lable at ScienceDirect

Behaviour Research and Therapy 59 (2014) 40e51

Contents lists avai

Behaviour Research and Therapy

journal homepage: www.elsevier .com/locate/brat

Dialectical behavior therapy skills for transdiagnostic emotiondysregulation: A pilot randomized controlled trial

Andrada D. Neacsiu a, b, *, Jeremy W. Eberle b, Rachel Kramer b, Taylor Wiesmann b,Marsha M. Linehan b

a Cognitive-Behavioral Research and Therapy Program, Duke University Medical Center (3026), 2213 Elba Street, Room 123, Durham, NC 27710, USAb University of Washington, USA

a r t i c l e i n f o

Article history:Received 3 October 2013Received in revised form12 April 2014Accepted 8 May 2014Available online 27 May 2014

Keywords:Dialectical behavior therapyEmotion dysregulationTransdiagnosticAnxietyDepression

* Corresponding author. Cognitive-Behavioral ReseDuke University Medical Center (3026), 2213 Elba St27710, USA.

E-mail address: andrada.neacsiu@duke.edu (A.D. N

http://dx.doi.org/10.1016/j.brat.2014.05.0050005-7967/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

Difficulties with emotions are common across mood and anxiety disorders. Dialectical behavior therapyskills training (DBT-ST) reduces emotion dysregulation in borderline personality disorder (BPD). Pre-liminary evidence suggests that use of DBT skills mediates changes seen in BPD treatments. Therefore, weassessed DBT-ST as a stand-alone, transdiagnostic treatment for emotion dysregulation and DBT skills useas a mediator of outcome. Forty-four anxious and/or depressed, non-BPD adults with high emotiondysregulation were randomized to 16 weeks of either DBT-ST or an activities-based support group (ASG).Participants completed measures of emotion dysregulation, DBT skills use, and psychopathology every 2months through 2 months posttreatment. Longitudinal analyses indicated that DBT-ST was superior toASG in decreasing emotion dysregulation (d ¼ 1.86), increasing skills use (d ¼ 1.02), and decreasinganxiety (d ¼ 1.37) but not depression (d ¼ 0.73). Skills use mediated these differential changes. Partic-ipants found DBT-ST acceptable. Thirty-two percent of DBT-ST and 59% of ASG participants droppedtreatment. Fifty-nine percent of DBT-ST and 50% of ASG participants complied with the research protocolof avoiding ancillary psychotherapy and/or medication changes. In summary, DBT-ST is a promisingtreatment for emotion dysregulation for depressed and anxious transdiagnostic adults, although moreassessment of feasibility is needed.

© 2014 Elsevier Ltd. All rights reserved.

Successful psychosocial treatments exist for a large range ofmental disorders; nevertheless, it is becoming increasingly difficultto offer clients the most effective treatment in the shortest amountof time (Barlow, Allen, & Choate, 2004; Fava, Evins, Dorer, &Schoenfeld, 2003). One potential solution to this problem is toidentify transdiagnostic mental health problems and interventionsthat successfully reduce them.

One transdiagnostic problem in need of treatment is emotiondysregulation, defined as lacking the skills needed or using mal-adaptive strategies to regulate emotional responses (Kring &Sloan, 2010; Neacsiu, Bohus, & Linehan, 2013). Difficulties withemotion regulation impede treatment (e.g., Ciarrochi & Deane,2001; Vogel, Wade, & Hackler, 2008) and are relevant to the

arch and Therapy Program,reet, Room 123, Durham, NC

eacsiu).

majority of psychological disorders. Over 85% of diagnoses in theDiagnostic and Statistical Manual of Mental Disorders involve ex-cesses or deficits of emotions or a lack of coherence amongemotional components (Kring & Sloan, 2010). When assessed fordifficulties with emotion regulation, adults with binge-eatingdisorder (Whiteside et al., 2007), generalized anxiety disorder(GAD; Salters-Pedneault, Roemer, Tull, Rucker, & Mennin, 2006),substance use disorder (Fox, Axelrod, Paliwal, Sleeper, & Sinha,2007), or anorexia nervosa (Harrison, Sullivan, Tchanturia, &Treasure, 2009) report more difficulty regulating emotions thanhealthy controls.

Affective disorders in particular have been strongly connected tomaladaptive emotional responses (Aldao, Nolen-Hoeksema, &Schweizer, 2010; Taylor & Liberzon, 2007), and their developmentand maintenance have been theoretically and empirically linked tochronic emotion dysregulation (e.g., Cisler, Olatunji, Feldner, &Forsyth, 2010; Kring & Bachorowski, 1999). Individuals who meetcriteria for anxiety or depression report and demonstrate frequent

Table 1Dialectical behavior therapy skills training curriculum and targeted components ofemotion dysregulation.a

Week Moduleb Selected skills Target problems witha

1 Mindfulness Wise Mind, Observe2 Describe, Participate,

Nonjudgmental, OneMindful, Effective

All components

3 Emotion Regulation Understand, Identify &Label Emotions

Processing emotions

4 Check the Facts Cognitive Change5 Opposite Action Managing emotional

expressions/actions6 Problem Solving Managing situations

that cue emotions7 Accumulate Positives,

Build MasteryManaging vulnerabilityto emotions

8 Cope Ahead, PLEASE9 Mindfulness Mindfulness review All components10 Distress Tolerance TIP Body Chemistry,

Managing ExtremeEmotions

Managingbiological/experientialchanges

11 Distract, Self-Soothe,Improve Moment

Controlling attention

12 Radical Acceptance,Turning the Mind

Cognitive Change

13 Willingness, Half-Smile,Mindfulness ofThoughts

Managingbiological/experientialchanges, cognitive change

14 InterpersonalEffectiveness

DEAR MAN GIVE FAST Managing situationsthat cue emotions

15 Interpersonal Validation Cognitive change16 Behavioral principles in

relationshipsManaging situationsthat cue emotions

Note. Weeks in which participants could join the group are in boldface.a See Neacsiu, Bohus, and Linehan (2013) for a full description of the treatment

model of emotion dysregulation.b DBT skills training module in the original skills manual (Linehan, 1993b).

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e51 41

use of maladaptive emotion regulation strategies, such as experi-ential avoidance, suppression, rumination, and problematic goalsetting (Aldao et al., 2010; Campbell-Sills, Barlow, Brown, & Hof-mann, 2006; Kring & Sloan, 2010). Furthermore, participants witheither a mood or an anxiety disorder report limited emotionalclarity, fear of experiencing emotions (Campbell-Sills et al., 2006;Mennin, Heimberg, Turk, & Fresco, 2005), inappropriateemotional intensity (Etkin & Wager, 2007), inability to modulateemotions based on contextual demands (Koole, 2009; Rottenberg,Kasch, Gross, & Gotlib, 2002), and intense reactions to nonthreat-ening cues (Kross, Davidson, Weber, & Ochsner, 2009; Schmidt &Keough, 2010). Researchers have also argued that underlyingproblems with affect are common in depression and anxiety(Barlow et al., 2004).

Transdiagnostic treatments for problems with emotions areemerging (Ellard, Fairholme, Boisseau, Farchione, & Barlow, 2010).Nevertheless, more research is needed to characterize howbehavioral treatments change emotion dysregulation across dis-orders. Our aim was therefore to test an intervention designed toreduce transdiagnostic emotion dysregulation. Neacsiu et al. (2013)presented a transdiagnostic treatment model for emotion dysre-gulation. Themodel includes teaching skills that help the individualreduce vulnerability to emotions; manage situations that cueemotions; control attention toward or away from emotional stim-uli; interpret emotional cues; manage biological, experiential, andaction changes; and process emotions.

This treatment model was derived from dialectical behaviortherapy (DBT; Linehan, 1993a), an empirically supported treatmentfor suicide and for borderline personality disorder (BPD; for a re-view see Neacsiu & Linehan, 2014). DBT is based on a skills deficitmodel that views dysfunctional behavior as either a consequence ofdysregulated emotions or a maladaptive approach to emotionregulation (Linehan, 1993a, 1993b). Consequently, DBT includesmore than 60 concrete skills (translated from behavioral researchand other evidence-based treatments) that are grouped into fourmodules: (a) mindfulness skills, which emphasize observing,describing, and participating in the present moment effectively andwithout judgment; (b) emotion regulation skills, including strate-gies for changing emotions and the tendency to respondemotionally; (c) interpersonal effectiveness skills, ranging fromacting assertively to maintaining self-respect; and (d) distresstolerance skills, including strategies to control impulsive actionsand to radically accept difficult life events (Linehan, 1993b). Theseskills map onto the treatment model for emotion dysregulation(Table 1), offering a comprehensive intervention for the lack ofadaptive skills and use of maladaptive strategies that defineemotion dysregulation (Neacsiu et al., 2013).

Emerging evidence suggests that DBT skills training (DBT-ST)reduces problems with emotions and is feasible to implement witha variety of mental disorders. DBT-ST outperformed treatment asusual in decreasing depression in treatment-resistant individuals(Harley, Sprich, Safren, Jacobo, & Fava, 2008) and in decreasingdepression, anxiety, and anger in a BPD sample (Soler et al., 2009).When compared to an active control condition, DBT-ST equallyreduced emotion dysregulation and problems with anger, anxiety,and depression for participants diagnosed with eating disorders(Safer, Robinson, & Jo, 2010). Evidence also suggests that increaseduse of DBT skills mediates the relationship between time in treat-ment and changes in depression, anger control, and suicidalbehavior acrossmultiple treatments in BPD samples (Neacsiu, Rizvi,& Linehan, 2010). Thus, DBT skills use may be a mechanism ofchange for emotion dysregulation.

In the current outcome study, we pilot tested DBT-ST as atransdiagnostic intervention for emotion dysregulation using arandomized controlled trial (RCT) designed to control for common

therapy factors. We targeted non-BPD adults with high emotiondysregulation who met criteria for at least one anxiety or depres-sive disorder. We chose to target adults with clinical anxiety ordepression because emotion dysregulation had been most stronglyconnected with depression and anxiety (see Aldao et al., 2010;Cisler et al., 2010; Kring & Bachorowski, 1999) and because wewanted to ensure the sample had significant clinical distress.

We had two primary aims. First, we assessed the unique effectsof DBT-STon emotion dysregulation.We hypothesized that (a) DBT-ST would reduce emotion dysregulation significantly more than anactivities-based support group (ASG) and (b) DBT skills use wouldmediate the differential changes between conditions. Second, weexplored (a) the unique effects of DBT-STon depression and anxietyseverity, (b) whether DBT skills use mediated differential changes,and (c) whether confounding effects explained any significantfindings. As a supplementary aim, we also examined the feasibilityof DBT-ST for a transdiagnostic sample based on (a) retention rates,(b) treatment credibility and satisfaction, and (c) compliance withthe ancillary treatment protocol.

Method

Participants and design

Intent-to-treat (ITT) participants were 44 men and women fromametropolitan area in the Northwestern United States. Participantswere included if theywere older than 18 years of age, scored high inemotion dysregulation (above 96 on the Difficulties in EmotionRegulation Scale; DERS; Gratz& Roemer, 2004), andmet criteria forat least one current depressive or anxiety disorder on the

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e5142

Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I;First, Spitzer, Gibbon,&Williams,1995). Participants were excludedif they scored above 2.5 on the Borderline Symptom List-23 (BSL-23; Bohus et al., 2009) or met full criteria for BPD on the StructuredClinical Interview for DSM-IV Axis II Personality Disorders (SCID-II;First, Gibbon, Spitzer, Williams, & Benjamin, 1997). Participantswere also excluded if they (a) were at high risk for suicide (hadattempted suicide within the past year or had current suicidalideation with a specific plan), (b) were mandated to psychologicaltreatment, (c) were homeless, (d) could not attend group, (e) hadreceived more than five sessions of outpatient DBT, (f) could notcommunicate in English, (g) met criteria for a bipolar or psychoticdisorder, (h) had a verbal IQ of less than 70 on the Peabody PictureVocabulary Test-Revised (PPVT-R; Dunn, 1981), or (i) met criteriafor a life-threatening disorder (e.g., severe anorexia). To beincluded, participants had to verbally agree to remain on the same

Fig. 1. Participant flowchart. DBT-ST ¼ dialectical behavior therapy skills training; ASG ¼ actithe intervention but continued participating in assessments; lost to follow-up ¼ participantparticipate in any additional assessments); DERS ¼ Difficulties in Emotion Regulation Scale

dosage (if any) of psychotropic medication and to refrain fromparticipating in ancillary psychotherapy. The University of Wash-ington (UW) Institutional Review Board approved all researchprocedures.

ScreeningTo recruit participants, we distributed study flyers and bro-

chures in the community and contacted clinicians to offer the studyas a potential low-cost referral. We screened 406 participants byphone and further screened qualified participants in person. Fifty-seven participants met all eligibility criteria (Fig. 1). The DERS cutoffwas set at one standard deviation above the pooled grand mean forcontrol samples (GM ¼ 77.33, pSD ¼ 19.52) in experimental studiespublished before July 2010. These samples had 1365 total partici-pants who (a) did not meet criteria for any disorder (Fox et al.,2007; Harrison et al., 2009), (b) did not meet criteria for the

vities based support group; discontinued intervention ¼ participants who discontinueds who dropped out of the study completely (discontinued the intervention and did not; BSL ¼ Borderline Symptoms List

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e51 43

disorder of interest in the study (Salters-Pedneault et al., 2006;Whiteside et al., 2007), or (c) were from the general population(Cohn, Jakupcak, Seibert, Hildebrandt, & Zeichner, 2010; Gratz &Roemer, 2004).

Four assessors who were blind to treatment assignment andtrained to reliability with the UW Behavioral Research and TherapyClinics gold standard conducted the in-person interviews. Inter-rater reliability for 20% of SCID-I interviews (14/67) was moderateto outstanding (k range: .44e1.00; Landis & Koch, 1977). The onlydiagnosis with low reliability was specific phobia. All other di-agnoses had acceptable reliabilities (>.60). To improve reliability, adifferent assessor subsequently coded all remaining ITT SCID-I in-terviews. Discrepancies were resolved through discussion with thefirst author, leading to 100% agreement. Interrater reliability for20% of SCID-II-BPD interviews (19/91) was outstanding (k ¼ .88,SD ¼ 0.11).

Power and randomizationWe determined sample size based on power analyses aimed at

detecting longitudinal emotion dysregulation differences betweenconditions. Data from a recently completed RCT and a publishedRCT suggested that we could expect an effect size ranging from 0.89to 1.53 (Gratz & Gunderson, 2006; Linehan, 2013). We conductedpower analyses in Optimal Design (Raudenbush et al., 2011), asoftware package designed for hierarchical linear modeling (HLM).In HLM, a slope is created for each participant completing at leastone assessment time point. The differences between these slopes(nested within conditions) are subsequently analyzed. An HLMpower analysis helps determine how many individual slopes areneeded to detect the expected effect size. Our power analysisindicated that, to reach 80% power with an effect size of 0.89, 42participants would need to complete the DERS at least once. Wetherefore recruited 48 participants, allowing for a 10% loss of data(from participants who do not complete any assessments). Using aminimization randomization algorithm (Pocock & Simon, 1975) tomatch participants on gender, primary diagnosis (depressive/anx-iety disorder), and psychotropic medication use (present/absent),we assigned the 48 participants equally to either DBT-ST or ASG.Four participants withdrew before treatment started and did notcomplete any assessment, leaving 44 participants for ITT analyses(Fig. 1).

Assessment scheduleTreatment and data collection occurred between November

2010 and January 2012. Assessments were conducted (a) at pre-treatment, (b) after 2 months, (c) at the end of treatment, and (d) ata 2-month follow-up. Participants completed brief interviews andself-reports measuring skills use, emotion dysregulation, psycho-pathology, feasibility, and confounding factors at each assessmentperiod, except where noted below. Participants received up to $100in compensation for completing all the study assessments.

Outcome measures

DERS (Gratz & Roemer, 2004)The DERS, our primary outcome measure, is a 36-item self-

report of six facets of emotion dysregulation: (a) nonacceptanceof negative emotions, (b) difficulty in emotional situations withpursuing goal-directed behaviors and (c) controlling impulses, (d)lack of regulation strategies, and (e) problems with emotionalawareness and (f) emotional clarity. The DERS has adequate inter-nal consistency (Cronbach's a ¼ .93), test-retest reliability (r ¼ .88),and construct and predictive validity. In the present study, internalconsistency at screening was acceptable (Cronbach's a ¼ .82).

DBT Ways of Coping Checklist (DBT-WCCL; Neacsiu, Rizvi, Vitaliano,Lynch, & Linehan, 2010)

The DBT-WCCL is a 59-item self-report of the frequency ofadaptive and maladaptive skills used to manage difficult situationsover the past month. We used only the DBT Skills Subscale (DSS),which includes general descriptions of skillful behavior withoutusing DBT-specific language. In several BPD samples, the DSS hadexcellent internal consistency (Cronbach's as ¼ .92 � .96), goodtesteretest reliability (r ¼ .71), and evidence for criterion validity(Neacsiu, Rizvi, & Linehan, 2010; Neacsiu, Rizvi, Vitaliano, et al.,2010). In the present study, internal consistency at pretreatmentwas high (Cronbach's a ¼ .86).

Patient Health Questionnaire-9 (PHQ-9; Kroenke, Spitzer, &Williams, 2001)

Given that some established measures of depression (e.g., BeckDepression Inventory; BDI; Beck, Steer, & Brown, 1996; HamiltonDepression Rating Scale; HAM-D; Hamilton, 1960) are not free forclinicians and that recent findings have led to questions about theiraccuracy (e.g., Bagby, Ryder, Schuller, & Marshall, 2004), we usedthe PHQ-9, a newer measure that is easy to administer, psycho-metrically strong, and clinician friendly. The PHQ-9 is a 9-item self-report that is sensitive to clinical change and that closely resemblesDSM criteria. The PHQ-9 has shown good test-retest reliability(r ¼ .84) over 2 days, and a cutoff of 9 had high sensitivity andspecificity in identifying major depressive disorder (MDD; McMil-lan, Gilbody, & Richards, 2010).

A growing body of evidence supports that the PHQ-9 has val-idity and reliability comparable to those of the BDI or the HAM-D.The PHQ-9 is strongly correlated with both measures (e.g., rBDI-II ¼ .84; Dum, Pickren, Sobell, & Sobell, 2008; rHAM-D ¼ .79;Cameron et al., 2011), as responsive as the BDI to change (Titovet al., 2011), and recommended over the BDI-II (Titov et al., 2011).Both the PHQ-9 and the BDI-II tend to overestimate depressionseverity when compared with the HAM-D (Cameron et al., 2011).Internal consistency at pretreatment was good (Cronbach's a¼ .74).

Overall Anxiety Severity and Impairment Scale (OASIS; Norman,Hami-Cissell, Means-Christensen, & Stein, 2006)

Using a similar rationale as for depression, we identified a free,easy-to-interpret, psychometrically strong self-report for anxietyseverity. The OASIS is a 5-item measure of general anxiety over thepast week with strong testeretest reliability (r ¼ .82), strongconvergent and discriminant validity, good internal consistency(Cronbach's a¼ .80), and adequate sensitivity and specificity. In onestudy, a cutoff of 7 on the OASIS correctly classified 87% of patientswith an anxiety disorder, and scores did not significantly vary withthe anxiety disorder identified as most distressing (Campbell-Sillset al., 2009). Other research also supports this cutoff (e.g.,Norman et al., 2011). Internal consistency in the present study atpretreatment was acceptable (Cronbach's a ¼ .79).

Confound and feasibility measures

Brief Treatment History Interview (B-THI)The B-THI is a short version of the Treatment History Interview

(Linehan & Heard, 1987) that contains face-valid questions abouttreatments received in the past 2 months. Use of psychotropicmedication and receipt of ancillary psychotherapy (each yes/no)were tested as confounds.

Addiction Severity Index Self-Report Form (ASI-SR)The ASI-SR is a self-report version of the ASI (McLellan et al.,

1992). A subsection of the ASI-SR was used to assess illicit druguse over the previous month, which was tested as a confound.

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e5144

Credibility and Expectancy of Improvement Scales (CEIS)Treatment credibility and satisfaction were assessed with an

adaptation of Borkovec and Nau's (1972) scale. Participants re-ported on a 9-point scale how logical the treatment seemed (oneitem), how successfully it would reduce their anxiety and/ordepression (two items), and how confidently they would recom-mend it to a friend (two items). Expectancy of improvement(Borkovec & Mathews, 1988) in anxiety and/or depression wasassessed with two items on a 0e100% scale. Credibility and ex-pectancy after the first group session (Cronbach's as ¼ .91 and .94,respectively) and at 2 months were tested as confounds. Four-month ratings of confidence in recommending the therapy and ofattribution of improvement to the treatment were analyzed asfeasibility outcomes.

Treatments

Each participant attended a weekly 2-h group therapy sessionfor 16 weeks and paid for each session on a sliding scale of $0e65.The groups were open and ongoing, and participants could join atweeks 1, 2, and 9 of the curriculum (see Table 1). Each participanthad access to 16 sessions of group therapy and met individuallywith their group therapists once for a 30-min private orientationsession in which a crisis plan was established. Outside of this ses-sion, participants did not have any contact with an individualtherapist. Participants who missed 4 consecutive weeks wereconsidered treatment dropouts but could still participate in as-sessments. If they declined assessment participation, they wereconsidered study withdrawals (and were marked as “lost to follow-up” in Fig. 1). Each group had a leader (a trained master's-leveltherapist) and a coleader (an untrained bachelor's-level assistant).

DBT-STDBT-ST was designed to retain the essence of standard DBT,

adopting the same philosophical base, treatment strategies, andtreatment targets. The DBT-ST therapist focused on teaching par-ticipants DBT skills and utilized modeling, step-by-step in-structions, structured behavioral rehearsal exercises, feedback, andhomework assignments for this purpose.

DBT-ST included two modifications to standard DBT that thetreatment developer approved. First, the protocol was shortenedfrom 24 weeks to 16 (Table 1) to correspond in length with otherevidence-based treatments for this population (e.g., Dimidjianet al., 2006). To this end, skills from the interpersonal effective-ness module were taught together with distress tolerance in ashorter amount of time than recommended by the original manual(Linehan, 1993b). This decision was based on evidence that inter-personal skills may not be needed to reduce emotion dysregulation(Gratz & Gunderson, 2006). Second, because of financial con-straints, the group coleader was not a trained DBT therapist.

Each session started with a mindfulness practice and review ofhomework (~40 min) and ended with skills instruction andhomework assignment (~75 min). Sessions 1 and 9 included alonger instruction segment (~110 min). Participants tracked theiruse of skills each day with written diary cards, which werereviewed weekly. (Diary card data were not included in analyses.)

ASGASG was designed to control for common factors based on

principles of client-centered supportive therapy. ASG therapistsaimed to foster a sense of belongingness, maintain unconditionalpositive regard, and provide empathy, psychoeducation, and atreatment structure. Teaching DBT skills and using cognitive-behavioral strategies in ASG were prohibited.

Each session started with a review of participants' experiencesover the last week (~30 min) and a support-building activity(~30 min) and ended with an open group discussion about topicsparticipants chose each week (~50 min). The most commonly dis-cussed topics were anger, anxiety, communication, coping, goals,sleep, stress, and depression. For homework, participants wereasked to think about issues discussed and to track daily socialsupport on diary cards.

Therapists and supervisionDBT groups were led by the first author, a master's-level clini-

cian who had participated in several DBT trainings (including aweeklong intensive training) and had 3 years of experienceproviding DBT under the supervision of the treatment developer.An expert DBT supervisor watched sessions and provided super-visionweekly. In accordance with the DBT model, the therapist andcoleaders also attended weekly DBT team meetings.

Two clinicians who held Master of Social Work degrees and whohad 3 years of experience conducting group psychotherapy led theASG groups. Before the study started, both therapists studied theASG manual and were trained in the study protocols by an expertcommunity clinician who had prior experience conducting andsupervising ASG groups. ASG therapists also received training insuicide assessment and management from the first author. ASGtherapists and coleaders received weekly supervision from thecommunity expert.

AdherenceAn independent coder trained to reliability with the clinic gold

standard in the DBT Global Rating Scale (Linehan& Korslund, 2009)rated 10% of the DBT group sessions for adherence. Because noadherence scale for supportive therapy could be found, we devel-oped a coding system (available upon request) to assess adherenceto ASG. A coder naïve to DBTwas trained in the ASG adherence scaleand rated 20% of the ASG sessions. We decided to code more ses-sions for ASG adherence because the measure used had not beenvalidated.

Ancillary treatment changesParticipants were classified as compliant or noncompliant with

the ancillary treatment protocol at each assessment period. Par-ticipants were deemed noncompliant if they reported (a) adding,dropping, or changing the dosage of a psychotropic medication or(b) seeing a psychotherapist outside the study. Participants wereallowed to continue regardless of their compliance. Noncompliancewas tested as confound.

Statistical analysis

Longitudinal outcomesHLM (Bryk & Raudenbush, 1992) was used to assess differences

between conditions over time. Appropriate covariance structureswere analytically determined (Verbeke, 1997), and the analysesincluded a restricted estimated maximum likelihood model to ac-count for missing data (Schafer & Graham, 2002). Assuming thatthe main effects could be different during the active phase oftreatment vs. the follow-up phase of the study, we separated thetime variable into two legs. The first leg (time1) was coded ascontinuous throughout treatment and constant at follow-up,whereas the second leg (time2) was coded as constantthroughout treatment and continuous at follow-up. Time1 andtime2 were both used as fixed effects and together yielded twotime-by-condition interactions (one for treatment and one forfollow-up). Each analysis included four a priori contrastsdtwo foreach study phasedto assess the significance of each condition

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e51 45

slope. Effect sizes were computed using Feingold's (2009) formulaand interpreted using Cohen's (1988) specifications.

ConfoundsTo assess whether another variable better explained treatment

effects, we tested confounds as covariates. Each analysis added tothe model described above the potential confound and two time-by-confound interactions (one for time1 and one for time2) asfixed effects. If the covariate main effect was significant, it was keptin the outcome's HLM model as a moderator; if it was not signifi-cant, it was removed. We tested six confounds: (a) use of psycho-tropic medication, (b) receipt of ancillary psychotherapy, (c) illegaldrug use (each yes/no), (d) treatment credibility, (e) expectancy ofimprovement, and (f) age. In addition, we assessed whetherancillary treatment changes (i.e., changing medication dose, start-ing/stopping a medication, or participating in ancillary psycho-therapy) explained treatment effects, by conducting independentanalyses on only participants who complied with the studyrequirement to maintain ancillary treatments unchanged.

Clinically significant changeFollowing Jacobson and Truax's (1991) method, we used change

scores, instrument reliability, and clinical and nonclinical distri-butions to classify each participant as deteriorated/unchanged,improved, or recovered.

MediationWe tested mediation hypotheses using Krull and MacKinnon's

(2001) statistical procedures and the specifications described byKramer, Wilson, Fairburn, and Agras (2002). Tofighi andMacKinnon's (2011) distribution-of-the-product method wasemployed to compute the mediation effect (a path*b path) and theconfidence interval (CI) that determined its significance. Mediationequations used a single-level HLM model in which the overallcondition effect throughout the study was the independent vari-able. We computed the percentage of the total effect explained bythe mediator to better understand these results.

Results

Sample characteristics and baseline differences

ITT participants were primarily single, heterosexual, Caucasianwomen who met criteria for multiple Axis I disorders

Table 2Participant demographic and clinical characteristics by condition (n ¼ 22 in each).

Demographic DBT-ST ASG Test p

Age (years)a 32.27 (10.50) 38.82 (13.55) t(42) ¼ 1.79 .08Femaleb 68.2% 63.6% c2(1) ¼ 0.10 .75Racial background FE 1.00Caucasian 95.4% 90.9%

Hispanic ethnicity 4.5% 9.1% FE 1.00LGBTQ 18.2% 13.6% FE 1.00Single/divorcedc 81.8% 61.2% c2(1) ¼ 0.30 .30Education U ¼ 221.00 .55� College graduate 63.6% 72.7%

No. BPD symptomsa 2.00 (1.20) 2.41 (1.30) U ¼ 199.00 .30Lifetime disorderDepressive disorder 95.5% 81.8% FE .35Anxiety disorder 72.7% 72.7% c2(1) ¼ 0.00 1.00SUD 40.9% 59.1% c2(1) ¼ 0.23 .23

Note. DBT-ST ¼ dialectical behavior therapy skills training; ASG ¼ activities-based suppotherwise specified; SUD ¼ substance use disorder; FE ¼ Fisher's Exact test.

a M (SD) reported.b Between-condition analysis compares males vs. females.c Between-condition analysis compares married vs. not married.

(Mdiagnoses ¼ 2.68, SD ¼ 1.21 in DBT-ST; Mdiagnoses ¼ 2.59, SD ¼ 1.44in ASG). Current GAD, MDD, and dysthymiawere the most frequentdiagnoses. Randomization successfully matched participants ongender, psychotropic medication use, and primary diagnosis. Nosignificant demographic differences emerged (Table 2). At pre-treatment,18 participants (40.9% in each condition) reported takinga mean of 1.39 (SD ¼ 0.61) medications for anxiety, depression, ormood (Supplementary Table S1). Participants joined their assignedtherapy groups, on average, 13.66 days after pretreatment(SD ¼ 12.72), and no significant difference arose between condi-tions, t(42) ¼ �0.18, p ¼ .86.

The average adherence score for a random sample of 6 out of 64DBT-ST sessions was 4.0 (SD ¼ 0.18), indicating that the treatmentwas adherent to the DBT model. In addition, only 1 out of 9randomly selected ASG sessions (43 total) was nonadherent.

Preliminary analyses

A preliminary examination of outcomes showed that, for theentire ITT sample at pretreatment, depression severity moderatelycorrelated with skills use (r ¼ �.32, p < .05) and anxiety severity(r ¼ .31, p < .05). We found no other significant correlations be-tween outcomes (ps > .05). HLM main effects, slope estimates, andstandard errors for ITT and completer analyses are included inTable 3. (Supplementary Fig. 1 shows raw score averages for the ITTsample.) Confound analyses showed that use of psychotropicmedications significantly predicted change in emotion dysregula-tion on the DERS, F(1, 152.91) ¼ 5.87, p < .05; therefore, thisconfound was included as a covariate in DERS analyses. No otherconfound was significant (Supplementary Table S2).

For clinical significance analyses, we computed cutoffs based onJacobson and Truax's (1991) specifications using DERS data fromnonclinical (Fox et al., 2007; Harrison et al., 2009) and clinical (Foxet al., 2007; Harrison et al., 2009; Salters-Pedneault et al., 2006;Whiteside et al., 2007) samples and PHQ-9 and OASIS data fromnonclinical (Kroenke et al., 2001), depressed (McMillan, Gilbody, &Richards, 2010), and anxious (Norman et al., 2011) samples. Forreliable change indices (RCIs), we used the testeretest reliabilitiesreported in the original validation studies for all measures exceptthe DERS, for which we computed reliability using data from 30 ITTparticipants who completed the measure both on the phone and atpretreatment (r ¼ .47, p < .01). See Fig. 2 for results.

Current disorder DBT-ST ASG Test p

Depressive disorder 68.2% 68.2% Х 2(1) ¼ 0.00 1.00Major depression 59.1% 40.9%Dysthymic disorder 4.5% 45.5%NOS 4.5% 0.0%

Anxiety disorder 90.9% 86.4% FE 1.00Panic 9.1% 18.2%Agoraphobia 4.5% 9.1%Generalized anxiety 77.3% 54.5%Social phobia 36.4% 36.4%Specific phobia 13.6% 22.7%Obsessive-compulsive 18.2% 4.5%Posttraumatic stress 13.6% 4.5%NOS 4.5% 13.6%

SUD 13.6% 0.00% FE .23

ort group; LGBTQ ¼ Lesbian, Gay, Bisexual, Transgender, Questioning; NOS ¼ not

Table 3HLM fixed effects and slope estimates for the intent-to-treat (ITT) and the completer analyses by condition.

ITT (n ¼ 44) Completer (n ¼ 24)

Outcome Phase Effect df F p DBT-ST slope ASG slope df F p DBT-ST slope ASG slope

DERS TX Time 1, 124.91 64.49 *** �19.44 (2.40)*** �8.33 (2.48)** 1, 70.18 30.21 *** �19.99 (3.14)*** �8.28 (4.08)*Interaction 1, 123.22 10.42 ** 1, 69.99 30.21 *

FU Time 1, 143.03 0.15 .70 4.28 (3.54) �5.62 (3.73) 1, 81.73 0.53 .47 6.44 (4.60) �10.3 0 (5.97)Interaction 1, 143.18 4.15 * 1, 82.67 6.05 *

DBT-WCCLa TX Time 1, 115.04 16.18 *** 0.23 (0.05)*** 0.06 (0.05) 1, 67.52 6.85 * 1.63 (0.46)*** 0.34 (0.60)Interaction 1, 115.04 5.87 * 1, 67.52 2.92 .09

FU Time 1, 138.63 1.79 .18 �0.12 (0.07) �0.03 (0.08) 1, 82.78 0.18 .67 �0.58 (0.67) 1.05 (0.87)Interaction 1, 138.63 0.75 .39 1, 82.78 2.23 .14

PHQ-9b TX Time 1, 95.89 45.25 *** �3.99 (0.73)*** �2.74 (0.69)*** 1, 53.00 10.79 ** �3.54 (0.88)*** �1.21 (1.15)Interaction 1, 95.89 1.57 .21 1, 53.00 2.58 .11

FU Time 1, 93.45 2.02 .16 3.18 (1.40)* �0.39 (1.38) 1, 53.00 0.36 .55 2.46 (1.68) �0.79 (2.20)Interaction 1, 93.45 3.29 .07 1, 53.00 1.37 .25

OASISc,d TX Time 1, 92.41 39.29 *** �2.96 (0.43)*** �1.13 (0.49)* 1, 63.95 8.96 **Interaction 1, 92.41 7.83 ** 1, 63.95 2.73 .10 �0.44 (0.12)*** �0.13 (0.15)

FU Time 1, 89.16 0.91 .34 2.04 (0.82)* �0.82 (0.98) 1, 65.71 0.02 .89 0.37 (0.19) �0.33 (0.23)Interaction 1, 89.16 5.08 * 1, 65.71 5.49 *

Note. HLM ¼ hierarchical linear model; TX ¼ treatment phase; DBT-ST ¼ dialectical behavior therapy skills training; ASG ¼ activities-based support group; FU ¼ follow-upphase; DERS¼ difficulties in emotion regulation scale; DBT-WCCL¼ dialectical behavior therapy ways of coping checklist; PHQ-9¼ patient health questionnairee depressionmodule; OASIS ¼ overall anxiety severity and impairment scale. When necessary completer analyses used transformations using aexponential or csquare-root functions. Onlyparticipants who at pretreatment scored over the bPHQ-9 cutoff (nITT ¼ 36; ncompleter ¼ 19) or dOASIS cutoff (nITT ¼ 20; ncompleter ¼ 35) were included in the PHQ-9 and OASISanalyses.*p < .05.**p < .01.***p < .001.

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e5146

Change in emotion dysregulation as a function of skills use

ITT analyses revealed that during treatment participants in bothconditions reported significantly less emotion dysregulation overtime but that those in DBT-ST improved significantly more andfaster (d ¼ 1.86). We found no significant difference in slopes forparticipants who were vs. were not taking psychotropic

Fig. 2. Classification of participants for each outcome over time by condition based on clinbased on a aPHQ-9 cutoff (n ¼ 36) or a bOASIS cutoff (n ¼ 35). DBT-ST ¼ dialectical behavi

medications; therefore, we present slope results per conditioncombining participants regardless of medication use (Table 3). Atfollow-up a significant time-by-condition interaction favored ASG.Although neither the DBT-ST nor the ASG slope was significant,participants in DBT-ST trended toward losing some of their gains,while participants in ASG trended toward continuing to improve.Completer analyses (n ¼ 24) yielded similar findings, except only

ical significance analyses. Includes only participants who were depressed or anxiousor therapy skills training; ASG ¼ activities-based support group.

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e51 47

ASG completers taking medication significantly improved inemotion regulation during treatment:slope estimateASGno meds

¼ �6:56; SE ¼ 4.49, t(75.53) ¼ �1.46,p ¼ .15; slope estimateASGmeds

¼ �10:57; SE ¼ 4.87, t(78.11) ¼ �2.17,p ¼ .03. Clinical significance analyses (Fig. 2) showed a significantdifference in recovery from emotion dysregulation favoring DBT-STat each time point (U2 months ¼ 135.00, U4 months ¼ 99.50, Ufollow-

up ¼ 117.00, ps < .05).Longitudinal ITT analyses using the DBT-WCCL indicated that

during treatment only DBT-ST participants significantly increasedtheir skills useover time (d¼1.02).During follow-up,participantsdidnot change significantly, suggesting that DBT-ST participants main-tained their gains. By the end of treatment, the skills use reported byITT participants increased by 16.0% in DBT-ST and 3.5% in ASG.Completer analyses showed similar results, although the interactioneffect during treatment was only a trend (Table 3). Significant dif-ferences in classification (ps < .05) based on clinical significance an-alyses favored DBT-ST at 4 months (U ¼ 92.00) and at follow-up(U ¼ 108.00; Fig. 2). Skills use significantly mediated the relation-ship between condition and improvement in emotion regulation(Table 4).

We conducted secondary analyses for the ITT sample to exploredifferential changes between conditions in the different facets ofemotion dysregulation (Supplementary Figure S2). During treat-ment DBT-ST participants increased their goal-directed behaviorand use of regulation strategies in emotional situations significantlymore than ASG participants: Finteraction(1, 130.14) ¼ 6.03, p < .05;Finteraction(1, 138.16) ¼ 10.90, p < .01; respectively. Gains weremaintained at follow-up for both DERS subscales: Finteraction(1,137.72) ¼ 3.87, p ¼ .05; Finteraction(1, 132.34) ¼ 3.09, p ¼ .08;respectively. For all other subscales, between-condition differenceswere not significant (ps > .05).

Changes in psychopathology

Depression severityThirty-six ITT participants who were clinically depressed at

pretreatment based on a score above the clinical cutoff (9) on thePHQ-9 were analyzed for differential changes over time in depres-sion severity. During treatment, participants in both conditionsimproved significantly and similarly (d¼ 0.73). During follow-up, nomaineffect of timewaspresent, but a trend for an interaction favoredASG. After treatment, DBT-ST participants slightly worsened,reporting a significant increase in depression severity from the endof treatment to follow-up, while ASG participants did not report asignificant change. Even with losses in depression gains, DBT-STparticipants still reported significant decreases during the study intheirdepressionseverity. Theeffect size forchange frompre- topost-treatment was 2.34 in DBT and 1.32 in ASG; from pretreatment to

Table 4Mediation of differences between conditions by skills use.

Outcome c (SE) a (SE) b (SE)

DERS �9.70 (4.80)* 0.20 (0.10)* �30.70 (4.67)***PHQ-9 �3.12 (1.10)** 0.20 (0.10)* �6.69 (1.07)***OASIS 0.06 (0.99) 0.20 (0.10)* �3.67 (0.83)***

Note. All analyses used the skills use subscale of the Dialectical Behavior Therapy Waysvariable. c ¼ coefficient estimate for the fixed effect of condition as a predictor of outcompredictor of skills use; b ¼ coefficient estimate for the fixed effect of skills use as a predictfixed effect of condition as a predictor of outcome (with skills use added to themodel alsoCI ¼ confidence interval for the mediated effect calculated by the distribution-of-the-proscale; PHQ-9 ¼ patient health questionnaire e depression module; OASIS ¼ overall anxi*p < .05.**p < .01.***p < .001.

a Represents the percentage of the total effect (direct þ indirect) explained by the me

follow-up it was 1.59 in DBT and 1.47 in ASG. Completer analysesindicated that only DBT-ST led to significant improvement indepression severity during treatment, although the time-by-condition interaction was not significant. At follow-up no signifi-cant change for treatment completers was present, suggesting DBT-ST completers maintained their gains (Table 3). Clinical significanceanalyses revealed no significant difference in classification betweenconditions (Fig. 2). Use of skills significantly mediated the relation-ship between condition and change in depression severity (Table 4).

Anxiety severityThirty-five ITT participants who were clinically anxious at pre-

treatment based on a score above the clinical cutoff (7) on the OASISwere analyzed for differential changes in anxiety severity. Duringtreatment, participants in both conditions reported a significantdecrease in their anxiety severity, but DBT-ST participantsimproved significantly faster (d¼ 1.37). Follow-up analyses showedno significant effect of time but revealed a significant interactionfavoring ASG. DBT-ST participants lost some of their gains from theend of treatment to follow-up. Even with the loss in some gains,DBT- ST participants still reported a significant decrease in theiranxiety severity from pretreatment to the end of the study. Theeffect size for changes from pre- to post-treatment was 1.34 in DBT-ST and 0.67 in ASG; from pretreatment to follow-up it was 1.98 inDBT-ST and 1.08 in ASG. Completer analyses revealed similar re-sults, with the interaction effect's approaching (although notreaching) significance during treatment (Table 3). Clinical signifi-cance analyses revealed no significant difference in classificationbetween conditions for anxiety (Fig. 2). Because the overall maineffect of condition was not significant for anxiety severity, we fol-lowed Kramer et al.'s (2002) guidelines and added the interactionbetween skills use and condition to themediationmodel. We foundsignificant condition and interaction effects, F(1, 147.65) ¼ 5.03, F(1,153.39) ¼ 4.43, respectively, ps < .05. Thus, skills use mediatedanxiety severity differentially by condition.

Feasibility of treatment and research protocol

More participants dropped treatment in ASG (n ¼ 13) than DBT-ST (n ¼ 7), a nonsignificant difference, c2(1, 44) ¼ 3.3, p ¼ .07.Clients attended, on average, two thirds of the sessions in DBT-ST(M ¼ 10.27, SD ¼ 4.72) and half in ASG (M ¼ 7.73, SD ¼ 5.21;U ¼ 177.50, p ¼ .13). At the end of treatment, participants in DBT-STattributed significantly greater improvement in depression and/oranxiety (M ¼ 53.33%, SD ¼ 21.14) to their treatment than partici-pants in ASG (M ¼ 26.88%, SD ¼ 25.81), t(32) ¼ 3.28, p < .01. Inaddition, DBT-ST participants reported significantly higher confi-dence in recommending their therapy to a friend (M ¼ 6.58,

c0 (SE) ME (SE) 95% CI Explainsa

�3.69 (5.33) �6.10 (3.16) [�12.68, �0.23] 62.31%�1.80 (1.01) �1.33 (0.69) [�2.78, �0.05] 42.50%0.80 (1.03) �0.73 (0.40) [�1.60, �0.03] 47.63%

of Coping Checklist (DBT-WCCL) as the mediator and condition as the independente (i.e., the direct effect); a ¼ coefficient estimate for the fixed effect of condition as aor of outcome (with condition added to the model); c0 ¼ coefficient estimate for thecalled the direct effect); ME¼mediated effect (i.e., the estimated indirect effect); 95%duct method (using PRODCLIN software); DERS ¼ difficulties in emotion regulationety severity and impairment scale.

diated effect.

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e5148

SD ¼ 2.35) than participants in ASG (M ¼ 4.06, SD ¼ 2.86),t(32) ¼ 2.82, p < .01.

During treatment and follow-up, 13 clients (5 in DBT-ST, 8 inASG) changed their medication, and 12 clients (7 in DBT-ST, 5 inASG) participated in ancillary psychotherapy. In total, 20 clients (9in DBT-ST,11 in ASG) did not comply with the research requirementof not engaging in ancillary psychotherapy or making changes topsychotropic medication regimens throughout the study. Of these,6 in DBT-ST and 5 in ASG were treatment completers. We con-ducted independent HLM analyses including only compliant ITTparticipants (13 in DBT-ST,11 in ASG) and found similar results withtwo exceptions: (a) compliant ASG participants did not reportsignificant reductions over time in difficulties with emotion regu-lation, depression severity, or anxiety severity (ps > .05) and (b) nosignificant interaction effect at follow-up for emotion dysregulationemerged (Supplementary Table S3).

Discussion

The present study is a preliminary examination of DBT skillstraining (DBT-ST) as a stand-alone treatment for emotion dysre-gulation in a transdiagnostic sample. Compared with a supportivetherapy control condition (ASG), DBT-ST was superior in reducingemotion dysregulation, increasing skills use, and reducing anxietyseverity in adults who met criteria for depressive and/or anxietydisorders. Although DBT-ST was comparable to ASG in reducingdepression severity, recovery from depression was twice as high inDBT-ST (71.4%) as in ASG (31.3%) based on clinical significance an-alyses. Differential improvements between conditions were notbetter explained by use of psychotropic medication or illicit drugs,participation in ancillary psychotherapy, expectancy of improve-ment, treatment credibility, or age. Moreover, use of DBT skillsmediated differential changes between conditions in emotiondysregulation and psychopathology. Participants also found theinterventions acceptable, although dropout from treatment washigh (32% in DBT-ST, 59% in ASG) and 45% of participants did notcomply with the research protocol to keep ancillary treatmentsconstant. Although neither noncompliance nor dropout impactedour results, the feasibility of DBT-ST as a stand-alone treatmentneeds further investigation. Taken together, these findings suggestthat (a) DBT-ST may be a promising transdiagnostic interventionthat contains an active mechanism of change for reducing emotiondysregulation, (b) targeting emotion dysregulation trans-diagnostically has a positive effect on anxiety and depressionseverity, and (c) future studies should more thoroughly examineand improve the feasibility of DBT-ST as a stand-alone intervention.

As hypothesized for our primary aim, DBT-ST participantsshowed a faster reduction in difficulties with emotion regulationthan those in ASG, although both treatments were efficacious. DBT-ST was more successful than ASG in increasing access to regulationstrategies and engagement in goal-directed behaviors duringemotional events, outcomes that directly map onto the emotionregulation and distress tolerance skills modules taught in DBT. Thelack of a significant finding for differential changes in emotionalawareness, clarity, or acceptance suggests that additional emphasison mindfulness and psychoeducation about emotions may beneeded. Nevertheless, DBT-ST helped 3 times as many participants(57.9%) score in a nonclinical range on emotion dysregulation bythe end of treatment. Moreover, only DBT-ST participants reportedusing significantly more skills over time in treatment and main-tained their gains at follow-up. By the end of the study, 47.6% ofDBT-ST participants reliably improved their skills usedonly 5.6% ofASG participants did the same. In addition to supporting the in-ternal validity of the study, this finding highlights that the directteaching and practice of skills is more effective than general

discussion and support at increasing skills use. Furthermore, skillsuse explained 62% of the variance in mediating the differentialimprovement in emotion dysregulation between conditions, con-firming our second hypothesis and suggesting that increased skillsuse may in turn reduce emotion dysregulation.

For our second primary aim, we explored the effects of targetingemotion dysregulation on the severity of psychopathology andfound promising results. The DBT-ST preepost effect sizes for im-provements in anxiety and depression severity are comparable tothose reported in trials for depression (d ¼ 1.34 � 2.92; Dimidjianet al., 2006; Westbrook & Kirk, 2005) and various anxiety disor-ders (d ¼ 0.92 � 2.06; McEvoy & Nathan, 2007; Norton & Price,2007; Stewart & Chambless, 2009). Teaching skills had a greatereffect than offering supportive therapy in reducing anxiety severitybut had a comparable effect in reducing depression severity. Thesefindings suggest that not directly addressing a primary disorder,but instead targeting a transdiagnostic problem, does not nega-tively impact psychopathology and may lead to significant im-provements. In addition, skills use may be a mechanism of changefor depression and anxiety severity. Future studies should morethoroughly investigate these hypotheses.

An unexpected finding was that at follow-up DBT-ST partici-pants lost some of their reductions in emotion dysregulation,anxiety, and depression. By contrast, ASG participants continued toimprove. We hypothesize that the motivational support or skillsguidance received during DBT-ST may explain this finding. In DBT-ST (unlike in ASG), participants were strongly encouraged to useskillful behavior through homework assignments, discussion, re-wards for homework completion, and punishment (behavioralanalyses) for homework noncompliance. In addition, use of specificskills was shaped and optimized via weekly feedback. Therefore,during the treatment participants in DBT-ST may have been moti-vated and coached in how to effectively use skills resulting ingreater relief from psychopathology and emotion dysregulation.Once treatment ended and the motivational support and skillsguidance was no longer available, participants may have reducedtheir efforts or may have become less effective in their use of skills,resulting in a loss in some of the relief they had gained. Futurestudies should test this hypothesis and provide solutions toimprove generalizability. Nevertheless, it is important to highlightthat participants did not lose all their gains at follow-up. Over thecourse of the entire study, DBT-ST participants improved signifi-cantly on all outcomes.

Our supplementary aimwas to assess the feasibility of offering agroup treatment to a transdiagnostic sample with high emotiondysregulation. We assessed feasibility by exploring treatmentretention rates, compliance with the research protocol, and satis-faction with the treatment. We considered our sample difficult totreat in light of evidence suggesting that difficulties with emotionregulation are connected with less willingness to engage in mentalhealth treatment (e.g., Ciarrochi & Deane, 2001; Vogel et al., 2008).In DBT-ST, 14% of participants dropped out because they wantedindividual therapy, 9% lost interest, and 9% could no longer attendgroup because of scheduling issues. The treatment retention rate inDBT-ST (68%) is lower than those of other treatments for depression(e.g., 77%; Kwan, Dimidjian, & Rizvi, 2010; Harley et al., 2008) oranxiety (e.g., 73%; van Ingen, Freiheit, & Vye, 2009), but consistentwith those of other treatments for transdiagnostic or difficult-to-treat samples, including a BPD group (66%; Soler et al., 2009), anoppositional defiant sample (60%; Nelson-Gray et al., 2006), a do-mestic violence group (67%; Iverson, Shenk, & Fruzzetti et al.,2009), and a group of anxious and depressed participants (59%;McEvoy & Nathan, 2007). Therefore, skills-only interventions forclients who are difficult to treat or transdiagnostic may have higherdropout rates than group therapies designed for specific diagnoses.

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e51 49

Although this finding suggests that DBT-ST is as feasible for atransdiagnostic sample as it is for other difficult-to-treat samples, italso highlights the need to reduce dropout from DBT-ST. Retentionmay improve by (a) allowing adjunctive individual therapy, (b)offering more groups to accommodate clients' schedules, or (c)permitting a flexible number of sessions to allow for faster/slowerimprovement.

Regarding compliance, 32% of all participants obtained addi-tional psychotherapy. The majority were noncompliant by eithernot ending previous psychotherapy relationships until afterbaseline or by seeking additional treatment during the follow-upperiod of the study. Analyses using only compliant participantsindicated that noncompliance with the protocol did not affect ourfindings. Nevertheless, the rate of noncompliance is concerning forthe feasibility of the intervention as a stand-alone treatment.Future research may improve compliance by allowing more timefor previous therapy to end before baseline or by providing moresessions for participants who still need psychotherapy after 16weeks.

Our results suggest that participants found DBT-ST acceptable.DBT-ST participants attributed their improvement in depressionand/or anxiety to the treatment and reported above-average con-fidence in recommending the treatment to a friend. Although thefeasibility of DBT-ST needs further evaluation, our treatmentdropout, compliance, and satisfaction findings suggest that DBT-SThas promise as a stand-alone, transdiagnostic treatment foremotion dysregulation.

Our findings also have important implications for assessment.Assessment burden is a well-known problem for clients and clini-cians in “real-world” mental health settings, and the structuredinterviews used in efficacy trials are rarely utilized in clinicalpractice (Garland, Kruse, & Aarons, 2003). We conducted struc-tured interviews to establish the diagnostic profile for participantswho scored over 96 on the DERS and completed an in-personscreening assessment. Out of 90 participants, only one did notmeet criteria for a current DSM-IV disorder on the SCID-I (but didmeet criteria for MDD in partial remission). The DERS took 7 min tocomplete, did not require reliability training, and yielded a group ofparticipants who responded to treatment. Thus, using the DERSwith a cutoff of 97 identifies a clinical sample for which an inter-vention with preliminary support exists.

It is important to highlight that our main goal was to assess DBT-ST as a treatment for transdiagnostic emotion dysregulation in aclinical sample and not to develop a new treatment for anxiety ordepression. Therefore, our study does not answer the importantquestion of whether DBT-ST improves upon existing evidence-based treatments for these disorders or whether such treatmentssufficiently change emotion dysregulation. Future studies shouldaddress such questions.

The present study has several limitations. First, therapist char-acteristics may have affected treatment effectiveness becausedifferent therapists conducted the interventions (Wampold &Serlin, 2000). Second, we had insufficient power to detect differ-ences in several analyses (e.g., depression, completer, mediation).Third, the mediation analyses were not designed to account for themediator-outcome temporal relationship (e.g., MacKinnon,Fairchild, & Fritz, 2007). Fourth, we had high dropout in ASG (59%)and a low number of sessions attended in both DBT-ST and ASG.Fifth, although both conditions contained equal numbers of par-ticipants with primary depressive disorders, ASG contained signif-icantly more dysthymic participants than DBT-ST.

In summary, DBT-ST is a promising treatment for trans-diagnostic emotion dysregulation, with a potentially active ingre-dient (skills use) and a positive effect on anxiety and depression inadults; nevertheless, DBT-ST can benefit from further assessment

and development to enhance retention and protocol complianceand to maintain durable gains.

Author statement

Marsha M. Linehan receives royalties from Guilford Press forbooks she has written on dialectical behavior therapy (DBT).Marsha M. Linehan and Andrada D. Neacsiu receive fees for DBTtrainings. Data was also presented as part of the first author'sdoctoral dissertation and in conference talks.

Acknowledgments

This research was supported by an American Psychological As-sociation Dissertation Research Award granted to the first author.The authors would like to thank the therapists and clients who tookpart in this study and to acknowledge Megan Smith, Laura Murphy,Sara Miller, Annika Benedetto, William Kuo, Dan Finnegan, KevinKing, Heidi Heard, Peter P. Vitaliano, and Nancy Whitney for theircontributions. We would like to thank Moria Smoski, M. ZacharyRosenthal, and Clive Robins for feedback on the manuscript.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.brat.2014.05.005.

References

Aldao, A., Nolen-Hoeksema, S., & Schweizer, S. (2010). Emotion-regulation strate-gies across psychopathology: a meta-analytic review. Clinical Psychology Review,30, 217e237. http://dx.doi.org/10.1016/j.cpr.2009.11.004.

Bagby, R. M., Ryder, A. G., Schuller, D. R., & Marshall, M. B. (2004). The HamiltonDepression Rating Scale: has the gold standard become a lead weight? Amer-ican Journal of Psychiatry, 161(12), 2163e2177. http://dx.doi.org/10.1176/appi.ajp.161.12.2163.

Barlow, D. H., Allen, L. B., & Choate, M. L. (2004). Toward a unified treatment foremotional disorders. Behavior Therapy, 35, 205e230. http://dx.doi.org/10.1016/S0005-7894(04)80036-4.

Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for the beck depression in-ventory. San Antonio, TX: The Psychological Corporation.

Bohus, M., Kleindienst, N., Limberger, M. F., Stieglitz, R., Domsalla, M.,Chapman, A. L., et al. (2009). The short version of the Borderline Symptom List(BSL-23): development and initial data on psychometric properties. Psychopa-thology, 42(1), 32e39. http://dx.doi.org/10.1159/000173701.

Borkovec, T. D., & Mathews, A. M. (1988). Treatment of nonphobic anxiety disorders:a comparison of nondirective, cognitive, and coping desensitization therapy.Journal of Consulting and Clinical Psychology, 56(6), 877e884. http://dx.doi.org/10.1037/0022-006X.56.6.877.

Borkovec, T. D., & Nau, S. D. (1972). Credibility of analogue therapy rationales.Journal of Behavior Therapy and Experimental Psychiatry, 3(4), 257e260. http://dx.doi.org/10.1016/0005-7916(72)90045-6.

Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications anddata analysis methods. Newbury Park, CA: Sage.

Cameron, I. M., Cardy, A., Crawford, J. R., du Toit, S. W., Hay, S., Lawton, K., et al.(2011). Measuring depression severity in general practice: discriminatory per-formance of the PHQ-9, HADS-D, and BDI-II. British Journal of General Practice,61(588), 419e426. http://dx.doi.org/10.3399/bjgp11X583209.

Campbell-Sills, L., Barlow, D. H., Brown, T. A., & Hofmann, S. G. (2006). Acceptabilityand suppression of negative emotion in anxiety and mood disorders. Emotion, 6,587e595. http://dx.doi.org/10.1037/1528-3542.6.4.587.

Campbell-Sills, L., Norman, S. B., Craske, M. G., Sullivan, G., Lang, A. J., Chavira, D. A.,et al. (2009). Validation of a brief measure of anxiety-related severity andimpairment: the Overall Anxiety Severity and Impairment Scale (OASIS). Journalof Affective Disorders, 112(1e3), 92e101. http://dx.doi.org/10.1016/j.jad.2008.03.014.

Ciarrochi, J. V., & Deane, F. P. (2001). Emotional competence and willingness to seekhelp from professional and nonprofessional sources. British Journal of Guidanceand Counseling, 29, 233e246. http://dx.doi.org/10.1080/03069880124843.

Cisler, J. M., Olatunji, B. O., Feldner, M. T., & Forsyth, J. P. (2010). Emotion regulationand the anxiety disorders: an integrative review. Journal of Psychopathology andBehavioral Assessment, 32, 68e82. http://dx.doi.org/10.1007/s10862-009-9161-1.

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

Cohn, A. M., Jakupcak, M., Seibert, L. A., Hildebrandt, T. B., & Zeichner, A. (2010). Therole of emotion dysregulation in the association between men's restrictive

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e5150

emotionality and use of physical aggression. Psychology of Men & Masculinity,11(1), 53. http://dx.doi.org/10.1037/a0018090.

Dimidjian, S., Hollon, S. D., Dobson, K. S., Schmaling, K. B., Kohlenberg, R. J.,Addis, M. E., et al. (2006). Randomized trial of behavioral activation, cognitivetherapy, and antidepressant medication in the acute treatment of adults withmajor depression. Journal of Consulting and Clinical Psychology, 74, 658e670.http://dx.doi.org/10.1037/0022-006X.74.4.658.

Dum, M., Pickren, J., Sobell, L. C., & Sobell, M. B. (2008). Comparing the BDI-II andthe PHQ-9 with outpatient substance abusers. Addictive Behaviors, 33(2),381e387. http://dx.doi.org/10.1016/j.addbeh.2007.09.017.

Dunn, L. M. (1981). Peabody picture vocabulary test-revised. Circle Pines, MN:American Guidance Service.

Ellard, K. K., Fairholme, C. P., Boisseau, C. L., Farchione, T. J., & Barlow, D. H. (2010).Unified protocol for the transdiagnostic treatment of emotional disorders:protocol development and initial outcome data. Cognitive and Behavioral Prac-tice, 17(1), 88e101. http://dx.doi.org/10.1016/j.cbpra.2009.06.002.

Etkin, A., & Wager, T. (2007). Functional neuroimaging of anxiety: a meta-analysis ofemotional processing in PTSD, social anxiety disorder, and specific phobia.American Journal of Psychiatry, 164(10), 1476e1488. http://dx.doi.org/10.1176/appi.ajp.2007.07030504.

Fava, M., Evins, A. E., Dorer, D. J., & Schoenfeld, D. A. (2003). The problem of theplacebo response in clinical trials for psychiatric disorders: culprits, possibleremedies, and a novel study design approach. Psychotherapy and Psychoso-matics, 72, 115e127. http://dx.doi.org/10.1159/000069738.

Feingold, A. (2009). Effect sizes for growth-modeling analysis for controlled clinicaltrials in the same metric as for classical analysis. Psychological Methods, 14(1),43e53. http://dx.doi.org/10.1037/a0014699.

First, M. B., Gibbon, M., Spitzer, R. L., Williams, J. B. W., & Benjamin, L. S. (1997).Structured clinical interview for DSM-IV Axis II personality disorders (SCID-II).Washington, DC: American Psychiatric Press.

First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. W. (1995). Structured clinicalinterview for DSM-IV Axis I disorders e patient Edition (SCID-I/P). New York, NY:Biometrics Research, New York State Psychiatric Institute.

Fox, H. C., Axelrod, S. R., Paliwal, P., Sleeper, J., & Sinha, R. (2007). Difficulties inemotion regulation and impulse control during cocaine abstinence. Drug andAlcohol Dependence, 89, 298e301. http://dx.doi.org/10.1016/j.drugalcdep.2006.12.026.

Garland, A. F., Kruse, M., & Aarons, G. A. (2003). Clinicians and outcome measure-ment: what's the use? The Journal of Behavioral Health Services and Research, 30,393e405. http://dx.doi.org/10.1007/BF02287427.

Gratz, K. L., & Gunderson, J. G. (2006). Preliminary data on an acceptance-basedemotion regulation group intervention for deliberate self-harm amongwomen with borderline personality disorder. Behavior Therapy, 37, 25e35.http://dx.doi.org/10.1016/j.beth.2005.03.002.

Gratz, K. L., & Roemer, L. (2004). Multidimensional assessment of emotion regula-tion and dysregulation: development, factor structure, and initial validation ofthe Difficulties in Emotion Regulation Scale. Journal of Psychopathology andBehavioral Assessment, 26, 41e54. http://dx.doi.org/10.1023/B:JOBA.0000007455.08539.94.

Hamilton, M. (1960). A rating scale for depression. Journal of Neurology Neurosur-gery and Psychiatry, 23, 56e62. http://dx.doi.org/10.1136/jnnp.23.1.56.

Harley, R., Sprich, S., Safren, S., Jacobo, M., & Fava, M. (2008). Adaptation of dia-lectical behavior therapy skills training group for treatment-resistant depres-sion. Journal of Nervous and Mental Disease, 196, 136e143. http://dx.doi.org/10.1097/NMD.0b013e318162aa3f.

Harrison, A., Sullivan, S., Tchanturia, K., & Treasure, J. (2009). Emotion recognitionand regulation in anorexia nervosa. Clinical Psychology and Psychotherapy, 16,348e356. http://dx.doi.org/10.1002/cpp.628.

van Ingen, D. J., Freiheit, S. R., & Vye, C. S. (2009). From the lab to the clinic:effectiveness of cognitive-behavioral treatments for anxiety disorders. Profes-sional Psychology: Research and Practice, 40(1), 69e74. http://dx.doi.org/10.1037/a0013318.

Iverson, K. M., Shenk, C., & Fruzzetti, A. E. (2009). Dialectical behavior therapy forwomen victims of domestic abuse: a pilot study. Professional Psychology:Research and Practice, 40, 242e248. http://dx.doi.org/10.1037/a0013476.

Jacobson, N. S., & Truax, P. (1991). Clinical significance: a statistical approach todefining meaningful change in psychotherapy research. Journal of Consultingand Clinical Psychology, 59, 12e19. http://dx.doi.org/10.1037/0022-006X.59.1.12.

Koole, S. (2009). The psychology of emotion regulation: an integrative review.Cognition & Emotion, 23, 4e41. http://dx.doi.org/10.1080/02699930802619031.

Kramer, H. C., Wilson, T., Fairburn, C. G., & Agras, W. S. (2002). Mediators andmoderators of treatment effects in randomized clinical trials. Archives of GeneralPsychiatry, 59, 877e883. http://dx.doi.org/10.1001/archpsyc.59.10.877.

Kring, A. M., & Bachorowski, J. A. (1999). Emotions and psychopathology. Cognitionand Emotion, 13, 575e599. http://dx.doi.org/10.1080/026999399379195.

Kring, A. M., & Sloan, D. M. (2010). Emotion regulation and psychopathology: Atransdiagnostic approach to etiology and treatment. New York, NY: Guilford Press.

Kroenke, K., Spitzer, R. L., & Williams, J. B. W. (2001). The PHQ-9: validity of a briefdepression severity measure. Journal of General Internal Medicine, 16(9),606e613. http://dx.doi.org/10.1046/j.1525-1497.2001.016009606.x.

Kross, E., Davidson, M., Weber, J., & Ochsner, K. (2009). Coping with emotions past:the neural bases of regulating affect associated with negative autobiographicalmemories. Biological Psychiatry, 65, 361e366. http://dx.doi.org/10.1016/j.biopsych.2008.10.019.

Krull, J. L., & MacKinnon, D. P. (2001). Multilevel modeling of individual and grouplevel mediated effects. Multivariate Behavioral Research, 36, 249e277. http://dx.doi.org/10.1207/S15327906MBR3602_06.

Kwan, B. M., Dimidjian, S., & Rizvi, S. L. (2010). Treatment preference, engagement,and clinical improvement in pharmacotherapy versus psychotherapy fordepression. Behaviour Research and Therapy, 48(8), 799e804. http://dx.doi.org/10.1016/j.brat.2010.04.003.

Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement forcategorical data. Biometrics, 33, 159e174. http://dx.doi.org/10.2307/2529310.

Linehan, M. M. (1993a). Cognitive-behavioral treatment of borderline personalitydisorder. New York, NY: Guilford Press.

Linehan, M. M. (1993b). Skills training manual for treating borderline personalitydisorder. New York, NY: Guilford Press.

Linehan, M. M. (2013). Effect size in a randomized controlled trial. Unpublished rawdata.

Linehan, M. M., & Heard, H. L. (1987). Treatment History Interview (THI). Unpublishedinstrument. Seattle, WA: Department of Psychology, University of Washington.

Linehan, M. M., & Korslund, K. E. (2009). Dialectical behavior therapy adherencecoding scale. Unpublished manuscript. Seattle, WA: Department of Psychology,University of Washington.

MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. AnnualReview of Psychology, 58, 593e614. http://dx.doi.org/10.1146/annurev.psych.58.110405.085542.

McEvoy, P. M., & Nathan, P. (2007). Effectiveness of cognitive behavior therapy fordiagnostically heterogeneous groups: a benchmarking study. Journal ofConsulting and Clinical Psychology, 75(2), 344e350. http://dx.doi.org/10.1037/0022-006X.75.2.344.

McLellan, A. T., Kushner, H., Metzger, D., Peters, R., Smith, I., Grissom, G., et al.(1992). The fifth edition of the Addiction Severity Index. Journal of SubstanceAbuse Treatment, 9, 199e213. http://dx.doi.org/10.1016/0740-5472(92)90062-S.

McMillan, D., Gilbody, S., & Richards, D. (2010). Defining successful treatmentoutcome in depression using the PHQ-9: a comparison of methods. Journal ofAffective Disorders, 127(1e3), 122e129. http://dx.doi.org/10.1016/j.jad.2010.04.030.

Mennin, D. S., Heimberg, R. G., Turk, C. L., & Fresco, D. M. (2005). Preliminary evi-dence for an emotion dysregulation model of generalized anxiety disorder.Behaviour Research and Therapy, 43, 1281e1310. http://dx.doi.org/10.1016/j.brat.2004.08.008.

Neacsiu, A. D., Bohus, M., & Linehan, M. M. (2013). Dialectical behavior therapyskills: an intervention for emotion dysregulation. In J. J. Gross (Ed.), Vol. 2.Handbook of emotion regulation (pp. 491e508). New York: Guilford Press.

Neacsiu, A. D., & Linehan, M. M. (2014). Dialectical behavior therapy for borderlinepersonality disorder. In D. H. Barlow (Ed.), Clinical handbook of psychologicaldisorders (5th ed.) (pp. 394e461). New York: Guilford Press.

Neacsiu, A. D., Rizvi, S. L., & Linehan, M. M. (2010). Dialectical behavior therapy skillsuse as a mediator and outcome of treatment for borderline personality disorder.Behaviour Research and Therapy, 48(9), 832e839. http://dx.doi.org/10.1016/j.brat.2010.05.017.

Neacsiu, A. D., Rizvi, S. L., Vitaliano, P. P., Lynch, T. R., & Linehan, M. M. (2010). TheDialectical Behavior Therapy Ways of Coping Checklist (DBT-WCCL): develop-ment and psychometric properties. Journal of Clinical Psychology, 66(6), 1e20.

Nelson-Gray, R. O., Keane, S. P., Hurst, R. M., Mitchell, J. T., Warburton, J. B., Chok, J. T.,et al. (2006). A modified DBT skills training program for oppositional defiantadolescents: Promising preliminary findings. Behaviour Research and Therapy,44(12), 1811e1820. http://dx.doi.org/10.1016/j.brat.2006.01.004.

Norman, S. B., Campbell-Sills, L., Hitchcock, C. A., Sullivan, S., Rochlin, A.,Wilkins, K. C., et al. (2011). Psychometrics of a brief measure of anxiety to detectseverity and impairment: the Overall Anxiety Severity and Impairment Scale(OASIS). Journal of Psychiatric Research, 45(2), 262e268. http://dx.doi.org/10.1016/j.jpsychires.2010.06.011.

Norman, S. B., Hami-Cissell, S., Means-Christensen, A. J., & Stein, M. B. (2006).Development and validation of an Overall Anxiety Severity and ImpairmentScale (OASIS). Depression and Anxiety, 23, 245e249. http://dx.doi.org/10.1002/da.20182.

Norton, P. J., & Price, E. C. (2007). A meta-analytic review of adult cognitive-behavioral treatment outcome across the anxiety disorders. The Journal ofNervous and Mental Disease, 195, 521e531. http://dx.doi.org/10.1097/01.nmd.0000253843.70149.9a.

Pocock, S. J., & Simon, R. (1975). Sequential treatment assignment with balancingfor prognostic factors in the controlled clinical trial. Biometrics, 31(1), 103e115.http://dx.doi.org/10.2307/2529712.

Raudenbush, S. W., Spybrook, J., Congdon, R., Liu, X., Martinez, A., Bloom, H., et al.(2011). Optimal design software for multi-level and longitudinal research(Version 3.01) [Software]. Retrieved from. http://www.wtgrantfoundation.org/.

Rottenberg, J., Kasch, K. L., Gross, J. J., & Gotlib, I. H. (2002). Sadness and amusementreactivity differentially predict concurrent and prospective functioning in majordepressive disorder. Emotion, 2, 135e146. http://dx.doi.org/10.1037/1528-3542.2.2.135.

Safer, D. L., Robinson, A., & Jo, B. (2010). Outcome from a randomized controlled trialof group therapy for binge eating disorder: comparing dialectical behaviortherapy adapted for binge eating to an active comparison group therapy.Behavior Therapy, 41(1), 106e120. http://dx.doi.org/10.1016/j.beth.2009.01.006.

Salters-Pedneault, K., Roemer, L., Tull, M. T., Rucker, L., & Mennin, D. S. (2006).Evidence of broad deficits in emotion regulation associated with chronic worry

A.D. Neacsiu et al. / Behaviour Research and Therapy 59 (2014) 40e51 51

and generalized anxiety disorder. Cognitive Therapy and Research, 30, 469e480.http://dx.doi.org/10.1007/s10608-006-9055-4.

Schafer, J. L., & Graham, J. W. (2002). Missing data: our view of the state of the art.Psychological Methods, 7(2), 147e177. http://dx.doi.org/10.1037/1082-989X.7.2.147.

Schmidt, N. B., & Keough, M. E. (2010). Treatment of panic. Annual Review of ClinicalPsychology, 6, 241e256. http://dx.doi.org/10.1146/annurev.clinpsy.121208.131317.

Soler, J., Pascual, J. C., Tiana, T., Cebri�a, A., Barrachina, J., Campins, M. J., et al. (2009).Dialectical behaviour therapy skills training compared to standard grouptherapy in borderline personality disorder: a 3-month randomized controlledclinical trial. Behaviour Research and Therapy, 47, 353e358. http://dx.doi.org/10.1016/j.brat.2009.01.013.

Stewart, R. E., & Chambless, D. L. (2009). Cognitive-behavioral therapy for adultanxiety disorders in clinical practice: a meta-analysis of effectiveness studies.Journal of Consulting and Clinical Psychology, 77, 595e606. http://dx.doi.org/10.1037/a0016032.

Taylor, S. F., & Liberzon, I. (2007). Neural correlates of emotion regulation in psy-chopathology. Trends in Cognitive Sciences, 11, 413e418. http://dx.doi.org/10.1016/j.tics.2007.08.006.

Titov, N., Dear, B. F., McMillan, D., Anderson, T., Zou, J., & Sunderland, M. (2011).Psychometric comparison of the PHQ-9 and BDI-II for measuring responseduring treatment of depression. Cognitive Behaviour Therapy, 40(2), 126e136.http://dx.doi.org/10.1080/16506073.2010.550059.

Tofighi, D., & MacKinnon, D. P. (2011). RMediation: an R package for mediationanalysis confidence intervals. Behavior Research Methods, 43, 692e700. http://dx.doi.org/10.3758/s13428-011-0076-x.

Verbeke, G. (1997). Linear mixed models for longitudinal data. In G. Verbeke, &G. Molenberghs (Eds.), Linear mixed models in practice (pp. 108e114). New York,NY: Springer. http://dx.doi.org/10.1007/978-1-4612-2294-1.

Vogel, D. L., Wade, N. G., & Hackler, A. H. (2008). Emotional expression and thedecision to seek therapy: the mediating roles of the anticipated benefits andrisks. Journal of Social and Clinical Psychology, 27(3), 254e278. http://dx.doi.org/10.1521/jscp.2008.27.3.254.

Wampold, B., & Serlin, R. C. (2000). The consequences of ignoring a nested factor onmeasures of effect size in analysis of variance. Psychological Methods, 5,425e433. http://dx.doi.org/10.1037/1082-989X.5.4.425.

Westbrook, D., & Kirk, J. (2005). The clinical effectiveness of cognitive behaviourtherapy: outcome for a large sample of adults treated in routine practice.Behaviour Research and Therapy, 43, 1243e1261. http://dx.doi.org/10.1007/978-1-4612-2294-1_3.

Whiteside, U., Chen, E., Neighbors, C., Hunter, D., Lo, T., & Larimer, M. (2007). Dif-ficulties regulating emotions: do binge eaters have fewer strategies to modulateand tolerate negative affect? Eating Behaviors, 8(2), 162e169. http://dx.doi.org/10.1016/j.eatbeh.2006.04.001.