Introduction
Social anxiety disorder is highly prevalent in adolescents and is reported as one of the most common forms of social distress in this population, also in adolescents with Mild Intellectual Disabilities (MID; Dekker and Koot
2003; Kessler et al.
2005). Social anxiety disorder often develops during adolescence and is marked by the persistent fear of social or performance situations (American Psychiatric Association
2013). Although treatments for child- and adolescent anxiety have shown good efficacy, at least 40% of children and adolescents continue to have a diagnosis after treatment (James et al.
2013). In particular, research suggests that socially anxious children and adolescents have the poorest outcomes following treatment when compared to other anxiety disorders, and they are only half as likely to remit as children and adolescents with other anxiety disorders, resulting in lifelong impairment (e.g., Hudson et al.
2010). Moreover, even though the prevalence rates of social anxiety are comparable in adolescents with MID (Dekker and Koot
2003), studies related to the treatment of anxiety disorders in adolescents with MID are scarce. However, the general conclusion is that treatments are even less effective in individuals with MID compared to individuals with an average IQ (see also Dagnan and Jahoda
2006). Generally, therapies like cognitive behavioral therapy, are often too complex and demanding for people with MID (De Wit et al.
2012). To develop more effective treatments for social anxiety disorder appropriate for adolescents with MID, it is important to find theoretical and practical innovations that might take current treatments into new directions.
Several underlying processes have been defined as maintaining and possibly causing anxiety disorders, including cognitive processes (for a review, see Mathews and MacLeod
2005). According to cognitive theories of anxiety disorders (e.g., Beck et al.
1985), socially anxious adults and children have anxiety-related schemata that direct processing resources towards threat-relevant information resulting in cognitive biases related to attention, interpretation and memory (e.g., Clark and Wells
1995; Rapee and Heimberg
1997). Numerous studies found evidence for the existence of cognitive biases, such as attention bias and interpretation biases, in anxious adolescents (for an overview, see Hadwin and Field
2010). A recent meta-analysis with regard to attention bias found a small positive association between anxiety and attention bias (
d = 0.21; Dudeney et al.
2015). Furthermore, they found a moderating effect of age; the relation between attention bias and anxiety increased with age. A meta-analysis with regard to interpretation bias found a medium positive association between anxiety and interpretation bias (
d = 0.62; Stuijfzand et al. in press). Furthermore, they found a moderating effect of the content of ambiguous scenarios in their meta-analysis; the relation between anxiety and interpretation bias was stronger when the ambiguous scenarios matched the anxiety subtype under investigation. However, the authors pointed out that this effect was mainly driven by studies that focused on social anxiety.
Even though there are numerous studies with regard to attention- and interpretation bias in typically developing children and adolescents, to the best of our knowledge, no studies were published on the relation between anxiety and attention bias in adolescents with MID, and only a few published studies that found evidence for the relation between anxiety and interpretation biases in adolescents with MID (Houtkamp et al.
2017; Van der Molen and Salemink
2016), For example, Van der Molen and Salemink (
2016) assessed interpretation bias using ambiguous scenarios in adolescents with varying levels of IQ (
min = 55,
max = 129,
Mean = 85). Results showed a medium positive significant relation between anxiety and interpretation bias, where higher levels of anxiety are associated with stronger threat-related interpretations, while IQ did not moderate this effect.
To investigate the role of threat-related interpretation biases in adolescents with MID in more detail and the potential to alleviate symptoms by reducing biases, the current study focused on Cognitive Bias Modification for Interpretation (CBM-I) training. In CBM-I, participants learn to restructure the way they interpret ambiguous related situations with the goal to reduce threat-related interpretation biases and reduce levels of anxiety symptoms (for meta-analyses in adolescents, see Cristea et al.
2015b; Krebset al. in press). When the first meta-analysis on the effects of CBM for interpretation and attention in adolescents came out (Cristea et al.
2015b), the conclusions were rather unfavorable towards CBM: CBM did not affect anxiety. This conclusion shed doubt on the clinical relevance of CBM techniques, while noticing that most studies were suboptimal. More recently, however, Krebs et al. (in press) conducted another meta-analysis, now specifically focusing on the effects of CBM-I, and found a moderate effect on negative- and positive interpretation bias (negative:
g = − 0.70/positive:
g = − 0.52), and a small, but significant effect on anxiety directly following training (
g = − 0.17). One of the reasons for these different findings is that Krebs et al. (in press) only included studies that focused on CBM-I. This might be an advantage, as there is some evidence that that CBM for interpretation bias is more effective than CBM for attentional bias (see also Cristea et al.
2015a; Lau
2015). Furthermore, a recent study by Grafton et al. (
2017) re-analyzed the meta-analysis by Cristea et al. (
2015b) showing that indeed CBM procedures do not always have an impact on mental health concerns. However, this is only correct when the cognitive bias has not changed during the CBM training. When CBM procedures successfully modify cognitive biases, this often results in a significant reduction in mental health concerns. In sum, several studies see the potential of CBM-I, but also acknowledge the fact that more research is needed and that several improvements are to be made (see also Krebs et al. in press). Furthermore, to the best of our knowledge, there are no published studies on the effects of CBM-I in adolescents with MID. However, CBM-I is a promising training for adolescents with MID, as the training does, compared to regular therapies, not appeal on reflection, meta-cognition, or other, for adolescents with MID, cognitive too demanding exercises.
In order to use CBM-I in adolescents with MID and to make improvements in our CBM-I procedures in general, it is important to focus on the details of the different studies that have been conducted thus far and on the recommendation made by these studies. Overall, interpretation bias training has proven to be capable of reducing biases in adolescents (e.g., Salemink and Wiers
2011) with long-term effects (De Hullu et al.
2017). Effects on anxiety (e.g., Reuland and Teachman
2014) and stress appraisal (e.g., Lau et al.
2012) have been less robust. CBM-I has shown to be specifically effective in adolescents with lower levels of cognitive control or working memory capacity (Salemink and Wiers
2012). As precisely adolescents with MID have difficulties in making use of their working memory (Van der Molen et al.
2010), CBM-I might be particularly relevant here as an alternative approach in treatment. Furthermore, the largest effect of CBM (on bias and symptoms) was found when training was performed within the school setting rather than, for example, via internet (Cristea et al.
2015b). Several ways forward have been formulated to improve CBM research. Hirsch et al. (
2016), for example, suggested that future studies should focus on disorders in which the negative resolving of ambiguity is a key aspect, such as social anxiety disorder, while paying attention that the used scenarios are idiosyncratic and address the disorder-specific ambiguity at the same time (see also Klein et al.
2015).
Many of the details and recommendations listed above were addressed in the current study that tested the efficacy of a CBM training in adolescents with MID. This study (1) focused solely on interpretation, (2) included five training sessions, and (3) used carefully selected stimuli focusing on social anxiety, (4) in a classroom setting. Based on earlier CBM-I studies in anxious adolescents with an average IQ, we hypothesized that adolescents in the positive training group would show significant reductions in interpretation biases and self-reported social anxiety after training, which we did not expect for adolescents in the neutral control-group.
Discussion
The goal of this study was to examine the effects of a Cognitive Bias Modification training for Interpretation (CBM-I) in socially anxious adolescents with MID. As predicted, we found that adolescents in the positive training group showed significant reductions in interpretation bias after training, which was not found in the control-training group. We found evidence for an effect of near and far transfer of learning; Adolescents in the positive training group showed a significant reduction on the IREC-T, which was very similar to the training task, as well as on the AST, which measured interpretation bias using a different format. Furthermore, as expected, we found a significant reduction on self-reported social anxiety in the positive group, but only after 10-weeks-follow-up and not directly following training. To the best of our knowledge, this is the first study in adolescents with MID that shows that interpretation biases and social anxiety can be reduced in highly socially anxious participants by a CBM training that is characterized by content-specific training materials, with one-on-one instruction within a school setting, and multiple training sessions. These results are in line with other studies on adolescent CBM-I (e.g., Hirsch et al.
2016) and reflect recent meta-analyses (Cristea et al.
2015a,
b; Grafton et al.
2017; Krebs et al. in press; for a comment on see Lau
2015) that suggest that CBM-I in adolescents might be effective under specific circumstances.
As learning difficulties are a central aspect of MID, we think it is remarkable that we were able to modify cognitive habits through a simple one-on-one CBM training paradigm provided by trained research assistants. Even though clearly more research is needed, this study indicates that a relatively simple technique can be used to lower anxiety symptoms in adolescents with MID. Moreover, CBM-I may be particularly beneficial, as previous studies show that regular treatments might be less effective to individuals with MID (see also Dagnan and Jahoda
2006). Our results are in line with the findings of Salemink and Wiers (
2012) who found that CBM-I was more effective in adolescents with lower levels of cognitive control or working memory capacity. As precisely adolescents with MID have difficulties in making use of their working memory (Van der Molen et al.
2010), our findings might implicate that cognitive bias modification (CBM) procedures are suitable for populations with MID. It might be interesting for future studies to examine the effect of a combined therapy including regular treatments such as cognitive behavioral therapy (CBT) and CBM procedures. It might well be that combining CBT with CBM increases the efficacy of CBT in adolescents with MID as adolescents with MID might simply need more practice than in standard CBT treatments. By combining CBT with CBM, adolescents with MID get more help in practicing new, helpful thoughts over and over again, without the support of a therapist.
Whereas we found training effects on both interpretation bias measures, it should be noted that only the effect of near transfer of learning (IREC-T) was maintained in the long term but not the effect of far learning (AST). These results might indicate that adolescents did learn how the training worked, they ‘learned the trick’, but that they potentially need more practice to also positively interpret ambiguity in different situations on the longer term. The fact that the effects on the AST disappeared on the longer term might indicate that booster sessions may be necessary to sustain the training effect. Moreover, the effects on self-reported social anxiety were only visible on the longer term, and not directly following training. In line with cognitive theory, anxiety would be reduced only after repeated exposure to ambiguous situations in which more positive interpretations are applied. Since self-report measures of anxiety probe experienced recently anxiety, effects could only be expected after the positive interpretation style has been applied over a longer period of time (see also Harmer et al.
2009). These results show that it is important to include a follow-up measurement. However, this does not match well with the results on the AST, where we did not find a significant effect from pre- to 10-weeks follow-up. One would expect that applying a new interpretation style implies that the effect on the AST should have still been there at 10-week follow-up. We have no clear explanation as to why the effects of the AST were not maintained at 10-week follow-up. It could be that the AST taps into a different interpretation process than what has been learned during training (see also Klein
2016), but this should clearly be investigated in a larger follow-up trial including booster sessions and a longer follow-up measurement.
A few limitations of our study should be mentioned. First, we included adolescents who scored above the clinical cut-off score on social anxiety, but we did not administer a diagnostic interview to find out if adolescents had a clinical diagnosis or not. Second, we were unfortunately not able to measure IQ during our study due to time limitations. In addition, we were also not allowed to collect the IQ reports from the schools, due to privacy restrictions. We therefore cannot say anything about the relation between the CBM-I outcomes and IQ. A recent study of Van der Molen and Salemink (
2016) studied interpretation bias in a sample of adolescents with varying IQ scores ranging in the MID range but also in the normal range, and they did not find an effect of IQ on the relation between interpretation bias and social anxiety. Also, the results of our study are comparable to samples including adolescents with an average IQ (e.g., Klein et al.
2015). Nevertheless, more research is needed that includes IQ as a moderator in the relation between anxiety and interpretation bias, and the effect of CBM-I before clear conclusions can be drawn. Third, we included a 10-week follow-up measurement, but we did not include a follow-up after a longer period, such as 6- or 12 months. Therefore, we cannot make predictions about the long-term effect of CBM-I on interpretation bias and anxiety symptoms in this population. Finally, our sample was relatively small and results showed small effects. Large-scale (clinical) trials including are needed before firm conclusions can be drawn with regard to the (clinical) implications of this CBM-I training.
In conclusion, this is the first study that examined the possible benefits of a CBM-I training on the decrease of anxiety symptoms in an understudied and often overlooked sample, namely adolescents with MID. In the current study, we addressed many of the theoretical and methodological limitations that were noted in recent meta-analyses of CBM (Cristea et al.
2015b) and CBM-I (Grafton et al.
2017; Krebs et al. in press). In a small but carefully designed randomized controlled trial, we showed that it is possible to modify interpretation biases in adolescents with mild intellectual disabilities with a CBM-I training, using social-anxiety specific stimuli, delivered in a controlled environment at school. This simple intervention resulted in a decrease in negative interpretations and in social anxiety 10-weeks following training. Even though clearly more research is needed, this first study in adolescents with MID shows the potential benefits of CBM-I for this population. CBM-I might be particularly beneficial combined with regular CBT, as adolescents with MID are able to practice many possible situations in a standardized environment without the need for a therapist.