Experience Versus Report: Where Are Changes Seen After Exposure-Based Cognitive-Behavioral Therapy? A Randomized Controlled Group Treatment of Childhood Social Anxiety Disorder

Social anxiety disorder (SAD) is one of the most common mental disorders associated with great impairment in the well-being and everyday life of affected children and youth [1]. With prevalence rates as high as 9% in youth [2], effective and efficient treatment is essential. Cognitive behavioral therapy (CBT) programs have generally proven effective for anxiety disorders in children, adolescents, and adults [3, 4]. For example, typical programs such as the Coping Cat program consists of identification of anxious feelings, cognitive restructuring, positive self-talk and exposure tasks, as well as rewards for efforts to cope with anxiety. However, in comparison to other anxiety disorders, a primary or comorbid diagnosis of SAD usually leads to less remission of symptoms and lower response rates in generic treatment programs [4-7]. Treatments based on theoretical models of SAD [4] and tailored to SAD-specific deficits may be needed. Furthermore, in contrast to the more common approach of measuring treatment success by reductions in SAD symptoms and decreases in the severity of a clinical diagnosis, additional relevant measures (e.g., experiences in social situations), should be taken into account.

As the core symptom of SAD is fear of social interactions often combined with difficulty performing adequately in social situations, a treatment specific to SAD may benefit from including peers to enhance possibilities for social interaction with peers in treatment (e.g., to provide continuous exposure). In addition to changing the treatment’s content (i.e., focus on cognitions), changing the treatment’s structure from individual to group may therefore likely be a second important adaptation to target SAD specifically. Group CBT programs have gained influence as a generic treatment approach for child and youth anxiety [8, 9]. In children and youth suffering from different anxiety disorders, group CBT has been shown to achieve benefits similar to those of individual CBT, and these have remained stable at 1-year follow-up [10]. A group approach allows almost constant exposure to other individuals and direct feedback from interaction partners. Interestingly, only few studies have focused on group treatment targeting SAD in adolescence [11-17]. Some pilot studies of small samples showed significant reduction in social anxiety symptoms after group treatment in adolescence [11, 12, 17]. A direct comparison of individual versus group therapy for SAD did, however, not result in a clear preference for either [13, 14, 18].

To date, even though the earliest onset of SAD has been reported at age 7 [19] to 9.2 years [2], almost all group CBT programs have been developed for adolescents starting at 12 years of age. Importantly, Halldorsson and Creswell [20] point out that preadolescents differ developmentally from adolescents. Only a few group treatments have focused on SAD in children [21, 22]. These studies showed substantial and stable therapeutic effects, but a large number of patients did not respond to the treatment. Thus, therapeutic effects may be enhanced if treatment programs include more exposure, which has been confirmed as the method of choice for adult patients with anxiety disorders [23-26]. The above mentioned treatments only used a low level form of exposure during social skills training, as homework [22] or as a short element in combination with cognitive restructuring [27]. Current studies suggest that exposure therapy is a key element in changing cognitions as negative expectations are challenged, attention biases corrected and positive cognitions applied [23].

CBT is based on the assumption that affective, cognitive, behavioral, and physiological responses are highly interrelated and are, thus, the basis for both psychopathological symptoms and treatment (e.g. [28].). Therefore, assessment of all responses would appear crucial. Rather than targeting these responses individually, as has been done in previous studies, a social stress task including public speaking (Trier Social Stress Test for Children; TSST-C [29]) could be used to evaluate these responses in SAD, as it induces disorder-similar stress. Concerning affective arousal, children with SAD report more social anxiety during social stress (e.g. [30].). While this heightened state anxiety is already apparent at baseline (i.e., anticipation anxiety), it increases during stress but decreases during recovery, showing a modulation back to baseline levels. Regarding cognitive responses, children with SAD report more negative post-event processing, that is, negative thoughts about their own failings after having experienced a social situation (e.g.[31], and negative anticipatory cognitions concerning an upcoming social situation (e.g.[32]. Concerning behavioral symptoms, children with SAD usually report a more negative perception of their social skills; that is, they perceive their own actions as more nervous and believe they make a negative impression (e.g. [33].). Finally, results on physiological arousal show a tonic hyperarousal during social situations that can be seen in both heart rate levels and electrodermal activity (e.g. [30].). Only few studies have addressed these variables as possible treatment outcome variables: Adult studies have shown that cognitions in SAD change as a result of CBT [34], but results are inconclusive about changes in heart rate [35]. Regarding behavior, an increase in parent-perceived social skills in children after CBT was shown [22]. Examining all aspects of the CBT model together might be useful for measuring treatment success.

Taking these findings into account and in line with recent work on exposure (e.g. [23].), we previously tested a SAD-specific exposure-based group treatment in a randomized controlled trial with 74 children (aged 8 to 12 year) with SAD [36]. Compared to parents of children in a waitlist control (WLC) group, parents of children in a CBT group reported a greater decrease in symptoms (CBT: d = 1.02, WLC: d = 0.06), but children did not differ on two measures of social anxiety. Still, an estimate of total treatment effects showed a steady decrease in social anxiety symptoms with medium to large effect sizes reported by both parents and children [36]. It is well known that parent–child agreement on (specific) anxiety disorders such as SAD [37] or anxiety symptoms [38] is low to moderate only. Although a meta-analysis also reported moderate to large agreement [39], the agreement on social anxiety in single studies is modest at best (e.g. [40-42].). This implies that treatment success may need to take both child and parent perspectives into account [41]. Additionally, to allow for a more differentiated picture after treatment, in this current study we have included both a structured interview and a social stress task to evaluate if state social anxiety (cognitions, behavior, physiology) changes even if this is not reflected in social anxiety reports.

For these reasons, we aimed to examine the effects of exposure-based CBT on children with SAD with both reports of social anxiety and an assessment of social anxiety during a laboratory task. The study was designed as a randomized controlled trial, in which half of the participants were allocated to an experimental group (CBT) receiving immediate treatment and the other half to a waitlist control (WLC) group receiving therapy about 16 weeks later. We tested laboratory and diagnostic data: We expected that compared to the WLC group and the first TSST-C before treatment, children in the CBT group would (a) report more positive and fewer negative cognitions (measured by the Social Interaction Self-Statement Test-Public Speaking, SISST-PS; [43]), (b) perceive their performance as less nervous (measured by the Performance Questionnaire for Children, PQ-C; [44]), and (c) show a change in heart rate. We did not expect differences concerning the affective part of social stress as the TSST-C is a very strong stressor, even inducing high social anxiety in nonclinical samples [45]. Further, (d) two different measures for children were used to examine a decrease in self-reported social anxiety symptoms in the CBT group after receiving treatment (questionnaires). This effect was expected to be confirmed by (e) parent report (questionnaire) and (f) a decrease in the severity index of a clinical diagnosis (interview).

All questionnaire measures were assessed at admission, pre-treatment/waiting, and post-treatment/waiting. Interview and laboratory measures were assessed at admission and post-treatment/waiting. A secondary analysis of stability of treatment effects is reported in the online supplements (S1). In addition, we exploratively assessed self-focused attention and emotion regulation. The pre-post results for these measures may be found in the online Supplements (S2).

This study aimed to extend findings from a previous study [36] on the efficacy of a SAD-specific group CBT by assessing change not only with parent and child reports but also with clinical interviews and a social stress test. It should be cautioned that affective, behavioral, and physiological responses during high social stress did not change as a result of treatment, thus indicating the situation was still highly stressful. However, a significant interaction between time and group appeared for positive cognitions, which resulted in a trend-significant increase of positive cognitions in children in the CBT group from pre- to post-treatment. This finding is in line with previous findings of a positive CBT effect on cognitions after a strong focus on exposure [23] and supports theories of the importance of cognitions in both the stability [50, 69, 70] and the treatment [71, 72] of the disorder. As state social anxiety values indicate, the experience of social stress remained high. Nevertheless—keeping in mind the trend significance—we can assume that children were able to think more positively about the situation (e.g., “It might be embarrassing, but I can cope with the situation”). One could assume that cognitive changes occur before other factors innate to SAD follow, such as the perception of behavior or a physiological response [72]. However, this mediational assumption needs further research. For the physiological response, it should be kept in mind that most research is inconclusive about an objective hyperarousal before treatment and it further tends to support the idea of a biased perception of an increased heart rate [30, 73]. Our findings show that even in the CBT group, the heart rate during social stress still increased compared to a baseline measure. While this might be a normal stress response [45], further studies are needed to clarify the stability of this result. Similar to the physiological results, those on behavioral measures such as social performance indicate that it is more the perception of social performance than the social performance itself that is decreased in children with SAD [33, 61]. Considering these findings, we expected an improvement in perception of social performance. Even though social skills and positive self-feedback were trained in role-playing exercises in a group with peers during CBT treatment, this perception did not change in the CBT group. Our findings do not allow direct assumptions about the cause of this lack of change. It could be that treatment was too short, as only repeated exposure over a long time and—possibly—increased positive social feedback lead to a change in this bias [7].

For diagnostic measures, a significant decrease appeared in child-reported social anxiety as assessed with the SPAI-C, but not when assessed using the SASC-R. Although both questionnaires assess typical dimensions of SAD, they have different foci: The SPAI-C assesses behavioral, physiological, and cognitive features across different types of social situations, while the SASC-R is more narrowly focused on fear of negative evaluation and social distress experience. It may be that—similar to the laboratory response—children rather perceived change in their cognitions than in other associated symptoms that are measured by the SASC-R. Quite a large number of children still met a diagnosis of SAD after treatment, which is in line with previous studies [7]. However, responder analyses based on full remission should be considered carefully because of, for example, their lack of power [74].

Extending the findings of an earlier study [36], the decrease in the severity index of the SAD diagnosis coded by blind interviewers supports the efficacy of the current group treatment. In the months following treatment, anxiety symptoms further decreased continuously on all three questionnaires (see online Supplement 1). In contrast to the earlier study [36], no decrease in dimensional social anxiety symptoms was found for the parent report, but a decrease was found for the child report. One possible explanation could be the slightly greater average age of the children in the current study. Insight into the relationship between anxiety (i.e., emotion) and avoidance (i.e., behavior or coping) is still limited in younger children [75]. Therefore, the significant reduction in child-reported SAD symptoms could be a result of older children’s greater cognitive insight. They might acknowledge both avoidance and anxiety (“I am afraid of others, so I do not talk to them”), and treatment helps them engage in less avoidance (“Even though I am afraid of others, I do talk to them”). While self- and parent-reported anxiety symptoms are both important criteria for treatment success, their reliability can be questioned, as several studies have found inconsistencies between parent- and child-reported anxiety [76]. Still, disagreements might be the result of different perspectives (e.g. [77].); in the current case, children may have already perceived a change in their SAD symptoms while parents had not yet rated these as substantial (possibly because they were not present during treatment). We have to acknowledge that questionnaires allow for only a limited assessment of SAD’s multiple facets. Thus, combined with our laboratory findings, results of the diagnostic measures indicate it might be possible that children still experience anxiety after treatment but have learned to cope better with their anxiety.

Even though CBT—in both group and individual set-up [14]—is the gold standard treatment for anxiety disorders, outcomes for SAD have repeatedly been shown to be inferior to outcomes for other anxiety disorders (e.g. [7].). One possible way to improve outcomes might be to extend treatment to more exposure sessions (see [78]). This recommendation is based on issues concerning the treatment of both temperamental, that is, behavioral inhibition, and interactional difficulties [4]: First, behaviorally inhibited children and youths—even before developing SAD—behave in a socially avoidant manner from an early age and as such might lack experience approaching social situations. Because of this avoidance, they receive limited positive social feedback, and anxiety in unstructured social interactions increases. Second, a general first tendency not to actively approach social situations might lead to a social skills deficit. A vicious circle can develop in which deficits in social skills increase the chance of negative social outcomes. Subsequently, expected negative outcomes of future social situations lead to thoughts of social inadequacy and the avoidance of social situations. Thereby, the development of social skills is further impeded, as few opportunities arise to practice coping with social situations [79]. Even though a social skills deficit is not apparent in all patients with SAD [60], the subjective perception of oneself as being (socially) incompetent might lead to further increases in anxiety and subsequent difficulties in social performance. Thus, given this lengthy developmental process and complex interactional demands, treatment over 12 sessions targeting all deficient elements (social skills deficits, negative cognitions, avoidance of social situations, etc.) can only be seen as an initiation of change. An efficient solution to enhance treatment effects could be a booster session model that allows flexible extension of treatment for those who need more than standard treatment (e.g. [80, 81].). Furthermore, a break after the first 12 sessions would allow treatment effects to stabilize in everyday life. The booster sessions would then provide the possibility to refresh learned skills to avoid relapse to old avoidant behavior.

While our study was carefully planned, several limitations apply. A comparison to individual treatment, not examined in this study, should be examined in future research. Previous studies comparing individual to group CBT did not show a clear preference for either [13, 14]. Still, our aim was not to demonstrate the superiority of group CBT over individual CBT but rather to provide empirical evidence for an efficient group treatment program. Additionally, the TSST-C is a highly potent stressor and, therefore, possibly not the best choice to examine treatment success. Previous studies with adult participants showed that even healthy people do not easily adapt or habituate to a second exposure to the TSST (for an overview see [45]). As mentioned before, it is even more remarkable that our results can be cautiously interpreted into the direction that children with SAD were able to change their cognitive coping with this highly stressful situation. To understand moderators and mediators of change, a follow-up TSST or other social stress task could provide insight on the mechanisms: Possibly, cognitions change after treatment while changes in behavioral, physiological, and affective factors follow several months later.

In treatment research, more randomized controlled trials including experimental designs are necessary to shed further light on the current findings, possibly varying setting (single vs. group therapy) and parental inclusion. Further, measures of success across studies differ widely even when targeting only disorder-specific psychopathology, with both general [22] and specific [21, 82] anxiety questionnaires having been used. The gold standard of a blind diagnostic interview before and after treatment should be applied to both parents and children to include both perspectives in the quality assessment of treatment.

The study aimed to assess CBT treatment success of child SAD not only by social anxiety reports but also by cognitive, behavioral, and physiological components of social stress. Children with SAD participated in a standardized social stress test before and after treatment or a waitlist control period. The CBT group showed a trend toward a significant increase in positive cognitions under social stress after treatment, while these cognitions decreased in the WLC group. No significant results appeared for behavior and physiology. Children in the CBT group, but not parents, further reported less social anxiety in one questionnaire from pre- to post-treatment. A structured interview confirmed a decrease in severity of SAD in the CBT group. While the gold standard of a blind interview showed efficacy of treatment, not all trait and state measures demonstrated similar success patterns. Therefore, this randomized controlled trial of an exposure-based treatment approach in a group setting showed this treatment as partly effective intervention for childhood SAD. A strong focus on exposure produced a trend toward significant change in cognitions during socially stressful situations. However, modifications of both the treatment group setting and the assessment of outcomes, including the use of multiple measures of social anxiety and experimental paradigms, warrant further research. Treatment of SAD needs etiologically based interventions, and possible effective modules in addition to exposure remain to be empirically verified.