International Journal of Obesity (1997) 21, 400±406
ß 1997 Stockton Press All rights reserved 0307±0565/97 $12.00
Investigation of the relationship between infant
temperament and later body composition
JCK Wells1, M Stanley1, AS Laidlaw1, JME Day1, M Stafford2 and PSW Davies3
1
Infant and Child Nutrition Group, Dunn Nutrition Unit, Downham's Lane, Milton Road, Cambridge CB4 1XJ; 2 Childhood Nutrition
Research Centre, Institute of Child Health, 30 Guilford Street, London WC1N 1EH; and 3 Queensland University of Technology, School of
Human Movement Studies, Kelvin Grove Campus, Victoria Park Road, Locked Bag No 2, Red Hill, Queensland 4059, Australia
OBJECTIVE: To investigate the ability of maternally-rated infant temperament to predict fatness and activity patterns
in early childhood.
DESIGN: Longitudinal investigation of infants studied at 12 weeks and followed up at 2±3.5 y of age.
SUBJECTS: Thirty healthy full-term infants from the general population.
MEASUREMENTS: Body composition, behavioural activity and temperament at 12 weeks; anthropometry, body
composition, diet and behavioural activity at follow-up.
RESULTS: Infant temperament predicted later behaviour and fatness. Easily soothable infants had leaner childhood
skinfold thicknesses (P < 0.02) and were more active in childhood (P < 0.025). Infant distress was also related to
childhood diet composition.
CONCLUSIONS: Infant temperament can predict later body composition and behaviour. Both energy intake and
energy expenditure may be mechanisms by which the relationship develops.
Keywords: irritability; behaviour; television
Introduction
Body composition in young children is a function of
the longitudinal balance between earlier energy intake
and energy expenditure, mediated by somatic growth.
In childhood and adolescence, when the energy cost of
growth is relatively low, physical activity is the most
variable component of energy expenditure. Thus in
children aged 3±5 y, both energy intake and energy
expended on physical activity have been linked with
subsequent changes in body composition1 and physical activity was found to show an inverse crosssectional relationship with fatness.2 Even in early
infancy, energy expended on physical activity shows
great variability between individuals,3 but no relationship between energy expenditure and fatness has been
identi®ed in infants4 and neither infant physical activity nor infant energy expenditure have been found to
predict subsequent fatness in the general population.5±8
This may be due to changes in individual activity
patterns that occur when mobility develops.
Nevertheless, infant physical activity is characterised by a number of related dimensions, including
motor activity, circadian rhythms and temperament.
The few studies seeking to relate infant activity to
Correspondence: Dr JCK Wells.
Received 18 April 1996; revised 27 September 1996;
accepted 28 January 1997
body composition have tended to concentrate on the
motor aspect.9,10 A notable exception is a study which
related infant weight gain to temperament,11 reporting
that infants gaining the most weight were perceived
by their mothers as more dif®cult. It was suggested
that dif®cult infants might be fed more often to
quieten them.11
The role of early temperament in the development
of both energy balance variables and parental behaviour merits more detailed investigation. In a previous
cross-sectional study, we found only one aspect of
infant temperament, irritability, to be signi®cantly
related to physical activity energy expenditure at 12
weeks.12 Activity energy expenditure, calculated as
the difference between total energy expenditure and
the costs of growth and minimal metabolism,3 was
increased in more irritable infants.12 However, in
longitudinal terms temperament might have considerable potential to in¯uence energy balance, since one
of the de®ning features of temperament is its stability
over time.13 Thus, although aspects of temperament
may be poorly related to physical activity at an initial
time point, they might be predictive either of subsequent behaviour patterns, or of subsequent parental
care patterns, both of which could in¯uence energy
balance.
In this study we examined the hypothesis that early
temperament predicts later behaviour patterns and
body composition in the general population. Irritability was studied in infants at 12 weeks of age using a
Infant temperament and later body composition
JCK Wells et al
parental activity diary and the Rothbart Infant Behaviour Questionnaire. The subjects were followed up in
early childhood and behaviour and body composition
again assessed.
Subjects and methods
A cohort of 50 infants was recruited from the Rosie
Maternity Hospital, Cambridge, UK, into a comprehensive study on energy metabolism. No reward was
offered to parents for participation at any stage of the
study. All infants were healthy, born at term and had
no known condition that might unduly in¯uence
health or development. Ethical permission for the
study was granted by Cambridge Health Authority
and the Medical Research Council Dunn Nutrition
Unit.
Measurements at 12 weeks
At or close to 12 weeks of age, various measurements
were made over an 8 d study period. Weight, supine
length and skinfold thicknesses at the biceps, triceps,
quadriceps and subscapular sites were measured on
the ®rst day of the study period. Weight was measured
again on the last day of the study period.
Total body water (TBW) was estimated by dilution
using the stable isotope 18O, given as water (H218O).
Subjects were dosed orally with 0.28 g H218O per kg
body weight, either as water if the infant was breastfed, or added to a small quantity of formula-milk if the
infant was formula-fed. The infants were allowed to
suck on a nasogastric tube attached to a syringe
containing the dose. A single urine sample was
collected immediately prior to the dose to determine
natural 18O enrichment of body ¯uids. Urine samples
were also collected for the following 7 d, allowing the
washout kinetics of the isotope to be determined.
Urine samples were collected using the method of
placing cotton wool in the infant's diaper, and checking the diaper frequently for urination.14 The time of
the sample was taken as the mid point between the
checks either side of urination.
Isotopic enrichment of the dose and the pre- and
post-dose urine samples was measured by isotope
ratio mass spectrometry (Aqua Sira model, VG
Isogas, Cheshire, UK). TBW was calculated as
described in detail previously, using the back extrapolation method.15 Values for TBW were combined
with age-speci®c and sex-speci®c values for the water
content of lean tissue16 to give fat free mass. Fat free
mass was subtracted from body weight to give fat
mass, and percentage fat was then calculated from fat
mass and weight. Both skinfold thicknesses and percentage fat were used in subsequent analyses, as they
represent different aspects of total body fatness.
Infant irritability was measured by two methods as
described below. Mothers were told that the resulting
data were intended to explain normal variation in total
energy expenditure, which was being measured at the
same time.12
Mothers were asked to complete a diary of infant
activity based on a previously published version.17
The chart distinguished ®ve categories of behaviour,
sleeping, awake and content, fussy, crying, and feeding, and the mothers were then asked to record the
behaviour categories as they occurred against a timescale, with a resolution of 15 min. The infant was
termed fussy if `your baby is unsettled, irritable,
restless or fractious and may be vocalising, but not
continuously crying', while crying referred to `periods
of prolonged distressed vocalisation'. Mothers were
asked to record the principal activity that characterised each 15 min period. The chart has been validated for the crying component only, with a high level
of agreement being demonstrated between parental
reports and audio recordings made over the same
period.17 The chart was completed for two periods
of 24 h during the study week. Mothers were asked to
avoid days unrepresentative of normal behaviour
patterns, such as those involving long car journeys
or vaccinations. Only the categories fussy and crying
have been used in the analysis described below.
Mothers were also asked to complete the Rothbart
Infant Behaviour Questionnaire, 1978 version18 at the
end of the study week. This questionnaire was a parent
report instrument designed to assess a number of
dimensions of infant temperament which has high
internal reliability18 and has been widely used over
the last decade. Our version contained 91 items
concerning the frequency of certain behaviours
observed over the previous week. The responses
were coded on a scale from 1±7: 1 ˆ never, 2 ˆ very
rarely, 3 ˆ less than half the time, 4 ˆ about half the
time, 5 ˆ more than half the time, 6 ˆ almost always,
7 ˆ always, x ˆ does not apply. An average level of
response for each category of behaviour was then
obtained by dividing in each case the sum of
responses by the number of questions answered.
The dimensions relevant to the present study are
de®ned as follows18: distress to limitations, a child's
fussing, crying, or showing distress while (a) waiting
for food, (b) refusing food, (c) being in a con®ning
place or position, (d) being dressed or undressed, (e)
being prevented access to an object toward which the
child is directing her/his attention; soothability, a
child's reduction of fussing, crying or distress when
soothing techniques are used by the caretaker or child.
Measurements at 2±3.5 y
Where possible, the infants took part in a follow-up
study at 2±3.5 y of age. Weight, height and skinfold
thicknesses at the triceps and subscapular sites were
recorded. Triceps and subscapular skinfolds were
combined to give sum of skinfolds, used in subsequent
analyses. Total body water (TBW) was calculated as
described above, except that the tracer was deuterium
rather than 18-oxygen and the plateau method of
401
Infant temperament and later body composition
JCK Wells et al
402
calculation was used.15 Deuterium dilution space
overestimates the body water pool by approximately
4%,19 so the dilution spaces were divided by 1.044 to
give values for TBW.20 These values were then
combined with values for body weight and the water
content of lean tissue for the relevant age group and
sex16 to give fat free mass, fat mass and percentage
fat.
Energy intake and diet composition were assessed
using a diet history by interview21 as has been used in
the same age group previously.22 The history combined a detailed record of the previous 48 h with
frequency measures for a 7 d period. Food quantity
was assessed against photographic categories of portion sizes. Data were coded by the Dido programme23
using standard tables for food composition.23,24 Diet
composition was estimated by food quotient, which
refers to the calculated amount of carbon dioxide
produced, divided by oxygen consumed, in the oxidation of all metabolisable fuels in the diet.25 A high
food quotient corresponds to a diet high in carbohydrate, while a low food quotient corresponds to a diet
high in fat.
Behaviour was again assessed using a parental diary
over two periods of 24 h. In this age group, the diary
distinguished the following categories of behaviour:
sleeping, awake and quiet; awake and active; watching TV; upset; feeding. Awake and active referred to
when the child was moving around, or remaining in
one place but in an excited state, while awake and
quiet referred to the child showing minimal limb
movement and not being in an excited state. Watching
TV was only recorded if the child was paying attention
to the television.
Statistics
The relationship between infant irritability and later
variables was investigated using simple correlation
analyses. Multiple regression analyses were also used
where possible to adjust outcome variables for baseline values and potential confounding factors. All
calculations were made using Minitab (Minitab
1985) software.
Results
Characteristics of the infants at the time of original
recruitment are given in Table 1. All infants were of
white ethnic status. All tables refer to the 30 infants
who participated in both stages of the study, except
where data points were lost. Parental body mass index
(BMI) ranged from 17.8±30.2 kg/m2 (mothers) and
18.2±34.2 kg/m2 (fathers), and 1 mother and 4 fathers
were obese using the BMI cut-off point of 30.0 kg/m2.
Anthropometric variables at 12 weeks of age are given
in Table 2, along with results of measurements of
activity and temperament. Successful measurements
of behaviour were achieved in all 30 infants.
Table 1 Characteristics of the 30 infants at recruitment
Mean
s.d.
Birthweight (kg)
Gestational age (weeks)
Maternal age (y)
Maternal BMI (kg/m2)
Paternal BMI (kg/m2)
3.59
40.3
31.8
23.6
24.7
0.29
1.1
4.1
3.6
4.0
Male/female
Breast-/Formula-fed
Social class I
II
IIIa
IV
Number
14/16
12/18
9
8
11
2
a
Manual and non-manual grouped together.
Table 2 Age, anthropometry and temperament at 12 weeks
(n ˆ 30)
Age (weeks)
Weight (kg)
Length (cm)
Triceps skinfold (mm)
Subscapular skinfold (mm)
Total body water (ml)
Percentage fat
Mean
s.d.
12.1
6.07
60.8
7.8
7.4
3510
25.3
0.5
0.63
1.7
1.1
1.3
320
7.2
Diary
Fussy (h/d)
Crying (h/d)
1.1
0.5
0.6
0.5
Temperament Questionnaire
Distress to limitations
Soothability
3.10
4.77
0.87
1.15
Of the 50 infants originally measured at 12 weeks,
20 could not be located or did not wish to participate
at the time of the follow up. The remaining 30
children were seen at between 2 and 3.5 y of age.
Children followed-up were signi®cantly shorter in
infancy than those not followed up (60.8 vs 61.7 cm;
t ˆ 2.02; P < 0.05) but there were no signi®cant differences in infant weight, percentage fat, skinfold
thicknesses, temperament, behaviour or parental BMI.
Results of anthropometry and deuterium dosing are
given in Table 3. Total body water was successfully
Table 3 Age, anthropometry, body composition and diet in
childhood
Age (y)
Weight (kg)
Height (cm)
BMI (kg/m2)
Triceps skinfold (mm)
Subscapular skinfold (mm)
Total body water (ml)
Percentage fat
Energy intake (kcal/d)
Carbohydrate intake (g/d)
Fat intake (g/d)
Food quotient
Mean
s.d.
2.64
14.0
92.2
16.4
8.3
5.9
9190
15.4
1380
188
54
0.863
0.31
1.4
3.5
1.2
1.8
1.5
1010
5.2
220
40
9
0.014
Diet variables: n ˆ 29; Skinfolds and percentage fat: n ˆ 25.
Infant temperament and later body composition
JCK Wells et al
403
Table 4 Correlations between infant irritability and childhood diet and fatness
Food quotient
Energy intake
Fat intake
Carbohydrate intake
Sum skinfolds
Percentage fat
0.14
0.58**
0.24
0.09
0.16
0.39
0.07
70.04
0.12
70.03
70.12
70.11
0.22
0.55**
0.15
0.03
70.08
70.16
70.05
70.66***
70.03
70.11
0.40*
70.15
Crying
Fussy
Distress to limitations
Soothability
Crying and fussy from activity diary; distress to limitations and soothability from Rothbart questionnaire. Diet variables: n ˆ 29. Skinfolds and
percentage fat: n ˆ 25.
* P < 0.05.
** P < 0.005.
*** P < 0.001.
Table 5 Correlations between infant temperament and childhood behaviour (n ˆ 28)
Crying
Fussy
Distress to limitations
Soothability
Sleep
Awake and quiet
Awake and active
TV
Feeding
Upset
70.02
0.08
0.26
70.16
70.13
70.34
70.24
70.47*
0.24
0.24
0.06
0.59**
70.32
70.17
70.46*
70.18
70.12
0.22
0.01
0.17
0.28
0.14
0.64**
0.22
Crying and fussy from activity diary; distress to limitations and soothability from Rothbart questionnaire.
* P < 0.02.
** P < 0.001.
estimated in 25 children. Percentage fat ranged from
6±24%, and none was classi®ed as obese by body
mass index.26 Activity patterns and temperament were
measured in 28 subjects. Food intake was assessed in
29 subjects. Sum of skinfolds, measured in 25 infants,
correlated poorly with percentage fat (r ˆ 0.21;
P ˆ 0.35) but better with fat mass (r ˆ 0.44;
P < 0.05). Percentage fat was related to age
(r ˆ 0.39; P ˆ 0.053), but sum of skinfolds was not
(r ˆ 0.12; P ˆ 0.58). Results of both activity diaries
have been described in detail previously.8
Preliminary analysis showed that parental BMI was
not related to any aspect of infant or childhood
behaviour. Parental BMI also showed no relationship
with childhood energy intake or food quotient. Therefore parental BMI was not included in the following
analyses. Time spent upset in childhood did not
correlate signi®cantly with childhood energy intake
(r ˆ 70.02) or food quotient (r ˆ 70.02), nor with
fatness (skinfolds: r ˆ 0.01; percentage fat: r ˆ 0.32;
P ˆ 0.13). Time spent awake and active was inversely
related to sum of skinfolds in childhood (r ˆ 70.41;
P < 0.05), but the same relationship for percentage fat
did not achieve signi®cance (r ˆ 70.24; P ˆ 0.28).
Correlations for awake and quiet were positive but
lower (sum of skinfolds: r ˆ 0.38; P ˆ 0.069; percentage fat: r ˆ 0.13; P ˆ 0.56).
Correlation coef®cients for the relationships
between infant irritability and childhood diet and
fatness are given in Table 4, and for the relationships
between infant irritability and childhood behaviour in
Table 5.
Multiple regression analyses were used to investigate further the signi®cant correlations, by adjusting
relationships for possible confounders. Childhood
percentage fatness was adjusted for infant percentage
fatness and childhood age. Childhood skinfolds were
adjusted for infant skinfolds and childhood activity
Table 6 Multiple regression analyses
Dependent variable
n
Independent variable
Childhood percentage fat
24
Infant percentage fat
Distress to limitations
Childhood age
Childhood sum of skinfolds
24
Awake and active in childhood
Regression
coefficient
Standard error
P value
0.390
1.959
6.673
0.114
0.978
2.498
0.003
0.059
0.015
Infant sum of skinfolds
Awake and active in childhood
Soothability
0.154
0.015
71.647
0.256
0.461
0.599
0.554
0.974
0.012
24
Infant sum of skinfolds
Childhood sum of skinfolds
Soothability
70.058
0.004
0.734
0.125
0.108
0.299
0.644
0.974
0.024
Childhood food quotient
28
Childhood weight
Upset in childhood
Fussy in infancy
70.002
70.002
0.014
0.002
0.005
0.004
0.421
0.620
0.003
Childhood carbohydrate intake
28
Childhood weight
Upset in childhood
Fussy in infancy
72.035
710.47
31.32
4.826
12.65
11.30
0.677
0.416
0.011
Childhood TV watching
28
Upset in childhood
Distress to limitations
70.445
70.200
0.285
0.178
0.131
0.273
Infant temperament and later body composition
JCK Wells et al
404
level. Childhood time spent awake and active was
adjusted for infant and childhood skinfolds. The
relationship between infant time spent fussy and childhood diet composition was adjusted for childhood
weight and the cross-sectional relationship between
childhood diet and time spent upset. A similar adjustment was made for the relationship between infant
distress to limitations and childhood TV watching.
These analyses are given in Table 6.
Discussion
Behaviour in early childhood is in¯uenced by many
factors. Activity patterns are a function both of
individual temperament and of parental care patterns.
Furthermore, a cross-sectional association between
activity level and body composition has been
observed in early childhood2 although it is not possible from the latter study to separate cause and effect
variables within this relationship. The development of
behaviour in children is a ®eld of study that has
received comparatively little attention from nutritionists, but which could aid signi®cantly in explaining
the development of fatness, and possibly the aetiology
of obesity, in this age group. Our study suggests that
measures of behavioural stability may be more successful in explaining the development of body composition than measurements of energy expenditure
made at a single time point.
There is more than one conceptualisation of temperament, but the most widely utilised11 is that of
Thomas and Chess,27 describing temperament as
behavioural style which has a constitutional basis.
Among the characteristics of behavioural style are
activity, rhythmicity, approachability and mood, with
parents consistently rating dif®cult children as having
low approachability and rhythmicity, and high activity
and negative mood.28,29 There are two ways in which
temperament could link with body composition.
Firstly, activity levels which show stability over
time30 might have a direct effect on energy balance,
such that low activity levels might predict greater
weight gain and fat deposition. Such a relationship
between initial activity level and subsequent weight
gain has been demonstrated in children1 but not in
infants.5±8 Secondly, persistent behaviour of the infant
which is perceived as dif®cult by the care-taker may
result in certain strategies of care. If these care
strategies were to in¯uence aspects of behaviour that
are known to affect energy balance, temperament
might in¯uence later body composition by this more
indirect route.
Previous studies in the ®eld of paediatric psychology have shown inconsistent results in attempting to
relate infant temperament to subsequent parental care
patterns. Several longitudinal studies have concluded
that high levels of infant irritability are associated
with a decrease in subsequent maternal attention.31±34
In contrast, other studies have reported the reverse,
with raised infant irritability eliciting greater maternal
response.35±37 This discrepancy is probably due to
several factors. Firstly, maternal response to infant
irritability may itself depend on maternal temperament.38 Thus motivated mothers may respond differently to unmotivated mothers. Secondly, the studies
focused on different age groups and time intervals,
and the infant-mother relationship may develop differentially over time.38
Finally, most such studies have assessed infant
temperament after the neonatal period, such that
maternal response may have already begun to shape
infant behaviour.38 In one study, for example, infant
irritability at three months was found to be predicted
more successfully by prenatal maternal attitude than
by infant irritability in the neonatal period.36
Thus the mechanism by which infant temperament
and maternal response are related remains poorly
understood. However, for the purposes of our study,
it is suf®cient to note that infant behaviour has
frequently been found to predict later maternal
response, whatever the nature of that response and
whatever the aetiology of infant temperament. Our
study was carried out in a population characterised by
unusually high levels of maternal education. It can be
assumed that they were also highly motivated
mothers, give that they have participated in a long
term research project without receiving any form of
®nancial reward. Our data are best interpreted within
this context.
Our study was intended to investigate the effect of
infant irritability on childhood behaviour and body
composition. In particular, it was intended to test the
hypothesis that infant irritability gives rise to behavioural patterns determined either by the infant, or by
inference by the care-taker in response to infant
temperament, which might in¯uence energy balance.
Our study utilised two ways of assessing infant
irritability, namely a diary record of time spent upset
(fussy or crying) and a questionnaire measure of
distress to limitations. It should be emphasised that
although these measures are expected to show agreement, they have different conceptual bases and measure different variables. The diary measured general
distress in units of time, using the categories fussy and
crying as described in the methods section. The
questionnaire measured distress arising from being
restrained or con®ned, in units of frequency which
were averaged over different types of constraint.
Signi®cantly, infant distress to limitations predicted
childhood time spent upset, while infant time spent
fussy and crying did not. These factors may explain
the difference in the way the two measures correlated
with childhood variables.
The amount of time spent fussy in infancy was
related to childhood food quotient, but not signi®cantly to childhood energy intake. Further analysis
showed that infant fussiness was strongly related to
Infant temperament and later body composition
JCK Wells et al
carbohydrate intake, but not to fat intake, indicating
that fussy infants consumed signi®cantly more carbohydrate in childhood. Multiple regression analyses
showed that these relationships were not due to the
cross-sectional relationship between diet composition
and being upset in childhood, nor to differences in
body size. Given that food intake is determined
mainly by the mother or other caretakers in this age
group,39 this suggests that sugary foods may be used
by caretakers to relieve infant distress. The ability of
sweetened ¯uids to reduce crying has been noted in
infants within hours of birth,40 and in older infants
intra-oral sucrose produces a calming effect.41
Furthermore, it has been observed previously that
children's preference for sweet foods may be reinforced by giving such foods in a positive context.39
We speculate that general distress, measure by infant
fussiness, may have given rise to such a reinforced
preference in this sample. However, general distress
did not predict later body composition.
In contrast, infant distress to limitations did predict
later percentage fat, but not subcutaneous fat. When
multiple regression analysis was used to adjust for
baseline fatness and childhood age, the relationship
between infant distress to limitations and childhood
percentage fat dropped marginally below the 5% level
of signi®cance (P ˆ 0.059). This may be due to the
small sample size, since infant distress to limitations
was not related to infant fatness (r ˆ 70.01), and a
similar relationship has been observed in a larger
sample of infants.11 Infant distress to limitations did
not predict childhood food quotient or energy intake,
but it is nevertheless possible that these variables were
in¯uenced during the intervening period. A link
between infant temperament and later body composition would involve either a reduction in energy
expenditure or an increase in energy intake at some
time between the measurement periods. Our study
does not enable us to consider the possible relationship between infant irritability and subsequent energy
expenditure, and such a relationship cannot be discounted.
Infant soothability was related to later subcutaneous
but not percentage fatness. Soothability was also
related to later behaviour, with easily soothable
infants spending less time quiet and more time
active in childhood. These data are consistent with
the relationship between distress to limitations and
later percentage fat, although they relate to a different
aspect of fatness. Distress to limitations and soothability are not direct opposites, since soothability
refers to the ease with which a distressed infant can
be calmed by various strategies. Thus soothability and
distress to limitations may differ in the way they relate
to later fatness, and in neither case does the relationship described here indicate the mechanism involved.
Easily soothable infants may have become leaner and
more active in childhood through having higher
energy expenditure in the intervening period, or
lower energy intake. A cross-sectional relationship
between fatness and activity has been reported previously in children2 and was found in the present
study, but it is not known if fatness in¯uences activity
in this age group, or if the converse is the case. The
regression analyses in Table 6 indicate that childhood
sum of skinfolds and awake and active were related
only because both could be predicted by infant soothability.
Distress to limitations was found to be inversely
associated with childhood TV watching. Multiple
regression analysis showed that this relationship
could be explained as an artefact of the inverse
cross-sectional relationship in childhood between
time spent being upset and watching TV, since distress to limitations was strongly related to time spent
upset in childhood. No cross-sectional relationship
was apparent between hours of TV viewing and any
measure of childhood fatness, although such a relationship has been reported in a cross-sectional study
of 5 y old children.2
Skinfold thicknesses and percentage fat were not
in¯uenced in the same way by early temperament.
This may be because the two aspects of fatness are
poorly correlated42 and develop differently over time.
For example, total body fatness undergoes marked
change in the ®rst two years of life,16 while a recent
study found subscapular and triceps skinfold thicknesses to be relatively stable over the same period.43
Both physical activity level and diet may affect
subcutaneous and total fatness in different ways, and
furthermore fatness has been shown to have a stronger
effect on activity level in infants than vice versa.7,8
The relationship between fatness and energy balance
in infants and young children is complex and remains
poorly understood.
Conclusions
Our study shows that infant irritability can predict
later fatness, diet composition and physical activity
patterns, but it does not distinguish the mechanism by
which later fatness is in¯uenced. Previous studies
suggest that both food intake and activity levels may
exert an effect on body composition in slightly older
children.1,2 However, our study differed from these
previous studies in investigating temperament before
childhood physical activity patterns and feeding behaviour have developed. Because temperament develops
from birth and shows continuity over time, it may
have an important role in the development of subsequent behaviour relating to both feeding and physical
activity patterns.
Whether temperament in¯uences later body composition through the mechanism of energy intake or
energy expenditure merits further study. Unpublished
data from our group show that total energy expenditure in free-living 9 and 12 month children is inversely
405
Infant temperament and later body composition
JCK Wells et al
406
related to the amount of time spent upset. However,
anecdotal evidence also suggests dif®cult children are
fed more to quieten them.11 Both aspects of childhood
behaviour require further investigation.
24
References
1 Klesges RC, Klesges LM, Eck LH, Shelton ML. A longitudinal analysis of accelerated weight gain in preschool
children. Pediatr 1995; 95: 126±130.
2 Fontvieille AM, Harper IT, Ferraro RT, Spraul M, Ravussin E.
Daily energy expenditure by ®ve-year-old children, measured
by doubly-labelled water. J Pediatr 1993; 123: 200±207.
3 Wells JCK, Davies PSW. Energy cost of physical activity in
twelve week old infants. Am J Hum Biol 1995; 7: 85±92.
4 Wells JCK, Cole TJ, Davies PSW. Total energy expenditure
and body composition in early infancy. Arch Dis Child 1996;
75: 423±426.
5 Berkowitz RI, Agras WS, Korner AF, Kramer HC, Zeanah
CH. Physical activity and adiposity: a longitudinal study from
birth to childhood. J Pediatr 1985; 106: 734±738.
6 Davies PSW, Day JME, Lucas A. Energy expenditure in early
infancy and later body fatness. Int J Obes 1991; 15: 727±731.
7 Li R, O'Connor L, Buckley D, Specker B. Relation of activity
levels to body fat in infants 6 to 12 months of age. J Pediatr
1995; 126: 353±357.
8 Wells JCK, Stanley M, Laidlaw AS, Day JE, Davies PSW.
The relationship between components of infant energy expenditure and childhood body fatness. Int J Obes 1996; 20: 848±
853.
9 Rose H, Mayer J. Activity, calorie intake, fat storage and the
energy balance of infants. Pediatr 1968; 41: 18±29.
10 Mack RW, Kleinhenz ME. Growth, caloric intake, and activity
levels in early infancy: a preliminary report. Hum Biol 1974;
46: 345±354.
11 Carey WB. Temperament and increased weight gain in infants.
Dev Behav Pediatr 1985; 6: 128±131.
12 Wells JCK, Davies PSW. The relationship between behaviour
and energy expenditure in 12 week infants. Am J Hum Biol
1996; 8: 465±472.
13 Rothbart MK. Longitudinal observation of infant temperament. Dev Psychol 1986; 22: 356±365.
14 Roberts SB, Lucas A. Measurement of urinary constituents
and output using disposable napkins. Arch Dis Child 1985; 60:
1021±1024.
15 Davies PSW, Wells JCK. Calculation of total body water in
infancy. Eur J Clin Nutr 1994; 48: 490±495.
16 Fomon SJ, Haschke F, Ziegler EE, Nelson SE. Body composition of reference children from birth to age 10 years. Am J Clin
Nutr 1982; 35: 1169±1175.
17 Barr RG, Kramer MS, Boisjoly C, McVey White L, Pless IB.
Parental diary of infant cry and fuss behaviour. Arch Dis Child
1988; 63: 380±387.
18 Rothbart MK. Measurement of temperament in infants. Child
Dev 1981; 52: 569±578.
19 Sheng HP, Huggins RA. A review of body composition studies
with emphasis on total body water and fat. Am J Clin Nutr
1979; 32: 630±647.
20 Racette SB, Schoeller DA, Luke AH, Shay K, Hnilicka J,
Kushner RF. Relative dilution spaces of 2H- and 18O-labeled
water in humans. Am J Physiol 1994; 267: E585±E590.
21 Bingham SA. The dietary assessment of individuals; methods,
accuracy, new techniques and recommendations. Nutr Abs Rev
1987; 57: 705±742.
22 Harbottle L, Duggan MB. Dietary assessment in Asian childrenÐa comparison of the weighed inventory and diet history
methods. Eur J Clin Nutr 1993; 47: 666±672.
23 Price GM, Paul AA, Key FB, Harter AC, Cole TJ, Day KC,
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Wadsworth MEJ. Measurement of diet in a large national
survey: comparison of computerised and manual coding of
records in household measures. J Hum Nutr Dietet 1995; 8:
417±428.
Holland B, Welch AA, Unwin ID, Buss DH, Paul AA,
Southgate DAT. McCance and Widdowson's The Composition
of Foods. 5th edn. Royal Society of Chemistry: Cambridge,
1991.
Black AE, Prentice AM, Coward WA. Use of food quotients to
predict respiratory quotients for the doubly labeleld water
method of measuring energy expenditure. Hum Nutr: Clin
Nutr 1986; 40C: 381±391.
Cole TJ, Freeman JV, Preece MA. Body mass index reference
curves for the UK, 1990. Arch Dis Child 1995; 73: 25±29.
Thomas A, Chess S. Temperament and Development. Brunner/
Mazell: New York, 1977.
McDevitt SC, Carey WB. The measurement of temperament
in 3±7 year old children. J Child Psychol Psychiatr 1978; 19:
245±253.
Hegvik RL, McDevitt SC, Carey WB. The middle childhood
temperament questionnaire. J Dev Behav Pediatr 1982; 3:
197±200.
Korner AF, Zeanah CH, Linden J, Berkowitz RI, Kraemer HC,
Agras WS. The relation between neonatal and later activity
and temperament. Child Dev 1985; 56: 38±42.
Campbell S. Mother-infant interaction as a function of maternal ratings of temperament. Child Psychiatr Hum Dev 1979;
10: 67±76.
Klein P. The relation of Israeli mothers toward infants in
relation to infants' perceived temperament. Child Dev 1984;
55: 1212±1218.
Lee C, Bates J. Mother-child interaction at age two years and
perceived dif®cult temperament. Child Dev 1985; 56: 1314±
1325.
Crockenberg S, Acredolo C. Infant temperament ratings: a
function of infants, or mothers, or both? Inf Behav Dev 1983;
6: 61±72.
Fish M, Crockenberg S. Correlates and antecedents of ninemonth infant behaviour and mother-infant interaction. Inf
Behav Dev 1981; 4: 69±81.
Crockenberg S, Smith P. Antecedents of mother-infant interaction and infant irritability in the ®rst three months of life. Inf
Behav Dev 1982; 5: 105±119.
Peters-Martin P, Wachs T. A longitudinal study of temperament and its correlates in the ®rst 12 months. Inf Behav Dev
1984; 7: 285±298.
Crockenberg SB. Are temperamental differences in babies
associated with predictable differences in care-giving? In:
Lerner JV, Lerner RM (eds). Temperament and Social Interaction During Infancy and Childhood. New directions for
child development No. 31. Jossey-Bass: San Francisco,
1986, pp 53±73.
Ritchey N, Olson C. Relationships between family variables
and children's preference for and consumption of sweet foods.
Ecol Food Nutr 1983; 13: 257±266.
Blass EM, Hoffmeyer LB. Sucrose as an analgesic for newborn infants. Pediatr 1991; 87: 215±218.
Barr RG, Quek VSH, Cousineau D, Oberlander TF, Brian JA,
Young SN. Effects of intra-oral sucrose on crying, mouthing
and hand-mouth contact in newborn and six-week-old infants.
Dev Med Child Neurol 1994; 36: 608±618.
Davies PSW, Jones PRM, Norgan NG. The distribution of
subcutaneous and internal fat in man. Ann Hum Biol 1986; 13:
189±192.
Dewey KG, Heinig MJ, Nommsen LA, Peerson JM, LoÈnnerdal
B. Breast-fed infants are leaner than formula-fed infants at 1 y
of age: the Darling study. Am J Clin Nutr 1993; 57: 140±145.