Cognitive–Behavioral Treatment of Anxiety Disorders in Children: Long-Term (6Year) Follow-Up

Abstract

Authors evaluated the long-term effectiveness of cognitive-behavioral therapy (CBT) for childhood anxiety disorders. Fifty-two clients (aged 14 to 21 years) who had completed treatment an average of 6.17 years earlier were reassessed using diagnostic interviews, clinician ratings, and self- and parent-report measures. Results indicated that 85.7% no longer fulfilled the diagnostic criteria for any anxiety disorder. On a majority of other measures, gains made at 12-month follow-up were maintained. Furthermore, CBT and CBT plus family management were equally effective at long-term follow-up. These findings support the long-term clinical utility of CBT in treating children and adolescents suffering from anxiety disorders.

Full text

Journal
of
Consulting
and
Clinical Psychology
2001, Vol.
69, No. 1,
135-141
Copyright
2001
by the
American Psychological Association, Inc.
0022-006X/01/S5.0Q
DOI:
10.1037//0022-006X.69.I.135
Cognitive-Behavioral Treatment
of
Anxiety Disorders
in
Children:
Long-Term
(6-Year)
Follow-Up
Paula
M.
Barrett
and
Amanda
L.
Duffy
Griffith
University, Gold Coast
Mark
R.
Dadds
Griffith
University,
Mt.
Gravatt
Ronald
M.
Rapee
Macquarie
University
Authors evaluated
the
long-term effectiveness
of
cognitive-behavioral
therapy (CBT)
for
childhood
anxiety
disorders.
Fifty-two clients (aged
14 to 21
years)
who had
completed
treatment
an
average
of
6.17
years earlier were reassessed using diagnostic interviews, clinician ratings,
and
self-
and
parent-report
measures. Results indicated that 85.7%
no
longer
fulfilled
the
diagnostic criteria
for any
anxiety disorder.
On
a
majority
of
other measures, gains made
at
12-month follow-up were maintained. Furthermore,
CBT
and
CBT
plus
family
management were equally
effective
at
long-term follow-up. These findings support
the
long-term clinical utility
of CBT in
treating children
and
adolescents
suffering
from
anxiety disorders.
A
growing body
of
evidence indicates that anxiety disorders
in
childhood
can be
successfully treated with relatively brief psycho-
social
interventions. Kendall
(1994)
conducted
the first
published
randomized clinical trial
of a
cognitive-behavioral
treatment
(CBT) with anxious children. This
study
involved
47 9- to 13-
year-old
children with overanxious disorder, separation anxiety,
or
avoidant
disorder. Children
who
received
the
16-session
treatment
displayed significant improvement
from
pre-
to
posttreatment
on
self-report, parent report,
and
behavioral observation measures.
In
addition,
at
posttreatment,
64% of
children
in the
treatment group
were
diagnosis free. These gains were maintained
at
1-year
follow-
up.
A
second clinical trial, utilizing
a
sample with similar charac-
teristics, showed comparable results, with
the CBT
group again
demonstrating significant improvements when compared with
the
wait-list group (Kendall
et
al.,
1997).
These
studies indicate that
CBT
treatment
for
children
is
effec-
tive
in
reducing anxiety,
and
attempts have increasingly focused
on
maximizing treatment gains. Specifically, recent years have seen
increasing interest
in the
role
the
family plays
in the
development
and
treatment
of
childhood disorders.
In
particular, several char-
acteristics appear
to be
more common
in
parents
of
anxious chil-
dren.
For
example,
in a
review
of the
literature, Rapee (1997)
reported that parental overcontrol
has
consistently been
found
to
be
associated with child anxiety problems. Findings
from
Paula
M.
Barrett
and
Amanda
L.
Duffy,
School
of
Applied Psychology,
Griffith
University, Gold Coast,
Southport,
Queensland, Australia; Mark
R.
Dadds, School
of
Applied Psychology,
Griffith
University,
Mt.
Gravatt,
Meadowbrook, Queensland, Australia; Ronald
M.
Rapee,
Department
of
Psychology,
Macquarie University, Sydney,
New
South Wales, Australia.
Correspondence concerning this article should
be
addressed
to
Paula
M.
Barrett,
School
of
Applied Psychology,
Griffith
University, Queens-
land
4111,
Australia. Electronic mail
may be
sent
to
p.barrett@mailbox.
gu.edu.au.
Siqueland, Kendall,
and
Steinberg's
(1996)
study supported this
result,
with independent observers rating parents
of
children with
anxiety
disorders
as
less granting
of
psychological autonomy than
parents
of the
control children.
In
addition, Siqueland
et al.
found
that
anxious children rated both
of
their parents
as
less
accepting
than
did
control children. Further research
by
Barrett, Rapee,
Dadds,
and
Ryan (1996) investigated
the
influence
of
family
discussion
on the
interpretations that anxious children made when
presented with ambiguous situations. They
found
that anxious
children made
a
relatively high number
of
threat interpretations
and
predominantly chose avoidant solutions, with
family
discus-
sion only provoking enhancement
of
these avoidant solutions.
Furthermore, rates
of
child avoidance were positively correlated
with
the
probability that parents reciprocated avoidance (Dadds,
Barrett, Rapee,
&
Ryan,
1996).
On
the
basis
of
such
findings, the
inclusion
of
parents
in
therapy
may
be an
important part
of
effectively treating child anxiety
problems,
and a
number
of
recent studies have addressed this
issue. Howard
and
Kendall
(1996)
used
a
multiple baseline across-
cases design
to
evaluate
a
family-based
CBT
program with
six
children
(aged
9 to 13
years)
who met the
criteria
for an
anxiety
disorder. They reported gains
at
posttreatment
on
diagnostic
and
questionnaire measures
for
four
of the
children
and,
with
the
exception
of one
child, these gains were maintained
at
4-month
follow-up.
Barrett,
Dadds,
and
Rapee
(1996)
compared child-only
CBT,
child
CBT
plus
family
anxiety management training
(CBT
+
FAM),
and a
wait-list control group. Participants were
79
children,
aged
7 to 14
years, with overanxious
disorder,
separation anxiety,
or
social
phobia.
The
family component
of the
program consisted
of
training
in
three areas:
(a)
child management,
(b)
parental
anxiety management,
and (c)
communication
and
problem-solving
skills.
Both
CBT and CBT + FAM
conditions showed greater
improvement
on a
variety
of
measures
at
posttreatment
and
12-
month
follow-up when compared with
the
wait-list. However,
the
135

136
BRIEF
REPORTS
clients that received
family
training also showed significantly
greater
improvement than
the
CBT-only
group
on a
number
of
measures.
In
addition,
at
12-month
follow-up,
70% of
those
in the
CBT
group
and 95% of
those
in the CBT + FAM
group
no
longer
met
the
diagnostic
criteria
for any
anxiety
disorder.
A
further
study
by
Cobham
(1998)
investigated
the
effective-
ness
of one
component
of the
family management program:
pa-
rental
anxiety management (PAM). Sixty-seven children, aged
between
7 and 14
years,
who met
diagnostic
criteria
for an
anxiety
disorder participated
in
this study.
Of
these,
32 had
parents
who
were
classified
as
nonanxious (child-anxiety-only group),
whereas
35 had
either
one or
both parents
who
reported high levels
of
anxiety (child
+
parental anxiety group). Children
from
both
of
these groups were then randomly assigned either
to
child-focused
CBT or to the
child-focused
CBT and
parental-anxiety manage-
ment (CBT
-I-
PAM).
Of
those children
who
received
only
CBT, 82.4%
of the
child-anxiety-only group were diagnosis
free
at
posttreatment,
compared with 38.9%
of the
child
+
parental anx-
iety
group.
Of
those
in the CBT + PAM
condition,
80% of the
child-anxiety-only
group
and
76.5%
of the
child
+
parent anxiety
group were diagnosis
free.
These results indicated that
the
children
with
two
nonanxious parents responded more favorably
to
child-
focused
CBT
than
did the
children
who had one or
more anxious
parents.
The
inclusion
of PAM
increased
the
efficacy
of
child-
focused
CBT for
children,
but
only
for
children
who had at
least
one
anxious
parent.
However,
at 6- and
12-month
follow-ups,
these
effects
became
less
evident, although trends
in the
expected
directions continued.
Although
these results
all
point
to the
effectiveness
of CBT in
treating children with anxiety
disorders,
longer term follow-up
of
clients
is a
vital next step (Kendall, 1998; Weisz
&
Hawley, 1998).
At
present, only
a
small number
of
such studies exist
in the
area
of
childhood anxiety.
One of the
earliest
was
conducted
in
1982
by
Graziano
and
Mooney,
who
investigated
the
long-term
effective-
ness
of a
behavioral treatment
of
children's nighttime fears. They
found
strong maintenance
effects
at
2.5-3-year
follow-up,
but the
study
did not
utilize standardized measures.
More
recently, Ken-
dall
and
Southam-Gerow (1996) reassessed
36 of the 47
children
treated
in
Kendall's (1994) original clinical trial.
The
length
of
time
from
completion
of the
treatment program
to
reassessment
ranged
from
2 to 5
years, with
an
average
of
3.35
years.
On
both
self-report
and
parent-report measures,
the
treatment gains seen
at
1-year
follow-up were maintained,
with
no
detectable
diminish-
ment.
In
terms
of
diagnostic status, improvements
at
1-year
follow-up
were also maintained.
The
present
study
furthers
research
in
this area
by
reassessing
the
clients
involved
in
Barrett,
Dadds,
&
Rapee's
(1996)
study
an
average
of 6
years
after
treatment completion.
It was
hypothesized
that treatment gains made
by
clients
at
12-month follow-up
(12-
month
FU)
would
be
maintained
at
long-term follow-up
(LT
follow-up).
This
would
be
evidenced
by no
significant
increase
in
anxiety
as
measured
by
diagnostic interview, self-report,
and
parent-report.
It was
also hypothesized that those
in the CBT +
FAM
group would continue
to
evidence
better
outcomes
than
those
in the CBT
group. Furthermore, additional analyses were
undertaken
to
explore
the
effects
of
diagnostic
comorbidity
on
long-term treatment outcome.
Method
Participants
The
participants
in the
present
study
had
previously completed treatment
as
part
of
Barrett,
Dadds,
&
Rapee's
(1996)
study.
These
children
had
been
referred
for
treatment,
and
parent
and
child diagnostic interviews con-
firmed the
presence
of a
DSM-IU
(Diagnostic
and
Statistical Manual
of
Mental
Disorders: American Psychiatric Association, 1980) anxiety dis-
order. Full details
of the
study
are
available
in
Barrett, Dadds,
&
Rapee.
Although
we
attempted
to
contact
all
79 of the
previous
participants,
23
(32.9%) could
not be
located.
Of the 56
participants that were located,
53
(94.6%;
67.1%
of the
original sample) agreed
to be
involved
in the
study.
AH
participants
who
were followed
up
were
also
asked
whether they
had
sought
alternative treatment since
the
original study. Only
1
participant
had
received
further
psychological treatment
for an
anxiety-related problem,
and
we
excluded this person
from
data analysis.
The
remaining
52
participants ranged
in age
from
13 to 21
years
(M
=
16.08,
SD —
2.26),
with
the
average length
of
time since treatment
completion being 6.17 years (range
=
5.33-7.08).
Twenty-three
had
orig-
inally
been diagnosed
with
overanxious disorder (OAD),
18
with
separa-
tion anxiety (SAD),
and 11
with social phobia (SP).
Further,
19.2% were
originally
diagnosed
as
comorbid with simple phobia, 3.8%
with
depres-
sion,
and
3.8%
with
oppositional defiant disorder (ODD). Thirty-one
participants
(17
boys,
14
girls) originally belonged
to the
child-only
CBT
condition
and 21
(11
boys,
10
girls)
to the CBT + FAM
condition.
Brief
Description
of
Treatment Conditions
In
the
initial study, participants were randomly
assigned
to
either
the
wait-list
CBT or CBT + FAM
condition, with those
in the
wait-list
receiving treatment
at a
later
date.
Both treatments consisted
of 12
sessions,
with
each session lasting
60-80
min. Treatment
sessions
were conducted
by
one of five
registered clinical psychologists
in the
Behavior Research
and
Therapy
Centre
of the
University
of
Queensland, Australia. Refer
to
Barrett
et
al.
(1996)
for
full
details.
Measures
Anxiety
Interview Disorder Schedule
for
Children
(ADIS-C;
Silverman
&
Nelles,
1988).
Children were administered
the
ADIS-C,
a
structured
interview
that
is
used
to
ascertain whether
a
child meets
the
DSM-1II
criteria
for any
anxiety disorder. This interview
was
administered over
the
phone,
by a
clinician
who was
unaware
of the
child's
original treatment
condition.
To
ensure reliable
diagnoses,
18
children were reinterviewed,
with
the
overall kappa agreement
for the
presence
of an
anxiety
disorder
being 0.85.
In
addition
to
making
a
diagnosis,
the
clinician rated improve-
ment
in the
child
and
family
on
seven
dimensions
of
adjustment:
(a)
clinical global impression,
(b)
overall
functioning,
(c)
overall anxiety,
(d)
avoidant
behaviors,
(e)
family disruption,
(0
parental perceived ability
to
deal
with
child's behavior,
and (g)
child's perceived ability
to
deal with
feared
situations.
These
ratings were based
on all
ADIS-C anxiety items,
as
well
as on
direct questioning
of
both
the
child
and
parent about each
dimension. Ratings were made
on a
7-point scale, where
0 =
markedly
worse,
3 = no
change,
and 6 =
marked improvement.
Revised
Children's
Manifest
Anxiety Scale
(RCMAS;
Reynolds
&
Rich-
mond,
1985).
The
RCMAS provides
a
measure
of a
child's chronic
anxiety.
The
questionnaire contains
37
items,
9 of
which
form
a Lie
scale.
For
each item,
the
child
is
asked
to
respond
yes or no.
This measure
has
been
found
to
have high internal consistency
and
test-retest
reliability,
as
well
as to
show convergent
and
divergent validity (Reynolds
&
Richmond,
1985).
Fear
Survey Schedule
for
Children—Revised
(FSSC-R;
Ollendick,
1983).
The
FSSC-R
assesses
specific fears
in
children.
It is 80
items
in
length,
with each item rated
on a
3-point scale. This questionnaire
has
also
been shown
to
have good
test—retest
reliability
and
internal consistency.

BRIEF REPORTS
137
Children's
Depression Inventory
(CDl;
Kovacs,
1992).
The
CDI
is 27
items
in
length
and
provides
a
measure
of
depressive symptomatology.
Each item consists
of
three descriptive statements,
of
which
the
child must
select
the one
that best characterizes
him or her
during
the
previous
2
weeks. This scale
has
been
found
to
have high internal consistency
and
moderate
test-retest
reliability,
as
well
as to
exhibit discriminant
and
concurrent validity (Kovacs, 1992).
Child
Behavior Checklist
(CBCL;
Achenbach
&
Edelbrock,
1991).
In
the
present study, both
the
child's
mother
and
father
completed
the
CBCL.
This measure
is
118
items
in
length,
with
parents'
rating each item
on a
3-point scale. From these items,
a
total problem-behavior
score
can be
derived,
as
well
as
several subscale scores,
and
scores
on two
dimensions
of
dysfunction: Internalizing
and
Externalizing. Only
the
Internalizing
and
Externalizing
scale scores were used
in
this study. Research
has
shown
these scales
to be
psychometrically sound, with high
test-retest
reliability
and
internal consistency
reported.
Support
for the
content, construct,
and
criterion-related validity
of the
CBCL
has
also been
found
(Achenbach
&
Edelbrock, 1991).
Results
To
determine whether
there
were significant demographic dif-
ferences
between those children involved
in the LT
follow-up
and
those
who
were
not,
we
conducted
a
number
of
/
tests
and
chi-
square
analyses. Results showed that
the two
groups
did not
differ
in
terms
of
gender,
^(1,
N = 79) =
0.60,
ns, or
age,
r(77)
=
—0.50,
ns, at
pretreatment.
A
significant difference
was
found
in
terms
of
severity
of
diagnosis
at
pretreatment, f(77)
=
—2.18,
p <
.05,
but
those children
who
were involved
in the LT
follow-up
had
the
more severe anxiety disorders. Furthermore,
no
significant
difference
was
found between participants
who
were involved
in
the
follow-up
and
those
who
were
not
involved
in
terms
of
diag-
nostic status
at
12-month
FU,
^(1,
N = 75) =
0.16,
ns.
Demographic variables were also examined
to
determine
whether
the CBT and CBT + FAM
groups differed.
No
significant
differences
between
the
conditions were
found
for
gender,
^(1,
N
= 52) =
0.03,
ns,
age,
/(50)
=
-0.31,
ns, or
severity
of
diagnosis,
r(49)
=
-1.58,
ns, at
pretreatment.
Diagnostic Status
At
LT
follow-up, diagnostic status
was
determined
on the
basis
of the
child interview only, whereas diagnoses
at
earlier assess-
ment
points were based
on
combined parent
and
child reports.
Consequently,
as
parents
and
children have been shown
to
disagree
about diagnostic status
(e.g.,
Rapee, Barrett, Dadds,
&
Evans,
1994),
comparing previous
combined
diagnoses
to
child-only
di-
agnoses
at LT
follow-up
may be
misleading. That
is,
some chil-
dren
may
have reported
no
problems
at
previous assessments
and
therefore,
although
not
meeting diagnostic criteria
at LT
follow-
up,
they
do not
qualify
as
having
"recovered."
Because
of
this concern, only
the
participants
who met
diagnos-
tic
criteria
at
pretreatment
on the
basis
of
child
report
were
in-
cluded
in the following
diagnostic status analyses. This restriction
led
to 3
participants being excluded.
Of the
remaining
49
partici-
pants,
21
were diagnosed with OAD,
18
with SAD,
and 10
with
SP. The
excluded cases
all
belonged
to the CBT
condition.
At
LT
follow-up,
42 of
these
49
participants (85.7%)
no
longer
met the
diagnostic criteria
for any
anxiety disorder.
In
comparison,
39'
of
these participants were diagnosisfree
at
12-month
FU
(79.6%). Five
of
these
39
participants
relapsed
and
again qualified
for
a
diagnosis
at LT
follow-up, while
a
further
6 who had
received
a
diagnosis
at
12-month
FU
were diagnosisfree
atLT
follow-up.
A
McNemar
analysis indicated
no
significant difference
in
diagnostic
status
between
the
assessment phases.
Differences
between
the CBT and CBT + FAM
groups
in
diagnostic status were also examined. Twenty-four
of the 28
children
(85.7%)
in the CBT
group
and 18 of the 21
(85.7%)
in
the CBT + FAM
group were diagnosis
free
at LT
follow-up.
Analysis
revealed
no
significant
difference
between
the
groups,
^(l,
N = 1) =
0.00,
ns.
In
addition,
an
analysis
was
conducted
to
determine whether
type
of
diagnosis
at
pretreatment
(i.e. OAD,
SAD,
or SP) was
associated
with
a
differential treatment
effect
at LT
follow-up.
No
significant
difference
was
found between
the
groups,
x*(2,
N
—
49) =
0.687,
ns,
with
81%
of the OAD
group, 88.9%
of the SAD
group,
and 90% of the SP
group
diagnosis-free
at LT
follow-up.
Clinical
Evaluations
Table
1
shows
the
means
for the
seven clinical evaluation scales
at
12-month
FU and LT
follow-up.
As the
clinical evaluations
represent seven interrelated dimensions,
we
analyzed data using
a
2
(condition: CBT,
CBT +
FAM)
X 2
(phase: 12-month
FU, LT
follow-up)
repeated-measures
multivariate
analysis
of
variance
(MANOVA). Results
of
this
analysis
revealed
no
significant
in-
teraction,
F(7,
41) =
1.33,
ns,
if
=
.18
2
,
or
condition
effect,
F(l,
41) =
0.65,
ns,
rf
=
0.10. However,
a
significant
effect
for
phas