Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Psychological Group-Treatments of Social Anxiety Disorder: A Meta-Analysis

  • Hanna Wersebe,

    Affiliation Department of Clinical Psychology, VU University, Amsterdam, The Netherlands

  • Marit Sijbrandij,

    Affiliations Department of Clinical Psychology, VU University, Amsterdam, The Netherlands, EMGO Institute for Health and Care Research, Amsterdam, The Netherlands

  • Pim Cuijpers

    Affiliations Department of Clinical Psychology, VU University, Amsterdam, The Netherlands, EMGO Institute for Health and Care Research, Amsterdam, The Netherlands, Leuphana University Lüneburg, Lueneburg, Germany


6 Dec 2013: Wersebe H, Sijbrandij M, Cuijpers P (2013) Correction: Psychological Group-Treatments of Social Anxiety Disorder: A Meta-Analysis. PLOS ONE 8(12): 10.1371/annotation/5f2f7ff4-ecfd-4a41-a162-34c1dd0c962a. View correction



A few meta-analyses have examined psychological treatments for a social anxiety disorder (SAD). This is the first meta-analysis that examines the effects of cognitive behavioural group therapies (CBGT) for SAD compared to control on symptoms of anxiety.


After a systematic literature search in PubMed, Cochrane, PsychINFO and Embase was conducted; eleven studies were identified that met the inclusion criteria. The studies had to be randomized controlled studies in which individuals with a diagnosed SAD were treated with cognitive-behavioural group therapy (CBGT) and compared with a control group. The overall quality of the studies was moderate.


The pooled effect size indicated that the difference between intervention and control conditions was 0.53 (96% CI: 0.33–0.73), in favour of the intervention. This corresponds to a NNT 3.24. Heterogeneity was low to moderately high in all analyses. There was some indication of publication bias.


It was found that psychological group-treatments CBGT are more effective than control conditions in patients with SAD. Since heterogeneity between studies was high, more research comparing group psychotherapies for SAD to control is needed.


Both population surveys [1] and studies [2]; [3]; [4] indicate that Social Anxiety Disorder (SAD) is a highly prevalent psychiatric disorder. The National Comorbidity Survey-Replication [1] for example, provides prevalence estimates of 12-month and lifetime DSM-IV SAD of 6.8% and 12.1%, respectively. SAD is defined by a marked fear of social performance situations and to possible scrutiny by other people [5]. As a result, phobic situations may be avoided or only endured with intense anxiety, and a pronounced distress and impairment in daily occupational and social life results. A diagnosis of SAD has been found to be associated to scholastic difficulties, as these individuals are more likely not to finish high school and to fail a grade [6]. Accordingly, SAD is related to reduced quality of life [7], impaired functioning and high (economic) costs [8]. However, SAD is still one of the least well-recognized mental disorders [e.g. 9] and frequently under-diagnosed [10].

To decrease the burden of patients various treatments have been developed. One of the most broadly researched and applied treatments for SAD is cognitive-behavioural therapy (CBT). Cognitive-behavioural therapies typically include a vast range of techniques, such as exposure to social stimuli and tasks and cognitive restructuring [11]. During exposure, the client is exposed to feared (social) situations despite experiencing distress. Cognitive restructuring is applied to correct maladaptive beliefs about the self and others. In addition, social skills training and applied relaxation are commonly used in treatment of SAD. Further, all techniques involve repeated practice both in form of homework and in the therapeutic setting.

Various past studies have supported the efficacy of cognitive-behavioural group therapy (CBGT) in diverse populations [12]; [13]; [14]; [15]; [16]; [17]. In CBGT, the therapy is carried out in a group setting. Group treatments, where patients with the same disorder are treated at the same time, may have certain benefits since social situations can be simulated more easily, mutual support might be given and exposure to social situations may occur naturally. Also, group therapies may be more cost-effective than individual therapies since they take less therapist time per patient [18].

Until now, only a few meta-analyses have examined the effects of psychological treatments for SAD. Specifically, five meta-analyses have been conducted in the last two decades that exclusively examined effects of psychotherapy on SAD [19]; [20]; [21]; [22]; [23]. Most of them were hampered by methodological limitations since, for example, only a few high-quality randomized controlled trials (RCTs) were available at the time they were conducted [20]; [21]; [22]; [23]. That could have resulted in an inflated effect size and therefore in a distorted overall picture. Although all previous meta-analyses report controlled effect sizes, which are known to be more conservative, only the meta-analysis of Acarturk et al. [19] applied tests for heterogeneity of included primary studies. Those state-of-the-art analyses are crucial in order to find differences between studies and their specific impact on the outcome of interest. Moreover, they can give an indication of treatment effects in specific population.

Unlike the earlier meta-analyses [20]; [21]; [22]; [23], Acarturk et al. [19] included only RCTs on the available psychological treatments for SAD. Earlier findings of the abovementioned meta-analyses could be replicated with the asset of detailed analyses of subgroups. Furthermore, all meta-analyses on SAD have combined both studies on individual and group treatment formats and no study until now had explicitly focused on CBGT contrasted to a control group only. Group-treatments and their particular effect-sizes are crucial to investigate as the limited resources of health care providers and their constant cutbacks on expenses for psychotherapies engage interest in both the most economical and effective treatment at the given point of time.

Therefore, the aim of our study was to investigate the effectiveness of CBGT based on the most recent published trials compared to waiting-list, placebo or treatment-as-usual (TAU) conditions in adults with SAD on reducing social anxiety symptoms. Hence, it is expected that CBGT is more effective in reducing social anxiety symptoms than control.


Study Eligibility

Studies were included when (1) the effects of group-treatments (2) in subjects aged 18 years or older (3) with a principal diagnosis of SAD (4) were compared with a control condition (5) in an RCT.

Moreover, studies had to define SAD either according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R [24] or DSM-IV [5]) or contain a fixed cut-off score on clinician-rated SAD questionnaire (Table 1). Control conditions could be treatment-as-usual (TAU), pill-placebo or waiting list. Studies were excluded, when the standardized mean difference could not be calculated (usually the case when no conducted statistical test examined the difference between the psychotherapy and the control condition).

Search Strategy

Studies were retrieved through systematic literature searches (from 1980 to 22 March 2012) in the databases of PubMed, Cochrane Central Register of Controlled Trials, PsychInfo and embase. Searches were conducted by means of databased keywords and text words indicative of SAD, while limiting the search to effect studies (randomized trials, controlled trials, clinical trials). No language restrictions were applied. The complete search strategy is available in the additional file A. Search Strings and results in the Appendix S1.

First, inclusion criteria were applied to titles and abstracts. Potential eligible studies were retrieved in full text. Second, criteria were applied to full articles. Two independent reviewers carried out the data extraction. Disagreement was solved in discussion and consensus.

Quality assessment

The validity and quality of the studies was assessed with the Cochrane risk of bias tool [25]. This tool allows to determinate the following four possible sources of bias in RCTs: 1. Adequacy of sequence allocation; 2. Allocation concealment; 3. Blinding of assessors and outcomes; 4. Incomplete outcome data (intention-to-treat analyses); as well as selective reporting and other biases. Two independent reviewers assessed the quality of the studies and disagreement was solved by discussion.


For each comparison between a psychological treatment and a control group, the effect size indicating the difference between the two groups at post-test (Hedges' g) and the 95% Confidence Intervals (CI's) were calculated. Hedges' g was chosen as the present meta-analysis includes several studies with a small sample size and this effect size adjusts for such small sample sizes. Effect sizes were calculated by subtracting (at post-test) the average score of the psychological treatment group from the average score of the comparison group, and dividing the result by the pooled standard deviations of the two groups. An effect size of 0.5 thus indicates that the mean of the experimental group is half a standard deviation larger than the mean of the control group. Effect sizes of 0–0.41 can be assumed to be small, of 0.40 to 0.70 moderate and 0.70 and above as a large effect [25].

Only those scales and measurements that explicitly measured symptoms of anxiety were used in the calculation of the effect sizes. If more than one measure was used to assess the change of SAD, the pooled effect size was calculated, so that each study only provided one effect size. When means and standard deviations were not reported, p-values were used to calculate effect sizes.

The Comprehensive Meta-Analyses computer program (CMA; version 2.2.021) was used to calculate the pooled effect size. Since considerable heterogeneity among the studies was expected, the pooled effect sizes were calculated with the random effects model. In the random effects model it is assumed that the included studies are drawn from ‘populations’ of studies that differ from each other systematically (heterogeneity). In this model, the effect sizes resulting from included studies not only differ because of random error within studies (as in the fixed effects model), but also due to true variation in effect size from one study to the next. Statistical effect sizes are generally not easy to comprehend and interpret from a clinical point of view. For this reason, the effect sizes were transformed into the numbers-needed-to-be-treated (NNT) by using the formulae provided by Kraemer & Kupfer [26]. The NNT indicates the number of patients that have to be treated in order to yield one additional positive outcome in one of them [27].

To assess heterogeneity between studies the I2 statistic was calculated, with a value of 0% indicating no heterogeneity, 25% low, 50% as moderate and 75% high heterogeneity [28].

Also, the Q- statistic was calculated to assess the level of homogeneity. A significant p-value indicates that the null-hypothesis of homogeneity has to be rejected and it is likely that the heterogeneity is not due to sample error.

Subgroup analyses were conducted according to the mixed effects model. In this model, studies within subgroups are pooled with the random effects model, which uses random effects within subgroups and fixed effects between subgroups.

Publication bias was tested by inspecting the funnel plot, and by means of Duval and Tweedie's trim and fill procedure [29], which gives an estimate of the effect size after the publication bias has been taken into account. Further, the Begg & Mazumdar rank correlation test [30] was applied to test whether the adjusted and observed effect sizes differed significantly from each other.


In Figure 1, a flowchart describing the inclusion of studies, is presented. Having examined a total of 1955 abstracts (1228 after removal of duplicates), 236 full-text papers that possibly met the inclusion criteria were examined for further consideration.

Figure 1. Flowchart of selection and Inclusion of Studies.

Two studies were excluded, as no data for the control conditions to which the subjects were randomized were given [15]; [16]. Further, one study had to be excluded since no data were provided for meta-analysis [31].

Characteristics of included studies

The eleven primary studies [32][41] included a total of 654 participants (315 in the treatment conditions and 290 in the control conditions). Selected characteristics of the eleven included studies are described in Table 1. In four studies, subjects were recruited from both community and clinical settings, in four studies from community settings only, whereas in three other studies participants were recruited from clinical settings only. In seven studies psychological treatment was compared with a waiting-list control group, in three studies with a pill-placebo and in one study with TAU.

CBGT was defined as a psychological intervention, in which the following elements were included: 6–16 sessions of therapy led by 2 therapists, 2 or 2.5 hours each and group sizes of at least 4 participants. Moreover, both exposure in vivo and cognitive elements such as cognitive restructuring or skills to identify negative thoughts were part of the treatment.

Quality Assessment

The quality of the eleven studies varied. In five studies [32]; [33]; [38]; [39]; [40] the allocation to conditions was reported adequately, whereas in the other studies it was not [11]; [34]; [35]; [36]; [37]; [41]. Only two studies described concealment of allocation [32]; [33]. Incomplete data was addressed in most studies [11]; [32]; [33]; [36]; [38]; [39]; [40]; [41]. Percentages of patients lost to follow-up ranged from zero to 37.5%. Intention-to-treat analyses were conducted in only four studies [32]; [33]; [38]; [39].

Effects of psychological treatments at post-tests

The effects of psychological group-treatment could be compared to a waiting-list, placebo or TAU in eleven studies (Table 2). The mean effect size was g = 0.54 (95% CI; 0.36–0.73), that corresponds with a NNT of 3.36. Heterogeneity was moderate (Q = 11.61; p = 0.31; I2 = 13,90%). Figure 2 presents the effect sizes and 95% confidence intervals of each study.

Table 2. Meta-analyses of studies examining the effects of group-therapy on social phobia.

Figure 2. Effect sizes of CBGT for adult SAD: Hedges’ g.

Subgroup analyses

Since the number of included studies was small, only the most basic subgroup analyses containing at least 4 studies were conducted. Specifically, these were analyses comparing the type of control group (waiting-list versus placebo/treatment as usual); type of recruitment (community sample versus both community and clinical sample) and type of analyses (intention-to-treat versus completers only) (Table 2). There were no statistical subgroup differences with respect to control groups, type of recruitment or type of analyses.

Heterogeneity varied between the subgroups. With the exception of two subgroups, placebo and treatment-as-usual (I2 = 0.00, g = 0.42) and intention-to-treat (I2 = 0.00, g = 0.45), all subgroup analyses resulted in I2 levels higher than 30%. Because of the small number of studies in each subgroup, the results of these analyses have to be interpreted with caution.

Publication Bias

The funnel plot and Duval & Tweedie's trim and fill procedure indicated some publication bias. After adjustment for missing studies, the effect size dropped from g = 0.54 to g = 0.48 (95% CI: 0.28–0.67; number of trimmed studies: 2) and the Begg & Mazumdar rank correlation test did not indicate a significant difference (p = 0.27, one-tailed).


This meta-analysis found that cognitive- behavioural group-treatments have a moderate, but significant effect in the treatment of SAD compared to control. Some individual studies did not have enough power to detect a significant effect, but after pooling of the studies the effect size became significant. However, the effect sizes differed highly between studies, ranging from zero (indicating no treatment effect) to above one (indicating a large effect size).

With the exception of Acarturk et al [19] the previous meta-analyses [20]; [21]; [22]; [23] included only few RCTs and did not apply tests for heterogeneity and other meta-analytic analyses since they may not have been available. As indicated, those state-of art analyses allow finding differences between studies and their influence on the outcome. Beyond that, none of the earlier meta-analyses investigated the effects of psychological group-formats, specifically CBGT, in SAD in adults. In short, this present meta-analysis extends the prior work by focussing explicitly on CBGT and including only recent RCTs, analysed with the latest available methodology.

This study has nevertheless several limitations. First, the number of included studies was rather small, which may have limited the possibilities to examine potential moderators of outcome. Second, there was a considerable difference between the included studies that may have contributed to heterogeneity. For instance, only few studies conducted intention-to-treat analyses, which are crucial to get a more reliable estimate of the effect size. Thirdly, a meta-analysis is always limited by the methodological quality of included studies. Since we also included studies with lower quality, it may have influenced our overall outcome.

More research in the field of group treatments for SAD is certainly needed. Future studies should give specific attention to important quality criteria, such as adequate randomization procedures and reporting, concealment of allocation, and the inclusion of all randomized respondents in the final analyses. Further, future trials should contain larger sample sizes, and may compare CBGT to other cost-effective therapies. A small study in 37 SAD patients suggested that individual CBT delivered through the internet was equally effective as face-to-face CBGT [42], but associated with substantially less clinician time.

Despite the limitations of this meta-analysis, evidence shows that CBGT seems to be an effective way of treating individuals with SAD. Thus, CBGT may be an attractive alternative to individual face-to-face CBT to decrease the burden of SAD, because of its advantages with respect to cost-effectiveness and since vivo exposure to social situations may occur naturally during group sessions.

Author Contributions

Analyzed the data: HW PC. Wrote the paper: HW. Designed and analyzed the study: PC HW. Interpretation of results, drafting and revising: HW MS PC.


  1. 1. Kessler RC, Chiu WT, Demler O, Walters E (2005) Prevalence, Severity, and Comorbidity of 12-month DSM-IV Disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry 62: 617–27.
  2. 2. Furmark T, Tillfors M, Everz P, Marteinsdottir I, Gefvert O, et al. (1999) Social Phobia in the general population: Prevalence and sociodemographic profile. Soc Psychiatry Psychiatr Epidemiol 34: 416–424.
  3. 3. Furmark T (2002) Social phobia: Overview of community surveys. Acta psychiatrica Scandinavica 105: 84–93.
  4. 4. Kasper S (1998) Social Phobia: The nature of the disorder. J Affect Dis 50 (Suppl 1)3–9.
  5. 5. American Psychiatric Association [APA] (1994). Diagnostic and statistical manual of mental disorders.American Psychiatric Association. (4th ed.). Washington, DC.
  6. 6. Magee WJ, Eaton WW, Wittchen HU, McGonagle KA, Kessler RC (1996) Agoraphobia, Simple Phobia, and Social Phobia in the National Comorbidity Survey. Arch Gen Psychiatry 53: 159–168.
  7. 7. Acarturk C, Graaf de R, Straten van A, Have MT, Cuijpers P (2008) Social Phobia and number of social fears, and their association with comorbidity, health-related quality of life and help seeking: A population-based study. Soc Psychiatry Psychiatr Epidemiolol 43: 273–279.
  8. 8. Andlin-Sobocki P, Bengt J, Wittchen HU, Olesen J (2011) Cost of disorders in the brain in Europe 2010. Eur Neuropsychopharmacol 21: 718–779.
  9. 9. Brunello N, Boer den J, Judd LL, Kasper S, Kelsey JE, et al. (2000) Social phobia: diagnosis and epidemiology, neurobiology and pharmacology, comorbidity and treatment. Jour Affective Disorders 60: 61–74.
  10. 10. Weiller E, Bisserbe JC, Boyer P, Lepine JP, Lecrubier Y (1996) Social phobia in general health care: an unrecognised undertreated disabling disorder. Br J Psychiatry 168: 169–174.
  11. 11. Heimberg RG, Liebowitz MR, Hope D, Schneier FR, Holt CS, et al. (1998) Cognitive Behavioral Group Therapy vs Phenelzine Therapy for Social Phobia. Arch Gen Psychiatry 55: 1133–1141.
  12. 12. Bjornsson AS. Bidwell LC, Brosse AL, Carey G, Hauser M, et al. (2011) Cognitive-behavioral group therapy versus group psychotherapy for social anxiety disorder among college students: a randomized controlled trial. Depress Anxiety 28: 1034–1042.
  13. 13. Herbert JD, Dalrymple K, Nolan EM (2005) Social Skills Training Augments the Effectiveness of Cognitive Behavioral Group Therapy for Social Anxiety Disorder. Beh Therapy 36: 125–138.
  14. 14. Rodebaugh TL, Holaway RM, Heimberg RG (2004) The treatment of social anxiety disorder. Clin Psy Review 24: 883–908.
  15. 15. Hofmann SG, Moscovitch DA, Hyo-Jim K, Taylor AN (2004) Changes in self-perception during treatment of social phobia. J Consult Clin Psychol 72: 588–96.
  16. 16. Hofmann SG, Suvak M (2006) Treatment attrition during group therapy for social phobia. Jour Anxiety Disorders 20: 961–972.
  17. 17. Piet J, Hougaard E, Hecksher MS, Rosenberg NK (2010) A randomized pilot study of mindfulness-based cognitive therapy and group cognitive-behavioral therapy for young adults with social phobia. Scand Jour Psychology 51: 403–410.
  18. 18. Silverman WK, Kurtines WM, Ginsburg GS, Weems CF, Lumkin PW, et al. (1999) Treating Anxiety Disorder in Children with Group Cognitive-Behavioural Therapy: A Randomized Clinical Trial. J Consult Clin Psychol 67: 995–1003.
  19. 19. Acarturk C, Cuijpers P, Straten van A, de Graaf R (2009) Psychological treatment of social anxiety disorder: a meta-analysis. Psychol Med 39: 241–54.
  20. 20. Federoff IC, Taylor S (2001) Psychological and Pharmacological Treatments of Social Phobia: A Meta-Analysis. J Clin Psychopharmacol 21: 311–324.
  21. 21. Feske U, Chambless DL (1995) Cognitive behavioral versus exposure only treatment for social phobia: A meta-analysis. Behav Therapy 26: 695–720.
  22. 22. Gould RA, Buckminster S, Pollack MH, Michael WO (1995) Cognitive-Behavioral and Pharmacological Treatment for Social Phobia: A Meta-Analysis (1995) Clin Psychol Sci Pract. 4: 291–306.
  23. 23. Taylor S (1996) Meta-analysis of cognitive-behavioral treatments for social phobia. Behav Therapy 27: 1–9.
  24. 24. APA (1987) Diagnostic and Statistical Manual of Mental Disorders 3rd edn, revised. American Psychological Association: Washington, DC.
  25. 25. Higgins JPT, Green S (2008) Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.1 [updated September 2008]. The Cochrane Collaboration. Available:
  26. 26. Kraemer HC, Kupfer DJ (2006) Size of treatment effects and their importance to clinical research and practice. Biol Psychiatry 59: 990–996.
  27. 27. Schünemann HJ, Oxman AD, Vist GE, Higgins JPT, Deeks JJ, et al. (2008) Interpreting results and drawing conclusions. In: Cochrane Handbook for Systematic Reviews of Interventions. Chichester (UK) Edited by Higgins JPT, Green S: John Wiley & Sons: 359–383.
  28. 28. Higgens JPT, Green S (2008) Guide to the Content of a Cochrane protocol and review. In: Cochrane Handbook for Systematic Reviews of Interventions. Chichester (UK) Higgins JPT, Green S. editors: John Wiley & Sons, 2008: John Wiley & Sons, Ltd.: Chichester, UK 2008: 51–78.
  29. 29. Duval S, Tweedie R (2000) Trim and Fill: A Simple Funnel-Plot-Based Method. Biometrics 56: 455–463.
  30. 30. Begg CB, Mazumdar M (1994) Operating Characteristics of a Rank Correlation Test for Publication Bias. Biometrics 50: 1088–1101.
  31. 31. Pishyar R, Harris LM, Menzies RG (2008) Responsiveness of measures of attentional bias to clinical change in social phobia. Cog Emotion 22: 1209–1227.
  32. 32. Blanco C, Heimberg RG, Schneier FR, Fresco DM, Chen H, et al. (2010) A placebo-controlled trial of phenelzine, cognitive behavioral group therapy, and their combination for social anxiety disorder. Arch Gen Psychiatry 67: 286–295.
  33. 33. Davidson JRT, Foa EB, Huppert JD, Keefe FJ, Franklin ME, et al. (2004) Fluoxetine, comprehensive cognitive behavioral therapy, and placebo in generalized social phobia. Arch Gen Psychiatry 61: 1005–1013.
  34. 34. Gruber K, Moran PJ, Roth WT, Taylor CB (2001) Computer –Assisted Cognitive Behavioural Therapy for Social Phobia. Behav Ther 32: 155–165.
  35. 35. Hofmann S, Moscovitch DA, Kim HJ, Taylor AN (2004) Changes in Self- Perception During Treatment of Social Phobia. 2004, J Consult Clin Psychology 72: 588–596.
  36. 36. Hope DA, Heimberg RG, Bruch MA (1995) Dismantling Cognitive-Behavioral Group Therapy for Social Phobia. Behav Res Ther 33: 637–650.
  37. 37. Mörtberg E, Karlsson A, Fyring C, Sundin Ö (2006) Intensive cognitive-behavioural group treatment (CBGT) of social phobia: A randomized controlled study. J Anxiety Disord 20: 646–660.
  38. 38. Mörtberg E, Clark DM, Sundin Ö, Wistedt A (2007) Intensive group cognitive therapy vs. treatment as usual in social phobia: a randomized controlled trial. Acta Psychiatr Scand 115: 142–154.
  39. 39. Rapee R, Abbott MJ, Baiillie AJ, Gaston JE (2007) Treatment of social phobia through pure self- help and therapist-augmented self-help. BJP 191: 246–252.
  40. 40. Stangier U, Heidenreich T, Peitz M, Lauterbach W, Clark DM (2003) Cognitive therapy for social phobia: individual versus group treatment. Behav Res Ther 41: 991–10007.
  41. 41. Wong DFK, Sun SYK (2006) A Preliminary Study of the Efficacy of Group Cognitive-behavioural Therapy for People with Social Anxiety in Hong Kong. HK J Psychiatry 2006 16: 1650–1656.
  42. 42. Andrews G, Davies M, Titov N (2011) Effectiveness randomized controlled trial of face to face versus Internet cognitive behaviour therapy for social phobia. Aust NZ J Psychiat 45: 337–340.