Parental separation in childhood and adult inflammation: The importance of material and psychosocial pathways

Article (PDF Available)inPsychoneuroendocrinology 38(11) · July 2013with 84 Reads
DOI: 10.1016/j.psyneuen.2013.05.007 · Source: PubMed
Cite this publication
Abstract
Childhood adversities are known to be associated with poorer health outcomes. A potential mechanism may be through changes in inflammatory processes. One such childhood adversity is separation of parents, however relatively little is known about the association between parental separation and inflammation in adulthood. The aims of this study were to (1) investigate whether parental separation is associated with inflammation in mid-life, (2) focus upon the mechanisms that may be involved in translating childhood adversities, such as parental separation, into poorer health outcomes in adulthood. We examine the association of parental separation in childhood, defined as the breakdown of the parent's partnership, and levels of C-reactive protein (CRP) in middle age. The role played by material (through material disadvantage and educational attainment), psychosocial (through parent-child relationship quality and psychological distress) and adiposity (through BMI) mechanisms is investigated using path analysis in a multiply-imputed dataset from a British birth cohort with concurrent measurements made throughout the life course (n=7462). Participants that report parental separation have higher CRP levels at age 44 than those who grew up with both parents (β=0.16, 95% CI: 0.06, 0.27). This association is largely explained by BMI, material and psychosocial factors. Material disadvantage after separation and educational attainment seem to be particularly important in this association. Parental separation increases CRP in adulthood via chains of disadvantage across the life course. This study points towards potential points for intervention and highlights a need to support separating families in order to minimise the long-term impact on adult health.
+
Models
PNEC-2426;
No.
of
Pages
9
Please
cite
this
article
in
press
as:
Lacey,
R.E.,
et
al.,
Parental
separation
in
childhood
and
adult
inflammation:
The
importance
of
material
and
psychosocial
pathways.
Psychoneuroendocrinology
(2013),
http://dx.doi.org/10.1016/j.psyneuen.2013.05.007
Parental
separation
in
childhood
and
adult
inflammation:
The
importance
of
material
and
psychosocial
pathways
Rebecca
E.
Lacey
*
,
Meena
Kumari,
Anne
McMunn
Department
of
Epidemiology
&
Public
Health,
University
College
London,
United
Kingdom
Received
12
March
2013;
received
in
revised
form
24
April
2013;
accepted
14
May
2013
1.
Introduction
Many
studies
have
reported
an
association
between
child-
hood
adversity
and
CRP
levels,
a
reliable
marker
of
low-grade
inflammation
(Pepys
and
Hirschfield,
2003),
across
the
life
course.
Chronic
inflammation
is
correlated
with
poorer
Psychoneuroendocrinology
(2013)
xxx,
xxx—xxx
KEYWORDS
Cohort
study;
Divorce;
Inflammation;
Material
disadvantage;
NCDS;
Path
analysis;
Relationships
Summary
Background:
Childhood
adversities
are
known
to
be
associated
with
poorer
health
outcomes.
A
potential
mechanism
may
be
through
changes
in
inflammatory
processes.
One
such
childhood
adversity
is
separation
of
parents,
however
relatively
little
is
known
about
the
association
between
parental
separation
and
inflammation
in
adulthood.
The
aims
of
this
study
were
to
(1)
investigate
whether
parental
separation
is
associated
with
inflammation
in
mid-life,
(2)
focus
upon
the
mechanisms
that
may
be
involved
in
translating
childhood
adversities,
such
as
parental
separation,
into
poorer
health
outcomes
in
adulthood.
Methods:
We
examine
the
association
of
parental
separation
in
childhood,
defined
as
the
breakdown
of
the
parent’s
partnership,
and
levels
of
C-reactive
protein
(CRP)
in
middle
age.
The
role
played
by
material
(through
material
disadvantage
and
educational
attainment),
psychosocial
(through
parent—child
relationship
quality
and
psychological
distress)
and
adiposity
(through
BMI)
mechanisms
is
investigated
using
path
analysis
in
a
multiply-imputed
dataset
from
a
British
birth
cohort
with
concurrent
measurements
made
throughout
the
life
course
(n
=
7462).
Results:
Participants
that
report
parental
separation
have
higher
CRP
levels
at
age
44
than
those
who
grew
up
with
both
parents
(b
=
0.16,
95%
CI:
0.06,
0.27).
This
association
is
largely
explained
by
BMI,
material
and
psychosocial
factors.
Material
disadvantage
after
separation
and
educa-
tional
attainment
seem
to
be
particularly
important
in
this
association.
Conclusions:
Parental
separation
increases
CRP
in
adulthood
via
chains
of
disadvantage
across
the
life
course.
This
study
points
towards
potential
points
for
intervention
and
highlights
a
need
to
support
separating
families
in
order
to
minimise
the
long-term
impact
on
adult
health.
#
2013
Published
by
Elsevier
Ltd.
*
Corresponding
author
at:
Department
of
Epidemiology
&
Public
Health,
University
College
London,
1-19
Torrington
Place,
London
WC1E
6BT,
United
Kingdom.
Tel.:
+44
0207
679
1676;
fax:
+44
0207
813
0242.
E-mail
address:
rebecca.lacey@ucl.ac.uk
(R.E.
Lacey).
Available
online
at
www.sciencedirect.com
j
our
na
l
h
omepa
g
e:
www.e
lse
vie
r.c
om/l
oca
te/
psyne
ue
n
0306-4530/$
see
front
matter
#
2013
Published
by
Elsevier
Ltd.
http://dx.doi.org/10.1016/j.psyneuen.2013.05.007
health
outcomes,
such
as
depression
(Danner
et
al.,
2003;
Howren
et
al.,
2009),
type
II
diabetes
(Bassuk
et
al.,
2004)
and
coronary
heart
disease
(Danesh
et
al.,
2004).
Therefore
investigating
risk
factors
for
inflammation
in
adulthood,
and
the
mechanisms
through
which
they
act,
is
likely
to
be
fruitful
in
mitigating
the
effects
of
low
grade
inflammation
and
subsequent
disease.
Childhood
adversities
previously
investigated
include
low
socioeconomic
position
(e.g.
Phillips
et
al.,
2009;
Chen
et
al.,
2011;
Miller
and
Cole,
2012),
maltreatment
and
abuse
(Danese
et
al.,
2007,
2009),
as
well
as
broader
adverse
childhood
event
(Slopen
et
al.,
2010,
2012)
and
family
environment
scores
(Taylor
et
al.,
2006).
These
studies
suggest
that
childhood
adversity
is
associated
with
higher
CRP
levels
in
adolescence
and
adulthood.
How-
ever,
this
association
may
vary
by
the
type
of
childhood
adversity
investigated.
The
disadvantages
of
these
studies
are
three
fold;
firstly
many
are
cross-sectional
in
design
and
therefore
use
retrospective
measures
of
childhood
adversi-
ties,
which
may
be
prone
to
recall
bias.
Secondly
many
studies
have
used
small
convenience
samples
and
finally
existing
studies
have
generally
ignored
missing
data
and
conducted
complete
case
analyses.
It
is
known
that
those
who
have
complete
information
on
all
variables
of
interest
differ
from
those
who
have
missing
data
(Klebanoff
and
Cole,
2008),
being
more
advantaged
with
regards
to
health
and
social
circumstances.
We
focus
on
one
specific
type
of
child-
hood
adversity
(parental
separation)
in
this
study
as
it
is
likely
that
the
mechanisms
involved
vary
for
different
adversities.
Parental
separation
in
childhood
has
been
linked
previously
to
poorer
physical
and
mental
health
in
adulthood
and
to
stress
biomarkers,
such
as
cortisol
(Kraft
and
Luecken,
2009).
Relatively
little
is
known
about
the
mechanisms
through
which
the
experience
of
childhood
adversity
might
influence
chronic
inflammation
in
adulthood.
Tw o
main
pathways
invol-
ving
material
and
psychosocial
factors
are
suggested.
Regard-
ing
material
pathways
parental
separation
is
associated
with
increased
material
disadvantage
and
a
reduction
in
living
standards
at
different
stages
of
the
life
course
(Elliott
et
al.,
1993).
This
disproportionately
affects
lone-mother
headed
households
(Aassve
et
al.,
2007),
the
most
common
outcome
of
parental
separation.
Parental
separation
has
also
been
associated
in
many
studies
with
reduced
educational
attain-
ment
(Amato
and
Keith,
1991;
Ely
et
al.,
1999;
Ross
and
Mirowsky,
1999).
Although
education
is
not
a
material
factor
per
se
it
is
thought
to
be
the
main
way
in
which
material
disadvantage
is
transmitted
across
the
life
course.
In
turn
both
material
disadvantage
and
educational
attainment
have
been
linked
to
increased
CRP
levels
(Hemingway
et
al.,
2003;
Owen
et
al.,
2003;
Alley
et
al.,
2006).
It
is
also
possible
that
the
level
of
material
disadvantage
in
childhood
affects
risk
of
parental
divorce,
as
suggested
by
the
Family
Stress
Model
(Conger
et
al.,
1992).
Regarding
psychosocial
pathways,
parent—child
relation-
ships
may
deteriorate
as
a
consequence
of
parental
relation-
ship
breakdown
and
there
is
much
evidence
to
suggest
that
parental
separation
is
associated
with
poorer
quality
parent—
child
relations
(Zill
et
al.,1993;
Amato
and
Booth,
1996).
Also
there
is
a
well-established
link
between
parental
separation
and
adult
psychological
distress
(Amato
and
Keith,
1991;
Rodgers
et
al.,
1997;
Lacey
et
al.,
2012).
Poor
parent—child
relationship
quality
has
also
been
associated
with
increased
reporting
of
psychological
distress
in
adulthood
(Morgan
et
al.,
2012).
Finally
adult
psychological
distress
has
been
linked
to
CRP
in
previous
work
(Surtees
et
al.,
2008;
Copeland
et
al.,
2012;
Miller
and
Cole,
2012).
Therefore
it
is
possible
that
these
factors
are
involved,
and
interlinked,
in
a
psy-
chosocial
pathway
linking
parental
separation
in
childhood
to
CRP
in
adulthood.
Material
and
psychosocial
mechanisms
are
also
linked
across
the
life
course
and
this
is
something
we
additionally
investigate.
Obesity
is
often
considered
a
condition
of
low-
grade
inflammation
(Das,
2001)
and
the
association
of
BMI
with
CRP
is
well
established
(Visser
et
al.,
1999).
Evidence
suggests
that
both
material
(Wang
and
Beydoun,
2007)
and
psychosocial
disadvantage
(Luppino
et
al.,
2010)
are
asso-
ciated
with
increased
BMI.
It
is
therefore
possible
that
material
and
psychosocial
factors
are
linked
to
increased
CRP
via
BMI
as
shown
by
others
(Taylor
et
al.,
2006).
The
aim
of
this
study
was
to
(1)
investigate
whether
parental
separation
in
childhood
was
associated
with
CRP
in
adulthood,
and
(2)
whether
this
association
could
be
explained
by
material
factors
and
psychosocial
factors
acting
across
the
life
course.
Our
study
represents
an
enhancement
of
previous
research
as
we
use
prospective
data
from
a
large
British
birth
cohort
in
which
concurrent
measures
were
made;
we
focus
on
a
single
measure
of
childhood
adversity
+
Models
PNEC-2426;
No.
of
Pages
9
Please
cite
this
article
in
press
as:
Lacey,
R.E.,
et
al.,
Parental
separation
in
childhood
and
adult
inflammation:
The
importance
of
material
and
psychosocial
pathways.
Psychoneuroendocrinology
(2013),
http://dx.doi.org/10.1016/j.psyneuen.2013.05.007
Parental
separatio
n
(0-16 yrs)
Parent
-child
relatio
nship
quali
ty
(16
yrs)
Educational
attainment
(23
yrs
)
BMI
(42
yrs)
Material
disadvantage
(16
yrs)
C-reacti
ve
protein
(44
yrs)
Psych
ological d
istres
s
(23-42 yrs)
Material
disad
vanta
ge
(33
yrs)
Figure
1
Conceptual
model
showing
the
role
of
material
factors,
psychosocial
factors
and
BMI
between
parental
separation
(0—16
yrs)
and
CRP
(45
yrs).
2
R.E.
Lacey
et
al.
+
Models
PNEC-2426;
No.
of
Pages
9
Please
cite
this
article
in
press
as:
Lacey,
R.E.,
et
al.,
Parental
separation
in
childhood
and
adult
inflammation:
The
importance
of
material
and
psychosocial
pathways.
Psychoneuroendocrinology
(2013),
http://dx.doi.org/10.1016/j.psyneuen.2013.05.007
Table
1
Descriptive
information
and
comparison
of
observed
and
imputed
data
(n
=
7462).
Missingness
(%)
Observed
(%)
Imputed
(%)
Imputed
data
Parental
separation
(%)
No
parental
separation
(%)
CRP
(45
yrs)
Median
(mg/l)
0
e
0.94
0.94
1.11
0.93
a
Parental
separation
(0—16
yrs)
Yes
27.8
91.7
90.9
No
8.3
9.1
Material
disadvantage
score
(16
yrs)
Mean
23.6
0.01
0.02
0.79
0.05
b
Educational
qualifications
(23
yrs)
No
qualifications
16.7
10.9
11.4
21.4
10.4
c
CSE
2—5/O-level
49.6
49.9
56.3
49.2
A-level
18.6
18.3
12.2
18.9
Higher
qualification/degree
21.0
20.4
10.1
21.5
Social
class
(42
yrs)
I
(highest)
15.4
5.7
5.5
3.5
5.7
c
II
39.0
38.0
33.2
38.5
IIINM
20.5
20.3
18.8
20.5
IIIM
20.1
20.4
23.2
20.1
IV
11.6
12.3
16.4
11.9
V
(lowest)
3.1
3.5
4.9
3.3
Parent—child
relationship
quality
(16
yrs)
Very
good
quality
23.8
28.6
28.4
20.5
29.2
c
1.5
12.4
12.3
11.4
12.4
2
37.1
37.1
34.3
37.4
2.5
10.1
10.2
13.5
9.9
Uncertain
7.6
7.8
12.8
7.3
3.5
2.3
2.3
3.9
2.1
4
1.6
1.6
2.5
1.5
4.5
0.1
0.1
0.5
0.1
Very
poor
quality
0.3
0.3
1.2
0.2
Adult
malaise
(23—42
yrs)
Mean
of
hierarchical
factor
score
26.0
0.03
0.001
0.24
0.02
b
Body
Mass
Index
(42
yrs)
Mean
(kg/m
2
)
9.5
25.6
25.7
25.9
25.7
d
Smoking
status
(42
yrs)
Never
smoker
3.2
45.7
45.6
33.8
46.8
c
Ex-smoker
26.1
26.1
26.3
26.1
Current
smoker
28.2
28.3
40.0
27.1
Problem
drinking
(CAGE)
(42
yrs)
No
4.3
69.2
69.2
65.4
69.6
c
Yes
(CAGE
score
>
1)
30.8
30.8
34.6
30.4
Gender
(0
yrs)
Men
0
50.5
50.5
46.1
50.9
c
Women
49.5
49.5
53.9
49.1
Father’s
social
class
(0
yrs)
I
(highest)
9.5
4.9
4.9
3.0
5.0
c
II
14.3
14.2
9.7
14.6
IIINM
10.2
10.1
11.3
9.9
IIIM
50.7
50.5
53.6
50.2
IV
12.0
12.1
13.2
12.1
V
(lowest)
7.9
8.2
9.2
8.1
Abbreviations:
CRP:
C-Reactive
Protein;
CAGE:
Cut
down,
Annoyed,
Guilty
and
Eye
opener
items
used
to
screen
for
problem
alcohol
consumption;
CSE:
Certificate
of
Secondary
Education.
NB
only
%s
given
as
Ns
vary
across
the
20
imputed
datasets.
a
p
<
0.001,
obtained
by
Wilcoxon
rank
sum
test
of
medians.
b
p
<
0.001,
obtained
by
t-test.
c
p
<
0.001,
obtained
from
x
2
test.
d
p
>
0.05,
obtained
by
t-test.
e
No
missing
values
as
analysis
is
on
those
with
CRP
values.
Parental
separation
and
adult
inflammation:
pathways
3
and
we
account
for
a
number
of
factors
that
may
confound
our
association
such
as
material
factors
before
parental
separation.
The
specific
pathways
investigated
are
presented
in
Fig.
1.
2.
Methods
2.1.
Sample
This
study
uses
data
from
the
National
Child
Development
Study
(NCDS).
This
study
was
initiated
as
the
Perinatal
Mor-
tality
Study
aiming
to
recruit
all
the
babies
born
during
a
single
week
of
1958
(achieved
sample
=
17,414,
98.2%
of
target)
in
Great
Britain
(Power
and
Elliott,
2005).
Partici-
pants
have
been
followed-up
at
the
following
ages
so
far:
7,
11,
16,
23,
33,
42,
44,
46
and
50
years.
The
study
is
multi-
disciplinary
collecting
information
on
educational,
eco-
nomic,
medical,
developmental
and
social
aspects
of
parti-
cipants’
lives
from
multiple
sources.
This
study
uses
information
from
birth
through
44
years
as
this
latter
wave
is
a
biomedical
survey
and
the
only
wave
in
which
blood
samples
were
collected.
This
wave
collected
data
on
a
sub-
sample
of
NCDS
participants
(achieved
sample
=
9377,
78%
of
target)
(Elliott
et
al.,
2008).
In
this
study
we
use
those
participants
from
whom
blood
samples
were
taken
from
this
survey
(n
=
8233,
87.8%
of
target).
2.2.
Measures
2.2.1.
C-reactive
protein
(CRP)
Blood
samples
were
collected
at
age
44
years
during
the
biomedical
survey.
CRP
was
measured
in
citrated
plasma
by
high-sensitivity
nephelometric
analysis
of
latex
particles
coated
with
CRP-monoclonal
antibodies
(Elliott
et
al.,
2008).
CRP
was
the
only
inflammatory
marker
collected
in
this
dataset.
Participants
with
levels
10
mg/L,
indicative
of
infection,
pathology
or
trauma,
were
excluded
from
this
analysis.
CRP
values
were
missing
for
136
people
who
were
taking
warfarin
or
were
pregnant,
9
participants
whose
results
were
not
recorded,
413
participants
who
provided
consent
but
blood
was
not
taken,
and
586
participants
who
refused
to
provide
a
blood
sample.
Those
with
missing
CRP
levels
were
more
likely
to
be
male
but
did
not
differ
on
other
analysis
variables.
Values
were
positively
skewed
and
were
therefore
log-transformed.
2.2.2.
Parental
separation
in
childhood
Parental
separation
in
this
study
covers
both
divorce
and
the
breakdown
of
cohabiting
relationships.
This
measure
was
prospectively
derived
from
birth
to
16
years.
At
age
7
this
information
was
provided
by
the
health
visitor
interviewing
the
family.
At
ages
11
and
16
this
was
derived
from
informa-
tion
on
the
identity
of
parental
figures
and
reasons
for
any
change
reported
by
the
mother.
If
there
was
a
change
in
parental
figures
and
the
reason
given
for
this
change
was
‘divorce’
or
‘separation’
these
children
were
recorded
as
having
experienced
parental
separation.
2.2.3.
Material
pathway
factors
Material
disadvantage
at
age
16
years
was
measured
using
several
variables
all
reported
by
the
participant’s
mother:
housing
tenure,
free
school
meal
receipt,
overcrowding
(1.5
person
per
room),
perceived
financial
hardship
in
the
past
year,
access
to
household
amenities
(inside
lavatory,
cooking
facilities
and
hot
water),
whether
a
child
shared
a
bed
with
others,
and
benefit
receipt.
A
single
scale
was
created
from
these
items
using
multiple
correspondence
analysis.
This
is
a
method
of
data
reduction
similar
to
prin-
cipal
components
analysis
but
is
more
appropriate
when
nominal
variables
are
involved.
High
scores
are
indicative
of
material
advantage.
Educational
attainment
was
measured
as
the
highest
qualification
achieved
by
age
23
years
(no
qualifications,
CSE/O-level
(NVQ1-2),
A-level
(NVQ3),
higher/degree
(NVQ4+)).
Adult
material
disadvantage
was
indicated
by
participant’s
occupational
social
class.
This
was
classified
using
the
Registrar
General’s
Social
Class
(RGSC)
schema,
based
upon
occupation,
and
has
the
following
categories:
I
(professional),
II
(managerial
and
technical),
IIINM
(skilled
non-manual),
IIIM
(skilled
manual),
IV
(semi-skilled
manual)
and
V
(unskilled).
Individuals
who
were
not
working
were
included
in
group
V
as
this
is
likely
to
be
associated
with
disadvantage.
2.2.4.
Psychosocial
pathway
factors
Parent—child
relationship
quality
was
measured
using
two
items
at
age
16
years
‘I
get
on
well
with
my
mother’
and
‘I
get
on
well
with
my
father’.
Responses
were
on
a
Likert
scale
ranging
from
(1)
‘very
true’
to
(5)
‘very
untrue’.
The
mean
of
these
two
items
was
taken.
If
participants
only
had
one
parent
only
the
score
for
that
parent
was
used.
Rutters
Malaise
Inventory
was
used
to
measure
psycho-
logical
distress
at
ages
23,
33
and
42.
This
comprises
24
yes/
no
items
on
both
emotional
and
somatic
symptoms
(Rutter,
1970).
A
hierarchical
factor
analysis
was
conducted,
firstly
deriving
two
factors
for
each
wave
(one
emotional
and
one
somatic,
as
have
been
identified
by
previous
work
(Rodgers
et
al.,
1999))
and
then
deriving
an
overall
factor
from
just
the
three
emotional
factors
obtained
from
the
first
step.
This
approach
is
preferable
to
measuring
psychological
distress
at
a
single
wave
and
reduces
the
chances
of
finding
a
strong
association
with
CRP
which
is
driven
by
physical
symptoms
captured
by
the
Malaise
Inventory.
2.2.5.
Body
Mass
Index
(BMI)
BMI
was
taken
from
age
42
using
self-reported
height
and
weight.
While
objective
measures
of
BMI
were
available
at
other
ages
in
NCDS,
BMI
at
age
42
years
fitted
best
in
our
conceptual
model
as
it
follows
adult
psychological
distress
and
material
disadvantage
and
is
prior
to
CRP
measured
at
age
44.
Sensitivity
analyses
showed
that
BMI
at
42
years
correlated
highly
with
measured
BMI
at
the
preceding
and
following
waves
(r
=
0.739
BMI33—BMI42,
r
=
0.821
BMI42—
BMI45,
r
=
0.783
BMI33—BMI45).
Also
associations
involving
BMI
in
our
conceptual
model
(e.g.
BMI
to
CRP)
were
almost
identical
regardless
of
the
BMI
measure
used.
2.2.6.
Covariates
Father’s
social
class
(RGSC)
from
the
birth
survey
was
used
as
an
indicator
of
material
disadvantage
prior
to
separation
in
order
to
control
for
selection
effects
(coding
is
the
same
as
for
participant’s
own
social
class
above).
This
is
the
only
available
indicator
of
socioeconomic
position
available
in
this
+
Models
PNEC-2426;
No.
of
Pages
9
Please
cite
this
article
in
press
as:
Lacey,
R.E.,
et
al.,
Parental
separation
in
childhood
and
adult
inflammation:
The
importance
of
material
and
psychosocial
pathways.
Psychoneuroendocrinology
(2013),
http://dx.doi.org/10.1016/j.psyneuen.2013.05.007
4
R.E.
Lacey
et
al.
wave.
We
also
controlled
for
smoking
status
at
age
42
(never
smoked,
ex-smoker,
current
smoker),
problem
drinking
at
age
42
(CAGE
score
>
1)
and
sex.
2.2.7.
Statistical
analysis
Missing
data
is
a
problem
for
longitudinal
studies,
potentially
resulting
in
reduced
statistical
power,
bias
and
unrepresen-
tativeness.
Indeed
complete
cases
in
this
dataset
were
found
to
differ
on
most
variables
used
in
this
analysis,
being
much
more
socially
advantaged
than
those
who
had
missing
infor-
mation.
Therefore
a
complete
case
analysis
would
be
inap-
propriate.
We
therefore
used
multiple
imputation
by
chained
equations
(MICE)
to
account
for
missing
data
and
account
for
differential
attrition.
This
approach
is
useful
where
there
is
missing
data
on
many
variables,
such
as
in
this
study.
The
imputation
model
contained
all
analysis
variables,
variables
predictive
of
missingness
(e.g.
indicators
of
disadvantage),
and
variables
thought
to
provide
useful
information
to
fill
in
gaps
(e.g.
same
measures
from
preceding
or
subsequent
waves).
We
imputed
20
datasets
for
those
with
a
CRP
value
(MICE
was
implemented
prior
to
removing
those
with
CRP
values
10
mg/L)
and
subsequent
analyses
were
run
across
all
20
datasets.
Table
1
shows
the
proportion
of
missing
information
for
each
variable
and
also
compares
the
observed
and
imputed
data,
suggesting
that
the
imputation
has
been
appropriately
conducted.
Our
final
sample
size
was
7462.
Chi-squared
tests,
t-tests
and
Wilcoxon
rank
sum
tests
were
used
to
assess
bivariate
associations
between
parental
separation
and
all
analysis
variables.
Linear
regression
was
used
to
test
the
unadjusted
and
adjusted
association
between
parental
separation
and
CRP.
Path
analysis
was
then
used
to
simultaneously
estimate
all
associations
in
our
con-
ceptual
model,
adjusted
for
all
control
variables.
The
model
was
then
refined
according
to
the
modification
indices
and
non-statistically
significant
associations
were
removed,
starting
with
the
association
with
a
p
value
closest
to
1.
The
final
model
therefore
represents
all
associations
which
are
statistically
significant
at
the
5%
level.
A
Wald
test
was
used
to
test
the
relative
weight
of
purely
material
and
psychosocial
pathways
in
order
to
assess
which
group
was
more
important.
3.
Results
3.1.
Sample
characteristics
8.4%
of
the
sample
had
experienced
parental
separation
during
childhood
(Table
1).
For
those
who
experienced
par-
ental
separation,
median
CRP
was
significantly
higher
at
1.11
mg/L,
compared
to
0.93
mg/L
for
those
who
did
not.
Those
who
experienced
separation
were
less
materially
advantaged,
had
lower
educational
attainment
and
were
+
Models
PNEC-2426;
No.
of
Pages
9
Please
cite
this
article
in
press
as:
Lacey,
R.E.,
et
al.,
Parental
separation
in
childhood
and
adult
inflammation:
The
importance
of
material
and
psychosocial
pathways.
Psychoneuroendocrinology
(2013),
http://dx.doi.org/10.1016/j.psyneuen.2013.05.007
Table
2
Association
between
parental
separation
(0—16
yrs)
and
adult
CRP
(45
yrs).
Standardised
regression
coeff.
(b)
95%
CI
p
Unadjusted
association
Parental
separation
(0—16
yrs)
No
separation
Ref
Separation
0.16
0.06,
0.27
0.002
Adjusted
association
a
Parental
separation
(0—16
yrs)
No
separation
Ref
Separation
0.10
0.01,
0.21
0.042
Smoking
status
(42
yrs)
Never
smoker
Ref
Ex-smoker
0.07
0.01,
0.14
0.018
Current
smoker
0.31
0.25,
0.37
<0.001
Problem
drinking
(CAGE)
(42
yrs)
No
Ref
Yes
(CAGE
score
>1)
0.01
0.06,
0.05
0.886
Gender
(0
yrs)
Men
Ref
Women
0.01
0.04,
0.06
0.592
Father’s
social
class
(0
yrs)
I
(highest)
0.30
0.42,
0.18
<0.001
b
II
0.31
0.38,
0.23
IIINM
0.20
0.29,
0.11
IIIM
Ref
IV
0.05
0.04,
0.13
V
(lowest)
0.03
0.06,
0.13
Abbreviations:
CAGE:
Cut
down;
AnnoyedGuilty
and
Eye
opener
items
used
to
screen
for
problem
alcohol
consumption;
CI:
confidence
interval
a
Association
between
parental
separation
and
CRP
controlling
for
sex,
father’s
social
class,
smoking
status
and
problem
drinking.
b
p
value
for
trend.
Parental
separation
and
adult
inflammation:
pathways
5
more
likely
to
be
in
a
manual
social
class.
Also
those
in
the
separation
group
were
more
likely
to
report
poorer
quality
parent—child
relations
and
psychological
distress.
Those
who
grew
up
with
two
parents
were
less
likely
to
be
a
smoker
or
a
problem
drinker
at
age
42,
and
were
less
likely
to
have
fathers
from
social
classes
I
and
II.
However
BMI
did
not
vary
by
parental
separation
status.
3.2.
Association
between
parental
separation
and
CRP
Table
2
shows
the
results
from
linear
regression
analyses.
Those
who
experienced
parental
separation
in
childhood
had
higher
CRP
levels
at
age
44
than
those
who
did
not.
This
association
only
slightly
attenuated
on
adjustment
for
control
variables
(father’s
social
class,
smoking,
problem
drinking
and
gender).
In
addition
father’
social
class
and
smoking
status
were
associated
with
CRP.
3.3.
Pathways
Results
of
testing
the
path
model
are
shown
in
both
Fig.
2
and
Table
3.
The
final
model
in
Fig.
2
differs
from
the
initial
path
model
as
associations
have
been
added
in
response
to
the
modification
indices
and
also
some
associations
have
been
removed
as
they
were
not
statistically
significant.
The
most
notable
association
that
was
removed
was
the
direct
path
between
parental
separation
and
CRP.
Upon
modelling
the
indirect
paths
through
material
and
psychosocial
factors,
this
direct
pathway
was
no
longer
statistically
significant,
sug-
gesting
that
these
factors
largely
mediate
the
association
between
parental
separation
and
CRP.
As
we
added
each
+
Models
PNEC-2426;
No.
of
Pages
9
Please
cite
this
article
in
press
as:
Lacey,
R.E.,
et
al.,
Parental
separation
in
childhood
and
adult
inflammation:
The
importance
of
material
and
psychosocial
pathways.
Psychoneuroendocrinology
(2013),
http://dx.doi.org/10.1016/j.psyneuen.2013.05.007
Parent-child
relationship
quality
(16 yrs)
j
h
Psychological distress
(23-42 yrs)
i
k
C-reactive
protein
(44 yrs)
g
BMI
(42 yrs)
f
d
e
Educational
attainment
(23 yrs)
c
Parental
separation
(0-16 yrs)
Material
disadvantage
(16 yrs)
a
b
Figure
2
Final
model
of
material
and
psychosocial
mechanisms
linking
parental
separation
and
adult
inflammation
in
the
NCDS.
Table
3
Path
analysis
results
for
final
model.
Path
label
Association
tested
Standardised
regression
coeff.
(b)
p
Value
a
Parental
separation
!
material
disadvantage
0.24
<0.001
b
Parental
separation
!
education
0.06
<0.001
c
Material
disadvantage
!
education
0.33
<0.001
d
Education
(no
qual)
!
psychological
distress
Ref
<0.001
a
Education
(CSE)
!
psychological
distress
0.26
Education
(A-level)
!
psychological
distress
0.30
Education
(Higher)
!
psychological
distress
0.33
e
Education
(no
qual)
!
BMI
Ref
<0.001
a
Education
(CSE)
!
BMI
0.14
Education
(A-level)
!
BMI
0.15
Education
(Higher)
!
BMI
0.21
f
Education
(no
qual)
!
CRP
Ref
<0.001
a
Education
(CSE)
!
CRP
0.27
Education
(A-level)
!
CRP
0.48
Education
(Higher)
!
CRP
0.61
g
BMI
!
CRP
0.33
<0.001
h
Parental
separation
!
parent—child
relationship
qual.
0.08
<0.001
i
Parental
separation
!
psychological
distress
0.04
0.001
j
Parent—child
relationship
qual.
!
psychological
distress
0.14
<0.001
k
Psychological
distress
!
CRP
0.04
0.002
Abbreviations:
BMI:
Body
Mass
Index;
CRPC-Reactive
Protein;
CSE:
Certificate
of
Secondary
Education.
Model
fit:
RMSEA
=
0.062;
CFI
=
0.469;
TLI
=
0.400;
Rsq
for
CRP
=
0.149;
NB
confidence
intervals
are
not
available
for
STDYX
estimates
with
imputed
data
in
MPlus.
Model
adjusted
for
sex,
smoking
status
(42
yrs),
problem
alcohol
consumption
(42
yrs),
father’s
social
class
(0
yrs).
a
p
Value
for
trend.
6
R.E.
Lacey
et
al.
group
of
pathway
variables
the
model
fit
indices
suggested
an
improved
model
fit
(i.e.
RMSEA
reduced
towards
0,
CFI
and
TLI
increased
towards
1).
Adolescent
material
disadvantage
and
educational
attain-
ment
appeared
to
be
important
in
mediating
the
association
between
parental
separation
and
adult
CRP.
In
particular
the
results
suggest
that
children
who
experienced
separation
tended
to
be
more
materially
disadvantaged
and
tended
to
do
less
well
in
education.
Social
class
at
age
42
was
removed
as
the
associations
passing
through
this
variable
were
not
statistically
significant.
The
associations
which
involve
edu-
cational
attainment
appear
to
be
particularly
strong
(com-
paring
standardised
estimates).
Psychosocial
factors
are
also
important
in
mediating
the
overall
association;
those
who
experienced
parental
separation
reported
poorer
quality
parent—child
relationships
and
were
more
likely
to
be
psy-
chologically
distressed
in
adulthood.
A
link
was
found
between
material
and
psychosocial
mechanisms;
those
who
had
a
lower
level
of
educational
attainment
were
more
likely
to
be
psychologically
distressed.
There
is
also
evidence
that
BMI
is
involved
in
the
relation-
ship
between
educational
attainment
and
inflammation
in
mid-life,
but
only
in
partially
mediating
the
effect
of
educa-
tional
attainment
on
CRP.
BMI
was
not
involved
in
translating
psychosocial
disadvantage
into
adult
inflammation.
All
the
associations
in
the
model
exist
after
controlling
for
father’s
social
class
at
birth
as
an
indicator
of
disadvantage
prior
to
parental
separation,
taking
account
that
families
who
separate
tend
to
be
more
disadvantaged
prior
to
the
separation
process.
We
additionally
tested
the
relative
importance
of
material
and
psychosocial
mechanisms
using
a
Wald
test,
and
find
that
material
factors
are
more
impor-
tant
(
p
<
0.001)
in
the
association
between
parental
separa-
tion
and
adult
CRP.
4.
Discussion
In
this
study
we
find
that
parental
separation
occurring
during
childhood
is
associated
with
increased
inflammation
in
adult-
hood
in
a
large
British
birth
cohort
study.
This
finding
is
supported
by
previous
work
which
finds
a
relationship
between
childhood
adversities
and
adult
inflammation
(e.g.
Taylor
et
al.,
2006;
Danese
et
al.,
2007,
2008,
2009;
Slopen
et
al.,
2010;
Appleton
et
al.,
2012;
Tietjen
et
al.,
2012).
The
mechanisms
through
which
parental
separation
leads
to
a
pro-inflammatory
profile
in
adult
life
include
material
factors,
psychosocial
factors
and,
more
proximally,
BMI.
The
findings
suggest
that
parental
separation
appears
to
set
individuals
down
a
path
of
disadvantage
characterised
by
increased
material
disadvantage,
reduced
educational
attainment
which
is
associated
directly
with
increased
inflammation
but
also
operates
through
increased
adult
psy-
chological
distress
and
increased
BMI.
These
findings
concur
with
others
who
find
that
parental
separation
is
associated
with
long-term
disadvantage
in
both
material
(Amato
et
al.,
1991;
Ely
et
al.,
1999)
and
psychosocial
(Cooney,
1994;
Rodgers
et
al.,
1997;
Chapman
et
al.,
2004)
domains
of
life,
and
link
to
Sweeting
and
West’s
concept
of
the
‘unhealthy
life
career’
hypothesis
connecting
disadvantage
in
the
family
of
origin
to
adult
health
(Sweeting
and
West,
1995).
Links
between
reduced
educational
attainment,
increased
psycho-
logical
distress,
and
increased
CRP
in
mid-life
corroborate
previous
work
(Hemingway
et
al.,
2003;
Copeland
et
al.,
2012).
While
the
effect
of
educational
attainment
on
CRP
was
partially
explained
by
increased
BMI
amongst
those
with
lower
qualifications,
BMI
was
not
linked
to
psychosocial
mechanisms
as
previously
found
by
others
(Taylor
et
al.,
2006).
Some
additional
pathways
were
suggested
by
the
model
modification
indices
which
were
added
to
the
final
model,
representing
additional
linkages
between
material
and
psy-
chosocial
mechanisms.
In
particular
a
link
between
educa-
tional
attainment
and
adult
psychological
distress
was
added,
and
it
is
well
established
that
socioeconomic
factors,
of
which
education
can
be
considered,
are
associated
with
psychological
health.
Upon
testing
the
relative
weight
of
material
and
psycho-
social
mechanisms,
material
factors
were
found
to
be
parti-
cularly
important
and
this
may
represent
the
more
objective
nature
of
the
measurements
used,
compared
to
those
used
to
capture
the
quality
of
parent—child
relationships
and
psy-
chological
distress.
Associations
involving
educational
attain-
ment
appear
to
be
the
strongest
and
this
may
suggest
that
supporting
children
through
education
may
help
to
limit
the
long-term
consequences
for
individual
health.
After
taking
into
account
all
mechanisms,
there
was
no
longer
a
statis-
tically
significant
association
between
parental
separation
and
adult
CRP,
suggesting
that
these
mechanisms
largely
explain
this
association.
The
results
of
this
study
suggest
that
social
disadvantages
investigated
in
our
model
may
lead
to
biological
change
through
alterations
in
low
grade
inflammation,
a
risk
factor
for
later
pathology.
The
potential
biological
mechanisms
involved
are
complex.
Material
and
psychosocial
stress
have
been
linked
to
changes
in
the
hypothalamic-pituitary-adrenal
(HPA)
axis,
and
chronically
raised
cortisol
levels
may
poten-
tially
decrease
glucocorticoid
sensitivity
leading
to
raised
inflammation
(Fagundes
et
al.,
2013).
Stress-induced
auto-
nomic
function
may
also
be
enhanced
in
those
who
experi-
ence
chronic
stress
and
this
may
be
an
additional
source
of
chronic
inflammation
through
raised
noradrenaline
(Fagundes
et
al.,
2013).
With
regards
to
timing
across
the
life
course
it
is
possible
that
inflammatory
marker
levels
are
chronically
disrupted
from
childhood
and
additionally
impacted
on
by
further
stressors
across
the
life
course.
It
is
also
possible
that
it
is
the
last
links
in
the
chain
of
disadvantage
disadvantaged
material
circumstances,
reduced
educational
attainment,
psychological
distress,
and
high
BMI
which
are
associated
with
increased
low-
grade
inflammation.
Unfortunately
the
design
of
this
dataset
does
not
allow
us
to
disentangle
the
timing
of
change.
Additional
disadvantages
and
advantages
of
this
study
need
to
be
addressed.
We
have
no
measures
of
family
conflict
or
maltreatment
in
childhood
in
this
cohort
and
it
may
be
that
parental
separation
is,
in
part,
a
proxy
for
these
additional
childhood
adversities.
Previous
work
has
shown
that
parental
conflict
and
not
divorce
is
associated
with
poorer
health
outcomes
(Amato
et
al.,
1995;
Morrison
and
Coiro,
1999;
Cummings
and
Davies,
2002;
McIntosh,
2003).
Our
approach
to
missing
data
assumes
that
data
are
missing
at
random,
meaning
that
missingness
depends
upon
observed
character-
istics
of
participants.
This
is
a
reasonable
assumption
because
+
Models
PNEC-2426;
No.
of
Pages
9
Please
cite
this
article
in
press
as:
Lacey,
R.E.,
et
al.,
Parental
separation
in
childhood
and
adult
inflammation:
The
importance
of
material
and
psychosocial
pathways.
Psychoneuroendocrinology
(2013),
http://dx.doi.org/10.1016/j.psyneuen.2013.05.007
Parental
separation
and
adult
inflammation:
pathways
7
this
is
a
multidisciplinary
study
and
we
included
many
variables
in
the
imputation
model.
The
approach
represents
an
advantage
over
complete
case
analysis
which
assumes
data
to
be
completely
missing
at
random.
Another
limita-
tion
is
that
we
were
not
able
to
adequately
investigate
whether
the
timing
of
parental
separation
is
important
for
the
associations
investigated
as
this
information
is
not
available
beyond
the
broad
inter-sweep
periods
in
which
it
occurred
(i.e.
between
0
and
7
years,
8
and
11
years
and
12
and
16
years).
Using
these
broad
age
bands
we
found
that
the
relationship
between
parental
separation
and
later
CRP
did
not
vary
significantly
by
the
age
ranges
at
which
it
occur red.
However,
this
sub-analysis
is
clearly
limited
by
available
information.
This
study
also
has
a
number
of
strengths.
Firstly
all
measures
that
we
used
were
prospectively
collected,
thereby
limiting
recall
bias.
A
large
British
birth
cohort
is
used
which
is
representative
of,
and
therefore
generalisable
to,
people
who
grew
up
in
Great
Britain
of
a
similar
age.
This
study
had
a
large
sample
size
and
this
was
maximised
by
accounting
for
missing
data.
A
limitation
of
many
previous
studies
is
that
they
use
simple
multivariate
regression,
con-
trolling
for
mediators
of
interest.
In
this
study
we
employ
a
more
appropriate
analytical
strategy
which
allows
for
the
temporal
ordering
of
variables
and
explicit
modelling
of
direct
and
indirect
pathways
of
interest.
This
allowed
for
the
investigation
of
how
material
and
psychosocial
factors
interacted
across
the
life
course.
In
conclusion
our
study
highlights
the
importance
of
sup-
porting
families
undergoing
separation
in
order
to
minimise
the
long-term
impact
upon
children’s
health
and
disease
vulnerability
as
we
show
that
experiencing
parental
separa-
tion
sets
children
on
a
life
course
trajectory
of
increased
disadvantage
compared
to
their
peers
who
grow
up
in
‘intact’
families.
This
study
also
provides
evidence
for
the
targeting
of
interventions
in
order
to
reduce
the
long-term
impact
of
separation;
for
example,
pathways
through
education
appear
to
be
particularly
important
and
supporting
children
through
education
may
be
beneficial.
Role
of
funding
source
The
funding
bodies
were
not
involved
in
the
analysis,
inter-
pretation
of
data,
writing
of
the
report,
or
decision
to
submit
this
paper
for
publication.
Conflict
of
interest
All
authors
report
no
conflict
of
interest
in
relation
to
this
work.
Acknowledgements
Thanks
to
all
participants
of
the
National
Child
Development
Study
and
to
the
study
team.
The
data
are
supplied
by
the
Economic
and
Social
Data
Service
(ESDS).
Those
who
carried
out
the
original
collection
and
analysis
of
the
data
bear
no
responsibility
for
its
further
analysis
and
interpretation.
Dr
Lacey’s
and
Dr
McMunn’s
time
on
this
study
was
supported
by
the
European
Research
Council
(ERC-2011-StG_20101124).
Dr
Kumari’s
time
on
this
manuscript
was
partially
supported
by
the
Economic
and
Social
research
council
(RES-596-28-0001)
and
the
NHLBI
(HL36310).
References
Aassve,
A.,
Betti,
G.,
Mazzuco,
S.,
Mencarini,
L.,
2007.
Marital
disruption
and
economic
well-being:
a
comparative
analysis.
J.
Roy.
Stat.
Soc.
A
Stat.
170,
781—799.
Alley,
D.,
Seeman,
T. ,
Kim,
K.,
Karlamangla,
A.,
Hu,
P. ,
Crimmins,
E.,
2006.
Socioeconomic
status
and
C-reactive
protein
levels
in
the
US
population.
NHANES
IV.
Brain
Behav.
Immun.
20,
498—504.
Amato,
P. ,
Booth,
A.,
1996.
A
prospective
study
of
divorce
and
parent—child
relationships.
J.
Marriage
Fam.
58,
356—365.
Amato,
P. ,
Keith,
B.,
1991.
Parental
divorce
and
adult
well-being:
a
meta-analysis.
J.
Marriage
Fam.
53,
43—58.
Amato,
P. ,
Spencer
Loomis,
L.,
Booth,
A.,
1995.
Parental
divorce,
marital
conflict,
and
offspring
well-being
during
early
adulthood.
Soc.
Forces
73,
895—915.
Appleton,
A.,
Buka,
S.,
McCormick,
M.,
Koenen,
K.,
Loucks,
E.,
Kubzansky,
L.,
2012.
The
association
between
childhood
emo-
tional
functioning
and
adulthood
inflammation
is
modified
by
early-life
socioeconomic
status.
Health
Psychol.
31,
413—422.
Bassuk,
S.,
Rifai,
N.,
Ridker,
P. ,
2004.
High-sensitivity
c-reactive
protein:
clinical
importance.
Cardiology
29,
439—493.
Chapman,
D.,
Whitfield,
C.,
Felitti,
V. ,
Dube,
S.,
Edwards,
V. ,
Anda,
R.,
2004.
Adverse
childhood
experiences
and
the
risk
of
depres-
sive
disorders
in
adulthood.
J.
Affect
Disorders
82,
217—225.
Chen,
E.,
Miller,
G.,
Kobor,
M.,
Cole,
S.,
2011.
Maternal
warmth
buffers
the
effects
of
low
early-life
socioeconomic
status
on
pro-inflammatory
signaling
in
adulthood.
Mol.
Psychiatr.
16,
729—737.
Conger,
R.,
Conger,
K.,
Elder,
G.,
Lorenz,
F. ,
Simons,
R.,
Whitbeck,
L.,
1992.
A
family
process
model
of
economic
hardship
and
adjust-
ment
of
early
adolescent
boys.
Child
Dev.
63,
526—541.
Cooney,
T. ,
1994.
Young
adults’
relations
with
parents:
the
influence
of
recent
parental
divorce.
J.
Marriage
Fam.
56,
45—56.
Copeland,
W.,
Shanahan,
L.,
Worthman,
C.,
Angold,
A.,
Costello,
E.,
2012.
Generalized
anxiety
and
c-reactive
protein
levels:
a
pro-
spective,
longitudinal
analysis.
Psychol.
Med.
42,
2641—2650.
Cummings,
E.M.,
Davies,
P. ,
2002.
Effects
of
marital
conflict
on
children:
recent
advances
and
emerging
themes
in
process-ori-
ented
research.
J.
Child
Psychol.
Psych.
43,
31—63.
Danese,
A.,
Moffitt,
T. ,
Harrington,
H.,
Milne,
B.,
Polanczyk,
G.,
Pariante,
C.,
Poulton,
R.,
Caspi,
A.,
2009.
Adverse
childhood
experiences
and
adult
risk
factors
for
age-related
disease.
Arch.
Pediat.
Adol.
Med.
163,
1135—1143.
Danese,
A.,
Moffitt,
T. ,
Pariante,
C.,
Ambler,
A.,
Poulton,
R.,
Caspi,
A.,
2008.
Elevated
inflammation
levels
in
depressed
adults
with
a
history
of
childhood
maltreatment.
Arch.
Gen.
Psychiat.
65,
409—415.
Danese,
A.,
Pariante,
C.,
Caspi,
A.,
Taylor,
A.,
Poulton,
R.,
2007.
Childhood
maltreatment
predicts
adult
inflammation
in
a
life-
course
study.
Proc.
Natl.
Acad.
Sci.
U.S.A.
104,
1319—1324.
Danesh,
J.,
Wheeler,
J.,
Hirschfield,
G.,
Eda,
S.,
Eiriksdottir,
G.,
Rumley,
A.,
Lowe,
G.,
Pepys,
M.,
Gudnason,
V. ,
2004.
C-reactive
protein
and
other
circulating
markers
of
inflammation
in
the
prediction
of
coronary
heart
disease.
N.
Engl.
J.
Med.
350,
1387—1397.
Danner,
M.,
Kasl,
S.,
Abramson,
J.,
Vaccarino,
V. ,
2003.
Association
between
depression
and
elevated
c-reactive
protein.
Psychosom.
Med.
65,
347—356.
Das,
U.,
2001.
Is
obesity
an
inflammatory
condition?
Nutrition
17,
953—966.
Elliott,
B.J.,
Richards,
M.P.M.,
Warwick,
H.,
1993.
The
consequences
of
divorce
for
the
health
and
well-being
of
adults
and
children.
Final
report
for
health
promotion
research
trust
no.
2.
Centre
for
Family
Research,
Cambridge.
+
Models
PNEC-2426;
No.
of
Pages
9
Please
cite
this
article
in
press
as:
Lacey,
R.E.,
et
al.,
Parental
separation
in
childhood
and
adult
inflammation:
The
importance
of
material
and
psychosocial
pathways.
Psychoneuroendocrinology
(2013),
http://dx.doi.org/10.1016/j.psyneuen.2013.05.007
8
R.E.
Lacey
et
al.
Elliott,
J.,
Johnson,
J.,
Shepherd,
P. ,
2008.
User
guide
to
the
biomedical
survey
2002—2004
dataset.
Centre
for
Longitudinal
Studies,
Institute
of
Education,
London.
Ely,
M.,
Richards,
M.P.M.,
Wadsworth,
M.E.J.,
Elliott,
B.J.,
1999.
Secular
changes
in
the
association
of
parental
divorce
and
chil-
dren’s
educational
attainment–—evidence
from
three
British
birth
cohorts.
J.
Soc.
Policy
28,
437—455.
Fagundes,
C.,
Glaser,
R.,
Kielcolt-Glaser,
J.,
2013.
Stressful
early
life
experiences
and
immune
dysregulation
across
the
lifespan.
Brain
Behav.
Immun.
27,
8—12.
Hemingway,
H.,
Shipley,
M.,
Mullen,
M.,
Kumari,
M.,
Brunner,
E.,