International Journal of Obesity (2000) 24, 435±442
ß 2000 Macmillan Publishers Ltd All rights reserved 0307±0565/00 $15.00
www.nature.com/ijo
Do we eat less fat, or just report so?
BL Heitmann1,2,3*, L Lissner4 and M Osler1,5
1
Unit for Dietary Studies at the The Copenhagen County Centre for Preventive Medicine, Department of Medicine CF, Glostrup
University Hospital, Denmark; 2Institute of Preventive Medicine, Copenhagen Hospital Corporation, H:S Copenhagen Municipal
Hospital, Denmark; 3Research Department of Human Nutrition, Royal Veterinary and Agricultural University, Copenhagen Denmark;
4
Centre for Nutritional Epidemiology, Department of Internal Medicine, GoÈteborg University, GoÈteberg, Sweden; and 5Institute of
Public Health, University of Copenhagen, Copenhagen, Denmark
OBJECTIVE: To examine secular trends in diet reporting error.
METHODS: Dietary information was obtained from 228 Danish men and women in 1987 ± 88, and from 122 men and
women in 1993 ± 94.
RESULTS: Bias in dietary reporting of energy and protein intake was assessed by comparing reported intake with
intake data, estimated from 24 h nitrogen output, validated by administering P-aminobenzoic acid, and estimated 24 h
energy expenditure. Total energy was under-reported more than energy from protein at both surveys, suggesting that
energy from other nutrients, like fat and=or carbohydrate, must have been under-reported too. There was a greater
under-reporting for energy than for protein in 1993 ± 94 (29%) than in 1987 ± 88 (15%). Obesity was positively
associated with under-reporting, both in 1987 ± 88 and in 1993 ± 94.
CONCLUSION: The higher macro-nutrient speci®c error in 1993 ± 94 compared to 1987 ± 88 may re¯ect a trend to
increasingly omitting fat and=or carbohydrate-rich foods in dietary reporting. This may be a consequence of increased
awareness of diet intake, which, in turn, may be related to intensi®ed public health campaigns to reduce intake of fat
and=or simple carbohydrate. These results may have consequences for our understanding of the apparent decline in
dietary fat and associated health bene®ts.
International Journal of Obesity (2000) 24, 435±442
Keywords: secular trends; dietary under-reporting; obesity; macro-nutrient intakes
Introduction
It is generally recommended that individuals consume
no more than 30 ± 35% of their total energy as fat,1,2
but most population studies suggest that fat consumption exceeds these levels.3,4,6 However, a number of
countries have reported a trend suggesting a reduction
in energy from fat over the past 20 y.3 ± 6 For instance,
recent data suggests that North Americans have
reduced their fat intake from about 40% energy
from fat to 33%,7 the Norwegians from 40% to
34%,8 and Finns from 38% to 34%9 during the years
between 1960 and 1990. This reversal of trend has
generally been attributed to intensi®ed of®cial recommendations, and to a delayed reaction to earlier public
health campaigns to reduce fat intake.
It may seem paradoxical that the population is
simultaneously decreasing their energy intake from
fat while getting fatter.10 Over-consumption with
increased energy, but a similar fat intake, is one
explanation, and there is some suggestion that the
reported absolute fat intake has, in fact, remained
constant and total energy intake increased.3 However,
other explanations include a less physically active
*Correspondence: Unit for Dietary Studies at the Centre for
Preventive Medicine, Glostrup University Hospital, DK-2600
Glostrup, Denmark.
E-mail: [email protected]
Received 26 April 1999; revised 23 August 1999; accepted
4 November 1999
lifestyle or under-reporting of fat. Indeed, several
reports have shown that many overweight individuals
under-report their food intake substantially,11,12 and
that, in particular, foods high in fat and=or carbohydrate may be under-reported.13 We, therefore, hypothesized that the apparent trend of decrease in fat
intake re¯ects an increasing tendency to under-report
dietary fat, rather than a true decrease in fat intake.
Denmark is one nation in which recommendations
to reduce dietary fat are believed to have had a recent
effect on intake.3 Several national campaigns have
been launched, especially since the beginning of the
1990s.14 Indeed, recent data from two independent
population-based dietary surveys3,15 suggests a reduction in fat intake from around 42% to 38% of total
energy. We used data from one of the above mentioned population studies to examine time-dependent
trends in fat intake and diet reporting error between
1987 ± 88 and 1993 ± 94.
Subjects and methods
The study was part of the Danish MONICA project
(an international study conducted under the auspices
of the World Health Organisation to monitor trends in
and determinants of mortality from cardiovascular
disease). A random sample of adult Danes aged 30,
40, 50, and 60 y was drawn from the national personal
register in 1982 comprising 4581 subjects.13 Of this
Do we eat less fat?
BL Heitmann et al
436
group 3608 (79%) participated in a baseline health
examination. In December 1987 to November 1988 a
dietary survey was performed inviting a subset of 552
subjects from the population sample to participate.16,17
A follow-up dietary survey was conducted from May
1993 to November 1994. All subjects invited in
1987 ± 88 were re-invited in 1993 ± 94.
The subjects who were examined in 1993 ± 94 were,
on average, 5 y older than those appearing in 1987 ±
88 (P > 0.01). Hence, all analyses were adjusted for
age differences. There were no differences in baseline
age, body weight or BMI for the men and women who
were included in the two diet surveys and the remaining group (3258=3608, all P > 0.33). Participation has
been described in detail elsewhere.11,17
The project was approved by the Ethics Committee
for Copenhagen County, and is in accordance with the
Helsinki II Declaration on Human Rights.
Questionnaire data
All participants answered questions about physical
activity during leisure (sitting most of the time; light
activity at least 4 h a week; active in sport at least 3 h a
week or heavy work during leisure; active in competitive sport several times a week), physical activity
during work (no work; sitting most of the time; light
activity, walking around; walking and carrying most of
the time; strenuous physical work), and about educational level ( < 7 y, 7 ± 12 y or > 12 y of schooling).
Diet
The same trained dietician interviewed all the subjects, both in 1987 ± 88 and in 1993 ± 94, by taking a
diet history interview. Average daily intakes were
calculated from responses describing the previous
month. Data on meal patterns, dishes and food items
were obtained with a precoded interview form. Quantities were assessed with food models, series of
photographs, cups and measures. Calculations of
nutrients were carried out with the DANKOST program, which is derived from the Danish food composition tables.18
was measured to nearest 0.5 cm, with subjects standing without shoes, heels together, and head in the
horizontal Frankfurt plane.
Body weight (BW) was measured to nearest 0.1 kg
(SECA weighing scales), with the subjects wearing
only underwear.
Measurements of electrical impedance
A BIA-103 RJL-system-analyser (RJL Detroit) was
used to measure electrical impedance according to the
instructions given by the manufacturer. The measurement was taken with tetra polar electrode placement,
with electrodes placed on the right hand and foot, at
the distal metacarpals and metatarsals, respectively,
and between the distal prominences of the radius and
the ulna at the wrist and the medial and lateral
malleoli at the ankle.
The algorithm used to estimate body fat from
impedance (BF) had been developed in a subgroup
of the same sample of Danes:22
BF …kg† ˆ 0:819BW …kg† ÿ 0:279Ht2 =R …cm2 =ohm†
ÿ 0:064sex BW …kg† ‡ 0:077age …y†
ÿ 0:231Ht …cm† ‡ 14:941
where BF ˆ body fat, BW ˆ body weight, Ht ˆ height
and R ˆ resistance. Sex is coded as 1 for men and 0
for women. Basal energy expenditure was calculated
according to the formula by Garby et al:23
BEEJoule ˆ 116.76FFM ‡ 26.88BF, where fat-free
mass (FFM) is the difference between body weight
and body fat.
Average 24 h energy expenditure can be expressed
as a multiple of basal energy expenditure, known as
the physical activity level. The physical activity level
was assumed to be 1.55 (men) and 1.56 (women)
among those who were unemployed, or whose work
was classi®ed as sedentary; 1.78 and 1.64, respectively among those whose work was classi®ed as light
activity; and 2.10 and 1.82, respectively, among those
engaged in heavy or strenuous work, or engaged
actively in sports.24
Collection and analyses of urine
All subjects were instructed orally and in writing on
the collection of 24 h urine samples. To monitor
completeness, each participant was given three tablets
containing 80 mg para-amino-benzoic-acid (PABA),
to be taken during the day of urine collection.19
Protein intake was calculated from the 24 h nitrogen
output according to the formula of Isaksson:20 protein intake (g) ˆ (nitrogen output in 24 h urine
(g=day) ‡ 2g)6.25. The 30 ml urine samples were
analysed for nitrogen by ¯ash combustion technique
(NA 1500 Nitrogen Analyser, Ciba Corning).
Anthropometric data
All anthropometric measurements were made in
accordance with the WHO standards.21 Height (Ht)
International Journal of Obesity
Statistical methods
Multiple linear regression analysis was used to
describe associations between reporting error for
energy and protein and degree of obesity. In the
regression analysis, gender, age, smoking habits, and
education were included as covariates. The F-test was
used to examine differences in the mean and in
regression coef®cients between groups, as described
earlier.25 Reporting bias of energy and protein,
adjusted for covariates, by percentage body fat in
either tertiles or octiles was calculated for men and
women separately, with analysis of variance
(ANOVA). Differences were considered signi®cant
when P < 0.05. The statistical analysis was carried
out with the SPSS.PC version 2.0 program.
Do we eat less fat?
BL Heitmann et al
Results
Participants
Of the 552 subjects invited in 1987 ± 88, 435 subjects
attended both the health examination and gave diet
information. Of these, 323 subjects returned complete
24 h urine samples. This group has been described
earlier.13 The remaining 112 subjects were excluded
from further analysis (Table 1).
Six years later in 1993 ± 94, a second diet survey
was launched. Here, 365 of 435 subjects appeared for
examination. In total, 248 (68%) gave both a 24 h
urine sample and the dietary interview, and about half
(n ˆ 122) returned complete urine samples that contained more than 85% of the administrated p-aminobenzoic acid. The other 126 subjects, with incomplete
24 h urine samples, were excluded from further analyses (Table 1).
Of the 323 subjects examined in 1987 ± 88, 228
appeared only at the baseline examination, and were
examined only at this time. Trends were, therefore,
analysed using the 228 and the 122 subjects examined
either in 1987 ± 88 or in 1993 ± 94 (Table 1).
Non-participants
We showed earlier that recovery of p-amino-benzoic
acid in 1987 ± 88 did not differ between age groups or
between men and women (all P > 0.63).13 Furthermore, within each sex, there were no differences in
age, body weight, percentage body fat, body mass
index or prevalence of obesity (body mass index
> 30 kg=m2) between those with complete, and
those with incomplete data (all P > 0.30).13 This was
also the case when comparing the sub-group of 228
subjects participating in 1987 ± 88, only, with the
remaining subjects (all P > 0.59). Nor were any differences seen at the examination carried out in 1993 ±
94, in body weight, percentage body fat, body mass
index or prevalence of obesity among: (a) the 118
subjects who were interviewed about their diet intake
but did not return the urine sample; (b) the 126
subjects who were interviewed about their diet
intake but gave an incomplete urine sample; and (c)
the 122 subjects who were both interviewed and gave
a complete urine sample (all P > 0.16). However, men
who did not return complete urine samples tended to
be 4 y younger, on average, than other men (52 vs 56 y
of age, P ˆ 0.006).
437
Characteristics of the subjects
Table 2 gives characteristics (mean and s.d.) of the
228 subjects examined in 1987 ± 88 and the 122
subjects examined in 1993 ± 94. Analysis of variance
showed that once age differences had been considered, all differences between measures taken at the
two surveys were independent of gender. Compared to
1987 ± 88 men and women examined in 1993 ± 94
were taller, and among men more had > 12 y of
schooling. There were more active and fewer smoking
men in 1993 ± 94 than in 1987 ± 88.
Trends in fat intake
Absolute fat intake was not different at the two
surveys (both P > 0.69), but as a percentage of total
energy, fat intake was lower at the examination in
Table 1 Number of participants with complete data on diet
and 24 h urine collection at the examinations in 1987 ± 88 and
1993 ± 94
Number of subjects
Examination
1987=88
Examination
1993=94
435
386
323
228
365
248
122
122
95
95
Diet history interview
24 h urine samples
Complete 24 h urine samples
Examined exclusively in 1987=88
or 1993=94
Participated in both examinations
Table 2 Age, anthropometric characteristics (mean and s.d.), number of smokers and subjects participating in physical
activity among subjects with complete urine and diet data in 1987 ± 88 (n ˆ 228) and 1993 ± 94 (n ˆ 122)
Examination1987=88
Age (y)
Weight (kg)
Height (m)
Body mass index (kg=m2)
Body fat (kg)
Fat-free mass (kg)
No. (%) of smokers
No. (%) of subjects physically active at worka
No. (%) of subjects taking part in sportb
No. (%) of subjects with > 12 y schooling
Men
(n=101)
Women
(n=127)
49.7 11.1
78.5 11.3
1.76 0.07
25.4 3.2
19.0 6.9
59.5 5.2
53/101 (52)
23/101 (23)
19/101 (19)
13/101 (13)
50.2 11.9
65.0 11.0
1.63 0.05
24.6 4.0
21.4 8.2
43.7 4.3
50/127 (39)
13/127 (10)
18/127 (14)
9/127 (7)
Examination1993=94
Men
(n ˆ 65)
55.2 10.5
80.4 11.8
1.77 0.07**
25.9 3.1
20.0 6.3
60.5 6.2
20/65 (31)*
6/65 (9)*
18/65 (28)
12/65 (18)**
Women
(n=57)
54.9 10.0
68.1 13.3
1.64 0.07**
25.3 5.1
23.8 8.2
44.2 4.6
24/57 (42)
4/57 (7)
9/57 (16)
4/57 (7)
*P < 0.05, **P < 0.0001. (P per trend)
Heavy or strenuous work.
b
Engaged actively in sports at least 3 h a week.
a
International Journal of Obesity
Do we eat less fat?
BL Heitmann et al
438
Table 3 Trends in total energy and protein intakes (mean and s.d.) calculated from diet and urine samples among
subjects with complete urine and diet data in 1987 ± 88 (n ˆ 228) and 1993 ± 94 (n ˆ 122)
Baseline examination (1987 ^ 88)
Protein intake based on diet (g)
Energy intake based on diet (MJ)
Protein intake based on urine sample (g)
24 h energy expenditure (MJ)
Fatdiet (g)
Percentage fatdiet
Follow-up examination (1993 ^ 94)
Men
(n=101)
Women
(n=127)
Men
(n=65)
Women
(n=57)
82.3 21.6
10.1 2.6
95.3 26.8
13.4 2.2
107.8 30.4
40.5 5.9
63.8 16.9
7.3 1.8
75.8 15.7
9.3 1.2
78.0 22.9
40.4 6.4
86.0 19.9*
10.5 2.6
90.6 20.3**
13.5 2.2
107.5 29.5
38.9 5.9**
66.7 11.7*
7.6 1.6
69.1 16.6**
9.4 1.2
76.8 24.5
38.2 7.0**
*P < 0.05. **P < 0.01. (P per trend)
samples, can be shown to have descreased signi®cantly (Table 3). Total energy intake was underreported more than energy from protein at both
surveys, suggesting that energy from the other
macro-nutrients must have been underreported too.13
This difference was greater in 1993 ± 94 (29%) than in
1987 ± 88 (15%) (P < 0.0001).
1993 ± 94 (both P < 0.01), Table 3. These differences
were independent of gender.
Total energy and protein intakes calculated from diet
and urine samples
In 1987 ± 88 total energy and protein intakes, calculated from the diet interview, were lower than those
intakes determined from either the 24 h urinary nitrogen output or the 24 h energy expenditure in both men
and women (all P < 0.001). In 1993 ± 94 only total
energy, calculated from the diet and the 24 h energy
expenditure differed (both P < 0.001), but protein
intake values were similar (both P > 0.10).
Trends in diet reporting and obesity
Reported energy and protein intakes were negatively
associated, and estimated 24 h energy expenditure and
intake of protein (calculated from urinary nitrogen
output) were positively associated with obesity in both
1987 ± 88 and 1993 ± 94 (Table 4). At both surveys,
reported protein relative to protein calculated from the
urine samples (reporting of protein) and reported
energy relative to estimated 24 h energy expenditure
(reporting of energy), were negatively associated with
obesity. Relative to total energy, intake of protein was
over-reported by obese subjects at both surveys, but
among the lean, total energy was over-reported more
so than energy from protein in 1993 ± 94, only. The
associations were similar at the two surveys (all
P > 0.09), Table 4. Results were similar whether
educational level was included as a covariate or not
(data not shown).
Trends in energy and protein intakes
Protein intake estimated from the 24 h urine samples
showed a lower intake among the 122 subjects measured in 1993 ± 94 than the 228 subjects measured in
1987 ± 88 (P < 0.002). This trend was independent of
gender. On the other hand, reported protein intake was
higher in 1993 ± 94 than in 1987 ± 88 (P < 0.05). Both
the reported energy and the estimated 24 h energy
expenditure were similar on the two occasions (both
P > 0.18). Hence, it seems that, on average, reporting
for protein intake has changed towards a slightly
higher comsumption, despite the ®nding that the
protein intake, as estimated from the 24 h urine
Table 4 Associations (regression coef®cients b and s.e.e) between percentage body fat and calculated and reported total energy and
protein intakes in 1987 ± 88 and 1993 ± 94
Number (n)
1987 ^ 88
b s.e.e.; P
1993 ^ 94
b s.e.e.; P
228
122
Absolute
Protein intake based on diet (g)
Protein intake based on urine (g)
Energy intake based on diet (J)
24 h energy expenditure (J)
7 0.05 0.02;
0.03 0.02;
7 0.34 0.16;
1.55 0.18;
Relative
Percentage protein (diet)/protein (urine)
Percentage energy (diet)/energy (24EE)
7 0.05 0.01; 0.0007
7 0.10 0.2; 0.0001
Underreported protein relativea to under-reported energyb
Adjustment was made for sex, age, smoking and educational level.
Percentage intake (diet)/protein intake (urine).
Percentage energy intake (diet)/energy (24 h energy expenditure).
a
b
International Journal of Obesity
0.03
0.13
0.04
0.0001
3.81 1.35; 0.005
7 0.05 0.03;
0.03 0.03;
7 0.76 0.22;
1.20 0.23;
P (difference)
0.03
0.14
0.0005
0.0001
0.60
0.62
0.09
0.12
7 0.05 0.02; 0.004
7 0.13 0.02; 0.0001
0.86
0.37
2.56 1.60; 0.11
0.54
Do we eat less fat?
BL Heitmann et al
Figure 1 shows the direct relationship between
percentage body fat (in octiles) and reporting of
either protein or energy at the two surveys. Associations were similar for the two gender (both P > 0.17),
and in Figure 1 results are given for men and women
grouped together. In 1987 ± 88 the more obese subjects under-reported total energy more than energy
from protein, suggesting that energy from other
macro-nutrients must have been under-reported too,
whereas the non-obese gave less biased information.
In 1993 ± 94 both obese and non-obese gave biased
information. In 1987 ± 88 an 18% greater underreporting was seen for energy than for protein
among those belonging to the fattest tertile, compared
to only 8% in the leanest tertile (P ˆ 0.06). In 1993 ±
94 this difference was 39% vs 22%, P ˆ 0.07.
There was a signi®cant age difference of 5 y, on
average, between the two population samples examined in 1987 ± 88 or 1993 ± 94. Hence, despite the fact
that all analyses were adjusted for age-differences, the
trends reported here may not be considered true
secular trends. We, therefore, repeated the analyses
by comparing data restricted to subjects aged from 45
to 65 y (mean age 55.1 8.0, n ˆ 165), participating in
1987 ± 88, with the 122 subjects aged from 41 to 71 y
(mean age 55.0 10.0) participating in 1993 ± 94. All
results were essentially similar both when diet-reporting information on the 122 subjects, examined in
1993 ± 94, was compared with information from the
subgroups of 165 subjects, and the total group of 228
subjects examined in 1987 ± 88. Figure 2 shows the
association between obesity and the degree of underreported protein (percentage protein intake (diet)=protein (urine)) relative to under-reported energy (percentage energy intake (diet)=24 h energy expenditure)
in the 165 and 228 subjects examined in 1987 ± 88,
and the 122 subjects examined in 1993 ± 94. All
results were essentially similar when the longitudinal
Figure 1 Relationship between percentage body fat (in octiles)
and reporting of either protein or energy at the two surveys in
1987 ± 88 and 1993 ± 94.
439
Figure 2 Relationship between percentage body fat (in octiles)
and the degree of under-reported protein (percentage protein
intake(diet)=protein(urine)) relative to under-reported energy
(percentage energy intake(diet)=24 h energy expenditure) in the
165 subjects examined in 1987 ± 88, and in the 323 and 122
subjects examined in either 1987 ± 88 or 1993 ± 94.
changes were examined in the 95 subjects who had
been followed up between 1987 ± 88 and 1993 ± 94
(data not shown).
Discussion
The present study shows a substantial under-reporting
bias for both energy and protein at the two surveys
conducted in 1987 ± 88 and 1993 ± 94. The study also
shows that the bias was greater in 1993 ± 94 than in
1987 ± 88. It has been discussed previously13 that the
greater under-reporting of total energy compared to
energy from protein implies that energy from other
macro-nutrients, e.g. fat and=or carbohydrates, must
have been under-reported too. This implies that snacktype foods may have been differentially omitted,
although we do not not know from the present study
if this was a consequence of lack of memory or a
deliberate process.
In 1987 ± 88 under-reported total energy was 15%
larger, on average, than underreported energy from
protein.13 In 1993 ± 94 this difference was 29%. This
difference in reporting bias between the two surveys
may offer one explanation for the apparent downward
trend in reported fat intake, seen in the present and in
other studies.
The fact that the under-reporting bias was greater in
1993 ± 94 than in 1987 ± 88 may be a direct consequence of an increased awareness of diet intake
obtained through intensi®ed public health campaigns
or of®cial recommendations to reduce the intake of fat
and=or simple carbohydrates and to increase the
intake of ®bre and complex carbohydrates. Denmark
launched several campaigns in the late 1980s and in
International Journal of Obesity
Do we eat less fat?
BL Heitmann et al
440
the early 1990s, and, especially since the beginning of
the 1990s, campaign activities have been intensi®ed.14
The fact that food supply statistics do not show a
decrease in fat consumption may also provide some
support for the present ®ndings.27 In addition, food
supply statistic also suggest a much higher total
energy intake per capita than is justi®ed by individual
data, or by total energy needs. Therefore, there may be
quite a large margin for how individuals compose
their diet from the surplus of energy available as
suggested by the food supply statistics.
Whereas in 1987 ± 88 total energy was underreported compared to energy from protein, particularly among the obese subjects, even non-obese subjects displayed substantial under-reporting in 1993 ±
94. Previous studies have reported that the obese
generally give biased information on both total food
intake11,12 and macro-nutrients.13 In this regard, it is
generally assumed that the non-obese have no obvious
social inclination to under-report their diet intakes.
However, the substantial under-reporting even among
non-obese subjects in the present study may suggest
that intensi®ed public health campaigns Ð not necessarily directed towards obesity prevention or treatment, but possibly towards other conditions such as,
for instance, cardiovascular disease and cancer Ð may
have stimulated these subjects to selectively underreport food rich in fat and=or simple carbohydrates.
Indeed, a recent study showed that from 1982 to 1992
knowledge about the prevention of disease, including
consumption of low-fat diets, improved among
Danish adults.26 In addition, the public health campaigns launched in Denmark between 1987 ± 88 and
1993 ± 94 were not particularly directed towards prevention of obesity, but focused on prevention of
cardiovascular disease and=or cancer.
A generally higher socio-economic class may offer
one explanation for the apparent higher reporting bias
in 1993 ± 94 than in 1987 ± 88. However, the similar
results in analyses with and without adjustment for
educational level does not support this. The essentially
similar reporting biases seen for the subset of 95
individuals, who were followed up during the study
period, also does not support differences in socioeconomic status explaining the higher reporting bias
in 1993 ± 94 than in 1987 ± 88.
The present ®ndings of a trend in the error in diet
reporting has implications for interpreting results from
dietary surveys suggesting that, for instance, fat intake
is declining. In this regard a number of reports have
suggested that apparent trends in dietary fat are related
to recent trends in, for instance, cholesterol levels or
coronary heart disease incidence.5,6 However, the
present ®ndings may raise questions concerning such
an ecological reasoning.
Possible limitations of the study
The focus of fat intake in the present study is less
objective than protein and energy intakes, since we
International Journal of Obesity
have no independent estimation of fat intake analogous to urinary protein or 24 h energy expenditure.
Hence, we do not know how much of the underreported energy came from fat and how much from
carbohydrate.
Another possible limitation to the study is that some
of the more generalized decreased reporting of the
non-protein fraction may have been in¯uenced by the
increased availability of new low-fat products on the
market. At the same time, everyone, not only the
obese, may now be consuming these products.
Although, in the present study, the dietician was
instructed to take care of noting when low-fat alternatives were chosen, we cannot exclude the possibility
that she missed some of these products. However, a
closer examination of the actual food choices behind
the nutrient intakes revealed that the reported consumption of, for instance, low-fat milk, low-fat cheese
or low-fat margarine in the present study was not
different in 1987 ± 88 and 1993 ± 94 (all P < 0.11).
Low-fat yoghurt intake was greater in 1993 ± 94 than
in 1987 ± 88 (P < 0.001), but consumption was not
more prevalent in the obese compared to the nonobese (P ˆ 0.48).
On average, subjects were more obese in 1993 ± 94
than in 1987 ± 88. Hence, the increasing similarities
between the obese and non-obese in qualitive misreporting over time may be explained by an increasing
prevalence of obesity. However, adjustment for this
secular increase in percentage body fat gave similar
results (Figure 3).
We have pointed out previously that an unbiased
calculation of 24 h energy expenditure is crucial when
estimating percentage energy from protein from the
urinary nitrogen output, and furthermore that bias in
Figure 3 Relationship between percentage body fat (in octiles)
and the degree of under-reported protein (percentage protein
intake(diet)=protein(urine)) relative to under-reported energy
(percentage energy intake(diet)=24 h energy expenditure) after
adjustment for increasing prevalence of obesity in 122 subjects
examined in 1987 ± 88 or 1993 ± 94.
Do we eat less fat?
BL Heitmann et al
the reporting of physical activity level may invalidate
our ®ndings.13 In the present study we cannot exclude
the occurence over-reporting of physical activity.
However, in order to invalidate the present ®ndings,
this tendency to over-report physical activity would
have to have been proportionally greater among nonobese than among obese subjects and, in addition, to
the increase from 1987 ± 88 to 1993 ± 94. Reported
leisure time activity was slightly higher in 1993 ± 94
than in 1987 ± 88 (16% vs 22% reported they were
actively engaged in sports at least 3 h a week).
Also, more non-obese subjects reported taking an
active part in sports in 1993 ± 94 (31%) than in 1987 ±
88 (21%), whereas the proportion of the obese reporting that they were taking actively part in sports
remained at 11 ± 12% at both surveys. We do not,
however, know if this difference in leisure time
activity represents a compensation for a lower activity
related to work and unemployment, since, during the
same period, reported activities related to work went
down (16% vs 8% reported having heavy or strenuous
work). In particular, the proportion of lean subjects
reporting heavy or strenuous work went down (from
19% to 5%). Furthermore, no work was reported by
relatively fewer people (19% (44=228)) in 1987 ± 88
than in 1993 ± 94 (32% (39=122)). In addition, both
the proportion of unemployed lean and unemployed
obese subjects was higher in 1993 ± 94 than in 1987 ±
88 (obese 32% vs 47%; lean 8% vs 23%). Hence, bias
by over-reporting of physical activity probably does
not explain the present ®ndings.
Finally, it cannot be excluded that the present
®ndings may, in part, be population speci®c since
tendencies to under-report seem to vary with study
population.11 ± 13
Conclusion
From the present study, we suggest that fat and=or
carbohydrate-speci®c under-reporting may be responsible for the apparent trend of decrease in fat intake in the
Danish, and possibly other populations, provided that
these populations display similar reporting patterns to
the Danes. The results also suggest that this pattern of a
higher macro-nutrient speci®c bias in 1993 ± 94 than in
1987 ± 88 may be a direct consequence of increased
awareness of diet intake, related to intensi®ed public
health campaigns and of®cial recommendations directed towards reducing intake of foods high in fat and=or
sugar, and increased intake of fruit and vegetables to
improve not only overweight but also cardiovascular
®tness and cancer risk. These ®ndings may have implications for our understanding of the health bene®ts of the
apparent decline in dietary fat, for instance in relation to
incidence of cardiovascular disease.
Acknowledgements
The project was supported by grants from the Danish
Agricultural and Veterinary Research Council, the
Danish Medical Research Council, the Danish Research
Agency (FREJA), the Danish Health Insurance Foundation and Wedell-Wedellsborgs Foundation.
441
References
1 National Food Agency. Nordic food recommendations, 1993.
2 The Scienti®c Committee for Food. Nutrient and energy
intake for the European Community, Brussels. European
Union: 1992.
3 Osler M, Heitmann BL. Food patterns associated with intakes
of fat, carbohydrate and dietary ®bre in a cohort of Danish
adults followed for six years. Eur J Clin Nutr 1997; 51:
354 ± 361.
4 Mela DJ The intake and acceptance of fat in the diet. Nutr
Food Sci 1994; 3: 19 ± 21.
5 E LoÈfdahl ed. The Journal of Agricultural Economics 1996;
58: 303 ± 305.
6 Stephen AM, Sieber GM. Trends in individual fat consumption in the UK 1900 ± 1985. Br J Nutr 1994; 71: 775 ±
778.
7 Department of Agriculture, Agricultural Research Service.
Data tables: results from USDA's 1995 continuing survey:
CSFI=DHKS 1995. Food Surveys Research Group; Riverdale,
Md, 1995 (CD-ROM).
8 Johansson L, Drevon CA, Bjùrneboe GEA. The Norwegian
diet during the last hundred years in relation to coronary heart
disease. Eur J Clin Nutr 1996; 50: 277 ± 283.
9 Fogelholm M, MaÈnnistoÈ S, Vartiainen E, Pietinen P. Determinants of energy balance and overweight in Finland 1982 and
1992. Int J Obes 1996; 20: 1097 ± 1104.
10 Heini A, Weinsier RL. Divergent trends in obesity and fat
intake patterns: the American paradox. Am J Med 1997; 102:
259 ± 264.
11 Heitmann BL. The in¯uence of fatness, weight change, slimming history and other lifestyle variables on diet reporting in
Danish men and women aged 35 ± 65 years. Int J Obes 1993;
17: 329 ± 336.
12 Black AE, Goldberg GR, Jebb SA, Livingstone MBE, Cole TJ,
Prentice AM. Critical evaluation of energy intake data using
fundamental principles of energy physiology: 2. Evaluating
the results of published surveys. Eur J Clin Nutr 1991; 45:
583 ± 589.
13 Heitmann BL, Lissner L. Dietary underreporting by obese
individuals Ð is it speci®c or non-speci®c? B Med J 1995;
311: 986 ± 989.
14 Peetz-Schou M. En beskrivelse af strategi, aktiviteter og
erfaringer fra Levnedsmiddelstyrelsens oplysningsarbejde paÊ
ernñringsomraÊdet (Nutritional education for health promotion
and disease prevention purposes). The National Food Agency,
Copenhagen; 1997.
15 Andersen NL, Fagt S, Groth M, Hartkopp HB, Mùller A,
Ovesen L, Warming DL (BD). Danskernes kostvaner 1995 Ð
hovedresultater. (National dietary survey of Denmark). Publication No. 235. The National Food Agency, Copenhagen;
1996.
16 Heitmann BL. Body fat in the adult Danish population aged
35 ± 65 years: an epidemiological study. Int J Obes 1991; 15:
535 ± 545.
17 Jensen HK, Jùrgensen T. Incidence of gallstones in a
Danish population. Gastroenterology 1991; 100: 790 ±
794.
18 Mùller A (ed). Levnedsmiddeltabeller 1985, 2nd ed (Danish
food composition tables 1985, nutrient composition of Danish
food), Publication no. 75, Ministry of the Environment.
National Food Agency: Sùborg, 1986.
19 Bingham S, Commings JH. The use of 4-aminobenzoic acid as
a marker to validate the completeness of 24 h urine collections.
Clin Sci (Colch) 1983; 64: 629 ± 635.
20 Isaksson B. Urinary nitrogen output as a validity test in dietary
surveys. Am J Clin Nutr 1980; 33: 4 ± 5.
International Journal of Obesity
Do we eat less fat?
BL Heitmann et al
442
21 World Health Organisation. Measuring obesity Ð classi®cation and description of anthropometric data. Report on a
WHO consultation of the epidemiology of obesity, Warsaw
21 ± 23 October 1987, Nutrition Unit Document, EUR=ICP=
NUT 125. WHO; Copenhagen, 1989.
22 Heitmann BL. Prediction of body water and fat in adult Danes
from measurement of electrical impedance. A validation
study. Int J Obes 1990; 14: 789 ± 802.
23 Garby L, Garrow JS, Jùrgensen B, Lammert O, Madsen K,
Sùrensen P et al. Relation between energy expenditure and
body composition in man: speci®c energy expenditure in vivo
of fat and fat-free tissue. Eur J Clin Nutr 1988; 42: 301 ± 305.
International Journal of Obesity
24 World Health Organisation. Energy and protein requirements.
Technical Report Series 724, Report of a joint FAO=
WHO=UNU Expert Consultation, 1985.
25 Heitmann BL. The effect of gender and age on associations
between blood lipid levels and obesity in Danish men
and women aged 35 ± 65 years. J Clin Epidemiol 1992; 45:
693 ± 702.
26 Osler M, Schroll M. Lifestyle and prevention of ischaemic
heart disease in Denmark. Eur J Publ Health 1995; 5:
109 ± 112.
27 Statistisk AÊrbog 1995. (Statistical Yearbook), Statistics Denmark; Copenhagen, 1995.