You are on page 1of 10

Journal of Anxiety Disorders 50 (2017) 103–112

Contents lists available at ScienceDirect

Journal of Anxiety Disorders


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

Metacognitive therapy versus disorder-specific CBT for comorbid anxiety MARK


disorders: A randomized controlled trial☆

Sverre Urnes Johnsona,b, , Asle Hoffarta,b, Hans M. Nordahlc,d, Bruce E. Wampolda,e
a
Modum Bad Psychiatric Center, Vikersund, Norway
b
University of Oslo, Department of Psychology, Norway
c
Norwegian University of Science and Technology, Institute of Mental Health, Norway
d
St. Olavs Hospital, Div of Psychiatry, Nidaros DPS, 7006 Trondheim
e
University of Wisconsin-Madison, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Few studies have compared the effects of Metacognitive therapy (MCT) and Cognitive behavioral therapy (CBT)
Metacognitive therapy for comorbid anxiety disorders. In the current study we compared CBT and MCT for heterogeneous anxiety
CBT disorders in a residential setting. Ninety patients with a primary diagnosis of Post Traumatic Stress Disorder,
Transdiagnostic Social Phobia or Panic disorder, with and without Agoraphobia, were randomized to either CBT or MCT. Patients
Comorbidity
were assessed at pre-treatment, post-treatment and one-year follow-up. Primary outcome measures were Beck
Anxiety
Anxiety Inventory and ADIS IV and secondary outcome measures were SCID II, Beck Depression Inventory, Penn
State Worry Questionnaire, The Symptom Checklist-90 and the Inventory of Interpersonal Problems–64.
Treatment fidelity was satisfactory and therapist credibility was equal in both treatments. There was a significant
difference in the level of anxiety favouring MCT at post-treatment (d = 0.7), but there were no differences at
one-year follow-up, mainly due to a further improvement in the CBT group during the follow-up period. Both
treatments were efficacious. No differences in effect on comorbid diagnoses and symptoms were found, but MCT
produced larger change in personality problems. MCT seems to have a more rapid effect on anxiety symptoms,
but there were no significant differences in the long term for patients with comorbid anxiety disorders.

1. Introduction research have shifted towards processes that are common across psy-
chological disorders (Aldao & Nolen-Hoeksema, 2010; Harvey, Watkins,
Cognitive behavioral therapy (CBT) has grown into a collection of Mansell, & Shafran, 2004; Moses & Barlow, 2006; Wells & Matthews,
treatments for specific disorders and problems that have one key aspect 1994).
in common − changing maladaptive thoughts and behaviours (Beck, Several transdiagnostic processes have been discussed, each with a
1976). Meta-analyses have indicated that CBT has solid empirical evi- different treatment model, such as Metacognitive therapy (Wells, 2009)
dence and works well for a wide range of different disorders (Butler, and the Unified Protocol (Barlow et al., 2011). However, few studies
Chapman, Forman, & Beck, 2006; Tolin, 2010). CBT has developed by have evaluated the effect of transdiagnostic treatment models com-
designing specific variants of CBT for particular disorders (e.g., Clark, pared to a gold standard of disorder-specific treatments, such as CBT
1986). The use of a treatment for a single disorder may be problematic (Norton & Paulus, 2015 Reinholt & Krogh, 2014). A study by Norton
because comorbidity is common in psychiatric patients, especially for and Barrera (2012) compared a transdiagnostic cognitive-behavioral
anxiety disorders (Kessler et al., 2012). Further, comorbidity is associated group treatment to 12-week disorder-specific CBT protocols for panic
with higher level of disease burden (Gadermann, Alonso, Vilagut, disorder, social phobia, and generalized anxiety disorder, and failed to
Zaslavsky, & Kessler, 2012) and lowers the likelihood of recovery from find differences between transdiagnostic and the disorder-specific CBTs.
anxiety disorders (Bruce et al., 2008). To address comorbidity, trans- Moreover, no differences were found between disorder-specific and
diagnostic treatment models have been emerging and clinical theory and transdiagnostic CBT delivered in internet-format for GAD (Dear et al.,


This study was supported by “The National Program for Integrated Clinical Specialist and PhD-training for Psychologists” in Norway. The results of this paper were presented at the
3rd international conference of Metacognitive Therapy in Milano, 2016.

Corresponding author at: Research Institute, Modum Bad, NO-3370, Vikersund, Norway.
E-mail addresses: Sverre.Johnson@modum-bad.no (S.U. Johnson), Asle.Hoffart@modum-bad.no (A. Hoffart), Hmor-n@online.no (H.M. Nordahl),
wampold@education.wisc.edu (B.E. Wampold).

http://dx.doi.org/10.1016/j.janxdis.2017.06.004
Received 18 November 2016; Received in revised form 5 June 2017; Accepted 12 June 2017
Available online 15 June 2017
0887-6185/ © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
S.U. Johnson et al. Journal of Anxiety Disorders 50 (2017) 103–112

2015), panic disorder with and without agoraphobia (Fogliati et al., Anxiety Disorder at Modum Bad Psychiatric Center in Norway in 2013.
2016) and depression (Titov et al., 2015). There is some evidence that Modum Bad is a specialist hospital conducting, which includes an in-
transdiagnostic CBT may reduce comorbid symptoms more than dis- patient program for treatment resistant patients with anxiety disorders.
order-specific CBT (Norton et al., 2013). However, this pattern was not Before admission all patients received treatment in their local com-
found in a trial comparing acceptance and commitment therapy (ACT) munity and had not responded to previous treatments. The patients
with disorder-specific CBT for mixed anxiety disorders (Arch et al., were admitted to the hospital in groups of eight, but were treated in-
2012). Thus, it seems preliminarily that transdiagnostic CBT produces dividually. Treatment took place from the autumn of 2013 to the au-
similar results as disorder-specific CBT, and there is inconclusive evi- tumn of 2014 and one-year follow up was finished in the autumn of
dence regarding the differential effect on comorbid diagnoses and 2015. Recruitment was designed to be liberal using the clinical criteria
symptoms. for treatment used at the department. We used the Anxiety Disorders
An innovative transdiagnostic treatment model is Metacognitive Interview Schedule (IV) (ADIS IV; Brown, Di Nardo, & Barlow, 1994) for
therapy (MCT). MCT is based on the transdiagnostic S-REF model diagnostic evaluation. To be eligible for participation in the study,
(Wells & Matthews, 1994, 1996). According to this theory, psycholo- participants had to meet criteria for a principal DSM-IV disorder, ex-
gical disorders are linked to the activation of self-regulatory strategies ceeding above four on the clinical severity rating (CSR) on the ADIS IV
called the cognitive attentional syndrome (CAS). CAS consists of ex- of PTSD, SAD or PD/A. Further, participants had to have failed to
tended thinking in the form of worry and rumination, threat monitoring benefit from at least one structured psychological treatment, be 18
and unhelpful coping behaviours, and is maintained by positive and years or older, able to speak Norwegian, and provide informed consent.
negative metacognitions. Positive metacognitions are positive beliefs Exclusion criteria followed the intake-procedures at the department of
about using the CAS (e.g., “If I worry I will be prepared.”) while ne- Anxiety Disorders at Modum Bad, which excluded patients who (a) in a
gative metacognitions are related to the subjective feeling of lack of clinical context would have required immediate treatment or simulta-
control or dangers of the CAS-processes (e.g., “My worrying is un- neous treatment that could interact with the treatment in unknown
controllable.”). In the MCT-treatment the therapist focuses on enhan- ways, (b) had current DSM-IV diagnoses of organic mental disorders, (c)
cing flexible thinking by explicitly challenging metacognitions (Wells, had clear and current suicidal risk or (d) had evidence of current sub-
2009). MCT models have shown promising results for specific disorders stance abuse. All participants had to terminate the use of psychotropic
like generalized anxiety disorder (GAD; McEvoy et al., 2015; Van der medications before treatment. The patients were contacted before
Heiden, Muris, & van der Molen, 2012), major depressive disorder treatment to ensure that medications were terminated or was being
(MDD; Hagen et al., 2017, Wells et al., 2012), post traumatic stress reduced. The study was approved by the Norwegian regional ethical
disorder (PTSD; Wells, Walton, Lovell, & Proctor, 2015) and comorbid committee (2013/209/REK South-East).
and complex anxiety disorders (Johnson & Hoffart, 2016). A recent Participants who were included in the trial (N = 90) were rando-
meta-analysis indicated that MCT was superior to both waitlist and CBT mized to treatment stratified on their principal disorder. There were six
(Normann, van Emmerik, & Morina, 2014). Further, a small study participants who did not arrive at the hospital or complete the first
(N = 30) comparing individual MCT with individual CBT found that assessment. The participants did not have any knowledge of their
MCT had a significantly better effect on anxiety and worry at post- treatment-group before arrival. After diagnostic screening six patients
treatment than CBT (Nordahl, 2009). However, to our knowledge no had a loss of eligibility and at start of treatment four patients were
other studies have investigated the effect of MCT across several dis- excluded due to a new therapist arriving at the hospital who was not
orders in a randomized controlled trial. A comparison of MCT with formally trained. The remaining 74 participants who started treatment
disorder-specific treatments in samples with high degree of comorbidity (n = 38 CBT, n = 36 MCT) were included in the final analysis. Of the
is needed. participants starting treatment, seven did not complete the treatment
The current study was designed to compare the best documented and program, leaving 67 who completed all the treatment sessions (n = 33
recommended treatments of CBT for post traumatic stress disorder in CBT, n = 34 in MCT). See Table 1 for final sample characteristics and
(PTSD), social phobia (SAD) and panic disorder with and without agor- Fig. 1 for flowchart.
aphobia (PD/A) with a generic version of individual MCT in routine
clinical practice. The disorders were grouped together because there is
high degree of comorbidity between PTSD and the other anxiety dis- 2.2. Baseline scores
orders (Brady, Killeen, Brewerton & Lucerini, 2000). Furthermore, social
phobia and PTSD symptoms are often comorbid (Collimore, Carleton, The participants had on average 3.7 diagnoses at the start of
Hofmann, Asmundson, 2010; McMillan, Sareen, & Asmundson, 2014), treatment with 90.5% having a comorbid disorder and 30 (41%) a
and panic attacks often occur together with PTSD and SAD (Cougle, personality disorder. Only four participants (5%) had been working full
Feldner, Keough, Hawkins, & Fitch, 2010; Kessler et al., 2005). The me- time when entering treatment (see Tables 1 and 3). That few partici-
tacognitive model is also based on a theory (S-REF model) in which pants were working regularly, which in combination with long duration
common processes (i.e., worry, rumination and threat monitoring) be- of the anxiety disorder (M = 16.1 years, SD = 11.8) indicates a sample
tween the different disorders is highlighted, and thus it is possible to of chronic and dysfunctional patients.
conceptualize a transdiagnostic case-formulation for PD/A, SAD and
PTSD.
Based on previous studies, it was hypothesized that MCT would 2.3. Study design
have larger effects on the anxiety and primary diagnoses than would the
disorder-specific CBT. Further, due to the more general focus of trans- Participants in the study were randomized to two different treatment
diagnostic processes it was hypothesized that MCT would have a larger modalities, generic MCT and disorder-specific CBT. Patients in both MCT
effect on comorbid diagnoses and other symptoms compared to dis- and CBT were assessed at evaluation, pre-treatment, at the end of
order-specific CBT. treatment, and after a one-year follow-up period. At evaluation the
participants were assessed by clinicians in the department who con-
2. Method firmed the presence of an anxiety disorder before randomization, and a
new formal assessment including two diagnostic interviews (ADIS IV and
2.1. Participants SCID II) and self report questionnaires were conducted at the start of
treatment. Assessors were blind to treatment conditions.
Participants were referred to treatment at the Department of

104
S.U. Johnson et al. Journal of Anxiety Disorders 50 (2017) 103–112

Table 1
Sample and group characteristics at pre-treatment and 1-year-follow-up (1YFW)f.

Pre treatment 1YFW

Characteristic Total (74) CBT (38) MCT (36) Total (59) CBT (30) MCT (29)

a,b
Occupational status (last 6 months)
Disabled 18 (24.7) 8 (21.6) 10 (27.8) 15 (25.4) 8 (26.7) 7 (24.1)
Unemployed 30 (41.1) 17 (45.9) 13 (36.1) 14 (23.7) 8 (26.7) 6 (20.7)
Sick leave 11 (15.1) 7 (18.9) 4 (11.1) 3 (5.1) 1 (3.3) 2 (6.9)
Partly employed 7 (9.6) 2 (5.4) 5 (13.9) 10 (16.9) 5 (16.7) 5 (17.2)
Employed 4 (5.5) 2 (5.4) 2 (5.6) 6 (10.2) 4 (13.3) 2 (6.9)
Other reasons 3 (4.1) 1 (2.7) 2 (5.6) 11 (18.6) 4 (13.3) 7 (24.1)

Treatment c,d
Former treatment 74 (100) 38 (100) 36 (100) 40 (67.8) 22 (73.3) 18 (62.1)
Outpatient psychiatric treatment 63 (85.1) 32 (84.2) 31 (86.1) 36 (61.0) 19 (63.3) 17 (58.6)
Inpatient psychiatric treatment 30 (40.5) 18 (47.3) 12 (33.3) 3 (5.1) 2 (6.6) 1 (3.4)
e
Use of medication before treatment
Yes/no 62/10 32/4 30/6 23/36 12/18 11/18
Minor tranquilizers 20 8 12 8 2 6
Antidepressants 38 20 18 15 10 5
Major tranquilizers 2 2 0 0 0 0
Other 2 2 0 0 0 0

Note.
a
Other reasons = founded by spouse/living on old fundings.
b
Percentages given in parenthesis. 1 person missing in CBT.
c
Patients had both inpatient and outpatient treatment.
d
Timeframe of pre-treatment is since the patient got sick.
e
Before treatment indicates all former uses of medication. 2 persons did not report use of medication before treatment.
f
The sample analysed consisted of primary diagnosis: PD/A N = 28, SAD N = 22, PTSD N = 24.

2.4. Random assignment 2.5.2. Cognitive behavioral therapy


Treatments in the disorder-specific CBT condition were the most
Following diagnostic assessment at evaluation, participants were extensively documented cognitive treatments of panic disorder with
randomly assigned using a randomization sequence generated by www. and without agoraphobia (Clark, 1986; Wells, 1997), of social phobia
random.org. Patients were stratified individually on primary diagnosis (Clark & Wells, 1995; Wells, 1997), and of PTSD by using prolonged
prior to randomization (i.e., SAD, PD/A or PTSD) and randomly as- exposure (PE) therapy (Foa, Hembree, & Rothbaum, 2007). The proto-
signed to either MCT or CBT. The randomization was conducted four cols where chosen because of wide use and strong documentation over
weeks before the start of treatment because the patients needed time to the past two decades (see NICE, 2013). The treatments followed their
plan travel and prepare for the treatment. Pragmatic considerations at respective manuals. However, due to the length of the program the
the hospital dictated that one treatment group began 3 weeks before the number of sessions was adjusted to 7–9 in PE.
other (viz., the MCT group started their treatment 3 weeks before the
CBT group). To avoid possible confounding related to order-effects the
2.5.3. Comparison of MCT and CBT
intake of the patients was switched in the middle of the trial so that the
MCT and CBT are used by therapists to change various aspects of
MCT group started three weeks after the CBT group.
cognitions, and both treatments are goal directed, short term and
structured. However, CBT is focuses mainly on the content of cogni-
tions, whereas MCT focuses on the meta-level (cognitions about cog-
2.5. Treatments
nitions). In MCT the goal is to change how patients respond to thoughts
by changing metacognitions that drive the CAS. Furthermore, in MCT
The number of sessions for completers were M = 9.4 (SD = 1.7).
exposure to trauma is not a part of the treatment.
The number of sessions were equivalent in both conditions. However
the sessions in CBT lasted longer, due to the protocols for SAD and
PTSD, which lasted 90 min. Standard session time for MCT was 50 min. 2.6. Therapists and supervision
In the CBT group, actual session time in minutes was M = 71.5 (SD
18.8). In the MCT group the actual session time was 51.8 (SD = 13.3). Two clinical psychologists served as therapists in MCT and one
The patients received individual treatment in a ward for eight weeks clinical psychologist and two psychiatrists served as therapists in CBT.
with other anxiety patients. The patients participated in the ward’s The therapists worked at the department at the time of the trial and had
common activity, consisting of one ward meeting and one physical at least two years of formal clinical training in MCT or CBT or were in
exercise session per week. The patients in the study were requested not progress of completing their training in the respective treatments. One
to talk about the treatment outside the therapy-room and instead focus of the psychiatrists in CBT was included during the trial. She received
on other aspects of the inpatient setting. intensive supervision from the CBT expert and the first author, and had
two training-cases documenting competency and adherence in CBT
prior to inclusion. Each therapist in the dataset treated 14.8 patients on
2.5.1. Metacognitive therapy average (SD = 5.5, Range = 5 − 18). In the CBT group the average
The MCT treatment consisted of a manualized treatment protocol years of clinical experience was 10.0 (SD = 10.8), and in the MCT
for the generic MCT model (Wells, 2009). The protocol deemphasizes group the average years of experience was 2.5(SD = 0.7). An expert in
disorder-specific aspects, and focuses instead on challenging positive MCT (the third author) and an expert in CBT (the second author) su-
and negative metacognitions that drive the use of worry, rumination, pervised the therapist groups weekly and ensured the therapists’ fide-
threat-monitoring and coping behaviours to regulate emotions. lity.

105
S.U. Johnson et al. Journal of Anxiety Disorders 50 (2017) 103–112

Fig. 1. Consort diagram of participant flow. MCT =


Metacognitive Therapy, CBT = Cognitive Behavioral
Therapy.

2.7. Treatment fidelity clinical psychology students trained by the MCT and CBT experts and
the first author. The two MCT trained students rated 313 MCT videos
Therapist competency for CBT was assessed using the Cognitive (across 33 patients) and the two CBT trained students rated 281 CBT
Therapy Scale (CTS; Young & Beck, 1980) and for MCT the Meta Cog- (across 34 patients). Patients’ average competence and adherence rat-
nitive Therapy Competency Scale (MCT-CS; Nordahl & Wells, 2009). ings across sessions were calculated. Every 15th to 18th video, chosen
CTS has shown good psychometric properties (Vallis, Shaw & Dobson, at random, was coded independently by the first author for inter-rater
1986), while MCT-CS has not undergone psychometric investigation reliability, and for giving necessary supervision to avoid rater drift.
(see supplementary A), but was developed to directly mirror the items Inter-rater reliability using a one-way random intraclass correlation
in the CTS. Competence was evaluated on a scale from 0 to 6 in both (ICC [1,1]) was 0.88 for MCT competence (N = 17) and 0.76 for ad-
MCT and CBT, and trial target for competence ratings was four or herence. The ICC (1, 1) for the CBT competence ratings (N = 18) was
above. Adherence to the treatment protocols was calculated based on 0.93 and 0.83 for adherence.
six items: setting agenda, checking homework, following the structure
of the treatment protocol, using CBT/MCT interventions, giving
2.8. Measures
homework and asking for feedback at the end of session. All available
videos from the trial (N = 594) were rated by one of four advanced
Measures that were not specific to a particular anxiety disorder, but

106
S.U. Johnson et al. Journal of Anxiety Disorders 50 (2017) 103–112

assessed anxiety as a generalized condition, were selected. The Beck 2.8.5. The inventory of interpersonal problems-64-circumplex (IIP-64-C;
Anxiety Inventory (BAI) and Anxiety Disorder Interview Schedule for Horowitz, Alden, Wiggins, & Pincus, 2000; Horowitz, Rosenberg, Baer,
DSM-IV (ADIS-IV) were used as primary outcomes of the trial. Ureño, & Villaseñor, 1988)
Secondary outcomes1 included the Structured Clinical Interview for IIP-64-C is a self-report measure of maladaptive relationship beha-
Diagnosis of DSM-IV Axis II (SCID II), Symptom Check List-90 (SCL-90), vior. The scale consisted of 64 items of which 8 subscales are con-
IIP-64, BDI, Penn State Worry Questionnaire (PSWQ) and use of the ceptually organized in a circumplex manner and the patients are asked
ADIS IV CSR-scores. To cover those diagnoses not measured by ADIS IV, to rate interpersonal behavior on a 5-point Likert scale. The IIP 64C has
the patients were screened for eating disorder and body dysmorphic shown good psychometric properties (Horowitz et al., 1988). The
disorder using the Mini-International Neuropsychiatric Interview Cronbach’s alpha coefficient for the IIP-64-C for this study at start of
(M.I.N.I.; Sheehan et al., 1998), both at pre-treatment and one-year treatment, at the end of treatment and at one-year follow-up was 0.96,
follow up. 0.98 and 0.96, respectively.

2.8.6. Beck depression inventory II (BDI II; Beck, Steer & Brown, 1996)
2.8.1. Beck anxiety inventory (BAI; Beck, Epstein, Brown & Steer, 1988)
BDI-II is a 21-item self-report scale assessing current level of de-
BAI is an instrument with 21 items, measuring anxiety symptoms
pression. The range of scores is from 0 to 63, and the items are scored
during the prior week. The items are rated on a Likert scale from 0 to 3,
on a Likert scale from 0 to 4. The psychometric properties of BDI are
and the total score ranges from 0 to 63. BAI has been found reliable and
adequate (Beck, Steer, & Garbin, 1988). The Cronbach’s alpha coeffi-
valid for measuring anxiety symptoms (Steer, Ranieri, Beck, & Clark,
cient for the BDI II for this study at start of treatment, at the end of
1993). We used the BAI as the primary self-report outcome because it
treatment and at one-year follow-up was 87, 0.93 and 0.93, respec-
was necessary to have a measure of anxiety that was appropriate across
tively.
the three anxiety disorders. The Cronbach’s alpha coefficients for BAI in
this study at start of treatment, at the end of treatment and at one-year
2.8.7. Penn state worry questionnaire (PSWQ; Meyer, Miller,
follow-up were 0.89, 0.92 and 0.91, respectively.
Metzger, & Borkovec, 1990)
PSWQ is a widely used 16- item self-report inventory assessing the
2.8.2. ADIS −IV (Brown, DiNardo & Barlow, 1994) pervasiveness, excessiveness, and uncontrollability of worry. Total
ADIS-IV is a semi structured diagnostic interview designed to assess scores on the PSWQ range from 16 to 80 and the items are scored on a
the presence, nature and severity of DSM-IV anxiety and mood dis- Likert scale from 1 to 5. The PSWQ is a reliable and valid instrument for
orders. Clinical Severity Ratings (CSR) are conducted for each disorder measuring worry (Meyer et al., 1990). The Cronbach’s alpha coefficient
on a scale from 0 to 8. ADIS-IV interviewers were advanced clinical for the PSWQ for this study at start of treatment, at the end of treatment
psychology students, trained to reliability standards, who had demon- and at one-year follow-up was 0.92, 0.94 and 0.92, respectively.
strated adequate diagnostic reliability on three consecutive interviews.
All interviewers were blind to treatment condition. Reliability assessors 2.8.8. Generalization of treatment effects
who were blind to the original diagnosis coded videos of 15 percent of Co-occurring mood and anxiety disorders (with CSR of 4 or more) at
randomly chosen diagnostic interviews (N = 20). Inter-rater reliability post-treatment and one-year follow-up were analysed as an index of
on the principal diagnosis, PD/A, SAD and PTSD, was 100%. Inter-rater generalization of treatment effects. We used number of diagnoses and
agreement on dimensional CSR ratings for the primary diagnosis, using the total score of CSR for each patient as the measure of generalization
a one-way random intraclass correlation (ICC [1,1]) was 0.96 with a of treatment effects. The procedure was similar to the one used in a
single-measure,. The ICC (1, 1) across all the anxiety and mood dis- recently published RCT (Arch et al., 2012). Further, we used total
orders (met DSM IV criteria vs subclinical vs. none) was 0.85 and the number of 3 criteria (SCID II) as a marker for the effect on personality
ICC (1, 1) for the sum score of dimensional CSR rating was 0.89. problems.

2.9. Common factors: treatment credibility and working alliance


2.8.3. SCID-II (First, Gibbon, Spitzer, Williams, & Benjamin, 1997)
SCID-II is a semi structured diagnostic interview designed to assess
We wanted to make sure that the two treatments were comparable
the presence, nature, and severity of DSM-IV Axis II disorders.
on common factors, such as treatment credibility and working alliance,
Interviews were conducted by advanced clinical psychology students
which are essential for psychotherapy outcome (Wampold & Imel,
early in treatment, and at one-year follow-up. To assess interrater re-
2015). A shortened three-item version of the five-item Credibility Scale
liability, blind raters coded 25 of the videos. Cohen’s kappa was 1.0 for
(CS-3) was used (Borkovec & Nau, 1972) to assess on 0–10 point scales
the personality disorder, reflecting 100% agreement about the presence
how logical participants find the treatment, how successful participants
of personality disorder or not. The ICC (1, 1) for the total score of all
think the treatment would be to reduce symptoms, and how much the
criteria (scoring 3 or not) was 0.97.
participants would recommend the treatment to a friend. The measure
has been used in several trials to assess the credibility of competing
2.8.4. The symptom checklist-90 (SCL-90; Derogatis, 1983) treatments and psychological placebos, and has sound psychometric
SCL-90 is one of the most widely used measures of multiple aspects properties (Devilly & Borkovec, 2000). Cronbach’s alpha coefficient for
of psychological distress in clinical practice and research. The measure the CS-3 at session three was 0.80. Working Alliance Inventory − Short
consists of 90 items, and gives a GSI score of general distress. The items Revised (WAI-SR) (Hatcher & Gillaspy, 2006) is a shortened 12-item
are rated on a Likert scale from 0 to 5, and the total GSI score varies version of the original 36-item WAI (Horvath & Greenberg, 1989),
from 0 to 5. SCL 90 has shown good psychometric properties (Schmitz which measures the quality of the therapeutic alliance. Items are rated
et al., 2000). The Cronbach’s alpha coefficient for SCL-90 for this study on a 1–7 Likert scale, and the total score is the average of the twelve
at start of treatment, at the end of treatment, and at one-year follow-up items. The WAI-SR has demonstrated good psychometric properties
was 0.96, 0.98 and 0.98, respectively. (Hatcher & Gillaspy, 2006). The Cronbach’s alpha coefficient for the
WAI-SR at session three was 0.92.
1
This Trial was registered with 16 measures (ClinicalTrials.gov identifier:
NCT01889342). We chose a selection of the secondary outcomes of the trial in advance of
2.10. Clinical significance
conducting any analysis, because the other measures were process-measures or disorder-
specific measures that will be reported elsewhere. The patient improvement was evaluated using the clinical

107
S.U. Johnson et al. Journal of Anxiety Disorders 50 (2017) 103–112

Table 2
Mean, Standard deviation and effect sizes from pre- to post-treatment and 1-year-follow-up (IYFW).

Measure and condition Pre M (SD) Post M (SD) IYFW M (SD) d pre-posta d pre-IYFW

BAI
CBT (N = 38) 24.8 (10.0) 20.8 (12.1) 15.8 (10.9) 0.36 CI [-0.10, 0.81] 0.86 CI [0.38, 1.32]
MCT (N = 36) 26.7 (11.7) 15.7 (10.2) 17.4 (10.4) 1.00 CI [0.5, 1.48] 0.84 CI [0.35, 1.31]

BDI
CBT (N = 38) 20.8 (9.3) 16.9 (10.9) 16.6 (10.9) 0.38 CI [-0.07, 0.83] 0.41 CI [-0.04, 0.86]
MCT (N = 36) 22.0 (9.2) 12.1 (9.6) 15.8 (10.9) 1.05 CI [0.55, 1.53] 0.61 CI [0.14, 1.08]

IIP 64c
CBT (N = 38) 1.43 (0.60) 1.24 (0.67) 1.20 (0.62) 0.30 CI [-0.16, 0.75] 0.38 CI [-0.07, 0.83]
MCT (N = 36) 1.32 (0.58) 1.06 (0.71) 1.15 (0.63) 0.40 CI [-0.07, 0.86] 0.28 CI [-0.19, 0.74]

SCL 90
CBT (N = 38) 1.52 (0.51) 1.15 (0.67) 1.10 (0.70) 0.62 CI [0.15, 1.08] 0.68 CI [0.22, 1.14]
MCT (N = 36) 1.53 (0.58) 0.88 (0.60) 1.05 (0.66) 1.10 CI [0.59, 1.58] 0.77 CI [0.29, 1.24]

PSWQ
CBT (N = 37) 58.03 (14.8) 54.73 (15.8) 52.39 (13.2) 0.22 CI [-0.24, 0.66] 0.40 CI [-0.06, 0.86]
MCT (N = 36) 61.22 (12.1) 50.24 (14.3) 52.12 (13.4) 0.83 CI [0.34, 1.30] 0.71 CI [0.23, 1.18]

Note. BAI = Beck Anxiety Inventory, BDI = Beck Depression Inventory, IIP 64 = Inventory of Interpersonal Problems 64, SCL–90 = Symptom Checklist 90, PSWQ = Penn State Worry
Questionnaire.
a
Cohen's d = M1 − M2/SDpooled.

significance criteria on the BAI (Jacobson & Truax, 1991). The cut off 2.12. Power estimation
score of 15 was used to demark the clinical range, which based on the
norms from Gillis, Haaga and Ford (1995). The norms from Beck et al. Power analysis, conducted with Optimal Design (Raudenbush & Liu,
(1988) were used to calculate the standard error of the difference be- 2000), indicated that to reach 80% power with an effect size of d = 0.6,
tween the two scores and the reliable change index (RCI). Using the cut required 78 participants. Therefore our total sample size (N = 90) was
off point and the RCI, each patient could be classified as recovered expected to be sufficient to detect between group differences of mod-
(passed both criteria), improved (passed only the RCI criterion in the erate size at each assessment point as found in previous research by
positive direction), unchanged (did not pass the RCI criterion), or de- Nordahl (2009) and Normann et al. (2014).
teriorated (passed the RCI criterion in the negative direction).
3. Results

2.11. Statistical methods 3.1. Characteristics of the patients

The data from all patients who started treatment were analysed. Chi Table 1 gives an overview of the characteristics of the patients at
square tests and t-tests (two-tailed) were conducted to check for dif- pre-treatment and one-year follow-up. The patients’ did not differ on
ferences at pre-treatment. Treatment differences were analysed using key variables at pre-treatment, including number of diagnoses, former
Hierarchical Linear Modelling (HLM; Raudenbush & Bryk, 2002). In treatment, use of medications, age and gender (p > 0.05 for all vari-
HLM all available data is used. Thus, a research participant with only ables). Further, no significant pre-treatment differences between con-
baseline data can be included in an analysis and contribute to the es- ditions emerged on the outcome variables (p > 0.05 for all variables;
timation of model parameters (Kwok et al., 2008). The models were see supplementary material B). The sample analysed included 74 pa-
built by starting with a model with only fixed intercept and no random tients, of whom 72 were Norwegians (Caucasian), 1 was African and 1
effects. Random intercepts and random slope were then added if they was Asian. The mean age was 42.0 (SD = 12.8), mean duration of ill-
significantly increased model fit. The data was modelled for hetero- ness was 16.1 years (SD = 11.8), 45 (60.8%) were female and 29
scedastic residual variance over time. A diagonal covariance structure (39.2%) were male, 39 (52.7%) lived alone. Only 9 (12.2%) had a
of the residuals gave the best model fit. Maximum likelihood (ML) was university degree, while the majority had only upper secondary school
used as the estimation method (Fitzmaurice, Laird, & Ware, 2004). All 28 (37.8%). Of the 74 patients entering treatment, seven patients
models were tested for model fit using log likelihood tests, and the most (9.5%) terminated treatment prematurely. Two patients dropped out of
parsimonious model was selected. Since a non-linear relationship was MCT late in therapy (i.e.,after 4 weeks), and five patients dropped out
expected, with a larger decrease from pre-treatment to post-treatment, of CBT early (i.e., before 4 weeks). Drop out rate in the two conditions
we used a piecewise HLM with two different time variables. did not significantly differ (p = 0.159; Fisher’s exact test). The reasons
The MCAR (Missing Completely at Random) test (Little, 1988) was for drop out in MCT were loss of motivation (n = 1) and use of alcohol
not significant on the BAI (χ2 = 3,2, p = 0.669) or the other measures, (n = 1). In CBT the reasons were lack of motivation (n = 4) and use of
indicating that the data could be considered to be missing at random. alcohol (n = 1). In total, sixty-seven patients (MCT, N = 34; CBT,
Multiple imputations with twenty-five datasets were used on all the N = 33) completed the post-treatment assessment. See Fig. 1 for par-
self-report measures (Graham, Olchowski, & Gilreath, 2007; Graham, ticipant flow.
2009; Rubin, 1996). Pre-treatment scores were used as predictors, and
the pooled estimates used in the calculation of clinical significant
3.2. Treatment effects
change, mean and standard deviation. Due to the number of analyses
and in an effort to minimise Type I error, the false discovery rate pro-
3.2.1. Primary and secondary outcomes
cedure (Benjamini & Hochberg, 1995) was applied to determine sig-
The means and standard deviations for patients in the two treat-
nificance. Effect sizes (Cohen’s d) for HLM-analysis are calculated based
ments across outcome assessment are reported in Table 2 (self-report
on Feingold (2009). Interpretations were based on the classifications of
measures) and Table 3 (interview-based ADIS). To test for treatment
Cohen (1988). SPSS version 21.0 was used.
differences we conducted HLM with linear splines on all outcome

108
S.U. Johnson et al. Journal of Anxiety Disorders 50 (2017) 103–112

Table 3 treatments at one-year follow-up as evident in Table 2 (see supple-


Number of diagnoses and CSR from pre treatment to 1-year-follow-up (1YFW). mentary materials C and Table S6 for data on the HLM-analysis at one-
year follow-up). Subanalyses based on diagnosis revealed that the lar-
Pre treatment IYFW
gest difference between MCT and CBT at post-treatment was found for
Diagnosis CBT (38) MCT (36) CBT (30) MCT (29) PD/A. Details concerning analyses on the three specific diagnoses are
Diag (SD) 3.7 (1.6) 3.7 (1.6) 2.1 (1.6) 2.3 (1.6) presented in supplementary materials D.
CSR (SD) 22.0 (10.6) 22.6 (9.5) 14.4 (9.4) 14.0 (9.5)
PTSD (%) 25 (65.8) 15 (41.7) 10 (33.3) 13 (44.8)
PD/A (%) 23 (60.5) 28 (77.8) 6 (20.0) 10 (34.5)
3.2.2. Generalization of treatments effect
SAD (%) 21 (55.3) 24 (66.7) 11 (36.7) 13 (44.8) Treatments effect on comorbid diagnosis was established with the
GAD (%) 23 (60.5) 22 (61.1) 12 (40.0) 11 (37.9) use of three separate HLMs: number of diagnoses, CSR for all diagnoses,
MDD (current) (%) 16 (42.1) 13 (36.1) 9 (30.0) 6 (20.7) and number of 3 criteria on the SCID II. The analysis revealed no sig-
Dysthymia (%) 11 (28.9) 17 (47.2) 3 (10.0) 7 (24.1)
nificant interaction with group, for number of diagnoses (p = 0.57,
OCD (%) 4 (10.5) 1 (2.8) 2 (6.7) 1 (3.4)
SF (%) 5 (13.2) 6 (16.7) 3 (10.0) 2 (6.9) d = 0.1), or CSR total (p = 0.35, d = 0.2). However there was a sig-
BP (%) 2 (5.3) 1 (2.8) 1 (3.3) 0 (0) nificant interaction between slope and group for SCID II total criteria,
Cyclothymia (%) 0 (0) 0 (0) 1 (3.3) 0 (0) (p = 0.03, d = 0.3), indicating that from pre-treatment to one-year
Hypoch. (%) 4 (10.5) 1 (2.8) 2 (6.7) 1 (3.4) follow-up, MCT produced larger change in personality disorder pro-
SD (%) 4 (10.5) 1 (2.8) 1 (3.3) 2 (6.9)
Drug abuse (%) 2 (5.3) 0 (0) 0 (0) 0 (0)
blems. See Table S8 in supplementary material for estimates of random
Alcohol abuse (%) 1 (2.6) 2 (5.6) 1 (3.3) 1 (3.4) and fixed effects.
Alcohol dep. (%) 1 (2.6) 1 (2.8) 0 (0) 1 (3.4)
PD (%) 15 (39.5) 15 (41.7) 8 (26.7) 4 (13.8) 3.3. Common factor threats to validity
PD 3 Criteria (SD) 8.3 (7.7) 9.2 (9.8) 5.6 (6.1) 4.6 (6.0)
Cluster C PD (SD) 10 (27.8) 15 (41.7) 5 (16.7) 3 (10.3)
Cluster B PD (SD) 5 (13.9) 5 (13.9) 3 (10.0) 2 (6.9) There were no significant differences between MCT (M = 23.3,
Cluster A PD (SD) 3 (8.3) 2 (5.6) 1 (3.3) 1 (3.4) SD = 6.4) and CBT (M = 24.0, SD = 4.9) in treatment credibility at the
start of treatment using an independent sample t-test (t(1, 68) = 0.51,
Note. Diag = Number of diagnoses, CSR = Clinical severity rating (ADIS IV total score), p = 0.613, d = −0.1) or at session three between MCT (M = 24.2,
PTSD = Post traumatic stress disorder, PD/A = Panic disorder with and without agor-
SD = 4.8) and CBT (M = 23.6, SD = 4.4) (t(1, 66) = −0.69,
aphobia, SAD = Social phobia, GAD = Generalized Anxiety Disorder, MDD = Major
p = 0.496, d = 0.1). Further, there was no significant difference in the
depressive disorder, OCD = Obsessive compulsive disorder, SF = specific phobia,
BP = Bipolar disorder, Hypoch. = Hypochondriasis, SD = Somatization disorder, working alliance at session three: MCT (M = 5.9, SD = 0.9) and CBT
Alcohol dep.= Alcohol dependence, PD = Personality disorder, PD 3 Criteria = Number (M = 5.8, SD = 0.9) (t(1, 68) = −0.60, p = 0.549, d = 0.1).
of 3 criteria on SCID I, SD = Standard deviation. Accordingly, differences between treatment credibility and alliance did
not appear to compromise internal validity.
measures. Thus, there were two time-sequences, from pre-treatment to
post-treatment (slope) and from post-treatment to one-year follow-up 3.4. Therapist competence and treatment adherence
(slope2). As evident in Table 4, there was an interaction between slope
and treatment from pre-treatment to post-treatment in favour of MCT The mean overall competency rating was 4.39 (SD = 0.56) and 4.00
on BAI (p = 0.01, d = 0.7), BDI (p = 0.008, d = 0.7), SCL-90, (SD = 0.70) for MCT and CBT, respectively. Treatment adherence was
(p = 0.03, d = 0.6) and PSWQ (p = 0.002, d = 0.6), but not on ADIS 4.69 (SD = 0.60) and 4.28 (SD = 0.64) for MCT and CBT, respectively.
IV CSR primary diagnosis (p = 0.5, d = 0.3) and IIP 64 (p = 0.6, As stated previously, these values reflect adequate adherence and
d = 0.2). As indicated by Table 4 there was also a significant interac- competence (i.e., ratings > 4) for both MCT and CBT. Therapist effects
tion between slope2 and group on the same measures in the opposite were assessed with an ICC (1,1) for BAI and was found to be negligible
direction, indicating that CBT had larger gains from post-treatment to ( < 0.1). Thus, therapist effects are not likely to influence the statis-
one-year follow-up. There were no differences in effect between the tical results.

Table 4
BAI, SCL-90, BDI, IIP-64, PSWQ and ADIS in HLM.

Parameter BAI BDI SCL-90 PSWQ IIP-64 ADIS

Fixed parameters
Intercept 24.6*** (1.8), 21.2*** (1.5), 1.5*** (0.9), 58.0*** (2.4), 1.4*** (0.1), 7.0*** (0.3),
[20.9, 28.3] [18.2, 24.2] [1.3, 1.7] [53.3, 62.8] [1.2, 1.6] [6.3, 7.6]
Slope −3.5 (2.1), −4.3** (1.6), −0.4*** (0.09), −2.7 (2.0), −0.2* (0.1), −2.2** (0.8),
[−7.7, 0.7] [−7.4, −1.2] [−0.5, −0.2] [−6.6, 1.2] [−0.4,−0.1] [−3.8, −0.6]
Slope2 −2.3 (3.7), 3.6 (3.0), 0.3 (0.2), −0.01 (3.2), 0.1 (0.2), 0.6 (1.8),
[−9.6, 5.1] [−2.4, 9.5] [−0.06, 0.6] [−6.4, 6.3] [−0.2, 0.4] [−3.0, 4.1]
Group 1.9 (2.7), 0.8* (2.2), 0.01(01) 3.2 (3.4), −0,1 (0.1), 0.1 (0.2),
[−3.4, 7.2] [−3.5, 5.2] [−0.2, 0.3] [−3.6, 10.0] [−0.4, 0.2] [−0.4, 0.4]
Group*slope −7.6* (3.0), −6.0** (2.2), −0.3* (0.13), −8.7** (2.7), −0.1 (0.1), −0.3 (0.5),
[−13.5, −1.6] [−10.4, −1.6] [−0.5, −0.04] [−14.1, −3.2] [−0.3, 0.2] [−1.3, 0.7]
Group*slop- 15.0** (5.2) 10.5* (4.2) 0.5* (0.24), 13.6** (4.4), 0.2 (0.2), 1.1 (1.1),
e2 [4.7, 25.3] [2.1, 18.8] [0.04, 1.04] [4.8, 22.4] [−0.2, 0.7] [−1.1, 3.4]
Random parameters
Intercept 42.3*** (12.1) 53.0*** (11.7) 0.20*** (0.04) 139.7*** (26.5) 0.26*** (0.05) 0.5** (0.2)
-2 LL 1462.5 1403.2 306.8 1463.2 291.6 732.2

Note. 2 LL = −2 Log Likelihood Standard error is given in parenthesis, 95% Confidence interval given in brackets.
*p < 0.05. **p < 0.01. ***p < 0.001. BAI = Beck Anxiety Inventory, BDI = Beck Depression Inventory, IIP 64 = Inventory of Interpersonal Problems 64, SCL–90 = Symptom
Checklist 90, PSWQ = Penn State Worry Questionnaire, ADIS = Anxiety Disorder Interview Scale (primary diagnosis clinical severity rating).

109
S.U. Johnson et al. Journal of Anxiety Disorders 50 (2017) 103–112

Table 5 high degrees of comorbidity. There is a negative correlation between


Clinical significant change for MCT, CBT and the total sample at post-treatment and 1- mental health and unemployment (Murphy & Athanasou, 1999) and
year-follow-up.
being out of work at the start of treatment has been found to predict
Post-treatment poorer long-term outcome (Durham et al., 2005; Hoffart, Øktedalen,
Svanøe, Hedley & Sexton, 2015). Thus, the effect sizes of the present
Treatment Recovered Reliable Change Unchanged Detoriated study should be interpreted in this context. Furthermore, attrition rates,
9.5% from pre- to post- and 20.2% from pre- to one-year follow-up were
MCT (n = 36) 15 (41.6) 5 (13.8) 14 (38.9) 2 (5.6)
CBT (n = 38) 7 (18.4) 7 (18.4) 20 (52.6) 4 (10.5) modest and no group differences emerged. The attrition rates were
Total (N = 74) 22 (29.7) 12 (16.2) 34 (45.9) 6 (8.1) lower than the weighted mean attrition (23%) reported in a meta-
1-year-follow-up
analysis of CBT studies for anxiety disorders (Hofmann & Smits, 2008).
MCT (n = 36) 10 (27.8) 6 (16.7) 17 (47.2) 3 (8.3) The number of diagnoses was reduced from pre- to post-treatment
CBT (n = 38) 14 (36.8) 5 (13.2) 16 (42.1) 3 (7.9) and maintained at one-year follow-up with no significant differences
Total (N = 74) 24 (32.4) 11 (14.9) 33 (44.6) 6 (8.1) between the treatment groups. The results replicate and extend pre-
vious work on the effect of CBT on comorbid symptoms and diagnosis
Note. Percentages given in parentheses.
for anxiety disorders (Norton et al., 2013; Tsao, Lewin & Craske, 1998).
The results did not support our hypothesis that working on transdiag-
3.5. Clinical significant change
nostic processes, such as worry and rumination (MCT), would lead to
larger effects on comorbid symptoms and diagnoses. However, MCT
At post-treatment, 55.4% and 36.8% of the patients in MCT and CBT
reduced personality-problems more then CBT, indicating a possible
were either recovered or improved, respectively (see Table 5). At one-
broader effect of MCT on comorbid personality problems, a result that
year follow-up, 44.8% and 50.0% of the patients in MCT and CBT were
needs further investigation.
either recovered or improved.
4.1. Strengths and limitations
4. Discussion
This study has several strengths. A new promising treatment was
The main purpose of the study was to compare the effect of MCT compared to the best documented form of treatment for anxiety dis-
with a gold standard treatment for anxiety disorders, namely CBT. We orders and the study was conducted in a naturalistic setting with
hypothesized that MCT as a transdiagnostic treatment would have treatment resistant comorbid anxiety disorder patients.
greater effect on anxiety and comorbid diagnoses and symptoms in a Several limitations should also be noted. First, we did not include a
mixed anxiety disorder sample than CBT. There was an interaction no-treatment control group, which prevents examination of whether
between slope and treatment from pre-treatment to post-treatment in the two treatments were superior to no treatment. However, we know
favour of MCT on anxiety (BAI), depression (BDI), worry (PSWQ) and that the level of spontaneous recovery in chronic, treatment resistant,
general symptoms (SCL-90) but not on ADIS IV and interpersonal and comorbid anxiety disorder patients is low (Bruce et al., 2008). Also,
functioning (IIP 64). From pre-treatment to one-year follow-up there comparing established treatments does not typically require a no
were no significant differences between the treatments. There were no treatment or waitlist control group, the use of which may then be
significant differences between MCT and CBT on treatment credibility considered unethical (Devilly & McFarlane, 2009). Second, four pa-
at session one or session three, indicating that both treatments had a tients were removed from the trial due to changes in the clinical staff,
rationale that seemed plausible and that therapists form alliances in the damaging random assignment. Moreover, due to drop out before
two treatments that were comparable. Moreover, the adherence and treatment the patients analysed do not fit the strict definition of an
competence ratings in both treatments were at a satisfactory level intent-to-treat sample. Third, the study was carried out at one clinic,
(> 4), strengthening the internal validity of the study. which suggests that there may have been contamination of the thera-
A conceivable reason for the more rapid effect of MCT at post- pies between the patients during the treatment phase, although an at-
treatment is that MCT directly challenges emotion regulation strategies tempt was made to minimise the likelihood of such contamination.
at the process level in the form of worry and rumination, while CBT Fourth, we have not controlled for the ad hoc and additional treatments
may have needed a longer time to work through the content of cogni- that the patients received post-treatment, which might favour CBT as
tion. However, during the one-year follow-up period CBT patients many patients in both conditions received CBT in the follow-up period.
continued to improve, while the MCT patients remained stable. One
possible explanation is that the patients in the CBT group needed more 5. Conclusion
time to make use of the tools learned in therapy. For example, beha-
vioral experiments are a common practice in all three CBT manuals, This is the first randomized controlled trial comparing MCT and
thus patients returning home and continuing exposure could have en- CBT for comorbid anxiety disorders, and both treatments produced
hanced treatment effects. Notably, 67.8% of the sample (73% CBT and satisfactory effects, particularly for this treatment-resistant population.
62% MCT) had treatment in the follow-up period, which is normal for There were significant differences in favour of MCT from pre-treatment
this patient group. Of those receiving additional treatment, 33% had to post-treatment, but these differences disappeared at one-year follow-
CBT treatment, while the majority (49%) of participants could not tell up. The CBT group used the best-documented and widespread form of
what kind of treatment they were given. The lack of continued im- CBT for each primary anxiety disorder, while the MCT group used a
provement of MCT at one-year follow-up could be due to a shift in generic model. There are advantages to having to learn only one
treatment that was not compatible with MCT. treatment for anxiety rather than several. Learning one model is less
The effect sizes on the BAI from pre-treatment to one-year follow-up time consuming than learning three specific ones, which has ad-
were large, though smaller then found in other trials for MCT and CBT vantages for training of therapists. This study suggests that adoption of
(cf., Nordahl, 2009). However, in studies with highly comorbid anxiety MCT for patients with comorbid anxiety disorders may be warranted.
disorders the effect sizes were significantly lower than those in the
present study, and duration of illness had a significant negative asso- Acknowledgement
ciation with the effect of treatment (cf., Sánchez-Meca, Rosa-Alcázar,
Marín-Martínez, & Gómez-Conesa, 2010). Our sample consisted of pa- This study was supported by “The National Program for Integrated
tients who all had failed previous treatments, were out of work and had Clinical Specialist and PhD-training for Psychologists” in Norway. We

110
S.U. Johnson et al. Journal of Anxiety Disorders 50 (2017) 103–112

thank the individuals who participated in the study, the department of Washington, DC: American Psychiatric Press.
Fitzmaurice, G., Laird, N., & Ware, J. (2004). Applied longitudinal analysis. New York: John
anxiety disorder at Modum Bad for data collection and the research Wiley and Sons.
assistants who assisted in data collection. Foa, E. B., Hembree, E. A., & Rothbaum, B. O. (2007). Prolonged exposure therapy for PTSD:
Emotional processing of traumatic experiences therapist guide (Treatments that work).
USA: Oxford University Press.
Appendix A. Supplementary data Fogliati, V. J., Dear, B. F., Staples, L. G., Terides, M. D., Sheehan, J., Johnston, L., ... Titov,
N. (2016). Disorder-specific versus transdiagnostic and clinician-guided versus self-
Supplementary data associated with this article can be found, in the guided internet-delivered treatment for panic disorder and comorbid disorders: A
randomized controlled trial. Journal of Anxiety Disorders, 39, 88–102. http://dx.doi.
online version, at http://dx.doi.org/10.1016/j.janxdis.2017.06.004. org/10.1016/j.janxdis.2016.03.005.
Gadermann, A., Alonso, J., Vilagut, G., Zaslavsky, A., & Kessler, R. (2012). Comorbidity
References and disease burden in the national comorbidity survey replication (NCS-R).
Depression and Anxiety, 29, 797–806. http://dx.doi.org/10.1002/da.21924.
Gillis, M. M., Haaga, D. A. F., & Ford, G. T. (1995). Normative values for the beck anxiety
Aldao, A., & Nolen-Hoeksema, S. (2010). Specificity of cognitive emotion regulation inventory, fear questionnaire, penn state worry questionnaire, and social phobia and
strategies: A transdiagnostic examination. Behaviour Research and Therapy, 48, anxiety inventory. Psychological Assessment, 7, 450–455. http://dx.doi.org/10.1037/
974–983. http://dx.doi.org/10.1016/j.brat.2010.06.002. 1040-3590.7.4.450.
Arch, J. J., Eifert, G. H., Davies, C., Vilardaga, J. C. P., Rose, R. D., & Craske, M. G. (2012). Graham, J. W., Olchowski, A. E., & Gilreath, T. D. (2007). How many imputations are
Randomized clinical trial of cognitive behavioral therapy (CBT) versus acceptance really needed? Some practical clarifications of multiple imputation theory. Prevention
and commitment therapy (ACT) for mixed anxiety disorders. Journal of Consulting and Science, 8, 206–213. http://dx.doi.org/10.1007/s11121-007-0070-9.
Clinical Psychology, 80, 750–765. http://dx.doi.org/10.1037/a0028310. Graham, J. W. (2009). Missing data analysis: Making it work in the real world. Annual
Beck, A. T., Steer, R., & Brown, G. (1996). Beck depression inventory-II manual. San Review of Psychology, 60, 549–576. http://dx.doi.org/10.1146/annurev.psych.58.
Antonio, TX: The Psychological Corporation. 110405.085530.
Beck, A. T. (1976). Cognitive therapy and the emotional disorders. New York: International Hagen, R., Hjemdal, O., Solem, S., Kennair, L. E. O., Nordahl, H. M., Fisher, P., & Wells, A.
Universities Press. (2017). Metacognitive therapy for depression in adults: A waiting list randomized
Beck, A. T., Epstein, N., Brown, G., & Steer, R. A. (1988). An inventory for measuring controlled trial with six months follow-Up. Frontiers in Psychology, 8(31), http://dx.
clinical anxiety: Psychometric properties. Journal of Consulting and Clinical doi.org/10.3389/fpsyg.2017.00031.
Psychology, 56, 893–897. http://dx.doi.org/10.1037/0022-006X.56.6.893. Harvey, A. G., Watkins, E. R., Mansell, W., & Shafran, R. (2004). Cognitive-behavioral
Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric properties of the Beck processes across psychological disorders: A transdiagnostic approach to research and
Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8, treatment. Oxford: Oxford University Press.
77–100. http://dx.doi.org/10.1037/0022-006X.56.6.893. Hatcher, R. L., & Gillaspy, J. A. (2006). Development and validation of a revised short
Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and version of the Working Alliance Inventory. Psychotherapy Research, 16, 12–25. http://
powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B dx.doi.org/10.1080/10503300500352500.
(Methodological), 57, 289–300. Hoffart, A., Øktedalen, T., Svanøe, K., Hedley, L. M., & Sexton, H. (2015). Predictors of
Borkovec, T. D., & Nau, S. D. (1972). Credibility of analogue therapy rationales. Journal of short- and long-term avoidance in completers of inpatient group interventions for
Behavior Therapy and Experimental Psychiatry, 3, 257–260. http://dx.doi.org/10. agoraphobia. Journal of Affective Disorders, 181, 33–40. http://dx.doi.org/10.1016/j.
1016/0005-7916(72)90045-6. jad.2015.04.015.
Brady, K. T., Killeen, T. K., Brewerton, T., & Lucerini, S. (2000). Comorbidity of psy- Hofmann, S. G., & Smits, J. A. J. (2008). Cognitive-behavioral therapy for adult anxiety
chiatric disorders and posttraumatic stress disorder. Journal of Clinical Psychiatry, disorders: A meta-analysis of randomized placebo-controlled trials. The Journal of
61(Suppl. 7), 22–32. Clinical Psychiatry, 69, 621–632.
Brown, T. A., Di Nardo, P. A., & Barlow, D. H. (1994). Anxiety disorders interview schedule Horowitz, L. M., Rosenberg, S. E., Baer, B. A., Ureño, G., & Villaseñor, V. S. (1988).
for DSM-IV (adult version). Albany, NY: Graywind. Inventory of interpersonal problems: Psychometric properties and clinical applica-
Bruce, S. E., Yonkers, K. A., Otto, M. W., Eisen, J. L., Weisberg, R. B., Pagano, M., ... tions. Journal of Consulting and Clinical Psychology, 56, 885–892. http://dx.doi.org/
Keller, M. B. (2008). Influence of psychiatric comorbidity on recovery and recurrence 10.1037/0022-006X.56.6.885.
in generalized anxiety disorder, social phobia, and panic disorder: A 12-year pro- Horowitz, L. M., Alden, L. E., Wiggins, J. S., & Pincus, A. L. (2000). Inventory of inter-
spective study. Focus, 6, 539–548. http://dx.doi.org/10.1176/foc.6.4.foc539. personal problems: San Antoni. TX: Psychological Corporation.
Butler, A. C., Chapman, J. E., Forman, E. M., & Beck, A. T. (2006). The empirical status of Horvath, A. O., & Greenberg, L. S. (1989). Development and validation of the working
cognitive-behavioral therapy: A review of meta-analyses. Clinical Psychology Review, alliance inventory. Journal of Counseling Psychology, 36, 223–233. http://dx.doi.org/
26, 17–31. http://dx.doi.org/10.1016/j.cpr.2005.07.003. 10.1037/0022-0167.36.2.223.
Clark, D. M., & Wells, A. (1995). A cognitive model of social phobia. Social Phobia: Jacobson, N. S., & Truax, P. (1991). Clinical significance: A statistical approach to de-
Diagnosis, Assessment, and Treatment, 41, 68. fining meaningful change in psychotherapy research. Journal of Consulting and
Clark, D. M. (1986). A cognitive approach to panic. Behaviour Research and Therapy, 24, Clinical Psychology, 59, 12–19. http://dx.doi.org/10.1037/0022-006X.59.1.12.
461–470. http://dx.doi.org/10.1016/0005-7967(86)90011-2. Johnson, S. U., & Hoffart, A. (2016). Metacognitive therapy for comorbid anxiety dis-
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Lawrence Erlbaum orders: A case study. Frontiers in Psychology, 7(1515), http://dx.doi.org/10.3389/
Associates. fpsyg.2016.01515.
Collimore, K. C., Carleton, R. N., Hofmann, S. G., & Asmundson, G. J. G. (2010). Kessler, R. C., Chiu, W. T., Demler, O., & Walters, E. E. (2005). Prevalence, severity, and
Posttraumatic stress and social anxiety: The interaction of traumatic events and in- comorbidity of 12-month DSM-IV disorders in the national comorbidity survey re-
terpersonal fears. Depression and Anxiety, 27, 1017–1026. http://dx.doi.org/10.1002/ plication. Archives of General Psychiatry, 62, 617–627. http://dx.doi.org/10.1001/
da.20728. archpsyc.62.6.617.
Cougle, J. R., Feldner, M. T., Keough, M. E., Hawkins, K. A., & Fitch, K. E. (2010). Kessler, R. C., Avenevoli, S., McLaughlin, K. A., Green, J. G., Lakoma, M. D., Petukhova,
Comorbid panic attacks among individuals with posttraumatic stress disorder: M., ... Merikangas, K. R. (2012). Lifetime co-morbidity of DSM-IV disorders in the US
Associations with traumatic event exposure history, symptoms, and impairment. national comorbidity survey replication adolescent supplement (NCS-A).
Journal of Anxiety Disorders, 24, 183–188. http://dx.doi.org/10.1016/j.janxdis.2009. Psychological Medicine, 42, 1997–2010. http://dx.doi.org/10.1002/mpr.1359.
10.006. Kwok, O.-M., Underhill, A. T., Berry, J. W., Luo, W., Elliott, T. R., & Yoon, M. (2008).
Dear, B. F., Staples, L. G., Terides, M. D., Karin, E., Zou, J., Johnston, L., ... Titov, N. Analyzing longitudinal data with multilevel models: An example with individuals
(2015). Transdiagnostic versus disorder-specific and clinician-guided versus self- living with lower extremity intra-articular fractures. Rehabilitation Psychology, 53,
guided internet-delivered treatment for generalized anxiety disorder and comorbid 370–386. http://dx.doi.org/10.1037/a0012765.
disorders: A randomized controlled trial. Journal of Anxiety Disorders, 36, 63–77. Little, R. J. A. (1988). A test of missing completely at random for multivariate data with
http://dx.doi.org/10.1016/j.janxdis.2015.09.003. missing values. Journal of the American Statistical Association, 83, 1198–1202. http://
Derogatis, L. R. (1983). The symptom checklist −90 revised: Administration. scoring and dx.doi.org/10.2307/2290157.
procedures manual II (1983). Baltimore: Clinical Psychometric Research. McEvoy, P. M., Erceg-Hurn, D. M., Anderson, R. A., Campbell, B. N. C., Swan, A.,
Devilly, G. J., & Borkovec, T. D. (2000). Psychometric properties of the credibility/ex- Saulsman, L. M., ... Nathan, P. R. (2015). Group metacognitive therapy for repetitive
pectancy questionnaire. Journal of Behavior Therapy and Experimental Psychiatry, 31, negative thinking in primary and non-primary generalized anxiety disorder: An ef-
73–86. http://dx.doi.org/10.1016/S0005-7916(00)00012-4. fectiveness trial. Journal of Affective Disorders, 175, 124–132. http://dx.doi.org/10.
Devilly, G. J., & McFarlane, A. C. (2009). When wait lists are not feasible, nothing is a 1016/j.jad.2014.12.046.
thing that does not need to be done. Journal of Consulting and Clinical Psychology, 77, McMillan, K. A., Sareen, J., & Asmundson, G. J. (2014). Social anxiety disorder is asso-
1159–1168. http://dx.doi.org/10.1037/a0016878. ciated with PTSD symptom presentation: An exploratory study within a nationally
Durham, R. C., Chambers, J. A., Power, K. G., Sharp, D. M., Macdonald, R. R., Major, K. representative sample. Journal of Traumatic Stress, 27, 602–609. http://dx.doi.org/
A., ... Gumley, A. L. (2005). Long-term outcome of cognitive behaviour therapy 10.1002/jts.21952.
clinical trials in central Scotland. Healt Technology Assessment, 9, 1–4. Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Development and
Feingold, A. (2009). Effect sizes for growth-modeling analysis for controlled clinical trials validation of the penn state worry questionnaire. Behaviour Research and Therapy, 28,
in the same metric as for classical analysis. Psychological Methods, 14, 43–53. http:// 487–495. http://dx.doi.org/10.1016/0005-7967(90)90135-6.
dx.doi.org/10.1037/a0014699. Moses, E. B., & Barlow, D. H. (2006). A new unified treatment approach for emotional
First, B. M., Gibbon, M., Spitzer, R. L., Williams, J. B. W., & Benjamin, L. S. (1997). User’s disorders based on emotion science. Current Directions In Psychological Science, 15,
guide for the structured clinical interview for DSM-IV Axis II personality disorders: SCID-II. 146–150. http://dx.doi.org/10.1111/j.0963-7214.2006.00425.x.

111
S.U. Johnson et al. Journal of Anxiety Disorders 50 (2017) 103–112

Murphy, G. C., & Athanasou, J. A. (1999). The effect of unemployment on mental health. Dunbar, G. C. (1998). The Mini-International Neuropsychiatric Interview (M.I.N.I.):
Journal of Occupational and Organizational Psychology, 72, 83–99. http://dx.doi.org/ The development and validation of a structured diagnostic psychiatric interview for
10.1348/096317999166518. DSM-IV and ICD-10. Journal of Clinical Psychiatry, 59, 22–33.
National Institute for Health and Care Excellence (2013). Social anxiety disorder: Steer, R. A., Ranieri, W. F., Beck, A. T., & Clark, D. A. (1993). Further evidence for the
Recognition, assessment and treatment. NICE clinical guideline 159. validity of the beck anxiety inventory with psychiatric outpatients. Journal of Anxiety
Nordahl, H. M., & Wells (2009). Metacognitive therapy competence scale. Metacognitive Disorders, 7, 195–205. http://dx.doi.org/10.1016/0887-6185(93)90002-3.
Therapy Institute Unpublished manuscript. Titov, N., Dear, B. F., Staples, L. G., Terides, M. D., Karin, E., Sheehan, J., ... McEvoy, P. M.
Nordahl, H. M. (2009). Effectiveness of brief metacognitive therapy versus cognitive- (2015). Disorder-specific versus transdiagnostic and clinician-guided versus self-
behavioral therapy in a general outpatient setting. International Journal of Cognitive guided treatment for major depressive disorder and comorbid anxiety disorders: A
Therapy, 2, 152–159. http://dx.doi.org/10.1521/ijct.2009.2.2.152. randomized controlled trial. Journal of Anxiety Disorders, 35, 88–102. http://dx.doi.
Normann, N., van Emmerik, A. A. P., & Morina, N. (2014). The efficacy of metacognitive org/10.1016/j.janxdis.2015.08.002.
therapy for anxiety and depression: A meta-analytic review. Depression and Anxiety, Tolin, D. F. (2010). Is Cognitive-behavioural therapy more effective than other therapies?
31, 402–411. http://dx.doi.org/10.1002/da.22273. A meta-analytic review. Clinical Psychology Review, 30, 710–720. http://dx.doi.org/
Norton, P. J., & Barrera, T. L. (2012). Transdiagnostic versus diagnosis-specific CBT for 10.1016/j.cpr.2010.05.003.
anxiety disorders: A preliminary randomized controlled noninferiority trial. Tsao, J. C. I., Lewin, M. R., & Craske, M. G. (1998). The effects of cognitive-behavior
Depression and Anxiety, 29, 874–882. http://dx.doi.org/10.1002/da.21974. therapy for panic disorder on comorbid conditions. Journal of Anxiety Disorders, 12,
Norton, P. J., & Paulus, D. J. (2015). Toward a unified treatment for emotional disorders: 357–371. http://dx.doi.org/10.1016/S0887-6185(98)00020-6.
Update on the science and practice. Behavior Therapy, 47, 854–868. http://dx.doi. Vallis, T. M., Shaw, B. F., & Dobson, K. S. (1986). The cognitive therapy scale:
org/10.1016/j.beth.2015.07.002. Psychometric properties. Journal of Consulting and Clinical Psychology, 54, 381–385.
Norton, P. J., Barrera, T. L., Mathew, A. R., Chamberlain, L. D., Szafranski, D. D., Reddy, http://dx.doi.org/10.1037/0022-006X.54.3.381.
R., & Smith, A. H. (2013). Effect of transdiagnostic CBT for anxiety disorders on Van der Heiden, C., Muris, P., & van der Molen, H. T. (2012). Randomized controlled trial
comorbid diagnoses. Depression and Anxiety, 30, 168–173. http://dx.doi.org/10. on the effectiveness of metacognitive therapy and intolerance-of-uncertainty therapy
1002/da.22018. for generalized anxiety disorder. Behaviour Research and Therapy, 50, 100–109.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data http://dx.doi.org/10.1016/j.brat.2011.12.005.
analysis methods (2nd ed.). Newbury Park, CA: Sage. Wampold, B. E., & Imel, Z. E. (2015). The great psychotherapy debate: The evidence for what
Raudenbush, S. W., & Liu, X. (2000). Statistical power and optimal design for multisite makes psychotherapy work (2nd ed.). Routledge.
randomized trials. Psychological Methods, 5, 199–213. Wells, A., & Matthews, G. (1994). Attention and emotion: A clinical perspective. Erlbaum
Reinholt, N., & Krogh, J. (2014). Efficacy of transdiagnostic cognitive behaviour therapy Associates.
for anxiety disorders: A systematic review and meta-analysis of published outcome Wells, A., & Matthews, G. (1996). Modelling cognition in emotional disorder: The S-REF
studies. Cognitive Behaviour Therapy, 43, 171–184. http://dx.doi.org/10.1080/ model. Behaviour Research and Therapy, 34, 881–888. http://dx.doi.org/10.1016/
16506073.2014.897367. S0005-7967(96)00050-2.
Rubin, D. B. (1996). Multiple imputation after 18+ years (with discussion). Journal of the Wells, A., Fisher, P., Myers, S., Wheatley, J., Patel, T., & Brewin, C. R. (2012).
American Statistical Association, 91, 473–489. http://dx.doi.org/10.2307/2291635. Metacognitive therapy in treatment-resistant depression: A platform trial. Behaviour
Sánchez-Meca, J., Rosa-Alcázar, A. I., Marín-Martínez, F., & Gómez-Conesa, A. (2010). Research and Therapy, 50, 367–373. http://dx.doi.org/10.1016/j.brat.2012.02.004.
Psychological treatment of panic disorder with or without agoraphobia: A meta- Wells, A., Walton, D., Lovell, K., & Proctor, D. (2015). Metacognitive therapy versus
analysis. Clinical Psychology Review, 30, 37–50. http://dx.doi.org/10.1016/j.cpr. prolonged exposure in adults with chronic post-traumatic stress disorder: A parallel
2009.08.011. randomized controlled trial. Cognitive Therapy and Research, 39, 70–80.
Schmitz, N., Hartkamp, N., Kiuse, J., Franke, G. H., Reister, G., & Tress, W. (2000). The Wells, A. (1997). Cognitive therapy of anxiety disorders. Wiley.
symptom check-list-90-R (SCL-90-R): A german validation study. Quality of Life Wells, A. (2009). Metacognitive therapy for anxiety and depression. The Guilford Press.
Research, 9, 185–193. Young, J., & Beck, A. (1980). Cognitive therapy scale: Rating manual. Philadelphia, PA:
Sheehan, D. V., Lecrubier, Y., Sheehan, K. H., Amorim, P., Janavs, J., Weiller, E., ... Center for Cognitive Therapy Unpublished manuscript.

112