SCIENTIST PRACTITIONER
CLINICAL INTERNSHIP RESEARCH MASTER
RESEARCH TEAM
Name : Annelieke Hagen
Student number : 10001503
Organisation : de Bascule
UvA supervisor : Elske Salemink
Date : 01-07-2015
Is Cognitive Bias Modification in the Treatment of Anxiety Disorders a More Effective Tool for Youth Compared to Adults?
Abstract
Recently a new type of intervention for anxiety disorders has been introduced and gained a lot of interest. Cognitive Bias Modification (CBM) is a computerized method used to retrain certain cognitive biases that have been implicated in the development and maintenance of psychopathology. Research so far has revealed mixed findings, which makes it important to investigate in which way and for whom CBM might be a viable treatment option. Theoretical arguments have been reported for the hypothesis that CBM might be more effective in youth populations compared to adult populations. In this paper two recent meta-analyses will be reviewed to investigate whether there is any empirical evidence for this hypothesis. It was concluded that to date there is no clear empirical evidence for this hypothesis. Criticism on CBM research and implications for future research are reviewed and discussed.
Introduction
It was reported that anxiety disorders affect nearly 20 percent of the adult population (Kessler, Chiu, Demler, Merikangas, & Walters, 2005). This ranges from specific phobias to
generalized anxiety disorders. There are efficacious treatment options such as
pharmacotherapy and cognitive behavioural therapy (CBT) to combat anxiety disorders in youth and adults (Hoffman & Mathew, 2008; Mitte, 2005). However, many individuals with anxiety problems do not utilize these treatment options because they have concerns about side effects or are unwilling to engage in exposure therapy (Gunther & Witthal, 2010). In addition many patients remain symptomatic after treatment (Weisberg, Dyck, Culpepper, & Keller,
2007). Therefore it seems important to keep looking for alternative, effective, easy-to-deliver, cost-efficient novel treatments.
One such tool that has gained recent attention is cognitive bias modification (CBM) training. CBM has been defined as the “direct manipulation of a target cognitive bias, by extended exposure to task contingencies that favour predetermined patterns of processing selectivity” (Macleod & Mathews, 2012). Interventions have, to date, mainly focused on modifying attention and interpretation biases. In Cognitive Bias Modification of attention (CBM-A) participants are implicitly trained to direct their attention to positive or neutral stimuli and thereby to avoid negative or threatening stimuli. Cognitive Bias Modification of interpretations (CBM-I) uses more complex stimuli. Often several ambiguous sentences are presented in which a word is left out. The participant is asked to disambiguate the story by filling in the gap with a word that makes the story positive or neutral. For research purposes participants can also be trained to disambiguate the story in a negative way (e.g. as a control condition). CBM-I is based on the premise that the misinterpretation of ambiguous
information in a negative way is one of the core problems in anxiety disorders in adults and youth (e.g. Butler & Mathews, 1983; Salkovskis, 1985; Ouimet, Gawronski, & Dozois, 2009). In order to reduce the stress that these negative misinterpretations can cause, avoidant
behaviour may be performed, which contributes to the development and maintenance of anxiety disorders. CBM-I is a computerized method that attempts to correct the negative interpretation bias of ambiguous information by practicing to interpret it more positively.
CBM has proven to be able to change the targeted bias, although effects of CBM on more clinically relevant outcome measures are mixed (e.g. Menne-Lothmann et al., 2014; Cristea, Mogoase, David, & Cuijpers, 2015a; Cristea, Robin, Kok & Cuijpers, 2015b). To date most CBM research has used adult population samples. However, theoretically there are several reasons why CBM might be more effective in youth than in adults (see Lau, 2013 for an extensive review on this topic). These reasons have at least partly inspired the recent interest in CBM interventions for children and adolescents. First, anxiety in children seems to be acquired through associative learning (Haddad, Lissek, Pine, & Lau, 2011). CBM uses a similar associative learning strategy to train participants to respond in adaptive manners to ambiguous stimuli, thereby making it an ecologically valid treatment method. Secondly, children and adolescents might benefit more from CBM since the levels of neuroplasticity in the young brain makes the brain more receptive to new learning experiences and changes might lead to long-term sustainable gains (Lau, 2013). Hankin et al. (2009) for example have shown that cognitive styles become more stable and trait like across adolescence, suggesting
that these are more flexible and easier to influence before that time. Moreover, prefrontal cortex functions implicated in executive functions including the use of appraisal strategies to control emotional responding, show graduate maturation until early adulthood (Gogtay et al., 2004). Furthermore, CBT relies on dialogue and explicit instructions CBT that requires a certain level of cognitive abilities. Regular CBT is validated and recommended for the ages eight and higher but might be to demanding for younger children (Kendall, Safford, Flannery-Schroeder, & Webb, 2004). CBM might therefore be an effective treatment for younger children due to its implicit nature.
So far research into CBM in youth samples has yielded some promising results. For example Telman, Holmes and Lau (2013) compared positive to negative CBM-I training in a group of healthy adolescents. It was found that the positively trained adolescents selected more benign resolutions of new ambiguous situations and in addition rated recent stressors in their lives as having less impact on their lives than the negatively trained adolescents. In another study it was found that after three sessions of CBM-I, 10 and 11 year old socially anxious children showed more positive interpretations and less social anxiety compared to a test-retest control (Vassilopoulos, Banerjee, & Prantzalou, 2009).
The research field of CBM in general is very young and this is even more the case for CBM in youth samples. There are theoretical arguments that CBM might be more effective in youth than in adolescents. In this paper two recent meta-analyses on the effectiveness of CBM, one using youth samples and one using adult samples, will be reviewed and compared to investigate whether there is any evidence supporting the hypothesis that CBM is more effective in children and adolescents than in adults.
CBM-I in Youth versus Adults
In a recent meta-analysis the efficacy of CBM in adult samples was assessed (Cristea et al., 2015b). Outcome measures were general anxiety, social anxiety and depression. In addition potential moderators were examined. Forty-nine randomized controlled trials (RCT) were included in this meta-analysis in which CBM-I or CBM-A was compared to a control condition.
The mean effect sizes of CBM interventions, when all anxiety measures were taken together (41 comparisons) and for general anxiety separately (34 comparisons) were moderate and significant. Heterogeneity was high. After removing outliers (95% of the confidence interval was outside the 95% confidence interval of the pooled studies) the effect sizes decreased to small effects and heterogeneity was no longer significant. For CBM on social
anxiety (10 comparisons) the mean effect size was moderate and significant. Heterogeneity was high. After removing outliers the effect size was reduced to a non-significant effect. The authors found significant publication bias, after adjusting for missing studies the effect size for social anxiety decreased (g=0.14, 95% CI -0.22-0.51) and became non-significant. Subgroup analyses revealed that CBM for anxiety seemed to be more effective when participants were provided with compensation, than if they were not. Furthermore, higher effect sizes were found for subclinical or analogue samples than for clinical samples in social anxiety. In addition CBM delivered exclusively in the laboratory seemed to be more effective than CBM including home-based sessions. Effect sizes for CBM-I were significantly higher than for CBM-A. A significant negative relationship between publication year and effect size and between number of sessions and effect size was found.
The analyses were repeated with just the studies that used clinical samples. When just including clinical samples the effect sizes were smaller. When also removing outliers and adjusting for publication bias, all effect sizes became non-significant. For general anxiety and depression evidence of publication bias was found. For the analyses with only clinical
samples no subgroup (moderator) analyses were performed since there were too few studies in each outcome category. These findings were unexpected since these CBM trainings are designed with the goal of treatment.
In sum, effect sizes were moderate to small, and mostly non-significant for clinical samples. High heterogeneity and significant publication bias were found. After excluding outliers and adjusting for publication bias, most effect sizes became small and non-significant. It was concluded that there might be small effects of CBM on mental health problems, but that caution must be taken with the implementation of CBM since the included studies were of suboptimal quality, there was high heterogeneity and publication bias.
In the same year the same research group published another meta-analysis (Cristea,et al., 2015a) focusing on CBM in youth samples. Outcome measures were mental health, anxiety, depression and targeted bias. In addition potential moderators were examined. Twenty-three RCT’s were included in the analysis.
The mean effect sizes of CBM interventions in youth samples, when all mental health measures were taken together (23 comparisons), and for anxiety separately, (15 comparisons) were small and non-significant and heterogeneity was moderate. The mean effect size for bias (19 comparisons) was significant; heterogeneity was moderate. Removing one outlier reduced the effect size (still significant), with still moderate heterogeneity.
targeted bias, delivery of the intervention or type of participant. The effect size for targeted bias remained only significant in the healthy subgroup but not in the clinical subgroup. In a subgroup analysis involving delivery in a school setting a small but significant effect was found of CBM-I on mental health outcomes (all measures) and bias. A borderline significant negative relationship between age and mental health was found. No evidence for publication bias was found.
The two meta-analyses did not use the exact same outcome measure and analyses, which makes comparing them more complicated. However, in both meta-analyses a mean effect size is calculated for all mental health outcome measures (anxiety and depression) and separately for anxiety and depression (see Table 1). Effect sizes in adult populations are small to moderate and significant. However they become smaller and sometimes non-significant after the removal of outliers and adjustment for publication bias. The effect sizes in youth populations are small and non-significant for mental health outcomes. A significant effect was found for targeted bias. Both meta-analyses have not shown large effects on clinical outcome measures for CBM. The effect sizes on clinical outcome measures in youth are smaller and non-significant than those in adults. Therefore comparing the effect sizes of these meta-analyses does not provide support for the hypothesis that CBM might be more effective in youth.
Table 1. Comparison of Effect Sizes on Anxiety and Bias for Adult vs. Youth Populations nComp a Hedges’ g Sign. b I2 c Anxiety, all measure-adult
Outliers removed
Mental health, all measures-youth
Outliers removed 41 37 23 22 0.37 (95% CI 0.20-0.54) 0.23 (95% CI 0.14-0.32) 0.12 (95% CI -0.02-0.26) 0.09 (95% CI -0.03-0.22) significant significant n.s. n.s I2=72.66% n.s. I2=34% I2=22% Anxiety (general+social)-adult Outliers removed Anxiety-youth 34 30 15 0.38 (95% CI 0.17-0.59) 0.18 (95% CI 0.08-0.28) 0.17 (95% CI -0.01-0.36) significant significant n.s. I2=76.98% n.s. I2=44% Bias- youth Outliers removed 23 19 0.53 (95% CI -0.33-0.73) 0.48 (95% CI 0.29-0.66) Significant significant (I2=53.40%) moderate a
nComp; amount of comparisons used in the analysis
b
Significant at a p level of 0.05, n.s.=non-significant
c
Percentage of heterogeneity
There were also differences in subgroup analyses between the studies. It must be mentioned that many subgroup analyses could not be performed for the youth samples due to a lack of power. In adult populations the effect sizes were higher when participants were
compensated and for CBM-I compared to CBM-A. In addition a negative relationship between number of sessions and effect size in adult samples was found. These differences were not found in youth samples. In both samples the delivery setting seemed to have an influence on the effect size. In adult samples delivery in the laboratory resulted in higher effect-sizes than delivery with a home-based component. In youth samples delivery in a school setting resulted in higher effect sizes than delivery in another setting. In addition, both in adult and in youth samples effect sizes were higher for subclinical and analogue samples compared to clinical samples.
Summary and Discussion
Research into the effects of CBM-I is a very young but growing field. To date many studies with promising results have been published. It seems possible to change unhealthy
interpretations with CBM, but the effects of CBM on clinical outcome measures are mixed (Menne-Lothmann et al., 2014; Cristea et al., 2015ab). There are theoretical arguments supporting the hypothesis that CBM might be more effective in youth than in adults (Lau, 2013). In this paper two recent meta-analyses have been described and compared to investigate if there is any empirical evidence for this hypothesis.
No evidence for the hypothesis has been found by comparing the two meta-analyses. Actually, larger effect sizes on emotional outcome measures were reported for adult samples than for youth samples. Moderator analyses have resulted in different results for adults than for youth. However, delivery setting and subclinical vs. clinical population have been found to be moderators in both types of population. For youth populations no other moderators have been identified. However, in the youth sample meta-analysis only 23 studies were included which, especially when subdivided over anxiety and depression can cause problems with power. This made it impossible to do many reliable subgroup analyses.
Both meta-analyses are very critical and bring forward the question of whether or not research into CBM as a clinical treatment should be continued. Non-significant small effect sizes were found in youth and after removing outliers and adjusting for publication bias the effect sizes in the adult meta-analysis were reduced to small or non-significant effects as well. Based on these analyses one might be inclined to write off CBM as an ineffective method and suggest to just stick with the effective treatments that already exist (i.e. CBT or
pharmacotherapy). However, there are several aspects about these meta-analyses and the research field that must be considered before doing so.
Significant heterogeneity means that the study results differ by more than just the sampling error (Huedo-Medina et al., 2006). With heterogeneity this high, one might wonder if it makes sense to combine these studies into one meta-analysis. It might not be possible to speak of the effect of CBM on mental health outcome measures because the individual studies differentiate too much in these respects. The authors predicted heterogeneity a priori and therefore used a random effects model for the analyses. However, in the adult meta-analyses the heterogeneity was often reduced and became non-significant after removing one or more outliers. The authors do not give a theoretical justification for removing these studies. The studies are removed solely on the basis of a sensitivity analysis. Of course when a study has a
disproportionate influence on the analysis one should consider leaving it out. However it is important to examine the study in question and see if there is a theoretical explanation to actually remove it. Since the inclusion criteria of these meta-analyses were rather strict it is not justified to remove an outlier as a fluke without any justification to do so. In the adult meta-analysis one of the studies that was removed as an outlier was a study by Lester et al. (2011). In this study a new approach to CBM, designed to manipulate cognitive errors known to characterize anxiety and depression, was used. The researchers tried to improve the
ecological and clinical validity of CBM by developing the modification materials in consultation with clinicians and by trying to modify biased cognitions identified by Beck’s cognitive error categories, which are typically targeted during therapy (Lester et al., 2011). The results indicated that this form of CBM promotes positive interferences, reduces vulnerability to stress and improves self-perceptions of performance, all with high effect sizes. It may be that this study was too easily removed as an outlier while this form of CBM, which is designed specifically for clinical practice instead of research, might be very
promising. In a study of Yiend et al. (2014) CBM for cognitive errors also yielded promising although non-conclusive results in a sample of clinically depressed patients. It is possible that this study was an outlier since it used a different kind of CBM. When inspecting the other studies that were removed as outliers no clear theoretical reason could be found to justify this.
Furthermore, both analyses struggled with power issues. For the youth meta-analysis a power-meta-analysis was performed showing that with the 23 studies a small effect could be found, however it is very likely that there was not enough power to test for moderator effects, since a lot less studies were used in these analyses. In the adult meta-analysis no power-analysis was reported. For at least some of the subgroup analyses power was very likely not high enough to detect small or even moderate effects. For social anxiety in clinical samples only seven comparisons were used. In addition very large confidence intervals were
reported. This is often a problem when sample sizes are relatively small and makes it hard to specifically interpret outcomes (Higgins & Green, 2011). Moreover, since the author expected heterogeneity a priori they used a random effects model. Random effects models require more data to achieve the same statistical power as fixed effects models (Ades, Lu, & Higgins, 2005). This makes it even more likely that the analyses (especially the subgroup analyses) were underpowered. So even though potential moderators were investigated it is possible that existing effects were not found.
Even so Cristea et al. (2015b) found that CBM in adults is more effective in subclinical than in clinical populations and the same was found in the youth meta-analysis (Cristea et al., 2015a). This may suggest that CBM might be more useful as a preventive tool than as a treatment. In adults, training in the lab was more effective than training at home but the reasons for this (concentration, motivation, computer malfunctions, distractions?) so far remain unclear. In youth, training in a school context was shown to be effective while no effects were found for other training settings. Since many of the tasks used in CBM training paradigms were originally developed as measuring tasks only, it is plausible that the tasks are boring when used in a prolonged version as a training. Often CBM training sessions are quite long and monotonous. It could be that participants in less controlled settings will lose their attention and do not train seriously which could explain the moderating effect of setting.
Finally, in the adult meta-analysis a negative relationship was found between number of sessions and effect sizes. This is surprising and is in contradiction with earlier meta-analyses. Menne-Lothmann et al. (2014) for example found a positive relationship between number of sessions and effect sizes. It is possible that in the RCT’s included in the current meta-analysis there is indeed a negative relationship, however, it is suggested that one
interprets these results with caution since most of the studies include used only one session of CBM which makes it plausible that statistic power in this analysis was low.
In sum, this comparison of two recent meta-analyses did not reveal evidence for the hypothesis that CBM might be more effective in youth than in adults. Instead it brought up the question of whether or not CBM is as interesting of a research venue as was previously thought. The two analyses that were discussed were very critical. However, the meta-analyses themselves struggled with some statistical (power, heterogeneity, outliers) issues themselves. Moreover, one should remember that the field of CBM research is relatively young. It should be acknowledged that only a limited amount of different CBM methods have been developed and tested to date. It can be argued that CBM, which is grounded in sound theory of interpretation (and other) biases, which are strongly implicated in psychopathology,
has a lot of potential for research purposes and clinical practice, even though it still has to be realized. Instead of writing it off as an ineffective method, adequate research should be executed in order to investigate in which situations and for whom CBM can be a valuable paradigm.
Future directions
The discussed meta-analyses are very critical, however, it seems premature to discard CBM as a possible treatment for anxiety disorders. Many studies have shown the potential of CBM, but there is still room for improvement and need for extensive clinical research into if, how and for whom CBM might be a valuable treatment option. In the final section of this paper a few pointers to improve future research in the field of CBM will be given.
First of all, null findings and unexpected findings from methodological strong studies should be published. Meta-analyses (e.g. Cristea et al., 2015b) have reported publication bias in the field of CBM. This could lead to an overly optimistic view of the effects of CBM. More importantly it can obscure relevant information as to which factors influence the efficacy of CBM.
Secondly, a better understanding of, if and how the change of in bias mediates the effect of CBM on clinically relevant outcome measures, is warranted. Many studies have found a mediating effect of change in bias (e.g. Amir, Bomyea, & Beard, 2010; See, MacLeod, & Bridle, 2009). A clear understanding of mediating processes could help guide research and establish the scientific plausibility of CBM. If change in bias mediates the effects of CBM it could be interesting to only include studies in a meta-analysis in which the bias was successfully changed. It must however be mentioned that in an effect of bias was found in the youth meta-analysis (Cristea, et al., 2015a) while no effects on mental health were found. This implies that the mediating process might not be as clear-cut as the theory implies. More and more extensive research will have to shed light on this, since there are not many studies to date investigating the mediating processes in youth.
Third, many different paradigms of CBM are presently used. It is plausible that not all paradigms are equally effective. This hypothesis is supported by the meta-analysis of Cristea et al. (2015b) in which larger effects were found for CBM-I compared to CBM-A. Therefore the search for the most effective way of administering CBM should continue. When doing so one should keep in mind that most tasks currently used in CBM training were originally designed as measurement tasks. To train with these tasks for a prolonged period of time could be very boring. So in the search for an effective way to administer CBM one should try to
design the trainings in a way that is motivating for the participant. Currently researchers are trying to design more interesting ways of training for example by including game elements (Boendermaker, Prins, & Wiers, in press).
Furthermore, the effectiveness of CBM has often been assessed as a ‘stand alone’ treatment. However, it is interesting to see if it could enhance treatment as usual (e.g. CBT) as a tag on treatment. The study of Lester et al. (2011), discussed earlier in this paper, used a form of CBM in which many topics of regular CBT are incorporated and it yielded very high effect sizes. So combining CBM and CBT might be an interesting research venue.
Finally, one of the big advantages of CBM could be its implicit nature. Some very basic computer skills are needed to complete the training sessions but far less verbal and cognitive abilities are needed to complete CBM training compared to participating in CBT. Whereas CBT is mostly recommended for children of ages eight or higher with adequate verbal capacities, CBM could be a viable treatment option in populations not meeting these cognitive criteria. Research in young populations or populations with insufficient verbal capacities could therefore be very interesting.
References
Ades, A. E., Lu, G., & Higgins, J. P. (2005). The interpretation of random-effects meta- analysis in decision models. Medical Decision Making, 25(6), 646-654.
Amir, N., Bomyea, J., & Beard, C. (2010). The effect of single-session interpretation modification on attention bias in socially anxious individuals. Journal of Anxiety Disorders, 24, 178-182.
Boendermaker, W. J., Prins, P. J. M., & Wiers, R. W. (2015). Cognitive Bias Modification for adolescents with substance use problems- Can serious games help? Journal of
Behavior Therapy and Experimental Psychiatry, in press.
Butler, G., & Mathews, A. (1983). Cognitive processes in anxiety. Advances in Behaviour Research and Therapy, 5, 51-62.
Cristea, I. A., Mogoase, C., David, D., & Cuijpers, P. (2015a). Practitioner review: Cognitive bias modification for mental health problems in children and adolescents: a meta-analysis. Journal of Child Psychology and Psychiatry, doi: 10.1111.jcpp. 12383. Cristea, I. A., Robin, N., & Cuijpers, P. (2015b). Efficacy of cognitive bias modification
interventions in anxiety and depression: meta-analysis. British Journal of Psychiatry, doi: 10.1192/bjp.bp.114.146761.
Gogtay, N., Giedd, J. N., Lusk, L., Hayashi, K. M., Greenstein, D., Vaituzis, A. C., …. (2004). Dynamic mapping of human cortical development during childhood through early adulthood. Proceedings of the National Academy of Sciences of the United States of America, 101(21), 8174-8179.
Gunter, R. W., & Whittal, M. L. (2010). Dissemination of cognitive-behavioral treatments for anxiety disorders: Overcoming barriers and improving patient access. Clinical
Psychological Review, 30, 194-202.
Haddad, A. D., Lissek, S., Pine, D. S., & Lau, J. Y. (2011). How do social fears in adolescence develop? Fear conditioning shapes attention orienting to social threat cues. Cognition and Emotion, 25(6), 1139 -1147.
Hankin, B. L., Oppenheimer, C., Jenness, J., Barrocas, A., Shapero, B. G., & Goldband, J. (2009). Developmental origins of cognitive vulnerabilities to depression: review of processes contributing to stability and change across time. Journal of
Clinical Psychology, 65(12), 1327 -1338.
Higgins, J. P. T., & Green, S. (Eds.). (2011). Cochrane handbook for systematic reviews of interventions: Version 5.1.0. The Cochrane Collaboration, 2011. Available from www.cochrane-handbook.org.
Hoffman, E. J., & Mathew, S. J. (2008). Anxiety disorders: a comprehensive review of pharmacotherapies. Mount Sinai Journal of Medicine, 75, 248-262.
Huedo-Medina, T. B., Sanchez-Meca, J., Marin-Martinez, F., & Botella, J. (2006). Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychological Methods, 11(2), 193-206.
Kendall, P. C., Safford, S., Flannery-Schroeder, E., & Webb, A. (2004). Child anxiety treatment: Outcomes in adolescence and impact of substance use and depression at 7.4-year follow-up. Journal of Consulting and Clinical Psychology, 62, 276-287. Kessler, R. C., Chiu, W. T., Demler, O., Merikangas, K. R., & Walters, E. E. (2005).
Prevalence, severity and comorbidity of 12-month DSM-IV disorders in the national comorbidity survey replication. Archives of General Psychiatry, 62, 617-627.
Macleod, C., & Mathews, A. (2012). Cognitive bias modification approaches to anxiety. Annual Review of Clinical Psychology, 8, 189-217.
Menne-Lothmann, C., Viechtbauer, W., Höhn, P., Kasanova, Z., Haller, S. P., Drukker, M., … Lau, J. Y. F. (2014). How to boost positive interpretations? A meta-analysis of the effectiveness of Cognitive Bias Modification for Interpretation. PloS one, 9, e100925. Mitte, K. (2005). Meta-analysis of cognitive–behavioral treatments for generalized anxiety
disorder: a comparison with pharmacotherapy. Psychological Bulletin, 131, 785–795. Lau, J. Y. F. (2013). Cognitive bias modification of interpretations: A viable treatment for
child and adolescent anxiety? Behaviour Research and Therapy, 51, 614-622. Lester, K. J., Mathews, A., Davison, P. S., Burgess, J. L., & Yiend, J. (2011). Modifying
cognitive errors promotes cognitive well being: A new approach to bias modification. Journal of Behavior Therapy and Experimental Psychiatry, 42, 298-308.
Ouimet, A. J., Gawronski, B., & Dozois, D. J. (2009). Cognitive vulnerability to anxiety: A review and an integrative model. Clinical Psychological Review, 29, 459-470. Salkovskis, P. M. (1985). Obsessional-compulsive problems: A cognitive-behavioural
analysis. Behaviour Research and Therapy, 23, 571-583.
See, J., MacLeod, C., & Bridle, R. (2009). The reduction of anxiety vulnerability through the modification of attentional bias: a real-world study using a home-based cognitive bias modification procedure. Journal of Abnormal Psychology, 118, 65-75.
Vassilopoulos, S., P., Banerjee, R., & Prantzalou, C. (2009). Experimental modification of interpretation bias in socially anxious children: Changes in interpretation, anticipated interpersonal anxiety, and social anxiety symptoms. Behaviour Research and Therapy,
Weisberg, R. B., Dyck, I., Culpepper, L., & Keller, M. B. (2007). Psychiatric treatment in primary care patients with anxiety disorders: a comparison of care received from primary care providers and psychiatrists. American Journal of Psychiatry, 164, 276-282.
Yiend, J., Lee, J., Tekes, S., Atkins, L., Mathews, A., Vrinten, M., Ferragamo, C., & Shergill, S. (2014). Modifying interpretation in a clinically depressed sample using ‘Cognitive Bias Modification-Errors’: A double blind randomised controlled trial. Cognitive Therapy and Research, 38, 146-159.
