Working alliance and competence as predictors of outcome in cognitive behavioral therapy for social anxiety and panic disorder in adults
Introduction
The effect of cognitive behavioral therapy (CBT) for anxiety disorders has been documented in recent decades in a large number of randomized controlled trials and meta-analyses (Hofmann and Smits, 2008, Norton and Price, 2007, Olatunji et al., 2010). However, there is less knowledge from research on the relationship between process factors and treatment outcome.
Research on process factors in psychotherapy can be divided into two traditions, one emphasizing specific and unique factors for a specific treatment orientation, and the other emphasizing trans-theoretical mechanisms, frequently termed common factors (Ahn and Wampold, 2001, Messer and Wampold, 2002). Researchers emphasizing the specific factors in the treatment hypothesize that the therapist's adherence to and competence in delivering treatment in accordance with the specified theoretical and technical model are important in explaining psychotherapy outcomes. However, this assumption has so far received limited support from research findings. A meta-analysis indicated very small summary effect sizes for the association between the outcome and adherence- (r = .03, ns), as well as competence (r = .07, ns) when aggregated across different treatment modalities for various conditions (Webb, DeRubeis, & Barber, 2010). Some evidence suggest that higher therapist competence is related to better treatment outcomes in cognitive therapy of depression (Kuyken and Tsivrikos, 2009, Shaw et al., 1999, Strunk et al., 2010); however, research on this relationship in CBT of anxiety disorders is limited. Some recent studies have indicated a positive competence–outcome association in CBT for mixed anxiety and social anxiety disorder whereas other studies have indicated no- or even a negative - association between competence/adherence and the outcome in the treatment of panic disorder (Boswell et al., 2013, Huppert et al., 2006).
Advocates of the common factors perspective argues in general that treatment outcomes in CBT, as in other therapies, are predominantly regulated by trans-theoretical mechanisms (Ahn and Wampold, 2001, Messer and Wampold, 2002). In this tradition, the concept of working alliance (Bordin, 1994) has a central position. Several meta-analyses, the last one including 190 separate studies (Horvath, Del Re, Flückiger, & Symonds, 2011), indicate a consistent positive association between the quality of the working alliance and the treatment outcome. No apparent moderators have so far been identified for this association, indicating that the alliance is of similar importance for different types of disorders, and across different theoretical models, treatment formats, study designs, and outcome measures (Del Re et al., 2012, Fluckiger et al., 2012, Horvath et al., 2011). However, the quality of the alliance explains only a moderate part of the total variance in the outcome, approximately 5–7%, according to major meta-analyses (Horvath et al., 2011, Horvath and Symonds, 1991, Martin et al., 2000). Moreover, most studies on this association have not controlled for temporal confound or other potential confounding factors (Barber et al., 2009, Barber et al., 2007, Webb et al., 2010).
A common perspective among CBT researchers is that the alliance primarily operates as an auxiliary factor that facilitates the utility of specific CBT techniques. Aggregated research findings indicate that a stronger alliance is associated with a better treatment outcome, also in CBT (Castonguay et al., 2010, Waddington, 2002). However, the assumed causal alliance-outcome relationship has been questioned. Several studies have reported a non-significant alliance–outcome association in cognitive therapy when the improvement of the patients prior to the alliance ratings is taken into account (DeRubeis et al., 2005, Derubeis and Feeley, 1990, Feeley et al., 1999, Webb et al., 2014). This suggests that the quality of the alliance may rather be a consequence than an antecedent of improvement from therapy. In comparison to depression, however; relatively few studies have investigated the alliance–outcome association in CBT for anxiety disorders, and the findings are more mixed (Newman & Stiles, 2006). While some studies have indicated a positive alliance–outcome association (Huppert et al., 2014), others report non-significant findings in CBT for panic disorder and social anxiety disorder (Mörtberg, 2014, Ramnerö and Öst, 2007, Woody and Adessky, 2002). Taken together, the current research on therapist competence/adherence and the working alliance as predictors of outcome from CBT for anxiety disorders is inconclusive.
High rates of dropout are a common problem in clinical trials on CBT for anxiety disorders (Taylor, Abramowitz, & McKay, 2012). Assuming that patients who terminate treatment prematurely fare less well than those who complete treatment, it is important to also identify factors that are associated with dropout. There are some studies indicating that poorer alliance (Taylor et al., 2012) and lower therapist competence (Brown et al., 2013) is associated with dropout from treatment, but the research is limited also on this issue.
Alliance and the therapist' competence and adherence are frequently conceptualized as stable characteristics within a therapist and within a treatment. This is reflected in the design of the majority of studies in this field as competence/adherence and/or alliance are assessed at one single time-point, typically early in the therapy (Horvath et al., 2011, Webb et al., 2010). Alliance and competence might, however, vary between sessions for the same patient as well as across different patients of the same therapist (Boswell et al., 2013). Obtaining ratings from more than one time point and across treatments has been recommended to investigate the stability of these constructs (Crits-Christoph et al., 2011, Dobson and Mintz, 2006).
Relatively few studies have simultaneously investigated the association between competence and alliance, and outcome from psychotherapy (Barber et al., 2008, Barber et al., 2006, Hoffart et al., 2005, Trepka et al., 2004). The findings from these studies are inconclusive; however, there are some indications that the alliance and therapist competence overlap to some degree and explain some shared variance (Webb et al., 2010). It is therefore important to investigate alliance and competence simultaneously, and to control for other potential confounding factors such as early treatment gain and patient characteristics at baseline (Barber, 2009, Feeley et al., 1999, Kazdin and Whitley, 2006). Finally, most studies on the relationship between therapist adherence/competence and the treatment outcome have used data from efficacy trials with strictly selected, intensively trained, and frequently supervised therapists, resulting in very high competence ratings with little variation in scores (Roth, Pilling, & Turner, 2010). This restriction of the range limits the possibility of detecting potential associations between competence and outcome. Therefore it has been recommended to study this relationships in samples with greater variability in therapists' competence (Crits-Christoph, Gibbons, & Mukherjee, 2013).
This study investigated therapist competence/adherence and alliance as predictors of outcome from an effectiveness trial on CBT of panic disorder (PD) and social anxiety disorder (SAD). With therapists less experienced in CBT, we expected a larger variation in therapist competence/adherence than what is typically reported from efficacy trials. Moreover, it was controlled for temporal confound and other potential confounding factors. The primary objectives were as follows: 1) to examine whether the therapists' competence/adherence and the patients ratings of the working alliance were overlapping and interrelated concepts; 2) to examine whether competence/adherence and alliance ratings were stable across sessions 3 and 8 in the treatment; 3) to investigate the associations between working alliance, therapist competence/adherence, and treatment outcome while controlling for prior symptom improvement and other potential confounding factors; 4) to examine whether the associations between alliance, competence/adherence, and outcome were the same in the early and late phases of the treatment, and if they were similar for PD and SAD; and 5) to investigate the associations between working alliance, therapist competence/adherence and dropout.
Section snippets
Procedure
This study used data from the adult part of a randomized controlled effectiveness trial of CBT for SAD and PD. The trial, “Assessment and Treatment – Anxiety in Children and Adults” (ATACA) study, was conducted from March 2008–June 2012 in western Norway (Nordgreen et al., 2015). The patients were recruited, assessed, and treated in nine public mental health outpatient clinics. Referred patients with an indication of PD or SAD symptomatology were screened for the primary diagnosis, substance
Diagnostics
The Structural Clinical Interview for DSM-IV Axis-I disorders (SCID-I) (First, Spitzer, Gibbon, & Williams, 1997) and Axis-II disorders (SCID-II) (First, Spitzer, Gibbon, Williams, & Benjamin, 1994) were used. The interviews were videotaped, and a blind rescoring of a random selection of 20% of the interviews provided perfect agreement for PD and SAD (Cohen's kappa = 1.0).
Clinical rating
The Clinician Severity Rating (CSR) (Di Nardo et al., 1994) is a 0–8 point scale that was used by the assessors to rate a combination of the severity of the disorder and the consequences for daily functioning. A score of 0 indicates no symptoms or impairment, whereas a score of 8 indicates very severe symptoms with major disabling consequences. Blind rescoring of 19 randomly selected videos were used to assess inter-rater reliability of CSR. A two-way mixed effects model using absolute
Data screening and preliminary analyses
The data were screened for outliers and normality by checking the leverage and using the Kolmogorov–Smirnov test. The predictors were plotted against the residuals to test for linearity and homoscedasticity. In the inequality of variance test, we followed Field (2005) recommendation that one should proceed with the analysis if the ratio of the highest and lowest variance value is less than 2, even if the Levene's test was significant. As recommended by Tabachnick and Fidell (2007), the
Outcome from the treatment
The results indicated large pre-post effect sizes on the outcome for both diagnoses on CSR (PD: d = 2.36, SAD: d = 2.28), and the self-report questionnaires (BSQ (PD): d = 1.03, SPS (SAD): d = .83) (Appendix A).
Stability of the competence/adherence and the alliance
The mean total WAI-S score across the rated sessions was in the upper positive end of the 0–7 rating scale (M = 5.59, SD = 0.84, range 3.42-7.00), whereas the mean total CTACS-R score was closer to the midpoint of the 0–6 rating scale (M = 3.36, SD = 0.96, range 0.76-5.34). The WAI-S
Discussion
This article is to the best of our knowledge the first to address the effect of the therapist competence/adherence and the working alliance simultaneously on the outcome of face-to-face CBT for panic disorder (PD) and social anxiety disorder (SAD), while controlling for prior symptom improvement and baseline predictors. The treatment was conducted in a naturalistic setting and the therapists had less training and prior experience in CBT, and a wider range in the CBT competence level compared to
Limitations
Several limitations in this study should be noted. The competence/adherence raters were not completely independent because two of the three raters were clinical supervisors, and one rater was a therapist in the project. To overcome this limitation, the raters who were supervisors only rated the video sessions of therapists whom they did not supervise, and the rater who was a therapist did not rate his own therapy sessions. The scale that was used for rating of competence/adherence was
Conclusion and future directions
Despite these limitations, the findings from this study suggest that competence/adherence and alliance are stable characteristics with significant and independent contributions to explaining the variance in outcome from CBT of PD and SAD. Future research on these process variables should be based on ratings from a larger number of sessions of the therapy. This may provide more reliable assessments of competence/adherence and alliance, and more knowledge regarding the stability and patterns of
Acknowledgments
This study is a part of the adult part of the research project “Assessment and Treatment -Anxiety in Children and Adults. Adult part” (ATACA) and has received support from the Western Norway Regional Health Authority, through project no. 911366 and project no. 911253. ClinicalTrials.gov Identifier: NCT00619138.
References (81)
- et al.
Sudden gains in group cognitive-behavioral therapy for panic disorder
Behaviour Research and Therapy
(2008) - et al.
Self-reported versus clinician-rated symptoms of depression as outcome measures in psychotherapy research on depression: a meta-analysis
Clinical Psychology Review
(2010) - et al.
Psychological treatment of generalized anxiety disorder: a meta-analysis
Clinical Psychology Review
(2014) - et al.
Therapist effects in the therapeutic alliance-outcome relationship: a restricted-maximum likelihood meta-analysis
Clinical Psychology Review
(2012) - et al.
Treatment specific competence predicts outcome in cognitive therapy for social anxiety disorder
Behaviour Research and Therapy
(2012) - et al.
Stepped care versus face-to–face cognitive behavior therapy for panic disorder and social anxiety disorder: predictors and moderators of outcome
Behaviour Research and Therapy
(2015) Cognitive-behavioral therapy for social anxiety disorder: current status and future directions
Biological Psychiatry
(2002)- et al.
The interaction of motivation and therapist adherence predicts outcome in cognitive behavioral therapy for panic disorder: preliminary findings
Cognitive and Behavioral Practice
(2006) - et al.
Therapist effects and the outcome–alliance correlation in cognitive behavioral therapy for panic disorder with agoraphobia
Behaviour Research and Therapy
(2014) - et al.
Development and validation of measures of social phobia scrutiny fear and social interaction anxiety
Behaviour Research and Therapy
(1998)