Behav Res
Ther. Author manuscript; available in PMC 2013 May 1. Published in final edited form as: PMCID: PMC3327793 NIHMSID: NIHMS360462
Abstract
Beliefs that are negatively biased, inaccurate, and rigid are thought to play a key role in the mood and anxiety disorders. Our goal in this study was to examine whether a change in maladaptive beliefs mediated the outcome of individual cognitive-behavioral therapy [CBT] for social anxiety disorder [SAD]. In a sample of 47 individuals with SAD receiving CBT, we measured maladaptive interpersonal beliefs as well as emotional and behavioral components of social anxiety, both at baseline and after treatment completion. We found that [a] maladaptive interpersonal beliefs were associated with social anxiety at baseline and treatment completion; [b] maladaptive interpersonal beliefs were significantly reduced from baseline to treatment completion; and [c] treatment-related reductions in maladaptive interpersonal beliefs fully accounted for reductions in social anxiety after CBT. These results extend the literature by providing support for cognitive models of mental disorders, broadly, and SAD, specifically.
Keywords: maladaptive beliefs, core beliefs, belief change, social anxiety, cognitive models
Cognitive models of mood and anxiety disorders posit that maladaptive beliefs that are central to one’s identity and are negatively biased, inaccurate, and rigid [e.g., “core beliefs”] play a causal role in generating the emotional disturbance that characterizes these disorders [Beck & Dozios, 2011]. Empirical research has provided evidence that these types of maladaptive beliefs are associated with psychopathology and are generally stable [Riso, du Toit, Stein, & Young, 2007] but sometimes vary with the severity of symptomatology [e.g., Lewinsohn, Steinmetz, Larson, & Franklin, 1981]. Many studies have also provided evidence for a potentially causal role of maladaptive beliefs in mood and anxiety disorders [e.g., Alloy, Abramson, Whitehouse, Hogan, Panzarella, & Rose, 2006] though this evidence is mixed [Ingram, Miranda, & Segal, 1998; Jarrett, Vittengl, Doyle, & Clark, 2007]. Our goal in this study was to examine the role of maladaptive belief change in cognitive-behavioral treatment for social anxiety disorder [SAD].
Contemporary cognitive models of SAD [e.g., Heimberg, Brozovich, & Rapee, 2010; Rapee & Heimberg, 1997] posit that individuals with SAD have maladaptive beliefs regarding themselves [as socially incompetent] and others [as critical judges]. When activated in a social situation, these maladaptive cognitions transform innocuous social cues [e.g., another person looking away during a conversation] into significant social threats. Whereas many studies have documented the role of less stable, surface-level maladaptive cognitions, such as appraisals, attributions, and thoughts in SAD [e.g., Schulz, Alpers, & Hofmann, 2008; Stopa & Clark, 1993], only five studies have investigated the role of maladaptive beliefs in SAD [Anderson, Goldin, Kuria, & Gross, 2008; Ball, Otto, Pollack, Uccello, & Rosenbaum, 1995; Pinto-Gouveia, Castilho, Galhardo, & Cunha, 2006; Rapee, Gaston, & Abbott, 2009; Wenzel, 2004].
Changing maladaptive cognitions is a primary focus of the best researched and most widely supported psychotherapeutic treatment for SAD, cognitive-behavioral therapy [CBT; Beck & Dozios, 2011]. However, studies investigating maladaptive beliefs in the context of CBT [DeRubeis et al., 1990; Jarrett et al., 2007] have generally been conducted in the context of depressive disorders, except for a study by Hofmann and colleagues [2007] who studied patients with panic disorder, and a study by Rapee and colleagues [2009], who studied patients with SAD. Rapee and colleagues [2009] compared maladaptive beliefs related to negative representations of appearance and performance [e.g., “I look attractive”] and “core” beliefs related to SAD [e.g., “I am dumb/stupid”] in patients receiving an enhanced cognitive-behavioral treatment, a standard cognitive-behavioral treatment, or stress management. Patients receiving the enhanced treatment demonstrated significantly greater decreases in negative representations than patients in the other two conditions, whereas changes in “core” beliefs did not significantly differ by condition. Furthermore, decreases in negative representations, together with decreases in patients’ estimates of cost associated with negative evaluations, partially mediated the link between treatment condition and reductions in diagnostic severity.
In addition to over-endorsement of maladaptive beliefs, several additional studies have investigated whether other types of cognitive biases associated with SAD are associated with SAD change during cognitive or cognitive-behavioral therapy for SAD and whether such changes are associated with or underlie treatment gains [Bruch, Heimberg, & Hope, 1991; Foa, Franklin, Perry, & Herbert, 1996; Hofmann, 2004; McManus, Clark, & Hackmann, 2001]. These studies have generally demonstrated that overestimation of the probability of occurrence of negative social events [e.g., “Someone you know won’t say hello to you”] and the costs associated with occurrence of these events [i.e., how bad it would be if the event happened] mediated the treatment gains for patients with generalized SAD. A final study by Vögele and colleagues [2010] provided evidence of cognitive mediation of symptom reduction among SAD patients, albeit in the context of high density exposure therapy over the course of 4–10 days, rather than CBT or cognitive therapy.
The aim of the present study was to examine the relations between maladaptive beliefs and SAD symptoms by investigating whether belief change accounts for the effects of CBT in the treatment of SAD. Based on several preliminary studies [Anderson et al., 2008; Pinto-Gouveia et al., 2006; Rapee et al., 2009; Wenzel, 2004], we conceptualized maladaptive beliefs as unhelpful evaluative cognitions related to the self in the domain of interpersonal interactions [henceforth referred to as “maladaptive interpersonal beliefs”]. Using a measure we developed for this purpose, the current research tested three hypotheses: [a] maladaptive interpersonal beliefs are related to social anxiety severity among individuals with SAD; [b] CBT changes these beliefs; and [c] belief change mediates the impact of treatment. To test the third hypothesis, we investigated relations between treatment-related changes in maladaptive interpersonal beliefs and social anxiety among participants receiving immediate and delayed CBT.
Method
Participants
Participants were 47 adults [Mage = 32.9, SDage = 9.2; 51.1% female] meeting DSM-IV [American Psychiatric Association, 1994] criteria for a principal diagnosis of generalized social anxiety disorder [SAD]. Diagnoses were based on the Anxiety Disorders Interview Schedule for the DSM-IV-Lifetime version [ADIS-IV-L; DiNardo, Brown, & Barlow, 1994]. A small percentage of participants met criteria for current comorbid conditions, including generalized anxiety disorder [14.9%], dysthymia [4.2%], panic disorder [4.2%], and specific phobia [4.2%]. The majority of participants reported their race/ethnicity as Caucasian [57.4%], followed by Asian [23.4%], Hispanic [10.6%], Multiracial [4.3%], African-American [2.1%], and Pacific Islander [2.1%].
Because this study was part of a larger neuroimaging study of the effects of CBT, participants were screened to ensure they [a] did not exhibit any lifetime evidence of psychosis, mania, hypomania, or bipolar disorder; [b] did not use selective serotonin reuptake inhibitors, benzodiazepines, beta-blockers, anti-psychotics, blood thinners, thyroid hormone influencing agents, diabetic medications, or anticonvulsants within three months prior to entering the study; and [c] had not previously received a full-dose [at least 8 sessions] of CBT for any mood or anxiety disorder. Eighty-four potential participants met study inclusion criteria and were invited to participate. Nine participants discontinued prior to the randomization phase [e.g., failed to complete all the baseline assessments]. A total of 75 patients agreed to complete the study and were randomly assigned to receive immediate CBT or serve as a waitlist control, after which they received CBT [i.e., delayed CBT condition]. Of these participants, 47 [immediate treatment condition, n = 20; delayed CBT condition, n = 27] completed both measures included in this study at baseline and following completion of waitlist and/or treatment, and thus, were included in this study.
The 47 participants included in this study did not differ from those randomized but not included in terms of age, gender, ethnicity, the presence of current or past Axis 1 psychopathology [other than SAD], or past psychotherapy or pharmacotherapy. Furthermore, treatment groups [immediate vs. delayed] did not differ in age, gender, ethnicity, the presence of current or past Axis-I psychopathology [other than SAD], or past psychotherapy, but did differ in past pharmacotherapy, with more participants in the delayed [38.4%] versus immediate [10.0%] group having received past pharmacotherapy [p < .05].
Procedure
Participants were recruited through use of electronic bulletin-board listings and referrals from mental health clinics. After passing an initial telephone screening, potential participants were administered the ADIS-IV-L [DiNardo et al., 1994] by a clinical psychologist [PG or KW], who was trained using video recordings and test cases. Previous research has demonstrated that the ADIS-IV-L demonstrates high reliability for diagnosing SAD [Brown, DiNardo, Lehman, & Campbell, 2001]. Individuals with a principal diagnosis of generalized SAD [operationalized as a score of 4 or more on the ADIS-IV-L Clinician’s Severity Rating for SAD and ratings for 4 or more for 5 or more distinct social situations; all scales 0–8] were enrolled. All participants were asked to refrain from seeking additional treatment while receiving CBT. The study protocol was approved by the Stanford University Institutional Review Board [IRB], and all participants provided informed consent in accordance with IRB regulations.
Intervention
Individual CBT was administered by Ph.D. level psychotherapists trained according to a manualized protocol [see Hope, Heimberg, Juster, & Turk, 2000; Hope, Heimberg, and Turk, 2006], with on-going, on-site-training and assessment of treatment fidelity provided by the developer of the protocol [RGH]. CBT was provided in 16 one-hour weekly sessions in which cognitive restructuring and exposure, both in-session and in vivo, were used to reduce theoretically identified causes and symptoms of SAD.
Measures
Liebowitz Social Anxiety Scale – Self-Report [LSAS-SR, Fresco et al., 2001]
We measured emotional and behavioral aspects of social anxiety with the self-report version [LSAS-SR] of the widely used clinician-administered LSAS [Heimberg et al., 1999; Liebowitz, 1987]. The LSAS and LSAS-SR include an identical set of 24 items that assess the degree to which social interactions [e.g., “Talking to people in authority”] and performance situation [“Writing while being observed”] separately evoke emotions [i.e., fear/anxiety] and behavior [i.e., avoidance] associated with SAD. Participants rate items on a 4-point scale indicating the degree to which they experienced fear/anxiety [0 = None to 3 = Severe] and avoidance [0 = Never [0%] to 3 = Usually [68–100%]] in response to each situation during the past week. We summed across all items to obtain a total score. The LSAS-SR has been shown to be reliable and have good convergent validity [e.g., Baker et al., 2002; Fresco et al., 2001] and demonstrated excellent internal reliability among the total SAD sample at intake and post-treatment [Cronbach’s α = .91, .97].
Maladaptive Interpersonal Belief Scale [MIBS]
Our initial step in developing the MIBS was to identify evaluative interpersonal beliefs relevant to SAD. We derived 16 beliefs items by: [1] noting common, relevant beliefs of patients with SAD; [2] consulting with experts and clinicians who treated patients with SAD; and [3] reviewing the literature on SAD. To establish the clinical significance of these items, we asked an expert panel of six clinical psychologists with extensive experience in research and/or treatment of SAD to provide ratings of the extent to which a particular item would be relevant to someone with SAD. All experts regarded six beliefs as “extremely relevant” [rather than “somewhat relevant”, or “not at all relevant”] in this regard. Because these six items were all keyed in a direction indicating increased endorsement of a given belief, we added five reverse-keyed belief items to create a balanced set of items that would not be subject to acquiescence response bias [e.g., Soto & John, 2009].
We tested whether these beliefs formed an internally consistent set by administering the initial 11 belief items to a sample of healthy adults who did not currently meet DSM-IV criteria for any Axis I or II disorders [n = 42; 52% male; M age = 32.9 years, SD = 9.2; 52.4% Caucasian], and college students [n = 198; 39% male; M age = 20.9 years, SD = 2.5; 31% Caucasian]. Participants indicated their agreement or disagreement with each belief item by providing a rating from 1 [“Definitely false” or “Strongly disagree”] to 5 [“Definitely true” or “Strongly Agree”]. In each sample, we conducted an exploratory factor analysis of the 11 items, to investigate whether items together comprised a single, general factor. In both samples, nine items had substantial and consistent loadings on a single general factor, whereas two [“I must do everything perfectly”; “Others' opinions of me matter a great deal”] did not. We excluded these two items from further consideration. The nine remaining items [see Appendix] all had a substantial loading on the first unrotated factor in both samples [with mean factor loadings of .61 for healthy adults and .67 for college students], and there was no evidence for a second factor. Internal reliability was good for the 9-item measure both in the healthy adults sample [α = .77], and in the college student sample [α = .85].
Similar to the healthy adults and college students samples, in our SAD sample we also found that all items had consistent loadings on the general factor [mean factor loading = .66]. Internal reliability for the 9-item measure in the SAD sample was good both at baseline and at post-treatment [Cronbach’s α = .85, .94]. Test-retest reliability over 5-months for the 27 SAD participants who were randomized to the delayed treatment condition was excellent [r =.85].
An analysis of variance [ANOVA] with planned comparisons revealed that, as expected, the mean for SAD participants [M = 3.4, SD = 0.6] was significantly higher than the mean of the healthy adult participants [M = 1.8, SD = 0.4] and college student participants [M = 2.3, SD = 0.6; Omnibus F [2, 286] = 83.8, p