Efficacy and mechanisms underlying a gamified attention bias modification training in anxious youth: protocol for a randomized controlled trial

Anxiety disorders, such as separation anxiety, specific phobia, social phobia, generalized anxiety and panic, occur in 15–20% of all children and adolescents [1]. This is particularly important because anxiety during childhood and adolescence predicts not only later anxiety, but also other psychiatric disorders during adulthood [2, 3]. Cognitive behavioral therapy (CBT) emphasizes behavioral exposures and cognitive restructuring that aim to change dysfunctional behaviors (e.g. avoidance of fear-provoking stimuli/situations) and thoughts (e.g., catastrophic thinking). CBT has been shown to produce medium-to-large symptom reductions in anxious children and adolescents [4]. However, only one in five clinically anxious youth receives treatment [5] and about 50% of treated patients do not respond to treatment [6, 7]. This has motivated researchers to develop Attention Bias Modification Trainings (ABMT), which specifically target one well-replicated mechanism in anxiety disorders: an attentional bias towards threat [8].

ABMT is typically employed using a modified dot-probe task [9]. The dot-probe task indexes attentional bias using the difference in reaction times to target stimuli presented at the previous location of either a threatening or neutral stimulus, often emotional faces [10], for details see Fig. 2b). During ABMT, the probe consistently appears in the location of the neutral face stimulus thereby implicitly training participants to attend away from the threatening face, whereas in the control condition the probe replaces the neutral and threatening face stimuli with equal probability [11]. Thus, the modified version of the dot-probe task is hypothesized to change the attentional bias ‘bottom up’ by directly retraining salience contingencies [11]. On a neurobiological level, the dot-probe task has been most robustly associated with amygdala-insula connectivity [12‐14], which according to a first report by White and colleagues does not only differentiate anxious from healthy youth, but also predicts the response to ABMT [15]. Of note, the visual search for non-threat targets in the context of threatening distractors has also successfully been used as ABMT [16, 17], but has been more related to activity in the fronto-parietal attention network and the amygdala [18, 19]. Given that the modified dot-probe task and the visual search training target different, albeit partial overlapping neural circuits, it is conceivable that a combination of both trainings would yield enhanced treatment effects.

To date, seven meta-analyses have investigated the effects of ABMT in terms of the modified dot-probe task in youth and adults showing small [20‐23] to moderate [24‐26] reductions in attention bias towards threat. ABMT effects on symptoms of anxiety were generally small [20‐26], albeit single reports of moderate [27], and no symptom reductions [28]. An additional six studies have investigated the efficacy of active vs. control ABMT in addition to CBT. Four studies including the one by White and colleagues (2017) showed an enhanced effect of the combination treatment on symptoms rated by clinicians [15, 29‐31]. An additional study showed improved effects of ABMT+CBT on symptoms rated by child and parent only [32]; one study reported no significant differences between active ABMT+CBT vs. control ABMT+CBT [33].

These mixed findings might be partially explained by differences in the study designs. Particularly, the vertical, rather than horizontal, presentation of stimulus pairs during the modified dot-probe task appears to yield larger effects [27]. Moreover, effect sizes of ABMT are larger in laboratory vs. home settings, and when anxiety symptoms are assessed by clinicians (rather than self-report) [34]. Inconsistencies may further relate to individual differences in task engagement and motivation to complete a specific task [35], as engagement has been shown to moderate the efficacy of cognitive trainings [36] such as the modified dot-probe task [37]. The use of game-like elements, henceforth referred to as “gamification”, [38] could make ABMT more engaging and motivating, thereby enhancing adherence to the training and improving ABMT efficacy [39]. As outlined by Boendermaker and colleagues (2015), the use of game elements in cognitive trainings such as ABMT should ideally target both extrinsic and intrinsic motivation, which might be achieved by using a performance-based point system and designing the training in a way that it is inherently interesting or enjoyable. The use of game-elements seems particularly promising in younger children (e.g. < 12 years), who show relatively small effects compared to older youth [40].

Finally, it has been argued that ABMT can only result in anxiety symptom reduction, when an attentional bias towards threat is initially present, i.e. the proposed target mechanism of ABMT [41]. Unfortunately, it is difficult to assess attentional bias behaviorally as reliability of the bias measure derived from the dot-probe task is poor [42]. However, reliability of amygdala connectivity as a neurobiological correlate of the attention bias is better [43].

Attentional bias towards threat is a key mechanism in anxiety disorders [8] that is not specifically targeted by CBT. ABMT procedures mostly using a modified version of the dot-probe task have been specifically developed to change this attentional bias ‘bottom up’ by directly retraining salience contingencies. Despite promising results of initial studies, this has not proved straightforward [28] and thus further research is warranted.

This study will be the largest randomized controlled trial in anxious youth that uses fMRI to assess the attention bias by focusing on amygdala connectivity – a neurobiological marker that has adequate test-retest reliability [43]. Of note, the intervention used in this study has been designed to maximize treatment response and therefore consists of a novel combination of the modified dot-probe task and a modified visual search task to more comprehensively target processes involved in the selective allocation of attention. Moreover, we also address the issue of low motivation/ low adherence by designing the visual search training as a puzzle that will be inherently interesting and additionally introducing game elements such as digital rewards, feedback, levels, and time pressure. Finally, we apply several strategies shown to enhance the ABMT effect including [1] using vertical presentation of stimuli pairs during the modified dot-probe task [27]; [2] training in a clinic setting to minimize distractions [8], and activate a relevant ‘fear structure’ to provoke the attentional bias that can then be modified [30]; and [3] using of independent clinician-based assessments of anxiety, as opposed to self-reported anxiety measures that are more prone to biased reporting [71]. Considering this framework, it can be expected that the results of this study will have a significant impact on the debate regarding the efficacy of ABMT on attentional biases. The study also can contribute to the development of personalized treatments and delineate the mechanisms underlying a predictive relationship between pretreatment attentional bias reflected not only in behavioral measures, but also increased amygdala-insula connectivity and response to ABMT.

Despite considerable strengths, there are several limitations inherent in the study design. First, we essentially combine three interventions (modified dot-probe, positive visual search, and CBT), and test them with two study arms. Thus, it will not be possible to dissociate effects of the modified dot-probe aiming at bottom-up attentional processes, the visual search potentially also targeting top-down attentional processes and CBT meant to reduce dysfunctional appraisal and avoidance of potential threats using exposure and cognitive strategies such as restructuring. In this regard, we also will not be able to answer whether the gamification led to an enhanced treatment response should we find one. A second related limitation is that the attention bias will be solely assessed with the dot-probe task. No task measuring goal-directed inhibitory control of attention potentially targeted by the visual search training will be administered as the sum of the ratings, interview and the dot-probe task that are conducted pre- and post-treatment already pose such a strain on the participating families that unfortunately the assessment of a second task pre-post treatment is practically not feasible. Thirdly, all patients will receive CBT with either active or control ABMT, so our findings may not be generalizable to other effective treatments, such as medication. Fourth, we enroll participants with any one of various anxiety disorders (generalized anxiety disorder, social phobia, separation anxiety, specific phobia). For this reason, we cannot tailor training stimuli to specific anxiety disorders. Fifth, the performance during the visual search task depends on the performance during the dot-probe task. Although the dot-probe task is easy (participants earn a point for correctly indicating whether the arrow is pointing to the left or the right) and participants always earn points, which they can use during the visual search task; this might negatively impact the motivation for some participants. However, only the numbers found during the regular time will be analyzed. Sixth, it is not the focus of the present study to assess the effect of the game elements. Therefore, engagement and adherence with regards to the game elements are not specifically assessed or evaluated but might be the focus of future studies. Lastly, the generalizability of our findings may be affected by the study’s exclusionary criteria (e.g., depression, OCD).

In summary, this is an important study that further tests the efficacy of a computer-based attention bias modification training aiming at reducing increased attention towards threat – a mechanism not specifically targeted by CBT. Given previously mixed findings, we employ a combination of the modified dot-probe and visual search to target both bottom-up and top-down attentional processes. Additionally, the training will be imbedded in a game to enhance motivation and adherence. Finally, we will use fMRI to delineate mechanisms relevant for future guided treatment selection in this largest randomized controlled trial in anxious youth to date.