Content uploaded by Ylva B Almquist
Author content
All content in this area was uploaded by Ylva B Almquist on Jan 19, 2016
Content may be subject to copyright.
.................................................................................................................................
European Journal of Public Health, Vol. 22, No. 3, 378–383
ßThe Author 2011. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved.
doi:10.1093/eurpub/ckr045 Advance Access published on 15 April 2011
.................................................................................................................................
Childhood friendships and adult health: findings from the
Aberdeen Children of the 1950s Cohort study
Ylva M. Almquist
Centre for Health Equity Studies (CHESS), Stockholm University/Karolinska Institutet, SE-106 91 Stockholm, Sweden
Correspondence: Ylva M. Almquist, Centre for Health Equity Studies (CHESS), Stockholm University/Karolinska Institutet, SE-106 91 Stockholm,
Sweden, Tel: +46 8 6747969, Fax: +46 8 162600, e-mail: ylva.almquist@chess.su.se
Background: Social relations are known to influence morbidity and mortality but few have studied this association from a life-course
perspective specifically targeting the importance of social relations in childhood for adult health. In childhood, a key aspect of children’s
relationships is the number of friendships a child has in the school class, i.e. friendship quantity. The overall aim of this study was to examine
the association between childhood friendships and adult self-rated health. Methods: Data from a longitudinal study of children born in
Aberdeen, Scotland, between 1950 and 1956 was used. Information on friendship quantity (1964) was derived from sociometric nominations
among classmates and defined as mutual choices. The health outcome was based on self-ratings derived from a questionnaire in 2001–03.
The study included various childhood and adult circumstances as possible explanatory factors. The analysis was based on ordinal logistic
regression, producing odds ratios (n= 5814). Results: The results demonstrated a gradient in women’s self-rated health according to the
number of friendships in the school class. A number of circumstances in childhood and adulthood were partial explanations. For men, only
those without friends reported poorer self-rated health in adulthood. This finding was explained by adult socioeconomic status. Conclusion:
It is concluded that childhood friendships are linked to health disparities in middle age, underlining the importance of such relationships and
the need for a life-course perspective on health that integrates a variety of mechanisms as they interact across life.
.................................................................................................................................
Introduction
Social relations have repeatedly been identified as important predictors
of morbidity and mortality.
1–4
Despite the great interest in this field,
however, the terminology has long been characterized by a lack of clarity.
5
One attempt to clarify the concepts adopts social relations as the key
notion, on the basis of which two dimensions—structure and
function—may be identified.
6
The former, which is in focus in the
present study, has commonly been defined by how many close relation-
ships an individual has.
4,6
Although some prospective studies on the
structural features of social relations and health have been conducted,
they have primarily been confined to adulthood. Studies which apply a
longer time span appear to be scarce. Given the decisive importance of
childhood it seems appropriate to focus on that phase of the life-course
when studying subsequent health outcomes.
7
A primary aspect of
children’s social relations is their interaction with peers and who they
form friendships with.
8
Making and keeping friends is a major concern
for young people, who invest a great deal of energy in group social life.
9
School, and particularly the school class, is perhaps the most important
forum for forming, maintaining and managing friendships in childhood.
The significance of friendships with classmates for child health has been
demonstrated in previous empirical work. Among other things,
friendlessness has been linked to depression, physical complaints and
psychological problems.
10–12
Not only are there differences between
having and not having friends but also the positive influences of friend-
ships appear to be cumulative.
13
Whether the number of friendships with classmates is also associated
with adult health remains to be further explored. An earlier study by
O
¨stberg and Modin
14
demonstrated that children with lower peer
status in the school class have higher risks of reporting poor health in
middle age. Peer status and friendships constitute two overlapping but
distinct dimensions of social interaction.
15
While peer status is an issue of
social hierarchy based on general acceptance and liking among
classmates, friendships may be interpreted in terms of social affiliation
that is primarily driven by homophily (attraction to similar others) and
involves reciprocity (mutual liking and confirmation).
8
Because of these
differences, aspects of childhood socioeconomic status, family
composition and individual characteristics that have previously been
linked to peer status
14
may not be as clearly associated with friendship
quantity. A related issue concerns the pathways linking friendship with
health. It has been suggested that peer status primarily influences
long-term health through various types of childhood circumstance that
may be linked to health either directly or via adult circumstances
(socioeconomic status, family conditions and social support).
14
Since
friendships, like peer status, involve advantages or disadvantages in the
classroom, one might expect that the similar pathways would apply here.
Some of the scientific literature on social relationships and health has
suggested that social support, for example, is an important mechanism.
1
Again, the major difference is that friendships are based on homophily,
making it difficult to predict how friends themselves shape the individ-
ual’s health development: peers are not only attracted to similar others
but they influence each other in terms of behaviours such as smoking
behaviour and alcohol use.
16
The uncertainties surrounding these issues
make it important to explore whether previously examined causes and
correlates of peer status as well as the pathways linking peer status with
adult health
14
may also apply to friendships. Additionally, since
health-related behaviours have previously been examined in the
scientific literature on both peer status
17
and friendships
16
they will
also be included in the present study.
The aim of the present study is to examine the association between
childhood friendships and adult health in a Scottish cohort born 1950–
56. Is friendship quantity in the school class related to adult self-rated
health? If so, can this association be accounted for by circumstances in
childhood (socioeconomic status, family composition and individual
characteristics) or adulthood (socioeconomic status, family composition,
health-related behaviours and social support)?
Methods
The data material was the Aberdeen Children of the 1950s Cohort study,
consisting of individuals who were born in Aberdeen Maternity Hospital
1950–56 and who attended Aberdeen schools in 1962 (n= 12 150). These
individuals participated in the Aberdeen Child Development survey in
1962 which included all children in school grades III-VII (ages 7–11
378 European Journal of Public Health
by guest on December 29, 2015http://eurpub.oxfordjournals.org/Downloaded from
years). The sociometric test from which information on friendship
quantity was derived, was carried out in 1964.
18
The new database was
created in 1999, when the cohort members were followed up for infor-
mation on mortality, sickness and health.
19
In 2001–03, those who were
alive and not lost to follow-up for other reasons such as emigration or
institutionalization, were asked to answer a postal questionnaire
(n= 11 282). Of these, 7183 individuals (64%) responded.
20
In the
present study, the study sample was defined as individuals for whom
there was information for all included variables (n= 5814).
Approximately 50% of the attrition is due to individuals attending
schools or classes that did not participate in the sociometric test as well
as those who were absent on the day of the data collection and for whom
no mutual nominations could be registered. The remaining attrition is
due to missing information for one or more of the other variables for
childhood and/or adulthood. This is likely to lead to an underestimation
of the number of mutual friendships and a possible overestimation of
friendless individuals. Moreover, response analysis (data not presented)
showed that the study sample was positively selected in terms of
friendship quantity and other childhood circumstances (whereby the as-
sociation between friendship quantity and adult health may be
underestimated).
Information on friendship quantity in childhood (1964) was derived
from the sociometric question ‘Which boy or girl in this class do you like
best?’ All the children were given a class roster in alphabetical order and
instructed by the teacher to underline three names. Friendship was
defined here as mutual nominations between any two individuals (0
friends, 18%; 1 friend, 35%, 2 friends, 31%; 3 friends, 16%). This
procedure has been used elsewhere.
21–23
Adult self-rated health (2001–03) was assessed by the question: ‘Over
the last 12 months would you say that your health on the whole has
been—(1) excellent, (2) good, (3) fair, (4) poor?’ Self-rated health has
been found to be a reliable measure of overall health status.
24–27
While the
measure is often collapsed into a dichotomous variable (good vs. less than
good health) it has been suggested that self-rated health forms a
continuum,
28
which is why all four response options were used in the
present study (excellent, 27%; good, 54%; fair, 14%; poor, 4%).
Childhood circumstances were defined first by social class. This infor-
mation was based on the father’s occupation (1962), classified into seven
categories (no corresponding information about the mother’s occupation
was available). Second, two types of information about family compos-
ition were included: mother’s marital status at childbirth, and number of
siblings (1962). Third, cognitive ability and behavioural problems were
included as individual characteristics. The child’s cognitive ability was
measured at age 7 years by means of the Moray House picture intelligence
test and age standardized with a mean of 100 (which was also used as the
cut-off point for the binary variable used in the present study).
Information about behavioural problems (1964) was derived from
teacher ratings of each child’s behaviour, using the Rutter B2 scale. The
scale consists of 26 statements with each answer rendering 0–2 points
depending on how well the statement applied to the child in question.
18
Two of the items (‘not liked’ and ‘solitary’) were, however, excluded since
they are obviously correlated with the friendship measure. In accordance
with recommendations,
29
individuals who achieved a total score of nine
or more were classified as having behavioural problems.
Adult circumstances were indicated through socioeconomic status,
family composition, health-related behaviours and social support,
derived from the postal questionnaire (2001–03). Measures of social
class, educational level and employment status were used to specify
socioeconomic status. Two types of information about family compos-
ition were included: marital status and whether the respondent had given
birth to/fathered any children. Concerning health-related behaviours, in-
formation on cigarette smoking and harmful consequences of alcohol
consumption was included. Alcohol consumption was addressed by the
question ‘In the last year how often have you had a hangover from
drinking alcohol’ which has previously been forward as a proxy of
binge drinking.
30
The alternatives ranged from ‘at least once a week’ to
‘not at all in the last year’. Since the vast majority of those who did not
respond to this question (approximately 6% of the study sample) had also
reported that they did not drink alcohol, these were coded into the latter
category. The question on cigarette smoking was: ‘Have you ever smoked
cigarettes regularly? (at least one cigarette a day for 12 months of more)’.
Respondents could answer ‘no’, ‘yes, current smoker’ or ‘yes, ex-smoker’.
Social support was established through the question: ‘People sometimes
need help and support from others. Do you have a relative or close friend
who is there for you: (i) if you are ill, (ii) if you need company, (iii) if you
need someone to talk to about personal problems?’ Those who responded
‘no’ to any of the items were classified as having low social support.
For a description of the explanatory variables according to friendship
quantity and self-rated health, see table 1.
The association between friendship quantity and adult self-rated health
was analysed by means of ordinal logistic regression using Stata/SEs (9.0)
ologit command, producing odds ratios (ORs). In order to correct for the
clustering of individuals into school classes (which creates within group
dependence in estimating standard errors), Stata’s cluster command was
used. As a first step, the ‘crude’ association between friendship quantity
and adult health was analysed. Next, in order to adjust for various child
and adult circumstances, eight models were generated. Since of gender
differences in the distributions of friendship quantity and self-rated
health, with women displaying a more compact distribution of number
of friends as well as a tendency to report poor health, men and women
were analysed separately. All analyses were adjusted for year of birth.
Results
Table 1 demonstrates the distribution of all the included child and adult
circumstances by number of friendships. Among those with a higher
friendship quantity it seems to be more common to have had a
married mother, few siblings, a high cognitive score and a lack of behav-
ioural problems in childhood. In adult life, these individuals are more
likely than those with fewer friends to have a formal education, be
employed, have a hangover at least once a month, be ex-smokers and
have high levels of social support. No clear tendency can be seen for child
or adult social class or for having children. Table 1 also presents the
prevalence of ‘less than good self-rated health’ according to the explana-
tory factors. In general, presumably adverse circumstances are related to a
higher prevalence of poor health.
The upper part of Table 2 presents the association between friendship
and adult health among men. The first column shows the crude associ-
ation and indicates that it is 34% more common among those without
friends than among those with three friends to rate their health as poor.
The corresponding result for those with one friend is 22%, but this
difference is not statistically significant. In Models 1–3, where
childhood circumstances are adjusted for, only marginal changes occur.
Model 4 incorporates the effect of adult socioeconomic status. This leads
to a major reduction in the estimates, leaving the association be-
tween friendship quantity and self-rated health statistically non-
significant. None of the remaining adult circumstances reduce the asso-
ciation to any large extent (Models 5–7). In the last model, all circum-
stances in childhood and adulthood are included simultaneously. Here,
the reduction caused by adult socioeconomic status is reflected again.
The lower part of Table 2 presents the corresponding association for
women. The first column demonstrates the crude association, where a
clear gradient in self-rated health by number of friendships can be
observed: it is 67% more common among those without friends, 49%
among those with one friend, and 19% among those with two friends to
have poor adult health than it is among those who have three reciprocal
relationships. The three first models adjust for childhood circumstances
which lead to some reduction of the estimates, especially when individual
characteristics are taken into consideration in Model 3 (this reduction is
primarily caused by adjusting for cognitive score). Including adult
socioeconomic status (Model 4) reduces the association further. The
estimates are only attenuated in Model 5, where family composition
in adulthood is adjusted for; they are then reduced slightly more when
health-related behaviours are introduced in Model 6 and social support in
Model 7. The last model takes all control variables into account simul-
taneously, resulting in a substantial reduction. However, it remains
Childhood friendships and adult health 379
by guest on December 29, 2015http://eurpub.oxfordjournals.org/Downloaded from
statistically significantly more common for people with fewer friends to
report poor health.
Discussion
The results of this study show that there is an association between
friendship quantity in childhood and adult self-rated health. Among
women, this association is a gradient where having fewer friendships is
related to poorer health. For men, on the other hand, only those without
friends reported significantly worse health. The results also indicate that
adult socioeconomic status is the most important mediator, of the cir-
cumstances investigated here, in the association for men. For women on
the other hand, a number of factors such as individual characteristics,
adult socioeconomic status and social support accounted for much of the
association. A noteworthy finding is the tendency towards a threshold
effect in the fully adjusted association between friendship quantity and
health: those with 0–1 friends seem to be worse off than those with 2–3
friends (although the estimate for men with one friend does not reach a
statistically significant level). While this might in part reflect a misclassi-
fication of reciprocal friendships because of the restrictions inherent to
the measurement procedure (i.e. the upper limit of three nominations) or
the absence of a possible friend on the day of the data collection, it is still
reasonable to suppose that those with only one friendship are as
vulnerable as those without friends. Hypothetically speaking, although
Table 1 Distribution (%) of number of friendships and prevalence (%) of ‘less than good self-rated health’ according to child and adult
circumstances (n= 5814). Statistical significance levels for ‘less than good self-rated health’ are adjusted for gender and year of birth
n(%) 0 Friends 1 Friend 2 Friends 3 Friends Less than good
self-rated health
Father’s occupational class
I (ref.) 185 (3) 3 3 3 4 9
II 703 (12) 11 12 12 12 13
IIINM 805 (14) 13 13 15 16 14
IIIM 2489 (43) 43 42 44 43 19**
IV 756 (13) 15 14 12 12 22***
V 595 (10) 10 11 10 9 26***
Not known 281 (5) 5 5 4 4 26***
Mother’s marital status
Married (ref.) 5687 (98) 97 97 98 99 19
Unmarried 127 (2) 3 3 2 1 26*
Number of siblings
0 595 (10) 11 10 10 9 16
1 (ref.) 1952 (34) 30 32 36 37 16
2–3 2443 (42) 41 43 41 42 19*
4- 824 (14) 17 15 14 11 26***
Cognitive score
100 (ref.) 4464 (77) 72 76 78 81 16
<100 1350 (23) 28 24 22 19 27***
Behavioural problems
No (ref.) 5546 (95) 93 95 96 97 18
Yes 268 (5) 7 5 4 3 31***
Adult social class
I (ref.) 352 (6) 6 6 6 7 8
II 2131 (37) 35 36 37 40 13**
IIINM 1471 (25) 24 26 26 23 18***
IIIM 1016 (18) 20 16 18 17 24***
IV 574 (10) 11 11 9 9 28***
V 214 (4) 4 4 3 3 46***
Not known 56 (1) 1 1 1 1 11
Educational level
Formal education (ref.) 4702 (81) 78 80 81 86 16
No formal education 1112 (19) 23 20 19 14 32***
Employment status
Employed (ref.) 5039 (87) 84 86 88 89 14
Unemployed 775 (13) 16 14 12 11 50***
Marital status
Never married 443 (8) 10 9 7 5 22**
Married (ref.) 4277 (74) 70 73 76 76 16
Widowed 90 (2) 2 2 1 1 33***
Divorced/separated 1004 (17) 19 17 17 18 27***
Children
Yes (ref.) 4938 (85) 85 84 85 86 19
No 876 (15) 15 16 15 14 20
Alcohol consumption (frequency of hangovers)
At least once a week 113 (2) 2 2 2 2 37***
2–3 times a month 231 (4) 4 4 4 5 22**
Once a month 400 (7) 5 6 7 9 18
Less than once a month (ref.) 2228 (38) 36 38 39 42 14
Not at all in the last year 2842 (49) 54 51 47 43 21***
Smoking
No (ref.) 2818 (48) 47 48 50 49 14
Yes, ex-smoker 1473 (25) 23 25 26 27 18**
Yes, current smoker 1523 (26) 30 26 25 24 28***
Social support
High (ref.) 5241 (90) 88 89 92 92 18
Low 573 (10) 12 11 8 8 27***
***P< 0.001, **P< 0.01, *P< 0.05
380 European Journal of Public Health
by guest on December 29, 2015http://eurpub.oxfordjournals.org/Downloaded from
individuals who have one friend presumably enjoy the benefits of having
a friendship, they may suffer from having narrow margins (for example, if
there are conflicts in the relationship or if the friend is absent from
school).
In a previous study, O
¨stberg and Modin
14
focused on childhood peer
status and adult health, using the same data material. As was discussed
earlier, peer status and friendships are different dimensions of social
interaction which have, for example, been shown to be independently
correlated with malaise in childhood.
31
The present study also gives
some support to this notion: while most of the child and adult circum-
stances included here are related to the peer status distribution,
14
they
were not as clearly linked in the present study to number of friendships
(even when gender-specific distributions were considered, data not
presented). This is most likely due to homophily: individuals who end
up at the bottom of the status distribution because of adverse circum-
stances are still capable of forming alliances with peers who are similar in
terms of, for example, social background.
22
However, where pathways to
health are concerned, peer status and friendships seem to be related to
adult self-rated health in a similar manner. In both cases, socioeconomic
status in adulthood emerged as a primary explanation for men while it
was the simultaneous adjustment that caused the largest reduction in the
association among women. Several lines of research may be combined to
explain why different pathways emerge for men and women. First of all, it
is reasonable to expect that a child’s success in forming friendships will be
mirrored in the subsequent personal networks that he or she
develops.
32,33
Furthermore, the social causes of health in adult women
and men are known to differ. Some studies have demonstrated that
women report higher levels of stress linked to their social networks,
while men’s stress tends to be associated with career issues such as job
loss and work problems.
34,35
These pathways may thus also be reflected in
the present study. One set of explanatory factors that neither was
included in the present study nor in the study by O
¨stberg and Modin
involves, for example, body composition (e.g. BMI), fitness and physical
activity. These factors may be assumed to reflect a potentially important
physiological pathway from social relationships to health and should be
examined in future studies.
Although friendship quantity appears to be influential for health de-
velopment, other dimensions of childhood friendships should be remem-
bered. Of these, friendship quality and the identity of friends have
previously been recognized as important.
36
Quality refers to the
function of social relations and is commonly assessed in terms of social
support. Identity of friends primarily refers to the attributes and
behaviours of the individuals who constitute the dyad. Since the
measure of friendship quantity in the present study could not differenti-
ate between friendships in terms of quality or functioning, the strength of
the association between friendships and subsequent health may have been
obscured. Moreover, these issues may also explain why neither social
support nor health-related behaviours explained more of the association
between friendship and health. If friendship had been assessed in terms of
supportive functions and shared behaviours (such as smoking or alcohol
consumption), a clearer picture might have emerged.
A life-course perspective requires a longitudinal data design, which
made the Aberdeen Children of the 1950s Cohort study highly suitable
for our purposes. This data material not only facilitated an objective
assessment of reciprocal friendships but also permitted the inclusion of
a broad range of child and adult circumstances. A number of issues do,
however, need to be recognized. The first point concerns the causal
nature of the association between friendships and health. This study
presumed that childhood friendships affect adult health, but a caveat
should be added: while health develops in a social context, an individual’s
advantages or disadvantages within that context are also influenced by his
or her health status. Social relations and health development are thus
probably mutual processes that evolve over time. Second, the school
class is not the only social context in which friendships emerge. Young
people who lack friends among classmates may form friendships
elsewhere (e.g. in the neighbourhood or through after-school activities)
which could have compensatory effects.
37
Nevertheless, the school class is
a social arena which is forced upon children and in which they spend a
great deal of time, making it a relevant focal point for the study of
childhood friendships. Finally, by implementing a life-course perspective
that draws upon the importance of conditions in childhood for adult out-
comes, the use of ‘historical’ data is unavoidable. While there is no reason
Table 2 ORs (95% CI) of adult self-rated health (2001–03) according to number of friendships in childhood (1964) among men (n= 2805) and women
(n= 3009)
Self-rated health (excellent = lowest, poor = highest)
Crude
(95% CI)
Model 1
a
(95% CI)
Model 2
b
(95% CI)
Model 3
c
(95% CI)
Model 4
d
(95% CI)
Model 5
e
(95% CI)
Model 6
f
(95% CI)
Model 7
g
(95% CI)
Model 8
h
(95% CI)
Number of friendships
Men
3 (ref.) 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
2 0.93 0.93 0.93 0.93 0.93 0.94 0.92 0.94 0.93
(0.78–1.11) (0.77–1.11) (0.78–1.10) (0.78–1.10) (0.78–1.10) (0.78–1.12) (0.77–1.11) (0.79–1.12) (0.78–1.11)
1 1.22 1.21 1.20 1.20 1.15 1.22 1.22 1.21 1.16
(0.95–1.58) (0.92–1.57) (0.92–1.56) (0.93–1.56) (0.89–1.49) (0.94–1.58) (0.94–1.61) (0.94–1.56) (0.89–1.52)
0 1.34 1.30 1.32 1.27 1.14 1.32 1.34 1.33 1.19
(1.12–1.59) (1.09–1.56) (1.10–1.59) (1.05–1.53) (0.95–1.38) (1.10–1.59) (1.09–1.64) (1.12–1.58) (0.96–1.48)
Women
3 (ref.) 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
2 1.19 1.18 1.17 1.15 1.14 1.19 1.20 1.17 1.14
(0.95–1.48) (0.95–1.48) (0.93–1.47) (0.93–1.44) (0.91–1.43) (0.95–1.49) (0.94–1.49) (0.93–1.47) (0.91–1.44)
1 1.49 1.46 1.46 1.43 1.40 1.47 1.45 1.42 1.33
(1.19–1.87) (1.16–1.84) (1.16–1.84) (1.13–1.79) (1.10–1.78) (1.17–1.84) (1.16–1.82) (1.13–1.79) (1.04–1.70)
0 1.67 1.62 1.61 1.55 1.50 1.62 1.59 1.54 1.38
(1.28–2.19) (1.22–2.15) (1.23–2.11) (1.18–2.04) (1.14–1.98) (1.23–2.13) (1.24–2.05) (1.18–2.00) (1.04–1.85)
a: Adjusted for socioeconomic status (childhood)
b: Adjusted for family composition (childhood)
c: Adjusted for individual characteristics (childhood)
d: Adjusted for socioeconomic status (adulthood)
e: Adjusted for family composition (adulthood)
f: Adjusted for health-related behaviours (adulthood)
g: Adjusted for social support (adulthood)
h: Adjusted for all circumstances (childhood and adulthood)
Results from ordinal logistic regression, adjusted for year of birth
Childhood friendships and adult health 381
by guest on December 29, 2015http://eurpub.oxfordjournals.org/Downloaded from
to believe that the significance of friendships among those born in the
1950s should be any different from children of today it is plausible that
new technological developments, such as text messaging, online
communities and instant messaging, have expanded the arena for daily
social interaction among young people.
38
Hypothetically, this progress
could favour those who already enjoy many friendships by increasing
the channels of social support while further accentuating the hardship
among those who lack friends.
To conclude, this study took its starting point in a life-course perspec-
tive that focused upon the importance of social relations in childhood for
future outcomes. The results indicate that number of friendships in the
school class plays a role in the understanding of disparities in adult
health. Moreover, the pathways from childhood friendships to adult
health seem to include mechanisms related to a variety of prerequisites,
resources and achievements across the life-course. If friendships are
indeed defining aspects of childhood
39
and if, in turn, childhood
conditions greatly influence life chances, this is clearly an important
area for policy making. Friendships are, by definition, voluntary and
cannot therefore be forced. Interventions which target the whole school
class in order to increase the prospects of positive peer interaction among
all children may be one answer. Such an approach, in combination with
specifically identifying those at risk, may not only result in a better func-
tioning peer group but could also improve the quality and functioning of
the friendships that already exist.
Acknowledgements
The author is grateful to Raymond Illsley for providing her with the
original data from the Aberdeen Child Development Survey. Heather
Clark managed the study at the Dugald Baird Centre, Aberdeen.
Funding
The follow-up of the Aberdeen ‘Children of the 1950s’ Study from 1998
was funded by the UK Medical Research Council and the Chief Scientists
Office, Scottish Executive Health Department. This study was financially
supported by the Swedish Council for Working Life and Social Research
(no. 2006-1637).
Conflicts of interest: None declared.
Key points
Previous studies have shown that friendships are highly
important for children’s health and well-being, but few have
examined the long-term health consequences of childhood
friendships.
This study revealed an association between the number of friend-
ships in childhood and adult self-rated health: subsequent
ill-health is more common among children who have few or no
friends in the school class. This is particularly clear among
women.
The results suggest that the pathways through which friendship
quantity may be linked to adult health differ between men and
women. For men, socioeconomic conditions in adulthood
emerged as the primary explanation whereas a number of cir-
cumstances explained some of the association among women.
References
1 Cohen S, Gottlieb BH, Underwood LG. Social relationships and health. In: Cohen S,
Underwood LG, Gottlieb BH, editors. Social Support Measurement and Intervention. A
Guide for Health and Social Scientists. New York: Oxford University Press, 2000: 3–28.
2 Uchino BN, Cacioppo JT, Kiecolt-Glaser JK. The relationship between social support and
physiological processes: a review with emphasis on underlying mechanisms and implica-
tions for health. Psychol Bull 1996;119:488–531.
3 Berkman LF, Syme SL. Social networks, host resistance, and mortality: a nine-year
follow-up study of Alameda County residents. Am J Epidemiol 1979;109:186–204.
4 Berkman LF. The role of social relations in health promotion. Psychosom Med
1995;57:245–54.
5 O’Reilly P. Methodological issues in social support and social network research. Soc Sci
Med 1988;26:863–73.
6 Due P, Holstein B, Lund R, et al. Social relations: network, support and relational strain.
Soc Sci Med 1999;48:661–73.
7 Keating D, Hertzman C, editors. Developmental Health and the Wealth of Nations: Social,
Biological, and Educational Dynamics. New York: Guilford, 1999.
8 Hartup WW. Social relationships and their developmental significance. Am Psychol
1989;44:120–6.
9 Cotterell J. Social Networks and Social Influences in Adolescence. London: Routledge, 1996.
10 Ueno K. The effects of friendship networks on adolescent depressive symptoms. Soc Sci Res
2005;34:484–510.
11 Brolin La
˚ftman S, O
¨stberg V. The pros and cons of social relations: an analysis of ado-
lescents’ health complaints. Soc Sci Med 2006;63:611–23.
12 Due P, Lynch J, Holstein B, Modvig J. Socioeconomic health inequalities among a
nationally representative sample of Danish adolescents: the role of different types of social
relations. J Epidemiol Community Health 2003;57:692–8.
13 Nangle DW, Erdley CA, Newman JE, et al. Popularity, friendship quantity, and friendship
quality: interactive influences on children’s loneliness and depression. J Clin Child
Adolescent Psychol 2003;32:546–55.
14 O
¨stberg V, Modin B. Status relations in school and their relevance for health in a life
course perspective: findings from the Aberdeen children of the 1950’s cohort study. Soc Sci
Med 2007;66:835–48.
15 Hartup WW. The company they keep: Friendships and their developmental significance.
Child Develop 1996;59:1–13.
16 Urberg KA, Deg
˘irmenciog
˘lu SM, Pilgrim C. Close friend and group influence on
adolescent cigarette smoking and alcohol use. Develop Psychol 1997;33:834–44.
17 Terre L, Drabman RS. Relationship between peer status and health behaviors. Adolescence
1992;27:595–602.
18 Batty GD, Morton SMB, Campbell D, et al. The Abdereen Children of the 1950s cohort
study: background, methods and follow-up information on a new resource for the study of
life course and intergenerational influences on health. Paediatr Perinatal Epidemiol
2004;18:221–39.
19 Leon DA, Lawlor DA, Clark H, Macintyre S. Cohort profile: The Aberdeen Children of the
1950s Study. Int J Epidemiol 2006;35:549–52.
20 Nishiwaki Y, Clark H, Morton SM, Leon DA. Early life factors, childhood cognition and
postal questionnaire response rate in middle age: the Aberdeen children of the 1950s study.
BMC Med Res Methodol 2005;5:16.
21 Kingery JN, Erdley CA. Peer experiences as predictors of adjustment across the middle
school transition. Educ Treat Children 2007;30:73–88.
22 Parker JG, Asher SR. Friendship and friendship quality in middle childhood: links with
peer group acceptance and feelings of loneliness and social dissatisfaction. Develop Psychol
1993;29:611–21.
23 Erdley CA, Nangle DW, Newman JE, Carpenter EM. Children’s friendship experiences and
psychological adjustment: theory and research. In: Nangle DW, Erdley CA, editors. New
Directions for Child and Adolescent Development: the Role of Friendship in Psychological
Adjustment. San Francisco: Jossey-Bass, 2001: 5–23.
24 Mossey JM, Shapiro E. Self-rated health: A predictor of mortality among the elderly. Am J
Public Health 1982;72:800–8.
25 Burstro
¨m B, Fredlund P. Self rated health: Is it as good a predictor of subsequent mortality
among adults in lower as in higher social classes. J Epidemiol Community Health
2001;55:836–40.
26 Kennedy BP, Kawachi I, Glass R, Prothrow-Stith D. Income distribution, socioeconomic
status, and self rated health in the United States: multilevel analysis. BMJ 1998;317:917–21.
27 Mackenbach JP, Simon JG, Looman CWN, Joung IMA. Self-assessed health and
mortality: could psychosocial factors explain the association. Int J Epidemiol
2002;31:1162–8.
28 Manor O, Matthews S, Power C. Dichotmous or categorical response? Analysing self-rated
health and lifetime social class. Int J Epidemiol 2000;29:149–57.
29 Rutter M. Classification and categorization in child psychiatry. J Child Psychol Psychiatr
1965;6:71–83.
30 Batty DG, Deary IJ, Macintyre S. Childhood IQ and life course socioeconomic position in
relation to alcohol induced hangovers in adulthood: the Aberdeen children of the 1950s
study. J Epidemiol Community Health 2006;60:872–4.
382 European Journal of Public Health
by guest on December 29, 2015http://eurpub.oxfordjournals.org/Downloaded from
31 O
¨stberg V. Children in classrooms: Peer status, status distribution and mental well-being.
Social Sci Med 2003;53:17–29.
32 Elicker J, Englund M, Sroufe LA. Predicting peer competence and peer relationships in
childhood from early parent-child relationships. In: Parke RD, Ladd GW, editors.
Family-peer Relationships: Modes of Linkage. Hillsdale, NJ: Erlbaum, 1992: 77–106.
33 Matthews SH. Friendships through the Life Course. Beverly Hills, CA: Sage, 1986.
34 Kessler RC, McLeod JD. Sex differences in vulnerability to undesirable life events. Am
Sociol Rev 1984;49:620–31.
35 Kendler KS, Thornton LM, Prescott CA. Gender differences in the rates of exposure to
stressful life events and sensitivity to their depressogenic effects. Am J Psychiatr
2001;158:587–93.
36 Hartup WW, Stevens N. Friendships and adaption in the life course. Psycholl Bull
1997;121:355–70.
37 Kiesner J, Poulin F, Nicotra E. Peer relations across contexts: individual-
network homophily and network inclusion in and after school. Child Develop
2003;74:1328–43.
38 Oksman V, Turtiainen J. Mobile communication as a social stage: meanings of
mobile communication in everyday life among teenages in Finland. New Media Soc
2004;6:319–39.
39 Doll B. Children without friends: implications for practice and policy. School Psychol Rev
1996;25:165–82.
.................................................................................................................................
European Journal of Public Health, Vol. 22, No. 3, 383–389
ßThe Author 2011. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved.
doi:10.1093/eurpub/ckr067 Advance Access published on 26 May 2011
.................................................................................................................................
Breastfeeding and obesity in Brazilian children
Juliana F. Novaes
1
, Joel A. Lamounier
2
, Enrico A. Colosimo
3
, Sylvia C. C. Franceschini
1
, Silvia E. Priore
1
1 Departamento de Nutric¸a
˜o e Sau
´de, Centro de Cie
ˆncias Biolo
´gicas e da Sau
´de, Universidade Federal de Vic¸osa. Vic¸ osa, MG, Brazil
2 Centro de Po
´s-Graduac¸a
˜o em Cie
ˆncias da Sau
´de, Faculdade de Medicina, Universidade Federal de Minas Gerais, Belo Horizonte, MG, Brazil
3 Departamento de Estatı
´stica, Instituto de Cie
ˆncias Exatas, Universidade Federal de Minas Gerais, Belo Horizonte, MG, Brazil
Correspondence: Juliana Farias de Novaes, Departamento de Nutric¸a
˜o e Sau
´de, Centro de Cie
ˆncias Biolo
´gicas e da Sau
´de, Universidade Federal de
Vic¸osa, Campus Universita
´rio, Av. P.H. Rolfs s/n. 36.570-000. Vic¸osa, Minas Gerais, Brazil, tel: +55 31 38993735, fax: +55 31 38992541,
e-mail: jnovaes@ufv.br
Background: The association between breastfeeding and obesity is inconsistent by the literature. This study aims to assess whether obesity is
associated to occurrence of breastfeeding and to duration of total and exclusive breastfeeding in Brazilian children. Methods: A
cross-sectional study was conducted with 764 children enrolled in public and private schools from Vic¸osa, Minas Gerais, Brazil. Obesity
(outcome variable) was defined as body mass index above the +2 standard deviations score using sex and age specific standards of
World Health Organization. Exposure was the occurrence and duration of breastfeeding. Potential confounders were controlled by
multiple logistic regression analysis and were divided in two groups: children (gender, age, birth weight, gestational age, order of birth,
number of siblings, number of persons in the residence, type of school, physical activity patterns and time watching television) and mothers
(age, nutritional status, level of education, weight gain during pregnancy, smokes currently and during the pregnancy). Results: Prevalence
of obesity was 10.7%; 6.8% of the children were not breastfed and 59.0% did not receive exclusive breastfeeding. After adjustment for
confounding variables by logistic regression analysis, no statistically significant association was observed between obesity and the occurrence
and/or duration of total and exclusive breastfeeding. There was no dose–response effect of duration of breastfeeding on prevalence of
obesity. Conclusion: Our results do not support the hypothesis that breastfeeding promotion would reduce obesity in this population.
Controversial findings regarding this association by literature indicate a need for further investigations.
.................................................................................................................................
Introduction
Breastfeeding is an interaction that creates a strong link between the
mother and the infant in the extra-uterine environment, similar to the
placental link between mother and fetus before birth.
1
Breastfeeding has
other advantages, including nutritional, immunological, psychological,
economic and environmental benefits.
2
The hypothesis that breastfeeding
has a protective effect against the development of obesity is biologically
sound and has received support from epidemiological studies.
Obesity is an emergent public health problem in Brazil and many other
countries. It is one of the most serious nutrition problems affecting
children, with potentially severe consequences for physical and mental
health. Obesity has multiple causes and consequences, representing a
challenge for health professionals who work with children. Preventive
measures can avoid long-term harmful consequences of organic or psy-
chosocial origin, so that low-cost preventive measures should be
favoured. If a protective role for breastfeeding is confirmed, it could
represent an effective weapon against obesity, adding to its many other
already known advantages.
3
Many hypotheses have been proposed to explain why breastfeeding
may protect children against obesity. Protective mechanisms would
involve since the specific and unique composition of human milk to
the influence of environmental and behavioural factors such as
socio-economic status (SES), maternal education, dietary patterns and
physical activity. Different factors, such as bioactive substances present
in human milk, the diminished intake of energy and/or protein and a
unique hormonal response, could contribute to decrease the risk of
childhood obesity.
4
The protective role of human milk against obesity
may be explained by different biological mechanisms involving its unique
composition as well as the metabolic and physiological responses it
induces. Human milk is qualitatively and quantitatively different from
any milk formula, due to the presence of bioactive substances that influ-
ence adipocyte proliferation and differentiation and, as such, tissue
growth and development.
3
Breastfeeding has also the advantage of
allowing the control of the amount of ingested milk, based on satiety
control.
5
As the flavor of breast milk is influenced by the mother’s diet,
breastfed children accept more easily different diets and tend to have
healthier dietary habits.
6
Some studies have shown that breastfeeding is protective against the
development of childhood obesity, in a dose-dependent manner.
7,8
Other
groups, however, did not observe a statistically significant association, so
that the relationship between breastfeeding and obesity is still under
debate.
1,9
The existence of contradictory findings in studies evaluating
the association between breastfeeding and childhood obesity may be due
Breastfeeding and obesity in children 383
by guest on December 29, 2015http://eurpub.oxfordjournals.org/Downloaded from
