International Journal of Obesity (1999) 23, 1074±1078
ß 1999 Stockton Press All rights reserved 0307±0565/99 $15.00
http://www.stockton-press.co.uk/ijo
Patterns of long-term weight changes in
overweight developing Danish men and
women aged between 30 and 60 years
BL Heitmann1* and L Garby2
1
Danish Epidemiology Science Centre at the Institute of Preventive Medicine, Copenhagen University Hospital, Kommunehospitalet; and
The Copenhagen County Centre of Preventive Medicine, Department of Medicine C, Glostrup University Hospital; 2Department of
Physiology, University of Odense, Denmark
OBJECTIVE: To determine, in a retrospective semi-longitudinal study, the pattern of weight changes over 11 y in a
population of subjects who developed their overweight as adults.
DESIGN: A randomly selected population of 2436 subjects from the County of Copenhagen with ages 30 ± 60 y was
followed by body weight determinations on three occasions between the years 1982 ± 83 and 1993 ± 94.
SUBJECTS: At the last examination, 660 subjects had attained a body mass index (BMI) of 28 kg=m2. Of these
subjects, 341 had a self-reported BMI at 25 y of less than 25 kg=m2.
RESULTS: In 1994, average yearly weight changes since 1983 ranged between 0.9 and 1.2 kg in the younger (30 ± 40 y)
0.4 and 1.2 kg in the older (50 ± 60 y) men and women during the preceding 11 y. These changes were similar over the
two periods of 5 and 6 y. More than half of the subjects had weight gains between 6 and 21 kg, about equally
distributed between the two periods. Women displayed larger and more frequent gains than men, as did the younger
compared to the older subjects.
CONCLUSIONS: The data show that weight changes to attain overweight are highly variable among subjects.
However, in most subjects, the weight changes were so small and continuous that simple theories of control by
means of ®xed 'set points' cannot explain the results. In fact, the weight changes were generally so small that they are
unlikely to be re¯ected in food consumption studies and in studies of macronutrient metabolism.
Keywords: long-term weight change; longitudinal data; set point theory
Introduction
The database
Knowledge of rates of body weight change in subjects
who develop overweight have implications for the
understanding of several pathophysiological aspects
of overweight, for example, the hypotheses of weight
control that include `set points' and the physiological
and clinical signi®cance of the rates of weight
changes.
Several studies of weight changes over time in
overweight subjects have been published,1,2 as well
as studies describing the development of obesity over
time,3 ± 5 but we have found only a few studies on
long-term weight change rates in the part of the
population that develops overweight as adults,6 and
none that included more than two measurement points
over time. We report here the results of observations
made in the context of the Danish MONICA
project.7,8
Subjects
*Correspondence: Berit L. Heitmann, Ph.D. Copenhagen County
Centre for Preventive Medicine, Glostrup University Hospital,
building 8,7 ¯oor, DK - 2600 Glostrup, Denmark.
E-mail: [email protected]
Received 20 July 1998; revised 19 February 1999; accepted 28
April 1999
The study population included 3608 Danish citizens
aged 30, 40, 50 and 60 y at the baseline examination in
1982 ± 1984. This group was re-examined 5 y later in
1987 ± 88, where 2987 participated, and again 6 y later
in 1993 ± 94, where 2656 subjects participated. The
population sample identi®ed in 1982 included 4807
subjects, and was an age- and sex-strati®ed random
sample of the Danish population, selected from the
Central Person Register, among citizens who all lived
in the Western part of the Copenhagen County.9
Subjects not born in Denmark were excluded
(n ˆ 226). The remaining 4581 subjects were found
to be reasonably representative of the total Danish
population with respect to sex, age, educational level,
occupation and housing, but people employed in
agriculture, horticulture, ®shery and self-employed
and unskilled workers were slightly under-represented. Non-participation has been described elsewhere8,9 but, in brief, a telephone interview among
non-responders revealed that among those who were
invited but did not attend, school education and
social class were lower, and a higher proportion
were smokers.10
Weight gain in Danish adults
BL Heitmann and L Garby
A total of 2436 subjects contributed with complete
data on weight and height in 1982 ± 84 and at two
follow-up studies, 5 and 11 y later in 1987 ± 88 and in
1993 ± 94. These subjects were also asked to give their
body weight at the age of 25 y.
Anthropometry
Height (Ht) was measured to the nearest 0.5 cm with
subjects standing without shoes, heels together and
head in the horizontal Frankfurter plane. Body weight
(BW) was measured to the nearest 0.1 kg, using a
SECA scale (Vogel & Halke, Hamburg, Germany),
and subjects wearing only hospital underwear.11
Statistical methods
Analyses of variance and multiple linear regression
analyses were performed to evaluate mean differences
between the two genders, and to adjust the weight
changes for age. Differences between mean or median
values were tested using analysis of variance or
Wilcoxon's two sample test, respectively.
The project was approved by the Ethical Committee
for the Copenhagen County and is in accordance with
the Helsinki II Declaration.
Results
Drop-outs: Compared to the 2436 men and women,
who attended all three examinations, the 1170 who
attended only at baseline were slightly older (both
P<0.001). However, there was no difference in
response rate for the men and women, and after
adjustment for age-differences in response, baseline
BMI was similar for responders and drop-outs.
Out of the total of 2436 subjects, approximately
15% (n ˆ 374) had attained a BMI 30 kg=m2 at the
third examination, that is the second follow-up in
1994. Of those who had attained a BMI 30 kg=m2,
165 (approximately 6% of the total group) had a BMI
<25 at the age of 25 y. De®ning overweight
as 28kg=m2, 660 subjects (365 men and 295
women) had attained a BMI of 28 kg=m2 or more at
the ®nal examination in 1994. Among these subjects,
341 (162 men and 179 women) had a BMI <25 at the
age of 25 y. The average yearly weight changes in
men and women aged either 30 ± 40 y or 50 ± 60 y, are
shown in Table 1. They ranged from 0.7 to 1.2 kg per
year in the younger men and women, and from 0.4 to
0.6 in the older men and women, with similar values
in the ®rst ®ve and the following six years. All values
were associated with large standard deviations, up to
about twice those of the mean values.
Figure 1 shows the distribution of weight changes
among those 341 men and women with an attained
BMI 28 kg=m2, and a BMI<25 kg=m2 at age 25 y.
For subjects of the same age, weight changes over 1 to
11 y period were generally larger for women than for
men, and younger individuals of both sexes were more
likely to gain than lose. The average weight changes
during the years from age 25 and the ®rst study was
10.9 9.6 kg (10.5 9.2 kg for men and 11.4
10.1 kg for women) in the group of 660 subjects
with an attained BMI 28, and 13.4 8.2 kg
(13.9 9.0 kg for men and 12.9 8.3 kg for women)
in the group of 341 subjects who in addition had a
BMI<25 kg=m2 at age 25 y.
The distribution of weight changes, and their relation to the average yearly weight changes, is shown in
Table 2 and 3. Very few subjects had weight changes
of more than 2 kg=y over the 11 year period (6 men
and 12 women). In both men and women, mean yearly
weight changes were greatest for those who were
initially leanest, whereas the variation in yearly
weight change was greatest in those with a higher
initial BMI, Figure 2.
Patterns of weight change were essentially similar
for the 165 men and women who reported a BMI <25
at age 25, but who had attained a BMI 30 kg=m2
(data not shown).
Discussion
In the present population based sample of adult Danes,
aged 41, 51, 61 and 71 y in 1993 ± 94, approximately
1=4 had a BMI 28 kg=m2, and about half of these
had developed their overweight as adults. It appears
that a considerable amount of the weight gain took
Table 1 Mean and s.d. of weight changes, kg per year, between 1983 and 1988 and between 1988 and 1994 in 365
men and 295 women with an attained BMI of 28 kg=m2
Men
30 ± 40y
50 ± 60y
Women
30 ± 40y
50 ± 60 y
All Subjects
Subjects with a BMI< 25 at age 25 y
1983 ! 1988 ! 1994
1983 ! 1988 ! 1994
n
mean s.d.
mean s.d.
n
mean s.d.
mean s.d.
175
190
0.70 1.17
0.40 1.01
0.84 0.89
0.40 0.83
76
86
0.95 0.85
0.39 0.88
0.96 0.78
0.43 0.72
126
169
0.98 1.38
0.58 0.98
1.10 1.17
0.53 0.89
72
107
1.25 1.10
0.59 0.88
1.15 0.91
0.61 0.80
1075
Weight gain in Danish adults
BL Heitmann and L Garby
1076
Figure 1 Cumulative distribution curves for 11 y weight changes between 1983 and 1994 in 76 men aged 30 ± 40 y, 86 men aged 50 ±
60 y, 72 women aged 30 ± 40 y and 107 women aged 50 ± 60 y, who all had an attained BMI 28 kg=m2 at follow up in 1993, and a BMI
< 25 at age 25 y.
Table 2 Patterns of weight change in 660 men and women with
BMI 28 kg=m2 in 1994
Body weight gain
over11y, kg
Mean weight change, kg=y
n (%)
Initial BMI 1983 ! 1988 1988 ! 1994
Table 3 Patterns of weight change in 341 men and women with
BMI 28 kg=m2 in 1994, and who reported a BMI of less than
25 kg=m2 at 25 y of age
Body weight gain
over11y, kg
Mean weight change, kg=y
n (%)
Men
Less than 0
0±3
3±6
6±9
9 ± 12
12 ± 15
15 ± 18
18 ± 21
More than 21
59
49
71
73
50
21
19
15
8
(16)
(13)
(19)
(20)
(14)
(6)
(5)
(4)
(2)
31.6
29.3
29.0
28.0
27.2
26.8
27.1
28.0
26.3
7 0.45
0.20
0.26
0.61
0.81
1.47
1.30
2.41
2.24
7 0.23
0.12
0.51
0.73
1.06
0.99
1.60
1.25
2.15
Women
Less than 0
0±3
3±6
6±9
9 ± 12
12 ± 15
15 ± 18
18 ± 21
More than 21
Men
Less than 0
0±3
3±6
6±9
9 ± 12
12 ± 15
15 ± 18
18 ± 21
More than 21
18
19
30
35
27
12
12
7
2
27
32
53
51
50
30
22
15
15
(9)
(11)
(18)
(17)
(17)
(10)
(7)
(5)
(5)
32.5
29.9
29.1
28.1
26.6
27.1
26.7
24.5
26.3
7 0.23
7 0.31
0.33
0.80
0.99
1.56
1.42
1.54
1.94
7 0.61
0.47
0.46
0.59
0.93
0.93
1.55
1.87
2.56
Women
Less than 0
0±3
3±6
6±9
9 ± 12
12 ± 15
15 ± 18
18 ± 21
More than 21
10
13
33
31
41
17
16
10
8
place in the years after age 25 y, and most of this in
the younger subjects. This observation is in agreement
with the study on adult Finns,2 where the greatest
variability in weight change was found for the age
group 15 ± 29 y. Furthermore, the data show that
among the 660 subjects, who had attained a
BMI 28 kg=m2, the weight changes during the
6 y ranged from about 0.4 ± 1.0 kg per year, with
Initial BMI
1983?1988
1988?1994
(11)
(12)
(19)
(22)
(17)
(7)
(7)
(4)
(1)
30.4
29.1
28.5
27.5
26.2
25.7
26.1
25.8
29.0
7 0.26
0.09
0.40
0.67
0.81
1.26
1.38
2.25
2.19
7 0.31
0.22
0.39
0.65
1.10
1.16
1.54
1.43
2.39
(6)
(7)
(18)
(17)
(23)
(9)
(9)
(6)
(4)
31.8
29.2
28.2
27.2
26.2
25.8
25.4
24.7
25.6
7 0.02
7 0.30
0.20
0.82
1.00
1.75
1.42
1.65
1.88
7 0.62
0.45
0.57
0.57
0.91
0.80
1.52
1.77
2.42
similar changes during the preceding 5 y. The 341
subjects, who, in addition, were lean at the age of 25 y,
had mean yearly weight changes of the same magnitude. All mean values were statistically different from
zero, and were characterized by very large standard
deviations. This observation has been reported previously.4 ± 6 In addition, we and others found that the
Weight gain in Danish adults
BL Heitmann and L Garby
1077
Figure 2 Mean yearly weight changes and variance by deciles of baseline BMI, in 365 men and 295 women aged 30 ± 60 y who all had
an attained BMI 28 kg=m2.
variation in subsequent weight change increased particularly in the obese.6. This observation suggests a
less accurately working control mechanism in the
obese than in the lean subjects. Among the 660 and
the 341 subjects, one fourth and one sixth, respectively, had experienced no or only small weight
changes during 11 y, and had apparently reached
their individual overweight already at the start of the
study in 1983. On the other hand, more than half of
the subjects gained between 6 and 21 kg during the
11 y period of observation, and their weight gains, of
between 1.0 and 1.5 kg per year, were about the same
in the two periods of 5 and 6 y. This latter result does
not support the idea that body weight is primarily
dominated by a control system with a ®xed `set point`,
where any weight gain must be assumed to take place
at a much higher rate until the `®nal' weight is
reached. Rather, the data are consistent with control
by means of a `set point' with a rather wide `error
band' or `slack', in which case external signals play a
major role.
The mean daily weight increase for men and
women was between 1 and 3 grams during the 11 y
of study. These weight gains correspond to an average
daily `overeating' of about the same weight of fat, or
doubly that amount of carbohydrate. Such quantities
cannot be measured in food consumption studies and
it is dif®cult to imagine that underlying primary
disturbances in the metabolism in these subjects
could be measurable. However, the relatively large
between-subject variation (Table 1) shows that a
considerable number of subjects have much larger
(and smaller) weight changes.
One possible limitation of the present study
includes a selective non-response to participation
and drop out to follow-up. Although this was not
apparent for those attending the baseline examination
only, compared to those attending all three examinations, it has been shown earlier that subjects who
never turn up are more obese than participants.12 This
phenomenon leads to an underestimation of both level
and variation of BMI in population studies. Hence, the
rates of weight gain reported in the present study may
be slightly conservative, but this does not affect the
overall results.
Conclusion
The present study showed that the yearly weight gains
among subjects who developed their overweight as
adults are small and continuous, and hence the present
results exclude the possibility of weight control by
means of a ®xed `set point'.
Acknowledgement
This study was supported by grants from the Danish
National Foundation, the Danish Health Insurance
Foundation and the Danish Heart Association.
Weight gain in Danish adults
BL Heitmann and L Garby
1078
References
1 Williamson PS, Levy BT. Long-term body weight ¯uctuation
in an overweight population. Int J Obes 1988; 12: 579 ± 583.
2 Rissanen A, HelioÈvaara M, Aromaa A. Overweight and
anthropometric changes in adulthood: A prospective study of
17000 Finns. Int J Obes 1988; 12: 391 ± 401.
3 Jeffery RW, Folsom AR, Luepker RV, Jacobs DR, Gillum RF,
Taylor HL, Blackburn H. Prevalence of overweight and weight
loss behavior in the metropolitan adult population: The Minnesota Heart Survey Experience. Am J Public Health 1984; 74:
349 ± 352.
4 Garn SM, Pilkington JJ, Lavelle M. Relationship between
initial fatness level and long-term fatness change. Ecol Food
Nutr 1984; 14 85 ± 92.
5 Rookus MA, Burema J, Van' T Hof M, Deurenberg P, Hautvast
JGAJ. The development of the body mass index in young adults,
II: Interrelationships of level, change and ¯uctuation, a four-year
longitudinal study. Hum Biol 1987; 59: 617 ± 630.
6 Sonne-Holm S, Sùrensen TIA, Jensen G, Schnohr P. Longterm changes of body weight in adult obese and non-obese
men. Int J Obes 1990; 14: 319 ± 326.
7 Kirchoff M., Schroll M, Kirkby H, Hansen BS, Sanders S, Sjùl
A, Jùrgensen T, Hansen PF. Screening I. Danmonica. CVD
Epidemiol Newsletter 1983; 34: 32.
8 Hagerup L, Eriksen M, Schroll M, Hollnagel H, Agner E,
Larsen S. The Glostrup Population studies. Collection of
Epidemiological Tables. Scand J Soc Med 1981; 20 (Suppl
1): 5 ± 112.
9 Jensen KM, Jùrgensen T. Incidence of gallstones in a Danish
population. Gastroenterol 1991; 100: 790 ± 794.
10 Jùrgensen, T. Gallstones. An epidemiological investigation.
Dan Med Bull 1990; 37: 336 ± 346.
11 WHO: Measuring obesity Ð classi®cation and description of
anthropometric data. Report on a WHO Consultation on the
Epidemiology of Obesity, Warshaw 21 ± 23 October 1987.
Unit Document, EUR=ICP=NUT 125, WHO Copenhagen
1989.
12 Sonne-Holm S, Sùrensen TIA, Jensen G, Schnohr P. In¯uence
of fatness, intelligence, education and sociodemographic factors on response rate in the health survey. J Epidemiol Comm
Health 1989; 43: 369 ± 374