PSY JOURNAL due 12/04 7 AM est

Arch.2012.RCTofCBTandACTformixedanx.JCCP.pdf

Home >Psychology homework help >PSY JOURNAL due 12/04 7 AM est

Randomized Clinical Trial of Cognitive Behavioral Therapy (CBT) Versus Acceptance and Commitment Therapy (ACT) for Mixed Anxiety Disorders

Joanna J. Arch University of Colorado Boulder

Georg H. Eifert Chapman University

Carolyn Davies University of California, Los Angeles

Jennifer C. Plumb Vilardaga University of Nevada, Reno

Raphael D. Rose and Michelle G. Craske University of California, Los Angeles

Objective: Randomized comparisons of acceptance-based treatments with traditional cognitive behavioral therapy (CBT) for anxiety disorders are lacking. To address this gap, we compared acceptance and commit- ment therapy (ACT) to CBT for heterogeneous anxiety disorders. Method: One hundred twenty-eight individuals (52% female, mean age � 38, 33% minority) with 1 or more DSM–IV anxiety disorders began treatment following randomization to CBT or ACT; both treatments included behavioral exposure. Assess- ments at pre-treatment, post-treatment, and 6- and 12-month follow-up measured anxiety-specific (principal disorder Clinical Severity Ratings [CSRs], Anxiety Sensitivity Index, Penn State Worry Questionnaire, Fear Questionnaire avoidance) and non-anxiety-specific (Quality of Life Index [QOLI], Acceptance and Action Questionnaire–16 [AAQ]) outcomes. Treatment adherence, therapist competency ratings, treatment credibil- ity, and co-occurring mood and anxiety disorders were investigated. Results: CBT and ACT improved similarly across all outcomes from pre- to post-treatment. During follow-up, ACT showed steeper linear CSR improvements than CBT ( p � .05, d � 1.26), and at 12-month follow-up, ACT showed lower CSRs than CBT among completers ( p � .05, d � 1.10). At 12-month follow-up, ACT reported higher AAQ than CBT (p � .08, d � 0.42; completers: p � .05, d � 0.56), whereas CBT reported higher QOLI than ACT (p � .05, d � 0.42). Attrition and comorbidity improvements were similar; ACT used more non-study psychotherapy at 6-month follow-up. Therapist adherence and competency were good; treatment credibility was higher in CBT. Conclusions: Overall improvement was similar between ACT and CBT, indicating that ACT is a highly viable treatment for anxiety disorders.

Keywords: cognitive behavioral therapy, acceptance and commitment therapy, anxiety disorders

Supplemental materials: http://dx.doi.org/10.1037/a0028310.supp

Several decades ago, the development of behavioral and cogni- tive behavioral therapy for anxiety disorders (Barlow & Cerny, 1988; Beck, Emery, & Greenberg, 1985) introduced time-limited, relatively effective treatments. Numerous randomized clinical tri- als and meta-analyses (e.g., Butler, Chapman, Forman, & Beck, 2006; Hofmann & Smits, 2008; Norton & Price, 2007; Tolin, 2010) have demonstrated the effectiveness of CBT for anxiety disorders, including panic disorder, generalized anxiety disorder, social anxiety disorder, obsessive-compulsive disorder, specific

phobia, and posttraumatic stress disorder, relative to waitlist and/or psychological control conditions. As a result of clinical efficacy and ease of implementation, cognitive behavioral therapy (CBT) has become the dominant empirically validated treatment for anx- iety disorders. However, a significant percentage of individuals with anxiety disorders do not respond to CBT (e.g., Barlow, Gorman, Shear, & Woods, 2000), relapse following successful treatment (Brown & Barlow, 1995), seek additional treatment (Brown & Barlow, 1995), or remain vulnerable to developing

This article was published Online First May 7, 2012. Joanna J. Arch, Department of Psychology and Neuroscience, University of

Colorado Boulder; Georg H. Eifert, Department of Psychology, Chapman University; Carolyn Davies, Department of Psychology, University of Cali- fornia, Los Angeles; Jennifer C. Plumb Vilardaga, Department of Psychology, University of Nevada, Reno; Raphael D. Rose and Michelle G. Craske, Department of Psychology, University of California, Los Angeles.

Thank you to Daniel Dickson and Milena Stoyanova for study coordination, to John Forsyth for help with initial study planning, and

to Chick Judd and Kate Wolitzky-Taylor for sound statistical advice. Thank you to the many research assistants (over one dozen) who assisted in data collection, the many graduate student therapists, and above all, the individuals who participated in the study. Special thanks to Kathleen McGrath and colleagues for conducting the treatment reliability ratings.

Correspondence concerning this article should be addressed to Michelle G. Craske, Department of Psychology, 405 Hilgard Avenue, Los Angeles, CA 90095-1563. E-mail: [email protected]

750

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

anxiety and mood disorders throughout the life span (see Craske, 2003). Further, the cognitive model, which posits that anxiety disorders stem from faulty cognitions and are treated by modifying the content of these cognitions, has been challenged by insufficient supporting evidence (see Longmore & Worrell, 2007) and theo- retical arguments (see Brewin, 1996).

In attempt to broaden, improve upon, and provide theoretically strong alternatives to traditional (cognitive model-based) CBT, clini- cal researchers have shown increasing interest in mindfulness and acceptance-based treatments for psychopathology. Multiple treatment approaches have developed out of this exploration, including accep- tance and commitment therapy (ACT; Hayes, Strosahl, & Wilson, 1999), a behavioral therapy that cultivates mindfulness, acceptance, cognitive defusion (flexible distancing from the literal meaning of cognitions), and other strategies to increase psychological flexibility and promote behavior change consistent with personal values. Within ACT, psychological flexibility is defined as enhancing the capacity to make contact with experience in the present moment, and based on what is possible in that moment, persisting in or changing behavior in the pursuit of goals and values (Hayes et al., 1999). Case studies, multiple-baseline treatment studies, and a single randomized clinical trial provide nascent evidence that ACT is an effective treatment for anxiety disorders, including obsessive-compulsive disorder (Twohig et al., 2010), social anxiety disorder (Dalrymple & Herbert, 2007), panic disorder (Eifert et al., 2009), generalized anxiety disorder (Wetherell et al., 2011), and posttraumatic stress disorder (Orsillo & Batten, 2005). However, no randomized clinical trials for anxiety disorders have yet compared ACT with the gold-standard therapy for anxiety disorders, CBT.

Several studies have compared ACT and CBT within general clinic samples. One randomized effectiveness trial (n � 101) in undiag- nosed anxious and depressed patients (Forman, Herbert, Moitra, Yeo- mans, & Geller, 2007) compared ACT and CBT,1 concluding that the two treatments were equally effective across symptom outcome mea- sures but operated via somewhat different mechanisms. A small (n � 28) randomized treatment study comparing ACT and CBT for undi- agnosed “general clinic patients” (Lappalainen et al., 2007) found that ACT reduced symptoms to a greater degree than CBT and affected treatment processes at a different rate. In neither study, however, were anxiety disorders formally diagnosed, significantly limiting the con- clusions for the treatment of anxiety disorders. Roemer and Orsillo (2007; Roemer, Orsillo, & Salters-Pedneault, 2008) developed an acceptance-based treatment for generalized anxiety disorder (GAD) that integrated several ACT principles and found it to be significantly superior to a waitlist control group for the treatment of generalized anxiety disorder. However, this promising treatment was developed for generalized anxiety disorder alone, and it has not yet been com- pared to another active treatment.

In summary, important work has begun to investigate ACT for anxiety disorders, but studies to date have not compared ACT to the most evidence-based psychotherapy for the majority of anxiety disorders, CBT. Directly comparing traditional CBT and ACT for anxiety disorders within a randomized clinical trial bridges a vital gap in the empirical treatment literature, fulfilling the gold- standard method for investigating the relative efficacy of two treatments for anxiety disorders (Chambless & Hollon, 1998). Further, it provides an opportunity to compare the efficacy of treatment packages that contain distinct strategies for dealing with maladaptive cognitions (change the content of thoughts in CBT vs.

change the context by challenging the need to respond rigidly and literally to cognitions in ACT) and uncomfortable internal expe- riences (to master and reduce anxiety in CBT vs. open toward and accept anxiety in ACT) and that promote different treatment goals (anxiety reduction in CBT vs. living a valued life in ACT).

The current study compares ACT and CBT in a mixed anxiety disorders sample for two reasons. First, ACT (Hayes et al., 1999) originally was developed for the treatment of psychopathology in general rather than a specific disorder in particular. The ACT protocol used in the current study (Eifert & Forsyth, 2005) was designed for application across all of the anxiety disorders, with the content of values-guided behavioral exercises tailored to spe- cific anxiety disorders. Second, CBT shares common treatment elements across the anxiety disorders with variation in content specific to each disorder. We thus designed a CBT manual that addressed all anxiety disorders via branching mechanisms specific to each disorder. Our CBT protocol included the same basic treatment elements across all of the disorders (e.g., psychoeduca- tion, cognitive restructuring, exposure—see Method section) but tailored the content of these elements to each specific disorder, an approach we have successfully tested in previous studies (e.g., Craske et al., 2011).

We assessed patients at four longitudinal measurement points, including pre-treatment, post-treatment, 6-month follow-up, and 12-month follow-up, providing a thorough assessment of treatment-related change over time. Due to limited extant data, we did not make specific predictions regarding whether one treatment would lead to greater reductions in anxiety disorder-related symp- toms than the other treatment. Rather, investigating this question represented a central study aim. However, given the emphasis in ACT on psychological flexibility and valued living, we hypothe- sized that these measures would improve to a greater degree following ACT than CBT.

Method

Participants

One hundred and forty-three participants meeting Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM–IV; American Psychiatric Association, 1994) criteria for a diagnosis of one or more anxiety disorders, including panic disorder with or without agoraphobia (PD/A), social anxiety disorder (SAD), spe- cific phobia (SP), obsessive-compulsive disorder (OCD), or gen- eralized anxiety disorder (GAD),2 were randomized to ACT (n � 65) or CBT (n � 78). All participants who began treatment (n � 128) were included in the intent-to-treat (ITT) sample (n � 57 ACT, n � 71 CBT). See Table 1 for ITT sample characteristics

1 The study called their approach “cognitive therapy,” which they de- fined as the Beck-based treatment subtype of CBT (see Forman et al., 2007).

2 Only four participants met principal diagnosis of posttraumatic stress disorder (PTSD), perhaps because our recruitment materials stated “anxiety disorders” but not “trauma,” therefore attracting fewer PTSD participants. Of these, one did not begin treatment, one dropped treatment, and two completed treatment. Due to the very small sample size for this disorder (three total PTSD participants who began treatment and only one in ACT), these participants were excluded from analyses.

751CBT VERSUS ACT FOR ANXIETY DISORDERS

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

and Figure 1 for patient flow. Fifteen of the original 143 partici- pants, blinded to treatment randomization (unaware if they had been randomized to ACT or CBT), dropped prior to treatment initiation. Because pre-treatment attrition gave us no information about treatment preference or response, we did not analyze those participants further. Participants who dropped prior to treatment did not differ significantly from participants who began treatment on any sociodemographic variable from Table 1 ( ps � .20), nor did not differ by blind assignment to ACT versus CBT (n � 8 each, p � .66). Participants who dropped treatment showed some- what higher Clinical Severity Ratings (CSRs; M � 6.07, SD � 0.96) relative to participants who began treatment (M � 5.63, SD � 0.92), but group differences were small and did not reach statistical significance ( p � .08, �p

2 � .02). Eleven of the 15 participants (73%) dropped prior to completing the pre-treatment questionnaire assessment; therefore, we could not determine if they differed from participants who initiated treatment on ques- tionnaire measures.

Participants were recruited from the Los Angeles area in re- sponse to local flyers, Craigslist and local newspaper advertise- ments, and referrals. The study took place at the Anxiety Disorders Research Center at the University of California Los Angeles (UCLA), Department of Psychology.

Participants were either medication free or stabilized on psy- chotropic medications for a minimum standard length of time (1 month for benzodiazepines and beta blockers, 3 months for selec- tive serotonin reuptake inhibitors [SSRIs] or serotonin– norepinephrine reuptake inhibitors [SNRIs] and heterocyclics).

Also, participants were psychotherapy free or stabilized on alter- native psychotherapies other than cognitive or behavioral therapies that were not focused on their anxiety disorder, for at least 6 months prior to study entry. Participants were encouraged not to change their non-study medication or alternative psychotherapy during the course of the study. Exclusion criteria included active suicidal ideation; severe depression (CSR � 6 on ADIS–IV; see below); or a history of bipolar disorder, psychosis, mental retar- dation, or organic brain damage. Participants with substance abuse or dependence within the last 6 months or with respiratory, car- diovascular, pulmonary, neurological, muscular-skeletal diseases, or pregnancy were excluded.

Participants received 12 weeks of reduced-cost, sliding scale ($0 –100/session) individual treatment and received $15–$25 in cash or a gift certificate upon completion of the post-treatment and each follow-up assessment. Participants were reimbursed for UCLA parking fees for the assessments ($8 –$10). The study was fully approved by the UCLA human subjects protection commit- tee; full informed consent was obtained from all participants, including for video- and audiorecordings.

Design

Participants were assessed at pre-treatment (Pre), post-treatment (Post), and at 6 months (6mFU) and 12 months (12mFU) after Pre. Assessments included a diagnostic interview, self-report question- naires, and a 2- to 3-hr laboratory assessment (except at 6mFU) reported elsewhere. Assessors were blind to treatment condition.

Table 1 Demographic and Clinical Characteristics of the Intent-to-Treat Sample

Characteristic Total (n � 128) ACT (n � 57) CBT (n � 71) t or �2 p

Gender Female 52.3% (67/128) 50.0% (28/56) 54.9% (39/71) 0.43 .51

Reported race/ethnicity 0.82 .36a

White 67.2% (84/125) 71.4% (40/56) 63.8% (44/69) Hispanic/Latino/a 12.0% (15/125) 10.7% (6/56) 13.0% (9/69) African American/Black 8.8% (11/125) 7.1% (4/56) 10.1% (7/69) Asian American/Pacific Islander 8.0% (10/125) 8.9% (5/56) 7.2% (5/69) American Indian/Alaskan Native 0.08% (1/125) 1.8% (1/56) 0.0% (0/69)

Age in years, M (SD) 37.93 (11.70) range: 19–60 38.16 (12.41) 37.75 (11.19) �0.20 .85 Education in years, M (SD) 15.41 (2.07) range: 9–21 15.59 (2.01) 15.27 (2.12) �0.86 .39 Marital status 0.19 .91a

Married/cohabiting 32.3% (41/127)b 32.1% (18/56) 32.4% (23/71) Single 58.3% (74/127) 57.1% (32/56) 59.2% (42/71) Other 10.2% (13/127) 10.7% (6/56) 8.5% (6/71)

Children (1�) 28.0% (35/125) 30.4% (17/56) 26.1% (18/69) 0.28 .60 Currently on psychotropic medication 48.0% (61/127) 44.6% (25/56) 50.7% (36/71) 0.46 .50 Primary diagnosis 4.67 .32

Panic disorder (with or without agoraphobia) 41.7% (53/127) 32.1% (18/56) 49.3% (35/71) 3.79 .052 Social anxiety disorder 19.7% (25/127) 23.2% (13/56) 16.9% (12/71) 0.79 .37 Generalized anxiety disorder 20.5% (26/127) 25.0% (14/56) 16.9% (12/71) 1.26 .26 Obsessive-compulsive disorder 13.4% (17/127) 16.1% (9/56) 11.3% (8/71) 0.62 .43 Specific phobia 4.7% (6/127) 3.6% (2/56) 5.6% (4/71) .69d

Comorbid anxiety disorder (1�)c 33.1% (42/127) 39.3% (22/56) 28.2% (20/71) 1.75 .19 Comorbid depressive disorderc 23.6% (30/127) 23.2% (13/56) 23.9% (17/71) 0.01 .92 Principal disorder clinical severity rating at Pre, M (SD) 5.62 (0.92) 5.70 (0.89) 5.55 (0.94) 0.94 .35

Note. ACT � acceptance and commitment therapy; CBT � cognitive behavioral therapy; Pre � pre-treatment. a For race/ethnicity and marital status, analyses assessed group differences in minority versus White and married versus non-married sta- tus. b Demographic data were missing for one participant. c Comorbidity was defined as a clinical severity rating of 4 or above on the Anxiety Disorders Interview Schedule–IV (ADIS–IV). d Fisher’s exact test p value was reported due to small sample size.

752 ARCH ET AL.

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

Randomization sequences were produced by http://www.randomizer .org; therapists were requested not to inform participants of their assigned treatment condition. For the first two quarters of the study, participants were randomized 1:1 CBT to ACT. In the third quarter, participants were randomized 2:1 CBT to ACT due to the greater availability of CBT-trained therapists. In the final quarter, to equalize condition assignments, participants were randomized 1:2 CBT to ACT. Because fewer participants were recruited in the fourth com- pared to the third quarter, the total number of participants who began treatment in each condition was unequal (n � 57 to ACT; n � 71 to CBT). To maximize statistical power for the main group comparison hypotheses, we did not stratify patients on any variables.

Treatments

Following the Pre ADIS–IV and laboratory assessments, partic- ipants were randomized to treatment condition. Participants re- ceived 12 weekly, 1-hr individual CBT or ACT therapy sessions

based on detailed treatment manuals delivered by doctoral student therapists.3 If clients presented with multiple anxiety disorders, treatment focused on the principal disorder. Following the 12 sessions, therapists conducted follow-up phone calls once per month for 6 months, allowing 20 –35 min per call to check in and troubleshoot in a manner consistent with the assigned therapy condition, to enhance long-term outcomes (see Craske et al., 2006).

Therapists

Clinical psychology doctoral students at UCLA served as study therapists. The majority of therapists were relatively naı̈ve to CBT and ACT and inexperienced more generally (i.e., in their first or

3 See author Craske for a copy of the CBT treatment manual; the ACT manual has been published (Eifert & Forsyth, 2005).

Excluded (n = 87) � Did not meet inclusion criteria (n = 65) � Lost contact with study (n = 10) � Declined participation (n = 4)

o Reasons: schedule (n = 2), uncomfortable with being recorded (n = 1), no reason given (n = 1)

� Excluded from further analysis for diagnostic reasons: principal PTSD (n = 4)

Randomized (n = 143)

Enrollment

Assessed for Eligibility Completed phone screening (n = 294) Did not participate in intake interview (n = 69) � Did not qualify during phone screening (n = 14) � Invited to participate but chose not to (n = 55)

Participated in intake interview (n = 225)

Analyzed (n = 71)

Lost to Post (n = 23) � Drop by PI (n = 4) 2 due to new ineligible

medical conditions, 1 due to new meds, 1 due to new psychotherapy during CBT

� Transportation (n = 1), schedule (n = 1), not satisfied with treatment (n = 3), lost contact/unknown (n = 14)

Lost to Follow-Up 1 (n = 12) � Moved (n = 1), schedule (n = 1), declined

(n = 1), lost contact/ unknown (n = 9) Lost to Follow-Up 2 (n = 5) � Lost contact/ no reason given (n = 5)

Allocated to CBT (n = 71)

Lost to Post (n = 20) � Drop by PI (n = 1) referred out due to need

for more intensive treatment for suicidal depression

� No longer interested (n = 2), “too anxious” to complete treatment (n = 2), other (n = 3), lost contact/unknown (n = 12)

Lost to Follow-Up 1 (n = 10) � Moved (n = 1), schedule (n = 1), declined

(n = 1), lost contact /unknown (n = 7) Lost to Follow-Up 2 (n = 4) � Schedule (n = 2), lost contact/ no reason

given (n = 2)

Allocated to ACT (n = 57)

Analyzed (n = 57)

Allocation

Analysis

Follow-Up

Fewer than ½ Rx sessions (n = 15) Completed all 12 sessions (n = 49) Average sessions attended = 9.6

Fewer than ½ Rx sessions (n = 12) Completed all 12 sessions (n = 38) Average sessions attended = 9.4

Treatment

Lost to Pre-Assessment (n = 15) � Lost contact with study (n = 6), moved

(n = 1), began other treatment (n = 1), too anxious (n = 1), no longer interested (n = 1), no reason given (n = 5)

Figure 1. Patient flow for the study. PTSD � posttraumatic stress disorder; ACT � acceptance and commit- ment therapy; CBT � cognitive behavioral therapy; Rx � treatment; PI � principal investigator.

753CBT VERSUS ACT FOR ANXIETY DISORDERS

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

second year of treating patients).4 Therapists were assigned to ACT, CBT, or both (i.e., treated in both CBT and ACT, though never at the same time), depending on the need for therapists in a particular condition and the availability of training in that condi- tion (e.g., a multi-day training workshop with an ACT or CBT expert; see below). There were a total of 39 therapists; 18 thera- pists worked exclusively in CBT, 9 worked exclusively in ACT, and 11 treated both ACT and CBT participants (but never at the same time). Generally, therapists treated one to two patients at a time and three to six therapists worked within each treatment condition at a time. The mean number of patients treated by CBT-only therapists was 1.94 (SD � 1.16; range 1– 4) for a total of 35 participants, by ACT-only therapists was 1.89 (SD � 1.05; range 1– 4) for a total of 17 participants, and by therapists who treated in both conditions was 6.82 (SD � 1.60; range 4 –9) for a total of 75 participants. The mean patient number for therapists treating in both conditions was significantly higher than the mean for ACT- or CBT-only therapists ( ps � .001, �p

2 � .77 for each comparison; there were no differences between ACT- and CBT- only therapists). This is because therapists were allowed to gain training in the second treatment modality (e.g., in CBT if they started out in ACT) only if they had seen at least several patients in their original modality (e.g., in ACT). Consequently, therapists who treated in both conditions treated more patients overall. We tested the possibility, therefore, that therapists treating in different conditions would evidence systematic differences in competency that impacted study findings (see Results section).

Weekly, hour-long group supervision for study therapists was led separately by the principal authors of the treatment manuals and by advanced therapists from UCLA and from Dr. Hayes’ laboratory at the University of Nevada Reno, where ACT was originally developed. All sessions were videotaped for supervision purposes with a hidden video camera; sessions were also audio- taped for therapy adherence purposes with a discrete digital re- corder. Videos were generally played in supervision sessions or watched beforehand by supervisors. ACT supervision occurred by phone and Skype with offsite supervisors, supplemented by occa- sional face-to-face sessions, whereas CBT supervision was face- to-face. All therapists completed extensive training including an intensive 3-day workshop with the principal treatment manual author (author Craske for CBT, and author Eifert or Dr. Hayes for ACT) prior to treating participants. ACT and CBT manuals were matched on the number of sessions devoted to exposure but differed in coping skills and the framing of exposure intent.

Cognitive behavioral therapy (CBT). CBT for anxiety dis- orders followed a protocol authored by Craske (2005), which involved a single manual with branching mechanisms that listed cognitive restructuring and behavioral exposure content for each anxiety disorder. Session 1 focused on assessment, self- monitoring, and psychoeducation. Sessions 2– 4 emphasized cog- nitive restructuring with hypothesis testing, self-monitoring, and breathing retraining. Exposure (e.g., interoceptive, in-vivo, and imaginal) was tailored to the principal diagnosis and focused on empiricism and anxiety reduction over time; it was introduced in Session 5 and emphasized strongly in Sessions 6 –11. Session 12 focused on relapse prevention.

Acceptance and commitment therapy (ACT). ACT for anx- iety disorders followed a manual authored by Eifert and Forsyth (2005). Session 1 focused on psychoeducation, experiential exercises,

and discussion that introduced acceptance, creative hopelessness, and valued action. Creative hopelessness involved a process of exploring whether efforts to manage and control anxiety had “worked” and experiencing how such efforts had led to the reduction or elimination of valued life activities. Participants were encouraged to behave in ways that enacted their personal values (“valued action”), rather than spend time managing anxiety. Acceptance was explored as an alter- native to controlling anxiety. Sessions 2–3 further explored creative hopelessness and acceptance. Sessions 4 and 5 emphasized mindful- ness, acceptance, and cognitive defusion, or the process of experienc- ing anxiety-related language (e.g., thoughts, self-talk, etc.) as part of the broader, ongoing stream of present experience rather than getting stuck in responding to its literal meaning. Sessions 6 –11 continued to hone acceptance, mindfulness, and defusion and added values explo- ration and clarification with the goal of increasing willingness to pursue valued life activities. Behavioral exposures, including intero- ceptive, in-vivo, and imaginal, were employed as needed to provide opportunities to practice making room for, mindfully observing, and accepting anxiety (all types of exposure) and to practice engaging in valued activities while experiencing anxiety (in-vivo exposures). Ses- sion 12 reviewed what worked and how to continue moving forward. See the supplemental materials for additional details.

Outcome Measures

Because CBT emphasized symptom reduction, whereas ACT emphasized broader aims of psychological flexibility and valued living, we investigated two sets of primary outcomes across both treatments: anxiety-specific (i.e., symptom reduction related) and non-anxiety-specific, or broader, outcomes. For the anxiety- specific measures, we included the severity of each principal anxiety disorder. In addition, the mixed anxiety disorder nature of our sample required utilization of anxiety-specific outcome mea- sures that were relevant across the anxiety disorders. We selected measures of worry, fear, and behavioral avoidance, features that characterize all anxiety disorders (Craske et al., 2009), and empir- ically tested them to ensure that they changed following treatment across the entire sample (not merely within a single disorder). For the non-anxiety-specific, broader measures we assessed quality of life and psychological flexibility.

Anxiety-Specific Primary Outcomes

Diagnostic interview assessment. Clinical diagnoses were ascertained using the Anxiety Disorders Interview Schedule–IV (ADIS–IV; Brown, DiNardo, & Barlow, 1994). Doctoral students in clinical psychology or research assistants served as interviewers after completing 15–20 hr of training and demonstrating adequate diagnostic reliability on three consecutive interviews. “Clinical severity ratings” (CSRs) were made for each disorder by group consensus on a 0 to 8 scale (0 � none, 8 � extremely severe). Ratings of 4 or higher indicated clinical significance based on symptom severity, distress, and disablement and served as the cutoff for study eligibility (see Craske et al., 2007). ADIS–IV interviews were audiorecorded, and 15% (n � 22) were randomly

4 Prior to study training, therapists had received only one or two lectures on CBT and one lecture incorporating third-wave behavioral therapies at the point of initial study involvement.

754 ARCH ET AL.

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

selected for blind rating by a second interviewer.5 Inter-rater reliability on the principal diagnosis was 100%. Inter-rater agree- ment on dimensional CSR ratings was .65 with a single-measure, one-way mixed intraclass correlation6 coefficient (ICC) across the anxiety disorders. Inter-rater agreement for each specific disorder (met DSM–IV criteria vs. subclinical vs. none) was as follows: SAD (10 subclinical or clinical cases) and OCD (3 cases) ICC � 1.00 (100% agreement); PD/A (11 cases) ICC � .91; GAD (8 cases) ICC � .85; and SP (7 cases) ICC � .75.

The Anxiety Sensitivity Index (ASI; Peterson & Reiss, 1993; Reiss, Peterson, Gursky, & McNally, 1986)7 assesses fear of anxiety-related sensations (e.g., shortness of breath) based on the belief that such sensations are harmful. Although particularly elevated in panic disorder (Taylor, Koch, & McNally, 1992), the ASI shows elevation across most anxiety disorders (Rapee, Brown, Antony, & Barlow, 1992) relative to non-anxious controls (Peter- son & Reiss, 1993). Current sample alphas were .85 (Pre) and .93 (Post). The Penn State Worry Questionnaire (PSWQ; Meyer, Miller, Metzger, & Borkovec, 1990) assesses clinically relevant worry. Although particularly elevated in GAD, the PSWQ shows elevations across all anxiety disorders relative to non-anxious controls (Brown, Antony, & Barlow, 1992). Current sample alphas were .90 (Pre) and .93 (Post). The Fear Questionnaire’s (FQ; Marks & Mathews, 1979) Main Target Phobia Scale, a single-item avoidance rating for each participant’s “main phobia,” was used as the behavioral avoidance outcome.

Broader (Non-Anxiety-Specific) Primary Outcomes

The Quality of Life Inventory (QOLI; Frisch, 1994a, 1994b) assesses values and life satisfaction across 16 broad life domains and has good test–retest reliability and internal validity (Frisch et al., 2005). Current alphas were .86 (Pre) and .84 (Post). The Acceptance and Action Questionnaire–16 (AAQ; Bond & Bunce, 2000; Hayes et al., 2004) assesses psychological flexibility (Bond et al., 2011). The AAQ is a 16-item version of the seven- and nine-item AAQ that is hypothesized to be more sensitive to clinical change (Hayes et al., 2004). Both one- and two-factor solutions have been fit to the 16-item AAQ (Bond & Bunce, 2000, 2003). Herein, a one-factor scale was used, with higher scores indicating greater psychological flexibility. Current alphas were .78 (Pre) and .86 (Post). See supplemental materials for additional psychomet- rics on primary outcomes.

Secondary Outcomes

Use of additional treatment. As a behavioral indication of the degree to which each treatment met clients’ needs, we assessed participants’ reported use of additional (non-study-related) psy- chotherapy and psychotropic medication at Post-12mFU with questions on the ADIS–IV. At each assessment point, we com- pared groups on the portion of participants who initiated new, dropped (e.g., among participants using therapy or medication at the previous assessment point), or were using any form (e.g., either new or continued from the previous assessment) of non-study psychotherapy or psychotropic medication.

Generalization of treatment effects. Based on the ADIS–IV interview, co-occurring mood and anxiety disorders (with CSRs of 4�) at Post-12mFU were analyzed as an index of the generaliza-

tion of treatment effects. We examined the number of co-occurring anxiety disorders, the presence of co-occurring mood disorders, and the total number of co-occurring anxiety and mood disorders as indices of the generalization of treatment effects.

Treatment Credibility

Prior to the second therapy session, after the treatment rationale had been fully described, participants completed a six-item treat- ment credibility questionnaire adapted from Borkovec and Nau (1972; see supplemental materials). Total sample alpha was .94, ACT alpha was .92, and CBT alpha was .95.

Treatment Adherence and Therapist Competence

Treatment sessions were audiotaped, and 143 sessions from 91 participants (50 in CBT, 41 in ACT) were randomly selected for treatment adherence and therapist competency ratings using the Drexel University ACT/CT Therapist Adherence and Competence Rating Scale (DUACRS; McGrath, Forman, & Herbert, 2012). The treatment adherence items (n � 49) included five scales: General Therapy Adherence (12 items), General Behavioral Therapy Adherence (11 items), Cognitive Therapy Adherence (10 items), ACT Adherence (16 items), and Therapist Competence (5 items; see sup- plemental materials). The first author of the DUACRS (McGrath), who had no involvement with the current study and extensive training in both ACT and CBT, completed adherence ratings. To check treatment integrity, the blind rater (McGrath) noted which of 49 therapist adherence items (e.g., specific therapist behaviors or therapy content) occurred in each 5-min segment of therapy. Treatment-specific subscale scores (indicating adherence to ACT or CBT) were calculated by dividing the number of segments during which subscale-specific therapist behavior was present (i.e., at least one of the items composing a subscale was coded for that 5-min segment) by the total number of segments in the session, yielding an estimate of the percentage of time spent by the ther- apist on treatment-specific behavior. The General Therapy Adher- ence and General Behavioral Therapy Adherence scales were computed in similar manner. At the end of each recording, thera- pist competence was rated and the mean of the scale items repre- sented the therapist competency rating for that session.

Statistical Analyses

Raw data were inspected graphically; outliers (3 SD) were replaced with the next higher value, following the Winsor method (Guttman, 1973), prior to data analysis. In full hierarchical linear modeling (HLM) models (see below), Level 1 and 2 residuals were examined for model outliers and fit, and outliers (3 SD) were

5 Given the mixed anxiety disorder sample and subsequently low sample size per disorder, ICCs for individual disorders should be interpreted cautiously. Note, however, that agreement was based on all 22 rated audiotapes, not just the audiotapes of participants with clinically significant symptoms.

6 This test was selected because the second interviewers included several different trained assessors who rated several tapes each.

7 We used the original ASI because the revised ASI–3 (Taylor et al., 2007) had not yet been published at study initiation.

755CBT VERSUS ACT FOR ANXIETY DISORDERS

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

treated in the Winsor method and on two occasions, eliminated due to particularly strong and uncorrectable influence. Less than 5% of the data were modified or eliminated during outlier correction.

Longitudinal data were analyzed with HLM and hierarchical multiple linear modeling (HMLM) in HLM 6.0 (Raudenbush, Bryk, & Congdon, 2004). HLM/HMLM random effects models examined within- and between-group change across time (Pre, Post, 6mFU, 12mFU) and by condition (ACT and CBT). HLM/ HMLM incorporates participants with missing data by estimating the best fitting model from the data available for each participant (Hedeker & Gibbons, 2006). Therefore, for the intent-to-treat (ITT) analyses, all data points for participants who entered treat- ment were entered into the model. For completer results, which are reported when different from ITT results, analyses included par- ticipants who completed treatment and at least one subsequent assessment (Post, 6mFU, 12mFU). In fitting models to the longi- tudinal data, different variance– covariance structures were as- sessed starting with the simplest, the HLM compound symmetry model. HLM model fit was compared with multiple HMLM Level 1 variance– covariance options (homogenous, heterogeneous, first- order autoregressive, unstructured) using restricted log likelihood values (–2LL) in chi-square comparisons (see Raudenbush, Bryk, Cheong, & Congdon, 2004). The model with the best fit was selected; if models were not significantly different, the model with the fewest parameters was selected.

In the HLM/HMLM models, assessment time points (Pre, Post, 6mFU, 12mFU) were entered on Level 1 and nested within indi- viduals on Level 2. Demographic and clinical covariates and group (dummy coded ACT vs. CBT and centered at �.5, .5) were entered on Level 2. Between-group differences focused on group differ- ences in change slopes over time and at Post and 12mFU time points. In all HLM/HMLM analyses, all data points were included in the models. For the Post and 12mFU analyses, the intercept was shifted to Post or 12mFU to facilitate group comparisons at those time points. Due to curvilinear change patterns over time, qua- dratic and cubic time terms were tested on Level 1 and kept in final models if they were significant and significantly reduced model deviance (–2LL) according to chi-square-based model compari- sons. Cubic terms were fixed in order to not overestimate the model’s random effects. For analyses of non-CSR outcomes, pre- CSR was covaried on the intercept to account for pre-treatment diagnostic severity. Effect sizes for within-group change at each assessment point were examined in models that included linear and quadratic slopes only for the sake of clarity and brevity and to avoid overinflating within-group linear effect sizes, which can occur when both quadratic and cubic terms are in the model. We computed d effect sizes that accounted for the number of repeated measurement periods as needed (Feingold, 2009)8 and used Co- hen’s (1988) guidelines for interpretation. For ease of effect size comparisons between the ITT and completer samples, we used the ITT standard deviations in all effect size computations.

Differences in rate of improvement should translate into differ- ent outcomes at post-treatment or follow-up. Therefore, the Post and 12mFU time points represented our main time points of interest for examining cross sectional group differences.

For comorbidity analyses, we compared groups at Pre using chi-square and over time using generalized hierarchical linear modeling (GHLM) random effects repeated-measure models (see

Raudenbush, Bryk, Cheong, & Congdon, 2004), which utilize participants with incomplete data.

Three separate indices examined treatment response in terms of the percentage of responders at each assessment point (Post, 6mFU, 12mFU); chi-square analyses examined between-group differences. Diagnostic status improvement was defined in accordance with recent clinical trials (Newman et al., 2011; Roemer et al., 2008) as a principal diagnosis CSR of 3 or below. Reliable change was com- puted using the Jacobson and Truax (1991) method, using the more conservative denominator recommended by Maassen (2004). To re- main consistent with previous randomized clinical trials, we focused the response indices on anxiety-specific outcomes. Due to the range of principal disorders and conservative method of defining change, we examined group differences in reliable change on at least two of four anxiety-specific primary outcomes (reliable change status), as well as reliable change plus falling within 1 SD of the non-clinical normative range (or 3 or less on CSR) on at least two of four anxiety-specific primary outcomes (high end-state functioning status) and reliable change on the principal disorder CSR plus CSR of 3 or below (diagnostic change status). See supplemental materials for employed norms and computations details and Table 6 for reliable change critical values.

To assess if data were missing at random, we conducted chi-square comparisons on primary outcomes comparing participants who dropped out versus finished treatment and treatment finishers with complete versus incomplete data at 6mFU and 12mFU. For dropouts versus finishers, no significant differences emerged at Pre on any primary outcome variable. For treatment finishers with complete versus incomplete data, no significant differences emerged at Pre or Post, with the minor exception that participants with incomplete FU data had higher ASI scores at Pre ( p � .04, �p

2 � .05) but not at Post. The findings suggest that the data were missing at random.

Power analyses, conducted in Optimal Design (see Raudenbush & Liu, 2000), indicated that to reach 80% power, a cross-sectional between-group difference (e.g., at 12mFU) with an effect size of 0.70 required 67 total participants, whereas a between-group effect size of 0.50 required 126 total participants. Therefore, our total sample size (n � 128) was sufficient to detect between-group differences of moderate size at each assessment point.

Results

Pre-Treatment Group Differences

At pre-treatment, ACT and CBT evidenced no significant dif- ferences on anxiety-specific or broader outcome measures, al- though ASI differences approached significance, t(111) � 1.91, p � .058 (all other outcomes, ps � .2). Further, ACT (8.82%, 3/34) and CBT (9.68%, 3/31)9 showed no differences in use of non-study psychotherapy at Pre, �2 � 0.01, p � .91, nor in use of

8 We did not report 95% confidence intervals for our Feingold (2009) effect sizes because there is not yet a method for doing so (see Feingold, 2009; Odgaard & Fowler, 2010).

9 To comply with institutional review board requirements, the original ADIS–IVs had already been shred for the remaining half of the sample by the point at which we extracted these data. There is no reason to believe, however, that the remaining half differed from the first half in use of non-study psychotherapy.

756 ARCH ET AL.

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

psychotropic medication (see Table 1). ACT and CBT showed no significant differences on socio-demographic or clinical character- istics in Table 1, with one exception. Despite randomization pro- cedures, there was a trend for higher rates of a principal diagnosis of PD/A in CBT (49.3%) than in ACT (32.1%), �2 � 3.79, p � .052. In HLM analyses, principal PD/A diagnosis predicted supe- rior CSR outcomes compared to non-PD/A principal diagnoses, at Post (B � –1.14, SE � 0.45), t(126) � –2.53, p � .01, d � 1.24, and at 6mFU (B � –1.01, SE � 0.51), t(126) � –2.11, p � .04, d � 1.10, but not at 12mFU (B � – 0.62, p � .26). Despite the significant impact of PD/A on CSR outcomes, we did not covary PD/A in further HLM/HMLM analyses to avoid further stratifica- tion of participants into additional subgroups that lacked a priori significance and to avoid reduced statistical power to examine our principal comparison of CBT versus ACT. Principal PD/A did not predict other primary outcomes for which we found significant CBT versus ACT differences.

Treatment Credibility

Treatment credibility scores (immediately prior to Session 2) differed significantly by group, F(1, 76) � 9.08, p � .004, �p

2 � .11, with CBT evidencing higher scores (M � 6.08, SD � 1.44, n � 51) than ACT (M � 4.92, SD � 1.91, n � 27). Missing treatment credibility data from the first 24 ACT participants10

resulted in a lower sample size in this group. When the first 24 CBT participants were excluded from analyses, group differences held: F(1, 62) � 6.34, p � .01, �p

2 � .09, with CBT again showing significantly higher scores (M � 6.00, SD � 1.53, n � 37) than ACT (M � 4.92, SD � 1.91, n � 27).

Therapist Competence and Treatment Integrity

Therapist Competence scale scores (e.g., “knowledge of treat- ment,” “skill in delivering treatment,” and “relationship with client”; 1 � poor, 3 � good, 5 � excellent) indicated “good” therapist skills in CBT (M � 3.08, SD � 0.64) and ACT (M � 3.25, SD � 0.77). ACT therapists and CBT therapists did not significantly differ on competency ratings, F(1, 77) � 1.17, p � .28, �p

2 � .02. Cognitive therapy adherence scores were higher for CBT (M �

62.23, SD � 18.07) than ACT (M � 5.03, SD � 9.83), F(1, 87) � 316.88, p � .001, �p

2 � .76. Conversely, ACT adherence scores were higher for ACT (M � 82.26, SD � 18.04) than CBT (M � 3.94, SD � 6.40), F(1, 87) � 813.58, p � .001, �p

2 � .90. On the Behavioral Adherence scale, CBT (M � 47.41, SD � 23.59) scored significantly higher than ACT (M � 25.35, SD � 18.83), F(1, 87) � 22.77, p � .001, �p

2 � .21; however, this scale included a range of behavioral items such as therapist modeling that are more commonly used in CBT. We explored group differences on behavioral exposure-related items from the Behavioral Adherence scale; differences between CBT (M � 14.25, SD � 18.34) and ACT (M � 8.01, SD � 11.77) on behavioral exposure items ( p � .07, �p

2 � .04) did not reach full significance. The General Therapy Adherence scale did not differ significantly between CBT (M � 96.89, SD � 9.86) and ACT (M � 99.10, SD � 4.27), F(1, 87) � 1.72, p � .19, �p

2 � .02. The combined results show that therapists exhibited strong adherence to their assigned treatment.

To test the possibility that therapists treating in different condi- tions varied in competence, we compared CBT-only, ACT-only,

and both-type (e.g., treated participants in both CBT and ACT) therapists on competence. CBT-only (M � 2.96, SD � 0.64) and ACT-only therapists (M � 3.02, SD � 0.62) showed no differ- ences in competence ( p � .81), nor did ACT-only and both-type therapists ( p � .13). Both-type therapists, however, showed sig- nificantly higher competence (M � 3.36, SD � 0.72) than CBT- only therapists (M � 2.96, SD � 0.64), F(1, 73) � 5.18, p � .03, �p

2 � .07. To determine if therapists who treated in a single condition impacted study findings, we reran the CSR analyses twice, once without CBT-only therapists and once without CBT- and ACT-only therapists. Although reduced power from a lower sample size (which meant the analyses fell below the 80% statis- tical power level) meant that group differences were not statisti- cally significant, the pattern of findings for group differences matched those reported for the full sample below, suggesting that therapist assignment did not impact study findings. See supple- mental materials for an example of these results.

Treatment Attrition

Eighty-five of 128 participants (66%) completed the full 12 ses- sions of therapy, including 68% (48/71) in CBT and 65% (37/57) in ACT. The additional 43 participants (34%) received a partial dose of therapy: 13 participants attended one session (5 ACT, 8 CBT) and 30 participants (15 in each ACT and CBT) attended two to 11 therapy sessions. The portion of participants who did not complete the full 12 sessions did not differ by group, �2(1) � 0.40, ns. Finally, the total number of treatment sessions attended by the ITT sample did not differ by group: CBT M � 9.62 (SD � 4.08), ACT M � 9.37 (SD � 4.05), F(1, 126) � 0.12, p � .73, �p

2 � .00.

Primary Outcomes

Table 2 provides the means and standard deviations for primary outcomes at each assessment point. We conducted separate ITT and treatment completer analyses. Completer analyses are reported when they differ in significance or effect size from the ITT analyses. Effect sizes for within-group and between-group change are listed by group in Table 3. Table 4 provides the means, standard deviations, effect sizes, and diagnostic response rates for CSR outcomes at Post for each anxiety disorder. Individual disor- der outcomes were not analyzed or discussed further, however, because we did not design or power this study to examine group differences in outcomes for individual anxiety disorders.

Primary Outcome Change Slopes: ITT Sample

Anxiety-specific outcomes. With the intercept (0) represent- ing Pre, the HLM ITT model for CSR outcomes showed signifi- cant effects of linear, quadratic, and cubic change over time (all ps � .001), but no significant Group � Time interactions. How- ever, after treatment, the groups showed significant differences in CSR linear slope (within the full model accounting for group on higher order change terms and all time points with intercept representing Post) such that ACT continued to improve from Post

10 This occurred due to administrative error in which it was not under- stood that the treatment credibility measure should be administered to both the CBT and ACT participants.

757CBT VERSUS ACT FOR ANXIETY DISORDERS

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

to 12mFU, whereas CBT maintained but did not continue to improve as much, B � 0.58, SE � 0.28, t(126) � 2.03, p � .04, d � 1.26 (effect size of group slope difference from Post to 12mFU; see Figure 2A). The CSR slopes after treatment were best fit within a HMLM unrestricted covariance model.

The HLM ITT model for ASI, PSWQ, and FQ outcomes showed significant effects of linear, quadratic, and cubic change over time (all ps � .01), but no significant Group � Time inter- actions from pre- to post-treatment or after treatment.

Broader outcomes. The HLM ITT models for AAQ and QOLI outcomes showed significant effects of linear, quadratic, and cubic change over time (all ps � .001), but no significant Group � Time interactions from pre- to post-treatment or after treatment.

Primary Outcome Change Slopes: Completer Sample

Anxiety-specific outcomes. Completers evidenced a similar but smaller Group � Linear Slope interaction for CSR after treatment, B � 0.43, SE � 0.21, t(77) � 2.05, p � .04, d � 0.93, favoring ACT. A HMLM homogenous covariance structure best fit the data.

Primary Outcomes at Post: ITT Sample

Anxiety-specific and broader outcomes. At Post, ACT and CBT did not differ significantly on any anxiety-specific or broader outcome measures.

Primary Outcomes at 12mFU: ITT Sample

Anxiety-specific outcomes. At 12mFU, ACT and CBT did not differ on any anxiety-specific outcomes.

Broader outcomes. At 12mFU, CBT demonstrated higher QOLI ratings than ACT, B � 0.83, SE � 0.41, t(113) 2.05, p � .04, d � 0.42 (see Figure 2B). In addition, group differences in AAQ approached significance, B � 5.10, SE � 2.86, t(117) � 1.78, p � .08, d � 0.42, with ACT showing greater psychological

flexibility than CBT (see Figure 2C). An HLM covariance struc- ture best fit the data for these 12mFU outcomes.

Primary Outcomes at 12mFU: Completer Sample

Anxiety-specific outcomes. At 12mFU, ACT was assigned significantly lower CSR ratings for the principal anxiety disorder than CBT, B � 1.01, SE � 0.49, t(83) � 2.04, p � .04, d � 1.10.11 A homogenous Level 1 variance HMLM model best fit the CSR data. At this assessment point (12mFU), PD/A status was unrelated to CSR outcomes (see Results section), bolstering the significance of this finding. The groups did not differ significantly on the ASI, PSWQ, or FQ.

Broader outcomes. At 12mFU, CBT showed higher QOLI values than ACT, B � 0.68, SE � 0.32, t(65) � 2.12, p � .03, d � 0.34. ACT demonstrated higher AAQ scores than CBT, B � 6.76, SE � 2.65, t(66) � 2.65, p � .01, d � 0.56.

Secondary Outcomes

Use of additional psychotherapy and medication. There were no group differences at Post, 6mFU, or 12mFU in use of new or any (e.g., new or continued) psychotropic medication (all ps � .44; see supplemental materials). For dropped medication, there was no

11 Due to the fact that study selection criteria required a CSR of 4 or above, the pre-treatment CSR range was restricted. Thus, the pre-treatment standard deviation, which serves as the denominator in the Feingold (2009) effect size formula, was less than half the standard deviation of the 12mFU CSRs in magnitude (whereas pre-treatment standard deviations on other outcome measures were greater than or equal to their standard deviations at 12mFU). If we use the standard deviation at 12mFU to compute the effect size for 12mFU group differences on CSR, d � 0.44. Similarly, if we use the post-treatment CSR standard deviation to compute the group difference in ITT change slopes from Post-12mFU, d � .54.

Table 2 Means (SDs) for Primary Outcomes at Each Assessment Point

Measure and condition Pre-treatment Post-treatment 6-month follow-up 12-month follow-up

Anxiety-specific outcomes Clinical severity rating

ACT 5.70 (0.89) 3.11 (2.21) 2.77 (2.39) 2.33 (1.98) CBT 5.55 (0.94) 2.90 (2.12) 2.67 (2.24) 2.94 (2.52)

Anxiety Sensitivity Index ACT 31.81 (11.25) 18.65 (11.89) 14.56 (10.14) 17.05 (12.62) CBT 27.60 (11.81) 18.68 (11.16) 20.47 (12.90) 15.64 (8.04)

Penn State Worry Questionnaire ACT 46.52 (11.93) 39.89 (11.01) 37.79 (10.87) 39.32 (12.26) CBT 45.00 (12.82) 37.63 (15.22) 37.72 (13.04) 37.14 (12.72)

Fear Questionnaire (avoidance) ACT 5.84 (2.34) 4.13 (2.37) 4.00 (2.66) 4.28 (2.72) CBT 5.34 (2.95) 4.06 (2.96) 4.22 (3.12) 3.82 (2.70)

Broader outcomes

Quality of Life Index ACT 0.19 (1.85) 1.42 (1.88) 0.50 (1.43) 1.17 (1.51) CBT 0.55 (2.10) 1.78 (1.35) 1.45 (1.52) 1.86 (1.88)

Acceptance and Action Questionnaire–16 ACT 59.01 (12.35) 70.82 (13.14) 72.14 (10.86) 71.71 (11.42) CBT 58.49 (11.84) 69.43 (14.75) 68.38 (13.76) 68.43 (11.65)

Note. ACT � acceptance and commitment therapy; CBT � cognitive behavioral therapy.

758 ARCH ET AL.

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

group difference at 6mFU ( p � .69), with sample sizes at 12mFU too small to compare. At Post, however, CBT resulted in borderline greater dropped medication than ACT, B � 2.13, SE � 1.10, p � .05 ( p � .053), Exp(B) � 8.42 (95% CI [0.97, 73.06]), with 37.04% (10/27) of CBT versus 7.14% (1/14) of ACT participants dropping medication from Pre to Post. Because there were no group differences in overall medication use at Post ( p � .86), however, we did not further analyze this borderline significant finding.

The groups did not differ in new, dropped, or any outside psycho- therapy use at either Post or 12mFU ( ps � .26). At 6mFU, groups did not differ on new ( p � .13) psychotherapy; however, ACT reported greater use of any psychotherapy (e.g., new or continued) than CBT, B � 1.29, SE � 0.63, Wald (1) � 4.28, p � .04, Exp(B) � 3.65 (95% CI [1.07, 12.42]): 39% (11/28) of ACT participants versus 19% (5/27) of CBT participants. To explore the clinical impact of this finding, we reran the CSR analyses dropping the participants who reported any psychotherapy use at 6mFU and found that the results followed the same pattern as the ITT analysis reported above. Further, we assessed whether psychotherapy use at 6mFU predicted principal diagnosis CSRs at 6mFU or 12mFU and found that it did not ( ps � .4 for both ACT and CBT). Given these two sets of null findings, additional

psychotherapy at 6mFU was not covaried in subsequent analyses. Exploratory analyses showed, however, that ACT patients who re- mained severe (CSR 4�) following treatment were somewhat more likely to seek additional treatment than CBT patients who remained severe ( p � .07; see supplemental materials).

Generalization of treatment effects. The groups did not differ significantly in the number of co-occurring anxiety, mood, or anxiety and mood disorders combined at Pre ( ps � .18). Nor did the groups differ significantly in rates of reduction of co-occurring disorders over time, with both groups showing decreases in co- occurring disorders following treatment (see Table 5).

Treatment Response Rates

The groups did not significantly differ in treatment response rates (see Table 6).

Discussion

Within a randomized clinical trial, we aimed to test the efficacy of ACT relative to a gold-standard treatment for anxiety disorders,

Table 3 Effect Sizes (d) of Within- and Between-Subject Effects for Primary Outcomes

Outcome ACT CBT Between groups difference in d

Quadratic: Quadratic: Effect size

(d)

ACT CBT ACT CBT

Anxiety-specific outcomes CSR

Linear Pre–Post �3.74 �3.46 0.28 0.71 0.73 0.82 0.84 Linear Post–6mFU �1.98 �1.78 0.20 Linear 6–12mFU �0.46 �0.08 0.38

ASI Linear Pre–Post �1.18 �0.63 0.55 3.08 1.39 0.26 0.12 Linear Post–6mFU �0.65 �0.39 0.26 Linear 6–12mFU �0.12 �0.15 0.03

PSWQ Linear Pre–Post �0.74 �0.64 0.10 2.00 1.64 0.17 0.14 Linear Post–6mFU �0.40 �0.36 0.04 Linear 6–12mFU �0.06 �0.07 0.01

FQ Linear Pre–Post �0.83 �0.52 0.31 0.34 0.61 0.13 0.22 Linear Post–6mFU �0.38 �0.27 0.11 Linear 6–12mFU 0.05 �0.02 0.08

Broader outcomes

AAQa

Linear Pre–Post 1.16 0.90 0.26 �3.20 �2.70 �0.27 �0.22 Linear Post–6mFU 0.63 0.45 0.18 Linear 6–12mFU 0.10 0.00 0.10

QOLIa

Linear Pre–Post 0.43 0.43 0.00 �0.15 �0.21 �0.08 �0.11 Linear Post–6mFU 0.21 0.26 0.05 Linear 6–12mFU 0.00 0.12 0.12

Note. For the sake of clarity, effect sizes reflect hierarchical linear modeling (HLM) models with linear and quadratic time terms only. These partial model effect sizes differ in magnitude from the effect sizes reported in the text based on the full cubic models. Quadratic terms are invariant across assessment points. ACT � acceptance and commitment therapy; CBT � cognitive behavioral therapy; CSR � Anxiety Disorders Interview Schedule–IV (ADIS–IV) principal disorder clinical severity ratings; ASI � Anxiety Sensitivity Index; PSWQ � Penn State Worry Questionnaire; FQ � Main Target Phobia Scale (behavioral avoidance rating) from the Fear Questionnaire; AAQ � Acceptance and Action Questionnaire–16; QOLI � Quality of Life Inventory; Pre � pre-treatment; Post � post-treatment; FU � follow-up; 6 –12mFU � 6- to 12-month follow-up. a Higher scores indicate improvement.

759CBT VERSUS ACT FOR ANXIETY DISORDERS

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

CBT. Because ACT represents a transdiagnostic treatment ap- proach (Hayes et al., 1999), we focused on a mixed anxiety disorder sample. We explored the degree to which each treatment reduced anxiety symptoms and tested the hypothesis that ACT would improve on measures of quality of life and psychological flexibility to a greater extent than CBT.

Overall, the findings demonstrated that ACT and CBT did not differ significantly at post-treatment on either anxiety-specific or broader outcomes. Group differences emerged during the follow-up interval, however, with ACT showing superiority over CBT on prin- cipal disorder severity and psychological flexibility outcomes. How- ever, group differences are complicated by the fact that significantly more ACT participants utilized outside psychotherapy during the initial follow-up interval than CBT participants. Our hypothesis that ACT would improve more than CBT on broader outcomes met with limited support; on one broad outcome, ACT improved more than CBT but on the other, CBT improved more than ACT.

Primary Outcomes

On all primary outcomes, ACT and CBT showed substantial improvement from pre- to post-treatment. On anxiety-specific outcomes, within-group linear effect sizes in ACT and CBT from pre- to post-treatment ranged from very large for principal disorder severity (CSR) to moderate or large for other anxiety outcomes. Thus, both treatments were highly efficacious. Anxiety-related outcomes continued to improve through the 6-month follow-up assessment within both ACT and CBT. From 6 to 12 months, improvement slowed but treatment gains endured. On broader outcomes, both groups showed large linear improvements in psy- chological flexibility and more moderate linear improvements in quality of life from pre- to post-treatment. Broader outcomes continue to improve or endure through follow-up. In summary, ACT and CBT resulted in significant improvements from pre- to post-treatment that were maintained or improved upon during

Table 4 Means (SDs) for Primary Outcomes by Disorder

Anxiety disordera and measure Pre Postb 6mFUb 12mFUb

Linear/quadratic d effect size for Pre–Postc % CSR � 3 at Post

Panic disorder with or without agoraphobia

Clinical Severity Rating ACT (n � 18 Pre) 5.56 (0.78) 2.40 (2.27) 2.00 (2.74) 2.00 (2.08) �4.24/0.93 60.00% (6/10) CBT (n � 35 Pre) 5.57 (0.88) 2.22 (2.15) 2.59 (2.45) 2.58 (2.84) �4.02/1.02 69.56% (16/23)

Anxiety Sensitivity Index ACT (n � 17 Pre) 37.29 (6.73) 25.56 (12.39) 26.67 (12.42) 23.60 (14.12) �0.98/0.19 CBT (n � 33 Pre) 32.03 (9.57) 17.63 (10.87) 18.72 (8.96) 15.29 (8.88) �1.31/0.32

Social anxiety disorder

Clinical Severity Rating ACT (n � 13 Pre) 5.46 (0.88) 4.33 (1.73) 3.43 (2.07) 2.60 (2.07) �0.93 44.44% (4/9) CBT (n � 12 Pre) 5.75 (1.14) 3.56 (1.24) 3.33 (1.32) 4.00 (2.00) �0.59 66.67% (6/9)

FQ–Social Phobia subscale ACT (n � 11 Pre) 17.45 (9.78) 16.63 (7.60) 14.80 (9.68) 17.20 (9.52) �0.56/0.19 CBT (n � 10 Pre) 20.60 (11.01) 14.67 (8.26) 14.67 (6.67) 17.86 (6.96) �0.54/0.16

Generalized anxiety disorder

Clinical Severity Rating ACT (n � 14 Pre) 6.07 (0.73) 1.80 (1.69) 1.75 (2.19) 2.00 (1.55) �5.70/1.47 80.00% (8/10) CBT (n � 12 Pre) 5.25 (0.97) 3.00 (2.45) 0.00 (0.00) 1.50 (3.00) �2.97/0.58 44.44% (4/9)

Penn State Worry Questionnaire

ACT (n � 12 Pre) 50.50 (10.83) 35.26 (8.81) 35.17 (5.64) 36.09 (17.64) �0.58 CBT (n � 11 Pre) 52.14 (10.34) 42.43 (14.65) 37.20 (13.81) 40.75 (12.09) �0.25

Obsessive-compulsive disorder Clinical Severity Rating

ACT (n � 9 Pre) 5.67 (1.23) 4.00 (2.24) 4.00 (2.74) 2.00 (2.83) �2.67/0.52 28.57% (2/7) CBT (n � 8 Pre) 5.88 (0.84) 5.00 (1.58) 4.50 (2.38) 3.50 (2.52) �1.03/0.15 20.00% (1/5)

PI–WSUR ACT (n � 7 Pre) 36.69 (17.61) 37.50 (24.37) 28.25 (18.95) 26.98 (13.69) �0.07 CBT (n � 5 Pre) 40.60 (23.93) 27.25 (11.47) 35.00 (22.63) 28.00 (18.38) �0.25

Note. ACT � acceptance and commitment therapy; CBT � cognitive behavioral therapy; Pre � pre-treatment; Post � post-treatment; 6mFU � 6-month follow-up; 12mFU � 12-month follow-up; CSR � Clinical Severity Rating; FQ � Fear Questionnaire; PI–WSUR � Padua Inventory—Washington State University Revised. a We did not compute separate means or standard deviations for specific phobia participants because of their low numbers (n � 6). b These represent the mean and standard deviation for participants with data at each assessment point. c For ease of comparison with full-sample data reported in Table 3, we employed the standard deviation from the entire sample as the denominator in the Feingold (2009) effect size magnitude formula. This resulted in somewhat more conservative (lower) effect sizes for some measures. Further, slopes are based on change over time with all assessment points in the model (Pre through 12mFU). Therefore, effect sizes from Pre to Post reflect the slope of change across all assessment points which can differ from the slope of Pre to Post change alone.

760 ARCH ET AL.

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

follow-up, on both anxiety-specific and broader outcomes. Further, improvements were evident across two different dimensions of treatment response, namely, more objective, clinician-rated CSR outcomes as well as subjective, patient-rated self-report outcomes.

From the end of treatment through the 12-month follow-up, several noteworthy group differences emerged. On anxiety- specific outcomes, ACT demonstrated a steeper linear improve- ment rate than CBT in the principal disorder severity rating, a difference of large effect size. ACT’s steeper improvement rates resulted in lower principal disorder severity ratings than CBT at the 12-month follow-up, again of large effect size, although sta- tistically significant effects were limited to the completer sample. Over the long term, therefore, ACT more effectively reduced

principal anxiety disorder severity than CBT among those who completed treatment. This finding is consistent with a previous study (Lappalainen et al., 2007) that found that ACT resulted in more symptom improvement that CBT, albeit in a much smaller sample (n � 28) of unselected outpatients. However, since more ACT than CBT patients in the current study utilized outside therapy during the initial follow-up interval, we cannot fully de- termine whether ACT’s superiority resulted from the ACT treat- ment alone or ACT plus additional psychotherapy. Excluding patients using non-study therapy from the principal disorder se- verity analyses, however, did not change the pattern of results, suggesting that use of non-study therapy did not influence the principal disorder severity findings.

For broader outcomes, one unexpected finding was that CBT participants reported significantly higher quality of life than ACT at 12-month follow-up, a difference of moderate effect size. It had been hypothesized that the explicit focus on valued living in ACT would lead to greater improvements in quality of life. Conceiv- ably, our measure of quality of life was too general to capture values-specific improvements. Consistent with hypotheses, how- ever, ACT participants reported higher levels of psychological flexibility than CBT at 12-month follow-up on a measure specif- ically designed to capture ACT-related improvement (Hayes et al., 2004). Please refer to the supplemental materials for further dis- cussion of primary outcomes.

Finally, ACT and CBT produced similar rates of reliable change, diagnostic improvement, and high end-state functioning, comparable to our recent review showing that on average the mean response rate for CBT across anxiety disorder studies from 2000 –2011 was 51.36% at post-treatment (180 studies) and 54.80% at follow-up (71 studies; Loerinc, Meuret, Twohig, Rosenfield, & Craske, 2012). Very few studies (of the 180) used reliable change index methods to compute treatment response rates, but the few that did (e.g., Addis et al., 2004; Carter, Sbrocco, Gore, Marin, & Lewis, 2003) evidenced response rates comparable to those in the present study.

Secondary Outcomes

As noted above, more ACT than CBT patients reported non-study psychotherapy (new or continued psychotherapy) at the 6-month follow-up assessment, although there were no group differences in the initiation of new psychotherapy during this period nor in medication use, nor any group differences on these variables at 12-month follow- up. Reasons for this group difference, nonetheless, were explored. The data showed some support for the notion that ACT patients who remained distressed following treatment were more likely to seek additional psychotherapy than CBT patients who remained distressed. Another possibility is that the broader focus on exploring personal values, pursuing meaningful life behaviors, and contacting the full range of emotions in ACT inspired patients to continue engaging in psychotherapy.

Both ACT and CBT resulted in robust reductions in co-occurring mood and anxiety disorders. This finding demonstrates that treatment effects generalized in both groups, replicating and extending previous work on the broader effects of CBT for panic disorder and generalized anxiety disorder (e.g., Borkovec, Abel, & Newman, 1995; Tsao, Mystkowski, Zucker, & Craske, 2005).

A. Principal disorder severity (CSR)

B. Quality of Life Index (QOLI)

C. Acceptance and Action Questionnaire-16 (AAQ)

pre post 6m 12m

ACT

CBT

0.5

1.5

pre post 6m 12m

ACT

CBT

pre post 6m 12m

ACT

CBT

Figure 2. A–C: Primary outcomes in the intent-to-treat (ITT) sample with acceptance and commitment therapy (ACT) versus cognitive behavioral therapy (CBT) differences. CSR � Clinical Severity Rating; m � month.

761CBT VERSUS ACT FOR ANXIETY DISORDERS

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

Treatment and Therapist Variables

Attrition rates were relatively high across both ACT and CBT; however, no group differences emerged. Attrition was comparable to some large trials of CBT for anxiety disorders (e.g., Barlow et al., 2000) but was higher than the mean attrition (23%) reported in a meta-analysis of CBT studies for anxiety disorders (Hofmann & Smits, 2008). Although attrition reasons for many patients remain unknown, we suspect that four features of our study may have contributed to attrition. First, the study took place in a difficult to locate, high-traffic, parking-challenged clinic with limited public

transportation options, and travel times to and from treatment often exceeded 45 min each way. Second, our assessments included a 2- to 3-hr physiological laboratory session that was strongly anxiety provoking for many patients (e.g., involving hyperventilation, neg- ative picture slide viewing), which may have discouraged patients from study completion. Third, unlike many studies where treat- ment is free or low cost for all, most patients paid for treatment and incurred significant parking costs. Fourth, we offered few incen- tives for treatment completion and failed to sufficiently incentivize post-treatment and follow-up assessment completion. Higher in-

Table 5 Co-Occurring Disorders: Rates and Improvement Among Participants With Data at Each Time Point

% of participants with co-occurring disorders ACT CBT Between-group �2

ACT: slope over time, odds ratio

[CI]a‘

CBT: slope over time, odds

ratio [CI]

Pre-treatment Anxiety disorders 39.3% (22/56) 28.2% (20/71) p � .19 Mood disorders 23.2% (13/56) 23.9% (17/71) p � .92 Mood or anxiety 52.8% (29/56) 41.2% (28/68) p � .24

Post-treatment Anxiety disorders 10.8% (4/37) 8.7% (4/46) p � .75 0.38�� [0.24, 0.59] 0.47�� [0.29, 0.78] Mood disorders 8.1% (3/37) 2.2% (1/46) p � .21 0.43�� [0.23, 0.83] 0.36�� [0.17, 0.76] Mood or anxiety 18.9% (7/37) 8.7% (4/46) p � .17 0.41�� [0.26, 0.63] 0.44�� [0.26, 0.75]

6mFU Anxiety disorders 7.7% (2/26) 5.6%(2/36) p � .85 Mood disorders 3.8% (1/26) 2.8% (1/36) p � .85 Mood or anxiety 11.5% (3/26) 5.6% (2/36) p � .77

12mFU Anxiety disorders 5.9% (1/17) 7.4% (2/27) p � .85 Mood disorders 5.9% (1/17) 7.4% (2/27) p � .85 Mood or anxiety 11.8% (2/17) 14.8% (4/27) p � .77

Note. ACT � acceptance and commitment therapy; CBT � cognitive behavioral therapy; 6mFU � 6-month follow-up; 12mFU � 12-month follow-up. a The odds ratios are based on fixed generalized hierarchical linear modeling (GHLM) linear time slopes for within-group change over time for dichotomous outcomes, covarying pre-treatment Clinical Severity Rating (CSR) for the principal diagnosis at the model’s intercept. They are reported here at post-treatment but characterize the linear change rate over the pre-treatment to 12mFU period. �� p � .01. �� p � .001.

Table 6 Response Rates on Treatment Response Indices

Assessment CBT ACT �2 p Cramer’s v

Shows reliable changea on at least two of four anxiety-specific outcomes

Post-treatment 44.4% (16/36) 56.7% (17/30) 0.98 .32 .12 6-month FU 53.6% (15/28) 47.1% (8/17) 0.09 .76 .05 12-month FU 52.2% (12/23) 52.6% (10/19) 0.00 .98 .01

Shows reliable changea and high end-state functioning based on two of four anxiety-specific outcomes

Post-treatment 47.2% (17/36) 50.0% (15/30) 0.05 .82 .03 6-month FU 53.6% (15/28) 47.1% (8/17) 0.18 .67 .06 12-month FU 39.1% (9/23) 47.4% (9/19) 0.29 .59 .08

Shows reliable CSR changea and does not meet clinical diagnostic criteria (CSR � 3)

Post-treatment 51.0% (25/49) 44.7% (17/38) 0.34 .56 .06 6-month FU 59.5% (22/37) 44.4% (12/27) 1.41 .24 .15 12-month FU 50.0% (16/32) 54.5% (12/22) 0.11 .74 .05

Note. ACT � acceptance and commitment therapy; CBT � cognitive behavioral therapy; CSR � Clinical Severity Rating; FU � follow-up. a Reliable change required the following minimum improvement values from pre-treatment (see supplemental materials for computational details): principal disorder CSR � 2.75, Penn State Worry Questionnaire � 10.03, Anxiety Sensitivity Index � 10.48, Fear Questionnaire Main Target Phobia Avoidance rating � 1.97.

762 ARCH ET AL.

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

centives may have been particularly needed in a study with such significant treatment barriers (e.g., long travel times, parking fees and difficulties, treatment fees).

An important finding to emerge from blind treatment integrity ratings was that ACT and CBT were clearly distinguished from one another and that therapists strongly adhered to their designated treatment. Despite the use of novice student therapists, therapists averaged “good” overall skills across both treatments with no differences between treatment groups. Therapists treating patients in both CBT and ACT nonetheless evidenced significantly higher competence than therapists treating in CBT only. Principal diag- nostic severity analyses of patients treated only by “both-type” therapists showed the same pattern of outcomes, however, sug- gesting that these differences did not impact overall study findings.

Early in treatment, CBT was rated as a more credible treatment than ACT by a medium effect size. Thus, ACT therapists were not as successful as CBT therapists in convincing patients early on that they offered a credible treatment. This difference did not appear to influence attrition rates, which did not differ by group. Conceiv- ably, abstract ideas of acceptance and creative hopelessness in initial ACT sessions (rather than concrete skills in CBT) contrib- uted to a diminished sense of treatment credibility. Future studies should assess treatment credibility regularly throughout ACT to determine whether ACT follows a delayed trajectory in convincing patients that it offers something credible, or whether ACT patients remain skeptical throughout treatment but improve anyway. The latter would suggest that treatment credibility is relatively unim- portant to the success of ACT. Certainly, in the current study, the lowered credibility ratings relative to CBT did not appear to disadvantage ACT outcomes relative to CBT outcomes.

Study Limitations

Several study limitations should be noted. First, our mixed anxiety disorder sample limits the conclusions that may be drawn about any single anxiety disorder. Anxiety disorders typically co-occur at high rates with other anxiety disorders and share many common features (see Barlow, 2002; Craske et al., 2009), how- ever, strengthening the ecological validity of this approach. Sec- ond, relatively high attrition rates may have resulted in underesti- mated treatment effects or compromised ability to accurately assess treatment-related improvements in the ITT sample. On the other hand, we utilized a sophisticated statistical approach (HLM, HMLM) that utilized patients with incomplete data and drew upon all available data in the ITT analyses; we also conducted separate completers analyses. Third, we did not assess therapist allegiance, which may have impacted treatment results given that the study was conducted within a CBT-renowned research clinic. Based on the relatively inexperienced and junior nature of the therapists, however, allegiance is unlikely to be a significant factor. Therapist experience raises another limitation, which is that the results may differ in the hands of more experienced therapists. Fourth, CBT supervision was conducted onsite in a face-to-face manner, whereas most ACT supervision was conducted via phone or Skype with offsite supervisors. We did not assess supervision quality and thus could not investigate the impact of this group difference. Fifth, we did not systematically assess reasons for attrition and, thus, could not assess whether ACT and CBT differed in the extent to which patients dropped out because they were unsatisfied with

treatment. Future ACT/CBT studies should assess group differ- ences in stated reasons for attrition. Sixth, we utilized a single-item rating from the Fear Questionnaire for behavioral avoidance due to the lack of avoidance measures relevant to all anxiety disorders. Seventh, we used a website to generate the randomization se- quence, whereas use of an external agency would have been preferable. Eighth, we did not include a no-treatment or treatment- as-usual control group, which may have obscured our capacity to assess improvement due to treatment versus the passage of time. It has been argued, however, that comparing a newer to a well- established treatment does not require a no-treatment or waitlist control group and is more ethical without one (Kazdin, 2002). Also, the roughly equal number of treatment sessions devoted to behavioral exposure in ACT and CBT may have obscured treat- ment differences. The two treatment conditions were matched on exposure, albeit framed with different intents, given the potency of exposure as a change agent. This feature may have altered the way that ACT is typically done. In addition, the Penn State Worry Questionnaire scores at pre-treatment in the GAD subsample were considerably lower than those in recently published randomized trials for GAD (Newman et al., 2011; Roemer et al., 2008), which may hold implications for the interpretation of findings in this subgroup. Finally, regarding the generalizability of our findings, our sample largely reflected the racial, ethnic, and sex distribution of U.S. residents at the time of data collection (U.S. Census Bureau, 2012), supporting the broad generalizability of our find- ings. On the other hand, our sample was relatively educated (the average participant had completed 3.5 years of college), and thus, our findings may not generalize to less educated samples. Further, we cannot assume that treatment was equally efficacious across racial subgroups because we lacked the statistical power to exam- ine whether outcomes differed by race. This remains an important question for future research.

Summary and Conclusion

To our knowledge, this study represents the first randomized clin- ical trial comparing ACT and CBT for anxiety disorders. Despite differences in underlying treatment models, the overall findings are characterized by similarities in the immediate and long-term impact of both treatments. We have argued elsewhere (Arch & Craske, 2008) that ACT and CBT for anxiety disorders may represent different approaches to affecting common therapeutic changes. This study largely supports this hypothesis. On the other hand, some differences did emerge, in that CBT resulted in higher quality of life, whereas ACT resulted in greater psychological flexibility, and, among those who completed treatment, lower principal anxiety disorder severity, over the follow-up. Overall, our findings suggest that ACT is a highly viable treatment alternative to CBT, the current gold-standard psy- chosocial treatment for anxiety disorders. Further, they pave the way for future investigations of for whom each treatment approach is most effective and the shared versus unique mechanisms of therapeutic change.

References

Addis, M. E., Hatgis, C., Bourne, L., Krasnow, A. D., Jacob, K., & Mansfield, A. (2004). Effectiveness of cognitive– behavioral treatment for panic disorder versus treatment as usual in a managed care setting.

763CBT VERSUS ACT FOR ANXIETY DISORDERS

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

Journal of Consulting and Clinical Psychology, 72, 625– 635. doi: 10.1037/0022-006X.72.4.625

American Psychiatric Association. (1994). Diagnostic and statistical man- ual of mental disorders (4th ed.). Washington, DC: Author.

Arch, J. J., & Craske, M. G. (2008). Acceptance and commitment therapy and cognitive behavioral therapy for anxiety disorders: Different treat- ments, similar mechanisms? Clinical Psychology: Science and Practice, 15, 263–279. doi:10.1111/j.1468-2850.2008.00137.x

Barlow, D. H. (2002). Anxiety and its disorders: The nature and treatment of anxiety and panic (2nd ed.). New York, NY: Guilford Press.

Barlow, D. H., & Cerny, J. A. (1988). Psychological treatment of panic. New York, NY: Guilford Press.

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, 283, 2529 –2536. doi:10.1001/jama.283.19.2529

Beck, A. T., Emery, G., & Greenberg, R. L. (1985). Anxiety disorders and phobias: A cognitive perspective. New York, NY: Basic Books.

Bond, F. W., & Bunce, D. (2000). Mediators of change in emotion-focused and problem-focused worksite stress management interventions. Journal of Occu- pational Health Psychology, 5, 156 –163. doi:10.1037/1076-8998.5.1.156

Bond, F. W., & Bunce, D. (2003). The role of acceptance and job control in mental health, job satisfaction, and work performance. Journal of Applied Psychology, 88, 1057–1067. doi:10.1037/0021-9010.88.6.1057

Bond, F. W., Hayes, S. C., Baer, R. A., Carpenter, K. C., Guenole, N., Orcutt, H. K., . . . Zettle, R. D. (2011). Preliminary psychometric prop- erties of the Acceptance and Action Questionnaire—II: A revised mea- sure of psychological flexibility and experiential avoidance. Behavior Therapy, 42, 676 – 688.

Borkovec, T. D., Abel, J. L., & Newman, H. (1995). Effects of psycho- therapy on comorbid conditions in generalized anxiety disorder. Journal of Consulting and Clinical Psychology, 63, 479 – 483. doi:10.1037/0022- 006X.63.3.479

Borkovec, T. D., & Nau, S. D. (1972). Credibility of analogue therapy rationales. Journal of Behavior Therapy and Experimental Psychiatry, 3, 257–260. doi:10.1016/0005-7916(72)90045-6

Brewin, C. R. (1996). Theoretical foundations of cognitive-behavioral therapy for anxiety and depression. Annual Review of Psychology, 47, 33–57. doi:10.1146/annurev.psych.47.1.33

Brown, T. A., Antony, M. M., & Barlow, D. H. (1992). Psychometric properties of the Penn State Worry Questionnaire in a clinical anxiety disorders sample. Behaviour Research and Therapy, 30, 33–37. doi: 10.1016/0005-7967(92)90093-V

Brown, T. A., & Barlow, D. H. (1995). Long-term outcome in cognitive- behavioral treatment of panic disorder: Clinical predictors and alterna- tive strategies for assessment. Journal of Consulting and Clinical Psy- chology, 63, 754 –765. doi:10.1037/0022-006X.63.5.754

Brown, T. A., DiNardo, P. A., & Barlow, D. H. (1994). Anxiety Disorders Interview Schedule for DSM–IV. Albany, NY: Center for Stress and Anxiety Disorders.

Butler, A. C., Chapman, J. E., Forman, E. M., & Beck, A. T. (2006). The empirical status of cognitive-behavioral therapy: A review of meta- analyses. Clinical Psychology Review, 26, 17–31. doi:10.1016/ j.cpr.2005.07.003

Carter, M. M., Sbrocco, T., Gore, K. L., Marin, N. W., & Lewis, E. L. (2003). Cognitive– behavioral group therapy versus a wait-list control in the treatment of African American women with panic disorder. Cogni- tive Therapy and Research, 27, 505–518. doi:10.1023/A: 1026350903639

Chambless, D. L., & Hollon, S. D. (1998). Defining empirically supported therapies. Journal of Consulting and Clinical Psychology, 66, 7–18. doi:10.1037/0022-006X.66.1.7

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Craske, M. G. (2003). Origins of phobias and anxiety disorders: Why more women than men? Oxford, England: Elsevier.

Craske, M. G. (2005). Cognitive-behavioral treatment of anxiety disorders. Unpublished manuscript.

Craske, M. G., Farchione, T. J., Allen, L. B., Barrios, V., Stoyanova, M., & Rose, R. (2007). Cognitive behavioral therapy for panic disorder and comorbidity: More of the same or less of more? Behaviour Research and Therapy, 45, 1095–1109. doi:10.1016/j.brat.2006.09.006

Craske, M. G., Rauch, S. L., Ursano, R., Prenoveau, J., Pine, D. S., & Zinbarg, R. E. (2009). What is an anxiety disorder? Depression and Anxiety, 26, 1066 –1085. doi:10.1002/da.20633

Craske, M. G., Roy-Byrne, P., Stein, M. B., Sullivan, G., Hazlett-Stevens, H., Bystritsky, A., & Sherbourne, C. (2006). CBT intensity and outcome for panic disorder in a primary care setting. Behavior Therapy, 37, 112–119. doi:10.1016/j.beth.2005.05.003

Craske, M. G., Stein, M. B., Sullivan, G., Sherbourne, C., Bystritsky, A., Rose, R. D., . . . Roy-Byrne, P. (2011). Disorder-specific impact of coordinated anxiety learning and management treatment for anxiety disorders in primary care. Archives of General Psychiatry, 68, 378 –388. doi:10.1001/archgenpsychiatry.2011.25

Dalrymple, K. L., & Herbert, J. D. (2007). Acceptance and commitment therapy for generalized social anxiety disorder: A pilot study. Behavior Modification, 31, 543–568. doi:10.1177/0145445507302037

Eifert, G. H., & Forsyth, J. P. (2005). Acceptance and commitment therapy for anxiety disorders: A practitioner’s treatment guide to using mind- fulness, acceptance, and values-based behavior change strategies. New York, NY: Guilford Press.

Eifert, G. H., Forsyth, J. P., Arch, J. J., Keller, M., Langer, D., & Espejo, N. (2009). Acceptance and commitment therapy for anxiety disorders: Case studies using a unified treatment protocol. Cognitive and Behav- ioral Practice, 16, 368 –385. doi:10.1016/j.cbpra.2009.06.001

Feingold, A. (2009). Effect sizes for growth-modeling analysis for con- trolled clinical trials in the same metric as for classical analysis. Psy- chological Methods, 14, 43–53. doi:10.1037/a0014699

Forman, E. M., Herbert, J. D., Moitra, E., Yeomans, P. D., & Geller, P. A. (2007). A randomized controlled effectiveness trial of acceptance and commitment therapy and cognitive therapy for anxiety and depression. Behavior Modification, 31, 772–799. doi:10.1177/0145445507302202

Frisch, M. B. (1994a). Manual and treatment guide for the Quality of Life Inventory. Minneapolis, MN: Pearson Assessments.

Frisch, M. B. (1994b). Quality of Life Inventory (QOLI). Minneapolis, MN: National Computer Systems.

Frisch, M. B., Clark, M. P., Rouse, S. V., Rudd, M. D., Paweleck, J., & Greenstone, A. (2005). Predictive and treatment validity of life satisfac- tion and the Quality of Life Inventory. Assessment, 12, 66 –78. doi: 10.1177/1073191104268006

Guttman, I. (1973). Premium and protection of several procedures for dealing with outliers when sample sizes are moderate and large. Tech- nometrics, 15, 385– 404.

Hayes, S. C., Strosahl, K. D., & Wilson, K. G. (1999). Acceptance and commitment therapy: An experiential approach to behavior change. New York, NY: Guilford Press.

Hayes, S. C., Strosahl, K., Wilson, K. G., Bissett, R. T., Pistorello, J., Toarmino, D., . . . McCurry, S. M. (2004). Measuring experiential avoidance: A preliminary test of a working model. Psychological Record, 54, 553–578.

Hedeker, D., & Gibbons, R. D. (2006). Longitudinal data analysis. Hobo- ken, NJ: Wiley-Interscience.

Hofmann, S. G., & Smits, J. A. J. (2008). Cognitive-behavioral therapy for adult anxiety disorders: A meta-analysis of randomized placebo- controlled trials. Journal of Clinical Psychiatry, 69, 621– 632.

Jacobson, N. S., & Truax, P. (1991). Clinical significance: A statistical approach to defining meaningful change in psychotherapy research. Journal of Consulting and Clinical Psychology, 59, 12–19. doi:10.1037/ 0022-006X.59.1.12

764 ARCH ET AL.

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .

Kazdin A. E. (2002). Research design in clinical psychology (4th ed.). Needham Heights, MA: Allyn & Bacon.

Lappalainen, R., Lehtonen, T., Skarp, E., Taubert, E., Ojanen, M., & Hayes, S. C. (2007). The impact of CBT and ACT models using psychology trainee therapists: A preliminary controlled effectiveness trial. Behavior Modification, 31, 488 –511. doi:10.1177/ 0145445506298436

Loerinc, A. G., Meuret, A. E., Twohig, M. P., Rosenfield, D., & Craske, M. G. (2012). CBT for anxiety disorders: Treatment responder criteria and response rates. Manuscript submitted for publication.

Longmore, R. J., & Worrell, M. (2007). Do we need to challenge thoughts in cognitive behavioral therapy? Clinical Psychology Review, 27, 173– 187. doi:10.1016/j.cpr.2006.08.001

Maassen, G. H. (2004). The standard error in the Jacobson and Truax Reliable Change Index: The classical approach to the assessment of reliable change. Journal of the International Neuropsychological Soci- ety, 10, 888 – 893. doi:10.1017/S1355617704106097

Marks, I. M., & Mathews, A. M. (1979). Brief standard self-rating for phobic patients. Behaviour Research and Therapy, 17, 263–267. doi: 10.1016/0005-7967(79)90041-X

McGrath, K. B., Forman, E. M., & Herbert, J. D. (2012). Development and validation of the ACT/CT Adherence and Competence Rating Scale. Manuscript in preparation.

Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Development and validation of the Penn State Worry Questionnaire. Behaviour Research and Therapy, 28, 487– 495. doi:10.1016/0005- 7967(90)90135-6

Newman, M. G., Castonguay, L., Borkovec, T. D., Fisher, A. J., Boswell, J. F., Szkodny, L. E., & Nordberg, S. S. (2011). A randomized controlled trial of cognitive-behavioral therapy for generalized anxiety disorder with integrated techniques from emotion-focused and interpersonal ther- apies. Journal of Consulting and Clinical Psychology, 79, 171–181. doi:10.1037/a0022489

Norton, P. J., & Price, E. C. (2007). A meta-analytic review of adult cognitive-behavioral treatment outcome across the anxiety disorders. Journal of Nervous and Mental Disease, 195, 521–531. doi:10.1097/ 01.nmd.0000253843.70149.9a

Odgaard, E. C., & Fowler, R. L. (2010). Confidence intervals for effect sizes: Compliance and clinical significance in the Journal of Consulting and Clinical Psychology. Journal of Consulting and Clinical Psychol- ogy, 78, 287–297. doi:10.1037/a0019294

Orsillo, S. M., & Batten, S. V. (2005). Acceptance and commitment therapy in the treatment of posttraumatic stress disorder. Behavior Mod- ification, 29, 95–129. doi:10.1177/0145445504270876

Peterson, R. A., & Reiss, S. (1993). Anxiety Sensitivity Index Revised test manual. Worthington, OH: International Diagnostic Systems.

Rapee, R. M., Brown, T. A., Antony, M. M., & Barlow, D. H. (1992). Response to hyperventilation and inhalation of 5.5% carbon dioxide- enriched air across the DSM–III–R anxiety disorders. Journal of Abnor- mal Psychology, 101, 538 –552. doi:10.1037/0021-843X.101.3.538

Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., & Congdon, R. (2004).

HLM 6: Hierarchical linear and nonlinear modeling. Lincolnwood, IL: Scientific Software International.

Raudenbush, S. W., Bryk, A. S., & Congdon, R. (2004). HLM 6 for Windows [Computer software]. Lincolnwood, IL: Scientific Software International.

Raudenbush, S. W., & Liu, X.-F. (2000). Statistical power and optimal design for multisite randomized trials. Psychological Methods, 5, 199 – 213. doi:10.1037/1082-989X.5.2.199

Reiss, S., Peterson, R., Gursky, D., & McNally, R. (1986). Anxiety sensitivity, anxiety frequency, and the prediction of fearfulness. Behav- iour Research and Therapy, 24, 1– 8. doi:10.1016/0005- 7967(86)90143-9

Roemer, L., & Orsillo, S. M. (2007). An open trial of an acceptance-based behavior therapy for generalized anxiety disorder. Behavior Therapy, 38, 72– 85. doi:10.1016/j.beth.2006.04.004

Roemer, L., Orsillo, S. M., & Salters-Pedneault, K. (2008). Efficacy of an acceptance-based behavior therapy for generalized anxiety disorder: Evaluation in a randomized controlled trial. Journal of Consulting and Clinical Psychology, 76, 1083–1089. doi:10.1037/a0012720

Taylor, S., Koch, W. J., & McNally, R. (1992). How does anxiety sensi- tivity vary across the anxiety disorders? Journal of Anxiety Disorders, 6, 249 –259. doi:10.1016/0887-6185(92)90037-8

Taylor, S., Zvolensky, M. J., Cox, B. J., Deacon, B., Heimberg, R. G., Ledley, D. R., . . . Cardenas, S. J. (2007). Robust dimensions of anxiety sensitivity: Development and initial validation of the Anxiety Sensitivity Index—3. Psychological Assessment, 19, 176 –188. doi:10.1037/1040- 3590.19.2.176

Tolin, D. F. (2010). Is cognitive behavioral therapy more effective than other therapies? A meta-analytic review. Clinical Psychology Review, 30, 710 –720. doi:10.1016/j.cpr.2010.05.003

Tsao, J. C., Mystkowski, J. L., Zucker, B. G., & Craske, M. G. (2005). Impact of cognitive-behavioral therapy for panic disorder on comorbid- ity: A controlled investigation. Behaviour Research and Therapy, 43, 959 –970. doi:10.1016/j.brat.2004.11.013

Twohig, M. P., Hayes, S. C., Plumb, J. C., Pruitt, L. D., Collins, A. B., Hazlett-Stevens, H., & Woidneck, M. R. (2010). A randomized clinical trial of acceptance and commitment therapy vs. progressive relaxation training for obsessive compulsive disorder. Journal of Consulting and Clinical Psychology, 78, 705–716. doi:10.1037/a0020508

U.S. Census Bureau. (2012). Statistical abstract of the United States: 2012. Table 10. Resident population by race, Hispanic origin, and age: 2000 – 2009. Retrieved from http://www.census.gov/compendia/statab/2012/ tables/12s0010.pdf

Wetherell, J. L., Afari, N., Ayers, C. R., Stoddard, J. A., Ruberg, J., Sorrell, J. T., . . . Patterson, T. L. (2011). Acceptance and commitment therapy for generalized anxiety disorder in older adults: A preliminary report. Behavior Therapy, 42, 127–134. doi:10.1016/j.beth.2010.07.002

Received August 12, 2011 Revision received March 12, 2012

Accepted March 13, 2012 �

765CBT VERSUS ACT FOR ANXIETY DISORDERS

T hi

s do

cu m

en t i

s co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

tio n

or o

ne o

f i ts

a lli

ed p

ub lis

he rs

. T

hi s

ar tic

le is

in te

nd ed

s ol

el y

fo r t

he p

er so

na l u

se o

f t he

in di

vi du

al u

se r a

nd is

n ot

to b

e di

ss em

in at

ed b

ro ad

ly .