Introduction
Cognitive models of social anxiety disorder (SAD) emphasize the central role of fear of negative evaluation and embarrassment in the aetiology and maintenance of the disorder (Clark and Wells, Reference Clark, Wells, Heimberg, Liebowitz, Hope and Schneier1995; Leary and Kowalski, Reference Leary, Kowalski, Heimberg and Liebowitz1995; Rapee and Heimberg, Reference Rapee and Heimberg1997). These models propose that upon entering social situations, maladaptive assumptions about themselves (i.e. as socially inadequate and unlikeable) and their social world (i.e. as inherently critical) are activated in individuals with SAD. In turn, these assumptions lead individuals with SAD to perceive even potentially innocuous social situations as dangerous. Furthermore, it is proposed that a shift in attentional resources towards a detailed monitoring of themselves follows. Consequent interoceptive cues are thought to automatically and reflexively lead to the construction of distorted mental representations of their perceived negative appearance to others. In order to avoid negative evaluation, individuals with SAD employ a variety of avoidance and safety behaviours (Clark and Wells, Reference Clark, Wells, Heimberg, Liebowitz, Hope and Schneier1995).
Experimental data provide some support for these contemporary models of SAD with particular emphasis on biased information processing (see Musa and Lépine, Reference Musa and Lépine2000, for a review). Cognitive behavioural interventions are commonly employed to treat SAD based on this cognitive-behavioural conceptualization of the disorder. Cognitive behavioural group therapy (CBGT; Heimberg and Becker, Reference Heimberg and Becker2002), which integrates cognitive restructuring and exposure techniques, is one of the most empirically investigated treatment for individuals with SAD. Randomized clinical trials have demonstrated substantive efficacy of CBGT in treating SAD relative to control conditions, with enduring effects following treatment endpoint through to follow-up (Acarturk et al., Reference Acarturk, Cuijpers, Van Straten and De Graaf2009; Canton et al., Reference Canton, Scott and Glue2012).
Although fear of negative evaluation is considered to be the central negative cognitive content in SAD, there has been increasing recognition of the importance of additional aspects to fear content identified in patients with SAD that are not traditionally conceptualized within contemporary CBT accounts or treatments. Specifically, in contrast to focus on aversive consequences for oneself, some patients with SAD also report fear content relating to the perceived negative consequences of their bodily symptoms or anxiety for others in their social environment (Takahashi, Reference Takahashi1989; Rector et al., Reference Rector, Kocovski and Ryder2006a,Reference Rector, Kocovski and Ryderb). The importance of causing discomfort to others is also captured in well-developed interpersonal models of SAD (Alden, Reference Alden, Crozier and Alden2001), cross-cultural models of SAD (Takahashi, Reference Takahashi1989; Dinnel et al., Reference Dinnel, Kleinknecht and Tanaka-Matsumi2002), and has become a defining feature of SAD in the description of the condition in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5, fifth edition; APA, 2013). Specifically, the diagnostic criteria for SAD in the recent publication of the DSM-5 (APA, 2013) have broadened the scope of defined fears of SAD to include ‘fears of being rejected by or offending others’ (APA, 2013; p. 202). Given the growing recognition of this distinct allocentric fear content to SAD, further elucidation of their specificity to SAD, cognitive correlates and treatment changes in CBGT are warranted.
Study 1: interpersonal and cultural models of SAD
Interpersonal theories of SAD (Schlenker and Leary, Reference Schlenker and Leary1982; Alden, Reference Alden, Crozier and Alden2001) extend motivations of individuals with SAD to include the desire to maintain affiliation and closeness with others while also avoiding negative evaluation. Early relationships are thought to shape an individual's maladaptive expectations of how other people will treat them interpersonally. In turn, developed maladaptive social assumptions, expectations and behavioural patterns are proposed to be maintained by eliciting confirming negative responses from others, further shaping an individual's sense of self and others (Coyne, Reference Coyne1976; Alden and Taylor, Reference Alden and Taylor2004). The dysfunctional interpersonal styles of SAD are characterized by submissiveness, suppression of emotions and interpersonal dependency (Davila and Beck, Reference Davila and Beck2002; Grant et al., Reference Grant, Gayle Beck, Farrow and Davila2007).
Cultural models of SAD similarly underscore interpersonal aspects in the onset and maintenance of the disorder. While the main concern for SAD has been conceptualized as concerns around negative evaluation in Western cultures, the principle concern among individuals experiencing social anxiety in East Asian cultures is the disruption of social harmony (Marques et al., Reference Marques, Robinaugh, LeBlanc and Hinton2011). For instance, individuals with Taijin-Kyofusho (TKS), a cultural syndrome predominantly prevalent in East Asian cultures, are preoccupied with fear of offending others through their bodily characteristics rather than embarrassing themselves in social situations (Takahashi, Reference Takahashi1989). Concerns pertaining to the perceived consequences of one's bodily symptoms (Kleinknecht et al., Reference Kleinknecht, Dinnel, Kleinknecht, Hiruma and Harada1997; Choy et al., Reference Choy, Schneier, Heimberg, Oh and Liebowitz2008) and one's anxiety (Rector et al., Reference Rector, Kocovski and Ryder2006a) for others are, however, not absent among individuals in Western cultures. Dinnel and colleagues (2012) found that irrespective of cultural background, individuals who construed themselves as high on interdependence endorsed higher TKS symptoms.
Extending this line of work, we (Rector et al., Reference Rector, Kocovski and Ryder2006a) previously presented evidence among a sample of treatment-seeking SAD patient participants that allocentric fears relevant to social anxiety are not confined to offence-related fears focused on bodily parts, but also extend to fears that one's anxiety in general would cause discomfort in others. The authors constructed a psychometrically valid unidimensional scale measuring these fears, the Social Anxiety–Discomfort to Others Scale (SA-DOS). Cross-sectional replication revealed that SA-DOS scores are significantly higher among SAD patients relative to their panic disorder with or without agoraphobia (PD/A) and student controls. Finally, the authors reported that while SAD patients receiving standard CBGT reported significantly reduced social anxiety on standard measures, SA-DOS scores remained stable across treatment. Subsequent investigations have further extended these findings, demonstrating that SA-DOS scores also account for unique variance on other dimensions of social anxiety including negative interpretation of positive social events (Laposa et al., Reference Laposa, Cassin and Rector2010) and post-event processing (Laposa and Rector, Reference Laposa and Rector2011).
It remains unclear whether allocentric SA-DOS fears should be regarded as unique to SAD or as a transdiagnostic factor relevant to other anxiety and mood disorders. For instance, studies have found that although both TKS and SA-DOS symptoms have higher correlations with social anxiety measures, they are also significantly associated with depressive symptoms (Rector et al., Reference Rector, Kocovski and Ryder2006a; Choy et al., Reference Choy, Schneier, Heimberg, Oh and Liebowitz2008). Consistent with these findings, Zhu and colleagues (Zhu et al., Reference Zhu, Yao, Dere, Zhou, Yang and Ryder2014) reported elevated levels of SA-DOS scores among Euro-Canadian and particularly among Chinese depressed psychiatric out-patients reporting social anxiety concerns. Finally, generalized anxiety disorder (GAD) is the most prevalent concurrent diagnosis for SAD (Mennin et al., Reference Mennin, Heimberg and Jack2000) and content analysis of reported worries among GAD patients demonstrate that interpersonal concerns are the most prevalent among all domains (Shadick et al., Borkovec, Reference Shadick, Roemer, Hopkins and Borkovec1991).
A related issue is the extent to which first-line psychological treatments for SAD are effective in reducing allocentric fears. In the only available study addressing treatment of SA-DOS concerns to date, Rector et al. (Reference Rector, Kocovski and Ryder2006a) reported preliminary findings that CBGT did not lead to significant changes on SA-DOS concerns amongst a small sample of patients with a principal diagnosis of SAD. The authors suggested the lack of changes may be due to the fact that standard CBGT does not explicitly outline step-by-step procedures to target and reduce this fear content. Another possibility, however, may relate to the influence of other individual differences, including cognitive vulnerabilities that may moderate treatment impact.
Study 2: cognitive moderators of SA-DOS symptom reduction in CBGT
Dysfunctional attitudes related to perfectionism/self-criticism and need for approval/dependency are considered to be central to cognitive conceptualization of SAD (Clark and Wells, Reference Clark, Wells, Heimberg, Liebowitz, Hope and Schneier1995), highlighting dysfunctional attitudes related to excessively high standards of social performance, critical social evaluation, and unconditional negative self-beliefs. Interpersonal models of SAD highlight the excessive need for acceptance, nurturance and approval of others as characteristics of SAD (Darcy et al., Reference Darcy, Davila and Beck2005; Grant et al., Reference Grant, Gayle Beck, Farrow and Davila2007).
Consistent with the above clinical conceptualizations, studies have found elevated scores on both these dimensions among individuals with SAD relative to controls (Antony et al., Reference Antony, Purdon, Huta and Swinson1998; Saboonchi et al., Reference Saboonchi, Lundh and Öst1999; Cox et al., Reference Cox, Rector, Bagby, Swinson, Levitt and Joffe2000; Davila and Beck, Reference Davila and Beck2002; Darcy et al., Reference Darcy, Davila and Beck2005; Grant et al., Reference Grant, Gayle Beck, Farrow and Davila2007). Furthermore, higher scores on both dimensions have been associated with greater levels of social anxiety (Leary, Reference Leary1983; Regev et al., Reference Regev, Shahar and Lipsitz2012; Kopala-Sibley et al., Reference Kopala-Sibley, Zuroff, Russell and Moskowitz2014; Lancu et al., Reference Lancu, Bodner and Ben-Zion2015). However, very few studies have examined the role of dysfunctional attitudes as potential predictors and/or moderators of response to treatment of SAD. Cox et al. (Reference Cox, Walker, Enns and Karpinski2002) found that even after controlling for depressive symptoms, as well as pre-treatment self-criticism and dependency scores, changes in self-criticism but not dependency predicted unique variance in change in social anxiety. Recently, Hawley et al. (Reference Hawley, Rector and Laposa2016) found that higher levels of pre-treatment perfectionism moderated (i.e. reduced) treatment outcomes to CBGT of SAD.
Notwithstanding the non-significant findings in studies to date, there are additional reasons to hypothesize a role of need for approval in moderating treatment outcome for SAD. Interpersonal and ethological theories posit that individuals with dependent SAD are often pre-occupied with themes of abandonment and threats to nurturance, affiliation and emotional security (Gilbert and Trower, Reference Gilbert, Trower, Crozier and Alden2001; Darcy et al., Reference Darcy, Davila and Beck2005; Kopala-Sibley et al., Reference Kopala-Sibley, Zuroff, Russell and Moskowitz2014). Therefore, need for approval may have particular relevance to interpersonal expressions of social anxiety such as SA-DOS concerns. Dysfunctional beliefs reflecting need for approval reflect excessive fears of being abandoned or not having needs met by others (Gilbert and Trower, Reference Gilbert, Trower, Crozier and Alden2001). In turn, SA-DOS appraisals and their dependent behavioural expressions (i.e. submissiveness and cooperation) may be directed at eliciting attachment, care and support from others (Kopala-Sibley et al., Reference Kopala-Sibley, Zuroff, Russell and Moskowitz2014).
Aims of study 1
The first study extends comparisons beyond the SAD, PD/A and student samples which were examined in Rector et al. (Reference Rector, Kocovski and Ryder2006a) to include additional anxiety disorders, OCD and a sample of mood disorder comparisons. Specifically, study 1 aimed to examine cross-sectional comparisons of the SA-DOS among a large sample with DSM-IV-TR diagnosed SAD, GAD, major depressive disorder (MDD), OCD and panic disorder with or without agoraphobia (PD/A). It was hypothesized that SA-DOS scores would be highest in the SAD group compared with all other groups, thus demonstrating the hypothesized disorder-specific associations with SAD.
Method (study 1)
Participants
Seven-hundred and forty five (n=745) participants meeting DSM-IV-TR (APA, 2000) criteria for primary SAD (n=223), GAD (n=213), PD/A (n=187), MDD (n=33) and OCD (n=89) based on the Structured Clinical Interview for Axis I Disorder (SCID-I/P version 2.0; First et al., Reference First, Gibbon, Spitzer and Williams1996) were recruited from a large university-based mental health facility and consented to participate in a clinical-research database consisting of a broad range of symptom and cognition rating scales. Demographic details of the cross-sectional sample are given in Table 1. Within each of the diagnostic groups, the majority of patients had one co-morbid psychiatric disorder with the following frequencies: MDD (n=16; 48.49%), SAD (n=88; 39.46%), GAD (n=74; 34.74%), OCD (n=31; 34.83%) and PD/A (n=58; 31.02%).
Table 1. Demographic characteristics of study 1
SD, standard deviation; §Monte Carlo estimate of Fisher's exact test excluding other/no report; athe categories here collapse achievement (i.e. none/some/completed) within the educational category; *p<.05; **p<.01.
Assessments were conducted by research staff who received extensive formal training in the administration and scoring of the SCID-I/P protocol. Prior to administration, assessors completed a rigorous inter-rater reliability-training programme. Assessors had weekly clinical case conference meetings supervised by senior psychologists with expertise in the assessment and treatment of anxiety disorders in order to establish consensus primary psychiatric diagnosis. A diagnosis was considered primary when it referred to a patient's main source of distress as determined by the SCID diagnostic severity rating and the disorder to which participants were seeking treatment. SCID-I/P interviews were performed after participants provided written informed consent.
Measures
The Social Anxiety–Discomfort to Others Scale (SA-DOS; Rector et al., Reference Rector, Kocovski and Ryder2006a) is a 14-item scale that was designed to assess fears focused on the perceived negative affective and behavioural consequences of one's anxiety on others’ well-being. Each item is rated on a 5-point Likert scale ranging from 1 (not at all) to 5 (extremely). The items refer to anxiety in general and not bodily characteristics (i.e. ‘I often think that if I don't create a comfortable environment for others, then I am to blame’). This scale has demonstrated good convergent and discriminant validity (Rector et al., Reference Rector, Kocovski and Ryder2006a). In the current study, SA-DOS demonstrated excellent internal consistency (Cronbach's α=.91).
Results (study 1)Footnote 1
Between-diagnostic group comparisons
A one-way ANOVA was performed to determine whether diagnostic groups significantly differed on age. There was no significant main effect for age (p>.05). Diagnostic groups significantly differed on proportion of gender, χ2 (4)=23.59, p<.01. Monte Carlo estimates of Fisher's exact tests revealed significant differences in proportion of categories within education, marital status and ethnicity between diagnostic groups (Fisher's exact test, all pvalues<.05).Footnote 2
A one-way ANOVA was performed to determine whether diagnostic groups significantly differed on SA-DOS scores. A significant main effect was found for diagnostic group; F (4, 740)=19.83, p<.001, µ=0.097. Consistent with recommendations for post hoc tests for unequal sample sizes (Toothaker, Reference Toothaker1993), Games–Howell post hoc tests were conducted to clarify differences between groups. Results indicated that the SAD group scored significantly higher than GAD, OCD, PD/A diagnostic groups (p <.01) but not the MDD group.Footnote 3 Furthermore, the PD/A group scored significantly lower than all other diagnostic groups except OCD (p<.01). A one-way ANOVA controlling for the most prevalent secondary diagnoses (MDD) revealed a significant main effect for diagnostic group; F (4, 554)=15.60, p<.001. Games–Howell post hoc tests indicated an identical patterning to the cross-sectional results reported above. Table 2 gives means, standard deviations and post hoc (Games–Howell) comparisons.
Table 2. Means, standard deviations and results of multiple comparison analysis of groups on the SA-DOS scores (study 1)
Values in parentheses denote standard deviation; means across rows that do not share superscript letters differ at p < .01 in the Games–Howell post hoc analysis.
Discussion (study 1)
The results of study 1 provided only partial support for the diagnostic specificity of the SA-DOS to SAD. While contrary to predictions, the finding that SA-DOS scores were equivalent between SAD and MDD groups is consistent with past research demonstrating a significant positive correlation between SA-DOS and depression (Rector et al., Reference Rector, Kocovski and Ryder2006a; Zhu et al., Reference Zhu, Yao, Dere, Zhou, Yang and Ryder2014). Furthermore, the finding may be attributable to the considerable overlapping personality and cognitive structural vulnerabilities underlying the two disorders. For example, anaclictic depression (Blatt and Zuroff, Reference Blatt and Zuroff1992), a depressive subtype focused on close and nurturing relations, and dependent subtypes of SAD (Darcy et al., Reference Darcy, Davila and Beck2005) have common elevations in dysfunctional attitudes related to need for approval and affiliation with others. Given previous findings of stable SA-DOS scores across treatment (Rector et al., Reference Rector, Kocovski and Ryder2006a), further examination of the potential role of need for approval in predicting and/or moderating treatment response of the fear dimension is warranted.
Study 2
In study 2, we sought to determine the extent to which SA-DOS fears improve following treatment with CBGT in a large treatment seeking sample of patients with SAD (n=175). Studies examining the role of perfectionism in CBGT treatment response among individuals with SAD have typically found that after controlling for pre-treatment social anxiety, changes in perfectionism, but not pre-treatment levels of perfectionism, predict treatment outcome (Cox et al., Reference Cox, Walker, Enns and Karpinski2002; Rosser et al., Reference Rosser, Issakidis and Peters2003; Ashbaugh et al., Reference Ashbaugh, Antony, Liss, Summerfeldt, McCabe and Swinson2007). However, there is a significant dearth of comparative empirical examination pertaining to the role of need for approval.
We selected the Dysfunctional Attitude Scale (DAS; Weissman and Beck, Reference Weissman and Beck1978) as a measure of dysfunctional attitudes because its interpersonal vulnerability subscale, need for approval, more appropriately resembles SA-DOS concerns relative to its counterparts (i.e. Depressive Experiences Questionnaire; DEQ-dependency; Blatt et al., Reference Blatt, D'Afflitti and Quinlan1976). Specifically, whereas the dependency scale of the DEQ primarily measures an immature dependent attitude, the need for approval dimension of the DAS is thought to measure a more mature ability to relate to others (Blatt et al., Reference Blatt, Quinlan, Pilkonis and Shea1995). Individuals high on this subscale are thought to feel genuinely saddened by the grief of losing a significant relationship (Blatt et al., Reference Blatt, Quinlan, Pilkonis and Shea1995), and are high in compliance, modesty and tender mindedness (Dunkley et al., Reference Dunkley, Blankstein and Flett1997; Bagby et al., Reference Bagby, Gilchrist, Rector, Dickens, Joffe and Levitt2001), thus more closely mirroring the interdependent aspects of SA-DOS concerns. Given that SA-DOS concerns have greater relation with and are oriented towards interpersonal and allocentric concerns relative to achievement and performance fears (Rector et al., Reference Rector, Kocovski and Ryder2006a,b), interpersonal vulnerabilities related to need for approval may have greater association with SA-DOS concerns compared with perfectionism.
In study 2, we sought to examine predicted associations between perfectionism and need for approval and treatment response. Consistent with previous literature, we hypothesized that changes in, rather than pre-treatment levels of, dysfunctional attitudes related to perfectionism would predict treatment response related to typically measured SAD-related performance fears. Furthermore, we hypothesized that changes in, rather than pre-treatment levels of, dysfunctional attitudes related to need for approval would predict treatment response related to interpersonal fears including SA-DOS and interaction fears.
Method (study 2)
Participants
One hundred and seventy-five patients meeting DSM-IV (APA, 2000) criteria for SAD based on the Structured Clinical Interview for Axis I Disorders (SCID-1/P version 2.0; First et al., Reference First, Gibbon, Spitzer and Williams1996) referred for cognitive behavioural treatment for SAD at a large university-based anxiety clinic were recruited for this study. All patients received 12 consecutive weeks of CBGT for SAD as per manualized treatment protocol at a large university-based mental health hospital consisting of psycho-education, behavioural exposures, cognitive restructuring and core belief interventions. Participant mean age was 34.32 years (SD=10.81), 50% were female, and the majority were Caucasian (83.77%) and single (67.79%).Footnote 4 Participants were excluded to enter treatment if they had substance abuse/dependence, psychosis and mania. In the current study, 70.4% of participants completed at least seven of 12 sessions, with t-tests revealing no differences in pre-treatment outcome measures between completers versus non-completers; all pvalues>.05.
Measures
The Social Anxiety–Discomfort to Others Scale (SA-DOS; Rector et al., Reference Rector, Kocovski and Ryder2006a). In the current study, SA-DOS demonstrated good internal consistency (Cronbach's α=.89).
The Social Interaction Anxiety Scale (SIAS; Mattick and Clarke, Reference Mattick and Clarke1998) is a 20-item measure that evaluates anxiety in social interactions. Each item is rated on a 5-point Likert scale ranging from 0=not at all, to 4=extremely. This scale has been shown to be reliable and valid (Mattick and Clarke, Reference Mattick and Clarke1998). In the current study, the SIAS evidenced very good internal consistency (Cronbach's α=.86).
The Social Phobia Scale (SPS; Mattick and Clarke, Reference Mattick and Clarke1998) is a 20-item scale that measures fears of being seen by others while engaging in mundane activities. Each item is rated on a 5-point Likert scale. The measure is comprised of three factors: general scrutiny concern, specific fears, and the fear of being seen as ill or as losing control (Mattick and Clarke, Reference Mattick and Clarke1998). This scale has been shown to have good validity and reliability (Mattick and Clarke, Reference Mattick and Clarke1998; Osman et al., Reference Osman, Gutierrez, Barrios, Kopper and Chiros1998). In the current study, the SPS evidenced excellent internal consistency (Cronbach's α=.93).
The Dysfunctional Attitude Scale (DAS; Weissman & Beck, Reference Weissman and Beck1978) is a 40-item self-report measure consisting of depressogenic attitudes and was designed to measure cognitive vulnerability to depression. The present study used the DAS-Form A. Each item is rated on a 7-point Likert scale. Scores on each scale range from 40 to 280, with higher scores reflecting increasing levels of depressogenic beliefs. The DAS has been shown to possess good reliability and validity (Hamilton and Abramson, Reference Hamilton and Abramson1983). Two underlying dimensions have been identified, one measuring an interpersonal vulnerability termed need-for-approval (DAS-NFA) and another measuring an achievement-related vulnerability termed perfectionism (DAS-P) (Cane et al., Reference Cane, Olinger, Gotlib and Kuiper1986; Imber et al., Reference Imber, Pilkonis, Sotsky, Elkin, Watkins and Collins1990). Beliefs include ‘my life is wasted unless I am a success’ (DAS-P) or ‘I am nothing if a person doesn't love me’ (DAS-NFA). Moderate to high correlations between the two dimensions have been reported in a number of studies (e.g. Blatt et al., Reference Blatt, Quinlan, Pilkonis and Shea1995) and the two dimensions were found to be correlated at r=.54 in this study. In the current study, the DAS-P evidenced excellent internal consistency (Cronbach's α=.90) and DAS-NFA evidenced good internal consistency (Cronbach's α=.83).
Results (study 2)
Associations between outcome measures at pre-treatment
Pearson correlation coefficients were calculated between SA-DOS, SIAS, SPS, DAS-P and DAS-NFA at pre-treatment (see Table 3). SA-DOS was significantly and positively associated with all baseline social anxiety symptoms (r=.31 to .43), all pvalues<.001. Furthermore, SA-DOS was significantly and positively associated with both DAS-P and DAS-NFA scores at baseline (p<.001). Both DAS-P and DAS-NFA were significantly and positively associated with SIAS (p<.05), but SPS was only associated with DAS-P; p<.05. However, partial correlations indicated that once controlling for BDI-II scores, SA-DOS remained positively associated with only SIAS (r=.33), p<.05; and DAS-P and DAS-NFA were no longer positively associated with SIAS and SPS, p>.05.
Table 3. Pearson correlations, means and standard deviations for variables at pre-treatment (study 2)
SA-DOS: Social Anxiety and Fear of Causing Discomfort to Others Scale; SIAS: Social Interaction Anxiety Scale; SPS: Social Phobia Scale; DAS-P: Dysfunctional Attitudes Scale – Perfectionism; DAS-NFA: DAS-Need For Approval. *p <.05; **p <.001.
Changes across CBGT on symptom measures, DAS-P and DAS-NFA
Means, standard deviations and effect sizes for each outcome measure at pre- and post-treatment are shown in Table 4. Effect sizes were computed as per Cohen (Reference Cohen1988), with recommended adjustments (Morris and DeShon, Reference Morris and DeShon2002). Paired t-tests were used to determine significant reductions pre–post treatment on each outcome measure.Footnote 5 The alpha rate was Bonferroni-corrected at p = 0.01 to adjust for multiple comparisons. Results indicated that scores on SIAS (t (152)=10.35) and SPS (t (161)=10.07), were significantly reduced pre- to post-treatment, all pvalues < .001, with large mean within-group effect sizes (d=.80 to .83). Furthermore, DAS-P (t (161)=3.64) and DAS-NFA (t (161)=3.06) significantly decreased following treatment, all pvalues < .01. Finally, SA-DOS scores also evidenced significant change pre–post treatment (t (174)=7.60; p<.001). Both DAS subscales and SA-DOS evidenced small to medium mean within-group effect sizes (d=.24 to .58).
Table 4. Pre-treatment and post-treatment scores for CBGT group analyzed with paired-sample t-tests (study 2)
SA-DOS: Social Anxiety and Fear of Causing Discomfort to Others Scale; SIAS: Social Interaction Anxiety Scale; SPS: Social Phobia Scale; DAS-P: Dysfunctional Attitudes Scale – Perfectionism; DAS-NFA: DAS-Need For Approval. *p<.01 (α adjusted; 0.05/5); Cohen's d (Cohen, Reference Cohen1988) effect size adjusted using Morris and DeShon's (2002) procedure.
Perfectionism and need for approval as treatment moderators to CBGT for SAD
A series of stepwise regression analyses were conducted to address whether pre-DAS-P and pre-DAS-NFA predicted post-treatment social anxiety symptom scores. For post-SA-DOS, the overall model was significant, F (2,165)=5.08, p<.01, R 2=.06. DAS-NFA was a trended significant predictor of post-treatment scores on SA-DOS scores (β=.18), p=.05, but was no longer significant once pre-SA-DOS symptoms were hierarchically entered prior to pre-DAS variables. Regression models were not significant for post-SIAS and post-SPS, all pvalues>.05.
Changes in perfectionism and need for approval as predictors to CBGT outcome for SAD
A series of hierarchical regression analyses were conducted to determine unique and cumulative effects of pre-treatment social anxiety symptom and standardized residualized (res) DAS-P and DAS-NFA change scores on SA post-treatment symptom scores (see Table 5). Res scores for variables were calculated as regression analysis with pre-treatment scores as independent variables and post-treatment scores as dependent variables. At step 1, pre-treatment (pre) social anxiety symptoms were entered. At step 2, resDAS-P and resDAS-NFA were entered stepwise for regression except for post-SPS regression where routine entry was used for reasons of model parsimony. For post SA-DOS, the overall model of the first block was significant, F (1,160)=71.17, p<.001, R2 =.31. Stepwise entry of resDAS variables in the second block led to significant change, F Δ (1,159)=31.32, p<.001; R 2 Δ=.11. The final model was significant, F (2,159)=57.99, p<.001, R 2=.42 with preSA-DOS symptoms (β=.53) and res-DAS-NFA (β=.33), as the significant predictors, all pvalues<.001.Footnote 6 For resSA-DOS, preDAS-NFA was entered at step 1 with preSA-DOS. The overall model of the first block was not significant, p>.05. Stepwise entry of resDAS variables in the second block led to significant change, F Δ (1,158)=32.18, p<.001; R 2 Δ=.17. The final model was significant, F (3,158)=11.59, p < 0.001, R 2=.18 with resDAS-NFA (β=.41) as the lone significant predictor, p<.001.
Table 5. Hierarchical regressions for social anxiety symptoms measures (study 2)
SA-DOS: Social Anxiety and Fear of Causing Discomfort to Others Scale; SIAS: Social Interaction Anxiety Scale; SPS: Social Phobia Scale; DAS-P: Residualized Dysfunctional Attitudes Scale – Perfectionism; DAS-NFA: Residualized DAS-Need For Approval. * p<.05; **p<.01; ***p<.001; pre = pre-treatment; pst=post-treatment.
For post SIAS, the overall model of the first block was significant, F (1,140)=72.29, p<.001, R 2=.34. Stepwise entry of resDAS variables in the second block led to significant change, F Δ (1,139)=11.80, p<.001; R 2 Δ=.05. The final model was significant, F (2,139)=44.84, p<.001, R 2 =.39 with preSIAS symptoms (β=.57) and resDAS-NFA (β=.23), as the significant predictors, all pvalues<.01. For postSPS, the overall model of the first block was significant, F (1,148)=118.38, p<.001, R 2=.44. resDAS variables entered in the second block led to significant change, F Δ (2,146)=15.25, p<.001, R 2 Δ=.10. The final model was significant, F (3,146)=57.22, p<.001, R 2=.54 with preSPS symptoms (β=.67), resDAS-NFA (β=.20), resDAS-P (β=.14) as the significant predictors, all pvalues<.05.
Within participants who achieved clinical remission based on SIAS of 34 or less (31%), both t-tests and Mann–Whitney U tests revealed significantly lower SA-DOS scores for those that achieved remission relative to those that did not, all p values<.05. Further, pre–post change scores on the SA-DOS were significantly greater among those that achieved clinical remission relative to those that did not using both t-tests and Mann–Whitney U tests, all p values<.05. Pearson's correlation also indicated resSA-DOS as positively correlated with all residualized change scores on all outcome measures, all p values<.05.
General discussion
Emerging evidence of the relevance of allocentric fears to the experience of social anxiety (i.e. Dinnel et al., Reference Dinnel, Kleinknecht and Tanaka-Matsumi2002; Rector et al., Reference Rector, Kocovski and Ryder2006a,b) has warranted expansion of the diagnostic criteria for SAD to include this fear content (APA, 2013). To extend this line of inquiry, study 1 sought to determine the diagnostic specificity of SA-DOS concerns to SAD among a large sample of DSM diagnosed anxiety and MDD participants. The results were partially supportive of predicted diagnostic specificity of SA-DOS concerns to individuals with SAD. Specifically, while SA-DOS concerns were significantly elevated among individuals with SAD relative to individuals with GAD, OCD and PD/A, equivalent scores were found for individuals with MDD. Furthermore, the equivalent scores were not attributable to secondary SAD among MDD patients.
While contrary to expectations, one potential source of equally elevated scores on the SA-DOS between SAD and MDD participants may relate to the shared cognitive and affective profiles between the two disorders. For instance, dependent subtypes of both depression (i.e. anaclitic; Blatt and Zuroff, Reference Blatt and Zuroff1992) and SAD (Darcy et al., Reference Darcy, Davila and Beck2005) have been described in the literature. Sanz and Avia (Reference Sanz and Avia1994) found little discriminating power of the DAS between social anxiety and depression. Furthermore, affective profiles (negative affectivity and low positivity) previously hypothesized to be specific to depression (i.e. tripartite model of anxiety and depression; Clark and Watson, Reference Clark and Watson1991) also characterize social anxiety (Kashdan, Reference Kashdan2007). Specifically, both disorders are marked by tendencies to suppress or express emotions (Kashdan et al., Reference Kashdan, Elhai and Breen2008). Inhibiting emotional experience and expression may serve as a strategy to minimize negative social outcomes such as causing discomfort to others among individuals with SAD and MDD.
Results of our second study demonstrated that CGBT led to significant reductions on not only standard measures of SAD, but also dysfunctional attitudes and SA-DOS concerns, although effect sizes for the latter were medium. The effect sizes found in our study (d=.80 to .83) for SPS and SIAS are consistent with those found in previous studies (Cox et al., Reference Cox, Ross, Swinson and Direnfeld1998; Heimberg et al., Reference Heimberg, Liebowitz, Hope, Schneier, Holt and Welkowitz1998). The joint findings of nearly equivalent effect sizes between DAS-P and DAS-NFA, as well as the significantly greater reduction of SIAS and SPS relative to SA-DOS across treatment, suggests that SA-DOS reflect more enduring sources of interpersonal threat. Therefore, the findings highlight the importance of addressing a unique fear dimension that is not incorporated in regnant models and treatment of SAD. Rector and colleagues (Reference Rector, Kocovski and Ryder2006a,b) have provided an overview of treating SA-DOS concerns by recommending augmenting existing cognitive-behavioural interventions for SAD (Heimberg and Becker, Reference Heimberg and Becker2002; Wells, Reference Wells1997) including strategies to empirical test the impact of patient anxiety on others.
As predicted, changes in DAS-P and DAS-NFA, rather than their pre-treatment scores, predicted treatment response, even after controlling for pre-treatment levels of social anxiety and depression. Findings pertaining to perfectionistic dysfunctional attitudes and treatment response are consistent with previous studies (Cox et al., Reference Cox, Walker, Enns and Karpinski2002; Rosser et al., Reference Rosser, Issakidis and Peters2003; Ashbaugh et al., Reference Ashbaugh, Antony, Liss, Summerfeldt, McCabe and Swinson2007). A more novel contribution to the literature, however, were the findings that consistent with hypotheses, changes in DAS-NFA were significantly associated with treatment response in interpersonal expression of social anxiety including SA-DOS concerns, whereas changes in DAS-P were only related to performance anxiety. However, contrary to prediction, both changes in DAS-P and DAS-NFA were associated with treatment response in SPS. This is, however, similar to previous findings by Robins et al. (Reference Robins, Bagby, Rector, Lynch and Kennedy1997) in a depressed clinical sample. The findings further lend credence to the importance of targeting deeper structural vulnerabilities to psychopathology in addition to manifest symptoms and appraisals as echoed by experts in cognitive therapy (Beck et al., 1979; Padesky, Reference Padesky1994; Rector et al., Reference Rector, Bagby, Segal, Joffe and Levitt2000). In the current study, we found only partial support for the role of pre-treatment levels of DAS-NFA in predicting treatment outcome, and their perfectionistic counterparts did not predict treatment outcome. Finally, the findings suggesting that changes in SA-DOS were related to clinical remission is particularly encouraging.
The identification of changes in DAS-NFA as having specific relevance to changes in interpersonal expression of social anxiety across treatment among SAD patients undergoing CBGT is noteworthy. The association between DAS-NFA and SA-DOS is consistent with the nomological relations found in the personality literature. For example, sociotropy, a dimension closely resembling DAS-NFA (Blaney and Kutcher, Reference Blaney and Kutcher1991), is significantly correlated with the agreeableness facets of compliance, modesty and tendermindedness, and inversely correlated with the assertiveness facet of extraversion (Dunkley et al., Reference Dunkley, Blankstein and Flett1997; Bagby et al., Reference Bagby, Gilchrist, Rector, Dickens, Joffe and Levitt2001). Additionally, patients with SAD present in the ‘high’ range of the tendermindedness facet of the agreeableness domain of the NEO-PI-R (Rector, Reference Rector, Kocovski and Ryder2006a; Rector et al., Reference Rector, Bagby, Huta and Ayearst2012). In turn, the latter was significantly negatively correlated with the assertiveness facet scores of the extraversion domain.
Interpretation of findings here should be tempered by several limitations. First, studies neither included a healthy control sample nor a treatment comparison group. As such, the results are insufficient to draw conclusions regarding diagnostic specificity and isolating effects as a result of treatment per se. It is also unclear whether the associations found between cognitive change and treatment outcome and change in study 2 are specific to cognitive interventions (specificity hypothesis; Garratt et al., Reference Garratt, Ingram, Rand and Sawalani2007). Furthermore, we cannot preclude the possibility that the changes in outcomes may have been due to the passage of time. Another limitation is that we did not control for higher-order personality factors (i.e. neuroticism) when testing the predictive contributions of dysfunctional attitudes on treatment change and outcome. It is unclear whether results would have held after controlling for potential confounding variables. Because of the association between need for approval and tendencies towards social desirability (Uziel, Reference Uziel2010), self-reported limitations of postSA-DOS scores as reflecting this motivation should be considered when interpreting study findings. The design of study 2 prohibited us from conducting mediational analyses, limiting the interpretation of dysfunctional attitudes. Burns and Spangler (Reference Burns and Spangler2001) found that dysfunctional attitudes as measured by the DAS did not mediate treatment changes in anxiety or depression. Rather, the results suggested a mediated effect through an unknown third variable. Causal conclusions regarding DAS have not been established in the current and previous studies (Tryon, Reference Tryon2014).
In conclusion, the current study demonstrated relevance of SA-DOS concerns to both anxiety and mood disorders, particularly among SAD and MDD patients. Furthermore, the current study also established that while SA-DOS concerns did change across treatment, the effects are relatively small. In addition to targeting surface cognitions and appraisals related to fear of causing discomfort to others, the results also suggest that treatment providers might also do well by targeting deeper dysfunctional attitudes related to need for approval to address SA-DOS concerns among SAD patients. However, in line with cross-cultural treatment recommendations for SAD, we uphold the caution that because tendermindedness, low assertiveness, and cooperativeness are culturally valued behaviours, it remains important not to over-pathologize these behaviours (Iwamasa, Reference Iwamasa and Fredman1997; Rector et al., Reference Rector, Kocovski and Ryder2006a,b).
Acknowledgements
The authors wish to thank Dr Michelle Leybman and Bethany Lerman for their editorial assistance with and comments on earlier drafts of this manuscript as well as Jane Yating Ding for assistance with data management.
Conflicts of interest: Yasunori Nishikawa, Judith M. Laposa, Rotem Regev and Neil A. Rector have no conflicts of interest with respect to this publication.
Comments
No Comments have been published for this article.