Current psychological formulations of GAD emphasise excessive worry as the core feature of the disorder and are concerned with understanding the nature of worry and the conditions under which it persists. A comprehensive overview of theory, treatment and research on worry can be found in Davey & Wells (2006). In this section we summarise the work of the most prominent theorists in the field. Borkovec and his colleagues have conducted extensive and influential research on the nature of worry (cf. Borkovec & Newman, 1999). Key findings are the predominance of verbal self-talk rather than imagery in worry, the inhibiting effect of worry on emotional processing of threat-related material and the significant connection between GAD and interpersonal factors. The relationships between these factors are quite complex but in essence worry is conceptualised as a form of cognitive avoidance that functions to decrease sympathetic arousal to perceived threat and to inhibit emotional processing thereby preventing extinction of the fear response (Borkovec, Ray & Stoeber, 1998). It is an attempt to avoid the anxious arousal or anticipated catastrophe associated with negative events or deeper level emotional concerns and to the extent that this succeeds its role as a coping response is strengthened through a process of negative reinforcement. The deeper level concerns are thought to be predominantly interpersonal in nature. Borkovec (1994) suggests that insecure attachments to primary caregivers, as well as psychosocial traumas in early childhood, may be strongly related to the sense of uncontrollability of negative events that is so characteristic of GAD.
The cognitive model of GAD developed by Adrian Wells is based on a distinction between two types of worry: Type 1 worries, which concern everyday events and bodily sensations, and Type 2 worries, which are focused on the act of worrying itself and reflect both positive and negative appraisals of worrisome activity (Wells, 1999). There is evidence that the content of Type 1 worries is very similar to normal worries (Craske et al., 1989) and that GAD is associated in particular with Type 2 worries. The theory proposes a particular sequence of events. Once triggered, the worry cycle persists initially through the activation of positive Type 2 worries (positive metacognitions such as 'worry helps me cope', 'worry prevents bad things happening'), which in turn increase the accessibility of, and sensitivity to, threat-related information and lead to more intense worrying. The balance of appraisal then shifts to predominantly negative Type 2 worries ('my worries are uncontrollable', 'I could go crazy with worrying'), which motivate attempts to reduce distress and regain control. These attempts may involve avoidance, reassurance seeking, suppression of upsetting thoughts or engagement in distracting activities all of which may, in fact, reinforce worry as a coping strategy rather than reducing it. Deliberate attempts to suppress unwanted thoughts, for example, may inadvertently lead to an increase in their intrusiveness (Clark et al., 1991).
The cognitive model of GAD developed by Dugas et al. (1998) consists of four components: intolerance of uncertainty, erroneous beliefs about worry, poor problem solving and cognitive avoidance. Clearly this draws on some of the same empirical and conceptual base as Wells and Borkovec. In addition, however, a distinction is made between worries that are amenable to problem solving and worries about situations that cannot be resolved or which may never occur. Cognitive exposure is thought to be ideally suited to the second but not the first type of worrying.
Although chronic worry is the primary focus of psychological theorising it is important to remember that this state of mind is not necessarily accompanied by the physiological symptoms of increased motor tension, vigilance and behavioural inhibition that are required for a diagnosis. Worry always carries some degree of emotional charge but it may be best viewed as a core feature of anxious temperament rather than as the core feature of GAD per se, that is, as a personality trait that results in an episode of GAD when somatic symptoms of arousal are added in response to a period of acute or chronic stress (cf. Akiskal, 1998; Rickels & Rynn, 2001). There are dangers of oversimplification, therefore, when GAD is seen as primarily a disorder of maladaptive cognitive processing in which worry is the essential defining feature, which drives the emotional response. Both somatic symptoms and chronic worrying are features of GAD, with varying emphases depending on the individual concerned. Theories of GAD need to account for both cognitive and biological maintaining factors and their joint relation to proximal environmental stressors and developmental vulnerabilities (Goldberg & Goodyear, 2005). Generalised anxiety disorder is a complex emotional response in which cognitive appraisal processes are likely to be one significant causal factor and associative learning processes linking events with anxious arousal another (cf. Power & Dalgleish, 1999).
PSYCHOLOGICAL TREATMENTS Standard Treatment Strategies
Guidance for therapists on the psychological treatment of GAD can be found in a number of reviews and treatment manuals (Andrews et al., 2002; Borkovec & Newman, 1999; Brown et al., 2002; Clark, 1989; Wells & Butler, 1997). Guidance for sufferers can be found in sections of general self-help books (such as Butler & Hope, 1996), treatment manuals (Andrews et al., 2002) and books focused solely on worry and how to cope with it (such as Hallowell, 1997). All of these sources adopt as their primary focus the broad 'coping skills' approach that is characteristic of cognitive behavioural therapy. Habitual, maladaptive cycles of worry and tension need to be reversed through systematic, repetitive focus on, firstly, understanding the personal triggers and patterns of thinking, behaving and feeling that underlie the disorder and, secondly, on acquiring active habits of reacting to these triggers with more balanced thinking, lower arousal, and active problem solving.
The standard therapies for GAD, evaluated in clinical trials from 1980 to 2000, have been broadly based on the procedures and principles of cognitive therapy as applied to the appraisal of threat (Beck, Emery & Greenberg, 1985) and behaviour therapy as applied to reducing muscle tension through relaxation training (Bernstein & Borkovec, 1973; Ost, 1987). These approaches will be familiar to therapists with a basic training in cognitive behaviour therapy (CBT). The primary emphasis is on a process of self-regulation in which the sufferer learns to understand and then interrupt his or her particular cycle of anxiety triggers, bodily responses and worries with coping strategies based on either reducing arousal and muscle tension or changing the beliefs, and appraisal processes that underlie worrisome thinking. Clinical and research evidence suggests that confidence in either approach can bring about significant reductions in the severity of GAD but confidence in both is probably most effective. As GAD sufferers are often socially anxious, lacking in self-esteem and demoralised by life circumstances, therapeutic strategies involving graded exposure, assertiveness training and problem solving may also be helpful.
At its most ambitious the overall goal of therapy is to bring about a fundamental shift in coping from an essentially passive, worried, inhibited reaction to potential threat to an active, problem-solving approach in which stressful events and emotional reactions are accepted and managed. For those GAD sufferers who accept worry and tension as an unchangeable aspect of their personality and have only limited insight into the maladaptive nature of the disorder, therapists face a considerable change in bringing about active engagement with the demands of therapy. Motivational issues are frequently complicated by the presence of comorbid disorders and effective therapy is likely to depend crucially on the collaborative development of a formulation that emphasises the central role of worry and tension in maintaining the problems perceived to be most distressing.
More than 30 clinical trials of the efficacy of psychological treatments for GAD were published between 1975 and 2002, about half of which used DSM-defined diagnostic criteria from DSM-III onwards. The evidence is largely limited to CBT as this form of psychotherapy has been the main focus of evaluation in randomised controlled trials. A few non-CBT psychotherapies have been evaluated but only in the context of control conditions for testing the efficacy of CBT against non-specific treatment effects Most of the early trials compared 'cognitive-behavioural' therapies (cognitive restructuring, relaxation training, biofeedback, systematic desensitisation, anxiety management training) with no treatment, waiting list or psychological placebo. More recent trials have employed more complex and sophisticated combinations of behavioural and cognitive therapies with a more specific focus on worry. Direct comparisons of psychological therapy with pharmacotherapy are very few in number, most notably a comparison of diazepam, CBT and placebo, each alone and in combination (Power et al., 1990), and a comparison of drug and psychological therapy for GAD, panic disorder and dysthymia (Tyrer et al., 1993). Systematic reviews of these clinical trials have addressed a number of questions - the most important, for initial consideration, being whether or not treatment is more effective than no treatment.
A meta-analytic review by Gould and colleagues (Gould et al., 1997) addressed treatment efficacy for 13 studies comparing psychological therapy with no treatment, wait-list or psychological placebo and 22 studies comparing pharmacotherapy with pill placebo. Both within- and between-group effect sizes were calculated for anxiety and depression measures at post-treatment and, where available, for follow-up. Length of treatment was fairly short (three to nine weeks for pharmacotherapy and six to 15 weeks for psychological therapy) and follow-up data limited to six months in six of the psychological treatment studies. The majority of studies allowed comorbid anxiety disorders as long as GAD was the primary disorder. The results indicated that, for severity of anxiety symptoms at post-treatment, both CBT and pharmacotherapy were superior to control conditions and of broadly similar efficacy with moderately large effect sizes (ES = 0.70 for CBT, ES = 0.61 for pharmacotherapy). For severity of depressive symptoms CBT was associated with a significantly greater antidepressant effect than pharmacotherapy (ES = 0.77 for CBT, ES = 0.46 for pharmacotherapy). Among CBT interventions there was evidence that the combination of cognitive and behavioural techniques was more efficacious than each used independently. The limited evidence on follow-up for CBT conditions suggested that therapeutic gains were largely maintained and the authors contrasted this finding with other evidence that the long-term efficacy of pharmacotherapy (mainly benzodiazepines in this review) was attenuated following medication discontinuation.
In the short term, therefore, it might be reasonable to conclude that CBT and phar-macotherapy are both efficacious but that CBT, importantly, may have the edge in the maintenance of treatment gains and in treating comorbid depression. Benzodiazepines, however, are no longer considered to be an appropriate first-line treatment for GAD because of dependency and withdrawal problems and an increased risk of sedation and industrial and road traffic accidents with prolonged treatment. In contrast, the newer antidepressants (such as the SSRI, paroxetine) can be effective with GAD and comorbid depression, in both the short and longer term, and carry a reduced risk of dependency and rebound withdrawal problems (Ballenger et al., 2001; Davidson, 2001). The safest conclusion is that both CBT and antidepressants are broadly equivalent in efficacy and more effective than no treatment.
A more recent review by Borkovec & Ruscio (2001) analysed effect sizes for 10 of the same studies as in the previous review plus three studies not included, one of which compared CBT with psychoanalytic therapy (Durham et al., 1994) and one of which evaluated CBT for GAD in older adults (Stanley et al., 1996). This review is of interest in including a detailed breakdown of the methodological characteristics of the studies (described as generally rigorous) and in analysing effect sizes for four aggregated comparison groups: CBT, CT or BT, placebo or alternative therapies, wait list/no-treatment. Effect-size calculations were based on the five most commonly used outcome measures for anxiety (assessor rated severity of GAD on a 0-8 scale, Hamilton Rating Scale for Anxiety, and the Trait version of the State-Trait Anxiety Inventory) and depression (Hamilton Rating Scale for Depression and Beck Depression Inventory). The outcome of the analysis was consistent across studies with CBT clearly associated with the largest within-group and between-group effect sizes relative to all other comparison conditions. The absence of change in the wait-list/no treatment conditions is particularly striking in this review with mean within-group effect sizes on anxiety measures at post-treatment being 2.48 for CBT conditions and 0.01 for wait-list no treatment conditions. In the absence of treatment GAD shows little change and the spontaneous remission rate is estimated at around 20-25 % (Ballenger et al., 2001).
Cognitive behaviour therapy is clearly established as an efficacious treatment for GAD (cf. Roth & Fonagy, 1996; Westen & Morrison, 2001), and is recommended as a first line treatment in the NICE guidelines (National Institute for Clinical Excellence, 2004), but what about the clinical significance of treatment effects? Fisher & Durham (1999) reviewed six clinical trials conducted between 1990 and 1998, all of which used rigorous selection criteria and employed the State-Trait Anxiety Inventory (STAI-T) (Spielberger et al., 1983) as a common outcome measure. Jacobson's methodology for determining clinically significant change (Jacobson et al., 1999) was applied to the STAI-T to allocate individual participants (total n = 404) to one of four categories: reliable deterioration, no change, reliable improvement within the dysfunctional population and recovery. A recovery rate of about 40 % was found for the sample as a whole with 12 of the 20 treatment conditions obtaining very modest recovery rates of 30 % or less. The best results were found with individual CBT and applied relaxation, which had recovery rates at six months follow-up of 50-60%.
Evidence of treatment efficacy over the longer term (several years or more) is limited. Seivewright and colleagues conducted a five-year follow-up of a cohort of 210 psychiatric outpatients suffering from GAD, panic disorder or dysthymic disorder and randomised to medication, CBT or self help (Seivewright et al., 1998). Sixty per cent had a broadly favourable outcome with the remainder handicapped either intermittently or continuously throughout the follow-up period. Of relevance to the current discussion is that neither initial diagnosis or treatment condition was found to be of predictive value. A 10 to 14 year follow-up of two clinical trials was recently completed in central Scotland using structured interview with an assessor blind to initial treatment condition (Durham et al., 2003). One of the studies (Power et al., 1990), had been based in primary care using DSM-III criteria, and had compared CBT with a benzodiazepine and placebo whereas the other (Durham et al., 1994) had been based in secondary care and compared CBT with analytic psychotherapy. Follow-up samples were relatively low (30 % and 55 % of trial entrants respectively) but broadly representative of the original cohorts. Overall, 30 % to 40 % of participants were recovered (free of symptoms) and 30 % to 40 % did poorly. Outcome was significantly worse for the study based in secondary care in which the clinical presentation of participants was more complex and severe. Treatment with CBT was associated with significantly lower overall severity of symptomatology and less interim treatment, in comparison with non-CBT conditions, but there was no evidence that CBT influenced diagnostic status, probability of recovery or patient perceptions of overall improvement.
Treatment gains following CBT may, therefore, be less enduring than had been assumed from clinical trials with six- to 12-month follow-up periods. Sustained reductions in vulnerability to episodes of GAD, or other anxiety and depressive disorders, may occur in only a minority of people following a course of therapy. Some do well over the shorter term but relapse following stressful events and still others suffer a chronic course with continuing disability and distress. It seems reasonable to conclude that both CBT and the complexity and severity of presenting problems have a significant influence on the long-term outcome of GAD. This raises two important questions for the future: can we identify the characteristics of subgroups of people with GAD who do well and who do poorly and can we improve the power of current treatments and the quality of service delivery?
The search for reliable factors that moderate response to therapy is a natural step given the evidence of variable outcome and is of considerable theoretical importance. Unfortunately, small sample sizes and diverse measures, to name just two methodological difficulties, have limited growth of knowledge in this area (Durham, 2006). It is probably helpful at the outset to distinguish between general prognostic factors indicative of the overall likelihood of change for a person having GAD, irrespective of treatment received, and more specific treatment response indicators, reflecting that person's willingness and ability to engage with the demands of therapy. The former concern the complexity and severity of the person's presenting symptomatology and underlying vulnerability, the quality of their social adjustment and the severity of associated psychosocial stressors, whereas the latter concern the perceived suitability and power of the treatment being offered, the quality of the therapeutic alliance and the degree of engagement in therapeutic tasks.
A conceptual framework for outcome prediction along these lines has been described by Durham, Swan & Fisher (2000) who suggest that general prognostic factors will be most strongly related to outcome over the longer term and specific treatment response indicators will be most closely related to short- and medium-term outcome. They devised scales to reflect these two factors on the basis of both general research on prognostic factors in anxiety and depressive disorders and GAD clinical trial data (cf. Durham et al., 1997). General prognostic indicators are reflected in the CASP index (Complexity and Severity of Presenting Problems) while treatment response indicators are reflected in the CAIR index (Collaborative Alliance and Initial Response). The CASP is a simple additive scale based on a yes/no rating of the following eight factors:
• previous psychiatric treatment;
• single, widowed or divorced;
• significant relationship difficulties;
• low socioeconomic status;
• high self-reported symptomatology;
• high clinician-rated clinical global severity.
Of these variables, the importance of comorbidity (axis 1 and 2), social adjustment/ marital tension, global severity of disorder and, to a lesser extent, low socioeconomic status/unemployment, have received independent support (cf. Borkovec et al., 2002; Yonkers et al., 2000). The CASP index was used in a clinical effectiveness study to test whether or not more intensive CBT (15 versus nine sessions) was more effective with poor prognosis GAD patients (Durham et al., 2004). A comparison group of good prognosis GAD patients (low scores on the CASP) was given brief CBT (five sessions). There were no differences in outcome between the levels of treatment for the poor prognosis patients and both did relatively poorly at post-treatment and six month follow-up. In contrast, the good prognosis condition did significantly better at six-month follow-up than poor prognosis patients despite receiving only brief treatment. This result lends some support for the validity of the CASP but further refinement and testing of the scale is clearly needed.
Generalised anxiety disorder is clearly a difficult condition to treat effectively on account of both its chronicity and complexity of presentation and our incomplete understanding of the essential vulnerabilities and psychological mechanisms that maintain the disorder. Although clinical trials have established that cognitive behavioural therapies, based on standard behavioural and cognitive strategies, are more efficacious than no treatment or alternative therapies, treatment effects are generally modest and may not be sustained over the longer term. Preliminary evidence on treatment efficacy in routine clinical practice suggests that recovery rates may be lower outside specialist treatment centres.
Three issues need to be addressed if we are to improve the effectiveness of psychological treatment. First, as argued persuasively by Ballenger et al. (2001), there is a need to increase awareness of the importance of the disorder as a potential source of long-term damage to physical and mental health, as well as quality of life, and as a suitable focus, therefore, for clinical assessment and intervention. Generalised anxiety disorder is frequently overlooked in healthcare settings, and consequently undertreated, either because presenting complaints such as fatigue or sleeping problems are given more weight than the underlying pattern of chronic worry and tension of which they are a part or because its presence is masked by other disorders, which are judged to be primary and which become the main focus of attention. The importance of GAD and the key questions required to make a diagnosis need greater emphasis in the education of healthcare professionals.
Second, it will be of value for both clinical and health economic reasons to develop prognostic indices that identify those people who are likely to respond positively to brief intervention and those who may require more intensive therapy and long-term management. Generalised anxiety disorder is a common condition, trained therapists are scarce and, for the large number of people with less severe or subthreshold forms of the disorder, self-help and brief CBT is likely to be the most effective and efficient clinical response. There are well-established programmes of this kind (White, 2000; Williams, 2003) that can be incorporated into a stepped care model of service delivery as advocated in the NICE guidelines (National Institute for Clinical Excellence, 2004). For those people whose prognosis is poor, or who fail to respond to brief intervention, more intensive therapy will be required. The exact nature of this more intensive therapy has yet to be established and this is the third issue that needs to be addressed.
The promise of theoretical advances in our understanding of the nature of chronic worry and tension is the development of treatment strategies that target more precisely those vulnerabilities and psychological processes that keep the disorder going. This has been one of the central themes of clinical researchers such as Adrian Wells and Tom Borkovec and there is no doubt that progress has been made. Candidates for future clinical trials include worry exposure and worry behaviour prevention (Brown, O'Leary & Barlow, 2002), cognitive therapy focussed on metacognition (Wells, 1999), therapy for interpersonal difficulties (Crits-Christoff et al., 1996; Newman etal., 2002), mindfulness training (Roemer & Orsillo, 2002) and a therapeutic focus on intolerance of uncertainty (Ladoucer et al., 2000). A recent analysis of the efficacy of those newer therapies by Fisher (2006) does suggest that they can deliver significantly higher recovery rates although large scale clinical trials have yet to be undertaken.
Was this article helpful?