questionnaire, a validated
Finnish FFQ assessing dietary habits over 12 preceding months
was sent to all participants
(20,21)
. The average daily food,
nutrient and energy intakes were calculated, using the Finnish
National Food Composition Database (Fineli
®
) and in-house
software
(22)
. Participants whose daily energy intake (cut-offs)
corresponded to 0·5 % at both ends of the daily energy intake
distributions were excluded from analyses
(23)
.
In the clinical examination (Group 1), trained study nurses
measured participants’anthropometrics according to inter-
nationally standardised protocols
(24)
. These included a request
for the participant to remove shoes and heavy outer garments
and to empty their pockets. During the visit, nurses also
measured blood pressure and drew blood samples. Height
was measured to the nearest 0·1 cm, using stadiometer, and
weight to the nearest 0·1 kg, using balance scale. BMI based on
102 N. Kanerva et al.
https://doi.org/10.1017/S0007114518001344
Downloaded from https://www.cambridge.org/core. YBP Library Services, on 20 Aug 2018 at 13:17:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

nurse-measured values was calculated. Furthermore, partici-
pants’body composition (e.g. fat mass) was measured, using
electric bioimpedance scale (TANITA TBF-300MA; Tanita
Corporation of America, Inc.).
During the health examination, participants belonging to the
WC subset were asked to measure their WC using similar
measurement tape and written instructions as was mailed
for those who were not invited to the health examination
(Group 2). Study nurses were neither allowed to assist nor to
advise in the self-measurement.
The instructions advised participants to do the measurement
on bare skin or wearing only light clothing, and standing in front
of a mirror to facilitate correct placement of the measuring tape.
Participants were advised to place the measuring tape around the
waist to the mid-point between the lowest rib and iliac crest, then
breathe normally and read the measuring tape during exhale.
Thereafter, study nurses measured WC to the nearest 0·5cm
by placing a soft measuring tape around the waist to the
mid-point of lowest rib and iliac crest, and reading was done
during light exhale
(24)
.
Calculation of the dietary score
The development of the Baltic Sea Diet Score (BSDS) that
illustrates adherence to the healthy Nordic diet has been
described in detail elsewhere
(25)
. In brief, the BSDS includes
nine dietary components: high intake of Nordic fruits (apples,
pears and berries); Nordic vegetables (tomatoes, cucumber,
leafy vegetables, roots, cabbages, peas); Nordic cereals (rye, oat
and barley); low-fat and fat-free milk; Nordic fish (salmon and
freshwater fishes); ratio of PUFA:SFA and trans-fatty acids; low
intake of red and processed meat; and total fat (E%); and
moderate or low intake of alcohol (ethanol).
All components, except alcohol, were scored according to
sex-specific population consumption quartiles. Points were
assigned according to the predictable health impact of the
component. For fruits and berries, vegetables, cereals, low-fat
and fat-free milk, fish and the fat ratio, the lowest quartile was
coded as 0, the second lowest as 1, the third one as 2 and the
highest quartile as 3. For meat products and total fat, the highest
quartile was coded as 0, the second highest as 1, the next one
as 2 and the lowest quartile as 3. For alcohol, the cut-offs were
assigned according to the moderate consumption level recom-
mended in Nordic countries. Men consuming 20 g or less and
women consuming 10 g or less of alcohol per d received
1 point; otherwise, 0 points were given. The resulting BSDS
ranged from 0 to 25 points, with higher score values repre-
senting greater adherence to the healthy Nordic diet. Category
boundaries for BSDS fifths in men were: 1st fifth 0–9 points, 2nd
fifth 10–12 points, 3rd fifth 13–14 points, 4th fifth 15–16 points
and highest fifth 17–25 points. In women, boundaries were:
1st fifth 0–9 points, 2nd fifth 10–12 points, 3rd fifth 13–14 points,
4th fifth 15–16 points and highest fifth 18–25 points.
Statistical methods
We used R statistical software version 3.0.2
(26)
and SAS statistical
software version 9.3 (SAS Institute) to analyse the data.
Regarding missing values, only complete cases were analysed.
Descriptive statistics by sex, and by the BSDS fifths were
calculated as means and standard deviations or percentages
(%). Distribution of the main continuous variables was checked
using QQ-plots and histograms. The original sample size for the
DILGOM study was calculated on the estimation that a clinically
significant 5 % increase in BMI would occur among 43 % of
study participants during the 7-year follow-up. This gives a 90 %
power for detecting an OR of 1·1 at the αlevel of 0·05 per 1 SD
increase in cytokine concentration (e.g. IL-1ra).
In the validation of self-reported anthropometrics, intra-class
correlations (ICC) between the self-reported and nurse-measured
anthropometrics were calculated as the ratio of the subject var-
iance and the sum of the subject variance, the rater variance and
the residual (the agreement version). Furthermore, linear regres-
sion analysis was performed to investigate whether the difference
in anthropometric measures ((self-report) −(nurse-measurement))
varied across the mean of estimates ((self-report+nurse-
measurement)/2), as suggested by Bland & Altman
(27)
.Theslope
of the regression line was tested for a significant deviation from
zero. The slopes significantly different from zero indicate a rela-
tionship between adiposity status and the measurement error in
the self-report relative to the nurse-measurement.
In the main analysis, the association between BSDS and weight
change was analysed keeping men and women together as strong
evidence of interaction between the BSDS and sex did not emerge
in the study population (P>0·05). Change in weight, BMI and WC
during 7-years follow-up was calculated as: follow-up value −
baseline value. Positive values for weight/BMI/WC change indi-
cated that participant had gained weight during follow-up,
whereas negative values indicated weight loss. The change values
were then used as main outcome variables in linear regression
analysis. Baseline adherence to the healthy Nordic diet, measured
with the BSDS, was used as the main explanatory variable.
Moreover, we calculated the change between follow-up BSDS
and baseline BSDS, and these values were also used as expla-
natory variables to study the association between changes in
dietary adherence and change in weight/BMI/WC.
Confounding variables used in the models were first selected
based on the literature (age, sex, energy intake, physical
activity, smoking, education, previous pregnancies, baseline
weight and WC) and then confirmed with linear regression
analysis
(28)
. Based on this analysis, we adjusted Model 1 for age
(years; continuous), sex (dichotomic) and energy intake (kJ;
continuous); model 2 we additionally adjusted for smoking
(never smoker, quit >6 months ago, quit <6 months ago,
current smoker; categorical), leisure-time physical activity (low,
moderate and high or very high; categorical). Model 3 was
additionally adjusted for baseline weight/BMI/WC. When a
change in WC was examined as the main outcome, the third
model was also adjusted for change in BMI in order to control
the distribution of fat. Furthermore, when a change in BSDS was
examined as the main explanatory variable, baseline BSDS was
added to the third model as confounding variable.
Sensitivity testing was done by excluding participants who
reported myocardial infarction (follow-up, n73) or stroke
(follow-up, n69) at follow-up, as well as participants reporting
type 2 diabetes (baseline, n61; follow-up, n33), and cancer
The healthy Nordic diet and weight change 103
https://doi.org/10.1017/S0007114518001344
Downloaded from https://www.cambridge.org/core. YBP Library Services, on 20 Aug 2018 at 13:17:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

(baseline, n156; follow-up, n73) at baseline or during
follow-up. Furthermore, we excluded those participants who
had been pregnant between 2007 and 2014 (n103).
Results
Validation of anthropometric measures
The online Supplementary Table S1 describes background char-
acteristics and anthropometrics at the time of the follow-up study
for those participants who were included in the validation. The
measured height, weight and WC were on average similar to the
self-reported values. This was also demonstrated by the high ICC
agreement between self-reported and nurse-measured height
(ICC coefficient 0·96 in men and 0·97 in women), weight (ICC
coefficient 0·99 in men and women) and WC (ICC agreement 0·96
in men and 0·95 in women). Bland–Altman plots demonstrated a
negative bias toward higher values (Figs. 1 and 2). In other words,
the heavier or taller individual, the lower self-reported values
were compared with the nurse-measured values. The mean
difference between self-reports and measurements was <0·1cm
(SD 1·8cm for men and 1·6 cm for women) for height and 0·6kg
(SD 2·0kg for men and 1·8 kg for women) for weight. In WC, the
negative bias toward higher WC values was more obvious than in
height and weight measurements. However, the mean difference
between self-measurement and nurse-measurement was 0·2cm
(SD 3·4cm)for men and0·4cm (SD 4·1) for women (Fig. 3). These
results imply that self-reported and measured height and weight
could be used simultaneously without any correction coefficient
in the main analyses of this study.
Association between Baltic Sea Diet Score and weight change
Mean age of men and women was approximately 53 years at
baseline (Table 1). Most of the participants were never smokers
and were physically moderately active in their spare time.
15
10
5
0
–5
–10
–15
150 155 160 165 170 175 180 185 190 195 200
150 155 160 165 170 175 180 185 190145140
15
10
5
0
–5
–10
–15
Difference between self-measured and nurse-measured height (cm)
y=–0
.0221x+3
.8976*
Upper limit of
agreement 3.52 cm
Lower limit of
agreement –3.46 cm
y=–0
.0452x+7
.4891**
Upper limit of
agreement 3.21 cm
Lower limit of
agreement –3.10 cm
(a)
(b)
Mean of the self-measured and nurse-measured height (cm)
Fig. 1. Agreement between self-repor ted and nurse-measured height in the
DIetary, Lifestyle and Genetic Determinants of Obesity and Metabolic
syndrome 2014 Study men (a) and women (b). Linear regression analysis
was performed to investigate whether the difference in anthropometric
measures ((self-report) −(nurse-measurement)) varied across the mean of
estimates ((self-report + nurse-measureme nt)/2), as suggested by Bland &
Altman
(27)
. The dots represent observation made on each participant. The solid
line represents the slope of the regression line, which was tested for a
significant deviation from zero. The slopes significantly different from zero
indicate a relationship between height status and the measurement error in the
self-report relative to the nurse-measurement. The analysis included 557 men
and 719 women. * P<0·05, ** P<0·001.
20
15
10
5
0
–5
–10
–15
–2050 70 90 110 130 150
20
15
10
5
0
–5
–10
–15
–2035 55 75 95 115 135
Mean of the self-reported and nurse-measured weight (kg)
Difference between self-reported and nurse-measured weight (kg)
(a)
(b)
y=–0
.0152x+0
.689*
y=–0
.0273x+1
.3915**
Upper limit of
agreement 2.87kg
Lower limit of
agreement –3.99kg
Upper limit of
agreement 3.33kg
Lower limit of
agreement –4.53kg
Fig. 2. Agreement between self-reported and nurse-measured weight in the
DIetary, Lifestyle and Genetic Determinants of Obesity and Metabolic
syndrome 2014 Study men (a) and women (b). Linear regression analysis
was performed to investigate whether the difference in weight measures ((self-
report) −(nurse-measurement)) varied across the mean of estimates ((self-
report + nurse-measurement)/2), as suggested by Bland & Altman
(27)
. The dots
represent observation made on each participant. The solid line represents the
slope of the regression line, which was tested for a significant deviation from
zero. The slopes significantly different from zero indicate a relationship
between adiposity status and the measurement error in the self-report
relative to the nurse-measurement. The analysis included 520 men and 672
women. * P<0·05, ** P<0·001.
104 N. Kanerva et al.
https://doi.org/10.1017/S0007114518001344
Downloaded from https://www.cambridge.org/core. YBP Library Services, on 20 Aug 2018 at 13:17:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

During the 7-year follow-up, in general, the participants had
gained weight, their BMI had risen and also their WC had
increased. Furthermore, adherence to the healthy Nordic diet
(BSDS) decreased approximately 1 point during the follow-up.
Participants’age, educational attainment as well as the pro-
portion of participants in the most physically active group and
never smokers increased along with the BSDS (Table 2).
Moreover, participants in the highest BSDS fifth had improved
their BSDS and maintained their weight, BMI and WC better
compared with the lowest BSDS fifth. Intake of energy was on
average largest in the highest BSDS fifth.
In the main linear regression analyses, the baseline BSDS was
inversely associated with change in weight, BMI and WC
(Table 3). In other words, those with higher BSDS either
maintained their weight better or lost weight during the follow-
up. For weight and BMI, but not for WC, this association was
statistically significant when participants’leisure-time physical
activity, smoking and anthropometric measures at baseline
were taken into account as confounders (model 3). In the
sensitivity analysis, when participants whose weight or diet may
have changed due to illness or pregnancy during follow-up
were excluded the inverse association between baseline BSDS
and change in body weight and BMI remained statistically
significant (model 4).
When the change in BSDS during follow-up was used as the
main exposure, increasing BSDS during the follow-up asso-
ciated with better-maintained or decreased weight and BMI
(Table 3). This association became statistically significant
after taking account baseline anthropometric measures and
remained when participants whose weight or diet may had
changed due to illness or pregnancy during follow-up.
Finally, we did analysis which combined the baseline values
and changes during follow-up in both BSDS and body anthro-
pometrics (Table 4). Among those participants who increased
their BSDS during follow-up (n1119; mean change 2·9points,
SD 1·9), the baseline BSDS was inversely associated with changes
in weight and BMI when the baseline anthropometrics were
adjusted for (P<0·01; model 1–3). These associations strength-
ened after excluding participants whose weight or diet may have
changed due to illness or pregnancy during follow-up (model 4).
No association was observed between BSDS and WC in these
models. In participants who decreased or maintained their BSDS
during follow-up, (n1938; mean change −3·0 points, SD 2·5) no
statistically significant associations between the baseline BSDS
and anthropometric measures emerged (Table 4). As the results
15
10
5
0
–5
–10
–15
15
10
5
0
–5
–10
–1560 70 80 90 100 110 120 130 140 150
Mean of the self and nurse-measured waist circumference (cm)
Difference between the self and nurse-measured waist circumference (cm)
70 80 90 100 110 120 130
(a)
(b)
Upper limit of
agreement 6.51cm
Lower limit of
agreement –6.92cm
Upper limit of
agreement 7.58 cm
Lower limit of
agreement –8.63 cm
y=–0
.1161x+10
.9500**
y=–0
.0975x+7
.8767**
Fig. 3. Agreement between self-measured and nurse-measured waist
circumference in the DIetary, Lifestyle and Genetic Determinants of Obesity
and Metabolic syndrome 2014 Study men (a) and women (b). Linear
regression analysis was performed to investigate whether the difference in
waist circumference measures ((self-repor t) −(nurse-measurement)) varied
across the mean of estimates ((self-report + nurse-measurement)/2), as
suggested by Bland & Altman
(27)
. The dots represent observation made on
each participant. The solid line represents the slope of the regression line,
which was tested for a significant deviation from zero. The slopes significantly
different from zero indicate a relationship between adiposity status and the
measurement error in the self-report relative to the nurse-measurement. The
analysis included 140 men and 141 women. ** P<0·001.
Tab le 1. Characteristics of the DIetary, Lifestyle and Genetic Determi-
nants of Obesity and Metabolic syndrome 2014 follow-up participants
(Mean values and standard deviations; percentages)
Men Women
Mean SD Mean SD
nat follow-up 1400 1657
Age at baseline (years) 53·812·752·512·8
Educational attainment
at baseline (years)
12·54·013·14·0
Smoking status at baseline (%)
Never smoker 47·867·6
Quit >6 months ago 33·518·0
Quit <6 months ago 2·11·6
Current smoker 16·512·7
Leisure-time physical activity
at baseline (%)
Inactive 15·316·7
Moderately active 52·756·7
Active or highly active 32·026·7
Height at baseline (cm) 175·76·7162·86·2
Weight at baseline (kg) 82·812·470·013·4
Weight at 7-years follow-up (kg) 83·512·970·914·0
Weight change (kg) 0·75·41·05·5
BMI at baseline (kg/m
2
)26·83·726·45·1
BMI at follow-up (kg/m
2
)26·83·726·65·1
BMI change (kg/m
2
)<0·11·70·22·1
WC at baseline (cm) 95·510·886·012·8
WC at follow-up (cm) 97·811·087·912·8
WC change (cm) 2·36·02·06·8
BSDS at baseline 13·14·013·34·3
BSDS at follow-up 12·44·012·34·0
BSDS change −0·73·7−0·93·7
BSDS, Baltic Sea Diet Score; WC, waist circumference.
The healthy Nordic diet and weight change 105
https://doi.org/10.1017/S0007114518001344
Downloaded from https://www.cambridge.org/core. YBP Library Services, on 20 Aug 2018 at 13:17:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

for those who increased and those who maintained or decreased
their BSDS was different, we also analysed the interaction
between change in BSDS and baseline BSDS. However, we did
not find support for the hypothesis that change in BSDS during
follow-up would modify the association between baseline BSDS
and weight change (likelihood ratio test P=0·14).
Discussion
The healthy Nordic diet associated with lower weight gain
over 7 years of follow-up
In our prospective study, adherence to the healthy Nordic diet,
measured with the BSDS, was associated with decreased
weight and BMI in general Finnish population after 7 years of
follow-up. Those with the highest adherence to the healthy
Nordic diet at baseline had (on average) 0·3 kg higher weight
loss over the follow-up compared with those in the lowest
adherence to the diet. This may be interpreted as better
weight maintenance achieved by the healthy Nordic diet. The
association was found to be stronger among participants who
increased their adherence to the diet during the follow-up. No
association between adherence to the diet and change in WC
was observed. Adjustment of participants’leisure-time physical
activity, smoking and anthropometric measures at baseline did
not change the result markedly. Furthermore, the results
were similar for the whole cohort and when participants
whose weight may have been affected by illness or pregnancy
were excluded.
Our results are generally in line with the literature indicating
that dietary pattern (e.g. Mediterranean diet or Healthy Eating
Index) which is rich in fruits and vegetables, wholegrain, fish,
oils, nuts and dairy products as well as low in red and processed
meat, sugar-sweetened beverages and high-fat foods may pre-
vent from weight gain
(6,29–33)
. Epidemiological studies on the
potential of the healthy Nordic diet to prevent weight gain are
still few. One longitudinal cohort study
(16)
and one case–
control
(17)
have been previously published in this research area.
In contrast to our findings, either of these studies found an
association between the healthy Nordic diet and change in
anthropometric measures. The study of Li et al.
(16)
consisted of
Swedish women and was based on self-reports. Moreover,
Table 2. Characteristics of the DIetary, Lifestyle and Genetic Determinants of Obesity and Metabolic syndrome 2014 follow-up participants by adherence to
the healthy Nordic diet
(Mean values and standard deviations; percentages)
BSDS fifths*
13 5
Characteristics Mean SD Mean SD Mean SD P†
n(3057) 785 583 395
Women (%) 55·054·947·10·002
Age (years) 48·812·955·212·258·510·5<0·001
Educational level (%) <0·001
Low 31·227·216·8
Middle 37·334·835·5
High 31·738·047·7
Smoking status (%) <0·001
Never smoker 48·659·165·8
Quit >6 months ago 24·726·825·8
Quit <6 months ago 2·01·41·5
Current smoker 24·612·76·8
Leisure-time physical activity (%) <0·001
Inactive 23·511
·29·5
Moderately active 51·757·154·7
Active or highly active 24·931·735·8
Height (cm) 168·98·5168·49·3168·99·60·06
Weight (kg) 76·315·375·813·976·913·70·07
Weight change (kg) 1·85·90·75·0−0·55·3<0·001
Those who gained weight (%) 63·553·346·3<0·001
BMI (kg/m
2
)26·74·82 26·74·18 26·94·15 0·41
BMI change (kg/m
2
)0·47 2·07 0·10 1·78 −0·38 1·94 <0·001
Those who increased BMI (%) 58·551·843·3<0·001
WC (cm) 90·813·690·312·491·611·90·17
WC change (cm) 2·86·72·06·21·26·4<0·001
Those who increased WC (%) 65·262·259·00·06
Energy intake (kJ/d) 9700 3450 10 810 3820 11 430 3810 <0·001
Baseline BSDS 10·03·513·73·517·23·2<0·001
BSDS change −2·83·4−0·23·51·83·0<0·001
Those who increased BSDS (%) 17·743·165·8<0·001
BSDS, Baltic Sea Diet Score; WC, waist circumference.
* Category boundaries for BSDS fifths in men were: 1st fifth 0–9 points, 2nd fifth 10–12 points, 3rd fifth 13–14 points, 4th fifth 15–16 points and highest fifth 17–25 points. In women,
boundaries were: 1st fifth 0–9 points, 2nd fifth 10–12 points, 3rd fifth 13–14 points, 4th fifth 15–16 points and highest fifth 18–25 points.
†Unadjusted Pvalues were derived from logistic regression for continuous variables and from χ
2
test for categorical variables.
106 N. Kanerva et al.
https://doi.org/10.1017/S0007114518001344
Downloaded from https://www.cambridge.org/core. YBP Library Services, on 20 Aug 2018 at 13:17:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

dietary data were collected at baseline. Compared with Li
et al.
(16)
, our study had most of the anthropometric data mea-
sured by trained nurses and dietary intake measured two times,
which may have lowered information bias (e.g. measurement
error) and allowed us to detect statistically significant differ-
ences. In the case–control study, participants were from five
European countries, including one Nordic country. It is likely
that adherence to the healthy Nordic diet was at a much lower
level in the non-Nordic countries, which may have contributed
to the null finding. The Mediterranean diet
(34)
has been asso-
ciated with weight loss and lower WC among Western popu-
lations
(6,7,35)
, but the association has been substantially stronger
among Mediterranean populations than other Europeans,
especially Nordic populations
(6)
. Lastly, both of the previous
epidemiological studies used the Nordic Diet Score
(10)
, which
includes six food items to characterise the healthy Nordic diet.
The BSDS used in our study is more comprehensive compared
with the Nordic Diet Score. This may explain the difference in
the findings to some extent.
Although not entirely comparable to our study, intervention
studies have also reported an association between the healthy
Nordic diet and decreased weight. In the Danish intervention,
of those who completed the 26-week study, ninety-one parti-
cipants in the intervention group had 3·2 kg greater weight loss
than the fifty-eight controls
(12)
. The weight loss was followed
by regaining towards pre-intervention level during the next
52 weeks
(14)
. In the end, from screening to week 78 there was
no difference in total weight loss between the diets. Adamsson
et al.
(11)
found a 4 % decrement in weight among forty-four
participants following the healthy Nordic diet for 6 weeks
compared with forty-two controls. In contrast to these trials, a
multi-centre intervention reported no changes in body weight
during 18–24 week intervention
(9)
. However, this study aimed
to isoenergetic design and not to weight loss
(9)
.
Self-reported and -measured anthropometrics reached high
agreement among adult Finns
Our results indicate that self-reports of height, weight and WC
have a very high agreement with nurse-measured values among
the adult population. These results support the rationale to use
the subjective values solely or simultaneously with objective
data in epidemiological analysis without correction equations.
The majority of studies have reported good consistency
between self-reported and measured height and weight
(36)
, but
also poor consistency has been observed in some studies
(37)
.
Generally, people tend to overrate their height or underrate
their weight
(38–40)
. Furthermore, the degree of underrating of
weight has been shown to be higher in individuals with over-
weight and obesity and in younger women whereas in men the
results on weight misreporting have been more incon-
sistent
(37,41–43)
. Furthermore, overrating of height has been
Table 3. Linear association between Baltic Sea Diet Score (BSDS) and
change in body anthropometrics during 7-years follow-up*
(β-Coefficients with their standard errors)
Main exposure
Baseline BSDS Change in BSDS
βSE PβSE P
Change in body weight (kg)
Model 1†−0·062 0·024 0·010 −0·044 0·026 0·092
Model 2‡−0·048 0·025 0·054 −0·041 0·026 0·117
Model 3§ −0·045 0·025 0·067 −0·103 0·030 0·001
Model 4|| −0·056 0·027 0·043 −0·115 0·033 0·001
Change in BMI (kg/m
2
)
Model 1 −0·024 0·009 0·006 −0·014 0·009 0·128
Model 2 −0·018 0·009 0·042 −0·014 0·009 0·143
Model 3 −0·017 0·009 0·055 −0·036 0·011 0·001
Model 4 −0·021 0·010 0·031 −0·042 0·012 <0·001
Change in WC (cm)
Model 1 −0·026 0·028 0·360 −0·044 0·031 0·148
Model 2 −0·021 0·029 0·480 −0·041 0·031 0·187
Model 3 −0·026 0·029 0·368 −0·059 0·030 0·048
Model 4 0·008 0·022 0·731 −0·017 0·027 0·523
WC, waist circumference.
* Estimates and Pvalues were derived from linear regression analysis, in which the
baseline or change in BSDS was used as the main exposure and change in
anthropometric measures (baseline−follow-up) as the outcome.
†Model 1 is adjusted for age, sex and baseline energy intake.
‡Model 2 is model 1 further adjusted for baseline leisure-time physical activity and
smoking.
§ Model 3 is model 2 fur ther adjusted for baseline weight/BMI/WC depending on
which of these measures was used as the main outcome. When weight was
examined as the main outcome, the model was adjusted for baseline height. When
the change in BSDS was examined as the main exposure, this model was also
adjusted for baseline BSDS.
|| Model 4 is model 3, excluding participants who reported myocardial infarction or
stroke during follow-up; participants reporting type 2 diabetes, and cancer at
baseline or during follow-up; and women who had been pregnant during follow-up.
Tab le 4 . Linear association between baseline Baltic Sea Diet Score
(BSDS) and change in body anthropometrics by changes in BSDS during
7 years of follow-up*
(β-Coefficients with their standard errors)
Stratification by a change in BSDS during follow-up
Increased BSDS
(n1119)
Decreased/unchanged BSDS
(n1938)
βSE PβSE P
Change in body weight (kg)
Model 1†−0·149 0·044 0·001 −0·102 0·056 0·068
Model 2‡−0·136 0·046 0·003 −0·084 0·057 0·142
Model 3§ −0·129 0·045 0·004 −0·075 0·057 0·186
Model 4|| −0·163 0·050 0·001 −0·052 0·036 0·152
Change in BMI (kg/m
2
)
Model 1 −0·061 0·016 <0·001 −0·037 0·020 0·059
Model 2 −0·055 0·017 0·001 −0·029 0·020 0·152
Model 3 −0·052 0·016 0·002 −0·024 0·020 0·233
Model 4 −0·061 0·018 0·001 −0·018 0·013 0·165
Change in WC (cm)
Model 1 −0·090 0·053 0·088 −0·084 0·066 0·205
Model 2 −0·089 0·055 0·106 −0·071 0·068 0·295
Model 3 −0·089 0·053 0·092 −0·055 0·067 0·411
Model 4 0·032 0·040 0·419 −0·018 0·031 0·561
WC, waist circumference.
* Estimates and Pvalues were derived from linear regression analysis, in which the
baseline BSDS was used as the main exposure and change in anthropometric
measures (baseline−follow-up) as the outcome.
†Model 1 is adjusted for age, sex and baseline energy intake.
‡Model 2 is model 1 further adjusted for baseline leisure-time physical activity and
smoking.
§ Model 3 is model 2 further adjusted for baseline weight/BMI/WC depending on
which of the measures was used as the main outcome. When weight was
examined as the main outcome, this model was adjusted for baseline height.
|| Model 4 is model 3 excluding participants who reported myocardial infarction or
stroke during follow-up; participants reporting type 2 diabetes, and cancer at
baseline or during follow-up; and women who had been pregnant during follow-up.
The healthy Nordic diet and weight change 107
https://doi.org/10.1017/S0007114518001344
Downloaded from https://www.cambridge.org/core. YBP Library Services, on 20 Aug 2018 at 13:17:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

found to be to higher among older people and especially
among older men
(37)
. Patterns of under- and overrating could
be seen in our results, although our data generally showed
good consistency between self-reported and measured height
and weight, mean difference between these values being <1cm
for height and 0·6 kg for weight.
Thus far, few studies have validated self-measured WC
against nurse-measured WC. Results have been conflicting as
studies have reported either systematic over-
(44)
or under-
rating
(45)
of self-measurements compared with nurse-measured
values. A Dutch study (with fairly similar population char-
acteristics to ours) found that the participants’self-measurement
overrated WC averagely by 6 cm compared with nurse-
measurement
(44)
. In contrast, the largest validation study pub-
lished so far that concerned the Norfolk cohort of the European
Prospective Investigation into Cancer and Nutrition (EPIC)
observed significant underestimation of self-reported
WC compared with the measured values (mean difference
4–6 cm)
(40)
. A similar trend, but lower degree of under-
estimation (mean difference 2–3 cm) was reported in the
Oxford cohort of the EPIC study
(46)
. These are fairly similar
results compared with our study, taking into account that our
results showed an even lower degree of underestimation
(mean difference 0·4 cm). However, this underestimation may
have led to the non-significant finding between the BSDS and
change in WC in our study.
Methodological considerations
The strengths of our data include high participation rate (82 %)
and a population-based sampling. Because we had extensive
data from the baseline, we were able to explore and then adjust
the analyses for relevant variables. We had trained and
experienced study nurses conducting the anthropometric
measurements according to standardised procedures
(24)
whose
work was monitored during the data collection.
There may have been some selection among those who came
to the follow-up. Comparing the baseline characteristics to the
whole DILGOM population, those who attended the follow-up
where older, less often smokers and physically inactive at their
leisure-time, and were more likely to have high education.
This limits the generalisability of our results to adult Finns.
Considering that those who did not participate had an unheal-
thier lifestyle, which makes them more likely to gain weight
during the follow-up, our results may be an underestimation of
the true association between the BSDS and weight change.
However, taking account that we do not know about the
possible changes in lifestyle that these participants have made
during the follow-up, estimating the effect of selection bias to
our results is at best speculative. It should also be pointed out
that non-participation is a problem concerning all health
surveys, thus our study is not unique in this sense.
Regarding weight and height measures, we did not define the
validation sample of these measures similarly to WC, but
we were able to compare the difference between height and
weight reported on the health questionnaire with the ones
measured by the study nurse. We do not know if the partici-
pants actually measured their weight before filling in the
questionnaire. Thus, some participants may have reported
values that they measured several weeks, months or even years
ago. Considering the level of agreement between measured
weight and self-reported weight, most of the participants had to
have measured their anthropometrics quite recently.
Using a validated FFQ to assess baseline as well as changes in
dietary habits is also a strength of our study
(20,21)
. The same
food composition database and nutrient intake calculation
software were also used. Validation studies have revealed that
our questionnaire has a good ability to rank subjects according
to their relative nutrient and food intakes
(20,21,47)
. It also pro-
vides a fairly good reproducibility. The BSDS has been proven
to be a valid measure of participants’adherence to the healthy
Nordic diet
(25)
, and it has been used in many epidemiological
studies thus far in both adult and paediatric populations over
the last 5 years
(13,48–50)
.
As a limitation, nutrition research, including the FFQ, gene-
rally involves overestimation of healthy and underestimation of
unhealthy food consumption. Health conscious people, women
and overweight participants are especially prone to misreport-
ing
(47)
. This may have led to some misclassifications in the
BSDS fifths or dragged the BSDS cut-off points higher, which
may have weakened the observed associations. The BSDS also
has its weaknesses. Although a predefined score, such as the
BSDS, better enables the capture of cumulative dietary expo-
sure, some confounding due to correlations in the intake of
various dietary factors still remains.
Conclusions
Our study showed that high initial adherence to the healthy
Nordic diet, as well as increase in the adherence, may promote
long-term weight maintenance. We also found that self-
reported/measured anthropometrics have high agreement
with nurse-measured values among the adult population
which adds the credibility of our results. Further longitudinal
studies among Nordic populations are needed to confirm
our results.
Acknowledgements
This study was funded by the Finnish Foundation for Cardio-
vascular Research (N. K.). The DILGOM 2014 study was funded
by Juho Vainio Foundation (P. J.).
P. J., S. M., A. H., N. E. K. and N. K. participated in planning
and conducting the baseline and follow-up studies. M. M. par-
ticipated in data clearance and checking. K. H. and N. K. ana-
lysed the data. N. K. wrote the manuscript and had the primary
responsibility of its final content. All authors have critically
revised the manuscript for important intellectual content and
approved the final version.
The authors declare that there are no conflicts of interest.
Supplementary material
For supplementary material/s referred to in this article, please
visit https://doi.org/10.1017/S0007114518001344
108 N. Kanerva et al.
https://doi.org/10.1017/S0007114518001344
Downloaded from https://www.cambridge.org/core. YBP Library Services, on 20 Aug 2018 at 13:17:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

References
1. NCD Risk Factor Collaboration (NCD-RisC) (2017) 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 390, 2627–2642.
2. Lim SS, Vos T, Flaxman AD, et al. (2012) A comparative risk
assessment of burden of disease and injury attributable to 67
risk factors and risk factor clusters in 21 regions, 1990–2010: a
systematic analysis for the Global Burden of Disease
Study 2010. Lancet 380, 2224–2260.
3. Willett WC & Stampfer MJ (2013) Current evidence on
healthy eating. Annu Rev Public Health 34,77–95.
4. Jacobs DR & Tapsell LC (2007) Food, not nutrients, is the
fundamental unit in nutrition. Nutr Rev 65, 439–450.
5. Buckland G, Bach A & Serra-Majem L (2008) Obesity and the
Mediterranean diet: a systematic review of observational and
intervention studies. Obes Rev 9, 582–593.
6. Romaguera D, Norat T, Vergnaud AC, et al. (2010) Medi-
terranean dietary patterns and prospective weight change in
participants of the EPIC-PANACEA project. Am J Clin Nutr 92,
912–921.
7. Esposito K, Kastorini C-M, Panagiotakos DB, et al. (2011)
Mediterranean diet and weight loss: meta-analysis of rando-
mized controlled trials. Metab Syndr Relat Disord 9,1–12.
8. Bere E & Brug J (2009) Towards health-promoting and
environmentally friendly regional diets –a Nordic example.
Public Health Nutr 12,91–96.
9. Uusitupa M, Hermansen K, Savolainen MJ, et al. (2013) Effects
of an isocaloric healthy Nordic diet on insulin sensitivity, lipid
profile and inflammation markers in metabolic syndrome –a
randomized study (SYSDIET). J Intern Med 274,52–66.
10. Olsen A, Egeberg R, Halkjaer J, et al. (2011) Healthy aspects of
the Nordic diet are related to lower total mortality. J Nutr 141,
639–644.
11. Adamsson V, Reumark A, Fredriksson IB, et al. (2011) Effects
of a healthy Nordic diet on cardiovascular risk factors in
hypercholesterolaemic subjects: a randomized controlled trial
(NORDIET). J Intern Med 269, 150–159.
12. Poulsen SK, Due A, Jordy AB, et al. (2014) Health effect of the
New Nordic Diet in adults with increased waist circumference:
a 6-mo randomized controlled trial. Am J Clin Nutr 99,35–45.
13. Kanerva N, Kaartinen NE, Rissanen H, et al. (2014) Associa-
tions of the Baltic Sea diet with cardiometabolic risk factors –a
meta-analysis of three Finnish studies. Br J Nutr 112, 616–626.
14. Poulsen SK, Crone C, Astrup A, et al. (2015) Long-term
adherence to the New Nordic Diet and the effects on body
weight, anthropometry and blood pressure: a 12-month
follow-up study. Eur J Nutr 54,67–76.
15. Kanerva N, Kaartinen NE, Schwab U, et al. (2013) Adherence
to the Baltic Sea diet consumed in the Nordic countries is
associated with lower abdominal obesity. Br J Nutr 109,
520–528.
16. Li Y, Roswall N, Strom P, et al. (2015) Mediterranean and
Nordic diet scores and long-term changes in body weight and
waist circumference: results from a large cohort study. Br J
Nutr 114, 2093–2102.
17. Roswall N, Angquist L, Ahluwalia TS, et al. (2014) Association
between Mediterranean and Nordic diet scores and changes in
weight and waist circumference: influence of FTO and
TCF7L2 loci. Am J Clin Nutr 100, 1188–1197.
18. Konttinen H, Männistö S, Sarlio-Lähteenkorva S, et al. (2010)
Emotional eating, depressive symptoms and self-reported
food consumption. A population-based study. Appetite 54,
473–479.
19. Vartiainen E, Laatikainen T, Peltonen M, et al. (2010)
Thirty-five-year trends in cardiovascular risk factors in Finland.
Int J Epidemiol 39, 504–518.
20. Männistö S, Virtanen M, Mikkonen T, et al. (1996) Reprodu-
cibility and validity of a food frequency questionnaire in a
case-control study on breast cancer. J Clin Epidemiol 49,
401–409.
21. Kaartinen NE, Tapanainen H, Valsta LM, et al. (2012) Relative
validity of a FFQ in measuring carbohydrate fractions, dietary
glycaemic index and load: exploring the effects of subject
characteristics. Br J Nutr 107, 1367–1375.
22. Reinivuo H, Hirvonen T, Ovaskainen ML, et al. (2010) Dietary
survey methodology of FINDIET 2007 with a risk assessment
perspective. Public Health Nutr 13, 915–919.
23. Meltzer HM, Brantsaeter AL, Ydersbond TA, et al. (2008)
Methodological challenges when monitoring the diet of
pregnant women in a large study: experiences from the
Norwegian Mother and Child Cohort Study (MoBa). Matern
Child Nutr 4,14–27.
24. Tolonen H, Koponen P, Aromaa A, et al. (2008) Recommen-
dations for the Health Examination Surveys in Europe. no. B
21/2008. Helsinki: Kansanterveyslaitos.
25. Kanerva N, Kaartinen NE, Schwab U, et al. (2014) The Baltic
Sea Diet Score: a tool for assessing healthy eating in Nordic
countries. Public Health Nutr 17, 1697–1705.
26. R Development Core Team (2011) R: A Language and
Environment for Statistical Computing, vol. 2012. Vienna: R
Development Core Team.
27. Bland JM & Altman DG (1986) Statistical methods for assessing
agreement between two methods of clinical measurement.
Lancet Lond Engl 1, 307–310.
28. Rothman KJ (editor) (1986) Modern Epidemiology, 6th ed.
Boston, MA and Toronto: Little, Brown and Company.
29. Buckland G, Bach A & Serra-Majem L (2008) Obesity and the
Mediterranean diet: a systematic review of observational and
intervention studies. Obes Rev Off J Int Assoc Study Obes 9,
582–593.
30. Gao SK, Beresford SA, Frank LL, et al. (2008) Modifications to
the Healthy Eating Index and its ability to predict obesity: the
Multi-Ethnic Study of Atherosclerosis. Am J Clin Nutr 88,
64–69.
31. Kimokoti RW, Newby PK, Gona P, et al. (2010) Diet quality,
physical activity, smoking status, and weight fluctuation are
associated with weight change in women and men. J Nutr
140, 1287–1293.
32. Romaguera D, Angquist L, Du H, et al. (2011) Food compo-
sition of the diet in relation to changes in waist circumference
adjusted for body mass index. PLoS ONE 6, e23384.
33. Mozaffarian D, Hao T, Rimm EB, et al. (2011) Changes in diet
and lifestyle and long-term weight gain in Women and men.
N Engl J Med 364, 2392–2404.
34. Trichopoulou A, Kouris-Blazos A, Wahlqvist ML, et al. (1995)
Diet and overall survival in elderly people. BMJ 311,
1457–1460.
35. Kastorini CM, Milionis HJ, Esposito K, et al. (2011) The effect
of Mediterranean diet on metabolic syndrome and its com-
ponents: a meta-analysis of 50 studies and 534,906 individuals.
J Am Coll Cardiol 57, 1299–1313.
36. Kuczmarski MF, Kuczmarski RJ & Najjar M (2001) Effects of
age on validity of self-reported height, weight, and body mass
index: findings from the Third National Health and
Nutrition Examination Survey, 1988–1994. J Am Diet Assoc
101,28–34.
37. Shields M, Connor Gorber S & Tremblay MS (2008) Estimates
of obesity based on self-report versus direct measures. Health
Rep 19,61–76.
The healthy Nordic diet and weight change 109
https://doi.org/10.1017/S0007114518001344
Downloaded from https://www.cambridge.org/core. YBP Library Services, on 20 Aug 2018 at 13:17:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

38. Connor Gorber S, Tremblay M, Moher D, et al. (2007)
A comparison of direct vs. self-report measures for assessing
height, weight and body mass index: a systematic review.
Obes Rev Off J Int Assoc Study Obes 8, 307–326.
39. Tolonen H, Koponen P, Mindell JS, et al. (2014) Under-
estimation of obesity, hypertension and high cholesterol by
self-reported data: comparison of self-reported information
and objective measures from health examination surveys.
Eur J Public Health 24, 941–948.
40. Park JY, Mitrou PN, Keogh RH, et al. (2011) Effects of body size
and sociodemographic characteristics on differences between self-
reported and measured anthropometric data in middle-aged men
and women: the EPIC-Norfolk study. Eur J Clin Nutr 65, 357–367.
41. Taylor AW, Dal Grande E, Gill TK, et al. (2006) How valid are
self-reported height and weight? A comparison between
CATI self-report and clinic measurements using a large
cohort study. Aust N Z J Public Health 30, 238–246.
42. Stommel M & Schoenborn CA (2009) Accuracy and usefulness
of BMI measures based on self-reported weight and height:
findings from the NHANES & NHIS 2001–2006. BMC Public
Health 9, 421.
43. Hattori A & Sturm R (2013) The obesity epidemic and changes
in self-report biases in BMI. Obesity (Silver Spring) 21,856–860.
44. Contardo Ayala AM, Nijpels G & Lakerveld J (2014)
Validity of self-measured waist circumference in adults at risk
of type 2 diabetes and cardiovascular disease. BMC Med
12, 170.
45. Khunti K, Taub N, Webb D, et al. (2012) Validity of self-
assessed waist circumference in a multi-ethnic UK population.
Diabet Med 29, 404–409.
46. Spencer EA, Roddam AW & Key TJ (2004) Accuracy of self-
reported waist and hip measurements in 4492 EPIC-Oxford
participants. Public Health Nutr 7, 723–727.
47. Paalanen L, Männistö S, Virtanen MJ, et al. (2006) Validity of a
food frequency questionnaire varied by age and body
mass index. J Clin Epidemiol 59, 994–1001.
48. Männikkö R, Komulainen P, Schwab U, et al. (2015) The
Nordic diet and cognition –The DR’s EXTRA Study. Br J Nutr
114, 231–239.
49. Haapala EA, Eloranta A-M, Venäläinen T, et al. (2015)
Associations of diet quality with cognition in children - the
Physical Activity and Nutrition in Children Study. Br J Nutr
114, 1080–1087.
50. Perälä M-M, von Bonsdorff MB, Männistö S, et al. (2017) The
healthy Nordic diet predicts muscle strength 10 years later in
old women, but not old men. Age Ageing 46, 588–594.
110 N. Kanerva et al.
https://doi.org/10.1017/S0007114518001344
Downloaded from https://www.cambridge.org/core. YBP Library Services, on 20 Aug 2018 at 13:17:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.