Efficacy of an Acceptance-Based Behavior Therapy for Generalized Anxiety Disorder: Evaluation in a Randomized Controlled Trial Lizabeth Roemer University of Massachusetts Boston Susan M. Orsillo Suffolk University and Boston University Kristalyn Salters-Pedneault Veterans Affairs Boston Healthcare System and Boston University School of Medicine Generalized anxiety disorder (GAD) is a chronic anxiety disorder, associated with comorbidity and impairment in quality of life, for which improved psychosocial treatments are needed. GAD is also associated with reactivity to and avoidance of internal experiences. The current study examined the efficacy of an acceptance-based behavioral therapy aimed at increasing acceptance of internal experi- ences and encouraging action in valued domains for GAD. Clients were randomly assigned to immediate (n 15) or delayed (n 16) treatment. Acceptance-based behavior therapy led to statistically significant reductions in clinician-rated and self-reported GAD symptoms that were maintained at 3- and 9-month follow-up assessments significant reductions in depressive symptoms were also observed. At posttreat- ment assessment 78% of treated participants no longer met criteria for GAD and 77% achieved high end-state functioning these proportions stayed constant or increased over time. As predicted, treatment was associated with decreases in experiential avoidance and increases in mindfulness. Keywords: generalized anxiety disorder, mindfulness, experiential avoidance, worry, RCT Supplemental materials: http://dx.doi.org/10.1037/a0012720.supp Although efficacious individual cognitive behavioral therapies (CBT) have been developed for generalized anxiety disorder (GAD), a large proportion of individuals treated fail to meet criteria for high end-state functioning (see Waters & Craske, 2005, for a review), suggesting that further treatment development may be needed. A range of novel approaches are being explored (see Heimberg, Turk, & Mennin, 2004, for reviews). Our efforts have focused on an individual acceptance-based behavior therapy (ABBT) that targets experiential avoidance (attempts to alter the intensity or frequency of unwanted internal experiences Hayes, Wilson, Gifford, Follette, & Strosahl, 1996) using strategies aimed at increasing awareness and intended action in important life domains. Research suggests that individuals with GAD negatively eval- uate internal experiences, such as thoughts, emotions, and physi- ological sensations, and use worry, along with other strategies, as a means of escaping or avoiding these experiences. Individuals with GAD report distress about a wide range of emotions (e.g., Mennin, Heimberg, Turk, & Fresco, 2005), view their worrisome thoughts as dangerous and uncontrollable (Wells & Carter, 1999), and report a lack of self-compassion toward their own internal experiences (Roemer et al., in press). Engaging in the worry process reduces autonomic reactivity and distracts worriers from more distressing topics (Borkovec, Alcaine, & Behar, 2004). Di- rectly targeting these problematic relationships and responses to internal experiences may improve the efficacy of GAD treatments. Although behavioral exposure has not been a focus of GAD treatment, individuals with GAD do avoid anxiety-provoking sit- uations (Butler, Gelder, Hibbert, Cullington, & Klimes, 1987). In addition, clients describe making behavioral choices aimed at decreasing anxiety, rather than maximizing satisfaction, and being Lizabeth Roemer, Department of Psychology, University of Massachu- setts Boston Susan M. Orsillo, Department of Psychology, Suffolk Uni- versity, and Department of Psychology, Boston University Kristalyn Salters-Pedneault, National Center for Posttraumatic Stress Disorder, Vet- erans Affairs Boston Healthcare System, and Department of Psychiatry, Boston University School of Medicine. This study was supported by National Institute of Mental Health Grant No. MH63208 to Lizabeth Roemer and Susan M. Orsillo. Portions of these data were presented at the 2004 and 2007 annual meetings of the Associ- ation for Behavioral and Cognitive Therapies (formerly Association for Advancement of Behavior Therapy). We thank Dave Barlow, Tim Brown, and the staff at the Center for Anxiety and Related Disorders for their support of this research, and Tim Brown for assistance with revisions of this article. We also thank Tom Borkovec and Steve Hayes for their helpful consultations. We thank our therapists, Laura Allen, Gabrielle Liverant, Jill Stoddard, Matthew Tull, and Yonit Schorr, as well as our clients, for sharing their experience and their wisdom with us. Finally, we also thank Laura Allen for her excep- tional management of the project, Heidi Barrett-Model, Darren Holowka, and Matthew Tull for their therapy integrity ratings, and Shannon Erisman, Cathryn Freid, Michael Treanor, Matthew Tull, and Pete Vernig for their invaluable assistance with data management. Correspondence concerning this article should be addressed to Liza- beth Roemer, Department of Psychology, University of Massachusetts Boston, 100 Morrissey Boulevard, Boston, MA 02125. E-mail: Lizabeth.Roemer@umb.edu Journal of Consulting and Clinical Psychology Copyright 2008 by the American Psychological Association 2008, Vol. 76, No. 6, 1083���1089 0022-006X/08/$12.00 DOI: 10.1037/a0012720 1083

distracted by worries when they are engaged in important activi- ties. Therefore, an explicit focus on mindful behavioral engage- ment in valued actions (Wilson & Murrell, 2004) may be benefi- cial (for an extensive review of the empirical and theoretical rationale for ABBT for GAD, see Roemer & Orsillo, 2005, 2007). We developed an ABBT for GAD, drawing explicitly from cognitive behavioral interventions for GAD (e.g., Borkovec, New- man, Lytle, & Pincus, 2002), as well as acceptance and commit- ment therapy (Hayes, Strosahl, & Wilson, 1999), dialectical be- havior therapy (Linehan, 1993), and mindfulness-based cognitive therapy (Segal, Williams, & Teasdale, 2002). A small open trial of ABBT for GAD (Roemer & Orsillo, 2007) revealed promising findings. The current study expands this work by comparing ABBT to a waiting list condition and examining the durability of effects over a 9-month follow-up period. Method Participants Thirty-one clients consented to participation and were ran- domized to treatment (n 15) or waiting list control (n 16).1 Two participants withdrew from therapy and 4 from the waiting list. The remaining waiting list participants received delayed treatment, except for 1 participant who no longer met criteria for GAD after getting off the waiting list. A participant diag- nosed with GAD in partial remission following the waiting list period whose remaining symptoms were rated above the clinical cutoff received treatment and was included in subsequent anal- yses. One waiting list participant withdrew from therapy. Indi- viduals with a principal diagnosis of GAD (excluding the Di- agnostic and Statistical Manual of Mental Disorders [DSM���IV American Psychiatric Association, 1994] hierarchical rule that GAD could not occur only within the course of a mood disor- der),2 who did not report current suicidal intent, did not meet criteria for current bipolar disorder, substance dependence, or psychotic disorders, and were at least 18 years old were eligible for the study regardless of previous treatment history.3 See Table 1 for participant characteristics conditions did not differ significantly on demographic variables. See Figure 1 for a diagram of client enrollment throughout the study. Measures All assessments included primary measures of anxiety and worry, secondary measures of depression and quality of life, and measures of proposed mechanisms of change (experiential avoidance and mind- fulness). The Anxiety Disorders Interview Schedule for DSM���IV��� Lifetime Version (ADIS-IV DiNardo, Brown, & Barlow, 1994) was used to determine current and lifetime DSM���IV diagnostic status (an abbreviated version focusing on current diagnoses was given for posttreatment and follow-up assessments). The ADIS-IV includes a clinical severity rating (CSR) for each diagnosis received ranging from 0 to 8, with 4 being the diagnostic cutoff. All assessments were administered by doctoral students at the Center for Anxiety and Related Disorders (CARD) who had undergone extensive training and had demonstrated reliability in diagnosis.4 Diagnoses were confirmed in consensus meetings with a doctoral-level psychologist (Dr. T. A. Brown) and by therapists in their initial meetings. In a study con- ducted at CARD, the ADIS-IV had a reliability for principal GAD diagnoses of k .67 and for CSR ratings of GAD of k .72 (T. A. Brown, DiNardo, Lehman, & Campbell, 2001). During the time period of this study at CARD, reliability for GAD diagnoses were k .56, and for CSR ratings of GAD, k .77. Participants also completed the Penn State Worry Questionnaire (PSWQ Meyer, Miller, Metzger, & Borkovec, 1990), a 16-item measure of trait levels of excessive worry ( .795 in the current sample), and the Depression Anxiety Stress Scales���21-item version (Lovibond & Lovibond, 1995), a measure that yields separate scores of depression, anxiety, and stress. In the current study, the anxiety and stress subscales were used as indicators of anxiety, s .79 and .87, respectively. The Beck Depression Inventory (BDI Beck, Rush, Shaw, & Emery, 1979), .87 in the current sample, and an abbreviated version of the Quality of Life Inventory (QOLI Frisch, Cornwell, Villanueva, & Retzlaff, 1992), a measure of life satisfaction with .836 in the current sample, were considered measures of secondary outcomes. Two mea- 1 The study was conducted in compliance with the institutional review boards of the University of Massachusetts Boston, Boston University, Suffolk University, and Veterans Affairs Boston Healthcare System. No adverse events were reported throughout the duration of the study. 2 Because prior trials have omitted individuals with comorbid major depressive disorder (MDD) and because the DSM hierarchical rules arti- ficially limit comorbidity (T. A. Brown, Campbell, Lehman, Grisham, & Mancill, 2001), we chose to include individuals who met criteria for a current, principal diagnosis of GAD when the rule-out regarding GAD occurring solely during the course of a mood disorder was suspended (i.e., a full 6 months of GAD symptoms without MDD was not required). These individuals did report that GAD symptoms caused them more severe distress and impairment than did MDD symptoms. 3 Nineteen clients received prior psychotherapy for anxiety, 3 for depres- sion, and 19 had taken prior psychotropic medications for anxiety or mood problems. Seven participants (4 in the treatment condition) reported receiving a previous trial of CBT at some time before enrolling in the current study (2 for anxiety, 1 for depression, 1 for panic disorder with agoraphobia, 1 for GAD and obsessive���compulsive disorder, 1 for obsessive���compulsive disorder, and 1 for unspecified reasons). One client maintained intermittent contact (with no focus on CBT or anxiety) with a long-term psychotherapist throughout treat- ment and follow-up (once every 2 or 3 months). 4 Training included instruction, observation of taped and live interviews, and administration of collaborative interviews. For certification, assessors had to match with senior assessors on (a) identification of principal diagnosis(es) and (b) CSR for principal diagnosis within 1 point (c) all additional diagnoses had to be considered clinically significant for three of five consecutive interviews and assessors must not have committed admin- istration errors. All assessors had to attend a weekly consensus meeting to reduce drift, and 10% of clients seen in the clinic received double inter- views in order to confirm and maintain reliability. 5 To establish internal consistency of measures within our sample, alphas were calculated from the pretreatment assessment administration. 6 Unfortunately, due to a clerical error, five domains were omitted from the measure (children, relatives [other than children or partners], home, neighbor- hood, and community), so the scores reflect responses to the 11 remaining domains (health, self-esteem, goals and values, finances, work, recreation, learning, creativity, social/community action, romantic relationship, and friends). In a separate sample of 381 individuals recruited on an urban uni- versity campus, the full version of the QOLI was given. Scores were calculated for the full and shortened version of the questionnaire, and these were corre- lated at .94, suggesting that scores from the version used in the current study can be seen as reliable estimates of full measure scores for this measure. 1084 BRIEF REPORTS