Cognitive-Behavioral Therapy for Adult Anxiety Disorders in Clinical Practice: A Meta-Analysis of Effectiveness Studies

Abstract

The efficacy of cognitive-behavioral therapy (CBT) for anxiety in adults is well established. In the present study, the authors examined whether CBT tested under well-controlled conditions generalizes to less-controlled, real-world circumstances. Fifty-six effectiveness studies of CBT for adult anxiety disorders were located and synthesized. Meta-analytic effect sizes are presented for disorder-specific symptom measures as well as symptoms of generalized anxiety and depression for each disorder, and benchmarked to results from randomized controlled trials. All pretest-posttest effect sizes for disorder-specific symptom measures were large, suggesting that CBT for adult anxiety disorders is effective in clinically representative conditions. Six studies included a control group, and between-groups comparisons yielded large effect sizes for disorder-specific symptoms in favor of CBT. Benchmarking indicated that results from effectiveness studies were in the range of those obtained in selected efficacy trials. To test whether studies that are more representative of clinical settings have smaller effect sizes, the authors coded studies for 9 criteria for clinical representativeness. Results indicate an inverse relationship between clinical representativeness and outcome, but the magnitude of the relationship is quite small.

Full text

Cognitive–Behavioral Therapy for Adult Anxiety Disorders in Clinical
Practice: A Meta-Analysis of Effectiveness Studies
Rebecca E. Stewart and Dianne L. Chambless
University of Pennsylvania
The efficacy of cognitive– behavioral therapy (CBT) for anxiety in adults is well established. In the
present study, the authors examined whether CBT tested under well-controlled conditions generalizes to
less-controlled, real-world circumstances. Fifty-six effectiveness studies of CBT for adult anxiety
disorders were located and synthesized. Meta-analytic effect sizes are presented for disorder-specific
symptom measures as well as symptoms of generalized anxiety and depression for each disorder, and
benchmarked to results from randomized controlled trials. All pretest–posttest effect sizes for disorder-
specific symptom measures were large, suggesting that CBT for adult anxiety disorders is effective in
clinically representative conditions. Six studies included a control group, and between-groups compar-
isons yielded large effect sizes for disorder-specific symptoms in favor of CBT. Benchmarking indicated
that results from effectiveness studies were in the range of those obtained in selected efficacy trials. To
test whether studies that are more representative of clinical settings have smaller effect sizes, the authors
coded studies for 9 criteria for clinical representativeness. Results indicate an inverse relationship
between clinical representativeness and outcome, but the magnitude of the relationship is quite small.
Keywords: effectiveness, dissemination, clinical practice, cognitive– behavioral therapy, anxiety disorders
Supplemental materials: http://dx.doi.org/10.1037/a0016032.supp
Cognitive– behavioral therapy (CBT) appears prominently
among the empirically supported treatments (ESTs) for adult anx-
iety disorders (Chambless & Ollendick, 2001). Several meta-
analyses of well-controlled clinical trials provide support for the
efficacy of CBT for panic disorder, social anxiety disorder,
obsessive– compulsive disorder (OCD), generalized anxiety disor-
der (GAD), and posttraumatic stress disorder (PTSD; see Deacon
& Abramowitz, 2004, for a review). In the most recent meta-
analysis, Norton and Price (2007) examined the efficacy of CBT
across the anxiety spectrum. Results indicated that treatments that
used CBT techniques showed significantly larger treatment out-
come effect sizes than no treatment or placebo across all of the
anxiety disorders. Taken together, these multiple meta-analyses
indicate that CBT is an efficacious treatment for adult anxiety
disorders.
How well does CBT for anxiety disorders hold up in actual
clinical practice? This general question of the transportability of
efficacious interventions into naturalistic settings has been one of
the most contentious issues in the ongoing debate of evidence-
based practice in clinical psychology (e.g., Jacobson & Chris-
tensen, 1996). Skeptics question whether the procedures used to
maximize experimental control in randomized controlled efficacy
trials seriously compromise the external validity of the results.
Specifically, there are questions about patients, clinicians, and
treatments used in research settings, and whether these character-
istics of clinical trials are representative, appropriate, or relevant to
routine clinical practice.
The claim is often made that research treatments will not work in
clinical practice settings because the clients in practice settings are
purported to be more severe or to have more comorbid conditions than
clients treated in research settings. According to this argument, in
research settings patients are recruited specifically for research, and
patients with comorbid disorders are often excluded to achieve ho-
mogeneous diagnostic samples (Westen & Morrison, 2001). It has
been suggested that these highly selected groups are not representative
of patients who present in outpatient practice (Silberschatz in Persons
& Silberschatz, 1998). It has also been suggested that patient assent to
randomization further limits the generalizability of the sample (Selig-
man, 1995). Moreover, patient expectations of specialist treatment in
a research trial might be higher than patient expectations in nonre-
search settings (Sanderson, Raue, & Wetzler, 1998), which may
further enhance motivation and outcome in research settings. There is
disagreement in the literature as to whether populations used in
randomized controlled efficacy trials are in fact as selected as as-
sumed (see Stirman, DeRubeis, Crits-Christoph, & Brody, 2003;
Stirman, DeRubeis, Crits-Christoph, & Rothman, 2005). Nonetheless,
Editor’s Note. Rick E. Ingram served as the action editor for this arti-
cle.—AML
Rebecca E. Stewart and Dianne L. Chambless, Department of Psychol-
ogy, University of Pennsylvania.
We thank David B. Wilson and Betsy Becker for assistance with
statistical analyses. We also thank John Paul Jameson for coding the
studies and Amber Calloway for literature searches, as well as the many
respondents to our listserv queries and those authors who provided addi-
tional data.
Correspondence concerning this article should be addressed to Rebecca
E. Stewart, Department of Psychology, University of Pennsylvania, So-
lomon Laboratories, 3720 Walnut Street, Philadelphia, PA 19104. E-mail:
restewar@psych.upenn.edu
Journal of Consulting and Clinical Psychology © 2009 American Psychological Association
2009, Vol. 77, No. 4, 595–606 0022-006X/09/$12.00 DOI: 10.1037/a0016032
595

this suggestion that patients in research settings are less severe, more
motivated, and somehow easier may limit the degree to which it is
believed that results from research settings can generalize to actual
clinical practice.
Another concern about the transportability of ESTs to real world
settings is the treatments themselves and the clinicians who pro-
vide them. Treatment protocols in randomized controlled trials are
manualized and strictly monitored with an emphasis on treatment
integrity. Therapy manuals are less likely to be used in clinical
practice, and their relevance to practice has been questioned
(Seligman, 1995). Furthermore, front-line practitioners typically
do not have access to the level of intensive training, monitoring,
and supervision available to therapists in research settings
(Chambless & Hollon, 1998). Clinicians in research settings are
more likely to be expert in the administration of particular treat-
ments and are motivated through adherence measures to stay
consistent with the protocol. Moreover, research therapists often
have the luxury of focusing exclusively on one type of problem or
disorder, whereas average practitioners carry large caseloads cov-
ering a wide range of focal problems. In summary, treatments
delivered in naturalistic settings may not be as rigorous in terms of
content or quality, and this may limit how well results of controlled
research trials can generalize to actual clinical practice.
Whether one believes these criticisms are valid or significant
(for responses, see Chambless & Ollendick, 2001; Persons in
Persons & Silberschatz, 1998), one major implication is that the
opponents of controlled psychotherapy research may be uncon-
vinced that efficacy findings are applicable in actual clinical prac-
tice. One solution to the perceived shortcomings of the traditional
controlled conditions of efficacy research is the treatment effec-
tiveness study (Hoagwood, Hibbs, Brent, & Jensen, 1995). Effec-
tiveness research explores the transportability of efficacious inter-
ventions (such as ESTs) to real-world service settings, to examine
whether these treatments result in similar, beneficial effects when
used in more naturalistic settings. Whereas efficacy studies focus
on minimizing threats to a study’s internal validity and determin-
ing the causal factors of therapeutic change, the emphasis in
effectiveness studies is placed on maximizing external validity.
Effectiveness studies focus on the effects of psychotherapy con-
ducted in the field, and can include pretest–posttest, quasi-
experimental, or experimental designs. External validity is
achieved by utilizing one or more of the following clinically
representative qualities: clinically representative settings (e.g., pri-
vate practice or mental health centers), clinically representative
therapists (e.g., practicing clinicians for whom provision of ser-
vices is a substantial part of the job), or clinically representative
patients (e.g., few exclusion criteria or patients who refuse ran-
domization).
It is often thought that efficacy and effectiveness studies are
mutually exclusive. However, it is more productive and accurate to
consider them as studies with different foci on internal and exter-
nal validity operating on a continuum (Hunsley & Lee, 2007). For
example, there is no reason why controlled efficacy trials cannot
take place in applied clinical settings with minimal exclusion
criteria and clinically representative therapists as hybrid
effectiveness– efficacy studies maximizing both internal and ex-
ternal validity (Chambless & Hollon, 1998; for an example, see
Blomhoff et al., 2001). Efficacy trials offer a particularly compel-
ling means of testing for causal agency, allowing for confident
conclusions regarding the efficacy of active treatments in compar-
ison with control treatments. However, demonstration of efficacy
is considered only a first step in treatment research (Argras &
Berkowitz, 1980). Effectiveness studies are required to demon-
strate the transportability and generalization of efficacious inter-
ventions into actual clinical practice.
Since the 1995 special section in the Journal of Consulting and
Clinical Psychology highlighted the importance of effectiveness
research, the literature on outcome research in clinical settings has
burgeoned. In an early review, Weisz, Donenberg, Han, and Weiss
(1995) proposed a collection of variables that pertain to clinical
relevance and utilized these criteria to locate nine clinically rep-
resentative child and adolescent therapy studies (see also Weisz,
Weiss, & Donenberg, 1992). Weisz et al. concluded the effective-
ness of clinic therapy was modest or nonsignificant when com-
pared with research therapy. Weisz et al. intended their conclusion
to hold only for child and adolescent therapy studies, and they also
noted that their results should be interpreted with caution given the
small number of studies they located. Moreover, it is critical to
note that Weisz et al. combined treatments of all types and disor-
ders of all types, and this finding may not hold when clinic and
research therapy are compared for a specific treatment (i.e., CBT)
for specific disorders (i.e., anxiety disorders).
Shadish, Matt, Navarro, and Phillips (2000) conducted a meta-
analysis of 90 therapy outcome studies drawn from published
meta-analyses located in a literature and including Weisz et al.’s
(1995) original nine studies. Building upon and expanding Weisz
et al.’s criteria for clinically representative studies, the authors
utilized a graduated scale of clinical representativeness. Their 10
criteria were based on use in past research (e.g., Weisz et al.,
1995), consistency with empirical literature on clinical practice,
and face validity. The criteria were as follows:
(a) clinically representative problems, (b) clinically representative
setting, (c) clinically representative referrals, (d) clinically represen-
tative therapists, (e) clinically representative structure, (f) clinically
representative monitoring, (g) clinically representative problem het-
erogeneity, (h) pretherapy training, (i) therapy freedom, and (j) flex-
ible number of sessions. (Shadish et al., 2000, p. 514)
Using multiple regression to predict effect size from the clinical
representativeness scale total score, Shadish et al. (2000) found
that after controlling for confounds—such as therapy dose and
outcome specific measures— clinical representativeness was unre-
lated to effect size. The authors concluded that this study supports
the effectiveness of psychotherapy under clinically representative
conditions.
We now return to our original question: How well does CBT for
anxiety disorders in adults hold up in actual clinical practice?
Although important and informative, the inferences that may be
drawn from Shadish et al.’s (2000) study for these questions are
limited because effectiveness studies of CBT of adult anxiety
disorders were not the primary focus of this work. Similar to Weisz
et al. (1995), Shadish et al. included psychotherapy treatments of
all types and disorders of all types. Accordingly, the first goal of
the present study is to conduct a meta-analysis of effectiveness
studies for anxiety disorders to determine whether the benefits of
CBT tested under well-controlled circumstances generalize to less-
controlled, more real-world circumstances. We report analyses of
pretest–posttest effect sizes as well as group contrast effect sizes
596 STEWART AND CHAMBLESS

for those studies that included a control group. We then use a
benchmarking strategy to assess whether the pretest–posttest effect
sizes achieved in effectiveness studies are comparable with effect
sizes obtained in controlled outcome efficacy trials. The second
goal of the present study is to expand on the research of Shadish
et al. by testing whether the degree of clinical representativeness is
related to the effect size of outcome.
Method
Studies
Studies included in this meta-analysis utilized CBT for any adult
anxiety disorder encompassed under the current Diagnostic and
Statistical Manual of Mental Disorders (4th ed.; American Psy-
chiatric Association, 1994) nomenclature. CBT was defined
broadly and included any treatment with cognitive, behavioral
(e.g., exposure), or a combination of components. Effectiveness
studies utilizing brief therapy (fewer than six standard sessions)
and transdiagnostic CBT were excluded because the efficacy of
these forms of CBT for anxiety disorders has not yet been well
established in efficacy studies. Although their effectiveness is an
important question in its own right, bibliotherapy and computer-
directed therapy were excluded because they were considered to be
too different from what goes on in actual clinical practice to be
pertinent to the goals of the present article. We also excluded
treatments that used psychotropic medication as part of the treat-
ment protocol because our interest was in the effectiveness of
psychosocial interventions.
We located studies via a search of abstracts in PsycINFO using
the following keywords: effectiveness, generalization, dissemina-
tion, naturalistic, transporting, private practice, managed care set-
ting, outpatient clinic, community clinic, community mental health
center, cognitive– behavioral therapy, cognitive therapy, and be-
havior therapy. In addition, the major journals publishing effec-
tiveness studies were checked by hand from 1995 to 2008: Behav-
ior Therapy,Behaviour Research and Therapy,Cognitive
Research and Therapy,Cognitive and Behavioural Practice,Jour-
nal of Anxiety Disorders, and the Journal of Consulting and
Clinical Psychology. The year 1995 was selected as the lower limit
for journal hand searching because the influential special section
on effectiveness research in the Journal of Consulting and Clinical
Psychology was published in that year. This special section defined
effectiveness studies, highlighted its importance for psychotherapy
research, and initiated a new direction in the field.
To avoid any studies being missed because of the heterogeneity
of descriptor and keyword items, as well as to locate any unpub-
lished work, conference presentations, or works in progress or
press, we sent networking e-mails to the electronic mailing lists of
the following societies: Association of Behavioral and Cognitive
Therapies, Society for the Science and Practice of Clinical Psy-
chology, Society for Psychotherapy Research, and the Academy of
Cognitive Therapy. Lastly, reference sections of located articles
and other relevant chapters and papers were reviewed for poten-
tially eligible studies. Fifty-four potential studies were located
from the PsycINFO, journal, and reference searches. Fifteen po-
tential studies were located through networking. Thirteen studies
were excluded on the basis of these criteria: 6 studies did not meet
our minimum clinically representative cutoff of three (see below),
and 7 studies did not provide sufficient data to include in the
meta-analysis. Efforts were made to contact the authors of these
studies, but (a) data were not available from the authors, or (b) we
were unable to contact or received no response from the authors.
In sum, a total of 56 studies were included in these analyses: 17 for
panic disorder; 11 each for social anxiety disorder, OCD, and
GAD; and 6 for PTSD. One study (Westbrook & Kirk, 2005)
included data on CBT outcomes for all disorders with the excep-
tion of PTSD, and therefore it is counted more than once. No
effectiveness studies were located for specific phobias.
Participants. Available reported patient characteristics of each
study are compiled and presented in the supplemental materials on
the journal’s website. In brief, the majority of patients were female
(range ⫽37%–100%, unweighted Mdn ⫽68.3%) and in their
mid-30s (range ⫽31–71 years, unweighted Mdn ⫽35 years). On
average, half or less of the sample had a college education
(range ⫽0%–71%, unweighted Mdn ⫽33%), and the majority of
the patients were employed full time (range ⫽8%– 88%, un-
weighted Mdn ⫽59%). When reported, Axis I comorbidity was
common (range ⫽32%– 85%, unweighted Mdn ⫽55.4%). Al-
though the average patient was Caucasian, African Americans or
Caribbean Americans of African descent made up at least 20% of
the sample in six studies. Latinos were represented at this level in
only two studies.
Coding clinical representativeness. Codes were based on Shad-
ish et al. (2000) and modified for the present study. Four of Shadish
et al.’s original codes were excluded in the present study on the basis
of practicality and theory. By nature of this study’s focus on effec-
tiveness studies with anxiety disorders, all studies would meet criteria
for clinically representative problems (a). Alternatively, no studies
would meet criteria for clinically representative heterogeneity (g) or
therapy freedom (i) because we only included data from anxiety
disorder patients treated with CBT. We also excluded Shadish et al.’s
criteria of flexible number of sessions (j) because we do not agree that
a flexible number of sessions is necessarily clinically representative.
Managed care often poses strict limitations on the number of
sessions a patient may receive. Moreover, given the great percent-
age of uninsured Americans, many patients pay out of pocket,
which may also restrict the number of sessions they can afford and
receive. We modified clinically representative structure to include
whether strict, flexible, or no manualization was utilized. We added three
criteria to the remaining six, on the basis of their usage in effectiveness
studies and partly for their face validity: no randomization, clinically
representative patients (i.e., no exclusion criteria aside from psychosis,
suicidality, organic brain disease, or substance abuse), and allowance of
medication.
1
The resultant nine criteria are as follows: clinically repre-
sentative settings, clinically representative referrals, clinically representa-
tive therapists, clinically representative structure, clinically representative
monitoring, no pretherapy training of therapists, no randomization, clin-
1
Although we excluded studies that used medication as part of the treat-
ment protocol, it is common for anxiety patients who present in clinical
practice to be on varying levels of psychotropic medications, as prescribed by
psychiatrists or general medical practitioners. As a result, we included the
allowance of medication as a criterion for clinical representativeness to express
this feature of clinical practice. In contrast, in most randomized controlled
trials of psychological treatments, patients are withdrawn from medication
before initiation of the study protocol or required to maintain a stable dosage
throughout treatment.
597
CBT IN CLINICAL PRACTICE

ically representative patients, and allowance of medication. Scores on
these criteria were summed to yield a total clinical representativeness
score for each study. A coding manual was developed and is presented in
the Appendix.
As noted earlier, studies fall on a continuum from efficacy to
effectiveness in nature. Any determination of where on that con-
tinuum a study must fall to be classified as an effectiveness study
is necessarily arbitrary. For the purposes of this study, we selected
an a priori cutoff score of three on the clinical representativeness
scale constructed. Such a score would be achieved, for example, if
a study was conducted in a clinically representative setting, with
clinically representative patients treated by clinically representa-
tive therapists.
Coding of the clinically representativeness criteria followed a
rigorous examination of the methods in each study. Many of the
studies reviewed were clear in their explanations of these charac-
teristics. In the few cases in which the information was not
reported, we contacted the study authors. Rebecca E. Stewart
coded all studies, and a second coder independently coded 24 out
of the 56 studies. Reliability for the total score was excellent,
␳
I
(3,1) ⫽.83 (Shrout & Fleiss, 1979). The codes for each study are
available in the supplemental materials on the journal’s website.
Effect Size Calculation and Statistical Procedures
Standardized mean gain. Standardized mean gain (pretest–
posttest) effect sizes were computed for diagnosis-specific out-
come measures. In addition, because generalized anxiety and de-
pression symptoms are common complaints of patients with
anxiety disorders and were often assessed, we computed effect
sizes for these measures as well. For panic disorder, there were
three disorder-specific symptom constructs: frequency of attacks,
fear of fear, and avoidance measures. For OCD, social anxiety
disorder, and PTSD, there was one disorder-specific construct. In
the special case of GAD, generalized anxiety measures were used
as diagnosis-specific outcome measures. Because of the paucity of
intent-to-treat data (4 studies out of 56), completer data were used
in this study. We calculated Cohen’s dfor the pretest–posttest
effect sizes using the pooled standard deviation (see Dunlap,
Cortina, Vaslow, & Burke, 1996).
The effect size was adjusted to yield Hedges’s g(Hedges, 1981)
and weighted to account for sample size. The weights were based
on the standard errors of effect size (Lipsey & Wilson, 2001). The
standard error formula for repeated measures requires the use of
the correlation rbetween pretest and posttest measures. This value
was never reported in the studies, although it is possible to derive
a value of rfrom the means, standard deviations, and the paired
t-test value with the following formula (D. Wilson, personal com-
munication, February 27, 2008):
r⫽
共s1
2t2⫹s2
2t2兲⫺共X1⫺X2兲2
2s1s2t2.
However, only one fourth of the studies reported a paired t-test
between pretest and posttest conditions. These rs were calculated,
converted to z, averaged, and converted back to rto deduce an
overall correlation rto be used in following calculations (r⫽.41).
We also did a sensitivity analysis by repeating the analyses with
correlation rs of .2 and .6, and we found that the results did not
differ substantially. The results of the sensitivity analysis are
presented in the supplemental materials on the journal’s website.
On the basis of the averaged r, the standard error for each effect
size for each study was calculated as follows (Lipsey & Wilson,
2001):
SE ⫽
冑
2共1⫺r兲
n⫹ES2
2n.
Lastly, we calculated the weights of each effect size using the
inverse variant weight, which is the reciprocal of the squared
standard error:
w⫽1
SE2.
Given the heterogeneity of the sample (see below), a priori random
effects meta-analyses proceeded as follows. The effect size, stan-
dard error, and inverse variance weights were calculated for each
construct measured in the study: the disorder-specific constructs,
as well as generalized anxiety and depression. The weighted mean
effect size for each construct was computed for each disorder,
according to the formula:
ES ⫽
¥共wiESi兲
¥wi
.
The standard error of each weighted mean effect size was also
calculated:
SE ⫽
冑
1
¥wi
.
Individual studies often reported multiple measures on a given
construct. Multiple measures on one construct would violate as-
sumptions of independence, inflate the sample size, and distort
standard error estimates. Therefore, a single effect size was cal-
culated for each construct for each study by averaging the multiple
measures to result in a single effect size for each construct for each
study.
Homogeneity analysis. Statistical tests based on the Qstatistic
(Hedges & Olkin, 1985) and I
2
(Higgins & Thompson, 2002)
indicated significant heterogeneity among panic disorder, social
anxiety disorder, and PTSD effect sizes. This was not surprising
given differences in methods across these studies, such as the lack
of common measures. Accordingly, we adopted a random effects
model for the analyses. Random effects analyses have the advan-
tage of allowing generalization to the potential population of
studies. Heterogeneity analyses for OCD and GAD indicated that
the distributions of observed effect sizes were homogenous. How-
ever, to gain the greater generalizability to a potential population
of studies permitted by the random effects approach, random
effects models were adopted for all analyses.
Standardized mean difference. The standardized mean differ-
ence effect size was calculated from posttreatment data to evaluate
between-groups differences for those studies that included a con-
trol group. This analysis was completed to examine whether CBT
treatment groups in clinically representative studies yield signifi-
cantly improved outcomes when compared with control groups.
These included waiting list (n⫽3), treatment as usual (n⫽2), and
contact control (n⫽1) groups. Hedges’s gwas calculated as
598 STEWART AND CHAMBLESS

before. A positive sign indicates the effect size favors the CBT
treatment group over the control group, whereas a negative sign
denotes that the control group has an advantage over the treatment
group. The standard error for these analyses was computed as
follows (Lipsey & Wilson, 2001):
SE ⫽
冑
ng1⫹ng2
ng1ng2
⫹ES2
2共ng1⫹ng2兲.
The inverse variance weights were calculated as above. Heteroge-
neity analyses based on the Qstatistic and I
2
indicated significant
heterogeneity, and random effects meta-analyses proceeded as
described above. Because there were only six studies that included
a control group, we collapsed across disorder and used the
disorder-specific symptom measures for each construct for each
disorder. In the case of panic disorder (for which there were three
constructs), these constructs were averaged to yield one disorder-
specific construct.
Selection of benchmarking studies. We adopted a benchmark-
ing strategy to assess the transportability of CBT in tightly con-
trolled experimental studies to clinically representative studies.
The strategy allows us to determine whether the magnitude of
improvement in clinically representative studies is comparable
with that obtained in research settings. We selected three efficacy
studies per disorder to use as benchmarks against which to com-
pare our standardized mean gain (pretest–posttest) effect sizes. We
culled all appropriate efficacy studies from the most recent meta-
analysis of CBT for adult anxiety disorders (Norton & Price, 2007)
and determined the sample size for each study. Many studies
included multiple treatment conditions that meet our definition of
CBT treatment (e.g., applied relaxation, stress inoculation train-
ing). In these cases, we included all CBT-like treatment conditions,
and we utilized a weighted average (dependent on sample size)
across treatments to yield one effect size for each disorder. Be-
cause larger samples provide the most stable estimates of effect
size, we selected for each disorder the three studies with the largest
samples—provided that the study yielded the appropriate data—
and calculated the pretest–posttest effect sizes for completer anal-
yses in the manner reported above. For each disorder, this yielded
a range of effect sizes from randomized controlled studies against
which to benchmark our results.
Clinical representativeness. The second goal of the present
study was to examine the relationship of clinical representative-
ness to outcome. One disorder-specific pretest–posttest effect size
was calculated for each study as described above. In the case of
panic disorder (for which there were three constructs), these con-
structs were averaged to yield one disorder-specific construct. We
used an a priori linear contrast to examine the relationship between
clinical representativeness and effect size (Rosenthal & Rosnow,
1991).
File Drawer Problem
One concern for meta-analyses is the so-called file drawer
problem, which describes systematic upward bias due to the omis-
sion of studies that have been conducted but never reported be-
cause of trivial or null results (Rosenthal, 1979). Our networking
strategy to locate unpublished work is one potential resolution to
the file drawer problem. Another solution is to examine the scat-
terplot of effect size by sample size. Examination of the funnel plot
revealed a classic funnel pattern, indicating greater variability in
effect sizes from studies with smaller samples. This suggests that
the effect sizes come from an unbiased distribution and that small
sample data with negative results have not been suppressed (Light,
Singer, & Willett, 1994).
Rosenthal (1979) responded to the file-drawer problem by de-
veloping a statistic to estimate the threat posed by potential un-
published studies. The fail-safe Nestimates the number of unpub-
lished studies reporting null results that would be necessary to
reduce the cumulative effect across the meta-analytic studies to the
point of nonsignificance. Orwin (1983) calculated the formula for
application to standardized mean difference effect sizes. This
formula can be applied to our pretest–posttest meta-analysis (Lip-
sey & Wilson, 2001), where k
0
is the number of effect sizes with
a value of 0 needed to reduce the mean effect size to ES
c,
the
criterion effect size level. ES
k
is the weighted mean effect size, and
kis the number of studies utilized to calculate the mean effect size.
k0⫽k
冋
ESk
ESc
⫺1
册
.
The criterion effect size level is the effect size at which we would
no longer consider our results significant. We selected a value of
0.40, which resulted from a between-groups meta-analysis (fixed
effects) of the four studies that included a no treatment (waiting
list) control. Results indicate that 122 studies with null effects
would be necessary to offset our overall pretest–posttest effect
size. It is important to note that our criterion effect size level of
0.40, although substantially smaller than our weighted mean effect
size (1.29), is still of a moderate size and statistically significant.
Thus, this is a conservative estimate of the fail-safe N. Hence, it is
likely that our conclusions are not in error because of publication
bias.
Results
How Effective Is CBT for Anxiety Disorders in Clinically
Representative Conditions?
Table 1 summarizes the uncontrolled pretest–posttest effect size
estimates for each disorder. Positive effect sizes indicate improve-
ment from pretreatment to posttreatment. All disorder-specific
effect sizes (range ⫽0.83–2.59) are large (Cohen, 1998). Effect
sizes for reductions in depression and generalized anxiety symp-
toms were only calculated if there was a sufficient number of
studies (five or more) reporting these data. Aside from GAD (in
which generalized anxiety is considered disorder-specific), only
the panic disorder studies reported enough data from generalized
anxiety measures to calculate an effect size. The effect size for
reduction in generalized anxiety symptoms for panic disorder is
large. All depression symptom effect sizes (range ⫽0.73–1.60) are
large, with the exception of social anxiety disorder, for which the
effect size for depression symptoms is medium. All effect sizes are
significant at p⬍.0001.
Is CBT More Effective Than Control Groups?
Six studies included a control group against which to compare
treatment outcomes: two panic studies, one OCD study, and three
599
CBT IN CLINICAL PRACTICE

GAD studies. For the disorder-specific measures, random effects
analyses produced an effect size of g⫽1.29, p⬍.001, confidence
interval ⫽0.76 –1.83. The positive sign indicates an advantage of
CBT treatments over control groups, with a large effect size.
Following Rosenthal and Rosnow (1991), we converted this effect
size into a binomial effect size display (BESD) to yield a more
clinically meaningful metric. The BESD translates a continuous
effect size into its dichotomous equivalent, in this case improved
versus not improved. The degree of superiority of CBT over
control conditions indicated by a gof 1.29 is comparable with a
78% improvement rate for CBT patients versus a 22% improve-
ment rate for patients in the control conditions.
Benchmarking Effectiveness Studies Against Efficacy
Studies
Three benchmarking studies for each disorder were identified,
and pretest–posttest effect sizes were calculated as described
above (see Table 2). Foa et al.’s (2005) study is noteworthy
because it included samples from both an academic and commu-
nity clinic for the treatment of PTSD. The community clinic data
were included in the meta-analysis, and the sample size at the
academic clinic designated this study as one of our benchmarks
against which to compare our meta-analytic results for PTSD.
Overall, the effectiveness effect sizes from this study were in the
range of the effect sizes obtained in our selected efficacy trials.
Specifically, the present study’s effect sizes for social anxiety
disorder, OCD, and GAD are contained within the range of the
efficacy benchmarking studies, whereas the effectiveness effect
size for PTSD exceeds the range of the controlled studies. How-
ever, the effect size for panic disorder falls somewhat below the
range of the benchmarking studies.
Does Clinical Representativeness Predict Effect Size?
Scores on the clinical representativeness scale ranged from 3 to
9(M⫽6.62, SD ⫽1.76). A linear contrast was significant with a
small effect size, t(2526) ⫽⫺2.03, p⬍.05, d⫽⫺0.08. As degree
of clinical representativeness gets larger, the outcome improve-
ment effect size becomes smaller, but the relationship is slight.
Given the overall significant relationship between clinical repre-
sentativeness and outcome, exploratory post hoc contrast analyses
were conducted to gain insight into which individual components
in the overall score might lead to this finding. Reliability was
examined for the individual ratings and deemed acceptable if the
intraclass correlation coefficient exceeded .65 (see the Method
section for coding specifics). Three of the nine variables (clinically
representative referrals, therapists, and patients) were not used for
further analyses because of unacceptable reliability. Reliabilities
for the remaining six variables was adequate to excellent
[␳
I
(3,1) ⫽.66 –.88, Mdn ⫽.77]. A Bonferroni correction was
applied to protect against Type I error, yielding an effective p
value of .008 for significance (.05/6).
Clinically representative setting was not significantly related to
outcome, t(2525) ⫽2.29, p⫽.03, d⫽0.06. Outcome effect sizes
increased significantly when medication was allowed, t(2525) ⫽
5.87, pⱕ.0001, d⫽0.17, and when patients were not randomized
to treatments, t(2392) ⫽3.69, p⬍.0002, d⫽0.11. In contrast,
outcome effect sizes decreased significantly when there was no
training for therapists, t(2392) ⫽⫺10.83, p⬍.0001, d⫽⫺0.31;
when therapists were not asked to follow a manual (clinically
representative structure), t(2392) ⫽⫺5.49, p⬍.0001, d⫽⫺0.16;
and when there was little or no monitoring to make sure the
treatment was followed (clinically representative monitoring),
t(2392) ⫽⫺6.24, p⬍.0001, d⫽⫺0.18. All effects are small.
Discussion
The primary objective of this study was to determine whether
CBT for anxiety disorders works in actual clinical practice. A
secondary goal was to test whether clinical representativeness in
these studies was related to the effect size in a partial replication of
Shadish et al. (2000).
Table 1
Effectiveness Pretest–Posttest Effect Sizes by Disorder
Disorder/symptoms n
Effect
size SE
95% confidence
interval zp
Panic disorder
Attacks 9 1.01 0.12 0.77–1.25 8.32 ⬍.0001
Avoidance 14 0.83 0.12 0.60–1.06 7.09 ⬍.0001
Fear of fear 11 1.23 0.16 0.92–1.54 7.84 ⬍.0001
Depression 17 1.01 0.08 0.86–1.17 13.01 ⬍.0001
Generalized anxiety 14 1.02 0.13 0.77–1.26 8.04 ⬍.0001
Social anxiety disorder
Social anxiety symptoms 11 1.04 0.13 0.79–1.29 8.14 ⬍.0001
Depression 8 0.73 0.09 0.55–0.91 7.90 ⬍.0001
Posttraumatic stress disorder (PTSD)
PTSD symptoms 6 2.59 0.27 2.06–3.13 9.50 ⬍.0001
Depression 5 1.62 0.32 0.99–2.25 5.06 ⬍.0001
Generalized anxiety disorder
Generalized anxiety 11 0.92 0.08 0.77–1.07 11.94 ⬍.0001
Depression 5 0.89 0.09 0.70–1.07 9.39 ⬍.0001
Obsessive–compulsive disorder (OCD)
OCD symptoms 11 1.32 0.07 1.19–1.45 20.15 ⬍.0001
Depression 9 0.89 0.06 0.77–1.01 14.65 ⬍.0001
600 STEWART AND CHAMBLESS

Does CBT for adult anxiety disorders work in actual clinical
practice? The results from this meta-analysis provide evidence
supporting the effectiveness of CBT techniques with the anxiety
disorders. All pretest–posttest effect sizes for disorder-specific
symptom measures were large, indicating that patients treated with
CBT in clinically representative studies improved significantly and
substantially from pretest when they completed treatment. More-
over, CBT for anxiety disorders produced significant pretest–
posttest reductions in depression symptoms with large effects
across panic disorder, PTSD, GAD, and OCD, and a medium
effect for social anxiety disorder. Six of our studies included a
control group against which to compare posttreatment outcomes,
and the effect size was large for disorder-specific symptom mea-
sures across the anxiety disorders. Although the latter findings are
limited (on the basis of six studies), the results suggest that
treatments using CBT techniques in clinically representative con-
ditions lead to significantly larger treatment effect sizes than
waiting list or treatment as usual control groups. The effect size
comparing CBT with control conditions, translated into a BESD,
was comparable with a substantial difference in improvement
rates—78% for CBT versus 22% for control groups. Overall, these
results indicate that CBT is effective in clinical settings.
We utilized a benchmarking strategy to evaluate whether the
magnitude of improvement in clinically representative studies is in
the range of selected efficacy studies. Overall, this was the case.
The clinically representative studies of social anxiety disorder,
OCD, and GAD generated effect size estimates that were within
the range of effect sizes from our selected efficacy trials. In the
case of PTSD, the effectiveness effect sizes exceeded the range of
the benchmarking studies. Only for panic disorder was the effect
size smaller than the smallest of the benchmarking effect sizes,
although it is important to note that this effect size is still large.
The discrepant finding for panic disorder may have been due to the
noise introduced by the great variety of symptom measures used
for this disorder. In general, these results provide initial evidence
for the generalizability of CBT for adult anxiety disorders from
highly controlled research settings to typical clinical settings.
Is clinical representativeness related to treatment outcome? At
face value, our results do not replicate Shadish et al. (2000), in that
we found a slight inverse relationship between clinical represen-
tativeness and effect size. These results suggest that effect sizes are
lower in the more clinically representative studies, which is con-
sistent with the results of Weisz et al. (1995). However, it is critical
to note that although the test was statistically significant, the effect
size in the present study was extremely small (d⫽⫺0.08),
suggesting that the impact of clinical representativeness on effect
size is minor. In contrast, Weisz et al. found an effect size of d⫽
0.62 in favor of research settings over clinic settings.
Post hoc exploratory analyses of the individual variables con-
tributing to the overall clinical representativeness score indicated
that outcome effect sizes increase when patients are not random-
ized to treatments and when medication is permitted. The latter is
not surprising given that the potential additive effects of anxiolytic
and antidepressant medication with psychotherapy. On the other
hand, our results indicate that when therapists are not trained, do
not use manuals, and are not monitored to ensure they are carrying
out the intended treatment, outcome effect sizes decrease. This
may be a result of patients improving less, greater error in these
studies because of the uncontrolled variance in therapists’ behavior
(Crits-Christoph et al., 1991), or some combination of both. These
results are consistent with prior controlled research indicating that
patients with anxiety disorders may improve more with standard-
ized treatments than with therapist-generated treatment plans
(Schulte, Kunzel, Pepping, & Schulte-Bahrenberg, 1992) and that
therapists become more skilled at delivering an EST when they
receive supervised training (Sholomskas et al., 2005). Nonetheless,
these post hoc analyses must be interpreted with caution because
of their exploratory nature and the correlational nature of these
results.
Limitations in this study suggest caution in interpretation. The
first limitation is a construct validity issue: Do our codes assess
legitimate and central measures of clinical representativeness?
Although our codes were built on past research (e.g., Shadish et al.,
2000; Weisz et al., 1995), it is likely that there will be disagree-
ments about key elements we omitted or added to build the current
coding schema. It is also important to note that the codes were
developed (in current and past research) by academics who value
empirical testing of psychological treatments. The real-world men-
tal health practitioner may not agree that studies included in this
meta-analysis are clinically representative. More research and in-
put is needed from front-line practitioners so that we can better
approximate and codify clinically representative conditions. In
addition, although reliability for the total clinically representative
score was excellent, three of the nine clinically representative
characteristics (clinically representative referrals, therapists, and
Table 2
Benchmark Efficacy Studies by Disorder Versus Current
Effectiveness Results: Pretest–Posttest Effect Sizes
Disorder/study CBT treatments nEffect size
Panic disorder
Kenardy et al. (2003) CBT 42 1.53
O
¨st and Westling (1995) AR, CT 36 1.23
Barlow et al. (2000) CBT 56 1.43
Effectiveness 1.02
Social anxiety disorder
Stangier et al. (2003) CBT, CBGT 40 0.89
Clark et al. (2006) CT, ERP 42 1.75
Davidson et al. (2004) CBT 48 1.47
Effectiveness 1.04
Posttraumatic stress disorder
Foa et al. (2005) PE, PE/CR 56 2.50
Resick et al. (2002) CPT, PE 81 2.49
Foa et al. (1999) PE, SIT, PE/SIT 64 1.90
Effectiveness 2.59
Generalized anxiety disorder
Borkovec and Costello (1993) AR, CBT 36 2.26
Borkovec et al. (2002) CT, SCD, CBT 58 1.96
Barlow et al. (1992) AR, CT, AR/CT 34 0.84
Effectiveness 0.92
Obsessive–compulsive disorder
McLean et al. (2001) CBT, ERP 63 1.15
Whittal et al. (2005) CBT, ERP 59 1.88
Van Oppen et al. (1995) CT, ERP 57 1.24
Effectiveness 1.45
Note. CBT ⫽cognitive– behavioral therapy; AR ⫽applied relaxation;
CT ⫽cognitive therapy; CBGT ⫽cognitive– behavioral group therapy;
ERP ⫽exposure and response prevention; PE ⫽prolonged exposure;
CR ⫽cognitive restructuring; CPT ⫽cognitive processing therapy; SIT ⫽
stress inoculation training; SCD ⫽self-control desensitization.
601
CBT IN CLINICAL PRACTICE

patients) were not reliably coded, limiting further analyses. It is not
clear why this was the case. However, studies differed on the
quality of reporting this information, most notably for these three
criteria, and coders likely made judgment calls on the basis of the
limited information provided. Improved reporting and greater de-
tailed methodology in effectiveness studies should be encouraged
so that the reader may understand how closely these studies
estimate clinically representative conditions.
Another concern in meta-analysis is the quality of included
research studies and the measures utilized to represent constructs.
Although we selected studies that met our inclusion criteria, there
was still variability in the overall quality of studies, the quality of
measures, and reporting from which we had to infer codes. In their
meta-analysis of efficacy studies, Norton and Price (2007) used
only established measures as selected by Antony, Orsillo, and
Roemer (2001) in their compendium of assessment instruments for
anxiety disorders. We found this inclusion criterion unnecessarily
prohibitive. Owing to the small number of clinically representative
studies in the literature, we prioritized including studies over
eliminating studies that may not have used measures included by
Antony et al. In the current study, we determined on a case-by-case
basis whether to include a measure if we found it to adequately
assess the construct of interest, and not all measures had undergone
extensive psychometric evaluation.
Another limitation of this sample of studies is the lack of
measures on clinical or applied significance of the effects of CBT
(e.g., reliable change statistics, quality of life, disability). Although
our BESD demonstration was one effort to display the practical
importance of our results in a manner that is meaningful to re-
searchers and clinicians, there remains little information in these
papers about what these changes mean for the average patient’s
functioning. Such measures are helpful in determining whether
CBT is making a genuine difference in the lives of recipients.
More research addressing the applied importance of CBT in clin-
ically representative conditions is sorely needed.
A fourth issue that complicates interpretation is the use of
pretest–posttest analyses. As previously mentioned, because of
limitations in cost, organizational structure, ethical concerns, fea-
sibility, and the focus on external validity, many effectiveness
studies do not include a control group. Hence, our findings must be
interpreted with some caution because we cannot rule out alterna-
tive explanations for patients’ improvement—such as external
events, the passage of time, regression to the mean, and the effects
of the assessment itself. One solution to this problem is our
analysis of the six studies that included a control group, thus
permitting the conclusion that the intervention was responsible for
the change. There is also a body of efficacy studies of anxiety
disorders indicating that patients receiving no treatment respond
very minimally. Norton and Price (2007) found an average effect
size of d⫽0.25 in the placebo conditions, which is substantially
less than any CBT effect size produced by our meta-analysis.
Taken together, these two pieces of evidence suggest that it was
likely the CBT, rather than alternative factors, that produced the
effects reported in this article.
Nonetheless, additional controlled research is needed to permit
a more definitive statement about the causal role of CBT in
patients’ change. Although such research may be difficult to con-
duct (Weisz & Addis, 2006), the importance of such studies in
evaluating arguments for or against dissemination of ESTs is clear.
At the same time, it would be a mistake to overlook the importance
of uncontrolled effectiveness research. One of the most conten-
tious issues in evidence-based practice is the extent to which
results from randomized controlled trials can be generalized to
routine clinical practice. Uncontrolled effectiveness research per-
mits the researcher to maximize external validity by testing treat-
ments (with prior supporting efficacy research) in all types of
naturalistic circumstances to evaluate whether these treatments
translate well to the clinical setting. For example, such studies
permit us to determine whether patients who have not and, in some
cases, would not agree to randomization in a research trial none-
theless improve when they receive an EST—a point that has been
in some dispute in the literature (e.g., Seligman, 1995). In our
view, an argument is most powerful when results from research
emphasizing external validity and research emphasizing internal
validity converge as they do here.
Another issue complicating interpretation of the results is the
use of completer data in this meta-analysis. Few studies (4 out of
56) included reports of intention-to-treat data. Completer analyses
may inflate the apparent results of treatment, in that those who
dropped out of treatment are not represented in such samples. It is
not the case that all of those who drop out of treatment are
treatment failures (Bados, Balaguer, & Saldan˜a, 2007); indeed,
some patients leave therapy early because they believe they are
doing well. Nonetheless, it is undoubtedly true that patients who
leave treatment frequently do so because it is unacceptable to them
or because they are not benefiting. Most high-quality journals now
require authors of randomized controlled trials to report intention-
to-treat analyses. Effectiveness researchers should be encouraged
to do so as well.
Many authors of the effectiveness studies included here also
failed to report key characteristics of their samples. It is critical
that researchers provide tables or summaries of patient character-
istics so that readers can determine how generalizable findings
might be to patients of interest to them. Such data as were reported
attest to the need for more research with African American and
especially Latino patients. The dearth of such research is of par-
ticular concern given the growing number presence of Latinos and
other ethnic minorities in the U.S. population (U.S. Census
Bureau, 2001).
A final limitation of this article is that we did not complete a
formal meta-analytic comparison of effectiveness effect sizes with
efficacy effect sizes. Such an effort was beyond the scope of the
present research, but a formal meta-analytic comparison between
efficacy and effectiveness data would be a valuable next step.
In summary, this meta-analysis provides initial evidence that
CBT in clinically representative conditions is robustly effective
across the adult anxiety disorders. Moreover, the effect sizes found
in conditions approximating actual clinical practice are, on the
whole, within the range of effect sizes produced in our selected
efficacy benchmark studies. More research is needed to determine
whether this effectiveness generalizes to CBT treatments for chil-
dren with anxiety disorders, other ESTs for anxiety, or to CBT for
other disorders in children and adults.
References
References marked with an asterisk indicate studies included in the meta-
analysis. References marked with a plus indicate benchmarking studies.
602 STEWART AND CHAMBLESS

*Addis, M. E., Hatgis, C., Krasnow, A. D., Jacob, K., Bourne, L., &
Mansfield, A. (2004). Effectiveness of cognitive– behavioral treatment
for panic disorder versus treatment as usual in a managed care setting.
Journal of Consulting and Clinical Psychology, 72, 625– 635.
American Psychiatric Association. (1994). Diagnostic and statistical man-
ual of mental disorders (4th ed.). Washington, DC: Author.
Antony, M. M., Orsillo, S. M., & Roemer, L. (2001). Practitioner’s guide
to empirically-based measures of anxiety. New York: Kluwer Academic/
Plenum.
Argras, W. S., & Berkowitz, R. (1980). Clinical research in behavior
therapy: Halfway there? Behavior Therapy, 11, 472– 488.
*Arntz, A. (2003). Cognitive therapy versus applied relaxation as treatment
of generalized anxiety disorder. Behaviour Research and Therapy, 41,
633– 646.
Bados, A., Balaguer, G., & Saldan˜a, C. (2007). The efficacy of cognitive–
behavioral therapy and the problem of drop-out. Journal of Clinical
Psychology, 63, 585–592.
*Bakker, A., Spinhoven, P., van Balkom, A. J. L. M., Vleugel, L., & van
Dyke, R. (2000). Cognitive therapy by allocation versus cognitive ther-
apy by preference in the treatment of panic disorder. Psychotherapy and
Psychosomatics, 69, 240 –243.
⫹
Barlow, D. H., Gorman, J. M., Shear, M. K., & Woods, S. W. (2000).
Cognitive– behavioral therapy, imipramine, or their combination for
panic disorder: A randomized controlled trial. Journal of the American
Medical Association, 19, 2529 –2536.
⫹
Barlow, D., Rapee, R. M., & Brown, T. (1992). Behavioral treatment of
generalized anxiety disorder. Behavior Therapy, 23, 551–570.
Blomhoff, S., Haug, T. T., Hellstrom, K., Holme, I., Humble, M., Madsbu,
H. P., et al. (2001). Randomized controlled general practice trial of
sertraline, exposure therapy and combined treatment in generalized
social phobia. British Journal of Psychiatry, 179, 23–30.
⫹
Borkovec, T. D., & Costello, E. (1993). Efficacy of applied relaxation
and cognitive– behavioral therapy in the treatment of generalized anxiety
disorder. Journal of Consulting and Clinical Psychology, 61, 611– 619.
⫹
Borkovec, T. D., Newman, M. G., Pincus, A. L., & Lytle, R. (2002). A
component analysis of cognitive– behavioral therapy for generalized
anxiety disorder and the role of interpersonal problems. Journal of
Consulting and Clinical Psychology, 70, 288 –298.
*Burke, M., Drummond, L. M., & Johnston, D. W. (1997). Treatment
choice for agoraphobic women: Exposure or cognitive– behaviour ther-
apy? British Journal of Clinical Psychology, 36, 409 – 420.
Chambless, D. L., & Hollon, S. D. (1998). Defining empirically supported
therapies. Journal of Consulting and Clinical Psychology, 66, 7–18.
Chambless, D. L., & Ollendick, T. H. (2001). Empirically supported
psychological interventions: Controversies and evidence. Annual Review
of Psychology, 52, 685–716.
*Chambless, D. L., Tran, G. Q., & Glass, C. R. (1997). Predictors of
response to cognitive– behavioral group therapy for social phobia. Jour-
nal of Anxiety Disorders, 11, 221–239.
⫹
Clark, D. M., Ehlers, A., Hackmann, A., McManus, F., Fennell, M., Grey,
N., et al. (2006). Cognitive therapy versus exposure and applied relax-
ation in social phobia: A randomized controlled trial. Journal of Con-
sulting and Clinical Psychology, 74, 569 –578.
Cohen, J. (1998). Statistical power analysis for the behavioral sciences.
Hillsdale, NJ: Erlbaum.
Crits-Christoph, P., Baranackie, K., Kurcias, J., Beck, A. T., Carroll, K.,
Perry, K., et al. (1991). Meta-analysis of therapist effects in psychother-
apy outcome studies. Psychotherapy Research, 1, 81–91.
⫹
Davidson, J. R. T., Foa, E. B., Huppert, J. D., Keefe, F. J., Franklin,
M. E., Compton, J. S., et al. (2004). Fluoxetine, comprehensive cogni-
tive behavioral therapy, and placebo in generalized social phobia. Ar-
chives of General Psychiatry, 61, 1005–1013.
Deacon, B. J., & Abramowitz, J. S. (2004). Cognitive and behavioral
treatments for anxiety disorders: A review of meta-analytic findings.
Journal of Clinical Psychology, 60, 429 – 441.
*Deacon, B., & Abramowitz, J. (2006). A pilot study of two-day
cognitive– behavioral therapy for panic disorder. Behaviour Research
and Therapy, 44, 807– 817.
Dunlap, W. P., Cortina, J. M., Vaslow, J. B., & Burke, M. J. (1996).
Meta-analysis of experiments with matched groups or repeated measures
designs. Psychological Methods, 2, 170 –177.
*Durham, R. C., Fisher, P. L., Dow, M. G. T., Sharp, D., Power, K. G.,
Swan, J. S., & Morton, R. V. (2004). Cognitive behaviour therapy for
good and poor prognosis generalized anxiety disorder: A clinical effec-
tiveness study. Clinical Psychology and Psychotherapy, 11, 145–157.
*Ellis, T. E., & Lehman, G. L. (1994). A naturalistic outcome study of
group cognitive behavioral treatment for panic disorder. Paper pre-
sented at the 28th Annual Convention of the Association for Advance-
ment of Behavior Therapy, San Diego, CA.
*Feske, U. (2001). Treatment of low-income and African-American
women with posttraumatic stress disorder: A case series. Behavior
Therapy, 32, 585– 601.
⫹
Foa, E. B., Dancu, C. V., Hembree, E. A., Jaycox, L. H., Meadows, E. A.,
& Street, G. P. (1999). Comparison of exposure therapy, stress inocu-
lation training, and their combination for reducing posttraumatic stress
disorder in female assault victims. Journal of Consulting and Clinical
Psychology, 67, 194 –200.
*
⫹
Foa, E. B., Hembree, E. A., Cahill, S. P., Rauch, S. A. A., Riggs, D. S.,
Feeny, N. C., et al. (2005). Randomized trial of prolonged exposure for
posttraumatic stress disorder with and without cognitive restructuring:
Outcome at academic and community clinics. Journal of Consulting and
Clinical Psychology, 73, 953–964.
*Franklin, M. E., Abramowitz, J. S., Kozak, M. J., Levitt, J. T., & Foa,
E. B. (2000). Effectiveness of exposure and ritual prevention for
obsessive– compulsive disorder: Randomized compared with nonran-
domized samples. Journal of Consulting and Clinical Psychology, 69,
594 – 602.
*Friedman, S., Braunstein, J. W., & Halpern, B. (2006). Cognitive behav-
ioral treatment of panic disorder and agoraphobia in a multiethnic urban
outpatient clinic: Initial presentation and treatment outcome. Cognitive
and Behavioral Practice, 13, 282–292.
*Friedman, S., Smith, L. C., Halpern, B., Levine, C., Paradis, C.,
Viswanathan, R., et al. (2003). Obsessive– compulsive disorder in a
multi-ethnic urban outpatient clinic: Initial presentation and treatment
outcome with exposure and ritual prevention. Behavior Therapy, 34,
397– 410.
*Garcı´a-Palacios, A., Botella, C., Robert, C., Ban˜os, R., Perpin˜a´, C.,
Quero, S., & Ballester, R. (2002). Clinical utility of cognitive–
behavioural treatment for panic disorder. Results obtained in different
settings: A research center and a public mental health care unit. Clinical
Psychology and Psychotherapy, 9, 373–383.
*Gaston, J. E., Abbott, M. J., Rapee, R. M., & Neary, S. A. (2006). Do
empirically supported treatments generalize to private practice? A
benchmark study of a cognitive– behavioural group treatment pro-
gramme for social phobia. British Journal of Clinical Psychology, 45,
33– 48.
*Gillespie, K., Duffy, M., Hackmann, A., & Clark, D. M. (2002). Com-
munity based cognitive therapy in the treatment of post-traumatic dis-
order following the Omagh bomb. Behaviour Research and Therapy, 40,
345–357.
*Hahlweg, K., Fiegenbaum, W., Frank, M., Schroeder, B., & von Witzle-
ben, I. (2001). Short- and long-term effectiveness of an empirically
supported treatment for agoraphobia. Journal of Consulting and Clinical
Psychology, 69, 375–382.
Hedges, L. V. (1981). Distribution theory for Glass’s estimator of effect
size and related estimators. Journal of Educational Statistics, 6, 107–
128.
603
CBT IN CLINICAL PRACTICE

Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis.
Orlando, FL: Academic Press.
*Heldt, E., Manfro, G. G., Kipper, L., Blaya, C., Maltz, S., Isolan, L., et al.
(2003). Treating medication-resistant panic disorder: Predictors and
outcome of cognitive– behavior therapy in a Brazilian public hospital.
Psychotherapy and Psychosomatics, 72, 43– 48.
*Herbert, J. D., Gaudiano, B. A., Rheingold, A. A., Myers, V. H., Dal-
rymple, K., & Nolan, E. M. (2005). Social skills training augments the
effectiveness of cognitive behavioral group therapy for social anxiety
disorder. Behavior Therapy, 36, 125–138.
*Herbert, J. D., Rheingold, A. A., Gaudiano, B. A., & Myers, V. H. (2004).
Standard versus extended cognitive behavior therapy for social anxiety
disorder: A randomized-controlled trial. Behavioural and Cognitive Psy-
chotherapy, 32, 131–147.
Higgins, J. P. T., & Thompson, S. G. (2002). Quantifying heterogeneity in
a meta-analysis. Statistics in Medicine, 21, 1539 –1558.
Hoagwood, K., Hibbs, E., Brent, D., & Jensen, P. (1995). Introduction to
the special section: Efficacy and effectiveness in studies of child and
adolescent psychotherapy. Journal of Consulting and Clinical Psychol-
ogy, 63, 683– 687.
Hunsley, J., & Lee, C. M. (2007). Research-informed benchmarks for
psychological treatments, efficacy studies, effectiveness studies, and
beyond. Professional Psychology: Research and Practice, 38, 21–33.
*Hunt, C., & Andrews, G. (1998). Long-term outcome of panic disorder
and social phobia. Journal of Anxiety Disorders, 12, 395– 406.
Jacobson, N. S., & Christensen, A. (1996). Studying the effectiveness of
psychotherapy: How well can clinical trials do the job? American
Psychologist, 51, 1031–1039.
⫹
Kenardy, J. A., Dow, M. G. T., Johnston, D. W., Newman, M. G.,
Thomson, A., & Taylor, C. B. (2003). A comparison of delivery methods
of cognitive– behavioral therapy for panic disorder: An international
multicenter trial. Journal of Consulting and Clinical Psychology, 71,
1068 –1075.
*Levitt, J. T., Malta, L. S., Martin, A., Davis, L., & Cloitre, M. (2007). The
flexible application of a manualized treatment for PTSD symptoms and
functional impairment related to the 9/11 World Trade Center attack.
Behaviour Research and Therapy, 45, 1419 –1433.
Light, R. J., Singer, J. D., & Willett, J. B. (1994). The visual presentation
and interpretation of meta-analyses. In H. Cooper & L. V. Hedges
(Eds.), The handbook of research synthesis (pp. 439– 453). New York:
Russell Sage Foundation.
*Lincoln, T. M., Rief, W., Hahlweg, K., Frank, M., von Witzleben, I.,
Schroeder, B., & Fiegenbaum, W. (2003). Effectiveness of an empiri-
cally supported treatment for social phobia in the field. Behaviour
Research and Therapy, 41, 1251–1269.
*Linden, M., Zubraegel, D., Franke, U., & Schlattmann, P. (2005). Effi-
cacy of cognitive behaviour therapy in generalized anxiety disorders.
Psychotherapy and Psychosomatics, 74, 36 – 42.
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousand
Oaks, CA: Sage.
*Marom, S., Gilboa-Schechtman, E., Aderka, I. M., Weizman, A., &
Hermesh, H. (2009). Impact of depression on treatment effectiveness
and gains maintenance in social phobia: A naturalistic study of cognitive
behavior group therapy. Depression and Anxiety, 26, 289 –300.
*Martinsen, E. W., Olsen, T., Tønset, E., Nyland, K., & Aarre, T. F.
(1998). Cognitive– behavioral group therapy for panic disorder in the
general clinical setting: A naturalistic study with 1-year follow-up.
Journal of Clinical Psychiatry, 59, 437– 442.
*McEvoy, P. M. (2007). Effectiveness of cognitive behavioural group
therapy for social phobia in a community clinic: A benchmarking study.
Behaviour Research and Therapy, 45, 3030 –3040.
⫹
McLean, P. D., Whittal, M. L., Thordarson, D. S., Taylor, S., So¨chting,
I., Koch, W. J., et al. (2001). Cognitive versus behavior therapy in the
group treatment of obsessive– compulsive disorder. Journal of Consult-
ing and Clinical Psychology, 69, 205–214.
*Mortberg, E., Berglund, G., & Sundin, O. (2005). Intensive cognitive
behavioural group treatment of social phobia: A pilot study. Cognitive
Behaviour Therapy, 34, 41– 49.
*Mortberg, E., Karlsson, A., Fyring, C., & Sundin, O. (2006). Intensive
cognitive– behavioral group treatment (CBGT) of social probia: A ran-
domized controlled study. Anxiety Disorders, 20, 646 – 660.
*Norberg, M. M., Calamari, J. E., Cohen, R. J., & Riemann, B. C. (2008).
Quality of life in obsessive– compulsive disorder: An evaluation of
impairment and a preliminary analysis of the ameliorating effects of
treatment. Depression and Anxiety, 25, 248 –259.
Norton, P. J., & Price, E. C. (2007). A meta-analytic review of adult
cognitive– behavioral treatment outcome across the anxiety disorders.
The Journal of Nervous and Mental Disease, 195, 521–531.
Orwin, R. G. (1983). A fail-safe Nfor effect size in meta-analysis. Journal
of Educational Statistics, 8, 157–159.
⫹
O
¨st, L., & Westling, B. E. (1995). Applied relaxation versus cognitive
behavior therapy in the treatment of panic disorder. Behavior Research
and Therapy, 33, 145–158.
Persons, J. B., & Silberschatz, G. (1998). Are results of randomized
controlled trials useful to psychotherapists? Journal of Counseling and
Clinical Psychology, 66, 126 –135.
⫹
Resick, P. A., Nishith, P., Weaver, T. L., Astin, M. C., & Feuer, C. A.
(2002). A comparison of cognitive–processing therapy with prolonged
exposure and a waiting list condition for the treatment of chronic
posttraumatic stress disorder in female rape victims. Journal of Consult-
ing and Clinical Psychology, 70, 867– 879.
*Rosenberg, N. K., & Hougaard, E. (2005). Cognitive– behavioural group
treatment of panic disorder and agoraphobia in a psychiatric setting: A
naturalistic study of effectiveness. Nordic Journal of Psychiatry, 59,
198 –204.
Rosenthal, R. (1979). The “file drawer problem” and tolerance for null
results. Psychological Bulletin, 86, 638 – 641.
Rosenthal, R., & Rosnow, R. L. (1991). Essentials of behavioral research:
Methods and data analysis (2nd ed.). New York: McGraw-Hill.
*Rothbaum, B. O. (2000). Behavioral treatment of obsessive– compulsive
disorder in a naturalistic setting. Cognitive and Behavioral Practice, 7,
262–270.
*Sanderson, W. C., Raue, P. J., & Wetzler, S. (1998). The generalizability
of cognitive behavior therapy for panic disorder. Journal of Cognitive
Psychotherapy: An International Quarterly, 12, 323–330.
Schulte, D., Kunzel, R., Pepping, G., & Schulte-Bahrenberg, T. (1992).
Tailor-made versus standardized therapy of phobic patients. Advances in
Behaviour Research and Therapy, 14, 67–92.
*Schultz, P. M., Resick, P. A., Huber, L. C., & Griffin, M. G. (2006). The
effectiveness of cognitive processing therapy for PTSD with refugees in
a community setting. Cognitive and Behavioral Practice, 13, 322–331.
Seligman, M. E. P.(1995). The effectiveness of psychotherapy: The
Consumer Reports study. American Psychologist, 50(12), 965–974.
Shadish, W. R., Matt, G. E., Navarro, A. M., & Phillips, G. (2000). The
effects of psychological therapies under clinically representative condi-
tions: A meta-analysis. Psychological Bulletin, 126, 512–529.
*Sharp, D. M., Power, K. G., & Swanson, V. (2000). Reducing therapist
contact in cognitive behaviour therapy for panic disorder and agorapho-
bia in primary care: Global measures of outcome in a randomized
controlled trial. British Journal of General Practice, 40, 963–968.
*Sharp, D. M., Power, K. G., & Swanson, V. (2004). A comparison of the
efficacy and acceptability of group versus individual cognitive behav-
iour therapy in the treatment of panic disorder and agoraphobia in
primary care. Clinical Psychology and Psychotherapy, 11, 73– 82.
Sholomskas, D. E., Syracuse-Siewert, G., Rounsaville, B. J., Ball, S. A.,
Nuro, K. F., & Carroll, K. M. (2005). We don’t train in vain: A
dissemination trial of three strategies of training clinicians in cognitive–
604 STEWART AND CHAMBLESS

behavioral therapy. Journal of Consulting and Clinical Psychology, 73,
106 –115.
Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in
assessing rater reliability. Psychological Bulletin, 86, 420 – 428.
⫹
Stangier, U., Heidenreich, T., Peitz, M., Lauterbach, W., & Clark, D. M.
(2003). Cognitive therapy for social phobia: Individual versus group
treatment. Behaviour Research and Therapy, 41, 991–1007.
*Stanley, M. A., Hopko, D. R., Diefenbach, G. J., Bourland, S. L.,
Rodriguez, H., & Wagener, P. (2003). Cognitive– behavior therapy for
late-life generalized anxiety disorder in primary care. American Journal
of Geriatric Psychiatry, 11, 92–96.
Stirman, S. W., DeRubeis, R. J., Crits-Christoph, P., & Brody, P. E. (2003).
Are samples in randomized controlled trials of psychotherapy represen-
tative of community outpatients? A new methodology and initial find-
ings. Journal of Consulting and Clinical Psychology, 71, 963–972.
Stirman, S. W., DeRubeis, R. J., Crits-Christoph, P., & Rothman, A.
(2005). Can the randomized controlled trial literature generalize to
nonrandomized patients? Journal of Consulting and Clinical Psychol-
ogy, 73, 127–135.
*Tolin, D. F., Hannan, S., Maltby, N., Diefenbach, G. J., Worhunsky, P.,
& Brady, R. E. (2007). A randomized controlled trail of self-directed
versus therapist-directed cognitive– behavioral therapy for obsessive–
compulsive disorder patients with prior medication trials. Behavior
Therapy, 38, 179 –191.
*Tolin, D. F., Maltby, N., Diefenbach, G. J., Hannan, S. E., & Worhunsky,
P. (2004). Cognitive– behavioral therapy for medication nonresponders
with obsessive– compulsive disorder: A wait-list-controlled open trial.
Journal of Clinical Psychiatry, 65, 922–931.
U.S. Census Bureau. (2001). Profile of general demographic characteris-
tics: 2000 census of population and housing, United States. Retrieved
February 16, 2009, from http://www.census.gov/population/www/
socdemo/race/race.html
*Van Noppen, B., Steketee, G., McCorkle, B. H., & Pato, M. (1997).
Group and multifamily behavioral treatment for obsessive compulsive
disorder: A pilot study. Journal of Anxiety Disorders, 11, 431– 446.
⫹
Van Oppen, P., de Haan, E., van Balkom, A. J. L. M., Spinhoven, P.,
Hoogduin, K., & van Dyck, R. (1995). Cognitive therapy and exposure
in vivo in the treatment of obsessive compulsive disorder. Behavior
Research and Therapy, 33, 379 –390.
*Wade, W. A., Treat, T. A., & Stuart, G. L. (1998). Transporting an
empirically supported treatment for panic disorder to a service clinic
setting: A benchmarking strategy. Journal of Consulting and Clinical
Psychology, 66, 231–239.
*Warren, R., Strand, J., & Thomas, J. C. (2008). Panic control treatment in
private practice: Immediate and long term results. Manuscript in prep-
aration.
*Warren, R., & Thomas, J. C. (2001). Cognitive– behavior therapy of
obsessive– compulsive disorder in private practice: An effectiveness
study. Anxiety Disorders, 15, 277–285.
Weisz, J. R., & Addis, M. E. (2006). The research-practice tango and other
choreographic challenges: Using and testing evidence-based psychother-
apies in clinical care settings. In C. D. Goodheart, A. E. Kazdin, & R. J.
Sternberg (Eds.), Evidence-based psychotherapy: Where practice and
research meet (pp. 179 –206). Washington, DC: American Psychologi-
cal Association.
Weisz, J. R., Donenberg, G. R., Han, S. S., & Weiss, B. (1995). Bridging
the gap between laboratory and clinic in child and adolescent psycho-
therapy. Journal of Consulting and Clinical Psychology, 63, 699 –701.
Weisz, J. R., Weiss, B., & Donenberg, G. R. (1992). The lab versus the
clinic: Effects of child and adolescent psychotherapy. American Psy-
chologist, 47, 1578 –1585.
*Westbrook, D., & Kirk, J. (2005). The clinical effectiveness of cognitive
behaviour therapy: Outcome for a large sample of adults treated in
routine practice. Behaviour Research and Therapy, 43, 1243–1261.
Westen, D., & Morrison, K. (2001). A multidimensional meta-analysis of
treatments for depression, panic, and generalized anxiety disorder: An
empirical examination of the status of empirically supported therapies.
Journal of Consulting and Clinical Psychology, 69, 875– 899.
Westen, D., Novotny, C. M., & Thompson-Brenner, H. (2004). The em-
pirical status of empirically supported psychotherapies: Assumptions,
findings, and reportings in controlled clinical trials. Psychological Bul-
letin, 4, 631– 663.
*Wetzel, C., Bents, H., & Florin, I. (1999). High-density exposure therapy
for obsessive– compulsive inpatients: A 1-year follow-up. Psychother-
apy and Psychosomatics, 68, 186 –192.
*White, J., & Keenan, M. (1992). Stress control: A controlled comparative
investigation of large group therapy for generalized anxiety disorder.
Behavioural Psychotherapy, 20, 97–114.
⫹
Whittal, M. L., Thordarson, D. S., & McLean, P. D. (2005). Treatment of
obsessive– compulsive disorder: Cognitive behavior therapy vs. expo-
sure and response prevention. Behaviour Research and Therapy, 43,
1559 –1576.
(Appendix follows)
605
CBT IN CLINICAL PRACTICE

Appendix
Clinical Representativeness Coding Manual
Criterion for clinical
representativeness Definition and code
Clinically representative setting 1: Setting where clinical services are commonly provided in nonuniversity
settings (e.g., outpatient mental health clinics, CMHCs, general
hospitals, private practices, prisons, school systems) in which the
primary purpose is clinical care.
1
⁄
2
: Patient setting affiliated with a university research program but
primary purpose of setting is clinical care.
0: University research setting.
Clinically representative referrals 1: Clients initially referred through usual clinical routes (e.g., MDs).
1
⁄
2
: Clients referred through usual clinical routes (e.g., MDs) and
recruitment solicitations/advertisements.
0: Clients referred by recruitment solicitations/advertisements.
Clinically representative therapists 1: Practicing and experienced clinicians with regular caseloads for whom
provision of services is a substantial part of job. The primary therapists
at the clinical setting—interns are included.
0: Research therapists.
Clinically representative structure 1: Treatment either with a structure used in clinical practice or not
structured in a detailed and uniform way according to a manual (e.g.,
general CBT).
1
⁄
2
: Treatment “based on” or “inspired by” a clinical manual, which can
be used flexibly (e.g., include treatment procedures not in the manual,
flexible number of sessions).
0: Strict manualization, or manual-driven treatment (e.g., set number of
sessions).
Clinically representative monitoring 1: Implementation of treatment was not monitored in any way that could
influence therapist behavior (no formal adherence checks and no
supervision).
1
⁄
2
: Only monitoring was supervision and/or outlines containing
information to be covered in session (no formal adherence checks).
0: Supervision and adherence checks.
No pretherapy training 1: Therapists did not receive special training immediately before study in
specific techniques to be used. Of course, therapists may have received
training in that treatment at some point in their career but not for
purposes of study in which they participated.
1
⁄
2
: Only novice therapists are trained, and they represent less than 25% of
therapists.
0: Pretherapy training; also give 0 when percentage of novice therapists
not specified.
No randomization 1: Patients were not part of a randomized trial, did not meet criteria for a
particular trial, or refused randomization.
0: Patients were randomized to treatments.
Clinically representative patients 1: No exclusionary criteria aside from psychosis, suicidality, organic brain
disease, or substance dependence if patient meets criteria for disorder
under study.
0: Exclusion criteria beyond those above, including comorbidity and
medication.
Medications allowed 1: Medication is allowed. If no specific mention in exclusion criteria,
assume medication is allowed.
0: Patients on medication are forced to go off medication or are excluded
from study.
Note. CMHCs ⫽community mental health centers; MDs ⫽medical doctors; CBT ⫽cognitive– behavioral therapy.
Received June 9, 2008
Revision received February 19, 2009
Accepted April 3, 2009 䡲
606 STEWART AND CHAMBLESS