Content uploaded by Paul Rohde
Author content
All content in this area was uploaded by Paul Rohde on Nov 05, 2014
Content may be subject to copyright.
Journal
of
Abnormal Psychology
1999, Vol. 108,
No. 4,
606-614
Copyright 1999
by the
American Psychological
Association,
Inc.
0021-843X/99/S3.00
Life
Events
and
Depression
in
Adolescence: Relationship Loss
as a
Prospective
Risk Factor
for
First Onset
of
Major
Depressive Disorder
Scott
M.
Monroe
University
of
Oregon
Paul
Rohde,
John
R.
Seeley,
and
Peter
M.
Lewinsohn
Oregon Research Institute
Although
stressful life events have consistently been linked
to the
onset
of
major depressive disorder
(MDD), most research
has not
distinguished
1st
episodes
from
recurrences.
In a
large
epidemiologic
sample
of
older adolescents
(N =
1,470)
assessed
at 2
time points,
the risk
conferred
by a
recent romantic
break-up
was
examined
as a
predictor
of 1st
onset versus recurrence
of
MDD. Results indicated
a
heightened
likelihood
of 1st
onset
of MDD
during adolescence
if a
recent break-up
had
been reported;
in
contrast,
a
recent break-up
did not
predict recurrence
of
depression. These results held
for
both genders
and
remained significant
after
controlling
for
gender. Additional analyses
to
determine
the
discriminant
validity
and
specificity
of
these
findings
strongly supported
the
recent break-up
as a
significant
risk
factor
for
a 1st
episode
of MDD
during adolescence. Implications
of
these
findings and
subsequent research
directions
are
discussed.
A
considerable body
of
evidence indicates
the
importance
of
severe
life
events
for the
onset
of
major depression (Brown
&
Harris, 1989a; Monroe
&
McQuaid,
1994).
The
bulk
of
empirical
work
on
this topic
has
involved adults, many
of
whom
have
had a
previous
history
of
depression. With
few
exceptions (e.g., Brown,
Harris,
&
Hepworth,
1994; Lewinsohn, Gotlib, Allen,
&
Seeley,
1996; Lewinsohn, Allen, Seeley,
&
Gotlib, 1999), most reports
have
not
examined separately
the
onset
of
initial episodes
of
depression
versus recurrences
in
relation
to
life
stress.
Yet
there
are
good reasons
to
suspect that
the
etiologic factors
associated
with
the
initial onset
of
major
depressive disorder (MDD)
differ
from
those associated with recurrences
of the
disorder (Hammen,
1991; Monroe
&
Depue, 1991; Post, 1992).
For
example, Post
(1992)
has
theorized that first-onset depressions
are
more strongly
linked
to
antecedent
life
stress. Once having incurred
the
stress
and
the
depression,
the
person
is
sensitized
to
future
stressors,
requir-
ing
a
lessened amount
of
stress
to
bring about another depression.
Also,
with
few
exceptions (e.g., Lewinsohn,
Hoberman,
&
Rosen-
baum,
1988), most reports have
not
studied
the
true incidence
of
depression prospectively.
Yet
there also
are
good reasons
to
sus-
pect
that retrospective
and
prospective research designs yield dif-
ferent
information
about
the
associations between
life
stress
and
depression (with greater validity accorded
the
prospective proce-
dures;
Hammen,
Mayol,
de
Mayo,
&
Marks, 1986; Monroe, 1982).
A
productive investigative strategy, therefore, would
be one in
which
first
episodes
and
recurrences could
be
studied separately
and
prospectively
in
relation
to
life
stress.
Scott
M.
Monroe, Psychology Department, University
of
Oregon; Paul
Rohde, John
R.
Seeley,
and
Peter
M.
Lewinsohn, Oregon Research Insti-
tute,
Eugene, Oregon.
Correspondence concerning this article should
be
addressed
to
Scott
M.
Monroe, Psychology Department,
Straub
Hall, University
of
Ore-
gon,
Eugene, Oregon
97403-1227.
Electronic
mail
may be
sent
to
smonroe@oregon.uoregon.edu.
In
terms
of
developmental considerations,
a
focus
on
adoles-
cence
may be
especially
useful
for
understanding
the
initial devel-
opment
of an
episode
of MDD and its
implications
for
subsequent
recurrence.
It has
become
increasingly
clear
from
recent research
that
many individuals experience
a first
episode
of
depression
during
adolescence,
and
experience subsequent recurrences during
this
stage
of
life
(Kovacs, 1989; Lewinsohn, Clarke, Seeley,
&
Rohde, 1994; Parry-Jones, 1989).
A
study
of
adolescents, there-
fore,
is
useful
to
obtain
an
adequate proportion
of
individuals
with
first
incidence
of
depression. Adolescence, too,
is a
life
stage
during which gender differences
in
depression prevalence emerge,
with
girls becoming those predominantly
afflicted
(Lewinsohn,
Hops,
Roberts, Seeley,
&
Andrews, 1993;
Nolen-Hoeksema
&
Girgus,
1994). Finally, early-onset depressions
may
forecast
a
more virulent lifetime course
of
depression
and
other psychopa-
thology
(Gilbert, 1992). Overall,
the
discovery
of
factors relating
to
depression onset during adolescence
may
ultimately
be
infor-
mative
about
the
developmental pathways through which depres-
sion initially emerges,
the
sources
of
gender differences
in
mor-
bidity,
and the
processes that
confer
vulnerability
to
future
depressive
recurrences.
There
are
several considerations necessary
to set the
stage
for a
methodologically sound investigation
of
life
stress
and its
impli-
cations
for
depression arising during adolescence.
Two
general
issues
are the
sampling
of
participants
and the
measurement
of
life
stress. With respect
to
sampling issues, studying major depression
once
it is
disaggregated into first-onset versus recurrence sub-
groups
requires relatively large numbers
of
participants
to
provide
sufficient
statistical power. Such work, too, would optimally
in-
volve
longitudinal assessments,
so as to
more
rigorously
investi-
gate
the
developmental antecedents
of the
disorder
uncontami-
nated
by
retrospective bias
or
current depressive state (Hammen
et
al.,
1986; Monroe, 1982). Finally,
a
focus
on
nonpatient popula-
tions
is
preferred, owing
to
concerns
associated
with
the
represen-
606
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
LIFE
EVENTS
AND
DEPRESSION
IN
ADOLESCENCE
607
tativeness
of
patient samples
and
possible biases associated with
treatment seeking (e.g., Goodman
et
al.,
1997; Monroe, Roberts,
Kupfer,
&
Frank,
1996).
With
respect
to the
assessment
of
life
stress,
two
major
ap-
proaches have dominated
the
literature (see Brown
&
Harris,
1978,
1989b;
Kessler,
1997; Monroe
&
McQuaid,
1994).
One
approach
uses self-report checklists
of
life
events, whereas
the
other—often
called investigator-based
procedures—uses
semistructured
inter-
views
and
independent ratings
for
defining
and
weighting
stres-
sors. Self-report checklists have been criticized
on
several grounds
(e.g., lack
of
consistent
definition
for
events
and
poor reliability;
Brown,
1974;
Dohrenwend,
Link, Kern,
Shrout,
&
Markowitz,
1990; McQuaid
et
al.,
1992),
as
have interview-based measures
(e.g.,
labor-intensive nature
of the
assessment,
and
possible
rater
biases; Kessler, 1997; Monroe
&
McQuaid, 1994).
For
some
purposes, though,
a
complementary
approach
using
features
of
both
systems might prove
useful
and
could
set the
stage
for
more
detailed
study
of
promising leads.
For
example,
Brugha
and
col-
leagues
(Brugha,
1985; Brugha
&
Cragg, 1990) developed
a
brief
checklist
of
events
that
had
been
found
by
prior research using
investigator-based methods
to be
severely stressful
and
strongly
predictive
of
depression onset. Adopting
a
somewhat similar strat-
egy, Kendler
and
colleagues (Kendler
et
al.,
1995;
Kendler,
Walters,
&
Kessler, 1997) examined
a
subset
of
self-reported
life
events
and
evaluated
the
degree
of
association with
depression
and
other clinical characteristics. Confining
the
stress inquiry
to
spe-
cific
self-report events that possess
a
high
likelihood
of
being
severe permits
the
application
of the
method
to
large samples
without
incurring
all of the
methodological liabilities commonly
associated with life-event
checklists.
Although
the
domain
of
experiences that constitute severe
life
events
is
broad,
a
majority
of
such events tends
to
reflect
the
concept
of
loss
in a
general sense
of the
term (Brown, 1974; Brown
&
Harris, 1978),
or
"social
exists" (Paykel,
1974).
These losses
typically
derive
from
the
ending
of key
interpersonal relationships,
either
a
break-up
with
a
significant
person
or the
death
of a
close
individual.
An
important determinant
of the
severity involves
the
degree
to
which
the
loss
is one of a
core relationship.
For
example,
confining
break-ups
to
those
of
romantic relationships
and
deaths
to
those
of
first-degree family members
of the
respondent
in-
creases
the
likelihood
of
ensuring
a
high
degree
of
severity
of the
event
and
corresponds
to
other work attempting
to
bridge
the gap
between self-report
and
investigator-based measures
of
life
stress
(Brugha,
1985; Kendler
et
al.,
1995).
In
terms
of
population
prevalence
rates, life
events
involving
romantic
upheavals
are
more common than parental deaths (espe-
cially
during adolescence; Kendler
et
al.,
1995). Given
a
sufficient
occurrence
of the
event
in a
large sample, focusing
on the
specific
event
of a
break-up
of a
romantic relationship
may
provide
a
useful
method
for
investigating possible influences
of
stress
on
depres-
sion incidence
and
recurrence.
In a
sense, this strategy also com-
plements
the
traditional life-events approach with
a
second
ap-
proach
commonly adopted
in the
stress literature:
focused
studies
on
single, fairly common
stressful
life
events (see Kessler, 1997).
Certainly relationship problems, separations,
and
divorce
are
fre-
quent
themes underlying ideas
of
stress
and
depression
in
adult-
hood (Beach,
Sandeen,
&
O'Leary,
1990).
It may be
that
during
adolescence,
when romantic interactional patterns develop
and lay
the
foundations
for
future
styles
of
initiating
and
terminating
intimate relations, relationship loss
may be
especially meaningful
psychologically
and
have important consequences
for
emotional
development
and
psychological
vulnerability.
Our
primary goal
for the
present study
was to
test
the
specific
life
event
of
romantic break-up
as a
prospective
risk
factor
for the
incidence
of
first
onset versus recurrence
of MDD in a
large
epidemiologic
sample
of
adolescent boys
and
girls.
On the
basis
of
prior literature,
we
hypothesized that
(a) a
recent break-up would
predict total incidence
of MDD
over
the
subsequent year (Lewin-
sohn
et
al.,
1988; Monroe, Bromet, Connell,
&
Steiner, 1986)
and
(b)
the
break-up would primarily predict
first-onset
cases
as op-
posed
to
recurrences (Post, 1992). Secondary goals
of the
research
were
to (c)
delineate
possible
gender
differences
in the
role
of
this
stressor
and in MDD
incidence
(first
onset
and
recurrence)
and (d)
establish
the
discriminant validity
of
breaking
up
with respect
to
other
predictors
of MDD
onset
and
other
diagnostic
outcomes.
Method
Participants
and
Procedure
Participants were randomly selected
in
three cohorts
from
nine senior
high schools representative
of
urban
and
rural districts
in
western Oregon.
Sampling
was
proportional
to
size
of the
school,
grade within
school,
and
gender within grade.
A total of
1,709 adolescents completed
the
initial
(Time
1,
orTl)
assessments (interviews
and
questionnaires) between 1987
and
1989,
with
an
overall participation rate
of 61%
(additional details
are
provided
in
Lewinsohn
et
al.,
1993). Roughly half (53%)
of the
Tl
sample
was
female, with
an
average
age of
16.6 years
(SD =
1.2).
The
represen-
tativeness
of the Tl
sample
was
assessed
by
using several approaches;
differences
between
the
sample
and the
larger population
and
between
participants
and
those
who
declined
to
participate were small.
No
differ-
ences were
found
between
the
sample
and
1980 census data (U.S. Bureau
of
the
Census, 1982)
on
gender, ethnic status,
or
parental education,
although
our
sample
had a
slightly higher proportion
of
two-parent
fami-
lies. Although decliners
had a
lower mean socioeconomic status (SES)
level
than
did
participants, both represented
the
middle class. Participants
and
decliners
did not
differ
on
gender
of
head
of
household, family size,
and
number
of
parents
in the
household. Adolescents were paid
for
their
participation,
and
written informed consent
was
obtained
from
both
the
participants
and
their legal guardians.
At
the
second assessment (Time
2, or
T2), 1,507 participants (88%)
returned
for a
readministration
of the
interview
and
questionnaire (Mean
T1-T2 interval
=
13.8 months,
SD =
2.3). Biases that might have emerged
due
to
attrition
in the
T1-T2 panel sample were examined
by
comparing
the
adolescents
who did not
participate
at T2
(n
=
202) with
the
1,507
participants
on
demographic characteristics
and
measures
of
psychopathol-
ogy. Small
but
statistically
significant
differences were
found.
Attrition
was
associated
with
lower parental
SES
(M
=
44.3,
SD =
11.7,
vs.
M
=
47.5,
SD =
11.3),
F(l,
1431)
=
11.6,
p <
.001; smaller household
number
(M
=
2.8,
SD =
1.4,
vs. M =
3.0,
SD =
1.5),
F(l,
1683)
=
4.0,
p
<
.05; being male (60%
of
decliners
vs. 46% of T2
participants),
^(1,
N
=
1,709)
=
13.2,
p <
.001;
and
lower educational level
of the
parents,
(M=
5.1,SD
= 1.1
vs.M
=
5.5,
SD =
1.1),
F(l, 1544)
=
14.1,
p<.
001.
Significandy
higher attrition rates also were noted
for
participants
who had
a
history
of
disruptive behavior disorders (17%
vs.
11%),
^(1,
N
=
1,709)
=
30.7,
p <
.01,
and for
young
men
with
a
history
of
substance
use
disorders (26%
vs.
14%),
^(1,
N =
819)
=
7.7,;?
<
.01. However,
the
nonparticipants
did not
differ
from
the
T1-T2 panel sample
on
other
measures
of
psychopathology,
such
as
number
of
suicide attempts, number
of
episodes
of
current
and
past disorders including depression,
and the
self-report depression measures,
or on
race
or
grade level.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
608
MONROE, ROHDE,
SEELEY,
AND
LEWINSOHN
For
methodological reasons
(see
Diagnostic Groups, below), individuals
who
were currently depressed
at the
Tl
assessment
(n = 37)
were
not
included
in the
present analyses. This provided
a
sample size
of
1,470
for
the
present report
of
persons
who
completed both
the Tl and T2
assess-
ments.
Approximately half
of
this
T1-T2
panel sample
was
female (53%),
with
an
average
Tl age of
16.5
years
(SD =
1.2).
A
total
of 9%
were
non-White,
74%
were
living
with
two
parents,
and 55%
were living
with
two
biological
parents. Parental education
(maximum
value
for
mother
or
father)
was 45%
possessing
an
academic
or
professional degree
and 55%
possessing
less
than
a
bachelor's degree.
Diagnostic Interview
Participants were interviewed with
a
version
of the
Schedule
for
Affec-
tive
Disorders
and
Schizophrenia
for
School-Age
Children (K-SADS) that
combined features
of the
Epidemiologic version (K-SADS-E;
Orvaschel,
Puig-Antich,
Chambers,
Tabrizi,
&
Johnson,
1982)
and the
Present
Epi-
sode version
(K-SADS-P)
and
included additional items
to
derive diag-
noses
of
past
and
current psychiatric disorders
as
outlined
in the
revised
third
edition
of the
Diagnostic
and
Statistical Manual
of
Mental
Disorders
(DSM-III-R;
American Psychiatric Association, 1987).
At
T2,
participants were diagnostically interviewed using
the
Longitu-
dinal
Interval
Follow-Up
Evaluation
(LIFE;
Keller
et
al.,
1987).
The
LIFE
procedure provides detailed information about
the
course
of
psychiatric
symptoms
and
disorders
since
the
initial K-SADS interview, with
rigorous
criteria
for
recovery
from
a
disorder (i.e., symptom
free
for 8 or
more
weeks). Therefore, diagnostic information
was
available regarding
the
occurrence, onset
age,
and
duration
of all
disorders before,
and
during
the
course
of, the
study.
Diagnostic
interviewers were
carefully
selected, trained,
and
supervised,
and
all
interviews were audio-
or
videotaped.
For
reliability purposes,
a
second interviewer reviewed
the
recordings
of
approximately
12% of the
interviews
and
made diagnoses.
Interrater
reliability
was
evaluated
by the
kappa statistic (Cohen,
1960).
With
six
exceptions (kappas
for
diagnoses
for
lifetime
dysthymia,
lifetime eating disorders, current
and
lifetime
anxiety
disorders, current
and
lifetime depression;
not
otherwise specified
[NOS] were
.58, .66, .60, .53, .71,
and
.67,
respectively),
all Tl
kappas
were equal
to or
greater than
.75.
Specifically, kappas
for the
following
classifications
were current MDD,
.85;
lifetime MDD,
.86;
current adjust-
ment
disorder,
not
available
(n = 3); and
lifetime adjustment disorder,
.76.
We
also computed kappas across
all
disorders
at T2
because
of the
low
frequency
of
many disorders.
The
kappas
for any
disorder versus
no
disorder
at T2 and
between
Tl and T2
were
.87 and
.72,
respectively.
Diagnostic Groups
For the
present
study,
the
major focus
was on the
implications
of a
recent
break-up
for the
subsequent onset
of
MDD, with
a
particular interest
in
differentiating
effects
of the
break-up
for a first
onset versus
a
recurrence.
Consequently,
the
sample
was
confined
to Tl
currently nondepressed
adolescents
who
completed
the T2
assessment
(N =
1,470).
Of
those
individuals
developing
MDD
between
Tl and T2 (n =
113; 8%),
cases
were
also
divided
into
those
with
a
history
of MDD
before
Tl
(the
recurrence group;
n — 43) and
those
without such
a
history
(the
first-onset
group;
n =
70).
Assessment Instruments
A
comprehensive assessment
of the
sample
was
performed
at
both
Tl
and
T2 to
obtain
detailed
information
on
social
and
psychological
charac-
teristics
(Lewinsohn
et
al.,
1993).
For the
purposes
of the
present study,
the
following
measures provided
the
bases
for
testing
the
primary
and
second-
ary
study hypotheses.
(For
specific details
of
psychometric properties
for
these measures
in the
current sample,
see
Lewinsohn,
Roberts,
et
al.,
1994.)
Demographic
Variables
For the
present
study,
the
variables
of
gender,
age,
household
compo-
sition
(number
of
parents
and
number
of
biological parents),
and
race were
analyzed
for the
adolescents;
the
variable
of
education level
was
analyzed
for
the
parents.
Life
Stress Measures
Total
major
life
events.
The
occurrence
of 14
negative
life
events
to
self
or
significant others (i.e., parent, sibling, other relative,
or
close
friend)
during
the
past year
was
assessed (selected
from
the
Schedule
of
Recent
Experiences, Holmes
&
Rahe,
1967;
Life
Events Schedule, Sandier
&
Block, 1979).
To
increase confidence
in the
validity
of
endorsing
the
events,
we
counted only those that happened
to the
respondent. Because
some items represented symptoms
of
psychopathology,
we did not in-
clude
3 of the 14
questions (i.e.,
"got
in a lot of
arguments
or
fights,"
"had
problems
with
drugs
or
alcohol," "tried
to
commit suicide")
as
stressful
events
to
self.
(One
additional item,
"died,"
cannot
be
endorsed
by the
participant.)
The
total major
life
events score used
for Tl
analyses
did not
include
the
event "romantic break-up"
(see
below).
Tl
values
of
this
measure were
M =
0.63,
SD =
0.96,
range
=
0-7;
T2
values were
M
=
0.76,
SD
=
1.04,
range
=
0-6.
Romantic
break-up.
The
list
of
major
life
events included
the
item
"Got divorced, separated,
or
broke
up
with
girlfriend/boyfriend."
There
was a
sufficient
incidence
of
this
event
in the
sample,
and it was an
event
that
possesses
a
relatively clear behavioral reference
in
terms
of
one's
social world (thereby likely minimizing misinterpretations
in
reporting
by
the
participant; Dohrenwend
et
al.,
1990). Given that very
few
(i.e.,
3) of
the
adolescents
were
married,
this item essentially referred
to
terminations
of
romantic relationships. Items such
as
this also
are
prominently repre-
sented
in
other checklists purporting
to
assess
life
events
with
a
high
depressogenic impact (e.g.,
Brugha,
1985;
Kendler
et
ai.,
1995).
Daily
Hassles. Daily Hassles
assessed
the
frequency
of
occurrence
of
unpleasant
social
and
nonsocial events
in the
past month
(20
items;
Unpleasant
Events Schedule; Lewinsohn,
Mermelstein,
Alexander,
&
MacPhillamy,
1985). Participants responded
to
each item
on a
scale rang-
ing
from
1
(Not
at all in the
last month)
to 5
(About
every
day).
Tl
values
of
this
measure were
M =
47.68,
SD =
12.14, range
=
20-97;
T2
values
were
M
=
45.45,
SD =
12.03, range
=
20-86.
Depressive
Symptoms
Center
for
Epidemiologic Studies-Depression Scale
(CES-D).
The
CES-D
(Radloff, 1977)
was
included
as a
self-report measure
of
depressive
symptomatology.
The
20-item
scale
assesses
the
occurrence
of
depressive
symptoms
during
the
past week.
The
scale
has
been shown
to
have
adequate psychometric properties
on an
adolescent sample (Roberts,
An-
drews, Lewinsohn,
&
Hops,
1990).
Beck Depression Inventory
(BDI).
The
BDI
(Beck, Steer,
&
Garbin,
1988),
another
self-report
measure
of
depression,
was
also included.
The
21-item
scale
assesses
the
presence
of
depressive symptoms
during
the
past
week
and has
good psychometric properties when used with adolescent
samples
(Teri,
1982;
Roberts
et
al.,
1990;
Roberts,
Lewinsohn,
&
Seeley,
1991).
Data Analyses
Preliminary analyses (analyses
of
variance
and
contingency table)
pro-
vided
basic descriptive data
and
associations between demographic char-
acteristics, depression measures,
and the
major study variables. Next,
we
used logistic regression procedures
to
address
the
primary study hypothe-
ses.
The
three main effects (break-up event, history
of
depression,
and
gender)
were entered into
the
equation
as a
block initially,
the
contribution
of the
three
resulting
2-way interactions were entered
as a
second
block,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
LIFE EVENTS
AND
DEPRESSION
IN
ADOLESCENCE
609
Table
1
Depression
Group
Differences
by
Demographic Variables
and
Depressive Symptoms
Depression group
Variable
%
(and
n):
Female
White
With
2
biological parents
in
home
With
parental education
a
BS
Mean
(and
SD)
for:
Age
Female
Male
Total
CES-D score
Female
Male
Total
BDI
score
Female
Male
Total
Total
53.5
91.4
54.6
45.0
16.54
16.50
16.52
17.60
15.34
16.54
7.26
6.01
6.68
sample
783
1,343
803
662
1.15
1.22
1.18
10.73
9.62
10.28
7.13
6.95
7.07
Never
depressed
(n
=
1152)
48.5a
92.2
57.6a
47.
5a
16.46
16.45
16.45
15.56
14.77
15.16
5.91
5.74
5.82
559
1062
664
547
1.17a
1.20
1.19a
9.77a
9.47
9.62a
6.20a
6.97a
6.60a
Formerly
depressed
(n
=
205)
70.2b
87.8
42.4b
35.6b
16.80
16.88
16.83
22.76
18.13
21.38
10.24
8.51
9.72
144
180
87
73
1.02b
1.24
1.09b
11.02b
9.14
10.69b
7.89b
7.10b
7.69b
First
(n =
61.4b
90.0
44.3b
37.1a.b
16.64
16.40
16.55
21.28
19.33
20.53
9.00
5.64
7.70
onset
=
70)
43
63
31
26
1.16ab
1.24
1.19a,b
11.02b
10.41
10.75b
6.36,,
4.81a,b
6.00a,b
Recurrence
(n
= 43)
86.0C
88.4
48.8ab
37.2a,b
16.72
17.82
16.88
24.03
24.83
24.14
14.08
9.17
13.40
37
38
21
16
1.14.*
1.13
1.89a,b
12.86b
13.82
12.82b
10.04C
6.74,b
9.73C
Test
statistic"
54.58**
4.93
20.10**
12.97*
3.93*
4.86*
7.27**
26.26**
6.03**
35.82**
30.24**
3.40**
33.86**
Note.
Means
and
percentages with different subscripts within
the
same
row
differ
significantly
at/7
<
.05.
BS =
bachelor
of
science; CES-D
=
Center
for
Epidemiological Studies-Depression Scale;
BDI =
Beck Depression Inventory.
a
Chi-square values
are
reported
for
dichotomous variables
(df
= 3); F
values
are
reported
for
numeric variables
(dfs
= 3,
1466).
*
p <
.01.
**p
<
.001.
and
finally
the
three-way interaction
was
tested. Subsequent specification
of
the
best-fitting model
was
dependent
on the
results
of the
initial model
(i.e.,
the
implications given
the
findings
for the
three-way interaction).
When multiple predictors within
a set or
interactions
are
tested (following
entry
of
main
effects),
overall significance
is
reported
with
the
improve-
ment
chi-square.
For
testing predictors within
a
significant
set,
or
testing
individual
predictors,
the
Wald
statistic
is
reported (along
with
the
asso-
ciated odds
ratio—i.e.,
the
probability
of
depression given
a
break-up
divided
by the
probability
of
depression
given
no
break-up).
To
describe
the
nature
of
significant
interaction
effects,
we
provided percentages based
on
simple contingency tables along with
the
respective
.likelihood
ratio
significance
tests
and
associated odds ratios. Finally,
we
performed
a
series
of
secondary analyses
to
clarify
the
relationship between
a
break-up
and
subsequent
first
onset
of MDD in
light
of
alternative predictors
of
depres-
sion
(general life event scores, daily hassles, depressive symptoms,
and T2
break-up
events)
and
other diagnostic outcomes (adjustment disorders,
anxiety
disorders, depression NOS).
Results
Preliminary
Analyses
Descriptive data
and
associations between demographic charac-
teristics,
clinical characteristics,
and
primary study variables
are
provided
in
Tables
1 and 2.
First-onset
cases,
recurrences,
and
nondepressed persons (broken down into those with
and
those
without
a
prior history
of MDD at
Tl;
i.e.,
the
never
depressed
and
the
formerly depressed groups) were compared
in
terms
of
demo-
graphic
and
clinical variables.
We
performed Omnibus tests
for the
four
groups; when significant differences were found,
we ran
contrasts
between
the
four
groups.
The
pertinent comparisons
for
present
purposes
are
those between first-onset
MDD and the
never
depressed
group,
and
recurrence
of MDD and the
formerly
de-
pressed
group.
If
these comparisons were significant,
we ran the
primary
and
secondary analyses once again
to
control
for
such
effects.
Overall, there
was
little
change owing
to
these influences.
(The
findings for
these
additional analyses
are
reported
as
foot-
notes
in the
relevant sections.)
Primary
Analyses
To
simultaneously test
the
impact
of a
romantic break-up,
the
history
of
major
depression
(i.e.,
first
onset
vs.
recurrence),
and
gender differences,
we ran the
logistic
regression
by
entering
the
three
main
effects first,
the
three 2-way interactions second,
and the
three-
way
interaction third
to
predict
the
dichotomous variable
of
depres-
sion
onset between
Tl and T2.
This model revealed
no
significant
three-way
interaction, improvement
^(1,
N =
1470)
=
0.293,
p >
.50.1
The
model
was
respecified
with
the
three main effects
in the first
step
and a
forward
entry procedure (based
on the
likelihood ratio
and
entry
criterion
of
p <
.05)
for the
three 2-way interaction terms.
For
this
model,
the
first
step
was
highly significant, model
^(3,
A'
=
1470)
=
50.46,
p <
.0001,
with
all
three main
effects
being
signifi-
cant
predictors
of
onset
of MDD
between
T1
and
T2
(break-up: Wald
statistic
=
10.13,p
<
.002, odds ratio [OR]
=
1.92,
95%
confidence
interval
[CI]
=
1.29-2.87;
sex: Wald statistic
=
6.52,
p
<
.01,
1
For the first
model,
the first
step
is
obviously identical
to the
respeci-
fied
model
as
described
next.
In
the
second step,
only
the
interaction
between
the
event
and the
history
of
depression entered into
the
equation
(Wald statistic
=
3.03,
p <
.09).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
610
MONROE, ROHDE, SEELEY,
AND
LEWINSOHN
Table
2
Break-Up
Group
Differences
by
Demographic
Variable
and
Depressive
Symptoms
Variable
%
(and
n):
Female
White
With
2
biological
parents
in
home
With
parental
education
a BS
Mean (and
SD)
for:
Age
Female
Male
Total
CES-D score
Female
Male
Total
BDI
score
Female
Male
Total
No
break-up
49.4 (524)
91.0 (965)
55.4 (587)
44.6 (473)
16.46 (1.17)
16.43
(1.23)
16.44 (1.20)
16.78 (10.30)
15.07 (9.58)
15.92 (9.97)
6.61 (6.77)
5.85
(6.99)
6.23
(6.89)
Recent break-up
63.2 (259)
92.2 (378)
52.7 (261)
46.1 (189)
16.71 (1.08)
16.75
(1.16)
16.72 (1.11)
19.25
(11.39)
16.30 (9.73)
18.16 (10.89)
8.58 (7.66)
6.57 (6.81)
7.84
(7.41)
Test
statistic"
22.41**
0.50
0.87
0.26
8.19*
8.25*
16.79**
9.28*
1.93
14.26**
13.46**
1.26
15.54**
Note.
For no
break-up,
« =
1,060;
for
recent
break-up,
n =
410.
BS =
bachelor
of
science;
CES-D
=
Center
for
Epidemiological
Studies-Depression
Scale;
BDI =
Beck Depression Inventory.
a
Chi-square values
are
reported
for
dichotomous
variables
(df
=
1);
F
values
are
reported
for
numeric variables
(rf/s
= 1,
1468).
*
p <
.01.
**p
<
.001.
OR
=
1.76,
95%
CI
=
1.14-2.71;
and
depression history:
Wald
statistic
=
21.73,
p <
.0001,
OR =
2.74,
95% CI =
1.79-4.18).
In
each instance,
the
direction
of the
main
effect
was as
predicted (i.e.,
the
break-up predicted greater likelihood
of
depression; girls were
more
likely
to
become depressed;
and a
history
of
depression pre-
dicted
a
higher probability
of
incurring
a new
episode
of
MDD).
In
the
second step,
the
interaction between
the
break-up
and
depression
history
was
statistically significant, improvement
x*(l,
N
-
1470)
=
3.85,
p <
.05. None
of the
other interaction terms
met
criteria
for
entry.2
Partitioning these
effects
for
descriptive
purposes,
we
noted
that
of the 70
onset cases,
32
(46%) reported
the
prior break-up
and
that
of the
1,152
nondepressed people (with
no
history
of
depression) between
Tl
and T2, 272
(24%) reported
a
recent break-up (likelihood ratio
=
15.25,
p <
.0001;
OR
=
2.72,
95% CI =
1.67-4.45).
(The
first
incidence rate
for
onset cases
was 11% for
those with
the
event
vs. 4% for
those
without.)
Of the 43
recurrences,
20
(47%) reported
the
event
compared with
86 of the 205
(42%)
who
were
not
depressed
between
Tl and T2
(and
who had a
history
of
prior depression;
likelihood ratio
=
.30,
p >
,50;
OR =
1.20,
95% CI =
0.62-
2.33).3
(The recurrence rate
was 19% for
those with
the
event
vs.
16%
for
those without.) Thus,
the risk
conferred
by the
report
of a
recent break-up
is
confined
to
those with first-onset depressions.
Secondary
Analyses
Given
the
strong prospective prediction
of first
onset
of MDD by a
past
break-up, several questions arise pertaining
to the
discriminant
validity
and
specificity
of the
effect.
Below
we
report
on
secondary
analyses designed
to
place
the
findings
from
the
primary analyses
in
a
more comprehensive interpretive context.
We
tested
the
specificity
of
the
prediction
afforded
by the
break-up
in
relation
to (a)
other
forms
of
stress (general
life
event totals
and
daily hassles),
(b)
prior
depressive symptoms,
(c)
other diagnostic outcomes,
(d)
later occur-
rences
of a
romantic break-up,
and (e)
gender.4
2
Controlling
for age in
this analysis (see Table
2)
slightly altered
the
interaction results, improvement
^(1,
N =
1470)
=
3.77,
p <
.052.
Controlling
for age and
number
of
biological
parents
in the
home
(see
Table
1)
again slightly
lessened
the
significance
of the
interaction,
im-
provement
/(I,
N =
1470)
=
3.58,
p
<
.059.
3
These
effects,
too, tended
to
hold across gender
(as one
would expect
from
the
lack
of a
significant Gender
X
Event interaction).
Of the 43
first-onset
cases
for
girls,
20
(47%) reported
the
prior break-up compared
with
154
(28%)
of
the 559 who
were
not
depressed
between
Tl and
T2
(likelihood
ra-
tio
=
6.44,
p <
.02;
OR =
2.27).
Of the 37
female recurrences,
17
(46%)
reported
the
event, compared
with
68
(47%)
of the 144 who
were
not
depressed between
Tl and
T2
and who had a
history
of
prior depression
(likelihood ratio
=
.02,
p >
.80;
OR =
.95).
Of the 27 first-onset
cases
for
boys,
12
(44%) reported
the
prior break-up, compared with
118
(20%)
of the
593
who
were
not
depressed between
Tl and
T2
(likelihood
ratio
=
7.87,
p <
.01;
OR =
3.22).
Of the 6
male recurrences,
3
(50%) reported
the
event
compared with
18
(30%)
of the 61 who
were
not
depressed between
Tl and
T2
and who had a
history
of
prior depression (likelihood ratio
=
1.00,
p >
.30;
OR
=
2.39). Also,
it is
noteworthy that
the
interaction
of the
break-up event
with
gender
was not
statistically significant.
4
As
noted previously,
older
adolescents were
found
to
have
a
greater
likelihood
of
reporting
a
romantic break-up (see Table
2), and the
never
depressed
group
differed
significantly
from
the
first-onset group with
regard
to the
percentage with
two
biological parents
at
home (see Table
1).
All of the
secondary analyses were
run
controlling simultaneously
for
these
two
variables.
In no
instance
did
their inclusion
as
covariates
alter
the
basic
findings.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
LIFE
EVENTS
AND
DEPRESSION
IN
ADOLESCENCE
611
Specificity
of
Break-Up
Versus
General
Stress
In
previous reports, both major
life
event totals
and
daily hassles
were
found
to
predict
MDD
(see
Lewinsohn,
Roberts,
et
al.,
1994).
To
conservatively test
the
primary hypothesis
of a
recent break-up
as
an
independent predictor
of
total
MDD
incidence between
Tl
and
T2, we ran a
logistic regression forcing
Tl and T2
total
life
event
scores
and Tl and T2
daily
hassles
scores (along with
gender)
in the
first step
and
relationship
loss
score
in the
second
step.
As a
set,
the
four
general stress measures
and
gender
are
highly
significant predictors
of
first
onset
of
major
depression
between
Tl and T2,
model
^(5,
N =
1202)
=
26.14,
p
<
.0001.
In
terms
of the
unique predictive significance,
T2
total events
tended
to
predict such onsets
(Wald
statistic
=
3.ll,p
<
.08).
T2
daily
hassles
significantly
predicted this outcome (Wald statis-
tic
=
4.10,
p <
.05),
as did
gender (Wald statistic
=
6.42,
p <
.02). Most important,
in the
second step
the Tl
break-up event
again
significantly increased
the
prediction, model improvement
1,
N =
1202)
=
8.21,
p <
.005.
Specificity
of
Break-Up
and
Prior
Depressive
Symptoms
The
presence
of
subclinical depressive symptoms might predis-
pose
to the
life
event occurring, mediate vulnerability
to the
later
episode,
or
both. Simultaneous entry
of the
break-up event,
Tl
BDI
symptoms,
and Tl
CES-D symptoms
in the
logistic regression
indicated that
the
break-up
and
CES-D independently predicted
subsequent
first
onset
of
depression (break-up: Wald statis-
tic
=
14.47,
p
<
.0001;
CES-D symptoms: Wald statistic
=
13.02,
p
<
.0003).
BDI
symptoms
did not
predict
in
this model (Wald
statistic
=
0.45,
p >
.50). Thus,
the
break-up event remains
a
strong
predictor independent
of
depressive symptoms
for a
subse-
quent
first
episode
of
depression.
Specificity
of
Break-
Up
and
Differential
Diagnosis
The
incidence
of new
cases
of
adjustment
disorder
(n
= 37) and
depression
NOS (n =
1
1
8)
between
Tl
and T2
were also examined
in
relation
to the
break-up
event reported
at
Tl.
In
each analysis,
cases
without
a
history
of the
respective disorder were included
to
test
the
relation
of a
break-up with
first
lifetime episode
of the
particular disorder.
In no
instance
did the
event approach
signifi-
cance
in
predicting
the new
onset
of
disorder
(adjustment
disorder:
likelihood ratio
=
0.29,
p >
.50,
OR =
1.22,
CI
=
.60-2.44;
depression NOS: likelihood ratio
=
0.32,
p >
.50,
OR =
1.31,
CI
=
.74-1.73).
Specificity
of
Break-Up
at Tl
Versus
T2
One
pathway through which
the
break-up
at Tl
might
result
in
MDD
could
be
through
a
subsequent break-up.
A
logistic regres-
sion
analysis
was
performed
in
which break-up events
at Tl and
T2
were entered into
the
equation simultaneously
to
predict
first
onset
of
depression.
The Tl
break-up remained
a
highly
significant
predictor
of
first-onset depression occurring between
Tl and T2
(Wald
statistic
=
12.40,
p <
.0005).
The
reporting
of a
break-up
at
T2
evidenced
a
trend
for
predicting
first-onset
depression
be-
tween
Tl and T2
(Wald statistic
=
3.75,
p =
.053).
The
interaction
between
Tl and T2
break-up events
was not
significant
(p >
.40).
(The test
with
T2
break-up events
is
conservative because
the
design does
not
assure that
all T2
break-ups predated
depression
onset.)
Specificity
of
Break-Up
and
Gender
As
noted previously, gender
was a
significant predictor
of
total
depression incidence between
Tl and T2, as
well
as of first-onset
depressions between
Tl and T2.
Gender
was
also
related
to the
likelihood
of
reporting
a
romantic break-up
(with
girls being more
likely).
A
logistic regression simultaneously entering
the Tl
break-up
event
and
gender
to
predict
first-onset
depressions indi-
cated
that
the
break-up remained
a
significant
predictor
of first-
onset
depression (Wald statistic
=
14.59,
p
<
.0001);
in
contrast,
gender only evidenced
a
statistical trend (Wald statistic
=
2.98,
p
<
.09). Consequently, break-up
not
only remains
an
independent
predictor
of
later
first-onset
depression,
it
also partially explains
the
gender differences
in
likelihood
of
becoming depressed
for the
first
time.
Discussion
The
severing
of a
romantic relationship
in the
past year
may set
the
stage,
or at
least
one
stage,
for
developing
MDD for the
first
time during
adolescence.
As
predicted,
the
prospective association
afforded
by a
recent romantic break-up
was
specifically
for a first
lifetime
episode
of
MDD,
and did not
predict recurrences. Inter-
estingly
and
importantly,
the
relationship
held
when numerous
control analyses were performed
to
test
for
alternative predictors
and
for
diagnostic specificity
of the
effects.
All in
all,
the findings
support
the
primary hypotheses, with
the
documented relationships
being both strong
and
robust.
Almost half (46%)
of the first
onsets
of MDD
were preceded
by
a
break-up
in the
previous year. This
is in
contrast
to an
incidence
of
about
one
quarter (24%)
of
recent break-ups
for
adolescents
who
did not
become depressed. Clearly such events
are not
infre-
quent
occurrences
in the
adolescent population
at
large,
and
many
who
report such experiences
do not go on to
develop MDD.
Yet of
the
participants
who
were newly depressed between
Tl
and T2, a
considerable number reported
a
break-up
in the
past year.
The OR
of
2.72 (with
95% CI =
1.67-4.45)
certainly supports
the
idea that
such
recent experiences
are an
important arena
for
future
inquiry
into
the
genesis
of MDD
episodes
during adolescence.
These
initial
findings lay the
groundwork
for a
more detailed
inquiry
into
the
qualities associated
with
reporting
a
recent
break-up
that lead
to
depression.
At
present,
we do not
know
precisely what
is
reflected
by the
endorsement
of the
item. There
is
likely
to be
heterogeneity
in
terms
of
what
actually happened,
the
contextual factors that moderate stress impact (Brown
&
Har-
ris,
1978),
and the
sequelae
that ensue
after
the
event.
For
example,
some
of the
experiences reported
may fit the
spirit
of the
event
(i.e., terminations
of
important romantic relationships), whereas
others
may not
(i.e.,
fleeting
crushes
or
misperceptions
of the
nature
of the
relationship).
In
terms
of
other
factors
that moderate
the
seriousness
of the
event
and its
aftermath,
such considerations
as how the
event came about
and was
resolved, whether
confidants
were available
to
provide support, whether subsequent "neutraliz-
ing" events occurred (i.e., events that compensate
for or
nullify
the
impact
of the
prior event; Tennant, Bebbington,
&
Hurry,
1981),
and
the
implications
of
different
coping strategies
all
represent
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
612
MONROE,
ROHDE,
SEELEY,
AND
LEWINSOHN
important directions
for
future
work.
A
useful
approach
for
future
study
would
be to
provide better guidelines
and
criteria
to
define
the
break-up event
and to
integrate these data with
the
other factors
that
moderate
the
immediate
or
subsequent impact
of the
stressor.
By
adapting
the
methods
of
investigator-based life-stress assess-
ment
procedures
(e.g.,
Brown
&
Harris, 1989b), researchers could
attain
a
better understanding
of the
nature
of the
experiences
involved.
Overall,
the
present
findings
provide strong empirical
support
for
moving
to
more detailed
and rigorous
stress assessment
of
specific individual
life
events such
as
recent relationship
loss.
Future researchers must
ask how a
break-up
may lay the
foun-
dation
for
developing vulnerability
and
ultimate morbidity. This
is
particularly
true given that
the
association
was a
prospective
one
that
operated over
a
relatively prolonged interval. Indeed,
for the
Tl
assessment there
was no
concurrent association between cur-
rent
depression
and the
reporting
of a
recent
break-up (which held
for
total current depressions
as
well
as for first-onset
current
depressions).
The
median onset
of
MDD
was 8
months following
Tl. In
this regard,
the
present findings contrast with other recent
work
that suggests
a
more immediate impact
of
serious
life
stres-
sors, wherein
the
relationship between
the
event
and
depression
onset
was
essentially confined
to the
month
of the
event (Kendler
et
al.,
1995). Other investigators, too, have noted that when
de-
pression occurs,
it
tends
to do so
rather quickly
following
a
severe
event
(Brown
&
Harris, 1989a). Thus,
in
accord
with
the
present
findings,
it
might
be
useful
to
think
of a
recent romantic break-up
as a
"marker"
or
vulnerability index
for
future
depression onset
in
adolescence
(as
opposed
to a
"maker"
of
depression),
and to
entertain
theoretical possibilities that account
for
such results.
There
is a
spectrum
of
theoretical possibilities worth consider-
ing
in a
model
for the
development
of MDD
that
is
informed
by a
strong prospective association with
a
recent break-up. First,
and
most
simplistically
and
parsimoniously,
a
romantic break-up
may
be an
independent
risk
factor
for
depression. That
is,
although
other forms
of
vulnerability
are
necessary
to
translate
the
experi-
ence into
MDD
(for
not
everyone with
the
event becomes
de-
pressed; e.g.,
an
independent diathesis), these other forms
of
vulnerability
are
uncorrelated
with
the
likelihood
of
experiencing
the
event. Second,
a
romantic break-up
may be due to a
vulner-
ability
that accounts
for
both
the
increased likelihood
of the
break-up
and
depression.
Yet it is
through
the
break-up that
the
diathesis
is
activated, transforming
the
potential
of
predisposition
to
the
presence
of
MDD.
For
example, interpersonal dependency
may
heighten
the
probability that
one may be
rejected
in an
intimate
relationship
and
that
one may
become depressed under
such
circumstances (Monroe
&
Simons, 1991).
In
this model,
the
vulnerability
factor
creates
the
conditions that permit
its
expres-
sion
as
depression. Such thinking appears
to be in
accord with
some psychoanalytic perspectives
on
dependency
and the
devel-
opment
of
depression (e.g.,
Chodhoff,
1972). Alternatively, there
may
be an
underlying vulnerability
that
accounts
for the
height-
ened likelihood
of
both
the
break-up
and the
depression. This latter
model revolves around
the
idea that
a
vulnerability
can
lead
to the
event happening,
but the
event
is not
necessary
for
precipitating
the
depression.
For
example,
an
instability
in
behavioral regulation
might
reflect
a
vulnerability
to
depression
and
create
interpersonal
contexts that
are
prone
to
strife
and
loss.
Yet
this vulnerability
factor
alone accounts
for the
depression, with
the
events appearing
as
essentially artifacts.
Our findings
favor
thinking along
the
lines
of
the
first
two
models,
for the final one
does
not
easily accom-
modate
the
prospective (and
not
concurrent) relations between
a
recent break-up
and
future
onset
of a first
episode
of
MDD.
It
is
also possible that
the
life
event sets
in
motion psychological
or
social processes that heighten
the
likelihood
of
incurring
de-
pression
at a
later point
in
time.
For
example, termination
of an
early
significant romantic relationship
(or a
particular type
of
termination; e.g., hostile
or
callous rejection)
may
reverberate
in
negative
ways through later relationships
or
impair
one's
abilities
to
select
or
maintain
an
appropriate partner
in the
future.
The
severing
of
important social ties also
has
strong implications
in
evolutionary theories
of
depression (e.g., Gilbert, 1992).
It may
also
be
that
the
break-up disrupts important behavior patterns;
when
this disruption leads
to a
reduction
of
positive reinforcement,
depression follows
(Lewinsohn,
Hoberman,
Teri,
&
Hautzinger,
1985).
Once
again,
given
the lag
between
the
occurrence
of the
event
and the
timing
of MDD
onset,
it
appears
that
other inter-
vening processes
are
necessary
to
translate
the
event-induced
po-
tential
for
depression into clinical actuality. With methodological
refinement
and a
better grasp
of the
nature
of the
experiences
involved, along with
a
conceptual broadening
of the
stress process
(Pearlin,
Lieberman,
Menaghan,
&
Mullan,
1981),
a
better
foun-
dation will
be
laid
for
understanding what about such experiences
and
their
aftermath
might
set the
stage
for
developing
MDD for the
first
time.
It is
interesting
to
contrast
the
findings
for first-onset
depres-
sions with those
for
recurrence. Adolescents with
a
past history
of
depression have relatively high rates
of a
recent break-up irrespec-
tive
of
future
depression status. Thus,
47% of
those with
a
recur-
rence reported such events compared with
42% of
formerly
de-
pressed adolescents
who did not
experience
a
recurrence. These
findings
suggest
that
the
lack
of
association between
a
break-up
and a
recurrence cannot
be
explained solely
on the
basis
of low
statistical power.
It is
possible,
however, that
the
smaller sample
of
recurrences
(n
= 43)
compared with
first
episodes
(n
= 70)
might
yield
a
less
representative sample
of the
recurrence population.
Bearing such qualifications
in
mind,
the
present results
still
sug-
gest that
the
likelihood
of
having
a
recent break-up
is
raised
for all
adolescents
who
have
had a
prior experience with MDD,
yet the
event
does
not
appear
to
play
a
role
in
bringing about recurrence
of
depression.
Of
course, with
a
more detailed assessment
of the
event,
a
possible role
for
recurrence might emerge (e.g., only loss
of
more serious relationships might predict recurrence).
It is
also
possible that
a
break-up
of a
romantic relationship predicts recur-
rence
only
for the
formerly
depressed adolescents
who had
never
experienced
a
serious break-up before. Noteworthy, though,
is the
finding
that
the
rate
of
recent relationship loss
is
raised
for
those
with
prior depressions, supporting
the
idea that previously
de-
pressed people
are at risk for
incurring,
or
generating, higher levels
of
stress
(Hammen,
1991).
Overall, these results
are
also
in
accord
with
Post's
(1992)
ideas
on the
role
of
stress
in first and
later
episodes
of
depression.
Not
only
was the
event
of a
recent break-up
a
strong prospective
predictor
of first-onset
depressive
episodes,
it was a
strong inde-
pendent
predictor when
several
other factors were taken into
consideration. Controlling
for
other measures
of
life
stress
(at
either
Tl or T2) or for
prior depressive symptoms
did not
alter
the
basic predictive scheme.
In
addition, this prediction appeared
to be
diagnostically
specific
to
MDD. This points
to the
unique
impor-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
LIFE EVENTS
AND
DEPRESSION
IN
ADOLESCENCE
613
tance
of a
relationship break-up
for
first-onset depressions during
adolescence,
at
least
in a
statistical
sense.
There
are a few
matters
to
consider
in
relation
to
these
results. First,
it may be
that other
forms
of
stress
can
also bring about depression during adolescence.
In
the
present sample, though,
our
assessment
of
other events
was
probably
less
valid
or the
frequency
of
such events
was
low.
For
example, deaths
of first-degree
family members were relatively
infrequent
occurrences
in the
present sample. Being
a
victim
of
violence,
too,
appeared
to
possibly
be an
important event
but was
also relatively infrequent
(and
rather open
to
wide variability
in
interpretation;
Dohrenwend
et
al.,
1990).
Also, functional impair-
ment
from
disease
was
relatively infrequent
but has
previously
been shown
to put
adolescents
at risk for
subsequent depression
in
the
present sample (Lewinsohn,
Seeley,
Hibbard,
Rohde,
&
Sack,
1996).
Second,
it
might
be
that much
of the
predictive variance
associated with traditional life-event checklists
in
depression
re-
search
is
"taken
up" by
events involving separations
and
break-
ups. Such events commonly impact other areas
of
functioning
and
thereby initiate
a
sequence
of
experiences that give
rise to
elevated
life-event
totals
from
checklist measures
(e.g.,
"serious
arguments"
lead
to
"break-up,"
which leads
to
"dating
events"
and so
on).
Finally,
the
potency
of a
break-up event might
be a
unique feature
for
first-onset
MDD
during adolescence. Clearly this time period
of
life
is a
challenging
one for the
maturing individual,
and
interpersonal relations constitute
one of the
major
arenas
in
which
development takes place.
It is
conceivable that
the
manner
in
which
such break-ups come about
or are
resolved—particularly,
perhaps,
if it is the
first
serious
relationship—has
enduring
con-
sequences
for
future
relationships
and
future
depressions.
This study possesses several strengths.
The
findings
are
based
on
a
large sample
and
prospective design,
the
diagnoses were
based
on
clinical criteria
(as
opposed
to
self-report
of
depressive
symptoms),
and the
distinction
is
drawn between
first
onset
of
MDD
versus recurrence
of the
disorder.
On the
other hand,
one
might question
the
reliability
and
validity
of
diagnoses over
pro-
longed periods
of
time.
For
example,
in the
adult literature similar
questions have been raised (e.g., Bromet, Dunn, Connell,
Dew,
&
Schulberg,
1986).
Although adolescents have
a
shorter period
of
retrospective recall than adults (which presumably would increase
reliability),
the
general question
of
reliability
of
lifetime diagnoses
is
an
important
one and
represents
a
potential limitation
of the
present work. Another limitation,
as
noted previously, pertains
to
the
measurement
of
life
stress.
These
findings, though, provide
a
strong empirical rationale
for
studying
the
role
of
life stress
in
adolescence
and MDD
with more
rigorous
life-stress methodolo-
gies.
To our
knowledge,
the
present study represents
the
first
time
the
impact
of a
major
life
event
has
been evaluated
in
such
a
manner with adolescents
for the
prediction
of
first
onset
and
recurrence
of
MDD.
Finally,
the
results
are
especially interesting with regard
to
gender issues
in the
development
of
MDD.
The
prediction
of
first-onset
depression
afforded
by a
recent break-up operated
sim-
ilarly
across
the
sexes (with girls being
the
more vulnerable
to
first-onset
depressions
and
recurrences).
It is
noteworthy, though,
that
girls overall
had
both
a
higher percentage reporting
the
event
(31%
vs.
22%)
and a
higher incidence
of
first-onset depression
(7%
vs.
4%).
Most intriguing
is the finding
that
the
event
of a
recent break-up predicts
MDD
above
and
beyond gender
and
that
the
association
of MDD
with gender
is
attenuated
by
inclusion
of
the
break-up event variable
in the
model. These results raise
several interesting possibilities. There
may be a
higher
frequency
of
the
basic event
for
girls. Although this
at
first
seems counter-
intuitive (i.e.,
the
event
was for
romantic relationships,
the
vast
majority
of
which
are
heterosexual,
so
there should
be
comparable
total
of
such events
for
boys
and for
girls), there
are a
couple
of
possible explanations.
For
example, there
may be
more multiple
break-ups
for
boys,
or
girls
may be
dating older boys
not in the
sample.
On the
other hand, such
data'may
point
to
important
gender differences
in
forming romantic attachments
or
perceptual
processes used
in
defining important relationships
or
experiences
as
break-ups (Monroe
&
Simons,
1991;
Simons, Angell, Monroe,
&
Thase,
1993).
In
other words,
different
criteria,
or
different
thresholds,
for
defining relationships
or
break-ups
may be
used
by
the
different
genders. Once again, incorporating more detailed
assessments
of the
adolescents' lives
and
experiences will
be
necessary
to
develop
a
better understanding
of the
aspects associ-
ated
with
a
recent break-up that lead
to
later depression,
and
whether
or not
this association contributes
to the
emergence
of
gender differences
in
depression during this period
of the
life
span.
References
American Psychiatric Association. (1987). Diagnostic
and
statistical man-
ual
of
mental disorders
(3rd
ed.,
rev.).
Washington,
DC:
Author.
Beach,
S. R. H.,
Sandeen,
E. E., &
O'Leary,
K. D.
(1990). Depression
in
marriage.
New
York:
Guilford
Press.
Beck,
A. T.,
Steer,
R.
A.,
&
Garbin,
M. G.
(1988). Psychometric properties
of
the
Beck Depression Inventory:
Twenty-five
years
of
evaluation.
Clinical
Psychology
Review,
8,
77-100.
Bromet,
E. J.,
Dunn,
L. O.,
Connell,
M.
M.,
Dew,
M.
A.,
&
Schulberg,
H.
C.
(1986). Long-term reliability
of
diagnosing lifetime
major
depres-
sion
in a
community sample. Archives
of
General
Psychiatry,
43,
435-
440.
Brown,
G. W.
(1974). Meaning, measurement,
and
stress
of
life
events.
In
B. S.
Dohrenwend
& B. P.
Dohrenwend
(Eds.),
Stressful
life
events:
Their
nature
and
effects
(pp.
217-243).
New
York: Wiley.
Brown,
G.
W.,
&
Harris,
T. O.
(1978). Social origins
of
depression:
A
study
of
psychiatric disorder
in
women.
New
York:
Free
Press.
Brown,
G.
W.,
&
Harris,
T. O.
(1989a). Depression.
In G. W.
Brown
&
T. O.
Harris
(Eds.),
Life
events
and
illness
(pp. 49-93).
London: Guil-
ford
Press.
Brown,
G.
W.,
&
Harris,
T. O.
(1989b).
Life
events
and
illness. London:
Guilford
Press.
Brown,
G.
W.,
Harris,
T.
O.,
&
Hepworth,
C.
(1994).
Life
events
and
endogenous depression:
A
puzzle reexamined. Archives
of
General
Psychiatry,
51,
525-534.
Brugha,
T.
(1985).
The
List
of
Threatening Experiences:
A
subset
of 12
life
event
categories
with
considerable
long-term contextual threat. Psycho-
logical
Medicine,
15,
189-194.
Brugha,
T.
S.,
&
Cragg,
D.
(1990).
The
List
of
Threatening Experiences:
The
reliability
and
validity
of a
brief
life
events questionnaire. Acta
Psychiatrica
Scandinavica,
82,
77-81.
Chodhoff,
P.
(1972).
The
depressive personality. Archives
of
General
Psychiatry,
27,
666-673.
Cohen,
J. A.
(1960).
A
coefficient
of
agreement
for
nominal scales.
Educational
and
Psychological Measurement,
20,
37-46.
Dohrenwend,
B. P.,
Link,
B.
G.,
Kern,
R.,
Shrout,
P.
E.,
&
Markowitz,
J.
(1990).
Measuring
life
events:
The
problem
of
variability
within
event
categories. Stress Medicine,
6,
179-188.
Gilbert,
P.
(1992). Depression:
The
evolution
of
powerlessness.
New
York:
Guilford
Press.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
614
MONROE, ROHDE, SEELEY,
AND
LEWINSOHN
Goodman,
S. H.,
Lahey,
B.
B.,
Fielding,
B.,
Dulcan,
M.,
Narrow,
W., &
Regier,
D.
(1997).
Representativeness
of
clinical samples
of
youths with
mental
disorders:
A
preliminary population-based
study.
Journal
of
Abnormal Psychology, 106, 3-14.
Hammen,
C.
(1991). Generation
of
stress
in the
course
of
unipolar depres-
sion.
Journal
of
Abnormal Psychology, 100,
555-561.
Hammen,
C.,
Mayol,
A.,
de
Mayo,
R., &
Marks,
T.
(1986). Initial symptom
levels
and the
life-event-depression relationship. Journal
of
Abnormal
Psychology,
95,
114-122.
Holmes,
T.
H.,
&
Rahe,
R. H.
(1967).
The
Social
Readjustment Rating
Scale.
Journal
of
Psychosomatic Research,
II,
213-218.
Keller,
M.
B.,
Lavori,
P.
W.,
Friedman,
B.,
Nielsen,
E.,
Endicott,
J., &
McDonald-Scott,
P. A.
(1987).
The
longitudinal interval follow-up eval-
uation.
Archives
of
General Psychiatry,
44,
540-548.
Kendler,
K.
S.,
Kessler,
R.
C.,
Walters,
E.
E.,
MacLean,
C.
J.,
Sham,
P.
C.,
Neale,
M.
C.,
Heath,
A.
C.,
&
Eaves,
L. J.
(1995). Stressful life events,
genetic liability
and
onset
of an
episode
of
major
depression
in
women.
American Journal
of
Psychiatry, 152,
833-842.
Kendler,
K.
S.,
Walters,
E.
E.,
&
Kessler,
R. C.
(1997).
The
prediction
of
length
of
major
depressive
episodes:
Results
from
an
epidemiologic
sample
of
female twins. Psychological Medicine,
27,
107-117.
Kessler,
R. C.
(1997).
The
effects
of
stressful
life
events
on
depression.
Annual Review
of
Psychology,
48,
191-214.
Kovacs,
M.
(1989). Depression
in
children
and
adolescents. American
Psychologist,
44,
209-215.
Lewinsohn,
P.
M.,
Allen,
N.
B.,
Seeley,
J.
R.,
&
Gotlib,
I. H.
(1999). First
onset
versus recurrence
of
depression:
Differential
processes
of
psycho-
social
risk. Journal
of
Abnormal Psychology, 108,
483-489.
Lewinsohn,
P.
M.,
Clarke,
G.
N.,
Seeley,
J.
R.,
&
Rohde,
P.
(1994). Major
depression
in
community adolescents:
Age at
onset,
episode
duration,
and
time
to
recurrence. Journal
of
the
American Academy
of
Child
and
Adolescent Psychiatry,
33,
809-818.
Lewinsohn,
P.
M.,
Gotlib,
I.
H.,
Allen,
N.
B.,
&
Seeley,
J. R.
(1996,
September).
The
specificity
of
antecedents
of first
episodes
of
depression
during
adolescence.
Paper
presented
at the
annual meeting
of the
Society
for
Research
in
Psychopathology,
Atlanta, Georgia.
Lewinsohn,
P.,
Hoberman,
H.
M.,
&
Rosenbaum,
M.
(1988).
A
prospective
study
of
risk factors
for
unipolar depression. Journal
of
Abnormal
Psychology,
97,
251-264.
Lewinsohn,
P.
M.,
Hoberman,
H.,
Teri,
L.,
&
Hautzinger,
M.
(1985).
An
integrative
theory
of
depression.
In I. S.
Reiss
& R.
Bootzin
(Eds.),
Theoretical
issues
in
behavior
therapy
(pp.
331-359).
New
York: Aca-
demic
Press.
Lewinsohn,
P.
M.,
Hops,
H.,
Roberts,
R.
E.,
Seeley,
J.
R.,
&
Andrews,
J. A.
(1993). Adolescent psychopathology:
I.
Prevalence
and
incidence
of
depression
and
other
DSM-III-R
disorders
in
high school students.
Journal
of
Abnormal Psychology, 102,
133-144.
Lewinsohn,
P.
M.,
Mermelstein,
R.
M.,
Alexander,
C.,
&
MacPhillamy,
D.
(1985).
The
Unpleasant Events Scale:
A
scale
for the
measurement
of
aversive
events. Journal
of
Clinical Psychology,
41,
483—498.
Lewinsohn,
P.
M.,
Roberts,
R.
E.,
Seeley,
J.
R.,
Rohde,
P.,
Gotlib,
I.
H.,
&
Hops,
H.
(1994). Adolescent psychopathology:
II.
Psychosocial
risk
factors
for
depression. Journal
of
Abnormal Psychology, 103,
302-315.
Lewinsohn,
P.
M.,
Seeley,
J.
R.,
Hibbard,
J.,
Rohde,
P., &
Sack,
W. H.
(1996).
Cross-sectional
and
prospective relationships between physical
morbidity
and
depression
in
older
adolescents. Journal
of
the
American
Academy
of
Adolescent Psychiatry,
35,
1120-1129.
McQuaid,
J.
R.,
Monroe,
S.
M.,
Roberts,
J.
R.,
Johnson,
S.
L.,
Garamoni,
G.,
Kupfer,
D.
J.,
&
Frank,
E.
(1992). Toward
the
standardization
of
life
stress
assessment: Definitional discrepancies
and
inconsistencies
in
methods.
Stress Medicine,
8,
47-56.
Monroe,
S. M.
(1982). Life events
and
disorder: Event-symptom associa-
tions
and the
course
of
disorder. Journal
of
Abnormal Psychology,
91,
14-24.
Monroe,
S.
M.,
Bromet,
E.
J.,
Connell,
M.
M.,
&
Steiner,
S. C.
(1986).
Social support,
life
events,
and
depressive symptoms:
A
1-year prospec-
tive
study. Journal
of
Consulting
and
Clinical Psychology,
54,
424—431.
Monroe,
S.
M.,
&
Depue,
R. A.
(1991).
Life
stress
and
depression.
In J.
Becker
& A.
Kleinman
(Eds.),
Psychosocial aspects
of
depression
(pp.
101-130).
New
York: Erlbaum.
Monroe,
S.
M.,
&
McQuaid,
J. R.
(1994). Measuring
life
stress
and
assessing
its
impact
on
mental health.
In W. R.
Avison
& I. H.
Gotlib
(Eds.),
Stress
and
mental health: Contemporary issues
and
prospects
for
the
future (pp.
43-73).
New
York: Plenum Press.
Monroe,
S.
M.,
Roberts,
J.
E.,
Kupfer,
D.
J.,
&
Frank,
E.
(1996).
Life
stress
and
treatment course
of
recurrent depression:
II.
Postrecovery
associa-
tions
with attrition, symptom course,
and
recurrence over
3
years.
Journal
of
Abnormal Psychology, 105,
313-328.
Monroe,
S.
M.,
&
Simons,
A. D.
(1991). Diathesis-stress theories
in the
context
of
life stress research:
Implications
for the
depressive disorders.
Psychological
Bulletin,
110,
406-425.
Nolen-Hoeksema,
S.,
&
Girgus,
J. S.
(1994).
The
emergence
of
gender
differences
in
depression during adolescence. Psychological Bulletin,
115,
424-443.
Orvaschel,
H.,
Puig-Antich,
J.,
Chambers,
W.,
Tabrizi,
M.
A.,
&
Johnson,
R.
(1982). Retrospective assessment
of
prepubertal
major depression
with
the
Kiddie-SADS-E. Journal
of the
American Academy
of
Child
Psychiatry,
21,
392-397.
Parry-Jones,
W. L.
(1989). Depression
in
adolescence.
In K.
Herbst
& E.
Paykel
(Eds.),
Depression:
An
integrative approach. Oxford, England:
Heinemann
Medical.
Paykel,
E. S.
(1974).
Life
stress
and
psychiatric disorder.
In B. S.
Dohren-
wend
& B. P.
Dohrenwend
(Eds.),
Stressful
life
events:
Their
nature
and
effects
(pp.
135-149).
New
York: Wiley.
Pearlin,
L. L,
Lieberman,
M.
A.,-Menaghan,
E.
G.,
&
Mullan,
J. T.
(1981).
The
stress
process. Journal
of
Health
and
Social Behavior,
22,
337-356.
Post,
R. M.
(1992).
Transduction
of
psychosocial
stress into
the
neurobi-
ology
of
recurrent
affective
disorder. American Journal
of
Psychiatry,
149,
999-1010.
Radloff,
L. S.
(1977).
The
CES-D Scale:
A
self-report depression scale
for
research
in the
general population. Applied Psychological Measure-
ment,
1,
385-401.
Roberts,
R.
E.,
Andrews,
J.
A.,
Lewinsohn,
P.
M.,
&
Hops,
H.
(1990).
Assessment
of
depression
in
adolescents using
the
Center
for
Epidemi-
ologic Studies Depression
Scale.
Psychological Assessment,
2,
122-128.
Roberts,
R.
E.,
Lewinsohn,
P.
M.,
&
Seeley,
J. R.
(1991). Screening
for
adolescent depression:
A
comparison
of
depression
scales.
Journal
of
the
American Academy
of
Child
and
Adolescent Psychiatry,
30,
58-66.
Sandier,
I.
N.,
&
Block,
M.
(1979). Life stress
and
maladaptation
of
children. American Journal
of
Community Psychology,
7,
425-439.
Simons,
A.
D.,
Angell,
K.
A.,
Monroe,
S.
M.,
&
Thase,
M. E.
(1993).
Cognitive
diatheses
and
stressors
in
depression: Relationship between
cognitive factors, rating biases,
and
stress generation. Journal
of
Abnor-
mal
Psychology, 150,
584-591.
Tennant,
C.,
Bebbington,
P.
E.,
&
Hurry,
J.
(1981).
The
short-term out-
come
of
neurotic disorders
in the
community:
The
relation
of
remission
to
clinical factors
and to
"neutralizing" events. British Journal
of
Psy-
chiatry,
13,
213-220.
Teri,
L.
(1982).
The use of the
Beck Depression Inventory with adoles-
cents. Journal
of
Abnormal Child Psychology,
10,
277-284.
U.S. Bureau
of the
Census. (1982). Summary characteristics
for
govern-
mental
units
and
standard metropolitan statistical
areas—Oregon.
Washington,
DC:
U.S. Government Printing Office.
Received October
8,
1997
Revision
received August
21,
1998
Accepted
December
22,
1998
•
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
