An attempt to target anxiety sensitivity via cognitive bias modification

Machine learning meta-analysis identifies individual characteristics moderating cognitive intervention efficacy for anxiety and depression symptoms

Cognitive training is a promising intervention for psychological distress; however, its effectiveness has yielded inconsistent outcomes across studies. This research is a pre-registered individual-level meta-analysis to identify factors contributing to cognitive training efficacy for anxiety and depression symptoms. Machine learning methods, alongside traditional statistical approaches, were employed to analyze 22 datasets with 1544 participants who underwent working memory training, attention bias modification, interpretation bias modification, or inhibitory control training. Baseline depression and anxiety symptoms were found to be the most influential factor, with individuals with more severe symptoms showing the greatest improvement. The number of training sessions was also important, with more sessions yielding greater benefits. Cognitive trainings were associated with higher predicted improvement than control conditions, with attention and interpretation bias modification showing the most promise. Despite the limitations of heterogeneous datasets, this investigation highlights the value of large-scale comprehensive analyses in guiding the development of personalized training interventions.

Effects of Training Body-Related Interpretations on Panic-Related Cognitions and Symptoms

Background

Interpretation biases (IBs) are central in panic disorder, and there is rich evidence showing that these are correlated with and predictive of panic-relevant symptomatology. However, experimental studies are needed to examine the potential causal effects of IBs, as predicted by cognitive models.

Methods

Panic-related IBs were manipulated via a sentence-completion Cognitive Bias Modification-Interpretation (CBM-I) training. The sample included N = 112 healthy participants reporting moderate levels of fear of bodily sensations. Participants were randomly allocated to a positive, negative, or control CBM-I condition. To test the trainings' effect on panic-relevant cognitive processing, IBs were assessed via proximal and distal measures. Symptom provocation tasks were applied to test transfer to panic-relevant symptomatology.

Results

Results on the proximal measure showed that positive CBM-I led to more positive IBs compared to negative, and control training. Further, positive CBM-I led to more positive IBs on the distal measure as compared to negative CBM-I. However, there were no differential training effects on panic-related symptomatology triggered via the provocation tasks.

Conclusion

The findings indicate a limited generalization of the effects of CBM-I on IBs and panic-related symptoms. Potential means to improve generalization, such as applying more nuanced measures and combining CBM-I with psychoeducation are discussed.

Anxiety Sensitivity as a Malleable Mechanistic Target for Prevention Interventions: A Meta-Analysis of the Efficacy of Brief Treatment Interventions

Anxiety sensitivity (AS) is a transdiagnostic risk factor and potential treatment target for prevention of associated psychopathology and negative health behaviors. We conducted a meta-analysis evaluating the efficacy of brief interventions in at-risk samples for reducing AS and associated clinical/behavioral outcomes (e.g., depression, alcohol use) across 28 studies (1,998 participants). AS targeted interventions, compared to control conditions, evidenced a significant moderate effect size for alleviating AS from pre- to post-treatment (d = 0.54) and approached a large effect size from pre-treatment to short-term follow-up (d = 0.78). The effect size for long-term follow-up did not reach significance (d = 0.29). For clinical/behavioral outcomes, AS interventions demonstrated significant small-to-moderate effect sizes for the three timepoints examined (d's = 0.20-0.41). Our findings help validate AS as a modifiable mechanistic target for prevention efforts.

A randomized controlled trial evaluating the efficacy of a brief computerized anxiety sensitivity reduction intervention for health anxiety

It is estimated that individuals with severe health anxiety (HA) utilize 41 %-78 % more healthcare resources than individuals with identified medical diagnoses. Thus, identifying targets for intervention and prevention efforts for HA that are appropriate for primary care or specialty clinic settings is imperative. The aim of the present investigation was to evaluate the effect of a single-session, computerized anxiety sensitivity (AS) intervention on AS and HA. Participants were 68 university students (79.4 % female; M_age = 19.68) with elevated levels of AS and HA. Participants were randomized to either the AS intervention condition or an active control condition and completed self-report and behavioral follow-up assessments at post-intervention, 1-week follow-up, and 1-month follow-up. Results indicated a significant Time x Condition interaction for ASI-3 at each follow-up assessment (all ps < .001), such that individuals in the active condition exhibited greater reductions in AS compared to the control condition. There was no significant Time x Condition interaction for HA at any follow-up. Mediation analyses revealed a significant indirect effect of Condition on changes in HA through changes in AS. No significant effects were observed for behavioral outcomes. Findings suggest that this intervention successfully reduces AS among those who are high in HA and AS and may indirectly contribute to reductions in HA over time through reductions in AS.

Personalized cognitive training: Protocol for individual-level meta-analysis implementing machine learning methods

Accumulating evidence suggests that cognitive training may enhance well-being. Yet, mixed findings imply that individual differences and training characteristics may interact to moderate training efficacy. To investigate this possibility, the current paper describes a protocol for a data-driven individual-level meta-analysis study aimed at developing personalized cognitive training. To facilitate comprehensive analysis, this protocol proposes criteria for data search, selection and pre-processing along with the rationale for each decision. Twenty-two cognitive training datasets comprising 1544 participants were collected. The datasets incorporated diverse training methods, all aimed at improving well-being. These training regimes differed in training characteristics such as targeted domain (e.g., working memory, attentional bias, interpretation bias, inhibitory control) and training duration, while participants differed in diagnostic status, age and sex. The planned analyses incorporate machine learning algorithms designed to identify which individuals will be most responsive to cognitive training in general and to discern which methods may be a better fit for certain individuals.

Investigating the therapeutic potential of cognitive bias modification for high anxiety sensitivity

BACKGROUND AND OBJECTIVES

Individuals with high anxiety sensitivity (AS) display negative interpretive biases in response to uncomfortable but nondangerous physical sensations. Research suggests that modifying interpretation biases associated with AS leads to changes in AS. The present study sought to replicate and extend this research by addressing limitations of previous studies, increasing the amount of training and adding a follow-up period.

METHOD

Participants high in AS were randomly assigned to four sessions of computerized interpretation bias modification (CBM-I) training or four sessions of computerized "sham" training (control condition) over a 2-week period. The outcomes were AS, interpretive biases, and reactions to induced physical sensations. Assessments occurred at baseline, during training, immediately after the final training session, and 2 weeks after the final training; number of re-assessments varied by outcome.

RESULTS

The CBM-I condition did not outperform the control condition. At the end of the training period, the CBM-I condition displayed limited reductions in AS and interpretation biases. There were no changes in reactions to induced physical sensations. Similar results were found in the control condition for all outcomes.

LIMITATIONS

The control task and the dose of training may have obscured potential effects of CBM-I.

CONCLUSIONS

When considered within the context of previous research, the experimental effects and therapeutic potential of CBM-I for high AS appear to be minimal. However, methodological questions need to be resolved before such a conclusion can be considered definitive.

Negative memory bias as a transdiagnostic cognitive marker for depression symptom severity

BACKGROUND

Negative memory bias is a strong risk factor for the development and maintenance of depression. Recent evidence also found negative memory bias in other mental disorders. Here, we aim to: 1) assess the presence and strength of negative memory bias in a range of (comorbid) mental disorders, 2) investigate which disorder-specific symptoms are associated with negative memory bias, and 3) test whether negative memory bias might be a transdiagnostic mechanism.

METHODS

Negative memory bias was measured in patients with at least one diagnosis of a stress-related disorder (n = 86), a neurodevelopmental disorder (n = 53), or both (n = 68), and 51 controls. Depression, anxiety, attention-deficit/hyperactivity disorder, and autism spectrum disorder symptom severity was assessed using questionnaires. Groups were compared on negative memory bias and the associations between negative memory bias and symptom severity were made using linear regression models.

RESULTS

All patient groups showed stronger negative memory bias than the controls. Negative memory bias was individually associated with all symptom severity indices, but when added into a single model, only the association with depressive symptom severity remained. This persisted after controlling for diagnostic group.

LIMITATIONS

Due to the cross-sectional sectional study design, we could only look at the associations between negative memory bias and disorder-specific symptoms and not at the direction of the effects.

CONCLUSIONS

Negative memory bias is characteristic of a depressotypic processing style and present in different mental disorders. It might play a mechanistic role in the development of (subclinical) co-occurrence between mental disorders.

A randomised controlled trial of computerised interpretation bias modification for health anxiety

BACKGROUND AND OBJECTIVES

Health anxiety is characterised by a tendency to interpret benign bodily sensations as a sign of serious illness. This study aimed to examine whether computerised cognitive bias modification (CBM-I) training to facilitate benign interpretations of bodily symptoms could reduce these negative interpretations of bodily symptoms, health anxiety and comorbid symptoms.

METHODS

Participants (N = 89) with clinical levels of health anxiety (Short Health Anxiety Inventory [SHAI] scores ≥20) were randomised to receive two internet-delivered 30-min sessions of either CBM-I interpretation training (Intervention) or control CBM-I training over 1-week. Participants were assessed at pre-treatment, post-treatment and 2-weeks follow-up on self-reported health anxiety, cognitions and attributions of bodily symptoms, and comorbid symptoms (e.g., depression).

RESULTS

Results showed significantly reduced catastrophic attributions, health anxiety and related symptoms at post-treatment in both groups, which were maintained at 2-week follow-up. Although there were no significant group differences at post-treatment or follow-up, we found small non-significant effect sizes in favour of the CBM-I Intervention group on health anxiety and cognitions (Cohen's d's between-group effect sizes of 0.28 at post-treatment and d = 0.35 at follow-up on the 18-item Short Health Anxiety Inventory).

LIMITATIONS

The study was conducted online, limiting control over the setting in which participants completed training. Additionally, training was not tailored to the specific symptoms or diseases that participants feared.

CONCLUSIONS

This study is the first to evaluate the effect of internet-delivered CBM-I training targeting benign interpretations of bodily symptoms on health anxiety. Further research is needed before this intervention is disseminated.

Associations among interpretation bias, craving, and abstinence self-efficacy in adults with substance use disorders

OBJECTIVE

Interpretation bias is a crucial therapeutic target in emotional disorders. However, few studies have examined the role of interpretation bias in substance use disorders (SUDs). Our specific aims were: (1) to examine whether interpretation bias was associated with craving and abstinence self-efficacy, and (2) explore potential moderators of these associations, including anxiety severity, sex, and substance type.

METHODS

Adults attending an inpatient SUD treatment program (N = 224; mean age = 38.95; 67% male/33% female; 68% primary alcohol use disorder/29% primary opioid use disorder) completed the Word-Sentence Association Paradigm (WSAP) with ambiguous situations related to general anxiety domains (e.g., daily stress, health, relationships), as well as measures of craving (Craving Scale), abstinence self-efficacy (Brief Situational Confidence Scale), and anxiety symptoms (Overall Anxiety Severity and Impairment Scale [OASIS] and Anxiety Sensitivity Index-3).

RESULTS

Negative interpretation bias was modestly associated with more craving (r = .23, p =  .001) and less confidence to resist using substances (r = -0.23, p =  .001). In multiple linear regression models that included the anxiety measures, interpretation bias was the most robust predictor of craving and abstinence self-efficacy. Sex (N = 224) and substance type (opioid vs. alcohol; n = 219) did not moderate these relationships.

CONCLUSIONS

These findings suggest that interpretation bias might be an important individual difference within SUD populations.

A systematic review of the word sentence association paradigm (WSAP)

BACKGROUND AND OBJECTIVES

The Word Sentence Association Paradigm (WSAP) was originally designed to assess and modify interpretive biases (IB) in socially anxious individuals. Researchers have since modified the WSAP for use across various populations. Despite its widespread use, no studies have systematically reviewed the WSAP to determine its validity and reliability.

METHODS

We review variations to the WSAP, populations in which the WSAP has been used, reliability data, and effect sizes across 41 studies published between 2008 and March 2018.

RESULTS

Results indicate that the WSAP has been utilized to target 18 disorders and symptoms in adults and children. Modifications include stimulus content, timing parameters, and presentation order of word and sentence pairs. Reported internal consistency and test-retest reliability suggest good to excellent reliability. Medium to large effect sizes were reported when comparing control samples to those with psychopathology and in pre-post comparisons of the modification version of the WSAP.

LIMITATIONS

Studies varied regarding which indices of the WSAP were presented and specific task parameters used, making it challenging to compare effects.

CONCLUSIONS

The WSAP is a reliable and valid instrument for assessing and modifying interpretive biases with unique characteristics compared with other IB assessment and modification tasks.

Translating CBM-I Into Real-World Settings: Augmenting a CBT-Based Psychiatric Hospital Program

Cognitive bias modification for interpretation (CBM-I) is an appealing augmentation to cognitive-behavioral therapy (CBT) because it targets cognitive bias efficiently via computerized training. Few studies have tested the combination of CBM-I and CBT, and none have translated lab-based CBM-I protocols to an acute psychiatric setting. The present study describes the development and implementation of CBM-I as an augmentation to a CBT-based partial hospital. We developed a transdiagnostic CBM-I based on the word-sentence association paradigm (WSAP), which reinforces individuals for endorsing benign interpretations and rejecting negative interpretations of ambiguous sentences. Over two iterations of development, we randomly assigned patients (N = 127; M age = 34.21; 58% female, 40% male, 2% nonbinary) to either CBM-I or a control group (Phase 1: neutral WSAP task; Phase 2: treatment as usual). CBM-I comprised daily sessions (10 minutes) completed during program hours, and number of sessions varied naturalistically according to patient length of stay. Primary outcomes included feasibility, acceptability, and target engagement (interpretation bias). CBM-I was feasible and acceptable to acute psychiatric patients, and successfully shifted interpretation for novel stimuli. Patient feedback suggested that participants viewed CBM-I as bolstering their primary CBT-based care. Exploratory analyses examining clinical benefit revealed a small between-group effect on anxiety severity (d = 0.378), but no group differences on depression outcomes (d = 0.008). Findings indicate that CBM-I is a feasible and acceptable augmentation to CBT-based partial hospital care. Future studies are warranted to determine who is most likely to benefit from this low-intensity approach.

Interactive effects of Anxiety Sensitivity and Attentional Control on Post-traumatic Stress in Community and Clinical Populations

OBJECTIVE

High anxiety sensitivity (AS) and poor attention control (AC) are established risk factors for posttraumatic stress symptoms (PTSS), but little is known about the combined influence of these variables. Consistent with dual-systems models suggesting facets of executive function (e.g., AC) will modulate the effects of other risk factors (e.g., AS), the current study evaluated the singular and interactive effects of these variables on posttraumatic stress disorder (PTSD) symptoms.

METHOD

In Study 1, latent variable modeling was used to examine the unique and interactive effects of AS and AC on PTSS, controlling for trauma history, sex, and age, in a sample of trauma-exposed community adults (N = 670). In Study 2, latent variable modeling was used to replicate these effects in a sample of trauma-exposed treatment-seeking adults (N = 207).

RESULTS

Findings from both studies demonstrated a significant and negative interaction between AS and AC predicting PTSS when controlling for trauma history, sex, and age. Moreover, results revealed that AS more strongly predicts PTSS among those with poor AC.

CONCLUSIONS

These findings suggest that impaired AC, one facet of executive function, may potentiate the effects of AS on PTSS, and increasing levels of AS may enhance the effects of AC on PTSS. Results are discussed within the context of a dual-systems model of PTSS.

Is computerized psychoeducation sufficient to reduce anxiety sensitivity in an at-risk sample?: A randomized trial

BACKGROUND

Anxiety sensitivity (AS), or a fear of anxiety-related sensations, has become one of the most well researched risk factors for the development of psychopathology and comprises three subfactors: physical, cognitive, and social concerns. Fortunately, research has demonstrated brief protocols can successfully reduce AS, and in turn improve psychopathological symptoms. Computerized AS reduction protocols have combined psychoeducation with interoceptive exposure (IE), but they have not been dismantled to evaluate the effects of psychoeducation alone.

METHOD

The current study sought to determine the efficacy of a brief single-session psychoeducation intervention for AS, compared to a control intervention, in a sample of at-risk individuals (N=54) with elevated AS cognitive concerns.

RESULTS

Individuals in the active condition displayed greater reductions in self-reported AS (β=.198, 95% CI [.065, .331]) and less fear reactivity (β=.278, 95% CI [.069, .487]) to the induction of AS cognitive-relevant sensations through a behavioral challenge compared to those in the control condition. Further, fear reactivity to the challenge was mediated by reductions in self-reported AS cognitive concerns.

LIMITATIONS

Study limitations include use of an at-risk nonclinical student sample, lack of a long-term follow-up assessment, and inability to discern whether AS reductions due to CAST psychoeducation prevent future, or improve current, psychological symptoms.

CONCLUSIONS

These results suggest that psychoeducation alone can produce significant AS reduction.

Combined "top-down" and "bottom-up" intervention for anxiety sensitivity: Pilot randomized trial testing the additive effect of interpretation bias modification

OBJECTIVE

Anxiety disorders contribute substantially to the overall public health burden. Anxiety sensitivity (AS), a fear of anxiety-related sensations, is one of the few known malleable risk factors for anxiety pathology. Previous AS reduction treatments have primarily utilized "top-down" (e.g., psychoeducation) interventions. The goal of the current study was to evaluate the effect of adding a "bottom-up" (interpretation bias modification; CBM-I) intervention to an AS psychoeducation intervention.

DESIGN

Single-site randomized controlled trial. Participants completed either a 1) Psychoeducation + active CBM-I or 2) Psychoeducation + control CBM-I intervention. Change in AS was assessed post-intervention and at a one-month follow-up.

PARTICIPANTS

Individuals with elevated levels of AS.

INTERVENTION

Single-session computer-delivered intervention for AS.

RESULTS

Accounting for baseline ASI-3 scores, post-intervention ASI-3 scores were significantly lower in the combined condition than in the psychoeducation + control CBM-I condition (β = 0.24, p < 0.05; d = 0.99). The active CBM-I plus psychoeducation AS intervention was successful in reducing overall AS (59% post-intervention; p < 0.05, Cohen's d = 0.99) and these reductions were maintained through one-month post-intervention (52%; p < 0.05, Cohen's d = 1.18). Participants in the active condition reported significantly lower rates of panic responding to a vital-capacity CO₂ challenge (OR = 6.34, 95% CI = 1.07-37.66). Lastly, change in interpretation bias significantly mediated the relationship between treatment condition and post-treatment AS reductions.

CONCLUSIONS

The current intervention was efficacious in terms of immediate and one-month AS reductions. Given its brevity, low-cost, low-stigma and portability, this intervention could lead to reducing the burden of anxiety disorders.

Randomized controlled trial of attention bias modification in a racially diverse, socially anxious, alcohol dependent sample

OBJECTIVE

Attention biases may be an important treatment target for both alcohol dependence and social anxiety. This is the first ABM trial to investigate two (vs. one) targets of attention bias within a sample with co-occurring symptoms of social anxiety and alcohol dependence. Additionally, we used trial-level bias scores (TL-BS) to capture the phenomena of attention bias in a more ecologically valid, dynamic way compared to traditional attention bias scores.

METHOD

Adult participants (N = 86; 41% Female; 52% African American; 40% White) with elevated social anxiety symptoms and alcohol dependence were randomly assigned to an 8-session training condition in this 2 (Social Anxiety ABM vs. Social Anxiety Control) by 2 (Alcohol ABM vs. Alcohol Control) design. Symptoms of social anxiety, alcohol dependence, and attention bias were assessed across time.

RESULTS

Multilevel models estimated the trajectories for each measure within individuals, and tested whether these trajectories differed according to the randomized training conditions. Across time, there were significant or trending decreases in all attention TL-BS parameters (but not traditional attention bias scores) and most symptom measures. However, there were not significant differences in the trajectories of change between any ABM and control conditions for any symptom measures.

CONCLUSIONS

These findings add to previous evidence questioning the robustness of ABM and point to the need to extend the effects of ABM to samples that are racially diverse and/or have co-occurring psychopathology. The results also illustrate the potential importance of calculating trial-level attention bias scores rather than only including traditional bias scores.

Training interpretation biases among individuals with body dysmorphic disorder symptoms

The current study provided an initial test of a Cognitive Bias Modification for Interpretations (CBM-I) training paradigm among a sample with elevated BDD symptoms (N=86). As expected, BDD-relevant interpretations were reduced among participants who completed a positive (vs. comparison) training program. Results also pointed to the intriguing possibility that modifying biased appearance-relevant interpretations is causally related to changes in biased, socially relevant interpretations. Further, providing support for cognitive behavioral models, residual change in interpretations was associated with some aspects of in vivo stressor responding. However, contrary to expectations there were no significant effects of condition on emotional vulnerability to a BDD stressor, potentially because participants in both training conditions experienced reductions in biased socially-threatening interpretations following training (suggesting that the "comparison" condition was not inert). These findings have meaningful theoretical and clinical implications, and fit with transdiagnostic conceptualizations of psychopathology.