cognitive therapy, 2004.pdf
TRANSCRIPT
-
7/28/2019 cognitive therapy, 2004.pdf
1/30
R E S E A R C H U P D A T E R E V I E W
Cognitive-Behavioral Psychotherapy for Anxiety andDepressive Disorders in Children and Adolescents:
An Evidence-Based Medicine ReviewSCOTT N. COMPTON, PH.D., JOHN S. MARCH, M.D., M.P.H., DAVID BRENT, M.D.,
ANNE MARIE ALBANO, PH.D., V. ROBIN WEERSING, PH.D., AND JOHN CURRY, PH.D.
ABSTRACT
Objective: To review the literature on the cognitive-behavioral treatment of children and adolescents with anxiety and
depressive disorders within the conceptual framework of evidence-based medicine. Method: The psychiatric and
psychological literature was systematically searched for controlled trials applying cognitive-behavioral treatment to
pediatric anxiety and depressive disorders. Results: For both anxiety and depression, substantial evidence supports the
efficacy of problem-specific cognitive-behavioral interventions. Comparisons with wait-list, inactive control, and active
control conditions suggest medium to large effects for symptom reduction in primary outcome domains. Conclusions:From an evidence-based perspective, cognitive-behavioral therapy is currently the treatment of choice for anxiety and
depressive disorders in children and adolescents. Future research in this area will need to focus on comparing cognitive-
behavioral psychotherapy with other treatments, component analyses, and the application of exportable protocol-driven
treatments to divergent settings and patient populations. J. Am. Acad. Child Adolesc. Psychiatry, 2004;43(8):930959.
Key Words: outcome studies, children and adolescents with major depression and dysthymic disorder, children and
adolescents with anxiety disorder, literature review.
Due in part to a productive interplay between researchand clinical practice (Rutter, 1999), many clinical re-searchers now believe that cognitive-behavioral therapy
(CBT) administered within an evidence-based, multi-modal, multidisciplinary practice model is the psycho-therapeutic treatment of choice for youth withinternalizing disorders (Geddes et al., 1997; March and
Wells, 2003). In this context, the past 10 years wit-nessed the emergence of diverse, sophisticated, and em-pirically supported CBTs covering the range ofchildhood-onset anxiety and depressive disorders
(Bernstein and Shaw, 1997; Birmaher et al., 1996a,b).Using the tools of evidence-based medicine (EBM)(Sackett et al., 1997), this article provides a critical
review of CBT for these conditions. We do not addressobsessive-compulsive disorder and posttraumatic stressdisorder, for which recent critical reviews are available(Cohen et al., 2000; Franklin et al., 2002; March,1995), or bipolar disorder, for which cognitive-behavioral interventions are just now emerging (Mc-Clellan and Werry, 1997). The reader interested in ahow-to-do-it perspective may wish to pursue recentoverviews of CBT (Hibbs and Jensen, 1996; Reineckeet al., 2003) interventions for childhood-onset anxiety(Kendall et al., 1999, 2000, 2003; March and Mulle,1998; Rapee et al., 2000; Silverman and Kurtines,
1996) and depressive disorders (Brent et al., 1997;Clarke et al., 1990).
GUIDING THEORY
Although a comprehensive review of the theoreticalrationale of CBT is clearly beyond the scope of thisarticle (for a still cogent prcis, see Kendall, 1993; Ken-
Accepted November 17, 2003.
Drs. Compton, March, and Curry are with the Department of Psychiatry and
Behavioral Psychology, Duke University Medical Center, Durham, NC; Dr.
Brent is with Western Psychiatric Institute and Clinic, Pittsburgh, PA; Dr.
Albano is with New York University School of Medicine, NY; and Dr. Weersing
is with the Yale Child Study Center, New Haven, CT.
Correspondence to Dr. Compton, Duke University Medical Center, Box
3527, Durham, NC 27710; e-mail: [email protected].
0890-8567/04/430809302004 by the American Academy of Child
and Adolescent Psychiatry.
DOI: 10.1097/01.chi.0000127589.57468.bf
J. AM . A CA D. CH IL D AD OL ES C. PS YC HI AT R Y, 43 :8 , AU GU S T 20 04930
-
7/28/2019 cognitive therapy, 2004.pdf
2/30
dall and Panichelli-Mindel, 1995), a short overview isheuristically valuable. Historically, behavior therapy(the BT in CBT) evolved within the theoretical frame-
work of classical and operant conditioning, with cog-nitive interventions (the C in CBT) assuming a moreprominent role with the increasing recognition that
personenvironment interactions are mediated by cog-nitive processes (Van Hasselt and Hersen, 1993).Looked at in the context of situational and/or cognitiveprocesses, BT is sometimes referred to as nonmedia-tional (emphasizing the direct influence of situationson behavior) and CT as mediational (emphasizing thatthoughts and feelings underlie behavior). Hence, be-havioral psychotherapists work with patients to changebehaviors and thereby to reduce distressing thoughtsand feelings. Cognitive therapists work to first changethoughts and feelings, with improvements in func-tional behavior following in turn.
Although CBT is often referred to as a unitary treat-ment, it is actually a diverse collection of complex andsubtle interventions that must each be mastered andunderstood from the social learning perspective. Sub-sequently, a cognitive-behavioral case formulationguides the therapist in administering treatment tech-niques in a flexible manner for the patient presenting
with any one disorder or comorbid presentation ofmental disorders (for an overview of a modular ap-proach to CBT interventions, see Curry and Reinecke[2003]). Nonetheless, despite their seeming differences,cognitive-behavioral interventions typically share fivequalities: (1) adherence to the scientistclinicianmodel, whereby treatments are chosen based on dem-onstrated evidence or are applied within a case evalu-ation format to determine efficacy; (2) a thoroughidiographic assessment (e.g., functional analysis) of tar-get behaviors and the situational, cognitive, and behav-ioral factors that have established or are maintainingthe symptoms of interest (for a detailed overview ofhow to conduct a functional analysis, see Haynes andOBrien [1990]); (3) an emphasis on psychoeducation;(4) problem-specific treatment interventions designed
to ameliorate the symptoms of concern; and (5) relapseprevention and generalization training at the end oftreatment. For example, using cognitive restructuringand exposure-based interventions, CBT for anxiety dis-orders encourages cognitions and behaviors designed topromote habituation or extinction of inappropriatefears. Likewise, CBT for depression directly confronts
maladaptive depressogenic cognitions, including help-lessness, hopelessness, and hostility, and aims behavior-ally to reconstitute pleasant relationships, be theyintrapsychic, interpersonal, school, or spiritual. As evi-dence-based therapies, each is supported by a more orless robust research literature, and manuals are usually
available to guide practitioners in using CBT for spe-cific problems. Thus, CBT fits nicely into the currentmedical practice environment that appropriately valuesempirically supported, brief, problem-focused treat-ments.
From this vantage point, CBT represents a develop-mentally sound approach to pediatric mental illness.Children normally acquire social-emotional (self andinterpersonal) competencies across time. The failure todo so, relative to age, gender, and culture-matchedpeers, may reflect capacity limitations, individualdifferences in the rate of skill acquisition for specific
competencies, environmental factors, and/or the devel-opment of a mental illness. In CBT, the task of themental health practitioner is to understand the present-ing symptoms in the context of child-specific con-straints to normal development and to devise a tailoredtreatment program that eliminates those constraints sothat the youngster can resume a normal developmentaltrajectory insofar as is possible.
To the extent that symptom relief occurs, it can beassumed that improvement reflects concurrent changes(e.g., learning) in the CNS (Andreason, 1997; Hyman,2000). Thus, the cognitive-behavioral treatment of pe-diatric mental illness can be thought of as partiallyanalogous to the treatment of, for example, juvenile-onset diabetes, with the caveat that the target organ, thebrain in the case of major mental illness, requires in-terventions of much greater complexity. Althoughmedications are of importancein diabetes, insulin,and in the anxiety or affective disorders, a serotoninreuptake inhibitorthe critical point is that each alsoinvolves crucial psychosocial interventions that work inpart by biasing the somatic substrate of the disordertoward more normal functioning (Hyman, 2000). In
diabetes, the behavioral intervention of choice is dietand exercise, and in the anxiety or affective disorders, itis cognitive-behavioral psychotherapy.
METHOD
EBM has emerged as a promising paradigm for medical practice(for a comprehensive review, see Sackett et al. [2000]) and is clini-
R ES EA R C H U PD A T E R EV I EW : C B T
J. AM . AC A D. CH IL D A DO LE SC . P SY CH I AT RY , 43 :8 , A UG US T 20 04 931
-
7/28/2019 cognitive therapy, 2004.pdf
3/30
cally akin to the scientistpractitioner model in academic psychol-ogy (Barlow, 1993). EBM deemphasizes the more typical relianceon unsystematic clinical experience as a sufficient ground for clini-cal decision making. Instead, EBM stresses the examination of evi-dence from systematic diagnostic assessment technologies andclinical research as a tool to inform clinical practice, and it providesa heuristically valuable organizing focus for the individual clinicianseeking to transition efficacy and effectiveness studies into clinical
practice at the level of the individual patient (Geddes et al., 1997).Using established EBM criteria for assessing the validity of treat-
ment studies as guides to clinical practice (Guyatt et al., 1993,1994, 1999), a search for relevant literature was conducted viaMedline and PsycINFO, using the following text terms: anxiety,depression, cognitive therapy, and behavior therapy. Only random-ized, controlled trials (RCTs) for individuals with a specific disorderwere included. Additionally, to be included, articles must have metthe following criteria: published in an English-language, peer-reviewed journal between 1990 and 2002; included children be-tween the ages of 8 and 18; included an outcome measure of knownclinical significance; and used an analytic strategy consistent withthe study design. A follow-up assessment was preferred but notrequired. Excluded from consideration were articles concerning thetreatment of obsessive-compulsive disorder, posttraumatic stress
disorder, or bipolar disorder; included were articles concerning thetreatment of specific phobias, social phobia, selective mutism, over-anxious disorder, separation anxiety disorder, panic disorder, gen-eralized anxiety disorder, major depression, and dysthymia.
The text of this article is supported by a series of tables thatsummarize the main findings of each study identified during theliterature search. The tables are organized by type of disorder (anxi-ety versus depression); within each disorder, separate tables sum-marize findings at post-treatment and at long-term follow-up.
The information presented in each table includes study citation(studies are listed in alphabetical order by first author), researchdesign (type, control condition, analysis sample), sample informa-tion (total number, age range, percentage of males, and ethnicity),the diagnoses targeted by the intervention, brief details about theintervention, primary dependent measures (both categorical andscalar), sample size in each treatment condition, proportion ofsample responding, magnitude of the treatment effect (portrayed interms of number needed to treat [NNT] and standardized effectsize estimates), and general comments by the authors.
The NNT is a measure of the average response, presented as theprobability of response in single patient units. Arithmetically, theNNT is the inverse of the absolute risk reduction (1/ARR), definedas the percentage of response in the experimental group minus thepercentage of response in the control condition. In practice, NNTrepresents the number of patients who need to be treated with theactive treatment to produce one additional positive outcome be-yond that obtainable with the control or comparison condition. Forexample, an NNT of 10 describes the number of patients whom aclinician would need to treat with the active treatment rather thanthe control treatment to see one additional positive outcome. A verysmall NNT (that is, an NNT that approaches 1) suggests that a
favorable outcome occurs in nearly every patient who receives thetreatment and in relatively few patients in the comparison group.An NNT of 2 or 3 indicates that a treatment is quite effective.
Standardized effect size estimates were calculated with the assis-tance of ES (Shadish et al., 1999), a computer software programdesigned to calculate effect size estimates from published studies. EScalculates the standardized mean difference statistic, commonly re-ferred to as Cohens d and computed as d= (M
t M
c)/SD, where
Mt
is the mean of the treatment group, Mc
is the mean of thecomparison group, and SD is the pooled within-group standarddeviation. All effect size estimates are reported such that positivescores indicate that the treatment group improved more than thecomparison group.
TREATMENT OF ANXIETY DISORDERS
To their advantage, cognitive-behavioral therapistshave a robust literature validating the effectiveness ofspecific psychological techniques for anxiety disordersand a steadily growing literature supporting the use ofprescriptive treatment protocols for these disorders.
Types of Investigations
Twenty-one RCTs evaluating a variety of cognitive-behavioral interventions for the treatment of child andadolescent anxiety disorders were identified (Table 1).
As a group, these studies are noteworthy for their meth-
odological rigor and the systematic way in which theyhave advanced the understanding of childhood anxietydisorders and how best to treat this important popula-tion. With respect to methodological rigor, all studiesused contrasting group designs in which active treat-ments were compared with either a wait-list or no-treatment control condition (Cornwall et al., 1996;Hayward et al., 2000; Kendall, 1994; Kendall et al.,1997; King et al., 1998; Shortt et al., 2001; Silvermanet al., 1999a) or an attention placebo-controlled con-dition (Beidel et al., 2000; Last et al., 1998; Muris etal., 2002). Moreover, several studies compared more
than one active treatment condition (Barrett, 1998;Barrett et al., 1996; Beidel et al., 2000; Cobham et al.,1998; Flannery-Schroeder and Kendall, 2000; Mend-lowitz et al., 1999; Menzies and Clarke, 1993; Muris etal., 2001; Nauta et al., 2003; Silverman et al., 1999b;Spence et al., 2000).
Investigators in this area have also systematicallyevaluated a variety of clinically relevant questions: forinstance, whether group CBT is more effective thanindividual CBT (Flannery-Schroeder and Kendall,2000; Manassis et al., 2002; Muris et al., 2001), wheth-er adding parental participation enhances treatmentoutcomes (Barrett, 1998; Barrett et al., 1996; Cobhamet al., 1998; Mendlowitz et al., 1999; Nauta et al.,2003; Shortt et al., 2001; Spence et al., 2000), whetherconcurrent treatment of parental anxiety enhancestreatment outcomes (Cobham et al., 1998), and
whether two active treatment components, which are
C O M P TO N E T A L .
J. AM . A CA D. CH IL D AD OL ES C. PS YC HI AT R Y, 43 :8 , AU GU S T 20 04932
-
7/28/2019 cognitive therapy, 2004.pdf
4/30
TABLE
1
Ran
dom
ized
Clinic
alT
rials
ofC
BT
for
Child
and
Adole
scen
tA
nxiety
Diso
rders:
Eff
ects
atP
ost-trea
tmen
t
Aut
hor
(s)
Des
ign
Sample
Informat
ion
Target
Diagnos
is
Treatment
Informat
ion
Primary
Depen
dent
Measures
SampleSi
ze
(Initial/
C
ompleted
)
Proportion
Responding
Posttreatm
ent
EBMNN
T
EffectSize
Comments
Barrett,
1998
RCT
,al
ternative
treatmentand
WL
control,
blind
assessment,
completer
analys
is
N=607
14
years,53
%
male
Ethnicity
unspec
ified
OAD(n=30
)
SAD(n=26
)
SOP
(n=4)
12se
ssions,gr
oup
CBT-C
12se
ssions,gr
oup
family
CBT-C
P
WL
(12
wee
ks,
thenof
fered
treatment)
FSSC-R
CBCL
Noan
xietydx
23/19
17/15
20/16
11/19
11/15
4/16
4 2 FSSC-R
GCBT
=1.
58
GCBT+=
2.53
CBCL-I(mo
ther
)
GCBT
=3.
37
GCBT+=
3.98
Bot
htr
eatments
associatedw
ith
sign
ificant
improvements;
GCBT+
associatedw
ith
marginal
ly
betterou
tcomes
Barrett,
Dad
ds,&
Rapee,
1996
RCT
,W
Lco
ntrol,
blind
assessment,
completer
analys
is
N=79
9.3
years
57%male
Ethnicity
unspec
ified
OAD(n=30
)
SAP
(n=30
)
SOP
(n=19
)
12se
ssions,
indivi
dualC
BT
12se
ssions,
indivi
dual
CBT-C
P
WL
(12
wee
ks,
thenof
fered
treatment)
RCMAS
FSSC-R
CBCL
Noan
xietydx
28/28
25/25
26/23
16/28
21/25
6/23
3 2 RCMAS
CBT
=0.40
CBT+=0
.94
FSSC-R
CBT
=0.49
CBT+=0
.73
CBCL-I(mo
ther
)
CBT
=0.96
CBT+=1
.19
Bot
hac
tive
treatments
showedpo
sitive
benef
it,C
BT+
wassu
perioron
several
outcomes
Bei
del,
Turner,&
Morris,
2000
RCT;n
onspeci
fic
treatment
control,bli
nd
assessment,
completer
analys
is
N=67
10.5ye
ars;
40%male,
70%w
hite
SOP
(n=67
)
12in
divi
dualan
d
12gr
oup
sess
ions,C
BT
12in
divi
dualan
d
12gr
oup
sess
ions,
Nonspec
ific
treatment
control
SPAI-C
,
C-G
AS,
ADIS-C
CSR
,
Noan
xietydx
36/30
31/20
20/30
1/20
2 SPAI-C=0.
91
C-G
AS
=1.4
6
ADIS-C
CSR
=
2.04
Act
ivetr
eatment
wasas
sociated
with
sign
ificant
improvements
acrossm
ultiple
domains
cont
inue
d
R ES EA R C H U PD A T E R EV I EW : C B T
J. AM . AC A D. CH IL D A DO LE SC . P SY CH I AT RY , 43 :8 , A UG US T 20 04 933
-
7/28/2019 cognitive therapy, 2004.pdf
5/30
TABLE
1
cont
inue
d
Aut
hor
(s)
Des
ign
Sample
Informat
ion
Target
Diagnos
is
Treatment
Informat
ion
Primary
Depen
dent
Measures
SampleSi
ze
(Initial/
C
ompleted
)
Proportion
Responding
Posttreatm
ent
EBMNN
T
EffectSize
Comments
Cob
ham,
Dad
ds,&
Spence,
1998
RCT
,al
ternative
treatment
control,bli
nd
assessment,
ITT
analys
is
N=67
9.6
years
51%male
Ethnicity
unspec
ified
SAD(n=8)
OAD(n=3)
GAD(n=40
)
SIP
(n=12
)
SOP
(n=3)
AGP
(n=1)
10se
ssions,
child-focused
group
CBT
(parents
participated
)
10se
ssionspl
us4
parent
anxiety
management
sess
ions,gr
oup
CBT+P
AM
(parents
participated
);
outcomesw
ere
alsocr
osse
don
parental
anxiety
(nonanxious
parent
vs.
anxiouspa
rent)
RCMAS
STAIC
Noan
xietydx
(usingA
DIS-P
CSR)
35/35
32/32
(no
te:3
fam
iliesin
eac
hco
ndition
com
pletedle
ss
tha
n50
%of
the
sess
ions)
CBT
NAP
=
14/17
CBT
AP
=
7/18
CBT+P
AM
NAP
=12
/15
CBT+P
AM
AP
=13
/17
(Calcu
lations
madeusin
g
CBT
APas
control
condition)
CBT
NAP=
2
CBT+P
AM
NAP
=2
CBT+P
AM
AP
=3
ESs
wereno
t
calculated
on
self-report
measures
dueto
lowco
mpletion
rate
ofthe
assessment
instruments
(n=29
)
Children
with
anxious
parent(s)
respon
ded
less
favorablyto
child-focuse
d
CBT
at
post-treatment;
thead
ditionof
PAMimproved
outcomesfo
r
children
with
anxious
parent(s)bu
t
notfo
rch
ildren
with
nonanxious
parents
Cornwal
l,
Spence,&
Schotte,
1996
RCT
,W
Lco
ntrol,
blindness
unclear,
completer
analys
is
N=24
8.25
years
Gen
der
unspeci
fied
Ethnicity
unspeci
fied
SIP
(dar
kness
phob
ia,n=
24)
6se
ssions,em
otive
imagery
WL
(3m
onthsin
duration
)
FSSC-R
RCMAS
FT
DFBQ
12/12
12/?
Dataun
avai
lable
Datau
navailab
le
toca
lculate
NNT
FSSC-R=
0.53
RCMAS=
0.52
DFBQ=1.
59
Act
ivetr
eatment
associatedw
ith
sign
ificant
improvement
in
all
outcomes
Flannery-
Schroe
der
& Ken
dall,
2000
RCT
,W
Lco
ntrol,
blind
assessment,
ITT
and
comple
ter
analys
is
N=45
814
years
46%males
89%w
hite
GAD(n
=21
)
SAD(n=11
)
SOP
(n=5)
18se
ssions,
indivi
dualC
BT
18se
ssions,gr
oup
CBT
WL
(9w
eeks
,
thenof
fered
treatment)
RCMAS
CBCL-I
Noan
xietydx
(usingA
DIS-P
)
18/13
13/12
14/12
Noan
xiety
dx(I
TT)
8/18
6/13
0/14
2 2 RCMAS
ICBT
=0.7
9
GCBT
=1.
11
CBCL-I
ICBT
=1.5
2
GCBT
=0.
84
Bot
hin
divi
dual
and
group
CBT
wereas
sociated
with
lowerra
tes
ofan
xiety
disordersan
d
enhance
d
copingab
ilities;
outcomesfo
r
ICBT
and
GCBT
were
comparable
cont
inue
d
C O M P TO N E T A L .
J. AM . A CA D. CH IL D AD OL ES C. PS YC HI AT R Y, 43 :8 , AU GU S T 20 04934
-
7/28/2019 cognitive therapy, 2004.pdf
6/30
TABLE
1
cont
inue
d
Aut
hor
(s)
Des
ign
Sample
Informat
ion
Target
Diagnos
is
Treatment
Informat
ion
Primary
Depen
dent
Measures
Sam
pleSi
ze
(Initial/
Com
pleted
)
Proportion
Responding
Posttreatment
EBMNNT
Effect
Size
Comments
Haywar
det
al.,20
00
RCT
,no-treatment
control,bl
ind
assessment,
completer
analys
is
N=35
15.8ye
ars10
0%
females
Ethnicity
unspec
ified
SOP
(n=35
)
16
sess
ions,gr
oup
CBT
N
o-treatment
control
ADIS
CSR
,
SPAI-C
Noan
xietydx
12/11
23/22
6/11
1/22
2 ADIS-C=1.23
ADIS-P=0.
67
SPAI-C=0.
29
Act
ivetr
eatment
wasas
sociated
with
sign
ificant
improvement;
55%ofsu
bjects
continue
dto
meet
diagnostic
criter
iafo
rdx
Ken
dall,
1994
RCT
,W
Lco
ntrol,
blindness
unclear,
completer
analys
is
N=47
913
years
60%males
76%w
hite
OAD(n=30
)
SAD(n=8)
AVD(n=9)
16
sess
ions,
indivi
dualC
BT
W
L(8w
eeks
,
thenof
fered
treatment)
RCMAS
FSSC-R
CBCL-I
Nopr
imary
anxietydx
(usingA
DIS-P
)
27/NR
20/NR
(Note:13
sub
jects
dro
pped,
not
reported
bygr
oup)
17/27
1/20
2 RCMAS
=0.87
FSSC-R=0.
38
CBCL-I=1.22
Indivi
dualC
BT
wasas
sociated
with
lowerra
tes
ofan
xiety
disordersan
d
enhance
d
copingab
ilities
Ken
dall,
Flanney-
Schroe
der,
Pan
ichelli-
Mindel,
Sout
ham-
Gerowet
al.,19
97
RCT
,W
Lco
nro
l,
blindness
unclear,
completer
analys
is
N=94
913
years
62%male
85%w
hite
OAD(n
=55
)
SAD(n=22
)
AVD(n=17
)
16
sess
ions,
indivi
dualC
BT
W
L(8w
eeks
,
thenof
fered
treatment)
RCMAS
STAIC
Noan
xietydx
byA
DIS-P
75/60
43/34
32/60
2/34
2 RCMAS
-0.
59
STAIC-T
A=0.7
2
STAIC-S
A=0.
40
Overall,
resu
lts
wereve
ry
similar
toea
rlier
studyof
indivi
dualC
BT
Kingetal.,
1998
RCT
,W
Lco
ntrol,
blindness
unclear,
completer
analys
is
N=34
11.0
3ye
ars
53%male
Ethnicity
unspec
ified
SR(n=34
)
6
sess
ions(o
ver
4
wee
ks;pl
us5
parent
sess
ions
and
1te
acher
meeting),
indivi
dualC
BT
W
L(4w
eeks
,
thenof
fered
treatment)
Schoolat
tendance
(%days
present)
FT
GAF
CBCL-I
17/17
17/17
No.w
ho
achieved90
%
school
attendance
15/17
5/17
2 FT
=1.
38
GAF
=1.
50
CBCL-I=0.59
Act
ivetr
eatment
wasas
sociated
with
sign
ificant
improvements
onal
lou
tcomes
except
teac
her
reports
cont
inue
d
R ES EA R C H U PD A T E R EV I EW : C B T
J. AM . AC A D. CH IL D A DO LE SC . P SY CH I AT RY , 43 :8 , A UG US T 20 04 935
-
7/28/2019 cognitive therapy, 2004.pdf
7/30
TABLE
1
continued
Aut
hor
(s)
Des
ign
Sample
Informat
ion
Target
Diagnos
is
Treatment
Informat
ion
Primary
Depen
dent
Measures
Sam
pleSi
ze
(Initial/
Com
pleted
)
Proportion
Responding
Posttreatment
EBMNNT
Effect
Size
Comments
Last, H
ansen,
&Franco,
1998
RCT
,al
ternat
ive
treatment
control,
blindness
unclear,
completer
analys
is
N=56
12.0
4ye
ars,
40%male
89%w
hite
ASR
(n=56
)
12
sess
ions,
indivi
dualC
BT
12
sess
ions,
indivi
dualE
S
Schoolat
tendance
GIS
FSSC-R
STAIC-M
Noan
xietydx
32/20
24/21
No.w
ho
attained95
%
attendance
13/20
10/21
6 95%attendance=
0.39
Clinician
GIS=
0.20
FSSC-R=0.
49
STAIC-M=0.
31
Nodx=0.
39
Bot
htr
eatments
wereeq
ually
effect
ivein
returning
children
to
schoo
l
Manassiset
al.,20
02
RCT
,al
ternat
ive
treatment
control,
blindness
unclear,
completer
analys
is
N=78
9.98
years
54%male
85%w
hite
GAD(n=47
)
SAD(n=20
)
SIP
(n=5)
SOP
(n=5)
PAD(n=1)
12
sess
ions,
indivi
dualC
BT
12
sess
ions,gr
oup
CBT
(Note:
parents
participatedin
both
treatments)
MASC
CGAS
41/NR
37/NR
Categor
ical
outcomes
weren
ot
prov
ided
NA
Relat
iveto
indivi
dualC
BT
MASC=
0.31
CGAS
=
0.64
Bot
htr
eatments
wereas
sociated
with
improvements
onch
ildan
d
parent
ratings;
clinician
CGAS
ratingsfa
vored
indivi
dualC
BT;
indivi
dualC
BT
wasm
ore
effect
ivefo
r
children
reportingh
igh
ratesofso
cial
anxiety
Men
dlow
itz
etal.,
1999
RCT
,al
ternat
ive
treatment
control,
blindness
unclear,
completer
analys
is
N=68
9.8
years
43%male
Ethnicity
unspec
ified
Children
with
DSM-I
V
anxietydx
(using
DICA-R-P
)
12
sess
ions,gr
oup
CBT
(child
only)
12
sess
ions,gr
oup
CBT-P
12
sess
ions,gr
oup
CBT-C
P
W
L(2to6
months,th
en
offered
treatment)
RCMAS
CCSC
GIS
23/23
21/21
18/18
Categor
ical
outcomes
weren
ot
prov
ided
NA
RCMAS
CBT-C=0.1
8
CBT-P=0.1
8
CBT-C
P=0
.35
CCSC(Act
ive
Cop
ing)
CBT-C=0.2
6
CBT-P=
0.6
5
CBT-C
P=0
.57
CCSC(Avo
idant
Cop
ing)
CBT-C=0.3
3
CBT-P=0.3
9
CBT-C
P=0
.39
All
three
treatments
were
associatedw
ith
sign
ificant
improvements
insy
mptomsof
anxietyan
d
depression;
children
in
CBT
(child
+
parent)
condition
reportedu
sing
moread
aptive
copingsk
ills
thanth
eot
her
twotr
eatment
conditions
continued
C O M P TO N E T A L .
J. AM . A CA D. CH IL D AD OL ES C. PS YC HI AT R Y, 43 :8 , AU GU S T 20 04936
-
7/28/2019 cognitive therapy, 2004.pdf
8/30
TABLE
1
continued
Aut
hor
(s)
Des
ign
Sample
Informat
ion
Target
Diagnos
is
Treatment
Informat
ion
Primary
Depen
dent
Measures
SampleSi
ze
(Initial/
C
ompleted
)
Proportion
Responding
Posttreatm
ent
EBMNN
T
EffectSize
Comments
Menzies&
Clarke,
1993
RCT
,al
ternative
treatment
control,bli
nd
assessment,
completer
analys
is
N=48
5.5
years
65%male
Ethnicity
unspec
ified
SIP
(water
phob
ia,
n=48
)
3se
ssions,IV
VE
3se
ssions,V
E
3se
ssions,IV
E
No-treatment
control
BRS
PCWP
CWP
OR
13/12
13/12
12/12
Categor
ical
outcomes
weren
ot
prov
ided
NA
Unab
leto
calculated
ueto
insu
fficient
data
Bot
hIV
VE
and
IVE
were
equa
llyef
fect
ive
and
more
effect
iveth
an
WL
inre
ducing
waterph
obia;
IVE
resu
ltedin
greater
generalizat
ion
to
novelsi
tuat
ions;
VE
showedn
o
benef
itov
er
no-treatment
control
Mur
is,
Mayer,
Bartelds,
Tierney,
&Bog
ie,
2001
RCT
,al
ternative
treatment
control,
blindness
unclear,
com-
pleteran
aly
sis
N=36
9.9
years
25%male
97%w
hite
GAD(n
=14
)
SAD(n=14
)
SOP
(n=7)
OCD(n=1)
12se
ssions,
indivi
dualC
BT
12se
ssions,gr
oup
CBT
SCARED-R
STAIC
17/not
reported
19/not
reported
Categor
ical
outcomes
weren
ot
prov
ided
NA
Relat
iveto
indivi
dual
CBT
SCARED-R
(total
)=
0.32
STAIC(trait
anxiety)=
0.14
Bot
htr
eatments
wereas
sociated
with
equal
improvements
insy
mptomsof
anxiety
Mur
is,
Meesters,
&van
Mel
ick
2002
RCT
,psyc
hological
PBOan
d
no-treatment
control,
blindness
unclear,
completer
analys
is
N=30
10.2ye
ars
33%male
90%w
hite
SAD(n
=10
)
GAD(n=7)
SOP
(n=3)
Diagnosticst
atus
ofn
o-treat-
ment
controls
notas
sessed
12se
ssions,gr
oup
CBT
12se
ssions,gr
oup
ED
No-treatment
control
RCADS
STAIC
10/10
10/10
10/10
Categor
ical
outcomes
weren
ot
prov
ided
NA
Com
binedact
ive
treatment
relativeto
no-treatment
control
RCADS
(total
anxiety)
CBT
=1.48
ED=0.1
7
STAIC(trait
anxiety)
CBT
=0.83
ED=0.4
6
CBT
relative
to
ED
RCADS
(total
anxiety)=
0.98
STAID(trait
anxiety)=
1.05
CBT
wassu
perior
toE
Dan
d
no-treatment
control;E
D
showedno
benef
itov
er
no-treatment
control
continued
R ES EA R C H U PD A T E R EV I EW : C B T
J. AM . AC A D. CH IL D A DO LE SC . P SY CH I AT RY , 43 :8 , A UG US T 20 04 937
-
7/28/2019 cognitive therapy, 2004.pdf
9/30
TABLE
1
continued
Aut
hor
(s)
Des
ign
Sample
Informat
ion
Target
Diagnos
is
Treatment
Informat
ion
Primary
Depen
dent
Measures
SampleSi
ze
(Initial/
C
ompleted
)
Proportion
Responding
Posttreatm
ent
EBMNN
T
EffectSize
Comments
Nauta,
Scholing,
Emmel
kamp,
& Minderaa,
2003
RCT
,W
Lco
ntrol,
blind
assessment,
ITT
analys
is
N=79
11.0ye
ars
49%male
Ethnicity
unspec
ified
SAD(n=26
)
SOP
(n=31
)
GAD(n=15
)
PAD(n
=7)
12se
ssions,
indivi
dualC
BT
12se
ssions,
indivi
dualC
BT
plus7se
ssions
CPT
WL
(durat
ion
not
spec
ified,
then
offered
treatment)
SCAS-c/p
FSSC-R
CBCL
Noan
xietydx
(ADIS-Can
d
P)
29/26
30/30
20/17
Com
binedvs.
WL
32/59
2/18
CBT
vs.C
BT+
20/37
23/39
Act
ivetr
eatm
ent
relativeto
WL
2 Unab
leto
calculated
ueto
insu
fficient
data
CBT
relative
to
CBT+20
SCAS-c=
0
.20
SCAS-p
=0
.33
FSSC-R=
0.12
Relat
ivetoW
L,
activetr
eatment
showedlo
wer
scoresonpa
rent
reportsan
d
morech
ildren
diagnosticfr
ee;
nodi
fference
between
WL
and
active
treatment
on
child
reports;
thead
ditionof
CPT
showedn
o
additional
benef
itac
rossal
l
outcomes
Shortt,
Barrett,&
Fox,20
01
RCT
,W
Lco
ntrol,
blind
assessment,
completer
analys
is
N=71
7.85
years
41%male
92%
Austral
ian
GAD(n=42
)
SAD(n=19
)
SOP
(n=10
)
10se
ssions(p
lus2
booster
sess
ions),gr
oup
CBT
WL
(10
wee
ks,
thenof
fered
treatment)
RCMAS
CBCL
Noan
xietydx
54/48
17/16
33/48
1/16
2 RCMAS
=0.9
9
Mot
her
CBC
L-I=
5.08
Fat
her
CBCL-I=
1.91
Act
ivetr
eatment
wasas
sociated
with
sign
ificant
improvements
acrossal
l
outcomes
Silvermanet
al.,19
99a
RCT
,W
Lco
ntrol,
blind
assessment,
completer
analys
is
N=56
9.66
years
61%males
45%w
hite
GAD(n=12
)
SOP
(n=15
)
OAD(n=29
)
12se
ssions,gr
oup
CBT
(concurrent
child
and
parent
groups
with
15
min
.co
njoint
meeting)
WL
(8to10
wee
ks,th
en
offered
treatment)
RCMAS
FSSC-R
CBCL-I
PGRS
Noan
xietydx
(ADIS-C
/P)
37/25
19/16
16/25
2/16
2 RCMAS
=0.5
8
FSSC-R=0.6
5
CBCL-I=1.2
5
PGRS
=1.7
8
GroupC
BT
was
associatedw
ith
sign
ificant
improvements
acrossal
l
primary
outcome
domainss
continued
C O M P TO N E T A L .
J. AM . A CA D. CH IL D AD OL ES C. PS YC HI AT R Y, 43 :8 , AU GU S T 20 04938
-
7/28/2019 cognitive therapy, 2004.pdf
10/30
-
7/28/2019 cognitive therapy, 2004.pdf
11/30
often combined in traditional cognitive-behavioral pro-tocols (e.g., behavioral contingency management versuscognitive self-control), are differentially effective (Sil-verman et al., 1999b). Moreover, several of the studiescited were replications and extensions of existing pro-tocols by independent researchers (Barrett, 1998; Bar-
rett et al., 1996; Manassis et al., 2002; Mendlowitz etal., 1999; Muris et al., 2001, 2002).
Assessment Issues
Diagnosis and Symptom Profile. Valid and reliableassessment is essential to the skillful application andevaluation of cognitive-behavioral treatments (Thyer,1991) and is a strength of the cited studies taken as a
whole. All but 2 of the 21 studies cited in Table 1(Cornwall et al., 1996; Menzies and Clarke, 1993) usedsemistructured clinical interviews to identify subjects ashaving an anxiety disorder as well as documenting di-
agnostic comorbidities and assessing treatment out-comes. By a significant margin (13 of 21), the most
widely used semistructured clinical interview was theAnxiety Disorders Interview Schedule, Child and Par-ents Versions (ADIS-C/P) (Silverman, 1987; Silver-man and Albano, 1996a,b; Silverman and Nelles,1988). This interview is most commonly administeredseparately to children and parents, and then data arecombined from both sources to derive a final com-posite diagnosis; however, several studies deviatedfrom this standard practice and relied solely on infor-mation obtained from parents to determine diagnosticstatus (Shortt et al., 2001; Spence et al., 2000) andtreatment outcome (Cobham et al., 1998; Flannery-Schroeder and Kendall, 2000; Kendall, 1994; Kendallet al., 1997; Shortt et al., 2001; Spence et al., 2000). Inaddition to providing a diagnosis, the ADIS requiresthe clinician to provide a clinician severity rating(CSR). The CSR is the clinicians estimate of the de-gree of functional impairment and distress engenderedby the disorder (Albano and Silverman, 1996). Unfor-tunately, only two studies characterized the sample interms of the CSR (Hayward et al., 2000; Silverman et
al., 1999a). Because the CSR may predict the natureand outcome of treatment, the failure of researchers toadequately characterize the baseline characteristics oftheir sample along this dimension is a notable defi-ciency.
Demographics and Severity. Both genders are largelyrepresented in the treated population, with only one
study containing a sample that was limited to females(Hayward et al., 2000). Although the majority of stud-ies attempted to recruit children and adolescents, theaverage age of subjects across all studies was approxi-mately 9.85 years. This leaves open the question ofgeneralizability of the research findings, as well as pro-
tocol-driven interventions, to older adolescent popula-tions. Other demographic variables, such as ethnicityor socioeconomic status, were generally well docu-mented. However, with the exception of two trials (Sil-verman et al., 1999a,b), most studies had extremelylow rates of ethnic minority participation (see Pina etal. [2003] who examined the differential treatment re-sponse of Hispanic/Latino youth and European-
American youth). A noted strength of the citedinvestigations was the clinical severity of the researchsample. All studies focused on subjects who soughtclinical services and whose impairment was severe
enough to warrant a psychiatric diagnosis. No studyincluded children who were simply endorsing symp-toms of anxiety on a self-report measure.
Outcome Measures. To their credit, the majority ofcited investigations relied on a multimethod (e.g., clini-cal interview, self-report measures), multiinformant(e.g., child, parent, clinician) approach to documenttreatment outcomes. Both scalar and dichotomousmeasures that sampled specific symptom domains wereregularly reported. Another strength of many of thecited investigations was that outcomes were not re-stricted to the simple reporting of statistically signifi-cant symptom improvement or symptom change.More clinically informative outcomes were commonlyreported, such as clinically significant improvement(defined as changes that return deviant subjects to
within nondeviant limits [Kendall and Grove, 1988])and posttreatment diagnostic status (defined as the per-centage of children who no longer meet criteria for acurrent anxiety disorder). For instance, 14 of the 21investigations reported the posttreatment diagnosticstatus of subjects. However, the methods used to quan-tify diagnostic status varied moderately from study to
study, which made it difficult to compare outcomesacross trials. For instance, some studies combined in-formation obtained from separate child and parentclinical interviews to determine posttreatment diagnos-tic status (Barrett, 1998; Barrett et al., 1996; Beidel etal., 2000; Last et al., 1998; Nauta et al., 2003; Silver-man et al., 1999a,b), whereas others relied solely on
C O M P TO N E T A L .
J. AM . A CA D. CH IL D AD OL ES C. PS YC HI AT R Y, 43 :8 , AU GU S T 20 04940
-
7/28/2019 cognitive therapy, 2004.pdf
12/30
information obtained from the parent (Cobham et al.,1998; Flannery-Schroeder and Kendall, 2000; Kendall,1994; Kendall et al., 1997; Shortt et al., 2001; Spenceet al., 2000). Moreover, some studies defined a subjectas diagnosis free if criteria for his or her primary anxietydiagnosis were no longer met (Last et al., 1998),
whereas others used a more restrictive definition anddefined a subject as diagnosis free if criteria for both hisor her primary and secondary (if present) anxiety di-agnoses were no longer met (e.g., Barrett, 1998).
Moderators of Outcome. Ten of the 21 cited investi-gations reported results of secondary analyses that at-tempted to determine whether basic demographic andclinical variables moderated treatment outcome (e.g.,age, sex, ethnicity, clinical severity, pretreatment diag-nosis, comorbidities). The most frequent finding is thatnone of the variables analyzed moderate treatment out-come (for a notable exception, see Barrett et al.
[1996]). However, the strength of this conclusion mustbe tempered because few studies were sufficiently pow-ered to adequately address this important question.
Long-Term Follow-up. Although the follow-up pe-riod varied widely across the cited investigations (from3 months to 6 years, with a modal length of 12months), the general conclusion that can be reached isthat CBT for anxiety disorders in children and adoles-cents is a durable intervention (Table 2). With fewexceptions (Cobham et al., 1998), posttreatment gains
were largely maintained at follow-up and showed littledeterioration. Interestingly, several studies that foundsignificant differences between two active treatmentspost-treatment reported that, at follow-up, the twotreatments were equally effective. However, because allcited studies lacked an adequate control group duringthe follow-up period, competing explanations for thepositive results reported cannot be dismissed.
Treatments
The behavioral treatment of fear and anxiety in chil-dren builds on early studies indicating that anxiety isreadily conceptualized as a set of classically conditioned
responses that can be unlearned or counterconditionedthrough associative pairing with anxiety-incompatiblestimuli and responses. For example, in systematic de-sensitization (SD), anxiety-arousing stimuli are system-atically and gradually paired (imaginally or in vivo)
with competing stimuli such as food, praise, imagery,or cues generated from muscular relaxation. SD with
children consists of three basic steps: (1) training inprogressive muscle relaxation, (2) rank ordering of fear-ful situations from lowest to highest, and (3) hierarchi-cal presentation of fear stimuli via imagery while thechild is in a relaxed state (Eisen and Kearney, 1995).SD appears to work well with older children and ado-
lescents. Younger children, however, often have diffi-culty with both obtaining vivid imagery and acquiringthe incompatible muscular relaxation. Strategies suchas using developmentally appropriate imagery and ad-
junctive use of workbooks may boost the effectivenessof these procedures with younger children.
Without encouragement, anxious children and ado-lescents often find it difficult to remain in the presenceof anxiety-arousing stimuli for a sufficient length oftime to allow habituation to occur in the natural en-vironment. In fact, in some cases, the process of nega-tive reinforcement maintains the anxiety response.
That is, when an individual initially confronts an anxi-ety-provoking situation (e.g., the assignment of an oralreport for the socially anxious youth), there is an in-crease in discomforting sensations and anxiousthoughts (e.g., rapid heart rate, sweating, thoughts suchas Ill look stupid to others). By escaping or avoidingthe situation, such as through complaints of feeling illand needing to leave class or the behavior of schoolavoidance/refusal, the individual feels immediate relieffrom the anxiety. This is the process of negative rein-forcement. The escape behavior is reinforced by therelief and sets the stage for cycles of anxiety arousalfollowed by escape or avoidance and relief.
After the adult treatment literature, the identifica-tion of the negative reinforcement paradigm led to thedevelopment of exposure-based interventions for a
wide range of pediatric anxiety disorders. Because es-cape and avoidance behaviors are negatively reinforcedby the cessation of anxiety, exposure-based proceduresrequire extended presentation of fear stimuli with con-current prevention of escape and avoidance behaviorsin order for the extinction of the conditioned responsesto occur. Unlike systematic desensitization, stimulus
presentation is not accompanied by progressive musclerelaxation. Rather, graduated imaginal and/or in vivoexposure to hierarchically presented fear stimuli is usedto attenuate anxiety to phobic stimuli. Gradual expo-sure, with the consent of the child, is generally consid-ered to produce less stress for the client (and therapist)and thus is often preferred over the use of more pre-
R ES EA R C H U PD A T E R EV I EW : C B T
J. AM . AC A D. CH IL D A DO LE SC . P SY CH I AT RY , 43 :8 , A UG US T 20 04 941
-
7/28/2019 cognitive therapy, 2004.pdf
13/30
TABLE
2
Ran
dom
ized
Clinic
alT
rialsof
CBT
for
Child
and
Adole
scen
tA
nxietyD
isor
ders:
Eff
ects
atF
ollo
w-u
p
Aut
hor
(s)
Fol
low-up
Citat
ion
Fol
low-up
Des
ign
Sample
Size
Prim
ary
Dependent
Measures
Treatment
Con
ditions
SampleS
ize
(Initial/F
U)
Proportion
Responding
FU
EBMNNT
Effect
Size
Comments
Barrett
et
al.,19
96
Inor
iginal
article
12-moF
U
(act
ive
treatments
only)
EN=79
FUN=53
RCMAS,
FSSC-R,
CBCL
,
independent
clinician
ratings;
noanx
iety
dx
12se
ssions,in
divi
dual
CBT
12se
ssions,in
divi
dual
child
and
parent
CBT
28/27
25/23
19/27
22/23
NA
NA
CBT+ha
dsi
gnificantly
morech
ildren
diagnosis
free,lo
werF
SSC-R
scores,an
dhi
gher
clinician
ratingsof
improvement;
no
sign
ificant
differenceon
CBCL;yo
unger
children
and
femalesre
spon
ded
bettertoC
BT+
Barrett
et
al.,19
96
Barrett
et
al.,20
016-yr
FU
EN=79
FUN=52
RCMAS,
FSSC-R,
noanx
iety
dx
12se
ssions,in
divi
dual
CBT
12se
ssions,in
divi
dual
child
and
parent
CBT
28/31,
18/21
(includes
only
subjects
who
me
tdx
statusa
t
pretreatment
bych
ild
interview
)
24/28
18/21
NA
NA
12-motr
eatment
gainsw
ere
largelym
aintainedat6-
yr
FU;co
ntrarytoau
thors
pred
ictions,C
BT+w
as
notm
oreef
fect
iveth
an
CBT
Barrett,
1998
Inor
iginal
article
12-moF
U
(act
ive
treatments
only)
EN=60
FUN=di
fficult
tode
termine
FSSC-R,C
BCL
,
independent
clinician
ratings,
noanx
iety
dx
12se
ssions,gr
oup
CBT
(child
only)
12se
ssions,gr
oup
family
CBT
(child
and
parent)
Difficu
ltto
determine
GCBT
=64.5
%,
GCBT+=
84.8
%
NA
NA
Bot
hac
tivetr
eatment
groups
continuedto
showim
provement;
no
sign
ificant
difference
between
2ac
tive
treatments
ondi
agnostic
status;G
CBT+gr
oup
reportedsi
gnificantly
lowerF
SSC-Rsc
oresan
d
CBCL
scores;G
CBT+
rece
ivedsi
gnificantly
higher
clinician
ratingsof
improvement
continued
C O M P TO N E T A L .
J. AM . A CA D. CH IL D AD OL ES C. PS YC HI AT R Y, 43 :8 , AU GU S T 20 04942
-
7/28/2019 cognitive therapy, 2004.pdf
14/30
TABLE
2
continued
Aut
hor
(s)
Fol
low-up
Citat
ion
Fol
low-up
Des
ign
Sample
Size
Prim
ary
Dependent
Measures
Treatment
Con
ditions
SampleS
ize
(Initial/F
U)
Proportion
Responding
FU
EBMNNT
Effect
Size
Comments
Barrett
et
al.,19
96
Inor
iginal
article
6-moF
U(act
ive
treatments
only)
EN=79
FUN=53
RCMAS,
FSSC-R,
CBCL
,
independent
clinician
ratings,
noanx
iety
dx
12se
ssions,in
divi
dual
CBT
12se
ssions,in
divi
dual
child
and
parent
CBT
28/28
25/25
20/28
21/25
NA
NA
Bot
hac
tivetr
eatment
groups
continuedto
showim
provement;
no
sign
ificant
difference
between
2ac
tive
treatments
ondi
agnostic
status,R
CMAS,
FSSC-R,
orC
BCL
scores;C
BT+
rece
ivedsi
gnificantly
higher
clinician
ratingsof
improvement
Bei
del
et
al.,20
00
Inor
iginal
article
6-moF
U
EN=67
FUN=22
(children
in
then
onspec
ific
treatment
condition
were
NA
for
FU
analys
is
SPAI-C,
CGAS,
ADIS-CCSR
,
noanx
iety
dx
12in
divi
dualan
d12
group
sess
ions,
CBT
12in
divi
dualan
d12
group
sess
ions,
nonspec
ific
treatment
control
36/22
19/22
NA
NA
Treatment
gainsw
ere
maintainedat6-
moF
U
Cob
hamet
al.,19
98
Inor
iginal
article
12-moF
U
EN=67
FUN=65
Noan
xiety
dx
10se
ssions,
child-focusedgr
oup
CBT
(parents
participated
)
10se
ssions+4pa
rent
anxiety
management
sess
ions,gr
oup
CBT=P
AM(parents
participated
);
groups
wereal
so
crosse
donpa
rental
anxietyN
AP
vs.A
P
33/35
32/32
CBT
NAP
=
12/16
CBT
AP
=
10/17
CBT+P
AM
NAP
=12
/15
CBT+P
AM
AP
=12
/17
NA
NA
Overall,
treatment
effects
wea
kenedby12
-moF
U;
nosi
gnificant
main
effect
for
anxietyco
ndition
(anxiouspa
rent
vs.
nonanxiouspa
rent);n
o
sign
ificant
main
effectfo
r
treatment
condition
(CBT
vs.C
BT+P
AM);
nosi
gnificant
interact
ions
between
parent
anxious
statusan
dtr
eatment
condition
continued
R ES EA R C H U PD A T E R EV I EW : C B T
J. AM . AC A D. CH IL D A DO LE SC . P SY CH I AT RY , 43 :8 , A UG US T 20 04 943
-
7/28/2019 cognitive therapy, 2004.pdf
15/30
TABLE
2
continued
Cob
hamet
al.,19
98
Inor
iginal
article
6-moF
U
EN=67
FUN=66
Noanx
iety
dx
10se
ssions,
child-focusedgr
oup
CBT
(parents
participated
)
10se
ssions
+4pa
rent
anxiety
management
sess
ions,gr
oup
CBT+P
AM(parents
participated
);
groups
wereal
so
crosse
donpa
rental
anxietyN
AP
vs.A
P
34/35
32/32
CBT N
AP
=14
/16
CBT
AP
=8/
18
CBT+P
AM
NAP
=12
/15
CBT+P
AM
AP
=12
/17
NA
NA
Children
with
anxious
parent(s)co
ntinue
dto
respon
dle
ssfa
vorably
toch
ild-focusedC
BT;
overal
l,ch
ildren
with
nonanxiouspa
rents
respon
ded
more
favorablytotr
eatment
regardlessoftr
eatment
condition
Cornwal
l
etal.,
1996
Inor
iginal
article
3-moF
U
EN=24
FUN=24
FSSC-R
,R
CMAS,
FT,D
FBQ
6se
ssions,em
otive
imagery
WL
(3m
oin
duration
)
12/12
12/12
NR
NA
FSSC-R=0.
90
RCMAS
=0.
79
DFBQ=1.
82
Treatment
gainsinth
e
activetr
eatment
condition
were
maintainedat3-
moF
U
Flannery-
Schroe
der
& Ken
dall,
2000
Inor
iginal
article
3-moF
U
EN=45
FUN=29
(includes
subjects
treated
after
WL)
RCMAS
,C
BCL-I,
noan
xietydx
18se
ssions,in
divi
dual
CBT
18se
ssions,gr
oup
CBT
WL
(9w
k,
thenof
fered
treatment)
18/14
18/15
Primarydx
:
11/14
8/15
Any
Anxdx
:
7/14
8/15
NA
NA
Treatment
gainsw
ere
maintainedat3-
mo
FU;n
osi
gnificant
differencesbe
tween
the
twoac
tivetr
eatments
onse
lf-report
and
parent
report
measures
Haywar
d
etal.,
2000
Inor
iginal
article
12-moF
U
EN=35
FUN=28
ADISC
SR,SP
AI
Noanxietydx
16se
ssions,gr
oup
CBT
Notr
eatment
control
12/10
23/18
4/10
10/18
6 SP
AI
=0.
07
Nosi
gnificant
between-group
differenceinra
tesof
socialph
obia
orSP
AI
mean
scoresat12
-mo
FU;ad
ditionalan
alyses
combiningso
cial
phob
iaan
dde
pression
diagnosespr
oduced
morero
bustbe
tween
group
treatment
changes
Ken
dall,
1994
Inor
iginal
article
12-moF
U
EN=47
FUN=38
(includesSs
treated
after
WL
period
)
RCMAS
FSSC
-R
CBCL-I
Noprimary
anxiety
dx
(ADIS-P)
16se
ssions,in
divi
dual
CBT
WL
(8w
k,th
en
offered
treatment)
47/38
Percent
dx-free
notre
ported
NA
NA
Treatment
gainsw
ere
maintainedat12
-mo
FUonse
lf-report
and
parent
report
measures
continued
C O M P TO N E T A L .
J. AM . A CA D. CH IL D AD OL ES C. PS YC HI AT R Y, 43 :8 , AU GU S T 20 04944
-
7/28/2019 cognitive therapy, 2004.pdf
16/30
TABLE
2
continued
Aut
hor
(s)
Fol
low-up
Citat
ion
Fol
low-up
Des
ign
Sample
Size
Primary
Dep
endent
Me
asures
Treatment
Con
ditions
Sample
Size
(Initial/
FU)
Proportion
Responding
FU
EBMNNT
Effect
Size
Comments
Ken
dall,
1994
Ken
dall
&
Sout
ham-
Gerow,
1996
25
yrF
U
EN=47
FUN=36
(includesSs
treated
after
WL
period
)
RCMAS
FSSC-R
CBCL-I
Noprim
ary
anxiety
dx
(ADIS-C)
Indivi
dualC
BT
47/36
Percent
dx-free
notre
ported
NA
NA
Treatment
gainsw
ere
largelym
aintainedat
long-termFUon
self-report
and
parent
report
measures
Ken
dall
et
al.,19
97
Inor
iginal
article
12-moF
U
EN=94
FUN=85
(includesSs
treated
after
WL
period
)
RCMAS
STAIC
Noanx
iety
dxby
ADIS
-P
16se
ssions,in
divi
dual
CBT
WL
(8w
k,th
en
offered
treatment)
85/94
Percent
dxfr
ee
notre
ported
NA
NA
Posttreatment
reduct
ions
werem
aintainedat
12-moF
Uw
ith
the
exception
thatC
BCL-I
(mot
her)
ratingsw
ere
sign
ificantlylo
wer
Kingetal.,
1998
Inor
iginal
article
3-moF
U
EN=34
FUN=17
(WL
not
assessed
)
Schoolat
tendance
(1%days
present)
FT
GAF
CBCL-I
6se
ssions(o
ver
4w
k;
plus5pa
rent
sess
ionsan
d1
teac
her
meeting),
indivi
dualC
BT
WL
(4w
k,th
en
offered
treatment)
17/17
No.w
ho
achieved90
%
schoo
l
attendance
14/17
NA
NA
Treatment
gainsac
rossal
l
primaryou
tcomesw
ere
maintainedat3-
moF
U
Last
etal.,
1998
Inor
iginal
article
2w
kin
toth
e
subsequent
schoolye
ar
EN=56
FUN=41
%repor
ting:
(1)
nod
ifficu
lty
returningto
schoolinne
w
year
(2)
mild
difficulty
(3)
mod
erate
difficu
lty
(4)
extreme
difficu
lty
12se
ssions,in
divi
dual
CBT
12se
ssions,in
divi
dual
educat
ionalsu
pport
32/20
24/21
CBT
vs.E
S
(1)
40%vs.
52%
(2)
30%vs.
19%
(3)
10%vs.5%
(4)
20%vs.
24%
NA
NA
Rough
ly30
%of
treatment
completersin
both
groups
reported
moderatetose
vere
difficulty
returningto
schoolth
efo
llow
ing
schoolye
ar
Last
etal.,
1998
Inor
iginal
article
4-w
kF
U
EN=56
FUN=29
%repor
ting:
(1)
main
tained
improvement
(2)
show
edfu
rther
improvement
(3)
relap
sed
(4)
neve
r
improve
d
12se
ssions,in
divi
dual
CBT
12se
ssions,in
divi
dual
educat
ionalsu
pport
32/14
24/15
CBT
vs.E
S
(1)
65%vs.
40%
(2)
14%vs.
13%
(3)
7%vs.7%
(4)
14%vs.
40%
4 NA
Them
ajor
ityofSs
continuedtosh
ow
improvement,
with
no
sign
ificant
between
group
differencesat
4-w
kF
U
Menzies&
Clarke,
1993
Inor
iginal
article
12-w
kF
U
EN=51
FUN=36
(WL
not
assessed
)
BRS
PCWP
CWP
OR
3se
ssions,IV
VE
3se
ssions,V
E
3se
ssions,IV
E
Notr
eatment
control
13/12
13/12
13/12
NA
NA
NA
Nonsign
ificant
deteriorat
ion
in
treatment
gainsn
oted,
IVVE
group
performed
betterth
anIV
Egr
oup
atF
U
continued
R ES EA R C H U PD A T E R EV I EW : C B T
J. AM . AC A D. CH IL D A DO LE SC . P SY CH I AT RY , 43 :8 , A UG US T 20 04 945
-
7/28/2019 cognitive therapy, 2004.pdf
17/30
TABLE
2
continued
Aut
hor(s)
Fol
low-up
Citat
ion
Fol
low-up
Des
ign
Sample
Size
Primary
Depen
dent
Measures
Treatment
Con
ditions
Sample
Size
(Initial/FU)
Proportion
Responding
FU
EBMNNT
Effect
Size
Comments
Nauta
et
al.,20
03
Inor
iginal
article
3
-moF
U
EN=79
FUN=73
SCAS-c/p
FSSC-R
CBCL
Noan
xietydx
12se
ssions,in
divi
dual
CBT
12se
ssions,in
divi
dual
CBT
plus7
sess
ionsC
PT
WL
(durat
ion
not
spec
ified,
then
offered
treatment)
37/34
39/39
23/34
27/39
0 Relat
ivetoC
BT
SCAS-c=
0.31
SCAS-p
=
0.24
FSSC-R=
0.24
Bot
hac
tivetr
eatment
groups
continue
dto
showim
provement
acrossal
lpr
imary
outcomesdu
ringF
U
period;th
ead
ditionof
CPT
conferredno
additionalbe
nef
it
acrossal
lou
tcomes
Shortt
et
al.,20
01
Inor
iginal
article
1
2-moF
U
EN=71
FUN=63
(includesSs
treated
after
WL
period
)
RCMAS
CBCL
Noan
xietydx
10se
ssions(p
lus2
booster
sess
ions),
group
CBT
WL
(10
wk,
then
offered
treatment)
47/63
32/47
NA
NA
Clinician
ratingsw
ere
maintainedatF
U,
RCMAS
scoresw
ere
sign
ificantlylo
werth
an
atpo
st-treatment
Silverman
etal.,
1999a
Inor
iginal
article
3
-,6-,12
-mo
FU(resu
lts
fromeachF
U
assessment
period
presented
and
analyzed
together
)
EN=56
FUN=31
(includes
pooledG
CBT
and
WL
data)
RCMAS
FSSC-R
CBCL-I
PGRS
Noan
xietydx
12se
ssions,gr
oup
CBT
(concurrent
child
and
parent
groups
with
15
min
.co
njoint
meeting)
WL
(8to10w
k,th
en
offered
treatment)
3-moFU
=
41/31
6-moFU
=
41/33
12-moFU
=
41/25
3-moF
U=
24/31
6-moF
U=
26/33
12-moF
U=
19/25
NA
NA
Overall
pattern
ofre
sults
showedala
rgepr
e-to
posttreatment
change
followedbygr
adua
lbu
t
continue
dim
provement
acrossal
lpr
imary
outcomesdu
ringF
U
period
Silverman
etal.,
1999b
Inor
iginal
article
3
-,6-,12
-mo
FU(resu
lts
fromeachF
U
assessment
period
presented
and
analyzed
together
)
EN=10
4
FUN=15
%
unava
ilablefo
r
FUassessments
RCMAS
FSSC-R
PGRS
Noan
xietydx
10se
ssions,in
divi
dual
child
and
parent
SC
10se
ssions,in
divi
dual
child
and
parent
CM
10se
ssions,in
divi
dual
child
and
parent
ES
NR
bytreatme
nt
group
NR
NA
NA
Overall
pattern
ofre
sults
showedala
rgepr
e-to
posttreatment
change
followedbygr
adua
lbu
t
continue
dim
provement
acrossal
lpr
imary
outcomesdu
ringF
U
period
Spence,
Donovan,
& Brechman-Toussaint,
2000
Inor
iginal
article
1
2-moF
U
EN=50
FUN=36
(num
ber
of
dropouts,if
any,w
eren
ot
spec
ified)
ADIS-PC
SR
RCMAS
Noan
xietydx
(ADIS
-P)
12se
ssions,ch
ildan
d
parent
group
CBT
12se
ssions,ch
ildon
ly
group
CBT
Notr
eatment
WL
16/17
17/19
13/16
9/17
NA
NA
Treatment
gainsw
ere
largelym
aintainedat
12-moF
Uacrossal
l
primaryou
tcomes;
invest
igatorsm
odified
ADIS
tofi
tDSM-I
V
criter
ia;on
lypa
rents
werein
terviewed;on
ly
phonein
terviews
conductedfo
rpo
stan
d
follow-up
assessments.
Results
for
self-report
and
behavioral
measuresal
sore
ported
continued
C O M P TO N E T A L .
J. AM . A CA D. CH IL D AD OL ES C. PS YC HI AT R Y, 43 :8 , AU GU S T 20 04946
-
7/28/2019 cognitive therapy, 2004.pdf
18/30
scriptive techniques, especially massed exposure orflooding.
Cognitive interventions, usually combined with ex-posure, also play a prominent role in CBT for anxiouschildren and adolescents. For example, Kendall andcolleagues developed a comprehensive cognitive-
behavioral protocol for anxious youth that focuses ontransmitting coping skills to children in need (Kendall,1994; Kendall et al., 1997). Based on the premise thatanxious children view the world through a templateof threat, automatic questioning (e.g., What if . . .),and behavioral avoidance, treatment is focused on pro-viding educational experiences to build a new copingtemplate for the child. Therapists assist the children toreconceptualize anxiety-provoking situations as prob-lems to be solved and situations with which to cope. Avariety of cognitive-behavioral components assist thetherapist and child in building the coping template:
relaxation training, imagery, correcting maladaptiveself-talk, problem-solving skills, and managing rein-forcers. Therapists use coping modeling, role-play re-hearsals, in vivo exposure, and a collaborativetherapeutic relationship with the child to facilitate thetreatment progress. As a rule, parents are actively in-volved in all facets of treatment as collaborators in thechange process.
For example, when significant others are trapped inthe childs anxiety symptoms, it is crucial that they stopparticipating in or reinforcing the childs avoidancestrategies or rituals. To test the hypothesis that addinga family anxiety management component would boosttreatment effectiveness, Barrett et al. (1996) developeda parallel family program to Kendalls Coping Catbased on behavioral family intervention strategiesfound effective for the treatment of externalizing dis-orders in youth. After the completion of each childsession with the therapist, the child and parents wouldparticipate in a family anxiety management session
with the therapist. The crux of the program is to em-power parents and children by forming an expertteam to overcome and master anxiety. Parents are
trained in reinforcement strategies, with an emphasison differential reinforcement and systematic ignoringof excessive complaining and anxious behavior. How-ever, unilateral extinction strategies, such as when aparent returns the school-phobic child to school byforce, have significant disadvantages relative to consen-sual child involvement: (1) lack of a workable strategy
TABLE
2
continued
Note:
ADIS
=A
nxiety
Dis
orders
Interv
iew
Schedule
;A
DIS
CSR
=A
nxiety
Dis
orders
Interv
iew
Schedule
forC
hildren
,cl
inician
seve
rity
rating
(sum
maryscore);
ADIS
-C=A
nxiety
Dis
orders
Interv
iew
Sched
ule
for
Children;
ADIS
-CCSR
=A
nxiety
Dis
orders
Interv
iew
Schedule
for
Children
,cl
inician
seve
rity
ratingch
ildba
sed;
ADIS
-P=A
nxiety
Dis
orders
Interv
iew
Sch
edule
for
Childr
en,P
aren
tV
ersion;A
DIS
-PC
SR=
Anxie
tyD
isor
ders
Interv
iew
Schedule
for
Children
,clinician
seve
rity
rating
parent
base
d;B
RS
=B
ehav
iour
Rat
ing
Scale;
CBCL
=C
hild
Behav
iorC
heck
list;
CBCL
-I=C
hild
Behav
ior
Che
cklist-
Internal
izin
gSu
bsca
le;
CBT
=co
gnitiv
e-behav
iora
lth
erapy;
CBT
AP
=C
BT
with
anx
ious
parent;
CBT
NAP
=
CBT
with
nonanxious
parent;C
BT
+PAM
=C
BT
plus
parenta
lanx
iety
managemen
t;C
BT
+PAMAP
=C
BT
plusp
aren
talanx
iety
managemen
tw
ithanxiousparen
t;C
BT
+PAMNAP
=
CBT
plus
parenta
lan
xiety
managemen
tw
ith
nonanxious
parent;
CM
=co
ntinge
ncy-
managemen
tth
erapy;
CWP
=W
ater
Phobi
aSu
rvey
Schedule
,Child
Ve
rsion;
DFBQ
=D
arkn
ess
Fea
rB
ehav
iour
Quest
ionna
ire;
EN
=en
try
num
ber;
ES
=ed
ucat
ionsu
ppor
t;F
SSC
-R=F
ear
Survey
Schedule
for
Children-Rev
ised;
FT
=fe
arther
mometer;
FU
=fo
llow-u
p;F
UN
=
follo
w-u
pnu
mber;
GAF=
Global
Assessmen
tofF
unct
ion
ing;
GCBT
=gr
oup
cogn
itiv
e-beh
aviora
lth
erapy;
IVE
=invi
voex
posu
re;
IVVE
=in
vivo
exposure
plus
vica
rious
exposu
re;
CGAS
=C
hildrens
Global
Assessmen
tSc
ale;
NA
=no
tav
ailabl
e;NR=
not
reported;O
R=ov
eral
lreact
ion
toph
ob
icsi
tuat
ion;
PCWP
=W
ater
Phobi
aSu
rvey
Schedule
,Paren
tV
ersion;
PGRS
=pa
rent
global
rating
ofse
verity;R
CMAS
=R
evis
edC
hildrens
Man
ifes
tA
nxiety
Scales;S
C=se
lf-con
trolt
herapy;S
PAI
=So
cial
Phobi
aan
dA
nxietyIn
ventory;S
PAI-C
=So
cial
Phobi
aan
dA
nxiety
Inventory
for
Children;
VE
=vi
carious
exposu
re;
WL
=w
ait-
list
control.
R ES EA R C H U PD A T E R EV I EW : C B T
J. AM . AC A D. CH IL D A DO LE SC . P SY CH I AT RY , 43 :8 , A UG US T 20 04 947
-
7/28/2019 cognitive therapy, 2004.pdf
19/30
for managing the childs distress, (2) disruption of thetreatment relationship, (3) inability to target symptomsthat are out-of-sight for parents and teachers, and (4),most important, failure to help the child internalize amore skillful strategy for coping with current and po-tential future anxiety symptomatology.
MAJOR DEPRESSION
At any one time, approximately 1 in 20 children andadolescents suffers from major depressive disorder, withrates of depression rising dramatically in adolescents,especially in girls. Although the economic burden ofdepression in youth is uncertain, the human burden isconsiderable, especially with teenage suicide. Hence, itis of critical importance to note that the empirical lit-erature is more supportive for problem-specific psycho-therapies, especially CBT, than for medicationmanagement of pediatric depressive disorders (Birma-
her et al., 1996a; Hoberman et al., 1996). In particular,several controlled trials have demonstrated that indi-vidual or group administered cognitive-behavioral psy-chotherapy is an effective treatment for depressedyouth (Brent et al., 1997; Lewinsohn et al., 1994), andsome investigators now consider CBT to be the treat-ment of choice for this disorder (Reinecke et al., 1998).
Types of Investigations
Twelve articles describing a variety of cognitive-behavioral intervention packages for the treatment ofchild and adolescent depression were identified (Table3). Although these depression trials are equally meth-odologically rigorous when compared with child andadolescent anxiety trials (e.g., contrasting group designscomparing one or more active treatments with eitherno treatment, wait-list, or attention placebo controls),the number of studies is significantly fewer, and theresearch agenda to date has been less coherent andsystematic. Moreover, several of the studies with nullfindings likely had insufficient power to detect a be-tween-group treatment effect due to the small samplesize of each treatment condition. This is a notable de-
ficiency and contributes to the widely held notionamong practitioners that all treatments for depressionare equally effective. It also makes it difficult, if notimpossible, to reach strong conclusions regarding thedifferential efficacy of the treatments evaluated.
Two studies addressed the question of whether add-ing a separate treatment module for parents incremen-
tally improves outcomes (Clarke et al., 1999;Lewinsohn et al., 1990). One study compared indi-vidual CBT to systemic behavioral family therapy(Brent et al., 1997). Another study evaluated the rel-evant question of whether adding CBT to usual care ina health maintenance organization is better than usual
care alone (Clarke et al., 2002). Five studies evaluatedthe efficacy of one or more CBT interventions in de-signs that included either an attention placebo condi-tion (Kahn et al., 1990; Liddle and Spence, 1990;Vostanis et al., 1996; Wood et al., 1996) or a no-treatment control (Weisz et al., 1997). One studycompared individual CBT with interpersonal psycho-therapy (Rossello and Bernal, 1999). One investigationevaluated the effects of maintenance CBT for depressedadolescents (Clarke et al., 1999). One study evaluatedthe acceptability and efficacy of a combined cognitive-behavioral family education treatment (Asarnow et al.,
2002). Finally, one study evaluated the efficacy of cog-nitive bibliotherapy for adolescents with mild to mod-erate depressive symptoms (Ackerson et al., 1998). Nopublished investigations compared components oftreatments, and there were no systematic replicationstudies by independent investigators.
Assessment Issues
Diagnosis and Symptom Profile. Six of the 12 studiesused semistructured clinical interviews to identify sub-
jects as having DSM major depressive disorder or dys-thymia (Brent et al., 1997; Clarke et al., 1999, 2002;Lewinsohn et al., 1990; Vostanis et al., 1996; Wood etal., 1996). The most commonly used interview was theSchedule for Affective Disorders and Schizophrenia forSchool-Age Children (Chambers et al., 1985; Orvas-chel and Puig-Antich, 1986; Puig-Antich and Cham-bers, 1978). The remaining six studies either failed tomention the specific assessment procedures used to de-termine inclusion criteria (Rossello and Bernal, 1999)or enrolled subjects solely on the basis of mild to mod-erate levels of self-reported depressive symptomatology(Ackerson et al., 1998; Asarnow et al., 2002; Kahn et
al., 1990; Liddle and Spence, 1990; Weisz et al., 1997).The same six investigations that used semistructuredclinical interviews also assessed comorbidity but failedto analyze whether comorbidity status was related totreatment outcome. Thus, failure to systematically as-sess the impact of comorbidity on outcome is a criticaldeficiency in both the anxiety and depression literature.
C O M P TO N E T A L .
J. AM . A CA D. CH IL D AD OL ES C. PS YC HI AT R Y, 43 :8 , AU GU S T 20 04948
-
7/28/2019 cognitive therapy, 2004.pdf
20/30
TABLE
3
Ran
dom
ized
Clinic
alT
rials
ofCB
Tfo
rC
hild
and
Adole
scen
tD
epress
ive
Diso
rders:
Eff
ects
atP
ost-trea
tmen
t
Aut
hor
(s)
Des
ign
Sample
Informat
ion
Target
Diagnos
is
Treatment
Info
rmat
ion
Primary
Depen
dent
Measures
SampleS
ize
(Initial/Comple
ted)
Proportion
Responding
Posttreatment
EBMNNT
Effect
Size
Comments
Ackerson
etal.,
1998
RCT
,W
L
control,
unblinded
assessment,
completer
analys
is
N
=30
1
5.9
yr
3
6%male
6
5%w
hite
Adolescents
with
mild
to
moderate
symptomsof
depression
4w
kto
complete
self-g
uided
CBT
bibliot
herapy
WL
(4
wk,
then
offered
treatment)
CDI
HAM-D
NormalC
DI
NormalH
AM-D
15/12
15/10
Categor
ical
outcomesn
ot
reported
NA
CDI
=1.
05
HAM-D
21=2.
57
CBT
bibliotherapy
superiortoW
Lac
ross
multiplem
easures;
parent
measure
of
depressionsh
owedn
o
sign
ificant
between-group
differences
Asarnowet
al.,20
02
RCT
,W
L
control,
blindness
unclear,
completer
analys
is
N
=234t
hto
6th
graders
3
5%male
5
7%w
hite
Children
with
elevated
symptomsof
depression
9se
ssions,ch
ild+
group
CBT
,+1
sessio
nfa
mily
group
CBT
and
psych
oeducation
5w
kW
L
CDI
Notre
ported
(1ch
ildha
d
missing
data,no
t
reported
by
group)
Categor
ical
outcomesn
ot
reported
NA
CDI
=0.
92
When
outlier
removed,
CBT
showedsu
perior
efficacytoW
Lon
multiplem
easures
(depress
ion,n
egat
ive
thoughts,co
ping)
Brent
et
al.,19
97
RCT
,alternat
ive
treatment
control,
blind
assessment,
ITT
analys
is
N
=10
7
1
5.6
yr
2
4%male
8
3%w
hite
DSM-I
IIRMDD
12
16se
ssions,
indiv
idua
lC
BT
12
16se
ssions,
SBFT
12
16se
ssions,
NST
BDI
Nom
ood
dxan
d
norma