Abstract
‘We are, perhaps uniquely among the earth's creatures, the worrying animal.’
Lewis Thomas [1]
Worry is a familiar private experience – a cognitive activity with which most people, if not all, can identify. It is also a particularly common symptom reported by patients with psychological problems. As a phenomenon, it can range from an innocuous activity possibly associated with positive consequences (i.e. solution finding), through to a distressing and uncontrollable process like the excessive and chronic worry recognized as the cardinal feature of generalized anxiety disorder (GAD) [2]. It has been defined broadly as repetitive thought activity, which is usually negative and frequently related to feared future outcomes or events [3].
In recent years worry has gained increasing research attention in its own right, both in non-clinical and in anxious populations, independent of more general anxiety research. This paper provides an overview and synthesis of several salient issues regarding worry.
Phenomenology of worry
Evidence suggests that worry is predominately comprised of cognitions most aptly defined as verbal or linguistic in nature, as opposed to images [3–5] and that worry involves a dynamic (and malleable) and narrative process in which themes are developed and elaborated rather than simply occurring as a string of negative thoughts [6]. It is also defined as the persistent awareness of or attention to possible danger which is rehearsed without successful resolution. Cognitive processes labelled as worry are usually described as difficult to dismiss by anxious subjects and self-labelled worriers [7].
Normal versus problematic worry
Can the act of worrying ever be beneficial, or is it always, by definition, a futile process? In a study of the phenomenology of ‘normal’ worry, Tallis et al. [6] found that most subjects considered worry to be a routine and mostly acceptable activity that occurred more or less daily, about various issues and mostly in the form of thoughts with a narrative course. Typically, worry was associated with real-life triggers, was both present and future-orientated, occurred more at night before sleep, and focused upon problems which were real or likely, rather than imaginary or remote. Worry was found to have potential benefits in relation to acting as a stimulant for action, with the majority of subjects perceiving their worry as a somewhat efficacious problem-solving activity.
However, various negative cognitive (e.g. pessimism, problem exaggeration) and affective (e.g. emotional discomfort, depression) consequences as well as behavioural (e.g. nail biting, restlessness) concomitants were also noted. A number of features distinguished subjects who were ‘high’ worriers, including: more daily worry episodes; greater difficulty in stopping worrying; rebounding worries; more mood disturbance and more perceived impairment in everyday functioning. ‘High’ worriers reported much greater indecision and doubt when worrying, were also more likely to perceive worry as having a negative effect on their health, but rarely reported worrying about their worry [6].
According to Mathews [8], worry may ‘mimic’ problem-solving, but the worry process is unlikely to achieve a satisfactory end or conclusion because of its characteristic rehearsal of feared scenarios. This is well illustrated in the case of ‘catastrophising’, whereby internal or external events are conceptualized as threats, with potential problems amplified by ‘what if…?’ questions, leading to increasingly worse outcomes. Undoubtedly, the very presence of catastrophising in worry would thwart any problem-solving process, complicating attempts at problem-orientated ‘worrying’, and also likely to lead to emotional discomfort [9]. We know that individuals who worry more, also more commonly engage in catastrophic thinking [10, 11].
Dugas et al. [12] found evidence to suggest that the immediate reactions of high worriers to problematic situations (e.g. problem avoidance, counterproductive expectations or beliefs), were likely to account for their poorer performance on problem solving tasks. This finding supports other research demonstrating that high worriers tend to use avoidance coping behaviours, be intolerant of ambiguity and uncertainty, and be hindered by elevated evidence requirements when faced with real-life problems [13–15]. Therefore, the relationship between problemsolving difficulties and worry may be best accounted for by problem orientation factors, that is, a host of unhelpful immediate responses to given problems.
Some research into personality correlates suggests that high worriers are more self-evaluative and self-conscious, more perfectionistic – placing excessive demands upon personal performance, tend to feel more time-pressured, and show more obsessional symptoms compared with low worriers [16, 17]. Also, chronic worriers and GAD patients make more elevated subjective probability judgements when evaluating the future likelihood of personal exposure to unpleasant events [18]. This evidence suggests that worriers may be less ‘effective’ in relation to task completion activities, but further personality research is required to better understand the relationship between worry and constructs such as conscientiousness (to test a possible inverse relationship) as well as perfectionism and obsessionality.
Is worry a distinct cognitive activity?
Several authors have argued that worry can be largely distinguished from other cognitive phenomena like obsessional ideations on the basis of content, with obsessional content considered to be more ego-dystonic, and with themes typically in conflict with the individual's personality [19, 20]. Worry tends to be a longer and more inconclusive process, whereas obsessions are held to be briefer and more ‘static’ thoughts that intrude into consciousness [21]. However, evidence as to whether or not each differs by degree of intrusiveness, controllability, dismissability and subjective distress has been less consistent [20, 22–24]. Furthermore, obsessions are more commonly accompanied by intrusive impulses and images [19]. Also, while various theoretical accounts of worry function have been proposed [25, 26], no functional significance has been attributed to obsessional ideation.
Worry is also similar to the cognitive process of ruminating, often described as depressive rumination [27–29]. Rumination is a verbal process which requires repetitive cognitive attention and is biased towards past or current problems. Ruminations are inert cognitions, being more fixed in theme or content (‘Why did this happen to me?’), as opposed to worry and are highly associated with negative affect. Sufferers of depression have been found to report benefits of ruminating in relation to improved self-awareness [30].
While obsessions and ruminations are similar processes, worry is a more fluid process, particularly because its future-orientated focus is conducive to the embellishment of hypothetical outcomes. While each of the phenomena discussed above are all repetitive cognitive processes, they may each possess differing qualities in relation to focus, structure and function.
Why worry? Theoretical accounts of worry function
The most common reasons for worrying given by both GAD and nonanxious individuals are, first, that it helps in discovering ways to avoid future adverse events; and second, that it somehow assists in preparing for the worst – if the worst cannot be avoided [31]. While the first reason denotes worry partly as an attempt to solve problems, the second justification is less cogent, and includes beliefs about worrying reducing the likelihood of bad events, and guilt over not worrying should the worst eventuate [14].
Worry as the inhibition of emotional processing
Borkovec et al. [25] argue for an important difference between thought and imagery in relation to worry and emotional disturbance. Based upon the evidence that worry is primarily(but not entirely) comprised of verbal thought activity (rather than imagery), Borkovec et al. [25] have conceptualized worry function as ‘the inhibition of emotional processing’.
The specific verbal–linguistic nature of worry is said to be responsible for reduced physiological activity and thus the inhibition of emotional reprocessing. Vrana et al. [32] found that verbal thoughts about emotional material elicit very little cardiovascular response, but images of the same material evoke a much greater response. Also, subjects have been found to use verbalization spontaneously as a strategy for abstraction, disengagement and emotion control that can decrease sympathetic arousal to aversive stimuli [33]. Borkovec and Hu [34] studied a group of subjects with a public speaking phobia and found that those assigned to a ‘worrisome thoughts’ group had no increase in heart rate upon exposure to subsequent phobic images, compared to subjects who used ‘relaxation’ or ‘neutral’ cognitive strategies. Thus, Borkovec et al. [25] argue that the verbal–cognitive system (which is less closely connected with affective, physiological and behavioural systems) plays an important role in maintaining pathological worry. On the one hand, this more ‘isolated’ verbal system is clearly adaptive, as it allows for mental experimentation (i.e. subvocalization, mental rehearsal). However, for any emotional disorder predominantly characterized by thought, such a system becomes problematic, because emotional processing will be inhibited, leading to the maintenance of anxiety-provoking meanings and emotional disturbance. For example, during exposure therapy for phobic stimuli, it is essential that full affective processing of fear-related material occurs for extinction to take place. Therefore, if worry is ‘thought activity’ that prevents emotional processing and is inadequate for processing and changing emotional meanings, its role in fostering emotional disturbance (e.g. anxiety and depression) is an important one. Worry as cognitive avoidance Borkovec et al.[25] also describe worry as a type of cognitive avoidance of perceived dangers. Via a ‘superstition reinforcement paradigm’, worrying is negatively reinforced by the non-occurrence of aversive outcomes – because many of the things worried about do not eventuate or result in negative outcomes. This environmental contingency may be responsible for the belief that worrying will lessen the likelihood of the feared outcome eventuating – a belief reported to exist even though people acknowledge its lack of logic [31].
Similarly, Eysenck [26] proposed a cognitive model of worry as an adaptive process which fulfils three major functions. First, when any form of ‘threat’ is detected, an alarm function operates to bring information about the threat into awareness. This is followed by a prompt function which retrieves additional threat-relevant thoughts and images from long-term memory. Finally, a preparation function becomes operative allowing the individual to ‘anticipate’ aversive outcomes. This preparation function allows for two possible coping-like mechanisms: (i) either the person will act in order to prevent the occurrence of the anticipated negative event, or (ii) a process of habituation will lead to a reduction in aversiveness of the anticipated outcome.
Much earlier, Janis and Leventhal [35] described ‘the work of worrying’ as preparation for subsequent threatening experiences. Here, worry is said to have (at least in part) desensitizing properties – increasing one's tolerance to real threats and possibly reducing anxiety in the process. Thus, perhaps some individuals are less surprised and better prepared to cope when faced with a threat, because of the desensitization achieved by anticipatory worrying. However, the more the worry process is perceived as ‘successful’ (i.e. if negative events do not eventuate), then the less likely one is to identify the process as worrying, but rather as ‘problem-solving’ or ‘preparatory coping’ [8]. Again, by this account, worrying becomes negatively reinforced and thus encouraged.
Worry, anxiety and depression
Is worry the cognitive component of anxiety? In factor analytic studies of anxiety measures, somatic symptoms are differentiated from cognitive ones (e.g. intrusive or unwanted thoughts) and the two factors appear to have different correlates [36, 37]. For example, worry is more related to test anxiety and a better predictor of poor test performance than is somatically-based emotionality [38]. Interestingly, reported episodes of worry are generally not associated with raised levels of physiological arousal [39]. Thus, worry seems to be verbally encoded and at least partly independent of somatic arousal, which could render it a specific cognitive process distinguishable from any broader construct of anxiety.
As mentioned earlier, evidence suggests that worrying (and its verbal–linguistic predominance) acts to disallow emotional processing via its inhibition of physiological arousal and its failure to achieve habituation or extinction [34, 40]. Foa and Kozak [41] noted that the degree of cardiovascular response to phobic exposures corresponded with the degree of emotional processing. This avoidance process leads to a maintenance of anxietyprovoking meanings and a more generalized anxiety and emotional disturbance. Hence, while worry and anxiety are distinguishable, in that physiological indicators of anxiety may be inhibited during worry episodes, ongoing worry clearly perpetuates chronic anxiety.
It is unclear however, exactly how worry actually works to suppress somatic anxiety, although three possible explanations have been proposed. First, Borkovec et al. [25] have speculated that, rather than directly suppressing aversive images, worry may be the immediate cognitive (avoidance) response to distressing images as they occur. Second, the suppression effect of worrisome thinking could be attributed to the fact that worry uses up large amounts of attentional resources [8]. Finally, Borkovec and Hu [34] have used Gray's [42, 43] model of the behavioural inhibition system (BIS) to explain the suppression of somatic anxiety and emotional processing. In Gray's neuro-psychological theory of anxiety, the discrepancy between expected and received information (e.g. novel or aversive stimuli) activates the BIS, eliciting behavioural responses such as the inhibition of ongoing activity, hyper-vigilance and increased arousal. If a person is preoccupied by worrying about aversive outcomes, then the likelihood of the BIS being ‘activated’ in response to semantically relevant cues is reduced (i.e. due to reduced discrepancy).
Worry and mood disturbance
Although worry-associated mood disturbance appears to be relatively minor in non-clinical populations, Tallis et al. [6] reported that ‘high’ (non-clinical) worriers experienced a greater degree of depressed mood during worrying compared to ‘low’ worriers and one study reported that ‘pathological’ worry existed to an equivalent degree in patients with either GAD or major depression [44]. Ruminations have been implicated in the maintenance and exacerbation of depression, both in studies examining induced [45] and naturally occurring depressed mood [46] when compared to a distraction coping strategy. Further, depressive rumination also increases subjects’ global negative attributions [47] and increases recall of negative memories [48].
Mathews [8] has argued that worry content may alternate between thoughts of future feared events (and generate anxiety) and thoughts about past negative events (and create depression). One recent study [49] examined this worry content issue in those with GAD, depression and both disorders and observed some differences in worry themes supportive of a content-specificity hypothesis. Although not conclusive, this study found that worry about ‘loss of control’ was more relevant to anxiety, whereas ‘aimless future’ worries were specifically associated with depression symptoms and severity, a finding consistent with the hopelessness aspect of depressive cognitions.
Depression and anxiety frequently coexist and, if worry is capable of generating and maintaining both anxiety and depression, then it may be responsible in part for their co-occurrence. Thus, worry appears to be a chief activity underlying the common clinical presentation of ‘anxious-depression’. The capacity of the selective serotonin re-uptake inhibitor antidepressants to reduce anxious worrying [50] and, by such means, reduce the onset and duration of depressive episodes, may further argue for the interdependence of these constructs.
Worry in generalized anxiety disorder
The frequency and uncontrollability of worry has been found to clearly distinguish GAD subjects from nonclinical worriers, whereas worry content is often viewed as less important [53, 54]. However, some studies have indicated specificity of worry content in relation to severity. For example, Mathews [8] argued that excessive worry is mainly concerned with more personal domains (threats to self) rather than ‘universal’ dangers. Craske et al. [51] found that GAD patients worried more about illness or injury and miscellaneous issues, whereas worries about family, home and interpersonal issues were equally reported by GAD patients and controls. Worry content in GAD subjects may be more generally diverse with worry about multiple minor matters, compared with both non-clinical subjects and other clinically anxious groups [55]. Roemer et al. [56] also compared the worry content of GAD patients and nonanxious control worriers. They found that worry in GAD was more pervasive and diffuse, with greater coverage of miscellaneous or minor worrisome topics – a finding consistent with the view that GAD patients tend to perceive themselves as ineffectual copers living within a dangerous world [3].
One study [57] found that worry about highly remote future events was a distinguishing feature of GAD, suggesting that high levels of worry about future events is a sensitive marker of both GAD and its severity – a finding that further highlights the exacerbating presence of catastrophizing in worry.
Thus, worrying about real immediate problems may have a useful function in arriving at reality-based solutions and therefore may well be adaptive [58]. Individuals who report these types of worries (e.g. related to work) may be describing the motivational type of worrying proposed by Tallis et al. [6]. Conversely, worrying about highly remote events is rarely adaptive, with little (if any) immediate problem-solving potential. Adaptive or ‘normal’ worry should therefore seldom, if ever, focus upon highly remote events [57].
Wells [54] has proposed a meta-cognitive model of worry in GAD, in which those who develop problematic worry possess a range of rigid beliefs about the process. Positive beliefs stimulate the usage of worry as a strategy to process material and as a coping strategy. However, when worry itself becomes the focus of negative appraisals, so called normal worry becomes transformed into pathological worry as in GAD. The model proposes that pathological worry arises when the individual begins to worry about their worrying. In a test of this model [59], meta-worry was positively associated with worry in GAD, but trait anxiety and general worry (i.e. non-meta-worry) was independent of that association – underscoring the potential specificity of meta-worry in GAD.
Generalized anxiety disorder worry has also been conceptualized as ‘anxious apprehension’ [52], a nonfunctional exercise leading to high self-focused attention and inevitably to performance deficits. Here, ‘anxious’ memories activated by certain cues lead to negative affective states, an attentional shift toward a self-evaluative focus, increased arousal and a perceived inability to predict or control situations. The attentional narrowing and hyper-vigilance for threat cues disallow the processing of other stimulation unrelated to worry.
Such anxious apprehension, whereby a person anticipates or attempts to cope with future negative events, is also a fundamental process common to all anxiety disorders [53, 60].
Treatment strategies
While it is beyond the scope of this review to address in detail issues relevant to the treatment of worry, it is necessary to address the psychological and behavioural factors relevant to certain interventions.
One dominant and conventional approach to controlling worry has been that of ‘thought suppression’ (a deliberate attempt to stop unwanted thoughts mostly by engaging in searches for distraction stimuli). Although intuitively appealing, thought suppression techniques are, for the most part, counter-productive and characteristically produce a rebound effect [61, 62] often resulting in an upsurge of unwanted or distressing thoughts.
Laboratory studies comparing the effectiveness of thought ‘suppression’ versus thought ‘expression’ in relation to worry activity (i.e. worry exposure vs distraction from worry) have yielded mixed results however, with some studies finding no evidence of suppression leading to a rebound of worry (see Purdon [63] for a comprehensive review). More research is needed in clinical samples as much of the mixed evidence relies on data collected in contrived experiments using non-clinical samples. Overall, the evidence suggests that chronic worriers can successfully postpone their worrying activity thereby reporting fewer intrusive thoughts and reducing anxiety in the short term. However, research that describes the ability of worry to divert emotional processing [25], highlights an important consideration for treatment approaches and reveals the need for (structured) exposure to emotional images depictive of worry content and also for strategies targeting the resolution of other emotional themes underlying worry content.
Treatment strategies using non-distraction approaches such as ‘symptom prescription’ [64] and other paradoxical techniques may prove effective in the management of worry. The rationale proposes that by bringing on and elaborating the symptom (e.g. worrying for a 20-minute period, even if difficult to complete) the individual gains greater control over it and thereby more able to reduce it. Symptom prescription is one briefstrategic intervention whereby the instruction to do more of an unwanted symptom, produces a paradoxical effect and ‘control’ is gained over what has previously been experienced as ‘uncontrollable’.
Most other studies interested in the treatment of worry in clinical samples have evaluated cognitive–behavioural therapies (CBT) for patients with GAD. There is an accretion of research examining CBT for anxiety disorders, which, in sum have found that both behavioural and cognitive treatments have been more effective than non-directive interventions, diazepam and waitlist control groups in reducing GAD symptoms, including worry [65]. However, a recent study [66] compared the effectiveness of cognitive therapy (CT) against applied relaxation (AR) (see Ost [67]) for the treatment of GAD with a number of outcome variables assessed at post-treatment and 12 month follow-up. They found that AR alone, was as equally efficacious as CT for reducing a number of variables including ‘worry severity’ and ‘number of worries’. Subjects who received AR experienced a further reduction in the number of worry themes from post-intervention to follow-up assessment. This study highlights the importance of reducing the physiological and somatic components of GAD and signifies the clinical importance of implementing relaxation training for the treatment of worry. The finding that AR alone (without CT) significantly reduced worry quantity and severity also challenges notions of causality about cognitions and physiological arousal and is also consistent with one study showing muscle tension (over other somatic anxiety symptoms) to be uniquely associated with pathological worry in worriers with and without GAD [68].
Conclusions
At least some degree of worrying is a commonplace experience for most people. Worry is not typically a pathological process, but can be described as a spectrum from functional, whereby worry-like activity of a mostly volitional nature proves useful in motivating and in problem-solving, to worry that is particularly difficult to stop, characterized by repetitive catastrophic speculation and which is generally detrimental to performance.
Worry is somewhat distinct from other cognitive processes like obsessions and ruminations and has functional significance, particularly in relation to its ‘inhibitory’, ‘avoidant’ and ‘preparatory’ mechanisms.
Worry plays an important role in perpetuating ongoing emotional disturbance, and this has considerable significance, especially when it appears to be a key symptom linking anxiety to mood disturbance. The tested benefits of relaxation training in the reduction of worry severity and frequency is of important note. Treatment approaches might also incorporate strategies to deal with factors associated with worry severity such as problem orientation factors and intolerance to uncertainty, as well as personality significant correlates (e.g. perfectionism).
Further research is needed to investigate some remaining questions regarding differences in worry as a pathological process in those with anxiety disorders (particularly GAD) compared with major depression and the severity and relevance of that worry for both disorders.
Footnotes
Acknowledgements
We acknowledge funding from NHMRC Program Grant (222708).