Cognitive Aspects of Generalised Anxiety Disorder in Adolescents: Exploring Intolerance of Uncertainty, Cognitive Avoidance, and Positive Beliefs About Worry

Anxiety disorders are common in childhood and adolescence [1, 2] and are associated with significant distress and functional impairment [3]. Generalised Anxiety Disorder (GAD) is characterised by uncontrollable worry and somatic symptoms (e.g., muscle tension [4]). It affects 1–3% of 11–19 year-olds [5], commonly co-occurs with other anxiety disorders [6], and is one of the most frequent anxiety diagnoses in treatment-seeking adolescent populations [7]. There is an increase in the prevalence [5] and clinician-rated symptom severity [7] of GAD from childhood into adolescence, perhaps reflecting the growth of meta-cognitive skills [8], and the heightened social-evaluative and academic concerns [9, 10] associated with this developmental stage. Yet despite its prevalence and negative effects, there is a lack of research into GAD relative to other anxiety disorders [11].

Across the lifespan, the hallmark feature of GAD is pervasive worry. This is defined as “a chain of thoughts and images, negatively affect-laden and relatively uncontrollable” (p.10) that is typically future-focused and reflective of an attempt at mental problem-solving [12]. It often presents as “what if…?” cognitions in which potential impending threats and uncertainties are identified [13]. Whilst worry exists on a continuum from normal to pathological [14], with the same content in individuals with GAD and those without [15], a GAD diagnosis is associated with significant emotional distress and impairment relating to worry [16]. Additionally, it has been suggested that clinical and non-clinical worriers are differentiated by the factors that underly their problematic cognitions, such as safety-seeking behaviours and threat beliefs [17]. Thus, it is vital to explore not what young people with GAD worry about, but how and why they worry. By furthering our understanding of the aspects that maintain this disorder, we can inform the development of effective, targeted psychological interventions [18]. This is necessary as most youth with GAD currently receive a disorder-general form of cognitive-behavioural therapy (CBT) which achieves remission rates of just 50% [19]. This leaves a concerning number of young people who continue to meet diagnostic criteria after treatment, and highlights the need for further research in this field.

A number of theoretical models of GAD in adults have been proposed to explain its aetiology and maintenance, and several have informed effective treatment approaches (see ​[20] for a review). For example, the Intolerance of Uncertainty Model [15, 21] suggests that individuals with GAD hold a set of negative beliefs about uncertainty and its consequences ​[22]​. Felt uncertainty is appraised as a sign of potential danger; it drives positive beliefs about worry, sets in motion a spiral of worrisome thinking, and feeds a negative problem orientation. Together, these cognitive, behavioural, and emotional processes exacerbate worry and motivate use of avoidance strategies (e.g., thought suppression). Treatments based on this model involve worry awareness training and re-evaluation of faulty beliefs about worry [22], and have been shown to result in large decreases in GAD severity, worry, and intolerance of uncertainty in adults [23]. Similarly, a separate extant model proposed by Wells [24] states that in GAD, positive beliefs about the usefulness of worry motivate its use as a coping strategy whilst concurrent negative beliefs about its danger (e.g., “I will lose my mind with worry”) cause anxiety to escalate. A vicious cycle is set in motion, as the cognitive and physiological symptoms associated with the worry reinforce problematic metacognitive beliefs and behaviours.

​Furthermore, Borkovec et al. [25] conceptualise worry as a cognitive avoidance strategy which, due to its verbal-linguistic nature, can inhibit negative mental images and their associated aversive somatic and emotional experiences. Via this process, worry hinders the emotional processing of fear-based information that is essential for extinction and habituation of the anxiety response [26], and positive beliefs about worry are reinforced by the avoidance of unwanted sensations [27].

​​Alternatively, Gustavsson et al.'s [17, 28] recent model of GAD in adults suggests that worry should be viewed as a safety-seeking behaviour which occurs in the context of negative core beliefs about oneself (e.g., “I’m a failure”). Its function is to reach a state of absolute certainty by foreseeing and taking steps to prevent all potential negative outcomes, which if they occurred, would confirm negative beliefs. GAD is maintained as the desired level of absolute certainty can never be realistically achieved, and failure of the feared outcome to take place reinforces positive beliefs about worry.

These models provide insight into the mechanisms underlying GAD in adults; however, little work has explored whether the same cognitive aspects are associated with the disorder in adolescence. This cannot be assumed, as adolescence is a unique developmental stage associated with distinct biological, social, and psychological factors [29]. Some preliminary evidence suggests that the aspects identified in adult cognitive models of GAD are also relevant in younger populations – for instance, relationships have also been found between cognitive avoidance and worry or anxiety in non-clinical adolescents [30, 31]​. However, other findings are mixed as demonstrated by a combination of significant [31, 32] ​and non-significant [33] results regarding positive beliefs about worry.

Furthermore, studies are often limited by their use of community populations (e.g., [25, 30, 32, 34]), or inclusion of clinical samples with broad age ranges spanning childhood and adolescence (e.g., [35, 36]). Of the clinical research, worry is frequently explored in the context of anxiety disorders in general, and therefore conclusions are not GAD-specific (e.g., [37]).

Additionally, questions remain regarding the specificity of these cognitive aspects to GAD (e.g., [38])—for example, intolerance of uncertainty explains additional variance in social anxiety symptoms after accounting for neuroticism, fear of negative evaluation, and anxiety sensitivity in adults [39] and individuals with specific phobias report greater efforts to engage with thought suppression than non-phobic controls [40]. It is vital to explore the question of specificity in order to understand the maintenance of GAD, facilitate accurate diagnoses [41], and inform more effective treatments approaches [18].

The present study examined levels of intolerance of uncertainty, cognitive avoidance, and positive beliefs about worry in adolescents. It addresses the shortcomings of previous studies with the inclusion of treatment-seeking young people who had GAD as their primary or secondary diagnosis (i.e., GAD was present in their diagnostic profile but was not the main presenting problem). The study also included adolescents with other anxiety disorders (without GAD in their diagnostic profiles). This ‘other anxiety disorder’ control group allowed us to assess whether the three cognitive aspects of interest are unique to GAD, or if they are general elements related to anxiety more broadly (with the former consistent with, although not sufficient evidence for, a causal role in GAD). To complete this analysis, scores were also compared to a community group who self-reported low levels of anxiety.

Our hypotheses were as follows:

This study explored whether three cognitive aspects that have been identified in adult models of GAD (intolerance of uncertainty, cognitive avoidance, and positive beliefs about worry) were elevated in adolescents with primary or secondary GAD, relative to anxiety control and community groups. Our hypotheses were partially supported. The GAD group was significantly more intolerant of uncertainty than both the anxiety control and community groups. Whilst cognitive avoidance was significantly higher in the GAD group compared to the community group, the comparison with the anxiety control group fell short of significance. In contrast, positive beliefs about worry were significantly higher in the GAD group compared to the anxiety control group, but not the community group. The three cognitive variables did not predict significant variance in GAD symptoms within the GAD group. Additionally, exploratory analyses revealed no significant group differences on cognitive variables when only those with a primary diagnosis of GAD were included in the GAD group.

Elevated intolerance of uncertainty in adolescents with GAD aligns with existing research in community populations [30, 32, 34, 65] and further supports a role for this construct in clinical samples. The significantly higher intolerance of uncertainty in the GAD group compared to the clinical control group can be interpreted as evidence for the specificity of this cognitive process to GAD. Mean intolerance of uncertainty scores were relatively similar for the anxiety control and community groups (M = 58.42 and M = 52.09, respectively) compared to the GAD group (M = 89.43), further supporting a unique relationship between intolerance of uncertainty and GAD. This supports previous research in adults showing that high intolerance of uncertainty distinguishes GAD from panic disorder with agoraphobia [66] and other anxiety disorders more broadly [67]. Moving forwards, studies may benefit from turning attention to the factors driving intolerance of uncertainty. For instance, one possible direction for research is the threat beliefs associated with GAD, which are suggested to centre around an intense fear of failure and a belief in one’s inability to cope [68, 69]. Such negative threat beliefs could underly an intolerance of uncertainty and associated positive views of worry [28]—for example, the notion that “Worry helps me cope” could be underpinned by a core negative belief about oneself (e.g., “I am incompetent” [70]). The role of such threat beliefs has received little research attention in GAD [17], but they are likely to be particularly relevant for adolescents who may be more vulnerable to fears of failure and negative evaluation, insecurity, and critical internal dialogues [71, 72].

Aligning with previous findings in adult populations [67], adolescents with GAD reported significantly greater use of cognitive avoidance than those in the community group. However, the difference between the GAD and anxiety control groups failed to achieve significance by a marginal amount. Whilst possible that this result is reflective of the ubiquity of avoidance behaviours across anxiety disorders, it may also be accounted for by our lack of statistical power. It might be that although avoidance is common across anxiety disorders, cognitive avoidance is more characteristic of GAD whereas other disorders are associated with greater behavioural avoidance. For example, whilst an individual with panic disorder who has previously had a panic attack on public transport may be able to avoid this situation, someone with GAD who worries about a great number of everyday events is likely to be less able to avoid anxiety-provoking circumstances completely. Instead, they may use cognitive avoidance strategies to control their anxiety within the situation. This is an area for future research, with larger samples and thus greater statistical power. The different patterns of significance between the intolerance of uncertainty and cognitive avoidance constructs imply that the former is more specific to GAD, whereas the latter is a process that is common across the anxiety disorders. Furthermore, both intolerance of uncertainty and cognitive avoidance are elevated in adolescent girls relative to boys [30]; given the gender imbalance in the GAD group, this may have contributed to stronger effects. This imbalance may be the result of boys experiencing greater barriers to accessing clinical services [73]. As such, future work must aim to overcome these obstacles. By increasing representation in clinical services (and by extension, research) we would be better placed to explore the effects of gender on anxiety.

Results showed significantly higher levels of positive beliefs about worry in the GAD group compared to the anxiety control group. This aligns with multiple cognitive models of GAD, which share the view that worry is reinforced by positive beliefs about its necessity [12, 15, 24, 28].

Interestingly, we found no significant difference between the GAD and community groups with respect to positive beliefs about worry, in contrast to previous research in adults (e.g., [67]). This finding may plausibly be the result of similar responses being given in different contexts across groups. For instance, “By worrying, I can find a better way to do things” may be true of someone who experiences time-limited and controlled worry. Yet equally, this belief could be held by an adolescent with GAD who worries excessively about finding the perfect solution to every perceived problem, and is unable to terminate the worry process until this unrealistic goal has been achieved. It could be inferred that where positive beliefs relate to pathological worry, this may contribute to the maintenance of GAD in young people. However, the small difference in mean scores between the anxiety control and community control group implies a relatively small effect of anxiety or community control group membership on this measure. This aligns with Wells' metacognitive model [24], which proposes that positive beliefs about worry are not unique to GAD. Rather, negative beliefs (e.g., that worry is uncontrollable and dangerous) differentiate GAD from social phobia and panic disorder [74]. Consequently, future studies may wish to explore whether negative beliefs about worry show greater specificity to GAD.

In all three analyses, depression was associated with significantly higher scores on cognitive variables. This is likely due to overlap between the mechanisms associated with worrying in the context of GAD, and those linked with depressive rumination. This can be understood by their shared characteristics (e.g., uncontrollability and negative content), and is reflected in the fact that both fall under the umbrella of ‘repetitive negative thinking’ (RNT; [75]). However, the finding that the presence of GAD predicted higher scores on cognitive aspects beyond the effects of depression supports their conceptualisation as separate disorders that may share cognitive patterns, but are nonetheless distinct [76]. This result has implications for debates regarding the nosology of GAD and depression [77], and provides avenues for further research into factors that distinguish GAD from other internalising disorders. Future studies may also wish to explore how GAD presents in isolation relative to when it co-occurs with depression—for example, with use of a two-by-two factorial design. Whilst extant research has assessed the differential impairments associated with pure and co-occurring GAD (e.g., [78]), whether cognitive processes differ between the two diagnostic groups remains relatively unexplored despite having clear clinical relevance.

Contrary to hypotheses, cognitive variables did not predict significant variance in GAD symptoms within the GAD group after controlling for gender, age, and depression. This may indicate that whilst these factors are relevant to GAD, they are not significant predictors of its severity. However, it is also possible that this result is due to a lack of statistical power or a failure of the outcome measure (RCADS-GAD subscale) to capture generalised anxiety symptoms. Exploratory post-hoc analysis revealed that of the participants with a diagnosis of primary or secondary GAD (determined by a gold-standard clinical interview), only 8 (17.4%) scored in the clinical range on this subscale. This may be because 5 of its 6 questions relate to pathological worry which, although a key feature of GAD, is not its only characteristic [16]. Thus, future research may benefit from using a different measure that reflects additional aspects of GAD (e.g., somatic symptoms).

Exploratory analyses comparing only participants with primary GAD to the control groups revealed no significant differences on any of the three cognitive measures—a result that may plausibly be accounted for by a lack of statistical power.

The strengths of this study include its clinical population, use of diagnostic interviews for the clinical groups, and ability to assess the extent to which these cognitive aspects are unique to GAD relative to other anxiety disorders (i.e., to establish broad specificity [41]). Yet despite these merits, it is important to interpret findings in light of their limitations.

Notably, the study was underpowered; while we were able to successfully recruit adolescents with GAD and adolescents from the community, it was challenging to recruit young people from clinical services with anxiety disorders where they did not have GAD in their diagnostic profiles, reflecting the high prevalence and pervasiveness of GAD in this age group. It is possible that had this group been larger, this would have changed the results, for example, the difference between this group and the GAD group for cognitive avoidance may have reached significance. This highlights the need for further research and to establish the relevance of our findings to a large proportion of young people. GAD is one of the most common anxiety diagnoses [5, 7], yet accounts for only 8% of anxiety disorder publications [79]. As such, moving forwards GAD should receive more research attention in line with its high prevalence and burden.

Furthermore, the strong gender bias in the GAD group was disproportionate to what would be predicted on the basis of prevalence rates [5]. Girls report higher levels of worry than boys [30], and thus it is possible that the greater proportion of girls in the GAD group compared to both control groups could partially account for our pattern of results. As such, future studies with larger and more balanced samples should be conducted to explore potential gender differences. Furthermore, due to resource constraints, we were unable to deliver diagnostic assessments to the community control group, instead screening on the basis of self-reported anxiety symptoms. The total anxiety score from the RCADS was used for this purpose; however, it does not assess all the anxiety disorders included in the DSM-5 [4]. As such, this assessment process was less rigorous than a clinical interview, leaving a possible presence of psychological disorders. Whilst this introduced a degree of uncertainty regarding the clinical characteristics of this subgroup, it also represents a more conservative comparison and thus may even lend stronger support to our significant results. We were also limited in the number of variables we could examine and therefore future research should consider additional factors, such as negative problem orientation [15].

Finally, our cross-sectional design means that we cannot determine whether, for example, intolerance of uncertainty has causal or consequential specificity to GAD [41]. Therefore, there is a need to replicate and build upon our findings in well-powered longitudinal studies with more diverse populations. It may also be beneficial to explore if and how the cognitive profile of GAD differs across the lifespan. This could involve directly comparing treatment-seeking samples of adolescents, children, and adults to provide a nuanced insight into any developmental differences and their effects on clinical characteristics.

Our novel findings indicate the relevance of intolerance of uncertainty and its specificity to GAD in adolescence, and may have clinical implications. Treatments for GAD that specifically aim to increase tolerance of uncertainty draw on Dugas et al.'s cognitive model [15], and include components such as worry awareness training and exposure to unsure situations. Preliminary studies of these interventions for adolescents have reported variable results ranging from reliable clinical improvement in 3 out of 7 participants [80] to 7 out of 12 participants [37]. Therefore, further work may include optimising treatments before comparing them in larger scale, more controlled trials. Overall, the picture of treatment for GAD in adolescents highlights the need for a better understanding of the mechanisms that maintain this anxiety disorder, and the refinement of targeted interventions that successfully address these processes.

This study provides preliminary evidence for the broad specificity of intolerance of uncertainty to GAD in a clinical sample of adolescents. Youth with primary or secondary GAD were significantly more intolerant of uncertainty than youth with other anxiety disorders, and non-anxious community participants. Meanwhile, cognitive avoidance was similarly elevated in both clinical groups relative to community participants. Meanwhile, cognitive avoidance was similarly elevated in both clinical groups relative to community participants. The GAD group reported significantly higher positive beliefs about worry compared to the anxiety control group, but the difference between the GAD and community groups on this measure was non-significant. Our study offers a theoretical basis for further research into the cognitive mechanisms driving GAD in adolescents, with particular support for the relevance of intolerance of uncertainty.