Cognitive–behavioral therapy modifies the naturalistic course of social anxiety disorder: Findings from an ABA design study in routine clinical practices
Correspondence: Toshi A. Furukawa, MD, PhD, Department of Health Promotion and Human Behavior, Kyoto University Graduate School of Medicine/School of Public Health, Yoshida Konoe-cho, Sakyo-ku, Kyoto 601-8501, Japan. Email: firstname.lastname@example.org
While randomized evidence appears to have established efficacy of cognitive–behavioral therapy (CBT) and some pharmacotherapy for social anxiety disorder (SAD), their real-world effectiveness has been called into question by long-term naturalistic cohort studies of patients with SAD as they show very low probability of recovery and sustained social dysfunctions despite some drug and psychological therapies.
The present study examines the effectiveness of group CBT for SAD in real-world settings (n = 62) by examining the course of patients' symptomatology and social functions through approximately 6 months on the waiting list, through 6 months receiving the manualized group CBT intervention consisting of 16 2-h sessions, and for 12 months after the treatment.
We found: (i) that the patients with SAD changed little or possibly worsened through the 6 months on the waiting list, although two in three of them were on antidepressants, benzodiazepines or both; (ii) that both their symptomatology and social function improved significantly and substantively through the group CBT; and (iii) that this improvement was maintained through the 3- and 12-month follow ups.
We can implement and must disseminate evidence-based, effective CBT for more patients with SAD to lessen their suffering and stop the perpetuation of their symptoms.
IT IS WIDELY accepted that cognitive–behavioral therapy (CBT) and pharmacotherapy with antidepressants and some benzodiazepines represent efficacious treatments for social anxiety disorder (SAD). Much of this evidence comes from short-term trials of a few months' duration at most, but some follow-up studies have also been conducted following acute-phase treatment trials. By and large, gains obtained during CBT appear to persist for 6–14 months after the treatment has ended.[2, 3] Acute-phase treatment efficacy by pharmacotherapy also appears to linger even when the medication is discontinued,[4, 5] although it is better maintained if the drug is continued. When pharmacotherapy and psychotherapy were directly compared after respective acute-phase treatment, both had some enduring effects but CBT was superior to drug therapy in the maintenance of the gains. Moreover, several studies have shown that treatment gains achieved in well-controlled research settings, such as randomized controlled trials (RCT), can also be achieved in more ‘naturalistic’ clinical populations outside of research institutions.[7, 8]
However, naturalistic observational studies of social phobia in the real world appear to be incongruent with this established efficacy and effectiveness of the treatments. They have shown that SAD follows a chronic unremitting course with debilitating consequences in the sufferers' social functionings, despite some apparently efficacious treatments that they receive. One of the longest longitudinal studies to date of the disorder, the Harvard/Brown Anxiety Research Project (HARP), has followed 176 patients with SAD for over 10 years now. SAD showed the lowest probability of recovery among anxiety and depressive disorders, although three-quarters of these SAD patients continued to receive some psychotropic medications and some 25–40% received behavioral or cognitive psychotherapies.[10, 11]
How can these apparently conflicting data translate into the real-world practices of CBT? We have been conducting group CBT for SAD since 2003 and, due to the growing demand, the waiting period has been extended up to 6–12 months for the past several years. In this context, we decided to examine the course of SAD before, during, and after the patients received evidence-based psychotherapy in the real-world settings. In total, the observational period spanned approximately 6 months before CBT, 6 months during CBT, and 12 months after CBT. In other words, the present study reports an ABA design examination of the real-world effectiveness of group CBT for SAD, where A represents a period of treatment-as-usual or no treatment and B the active CBT intervention. Our study is unique in enabling the comparison between the waiting period, in which the subjects received treatment-as-usual or no treatment, similar to the naturalistic observational studies, and the active treatment to post-treatment periods, in which the subjects have been treated with manualized CBT group treatment with established efficacy.
The subjects of the present study: (i) were patients who participated in the group CBT program for SAD at Nagoya City University Hospital Department of Psychiatry; (ii) had some assessment data (see below) at the time of their first consultation at our department some time before the commencement of the program; and (iii) provided written informed consent to have their clinical data analyzed and published.
All the patients were referrals from the general medical or psychiatric clinics and departments in the town or self-referrals. After the routine psychiatric consultation, they were placed on the waiting list for group CBT if: (i) their principal diagnosis was SAD; (ii) they wished to receive group CBT; and (iii) they did not suffer from psychotic disorder, mood disorder or personality disorder that was severe enough to interfere with group administration of CBT.
All the other psychiatric or medical comorbidities were allowed. There was no restriction on concomitant pharmacotherapy. At the time of their first consultation, the Liebowitz Social Anxiety Scale (LSAS) was administered by the psychiatrist, and the Social Phobia Scale (SPS), Social Interaction Anxiety Scale (SIAS), Fear Questionnaire (FQ) and Work, Home and Leisure Activities Scale (WHLS) were self-administered.
When their turn approached, they were re-invited for the baseline interview, in which:
- The anxiety and mood sections of the Structured Clinical Interview for DSM-IV (SCID-IV) were administered, and their SAD diagnosis was confirmed. In addition, the diagnosis of generalized subtype was established based on an ad hoc semi-structured interview based on Stein's work.
- All the above-mentioned assessments, as well as the Depression Subscale of the Symptom Checklist-90 Revised (SCL-Dep), were administered.
The group CBT for SAD at our department was originally based on the program developed by Andrews et al. and was later integrated with Clark and Wells' model of SAD. The program now consists of the following components: (i) psychoeducation about anxiety and social anxiety; (ii) individualization of the CBT model of SAD; (iii) experiments to drop safety behavior and self-focused attention; (iv) attention training; (v) cognitive restructuring; (vi) video-feedback of within-session role plays; and (vii) in vivo graded exposures. The number of patients in a group is limited to three in order to enable due attention to individual differences and yet to allow group interaction in the role-plays and video feedbacks. The program is run in 16 2-h weekly sessions by two therapists. All the therapies and the assessments were completed by the psychiatrists and clinical psychologists who had had at least 2 years of clinical experience in cognitive–behavioral therapy. There was no treatment integrity checks as the treatment took place in routine clinical settings.
The pharmacotherapy, if any if the patients were on such at the beginning of the CBT program, was kept constant through the CBT sessions. Upon completion of the CBT program, the patients are administered the same instruments, including LSAS, SPS, SIAS, FQ, SCL-Dep and WHLS.
At 3 and 12 months after the termination of the CBT program, the patients received (via mail) and were asked to complete the SPS, SIAS, FQ, SCL-Dep and WHLS. It was not possible to administer LSAS at these follow ups because the latter is an observer-rated instrument.
This study was approved by the Ethics Review Committee of the Graduate School of Medical Sciences, Nagoya City University. All the patients gave their written informed consent after full disclosure of the purposes and procedures of the study at the beginning of the CBT program. We have previously published a preliminary report from our program focusing on its acute-phase treatment effects. The present analyses include 17 of the former study's 57 subjects but represent a completely different cohort with 45 additional subjects as our program started to have a prolonged waiting list period since the study by Chen et al. was completed and as we started to assemble pre-baseline data more systematically.
SPS and SIAS
The SPS and SIAS are both 20-item self-report questionnaires, which in combination measure two aspects of social fears, namely those of being observed and those of social interaction. Each item is rated on a 4-point scale of 0 (not at all characteristic or true of me) through 4 (extremely characteristic or true of me). The score range is therefore between 0 and 80. Excellent internal consistency reliability and sufficient discriminant, predictive and concurrent validity have been demonstrated for the Japanese versions.
The FQ is a self-report questionnaire measuring avoidance due to three clusters of phobic fears, namely agoraphobia, blood-injury phobia and social phobia. Each cluster is represented by five typical situations, each of which is rated on a 9-point scale of 0 (would not avoid it) through 8 (always avoid it), with the resultant score range between 0 and 72. In this study we used the social phobia subscale. Good test–retest reliability and factor validity have been reported for the original version and its Japanese version.
The LSAS is the most frequently used clinician-administered instrument for the assessment of SAD. It is a 24-item scale that provides separate scores of fear (0–3, indicating none, mild, moderate and severe, respectively) and avoidance (0–3, indicating never, occasionally, often and usually, respectively) of social performance of interaction situations. Satisfactory reliability and validity have been reported for its Japanese version. The score ranges between 0 and 144.
Depression subscale of SCL-Dep
The Symptom Checklist-90 Revised is a widely used self-report symptom inventory that has been designed primarily to reflect the psychological symptom patterns of psychiatric and medical patients. We used its depression subscale, consisting of 13 questions, each rated on a 5-point scale of distress ranging from 0 (not at all) to 4 (extremely). The reliability and validity of the Japanese version of SCL-90-R has been established.
The WHLS is a self-report scale for measuring functional impairments in the areas of work, home management, social and private leisure activities. The patients' answers are graded between 0 (not at all impaired) to 8 (very severely impaired) for each of these four domains. Satisfactory reliability and construct validity have been reported. Its Japanese version has also been validated.
Our primary hypotheses concerned whether there were any changes in SAD symptomatology and social functioning before, during and after the CBT program. Accordingly we ran repeated-measures anova comparing the scores at the first visit and at the start of treatment, those at the start and the end of the treatment, and those at the end of the treatment and the follow ups. When Mauchly's test of sphericity was statistically significant at or below 0.05, we used Greenhouse–Geisser correction. In order to quantify the magnitude of change, we calculated within-subject effect sizes. An effect size of 0.2 is usually considered to represent a small change, 0.5 a moderate change and 0.8 a large change.
In order to determine possible prognostic factors, we ran multiple linear regression of the end-of-treatment LSAS, SPS and SIAS scores while controlling for their respective start-of-treatment scores. Each of the following possible prognostic factors[11, 27, 28] was entered into multiple regression separately one at a time: sex, age, age of onset, generalized subtype, comorbidity and drug treatment that the patients received through the CBT program.
Sixty-two patients had some data before the baseline upon their first consultation and at the baseline assessment at the beginning of the CBT program. Table 1 summarizes their baseline demographic and clinical characteristics. The sample consisted of equal numbers of men and women, their mean age was 32.5 and their mean age of onset of SAD was 18.0. The majority suffered from generalized SAD and almost half suffered from past or current major depression.
Table 1. Demographic and clinical characteristics of the sample
|Sex||31 M/31 F|
|Age of onset||18.0 (6.6)|
|Generalized subtype||51 (82%)|
|Major depression, current||6 (9.7%)|
|Major depression, history||20 (32.2%)|
|Bipolar disorder||2 (3.2%)|
|Generalized anxiety disorder||1 (1.6%)|
|Specific phobia||1 (1.6%)|
|Alcohol dependence||2 (3.2%)|
|Avoidant personality disorder||10 (16.1%)|
|Treatment received at baseline|| |
|Benzodiazepines, regular||14 (22.6%)|
|Benzodiazepines, p.r.n.||11 (17.7%)|
|No psychotropic medication||20 (32.3%)|
|Cognitive–behavioral therapy||0 (0%)|
The waiting period from the first visit to the start of the CBT program was 199 days on average (SD = 106, range: 36–638, interquartile range [IQR]: 125–244). During the waiting period, none received CBT, and over two-thirds (68%) were on some psychotropic medications, mostly antidepressants with or without benzodiazepines (18% and 42%, respectively).
Once they were enrolled in the program, the average number of sessions received was 13.4 (SD = 4.5, range: 2–20, IQR: 12–15) over the time span of 125 (SD = 22) days, with 10 (16.1%) patients dropping out of the treatment after receiving less than nine sessions. We set nine as the minimum number of sessions for the patients to be considered having received our CBT treatment because the program is so constructed that the main materials can only be covered with nine sessions. There were no statistically significant differences between these 10 early terminators and the remaining program completer in terms of age, sex, comorbidity or pharmacological treatment they were on at the beginning of the CBT program.
Information on the treatment received during the follow up was available for 46 patients of the original cohort (74%) who responded to the 3- and 12-month follow up. Of these, 29 were originally on pharmacotherapy when they started the CBT program and 17 were not. Of the former, seven (24%) stopped medication, while three (18%) of the latter group began pharmacotherapy. Overall, 25 (54%) were on some medications, mostly again on antidepressants with or without benzodiazepines (24% and 17%, respectively). None was on further cognitive–behavioral therapy.
Changes in SAD severity
Figure 1 depicts the changes in the patients' scores regarding SAD psychopathology through the five assessment points, namely upon their first visit, at the start of the CBT program, at the end of the program, at 3-month follow up, and at 12-month follow up. Figure 2 shows the same changes in the patients' social functions.
Table 2 tabulates the results of repeated-measures anova between the first visit and the start-of-treatment assessments, between the start-of-treatment and end-of-treatment assessments, and between the end-of-treatment and 3- or 12-month follow-up assessments. Except for the LSAS score from first visit to start-of-treatment, all the changes were small to less than small in the periods before treatment and after treatment. On the other hand, all the measurements except for the social functionings in the private leisure domain showed large to even larger changes through the treatment period.
Table 2. Changes in social anxiety disorder psychopathology of the sample, mean (SD)
|LSAS||66.5 (21.5)||78.7 (24.5)||0.53||<0.001|
|SPS||37.7 (13.3)||34.7 (13.2)||−0.23||0.063|
|SIAS||52.5 (13.6)||52.5 (12.6)||0.0||0.34|
|FQ social phobia subscale||23.1 (6.8)||22.8 (6.2)||−0.05||0.16|
|WHLS|| || || || |
|Work||4.9 (2.0)||4.9 (2.1)||0.0||0.87|
|Home||2.0 (1.7)||2.4 (1.8)||0.23||0.65|
|Social||4.8 (1.8)||4.7 (2.0)||−0.05||0.89|
|Private||1.7 (1.9)||1.6 (1.8)||−0.05||0.36|
|LSAS||75.8 (23.6)||53.5 (24.7)||−0.92||<0.001|
|SPS||34.1 (12.7)||22.3 (12.9)||−0.92||<0.001|
|SIAS||52.3 (13.2)||39.0 (13.9)||−0.98||<0.001|
|FQ social phobia subscale||22.8 (6.2)||16.3 (8.4)||−0.89||<0.001|
|WHLS|| || || || |
|Work||4.9 (2.2)||2.8 (1.8)||−1.05||<0.001|
|Home||2.3 (1.8)||1.1 (1.2)||−0.80||<0.001|
|Social||4.5 (2.0)||3.1 (2.0)||−0.70||<0.001|
|Private||1.6 (1.8)||1.0 (1.3)||−0.39||0.007|
|SCL-Dep||1.35 (0.88)||0.82 (0.62)||−0.71||<0.001|
|SPS||21.9 (12.8)||22.0 (11.8)||0.01||0.92|
|SIAS||39.1 (13.7)||39.5 (13.7)||0.03||0.72|
|FQ social phobia subscale||16.5 (8.6)||17.2 (7.6)||0.09||0.47|
|WHLS|| || || || |
|Work||2.8 (1.9)||3.1 (2.0)||0.15||0.13|
|Home||1.0 (1.1)||1.3 (1.6)||0.22||0.30|
|Social||3.1 (2.1)||2.9 (1.8)||−0.10||0.31|
|Private||0.9 (1.1)||0.9 (1.3)||0.0||0.69|
|SCL-Dep||0.85 (0.67)||1.03 (0.82)||0.24||0.06|
|SPS||21.6 (12.9)||19.6 (9.3)||−0.18||0.16|
|SIAS||38.5 (14.1)||37.8 (13.6)||−0.05||0.50|
|FQ social phobia subscale||16.7 (8.7)||16.7 (7.9)||0.0||0.92|
|WHLS|| || || || |
|Work||2.9 (1.8)||2.6 (1.9)||−0.16||0.32|
|Home||1.0 (1.1)||1.1 (1.4)||0.08||0.88|
|Social||3.1 (2.0)||2.9 (2.0)||−0.10||0.51|
|Private||0.9 (1.1)||1.1 (1.3)||0.17||0.35|
|SCL-Dep||0.89 (0.71)||1.02 (0.82)||0.17||0.33|
We also measured the participants' depression severity at start of treatment, at end of treatment and at 3- and 12-month follow ups, because almost half of them suffered from past or current major depression. The results are tabulated in Table 2, and followed the same pattern as SAD severity.
Predictors of treatment outcomes
None of the examined predictors at baseline, including sex, age, age of onset, generalized subtype, comorbidity and drug treatment before beginning the treatment, was significantly associated with the LSAS, SPS or SIAS scores at end of treatment when controlled for their respective baseline scores.
The present study represents a unique ABA design examination of the naturalistic course of SAD before the patients participated in the CBT program, the acute phase treatment effect of this program, and the naturalistic course after the program is over with possible enduring effects from the acute phase treatment. It found: (i) that the patients with SAD changed little or possibly worsened through the 6 months on average on the waiting list, although two in three of them were on antidepressants, benzodiazepines or both; (ii) that both their symptomatology and social function improved significantly and substantively through the group CBT consisting of 13, on average, 2-h sessions; and (iii) that this improvement was maintained through the 3- and 12-month follow ups while approximately half the patients continued with medication.
The weaknesses of the study include lack of experimental control and less-than-ideal follow-up rates after the end of the CBT program (88% at 3-month and 77% at 12-month follow ups). Also, we did not re-administer SCID after the CBT program was completed, so we do not know how many of our patients no longer satisfied the diagnostic criteria for SAD after the treatment. Its strengths, however, include the ABA design that measured the stability of the SAD symptoms before the active intervention, the comprehensiveness of the assessments encompassing both narrow symptomatology and social functionings, and the well-documented manualized nature of the CBT program provided. In a naturalistic open one-arm study like the present one, only the ABA design can provide a compelling argument that the changes observed through the acute-phase treatment are not totally attributable to the natural course or regression towards the mean of the disorder.
The fact that the patients' SAD psychopathology showed little change through the 6 months before the CBT program while 65% of them were on benzodiazepines and/or antidepressant medications is congruent with the existing observational studies. The naturalistic course of SAD, while possibly on some treatments without quality control, is indeed chronic and unremitting.
The CBT, however, can alter this naturalistic course in the real world. The magnitude of this improvement compares very favorably with the results obtained in more controlled randomized trials reported in the literature.[29-31] For example, the pre–post effect size of the changes in LSAS, SPS and SIAS were 0.75, 0.64 and 0.48, respectively, in the group studied by Mortberg et al. and were 0.92, 0.92 and 0.98, respectively, in our group. The fact that none of the examined variables could predict the outcomes is in line with some previous studies on SAD and may speak to the overriding influence of our intervention over the possible natural predictors.
Much to our satisfaction and again in accordance with the published reports, this treatment gain persisted in our sample up to 3 and 12 months of follow up, when one-third of those who had been on medications at the commencement of the CBT program had stopped medication and when none had booster CBT treatments. However, it must regrettably be noted that the average scores of the program completers were either only slightly below the recommended cut-off of 24 on SPS or slightly above those of 48 on LSAS and 34 on SIAS at the end of the CBT program and throughout the 3- and 12-month follow ups (see Table 2c).
The glass remains half full and half empty. The clinical implication of this study is clear. We must disseminate and can implement the evidence-based, effective cognitive–behavioral treatments for more patients with SAD to lessen their suffering and stop the perpetuation of their symptoms. The research implication of this study is again clear. The current treatment for SAD is good but not good enough. Although Clark et al. have reported excellent results in their controlled trials,[35, 36] direct dissemination efforts have failed to reproduce the same magnitude of effectiveness.[30, 31] We therefore need research not only into more effective treatment but also into more efficient training and dissemination methods.
This research was supported by Grants-in-Aid from the Ministry of Health, Labour and Welfare, Japan to T.A.F. The Ministry of Health, Labour and Welfare had no role in the design, conduct, interpretation and writing-up of this study.
T.A.F. has received speaking fees from Astellas, Dai-Nippon Sumitomo, Eli Lilly, GlaxoSmithKline, Janssen, Meiji, Otsuka, Pfizer, Schering-Plough and Yoshitomi. He is on the research advisory board for Sekisui Chemicals and Takeda Science Foundation. He has received royalties from Igaku-Shoin, Seiwa-Shoten, Nihon Bunka Kagaku-sha and American Psychiatric Publications. T.A. received research funds and speaking fees from Astellas, AstraZeneca, BMS, Daiichi-Sankyo, Dai-Nippon Sumitomo, Eisai, Eli Lilly, GlaxoSmithKline, Janssen, Kyowa-Hakko, Meiji, Otsuka, Pfizer, SanofiAventis, Shionogi and Yakult. N.W. has received speaking fees from Asahi Kasei, Dai-Nippon Sumitomo, GlaxoSmithKline, Pfizer and Schering-Plough. All the other authors report that they have no conflicts of interest to declare.