Wesleyan University
From the SelectedWorks of Charles A. Sanislow, Ph.D.
April, 1999
Vulnerability to depression:
Reexamining state dependence and
relative stability
David C. Zuro0,
McGill University
Sidney J. Blatt,
Yale University
Charles A. Sanislow,
Yale University
Colin M. Bondi,
Yale University
Paul A. Pilkonis,
University of Pittsburgh - Main Campus
Available at: https://works.bepress.com/charles_sanislow/
13/
Journal
of
Abnormal Psychology
1999,
Vol. 108,
No.
1,76-89
Copyright
1999
by the
American
Psychological Association, Inc.
0021-843Xy99/$3.00
Vulnerability
to
Depression: Reexamining State Dependence
and
Relative Stability
David
C.
Zuroff
McGill
University
Sidney
J.
Blatt,
Charles
A.
Sanislow III,
and
Colin
M.
Bondi
Yale
University
Paul
A.
Pilkonis
University
of
Pittsburgh
Treatment-related decreases
in
Dysfunctional Attitudes
Scale
(DAS;
Weissman
&
Beck, 1978) scores
have been interpreted
as
evidence that dysfunctional attitudes
are
state-dependent concomitants
of
depression. Data
from
the
National Institute
of
Mental Health Treatment
of
Depression Collaborative
Research Program were used
to
reexamine
the
stability
of
dysfunctional attitudes. Mean scores
for
Perfectionism, Need
for
Approval,
and
total
DAS
decreased
after
16
weeks
of
treatment. However,
test-retest
correlations
showed that
the DAS
variables displayed considerable relative stability. Structural
equation
models demonstrated that dysfunctional attitudes
after
treatment were significantly predicted
by
initial level
of
dysfunctional attitudes
as
well
as by
posttreatment
depression.
The
relative stability
of
dysfunctional
attitudes
was
even higher during
the
18-month follow-up period.
The
results were
consistent with
Beck's
(1967)
and
Blatt's
(1974)
theories
of
vulnerability.
Blatt's
(1974, 1990)
and
Beck's (1967) theories
of
vulnerability
to
depression
have
been
widely
influential,
yet
their basic assump-
tions
remain controversial.
Blatt
described
two
personality van-
David
C.
Zuroff,
Department
of
Psychology, McGill University, Mon-
treal, Quebec, Canada; Sidney
J.
Blatt, Departments
of
Psychiatry
and
Psychology, Yale University; Charles
A.
Sanislow III, Department
of
Psychiatry,
Yale University; Colin
M.
Bondi, Child Study Center, Yale
University;
Paul
A.
Pilkonis, Department
of
Psychiatry, University
of
Pittsburgh.
This research
was
supported
by
Grant
410-95-0948
from
the
Social
Sciences
and
Humanities Research Council
of
Canada
and by
Grant
ER0162
from the
Fonds Pour
La
Formation
De
Chercheurs
Et
L'Aide
A La
Recherche
(FCAR-Quebec).
The
National Institute
of
Mental Health (NIMH) Treatment
of
Depres-
sion
Collaborative Research Program
was a
multisite program initiated
and
sponsored
by the
Psychosocial
Treatments
Research
Branch,
Division
of
Extramural
Research Programs (now part
of the
Mood, Anxiety,
and
Personality
Disorders Research Branch, Division
of
Clinical Research),
NIMH.
The
program
was
funded
by
cooperative agreements
to six
partic-
ipating
sites: George Washington University,
MH
33762;
University
of
Pittsburgh,
MH
33753; University
of
Oklahoma,
MH
33760;
Yale Univer-
sity,
MH
33827; Clarke Institute
of
Psychiatry,
MH
38231;
and
Rush
Presbyterian-St.
Luke's
Medical Center,
MH
35017.
The
principal NIMH
collaborators
were Irene
Elkin,
Coordinator;
M.
Tracie
Shea, Associate
Coordinator (formerly
of
George Washington University); John
P. Do-
cherty
(now
at
Nashua Brookside Hospital);
and
Morris
B.
Parloff
(now
at
American University).
The
principal investigators
and
project coordinators
at
the
three participating research sites were,
at
George
Washington Uni-
versity, Stuart
M.
Sotsky
and
David Glass;
at the
University
of
Pittsburgh,
Stanley
D.
Imber
and
Paul
A.
Pilkonis;
and at the
University
of
Oklahoma,
John
T.
Watkins
and
William
Leber.
The
principal investigators
and
project
coordinators
at the
three research sites responsible
for
training
therapists were,
at
Yale University,
Myrna
Weissman (now
at
Columbia
University),
Eve
Chevron,
and
Bruce
J.
Rounsaville;
at the
Clarke Institute
ables,
dependency
and
self-criticism,
that
were
said
to be
associ-
ated
with
both chronic dysphoria
and
increased risk
for
experiencing
major
depression
in
response
to
stressful
events.
Beck's earliest statement
of his
cognitive theory
postulated
stable
cognitive
schemas, having their origin
in
childhood experiences,
whose
activation
by
stressful
events leads
to
depression.
The
concept
of
dysfunctional
attitudes
became more
prominent
in
later
statements
of the
theory (e.g., Beck,
Rush,
Shaw,
&
Emery,
1979),
with
dysfunctional
attitudes
playing
a
similar role
to
that
of
schemas.
Most recently, Beck (1983)
distinguished
two
"modes"
of
depression,
the
sociotropic
and
autonomous,
which
are
similar
to
Blatt's
(1974)
concepts
of
dependent
and
self-critical
depres-
sion.
Sociotropy
and
autonomy
are
believed
to be
associated
with
distinctive
clusters
of
dysfunctional
attitudes (Beck, 1983).
Despite
their origins
in
different
intellectual
traditions,
Blatt's
and
Beck's
theories
are
similar
in
many
respects
(Blatt
&
Marou-
das,
1992; Nietzel
&
Harris, 1990). Both theories postulate rela-
tively
stable personality
or
cognitive characteristics that
confer
vulnerability
to
stress. Cross-sectional studies comparing
de-
pressed
and
nondepressed persons' personalities
or
cognitions
have
provided weak tests
of
vulnerability
theories because
differ-
ences
may
have reflected concomitants
of the
depressed
state
of
Psychiatry, Brian
F.
Shaw
and T.
Michael
Vallis;
and at
Rush
Presbyterian-St.
Luke's Medical Center,
Jan A.
Fawcett
and
Phillip
Ep-
stein. Collaborators
in the
data management
and
data analysis aspects
of
the
program were
C.
James
Klett,
Joseph
F.
Collins,
and
Roderic
Gillis
of
the
Veterans Administration Studies Program, Perry Point, Maryland.
We
thank Yoshio
Takane
for his
expert statistical advice
and
Darcy
Santor
for his
helpful comments
on a
draft
of the
article.
Correspondence
concerning this article should
be
addressed
to
David
C.
Zuroff,
Department
of
Psychology, McGill University, 1205 Doctor Pen-
field
Avenue, Montreal, Quebec, Canada
H3A
1B1.
Electronic mail
may
be
sent
to
zuroff@ego.psych.mcgill.ca.
76
STATE DEPENDENCE
AND
RELATIVE STABILITY
77
rather than predisposing factors. Prospective designs
of
nonde-
pressed,
hypothetically
vulnerable individuals
are
preferable,
but
they
are
difficult
to
implement. Researchers have therefore turned
to
treatment studies
in
which depressed individuals'
scores
on
putative vulnerability
measures
can be
compared
before
and
after
treatment.
If
vulnerability
scores
decrease
as the
depressions
remit,
it
can be
argued that they
are
merely concomitants
of
depression.
Conversely,
if
these variables
are
stable over
the
course
of
treat-
ment,
they remain viable candidates
as
predisposing factors,
al-
though
their causal status remains
to be
established.
Because
the
stability
of
vulnerability variables
is
such
a
funda-
mental issue, numerous studies
of
treated depressives have been
conducted,
most using
the
Dysfunctional
Attitudes Scale (DAS;
Weissman
&
Beck, 1978)
to
operationalize
Beck's
(Beck
et
al.,
1979) concepts.
One
finding
was
consistent across
all of
these
studies:
Mean scores
on the DAS
were lower
after
treatment.
Because
DAS
scores
fluctuate
with clinical state,
the DAS is
frequently
described
as
"state dependent." Some studies have
found
that posttreatment
DAS
scores
of
remitted depressed
pa-
tients were
no
higher than those
of
nondepressed controls (e.g.,
Dohr,
Rush,
&
Bernstein, 1989, Study
1;
Hamilton
&
Abramson,
1983), whereas other studies have
found
that their
DAS
scores
remained elevated compared with controls' (Eaves
&
Rush, 1984;
Peselow, Robins, Block, Barouche,
&
Fieve, 1990). Small sample
sizes
and
inconsistencies
across
studies
in the
definition
of
remis-
sion
complicate
the
interpretation
of
this literature. Nevertheless,
reviewers
have
been
impressed
by the
changes
in DAS
scores
and
have
interpreted
the
literature
as
damaging
to the
cognitive theory,
or at
least
as
requiring clarification
of the
theory
(Barnett
&
Gotlib,
1988; Coyne
&
Gotlib, 1983; Haaga, Dyck,
&
Ernst, 1991; Segal
&
Ingram, 1994).
Added complexity
is
introduced
by the
fact
that stability over
time
can be
conceptually
and
operationally defined
in two
distinct
ways
(Santor, Bagby,
&
Joffe,
1997). Researchers
can
examine
changes
in
mean
score
by
using analysis
of
variance
(ANOVA),
or
they
can
examine relative stability (stability
in
participants' rela-
tive
standing
on a
trait)
by
using correlational techniques.
It is
possible
for a
group
to
show large, significant changes
in
mean
scores
on a
trait
in the
presence
of
perfect stability
in the
partici-
pants' relative standing
on the
trait.
This article examines changes
in
mean scores
and
relative
stability
of the DAS by
using data
from
the
National Institute
of
Mental Health (NIMH) Treatment
of
Depression Collaborative
Research
Program (TDCRP;
Elkin,
1994;
Elkin
et
al.,
1989).
The
DAS and
multiple measures
of
depression were administered
before
and
after
the
16-week
treatment period,
and 6, 12, and
18
months
after
the end of
treatment.
We
examined stability during
the
treatment period
(from
Week
0 to
Week
16) and the
follow-up
period
(from
Week
16 to the
follow-up
at 18
months).
We
used three data-analytic strategies. First,
we
examined
changes
in
mean
DAS
scores during treatment
and
follow-up.
Second,
we
calculated
zero-order
correlations between
the
mea-
sures
of
depression
and
dysfunctional attitudes. Concurrent corre-
lations
between depression
and
dysfunctional attitudes assessed
the
state dependence
effect;
test-retest
correlations assessed
rela-
tive
stability. Third,
we
calculated structural equation
models
(SEMs)
to
evaluate explicitly
the
magnitude
of
both
relative
sta-
bility
effects
and
state dependence
effects.
Relative stability
was
represented
by
autoregressive paths
from
scores
at one
point
in
time
to
scores
on the
same variable
at the
next point
in
time. State
dependence
was
represented
by
paths
from
depression
to
dysfunc-
tional attitudes measured
at the
same point
in
time,
for
example,
from
depression
at
termination
to
dysfunctional attitudes
at
termination.
Predictions
from
three
theoretical
models
were
evaluated.
The
pure
trait model requires that predisposing variables demonstrate
high
levels
of
relative stability
and no
significant state dependence.
In
terms
of
SEMs,
the
pure trait model predicts
a
large, significant
relative stability parameter
and a
nonsignificant parameter
for the
state
dependence
effect.
The
first
model
represents
cognitive the-
ories
as
interpreted
by
their early critics (e.g., Coyne
&
Gotlib,
1983).
The
pure state dependence model requires
the
opposite:
a
small, nonsignificant relative stability parameter
and a
large, sig-
nificant
state dependence parameter. This model represents
the
strong form
of the
critique
of
cognitive theories.
The
third model
is
based
on
interpretations
of
mood state
dependence
effects
advanced
by
Persons
and
Miranda (1992),
Riskind
and
Rholes
(1984), Segal
and
Ingram
(1994),
and
Teas-
dale (1983). This model rests
on two
assumptions. First, there exist
persistent individual differences
in the
availability
of
negative
cognitive-affective structures.
In
recent writings, such structures
have been referred
to as
cognitive-affective
schemas
(Blatt,
Auer-
bach,
&
Levy, 1997)
and
modes (Beck,
1996).
Individuals
differ
in
the
number, intensity,
and
interconnectedness
of
negative
cognitive-affective structures.
These
structures
are
stored
in the
brain
and
continue
to
exist regardless
of
clinical state.
The
persis-
tence
of the
structures
gives
rise to
consistency
in the
rank
ordering
of
individuals
on
measures
of
vulnerability, that
is, to
relative
stability. Second, continual fluctuations take
place
within individ-
uals
in the
accessibility
of
these structures. Accessibility
is
affected
by
multiple factors, including present mood, social context,
and
biological processes.
The
vulnerability
can
remain inaccessi-
ble—in
practice,
undetectable—until
psychological,
social,
or bi-
ological processes increase
its
accessibility
sufficiently
that
it can
be
detected
on
self-report measures such
as the
DAS. Thus,
it is
theoretically possible
for a
remitted depressed patient with
a
nor-
mal
score
on a
vulnerability measure
to
continue
to
possess
the
vulnerability. Because
the
third model postulates both stable dif-
ferences
in
availability
and
fluctuating differences
in
accessibility,
it
is
referred
to as the
state-trait
vulnerability
model.
1
The
state-trait
vulnerability model
is
more
consistent with
Beck's
(1967)
and
Blatt's
(1974) theories than
the
pure trait model
that
is
sometimes attributed
to
them.
In his
original theoretical
statement, Beck (1967) wrote, "Even though these attitudes
(or
concepts)
may not be
prominent
or
even
discernable
at a
given
time, they persist
in a
latent state like
an
explosive charge ready
to
be
detonated
by an
appropriate
set of
conditions. Once activated,
these concepts dominate
the
person's
thinking
and
lead
to the
typical depressive symptomatology"
(p.
277).
The
image
of
"la-
1
Drawing
on
both psychoanalytic developmental theory
and
Piagetian
cognitive-developmental
theory Blatt's (1974;
Blatt
et
al.,
1997) theory
is
more developmental
in
focus
than Beck's (1967). According
to
Blatt,
vulnerable individuals
are
characterized
not
only
by the
negative content
of
their
cognitive-affective schemas
but
also
by the
relative immaturity
of
those
schemas. Negative representations
of
self
and
other
are
often
less
differentiated,
less well integrated,
and at a
lower conceptual level.
The
associated
affects
are
more intense
and
less well modulated. Activation
of
these schemas implies
a
regressive
shift
toward less mature modes
of
response,
often
associated with
an
increase
in
negative content.
78
ZUROFF,
BLATT,
SANISLOW,
BONDI,
AND
PILKONIS
tent"
attitudes
and
schemas
are
prominent
in
Beck's
later
writings
as
well
(e.g., Beck, 1987;
Kovacs
&
Beck,
1978).
Most
recently,
Beck
(1996)
has
de-emphasized
the
causal
role
of
cognition,
suggesting
instead
that
the
cognitive,
affective,
motiva-
tional,
behavioral,
and
physiological
symptoms
of
depression
are
linked
together
as a
"mode."
Modes
are
said
to
vary
in
activation
(or
"charge"
or
"cathexis").
When
the
mode
is
highly
charged,
the
individual
experiences
an
episode
of
depression;
the
mode contin-
ues
to
exist,
but in a
deactivated state,
after
an
episode
of
depres-
sion
remits.
Beck
(1996)
distinguished
deactivating
a
mode
from
eliminating
or
modifying
a
mode.
Deactivation
can
occur
for
many
reasons,
including
pharmacological
treatment;
modification
of
modes
ordinarily
requires psychotherapeutic intervention.
Blatt
(Blatt
&
Maroudas,
1992)
has
placed greater emphasis
on
the
long-term
stability
of
personality
than
has
Beck. Nevertheless,
he
allowed that environmental stressors
can
activate
and
intensify
intrapsychic
conflicts
and
that,
conversely,
a
supportive environ-
ment
or
successful
psychotherapy
can
deactivate
or
modify
mal-
adaptive
character
traits.
The
activation
of
negative cognitive-
affective
structures,
derived
from
earlier experiences
and
characterized
by a
relatively
low
cognitive-developmental level,
would
be
described
in
psychoanalytic
terminology
as
regression
(Blatt,
1974;
Blatt
&
Shichman,
1983; Blatt
et
al.,
1997).
In
summary,
Blatt
and
Beck each
recognized
that
vulnerabilities
have
stable,
traitlike
components
and fluctuating,
statelike compo-
nents.
Their theories
are
most
consistent
with
the
state-trait vul-
nerability
model.
Consequently,
both
the
state dependence
and
relative
stability
parameters
would
be
expected
to be
significant.
We
planned
to
examine
the
relative stability
and
state depen-
dence
of
dysfunctional
attitudes
during
the
follow-up
period
as
well
as the
treatment
period
of the
TDCRP.
To
avoid
restricting
the
range
of
depression
at 18
months
and
consequently
limiting
the
size
of the
state
dependence parameter,
we
included
in
these
analyses
all
available
participants, regardless
of
whether
they
were
in
remission
or
relapsed.
Both
depression
and
dysfunctional
atti-
tudes
were
expected
to
show
greater mean
changes
during
the
treatment
period
than
during
the
follow-up
period. Larger mean
changes
in
depression
and
dysfunctional
attitudes
during
treatment
would
be
expected
to
lead
to
more
frequent
alterations
in
patients'
relative
standing
and
consequently
to
decreased relative
stability.
Therefore,
it was
predicted
that
the
relative
stability
parameters
would
be
larger
for the
follow-up
period
than
the
treatment
period.
Method
The
TDCRP tested
the
efficacy
of
cognitive-behavior
therapy, interper-
sonal therapy,
imipramine
plus clinical management,
and
placebo plus
clinical
management
for
outpatients with
nonbipolar,
nonpsychotic,
major
depressive disorders.
The
participants were
250
patients
who
were ran-
domly
assigned
to the
four conditions.
Of the 250
patients,
239
began
treatment,
and 162
were defined
as
"completers,"
having received
at
least
12
treatment sessions over
at
least
a
15-week
period.
Inclusion
and
exclusion criteria, sample
characteristics,
treatment pro-
cedures,
and
assessment procedures have been described
in
previous pub-
lications
(Elkin,
1994;
Elkin
et
al.,
1989;
Imber
et
al.,
1990; Sotsky
et
al.,
1991;
Watkins
et
al.,
1993). Patients
met
Research Diagnostic
Criteria
(RDC; Spitzer, Endicott,
&
Robins, 1978)
for a
current episode
of
definite
major
depression that
had
been present
for at
least
the
previous
2
weeks.
Among
the
patients
who
began treatment,
70%
were female,
38%
were
definite
endogenous
by RDC
criteria,
and 64% had had one or
more prior
episodes
of
major
depression.
The
average
age was 35.
Complete
data were
available
at
Weeks
0 and 16 for 154
patients. Because
of
attrition during
follow-up,
the
sample
for
analyses spanning termination (Week
16) to 18
months
was
reduced
to 142
patients.
Dysfunctional
Attitudes
The
40-item
DAS is
intended
to
measure cognitive vulnerability
to
depression. Previous analyses
of the
TDCRP data have used both
the
total
score
and the
Perfectionism
and
Need
for
Approval subscales.
The
sub-
scales were derived
from
the
intake data
by
principal-components analysis,
followed
by
varimax
rotation
(Imber
et
al.,
1990).
Need
for
Approval
is
conceptually related
to
Blatt's
(1974)
construct
of
dependency
and to
Beck's
(1983)
construct
of
sociotropy.
Perfectionism
is
conceptually
re-
lated
to
self-criticism
and to
autonomy. Although
these
constructs
are not
interchangeable (Blatt
&
Maroudas, 1992;
Zuroff,
1994), there
are
sub-
stantial correlations between measures
of
dependency, sociotropy,
and
need
for
approval
and
smaller,
but
still significant, correlations between
measures
of
self-criticism, autonomy,
and
perfectionism (Blaney
&
Kutcher,
1991; Zuroff, 1994).
Imber
et al.
(1990)
found
that
11
items loaded substantially
(>.40)
on
Need
for
Approval
and
that
15
items loaded substantially
on
Perfectionism.
The two
highest loading items
for
Need
for
Approval were
"What
other people think
of
me is
important"
and "I can find
happiness without being loved
by
another
person."
The
corresponding items
for
Perfectionism were
"If I do not do as
well
as
other people,
it
means that
I am an
inferior human
being"
and "If I
fail
at
my
work,
then
I am a
failure
as a
person."
The
items with high loadings were
summed,
and the
resulting
composites
had
high internal consistency
(a
=
.91
for
Need
for
Approval
and .82 for
Perfectionism).
The two
factors were
congruent with previous factor analyses
of the DAS
(Cane,
Olinger,
Gotlib,
&
Kuiper, 1986; Oliver
&
Baumgart,
1985)
and
with
Mongrain
and
ZurofFs
(1989a)
rationally
derived subscales.
The two
subscales were moderately correlated
at
intake
in the
total
sample
(r =
.59,
p <
.001). Need
for
Approval
and
Perfectionism therefore
share considerable variance,
but
each
has
unique variance
as
well.
Depend-
ing
on the
research question, investigators
can
focus
on the
shared variance
or
treat Need
for
Approval
and
Perfectionism
as
distinct variables.
Self-Reported
Depression
Two
self-report indicators
of
depression were used:
the
Beck Depression
Inventory
(BDI;
Beck, Ward,
Mendelson,
Mock,
&
Erbaugh, 1961)
and the
Depression subscale
of the
Hopkins Symptom
Checklist-90
(SCL-D;
Derogatis,
Lipman,
&
Covi,
1973).
The BDI is a
widely used
21-item
inventory
of the
affective, cognitive, motivational,
and
somatic symptoms
of
depression. Extensive research
has
demonstrated that
it is
reliable
and
correlates
well with self-report
and
interviewer-based measures
of
depres-
sion (Gotlib
&
Cane, 1989; Shaw, Vallis,
&
McCabe, 1985).
The
13-item
SCL-D
assesses primarily
the
affective
and
cognitive symptoms
of de-
pression.
It is
internally consistent
and
sensitive
to
changes
in
response
to
treatment
(Rabkin
&
Klein, 1987).
Clinical
Evaluator-Rated
Depression
At
each point
in
time, patients were assessed
by the
same clinical
evaluator
who had
screened them
for
participation
in the
study.
The
clinical
evaluators conducted
a
semistructured
interview, using
the the
Schedule
for
Affective
Disorders
and
Schizophrenia—Change
Version
(SADS-C;
En-
dicott, Cohen, Nee,
Fleiss,
&
Sarantakos,
1981).
On the
basis
of
this
interview, evaluators completed
the
Hamilton Rating Scale
for
Depression
(HRSD; Hamilton, 1960, 1967).
We
present data
for the
17-item
version
of
the
HRSD,
as
have previous researchers using
the
TDCRP data (Elkin,
1994).
The
evaluators also
recorded
patients' responses
to
each
of the
questions
on the
SADS-C.
The
original TDCRP researchers calculated
scores
for a
number
of
subscales.
We
used
two of
these subscales,
the
Depressive
Syndrome
(SADS-DS)
and the
Extracted Hamilton
STATE
DEPENDENCE
AND
RELATIVE STABILITY
79
Table
1
Means,
Standard Deviations,
and F
Tests
for
Change During Treatment
and
Follow-Up
in
Measures
of
Depression
and
Dysfunctional Attitudes
WeekO
Variable
Self-reported
depression
BDI
SCL-D
CE-rated
depression
HRSD
SADS-EH
SADS-DS
Dysfunctional
attitudes
Perfectionism
Need
for
Approval
Total
DAS
(N
=
M
26.84
2.50
19.12
17.85
41.74
50.62
45.81
141.56
154)
SD
7.89
0.65
4.06
4.76
5.93
17.31
12.06
36.04
Week
16
(N
=
M
8.74
0.79
7.43
7.21
22.74
39.05
37.01
113.29
154)
SD
8.76
0.72
5.73
4.93
8.68
16.69
11.65
36.01
F(l,
153)"
483.55
592.95
505.41
421.80
578.55
101.34
96.01
135.70
Week
16
(N
=
M
8.17
0.75
7.20
7.03
22.23
38.53
36.65
111.97
142)
SD
8.03
0.69
5.61
4.84
8.23
16.54
11.79
35.85
18
Months
(N
=
M
8.23
0.83
7.31
7.45
23.22
38.20
36.34
110.57
142)
SD
8.97
0.82
6.03
5.54
9.31
16.21
12.19
34.37
F(l,
141)
b
0.01
1.40
0.04
0.62
1.20
0.12
0.15
0.47
Note.
BDI =
Beck
Depression
Inventory;
SCL-D
=
Symptom
Checklist-90
Depression
subscale;
CE
=
clinical
evaluator;
HRSD
=
Hamilton Rating
Scale
for
Depression; SADS-EH
=
Schedule
for
Affective
Disorders
and
Schizophrenia—Extracted
Hamilton
subscale;
SADS-DS
=
Schedule
for
Affective
Disorders
and
Schizophrenia—Depressive
Syndrome subscale;
DAS =
Dysfunctional
Attitudes
Scale.
a
All Fs for
changes
from
0 to 16
weeks
were
significant
(p <
.001).
b
None
of the Fs for
change
from
16
weeks
to 18
months
were
significant.
(SADS-EH).
The
SADS-DS
included
16
items
that
sample
a
wide
range
of
depressive
symptoms
(e.g., self-reproach,
discouragement,
lack
of en-
ergy,
and
psychomotor retardation).
The
SADS-EH
was
developed
by
Endicott
et
al.
(1981)
as an
approximation
to the
HRSD.
It
combines
the 17
SADS-C
items
that
are
most
similar
to the
HRSD
items. Eight
items
occur
on
both
the
SADS-DS
and the
SADS-EH.
Results
We
begin
by
examining changes
in
mean scores during
the
treatment
and
follow-up periods.
We
then present zero-order cor-
relations, focusing
first
on
concurrent correlations between dys-
functional
attitudes
and
depression
as
indicators
of
state depen-
dence
and
then
on
test-retest
correlations
as
indicators
of
relative
stability. Finally,
we
test SEMs that postulate both relative stability
and
state dependence.
Changes
in
Mean
Scores
During
Treatment
and
Follow-Up
Table
1
presents means, standard deviations,
and F
tests
of the
significance
of
change.
There
were significant decreases during
the
treatment
period
on all
five
measures
of
depression
as
well
as on
Perfectionism,
Need
for
Approval,
and
total DAS.
These
results
are
consistent with previous reports that dysfunctional attitudes
decrease during treatment
of
depression.
However, total
DAS
scores
at
termination remained elevated compared with means
for
nondepressed community control participants reported
by
Peselow
et al.
(1990),
Eaves
and
Rush (1984),
and
Hamilton
and
Abramson
(1983).
2
The
depression scores
at
termination suggest
the
presence
of
residual depression
and
raise
the
possibility that total
DAS
scores
in
complete responders would
be
closer
to
normal levels.
Following
Elkin
et al.
(1989),
we
defined complete responders
as
those with Week
16
HRSD scores
of 6 or
less
and
partial responders
as
those with HRSD scores
of 7 to
11.
These criteria identified
76,46,
and
32
participants
in the
complete, partial,
and
nonresponse
groups,
respectively.
Total
DAS
scores were then subjected
to an
ANOVA
with
response group (complete, partial, none)
and time
(Week
0 to
Week
16) as
variables.
The
main
effect
for time was
significant,
indicating
a
marked
decrease
in
total
DAS
across
the
groups,
but it
was
qualified
by a
significant Response Group
X
Time interaction,
F(2,151)
=
3.78,
p
<
.05. Table
2
reveals that
the
complete respond-
ers
showed
the
greatest reduction
in
dysfunctional attitudes.
At
ter-
mination,
the
mean
DAS
score
for the
complete responders
was in the
same range
as
normal community controls.
There
were
no
significant changes
in
mean scores
for any of the
variables
from
the end of
treatment (Week
16)
to the
follow-up
testing
18
months later.
Correlational
Analyses
Concurrent
relations between depression
and
dysfunctional
at-
titudes.
Tables
3 and 4
present
the
zero-order correlations. Data
for
the
treatment period
are
presented above
the
diagonals; data
for
the
follow-up period
are
presented below
the
diagonals.
At
both
termination
(16
weeks)
and the
18-month
follow-up, correlations
between measures
of
depression
and
dysfunctional attitudes were
significant
and
moderately large,
as
would
be
predicted
by the
state
dependence
model.
Surprisingly,
the
concurrent correlations
at
intake
were smaller
and in
some
cases
not
significant.
3
It is
possible that
the
magnitude
of the
relation between depression
and
2
We
searched
for
studies
reporting
test-retest
data
on
nondepressed
participants
who
were,
if not
matched
to
depressed participants,
at
least
community-resident
adults
rather
than
college
students.
Surprisingly,
we
found
only
these three
studies.
Peselow
et al.
(1990)
used
Form
B of the
DAS,
whereas
the
TDCRP
used
Form
A of the
DAS.
Hamilton
and
Abramson
(1983)
used
both
forms.
However,
the two
forms
are
simply
subsets
of the
original
100-item
DAS. They
have
highly
similar
means
and
can
reasonably
be
compared
with
one
another
(Weissman,
1979).
3
The
smaller
correlations
at
intake
are
probably
not
attributable
to
restricted
variability because
the
standard
deviations
of the
measures
of
depression
and
dysfunctional
attitudes
were
similar
at the
three points
in
time.
The
unstandardized
regression
coefficients
for the
regressions
of
dysfunctional
attitudes
on
depression
displayed
the
same
pattern
as the
correlation
coefficients;
that
is, at
intake,
they
were
smaller
but
generally
significant.