International Journal of Obesity (2000) 24, 14±19
ß 2000 Macmillan Publishers Ltd All rights reserved 0307±0565/00 $15.00
www.nature.com/ijo
Lifetime dietary change and its relation to
increase in weight in Spanish women
P Montero1*, C Bernis1, C Varea1 and S Arias1
1
Unidad de AntropologõÂa, Departamento de BiologõÂa, Facultad de Ciencias, Universidad AutoÂnoma de Madrid, Spain
INTRODUCTION: Changes in dietary patterns and a decrease in physical activity have occurred in Western countries.
These are factors in the variation in body composition observed in populations, characterized by a progressive
accumulation of fat with age and a consequent increase in the risk of suffering from common chronic illnesses such as
obesity, cardiovascular disease and cancer.
OBJECTIVE: To investigate weight gain throughout the life-cycle and its relation to modi®cations in dietary patterns,
analyzing the causes of these modi®cations and their implications for patterns of adult overweight and obesity.
DESIGN: Cross-sectional sample of Spanish women from a socio-economically disadvantaged class.
SUBJECTS: 1037 healthy perimenopausal women (age: 45 ± 65 y).
MEASUREMENTS: Juvenile body mass index (BMI), current BMI, food frequency questionnaire, retrospective food
habits.
RESULTS AND CONCLUSIONS: Of these women, 48.8% had changed their dietary habits during their lifetime. A
change in diet due to migration or marriage occurred at approximately 20 years of age and was characterized by an
increased frequency of consumption of foods rich in protein and complex carbohydrates, while a change due to illness
occurred at around 50 years of age and was characterized by a decrease in the consumption of these types of food.
The change in dietary behavior due to migration was associated with weight gain. Weight gain was also inversely
associated with BMI during youth; women who in their youth had a BMI < 18.5 kg=m2 gained an average of 21.4 kg,
compared with those with a BMI > 27 kg=m2 in their youth, who gained an average of 5.4 kg.
International Journal of Obesity (2000) 24, 14±19
Keywords: food habits; juvenile BMI; current BMI; overweight; obesity; perimenopausal low-income women
Introduction
In the last 50 y there have been great changes in the
patterns of morbidity and mortality in industrialized
societies. Of the environmental factors that have
in¯uenced this change, nutritional factors are of particular importance.1 ± 3
Numerous epidemiological studies have demonstrated the relation between diet and widely prevalent
chronic illnesses in industrial societies, such as obesity, cardiovascular disease and cancer.4 ± 8 Diet is also
fundamental for proper infant growth and development, with all their consequences in adulthood and old
age.9,10 In addition, there has been a decrease in
physical activity in Western countries due to changes
related to the processes of urbanization and industrialization. These additional factors need to be taken into
account when considering the variation of body composition in such populations. This variation is characterized by a progressive increase in accumulated fat
with age and a parallel increase in the risk of suffering
the aforementioned diseases.6,11,12 This change has
been observed in migrants from the country to the
town. Such movement substantially modi®es their
*Correspondence: P Montero, Unidad de AntropologõÂa,
Departamento de BiologõÂa, Facultad de Ciencias, Universidad
AutoÂnoma de Madrid, Madrid 28049, Spain.
E-mail: [email protected]
Received 20 July 1998; revised 10 May 1999; accepted 22 June
1999
physical activity13,14 and dietary habits. Additionally,
some authors maintain that there is a relation between
prolonged infant caloric de®ciency and diseases such
as adult obesity due to the ®xation of metabolic
pathways that are ef®cient in dietarily deprived conditions. Subsequent nutrient-rich conditions disrupt
the established equilibrium. This could be one of the
causes of the problems of overweight and metabolism
seen in many rapidly urbanized populations.15,16
Cross-sectional questionnaires of a population can
provide an instant picture of the nutritional circumstances of a community. This may aid the understanding of the main nutritional problems and the
identi®cation of groups at risk of developing diseases
related to such problems.2,17 ± 19 The current study also
gathers retrospective information. This is very useful
for broadly identifying some aspects of the nutritional
and environmental conditions in which these women
developed. Anthropometric variables that can be used
as indicators of obesity, overweight and underweight
have the advantage of being easily and inexpensively
measurable. They can also be employed to detect
nutritional problems related to excesses or de®ciencies in the diet, and other associated diseases.15,19 ± 22
We set out to study weight gain throughout the lifecycle and its relation to modi®cations in dietary
patterns. We analyze the causes of these modi®cations
and their implications for patterns of adult overweight
and obsesity in a sample of Spanish women from a
socio-economically disadvantaged class.
Lifetime dietary change and increase in weight
P Montero et al
Materials and methods
The study was carried out within the framework of an
agreement between the Universidad AutoÂnoma de
Madrid (UAM), the Hospital Cantoblanco de Madrid
and the Ayuntamiento de Alcobendas, Madrid Province.
The sample examined here comprised all of the
1037 volunteer women between the ages of 45 and 65,
resident in Alcobendas, who took part in a women's
preventive health campaign, organized by the ConcejalõÂa de Salud del Ayuntamiento of Alcobendas, and
a study concerning reproductive ageing, carried out by
the Anthropology Unit of the UAM between 1995 and
1997.
Gynaecological examinations were conducted in
the Hospital Cantoblanco by the health workers
assigned to the project. In the same place, the team
from the Anthropology Unit (UAM) completed:
1. A food frequency questionnaire (FFQ) that considered a total of 13 foodstuffs.
2. A questionnaire regarding past and present dietary habits and lifestyles. Information was sought
concerning changes in food consumption
(yes=no), reasons for changes in food consumption, the age at which they took place and the
characteristics of any changes in dietary behaviour (for example, whether a foodstuff was eaten
as frequently or more or less frequently than
before the change).
3. A series of anthropometric measurements, of
which we concern ourselves here with weight
and height at the time of the study and the
weight at 18 ± 20 y. These were used to estimate
the current and juvenile body mass index (BMI).
Weight at 18 ± 20 y was taken to be that recalled
by the subjects. Women's responses were clear
and probably reliable as, for most of them, that
was the age when they married and the ®rst time
they had been weighed.
Weight gain since the age of 18 ± 20 y was also
calculated as a dichotomous variable, `weight
change', whose ®rst category corresponded to
women who lost weight or maintained the same
weight, and the second category corresponded to
those who gained weight.
In this article BMI is used as an indicator of fat
accumulation on the basis of the close relation with fat
accumulation in adult women that has been described
by several authors.23,24 The index has been used as a
continuous variable in some analyses and as an
ordered categorical variable in others. In the latter
case we have followed Garrow's24 classi®cation:
obesity, BMI > 30 kg=m2; overweight, BMI ˆ 25.0 ±
30.0 kg=m2; normal weight, BMI ˆ 20.0 ± 24.9 kg=m2.
As an indicator of the nutritional conditions in
women's youth the pre-reproductive BMI was calculated from their weight at age 18 ± 20 y and their
height at the time of the study. For some analyses,
these values were placed in ®ve ordered categories:
BMI < 18.5 kg=m2; BMI ˆ 18.5 ± 19.9 kg=m2; normal
weight; and combined overweight and obese (classes
are de®ned above). Overweight and obesity were
combined due to their infrequency in the youth of
these women. BMI was also calculated using armspan
instead of height to avoid the problem of the effect of
ageing on height.
Age was controlled for in all analyses, as it affects
dietary behavior and body measurements. It also
avoids the problem of taking into account the secular
change in height that has occurred in Western populations.25,26
Analyses were carried out using the SPSS 7.5
statistics package. First of all the relationship between
reasons for changing diet and dietary factors was
examined graphically in multidimensional space by
multiple correspondence analysis, using the SPSS
procedure HOMALS (homogeneity analysis by
means of alternating least squares; see the Appendix
for a detailed explanation). ANOVA and logistic
analyses were also carried out to examine these
relationships in greater detail.
15
Results
Frequency of consumption of foodstuffs
The frequencies of consumption of the different foodstuffs are shown in Table 1.
Changes in diet and reasons for the change
Responding to the question `Have you changed your
eating habits at any time in your life?', 48.8%
(n ˆ 504) of women replied af®rmatively (Table 2).
In response to the question `What was the reason for
the change in your way of eating?', these women most
frequently sited migration from their town of birth to
Madrid, marriage, illness, menopause, weight gain,
children's birth and children's habits. These are classi®ed here into three groups: (1) migration from their
Table 1
Food frequency consumption
Foodstuff
Milk=yogurt
Custard=creÁme caramel
Cheese
Eggs
Meat
Fish
Sausages
Vegetables
Salad
Fruit
Legumes
Bread
Sugar
n
Mean number of times=d
565
648
429
872
481
489
299
622
784
531
622
461
401
2.50
0.09
0.42
0.24
0.55
0.53
0.24
0.55
0.82
2.32
0.31
66.g.
1.68
International Journal of Obesity
Lifetime dietary change and increase in weight
P Montero et al
16
Table 2
Change in dietary habits since age 18 ± 20 by age at interview and reasons for change
Change in diet
%
49y
50 ± 54y
55 ± 59y
60y
Total
Mean age of change by reasons (years old)
44.1
52.2
53.8
47.6
48.8
n
No change in diet
%
173
55.9
156
47.8
106
46.2
696
52.4
504
51.2
w2 ˆ 6.842 d.f. ˆ 3 P ˆ 0.08
town of birth to Madrid, or marriage; (2) illness
(hypercholesterolemia, hypertension, hyperglycemia,
etc.); (3) menopause, weight gain and other reasons
(which have been combined for the purpose of analysis as they occurred at very similar ages and were
related to increases in weight). Table 2 also illustrates
the percentage of women who changed their diet for
the reasons cited and the mean age at which it took
place. Changes occurred signi®cantly earlier when
they were due to migration from the country to the
city, or marriage, than when they were the result of
illness or other reasons.
The manner in which the women's diet changed
depended greatly on their reason for changing, giving
rise to different behavior. A multiple correspondence
analysis was carried out to gain an overall view of the
relationships between the following variables: (1)
dietary behavioral variables related to the manner of
dietary modi®cation, the reason for the change, and its
nature (whether women were consuming the following foodstuffs at a greater, lesser or equal frequency at
the time of the study than before their diet changed:
bread, potatoes, pasta, legumes, fruit and vegetables,
meat, ®sh, eggs, milk, milk products, and cakes and
buns); (2) juvenile and current BMI; (3) weight gain
since youth. The results of this analysis are shown in
Figure 1. As can be seen in the graph, the change in
diet due to migration or marriage is associated with an
increase in the consumption of foods rich in protein
and simple carbohydrates and a decline in that of
n
219
143
91
76
52
Migration=marriage
(%)
Disease
(%)
Other
(%)
48.1
17.1
34.8
38.0
19.3
42.7
38.8
29.6
31.6
42.2
29.7
28.1
42.1
22.1
35.7
w2 ˆ 12.22 d.f. ˆ 6 P ˆ 0.06
23.1
50.1
37.7
F ˆ 40.61 P ˆ 0.000
foods rich in complex carbohydrates. Changes in diet
due to illness, menopause and weight gain are associated with decreased consumption of foods rich in
proteins, complex carbohydrates and fat and also with
low juvenile BMI and modest weight gain.
Prevalence of overweight and obesity
Overweight and obesity, measured by BMI, were very
widespread (57.6% of subjects; n ˆ 596) in the
sample. The prevalence was much greater than that
found in the Comunidad de Madrid16 and in the rest of
Spain (15.2% of women with BMI 30). However,
this ®gure was only 6.5% when these women were
young, when 27.8% (n ˆ 269) of the sample had had
BMIs of less than 20 kg=m2, and 9.2% (n ˆ 89) had
had BMIs of less than 18.5 kg=m2. Sixty-®ve percent
of women had worked on the land when they were
young, and 59.6% of the women whose BMI values
had been less than 18.5 had been agricultural workers.
Such low BMI values combined with strong physical
activity can be considered to be indicative of mile
chronic energy de®ciency.27,28
The mean current and youth BMI values are shown
in Table 3. Separate BMI values of women when they
were young were calculated using height and armspan
to investigate the effect of diminishing height with
age. A t-test showed signi®cant differences between
the two values, for which reason BMI in young
women was subsequently calculated using the armspan. This is closely correlated with height at maturity
and is a reliable alternative to height in the calculation
of BMI.29 ± 32 A high proportion of women were
currently overweight or obese in all age groups,
although the frequency of women of normal weight
was greatest in the under-50 age group. Overweight
increased slightly and obesity increased sharply after
50 years of age, the proportion of women older than
60 who were obese being twice that of those younger
than 50 (44.4% and 22.3%, respectively). The incidence of youthful overweight and obesity was very
low for all age groups.
Lifetime weight gain and its relation with juvenile BMI
and changes in dietary habits
Figure 1 Patterns of dietary changes.
International Journal of Obesity
More than 93% of the women in the sample
had gained weight since their youth. The highest
Lifetime dietary change and increase in weight
P Montero et al
17
Table 3 Juvenile BMI based on height (1) and armspan (2) and current BMI
BMIyouth(1) ˆ weight=height2
Age
49
50 ± 54
55 ± 59
60
Total
BMIyouth(2) ˆ weight=height2
n
x
s.d.
n
x
s.d.
n
x
s.d.
377
279
180
132
968
21.87
21.87
21.87
22.57
21.96
F ˆ 2.13
P ˆ 0.09
3.00
2.80
2.97
3.17
2.97
348
257
164
127
896
21.35
21.33
21.03
21.70
21.34
F ˆ 1.03
P ˆ 0.37
3.17
3.47
3.02
3.19
3.23
394
300
197
144
1035
27.37
28.18
28.56
29.67
28.15
F ˆ 12.36
P ˆ 0.00
4.16
3.95
3.84
4.11
4.10
ANOVA
percentages of women who had gained weight were
found in categories of low juvenile BMIs. This was
especially pronounced in the BMI < 20 group, in
which almost all women had gained weight. This
effect was statistically signi®cant in all age groups.
The increase in weight was inversely related to
juvenile BMI, being very much greater in the group
of women with the lowest values of juvenile BMI (a
mean of approximaely 18 kg in those with juvenile
BMI < 20, compared with 6.6 kg in those with a
juvenile BMI 25. Weight gain increased with age
in women with BMIjuvenile < 25, while there was no
such clear association for the case of women who
were overweight or obese when young. Differences in
weight gain were signi®cant among the different
BMIjuvenile groups at all ages.
The mean weight increase was slightly, but not
signi®cantly, greater among women who had changed
their dietary habits than those who had not, either
when considering the entire sample (mean weight in-
Table 4 Relation between weight gain and reason for change
in eating, juvenile BMI and age (ANOVA)
Weight
gain
Covariable
Age
Sum of
squares d.f.
F
P
Main effects
353.64 1 3.78 0.052
Reason for change 931.00 1 9.96 0.002
Juvenile BMI
2951.91 2 15.79 0.000
Interactions Reason for change* 327.78 1 3.51 0.062
Juvenile BMI
Model
4538.09 5 9.71 0.000
Table 5
BMI ˆ weight=height2
crease(had changed dietary habits) ˆ 15.29 kg, s.d. ˆ 10.5 kg;
mean weight increase(had not changed dietary habits) ˆ
13.87 kg, s.d. ˆ 9.6 kg; F ˆ 2.41, P ˆ 0.121), or when
age groups were considered separately.
An analysis of variance (ANOVA) was carried out
in which `weight gain' was taken as the dependent
variable, `juvenile BMI' and `reason for changing
eating habits' were independent variables, and `chronological age' was a covariable (Table 4). This clearly
demonstrates that women with lower juvenile BMI
put on the most weight, and that those who changed
their dietary behavior due to migration from the
country to the city or due to marriage experienced
greater weight gain than those whose habits did not
change.
Since the causes of weight gain are multifactorial,
the in¯uence of all the variables considered in this
study was evaluated by logistic regression on the
dichotomous factor of `weight gain'. The variables
included were: categorized values of juvenile BMI
(three levels: 19.9, 20.0 ± 24.9, 25.0); occurrence
of dietary change (two levels; yes, no); reason for
change (three levels: migration and marriage, illness,
other); and the continuous variables of age of change
and age at the time of the study. Reason for change
and juvenile BMI were included as categorical variables using the absence of change and overweight,
respectively, as reference groups. The analysis reveals
that juvenile BMI was the only variable associated
with weight gain: women with lower juvenile BMIs
were considerably more likely to gain weight than
those with higher values (Table 5).
Multifactorial causes of weight gain: results of logistic regression. Dependent variable: weight gain
Variables included in the model
Change of food habits
Reasons for change
Migration=marriage
Disease
Age
Age at change juvenile BMI (weight18years=armspan2)
BMI 19.9 y
BMI ˆ 20 ± 24.9 y
BMI 25 y
Constant
B
s.e.
d.f.
sig
exp(B)OR
95%Cif or exp(B)
7 0.2483
1.4501
0.0455 ± 13.3809
0.6052
0.5977
0.0440
2.7525
1.6764
1.0194
0.8405 ± 9.0138
0.5196 ± 5.4091
0.9351 ± 1.1112
3.8699
2.5271
7 0.2166
7 0.8612
1.4541
1.3115
1.2947
2.7010
0.864
0.222
0.094
0.387
0.663
0.000
0.008
0.054
0.867
0.749
0.7801
1.0125
0.5167
0.0192
1
2
1
1
1
3
1
1
1
1
47.9356
12.5167
0.8052
2.7730 ± 828.6286
0.9575 ± 163.6172
0.0637 ± 10.1841
International Journal of Obesity
Lifetime dietary change and increase in weight
P Montero et al
18
Discussion
The study deals with a sample of women on low
incomes whose pattern of consumption of foodstuffs
continues to be that typical of Mediterranean populations,1,16 despite having experienced changes in dietary patterns during their lives. These changes were
due to two main causes: an improvement in economic
conditions early on in their reproductive life, and the
appearance of nutrition-related illnesses later on in
life.
Changes following migration and marriage generally signify an improvement in living standards, at
least from a nutritional point of view, since most were
originally rural women who came to Madrid during a
period of industrialization when either they or their
husbands could work. Other behaviors, such as that
occurring around menopause when the tendency to
put on weight induces changes in dietary patterns (for
example less fat is eaten), are consequences rather
than causes of weight gain. Mean juvenile BMI values
of women who changed their diet due to migration or
marriage were not different from those of women who
did not change (BMI ˆ 21.13 weight18years=armspan2
and BMI ˆ 21.43 weight18years=armspan2, respectively). However, mean weight gain was greater in
those women whose dietary behavior changed due to
migration or marriage, although signi®cant differences are only apparent in women under 50, since
from this age there is the additional effect of age on
weight gain.
These women's BMI values at the time of the study
were very high due to the considerable weight
increases (a mean of 14 kg) experienced since their
youth. This lifetime weight gain was due not only to
nutritional factors, but also to others related to patterns of reproduction and lactation. The observed
phenomena may be explained in terms of the effect
of environmental factors operating during their developmental and later years. The women originally came
predominantly from rural populations (70.4%), where
they worked on the land from an early age (65%
agricultural workers), and lived through a time of
dietary de®ciencies as a consequence of the Spanish
Civil War (1936 ± 1939) and its aftermath.2 Around
the age of 20 they migrated in search of better living
conditions and changed the quantity and quality of
their diet by incorporating foodstuffs rich in protein
and simple carbohydrates that they had not previously
consumed with such frequency.33 As a result of
ceasing to work on the land, their physical activity
was greatly reduced.
The change in dietary behavior due to migration or
marriage clearly in¯uenced the large increases in
weight seen in this sample. However, it was also
in¯uenced by juvenile BMI. Women whose mean
juvenile BMI values were lower experienced a greater
mean weight gain. Although the change in eating
habits and the reduction of physical activity may be
International Journal of Obesity
suf®cient to explain the great weight increase
observed in these women throughout their lives, it is
still noteworthy that this increase is particularly pronounced in those women who were very thin in their
youth. These results are consistent with the hypothesis
of Ulijaszek34,35 concerning the ®xation of ef®cient
metabolic pathways during the primary stages of
development under conditions of prolonged caloric
scarcity. The subsequent disruption of these metabolic
processes under conditions of abundant energy, and
the decline in physical activity, would account for the
considerable weight gain experienced by women in
this sample, especially those who had very low
juvenile BMI values.
Acknowledgements
We would like to thank the constant institutional
support received from the Cantoblanco Hospital, and
the disinterested and ef®cient blood withdrawal, gynecological and laboratory services. We also especially
thank Paloma Arribas. We are also grateful to the
ConsejerõÂa de Salud del Ayuntamiento de Alcobendas
for their support. This study was supported by the
projects FIS94=0372 and FIS97=0487.
References
1 Graciani A, Rodriguez F, Banegas JR, HenaÂndez R and del
Rey Calero J. El consumo de alimentos en EspanÄa en el
perõÂodo 1940 ± 1988. Un estimacioÂn a partir de hojas de
balance alimentario. Documentos de trabajo. UAM Ed:
Madrid, 1996.
2 Moreiras O, Carbajal A and Perea IM. EvolucioÂn de los
haÂbitos alimentarios en EspanÄa. Madrid: Ministerio de Sandidad y Consumo, 1990.
3 Willet W. Nutritional epidemiology. Monographs in epidemiology and biostatistics, Vol 15. Oxford University Press:
Oxford, 1990.
4 Richardson S, Gerber M. Nutritional factors and breast cancer
in a French Mediterranean region. In Giacosa A, Hill MJ (eds).
The Mediterranean diet: Proceedings of a workshop organized
by European Cancer Prevention (ECP) and The Italian
League Against Cancer, Cosenza, Italy, pp 81 ± 103, 1991.
5 Gerber M. Olive oil and cancer. In Giacosa A, Hill MJ (eds).
The Mediterranean diet: Proceedings of a workshop organized
by European Cancer Prevention (ECP) and The Italian
League Against Cancer, Cosenza, Italy, pp 127 ± 139, 1991.
6 Williams CM, Dickerson JWT. Nutrition and cancer. Some
biochemical mechanisms. Nutr Res Rev 1990; 3: 75 ± 100.
7 Manson JE, Colditz GA, Stampfer MJ, Willet JC, Rosner B,
Monson RR, Speizer FE, Hennekens, CH. A prospective study
of obesity and risk of coronary heart disease in women. New
Engl J Med 1990; 29: 882 ± 889.
8 Hercberg S, Dupon H, Papoz L, GalaÂn P. NuricioÂn y Salud
PuÂblica. Abordaje epidemioloÂgico y polõÂticas de prevencioÂn.
Aula MeÂdica: Ed Cea, 1988.
9 Tanner JM. Growth as a mirror of conditions in society. In:
Lingren GW (ed). Growth as a mirror of conditions in society.
Stockholm Institute of Education Press: Stockholm, 1990, pp
3 ± 34.
10 Johnston FE. Physical growth and development and nutritional
status: epidemiological considerations. Fed Proc 1981; 40:
2583 ± 2587.
11 Sowers MF, Crutch®eld M, Jannausch ML, Russell-Aulet M.
Longitudinal changes in body composition in women approaching the midlife. Ann Hum Biol 1996; 23(3): 253 ± 265.
Lifetime dietary change and increase in weight
P Montero et al
12 Garrow JS. Obesity and related diseases. Churchill Livingston: London, 1988.
13 Singh LP, Harrison GA. Impact of migration, environment and
socioeconomic conditions on the physique of Sikhs. J Biosoc
Sci 1996; 28: 101 ± 116.
14 Singh LP, Kaur B. Impact of migration, environment and
socioeconomic conditions on blood pressure of Sikhs. J Biosoc
Sci 1998; 30(2): 155 ± 164.
15 Barker DJP. Prenatal in¯uences on disease in later life. In
Shetty PS, McPherson K (eds). Diet, nutrition and chronic
diseases. Lessons from contrasting worlds. Wiley: New York,
1977, pp 41 ± 53.
16 Ulijaszek SJ, Strickland SS. Nutritional studies in biological
anthropology. In Lsker GW, Mascie-Taylor CGN (eds).
Research strategies in human biology. Field and survey
studies. Cambridge University Press: Cambridge, 1993; pp
108 ± 139.
17 Aranceta J, PeÂrez C, Amela C, GarcõÂa Herrera, R. Encuesta de
nutricioÂn de la Comunidad de Madrid. Documentos teÂcnicos
de Salud PuÂblica, No. 18. DireccioÂn general de prevencioÂn y
promocioÂn de la salud. ConsejerõÂa de Salud. Comunidad de
Madrid, 1994.
18 Barret-connor E. Nutrition epidemiology: how do we know
what they ate? Am J Clin Nutr 1991; 54: 182S ± 187S.
19 Quandt SA. Methods for determining dietary intake. In:
Johnston FE (ed). Nutritional anthropology. Alan R Liss:
New York, 1987; pp 67 ± 84.
20 Lasker GW. The place of anthropometry in human biology. In:
Ulijaszek SJ, Mascie-Taylor CGN (eds). Anthropometry: the
individual and the population. Cambridge University Press:
Cambridge, 1994; pp 1 ± 6.
21 Davies PSW. Anthropometry and body composition. In: Ulijaszek SJ, Mascie-Taylor CGN (eds). Anthropometry: the
individual and the population. Cambridge University Press:
Cambridge, 1994; pp 130 ± 140.
22 Mascie-Taylor CGN. Nutritional status: its measurement and
relation to health. In: Mascie-Taylor CGN, Lasker GW (eds).
Application of biological anthropology to human affairs. cambridge University Press: Cambridge, 1991; pp 55 ± 83.
23 Wang MC, Bachrach LK. Validity of the body mass index as
an indicator of adiposity in an ethnically diverse population of
youths. Am J Hum Biol 1996; 8: 641 ± 651.
24 Garrow JS, Webster J. Quetelet's index (W=H2) as a measure
of fatness. Int J Obes 1985; 9: 147 ± 153.
25 Boldsen JL. The place of plasticity in the study of the secular
trend for male stature: an analysis of Danish biological
population history. In: Mascie-Taylor CGN, Bogin B (eds).
Human variability and plasticity. Cambridge University Press:
Cambridge, 1995; pp 75 ± 90.
26 Mascie-Taylor CGN, Boldsen JL. Regionla and social analysis
of height variation in a contemporary British sample. Ann Hum
Biol 1985; 14: 59 ± 68.
27 WHO Expert Committee. Physical status: the use and interpretation of anthropometry. WHO Technical Report Series,
1995, p 854.
28 James WPT, Ferro-Luzzi A, Waterlow JC. De®nition of
chronic energy de®ciency in adults. Eur J Clin Nutr 1988;
42: 969 ± 981.
29 Chilima DM, Ismail SJ. Anthropometric characteristics of
older people in rural Malawi. Eur J Clin Nutr 1998; 52(9):
643 ± 649.
30 Garshick E, Ashb J, Tun CG, Lieberman SL, Brown R.
Assessment of stature in spinal cord injury. J Spinal Cord
Med 1997; 20(1): 36 ± 42.
31 Kwok T, Whitelaw MN. The use of armspan in nutritional
assessment of the elderly. J Am Geriat Soc 1991; 39(5):
492 ± 496.
32 Allen SC. The relationship between height, armspan and
forced expiratory volume in elderly women. Age Ageing
1989; 18(2): 113 ± 116.
33 Montero P, Bernis C, Varea C, Arias S. HaÂbitos alimentarios
en mujeres espanÄolas: frecuencia de consumo de alimentos
y valoracioÂn del cambio en el comportamiento alimentario.
AtencioÂn Prim 1999; 23(3): 127 ± 131.
34 Ulijaszek SJ. Plasticity, growth and energy balance. In:
Mascie-Taylor CGN, Bogin B (eds). Human variability and
plasticity. Cambridge University Press: Cambridge, 1995, pp
91 ± 109.
35 Ulijaszek SJ. Energetics, adaptation, and adaptability. Am J
Hum Biol 1996; 8: 169 ± 182.
36 SPSS Categories. SPSS Inc.: Chicago, 1990.
19
Appendix
HOMALS derives optimal parameter estimates based
on the maximum separation of categories. Thus,
objects in the same category occur close to each
other while those in different categories are separated
as far as possible. HOMALS can compute a solution
for several dimensions. The maximum number of
dimensions equals the number of categories minus
the number of variables. The eigenvalues measure
how much of the categorical information is accounted
for by each dimension. The maximum eigenvalue for
each dimension is 1. Two dimensions together provide
an interpretation in terms of distances.36
International Journal of Obesity