The goal of the treatment of Generalized anxiety disorder is the reduction of impairment that results from both cognitive and somatic symptoms of anxiety: the worry or anxious expectation, and the accompanying symptoms of tension and overarousal. This chapter aims to summarize the evidence for the effectiveness of psychological and pharmacological treatments for Generalized anxiety disorder.
Since the first edition of this book was published, five reviews of the psychological treatment of Generalized anxiety disorder have been published. All agree that a cognitive behavioral approach is effective for this disorder, a finding that is consistent with previous quantitative and qualitative reviews. Hence, there is general agreement that cognitive behavioral therapy yields statistically and clinically significant improvement for the majority of patients, and that this change is maintained for up to a year following the end of treatment.
However there is less agreement on the extent to which different therapies produce differential effects. For example, Barlow et al. (1998: 311) stated “Until recently, most studies have not demonstrated differential rates of efficacy for active treatment techniques, although most studies have shown that active treatments are superior to non-directive approaches and uniformly superior to no treatment”. Fisher and Durham (1999) applied Jacobson’s criteria for clinical significance to six randomised controlled trials using the State-Trait Anxiety Inventory, Trait Version (STAI-T) as an outcome measure and showed that both cognitive behavioral therapy and applied relaxation produced recovery rates at 6-month follow-up of 50% to 60%. On the other hand, their results indicated that some individual therapies (behavior therapy and analytical therapy) were ineffective, with recovery rates of 11% and 4%, respectively. Of further interest was the finding that individual therapy produced higher recovery rates than the same treatment delivered in a group format. Fisher and Durham (1999) concluded that while cognitive behavioral therapy may not be the only treatment to deliver reasonable treatment outcome effects, it is unlikely that the effects observed were solely due to spontaneous remission, regression to the mean, or placebo factors.
Gould et al. (1997b) reported the results of a meta-analysis of 22 controlled trials of broadly defined “cognitive behavioral therapies”. They concluded that there were no significant differences between the eight trials of cognitive and behavioral techniques (mean anxiety effect size of 0.91) and the three trials of relaxation training (mean anxiety effect size of 0.64). Of note was the finding that the mean effect size for measures of depression was also substantial across the “cognitive behavior therapies” (mean depression effect size of 0.77). In the trials where follow-up data were available, the analysis confirmed that the beneficial effects of therapy were maintained over time. Analyses were conducted to statistically assess the influence of a range of variables – including sex, duration of disorder, use of concurrent medication, group versus individual format, or treatment duration – but no variable was found to be significantly associated with outcome.
Combining information from a range of available evidence, Gale and Oakely-Browne’s (2000) contribution to the British Medical Journal Clinical Evidence series concluded that the only intervention likely to be effective with a high degree of reliability is cognitive behavioral therapy. The approach used in this review grades intervention studies in their ability to predict treatment effectiveness (with systematic reviews with meta-analysis of randomized controlled trials seen to provide the most reliable evidence and expert opinion as the least reliable). Cognitive behavioral therapy was deemed to be effective on the basis of two systematic reviews of randomized controlled trials that found it to be more effective than remaining on a wait-list, anxiety-management training alone, or nondirective therapy. However, applied relaxation was deemed to be of unknown effectiveness as one systematic review had not established or excluded a clinically important difference between applied relaxation and cognitive therapy.
In conclusion, the reviews are largely in agreement despite the use of different methods to summarize the treatment outcome results. Cognitive behavioral therapy is certainly superior to no treatment and appears to be superior to nondirective treatments. Of interest, however, is a recent trial in adults over 55 years of age that found no differences between group-based cognitive behavior therapy or supportive psychotherapy at posttreatment and at the 6-month follow-up across all measures of anxiety, worry, and depression. As Stanley et al. pointed out, it remains to be seen whether potential differences in the phenomenology of Generalized anxiety disorder in older adults or the use of a group-based program account for this finding.
The evidence is less strong that cognitive behavioral therapy is superior to other active treatments, specifically applied progressive muscle relaxation. Six published controlled treatment studies comparing active treatments have used patient samples meeting DSM-III-R criteria, hence providing a closer match than earlier studies to current diagnostic criteria. Consistent with the conclusions of the reviews, all studies showed that the treatment conditions produced significantly greater improvements than wait-list or nondirective therapy controls.
Barlow et al. (1992) compared applied progressive muscle relaxation, cognitive restructuring, and a combination of relaxation and cognitive restructuring treatment conditions. Over the 2-year follow-up there was no significant difference observed between the three treatment groups, and, while medication use was substantially reduced, the patients remained considerably anxious. However, the results must be tempered by the high rates of drop out from the active treatments and high attrition at follow-up. As only 10, 13 and 11 patients completed treatment in the three conditions, there may have been insufficient power to detect significant group differences. White et al. (1992) compared a didactic “stress control” group therapy that was cognitive, behavioral (relaxation, exposure and respiratory control) or cognitive behavioral in focus. A placebo treatment (subconscious retraining) and a wait-list control were also employed. All treatment sessions were conducted in large groups (20 to 24 participants in each group) at a local health center. All groups with the exception of the wait-list showed improvements over the treatment duration, and with the exception of the placebo condition, showed further improvements over follow-up. No significant differences were found between the active treatment groups. Borkovec and Costello (1993) compared applied relaxation and cognitive behavior therapy (which included a substantive applied relaxation component and therefore relatively brief cognitive therapy) and found no difference in outcome between the two treatment groups at posttreatment or the 6 month follow-up. Lastly, Ost and Breitholtz (2000) compared 12 sessions of weekly cognitive therapy against applied relaxation. There were significant improvements on most indices across treatment and 1-year follow-up, but no differences between the two treatments. When there is a failure to find a statistically significant difference between treatment conditions, it is useful to consider whether studies have had adequate power to detect a clinically significant different treatment effect. The sample sizes in each treatment condition required for the detection of a large effect size (say 0.40) as statistically significant are 26 for two-group and 21 for three-group comparisons (given a significance criterion of 0.05 and power of 0.80). Only White et al. (1992) had sample sizes in each treatment condition large enough to detect such differences; it may be possible that the conduct of the therapy in such large-sized groups may have diluted the treatment effects.
Two studies have demonstrated that cognitive or cognitive behavioral therapy produced significantly greater improvement than the other active treatments tested. Butler et al. (1991) compared cognitive behavioral therapy to behavior therapy and found at the 6-month follow-up that cognitive behavioral therapy produced significantly greater improvements than behavior therapy on measures of anxiety, depression and cognition. Furthermore, there were fewer dropouts from therapy in the cognitive behavioral condition. Durham and colleagues compared cognitive therapy, analytical therapy (delivered by experienced therapists across two levels of intensity) and anxiety-management training (delivered by inexperienced therapists). At the 1-year follow-up cognitive therapy was significantly more effective than analytical therapy or the anxiety-management training. The results suggested that cognitive therapy delivered weekly showed greater improvement over the follow-up on some measures than when delivered fortnightly. The authors suggest that a higher intensity treatment may be important in a disorder that is often characterized by high rates of comorbidity and disablement, which in themselves appear to influence prognosis following treatment.
As worry has been given a central role in Generalized anxiety disorder, it seems important that specific therapeutic strategies be directed towards the cognitive aspects of this disorder. We saw that individuals who suffer from Generalized anxiety disorder describe their worries as difficult to control, excessive and hence out of proportion to the actual reality of the feared outcome. The characteristics of these cognitive aspects are amenable to cognitive therapy that aims to address unrealistic and erroneous beliefs, attitudes, and expectations. The efficacy of cognitive behavioral treatment packages for this disorder tend to support the case for such an approach to the amelioration of worrying thoughts, particularly in studies in which the measurement of worry has been included. In further support is the finding from one study that higher levels of apprehensive expectation or worry prior to cognitive behavioral treatment was significantly predictive of greater anxiety symptoms following treatment, irrespective of the level of anxiety symptoms prior to treatment.
More recently there has been an emergence of treatment packages specifically designed to address worry, with the rationale that a decrease in worry will produce concomitant changes in related subsystems (such as a decrease in somatic symptoms of anxiety) without these subsystems being specifically targeted (such as by relaxation training). Such treatments have been derived directly from research into the psychopathology of worry and conceptual models of Generalized anxiety disorder. Ladouceur et al. (2000a) conducted a wait-list controlled trial of a cognitive behavioral treatment that specifically targeted intolerance of uncertainty, erroneous beliefs about worry, poor problem orientation and cognitive avoidance. The treatment group showed significantly greater improvement than the wait-list group on all outcome measures that included self-report, clinician-rated and significant other ratings of Generalized anxiety disorder and associated symptoms. Furthermore, the treated sample showed statistically and clinically significant improvements following treatment, which was maintained at the 6- and 12-month follow-ups. Another specific approach to worry has been that of “worry exposure” in which the worry is targeted using an exposure-based paradigm. In short, repeated and controlled exposure to the imagery and thoughts associated with the “worst possible outcome” in each worry domain allows habituation of the anxiety associated with that worry. The results of clinical trials of this approach are eagerly awaited.
These newer approaches to the treatment of Generalized anxiety disorder are particularly important, given that the size of treatment effects and indices of clinically significant change have been modest for Generalized anxiety disorder, at least compared with the outcome for psychological treatment of other anxiety disorders. In discussing these modest treatment outcomes, Ost and Breitholtz (2000) pointed out that treatments that were developed for other disorders (such as cognitive therapy for depression and applied relaxation for panic and phobias) may not have specific relevance for Generalized anxiety disorder. Further research into the psychopathology of worry will help to further develop treatments that specifically target worry and hence improve the effectiveness of current treatments for this disorder.
Commenting on the high proportions of individuals with Generalized anxiety disorder failing to achieve high end-state functioning following treatment, Newman (2000) has argued that more attention needs to be paid to the predictors of treatment nonresponse, including symptom severity, comorbid depression, interpersonal problems, and avoidance of emotional processing. Newman has argued for the need for outcome studies to incorporate strategies aimed at interpersonal difficulties or depth of emotional processing within cognitive behavioral treatments. These strategies might be specifically targeting individuals who, because of these additional difficulties, may not fully respond to standard cognitive behavioral intervention. At the other end of the spectrum, the evaluation of low-cost minimal interventions will be important to determine whether there are a significant number of individuals who can learn to control their symptoms without specialist-level care. While yet to be compared with an active treatment condition, a self-help package requiring individuals to examine what was important in their lives has been found to produce significant improvement in anxiety symptoms in Generalized anxiety disorder compared to a wait-list control. Findings such as these may be particularly important in providing treatment options for a disorder that is common in primary care and whose sufferers tend not to reach specialist treatment programs.
Pharmacological Treatments of Generalized Anxiety Disorder
Combining psychological and pharmacological treatments
The common finding of relapse following discontinuation of benzodiazepine therapy has led some authors to address the issues of long-term use of benzodiazepines and adjunctive teaching of behavioral and cognitive coping strategies. Rickels and Schweizer (1990) have speculated that constant use of benzodiazepines may preclude patients from developing their own coping skills for anxiety relief, and suggest the concurrent use of such interventions as behavior modification or cognitive therapy. The treatment study of Power et al. (1990) tended to support the view that psychological and pharmacological treatments can be combined without a reduction in effectiveness. Power’s study included a trial of combined cognitive behavioral therapy and fixed low dose (3×5 mg daily) diazepam. They found no significant differences posttreatment between the combined treatment and cognitive behavioral therapy alone or with placebo on a number of symptom measures of anxiety. No significant withdrawal symptoms or re-emergence of anxiety symptoms were reported following careful and gradual withdrawal from the diazepam. At 6-month follow-up over 85% of the combined group showed “clinically significant change” (greater than 2 standard deviations from the pretreatment mean) and 84% had not sought further treatment. This group made an interesting contrast from the diazepam-alone group, which showed higher rates of subsequent treatment (54% of the group) and lower rates of “clinically significant change” (30% to 70%) at 6 months following treatment. While the findings of Power et al. (1990) require replication, they do suggest that the judicious use of long-acting benzodiazepines may not significantly interfere with cognitive behavioral therapy in individuals who have sought treatment for Generalized anxiety disorder. However, the cognitive behavioral treatments achieved similar clinical change in the absence of active pharmacological treatment. Therefore, while the use of benzodiazepines may not interfere with cognitive behavioral therapy, they do not appear to add to the treatment effect and the potential for adverse effects must be taken into consideration. In contrast, a review of psychological treatments in Generalized anxiety disorder found that, across a number of studies, outcomes indicating above-average improvement in symptom measures and clinically significant changes were associated with patient samples that were free from anxiolytic medication. No data appear to be available concerning the combined use of psychological treatment and antidepressant medication. Clearly, the role of pharmacotherapy as an adjunct to psychological treatment in Generalized anxiety disorder has yet to be established.
The cognitive behavioral therapies appear to be at least as effective in the short term as pharmacotherapy, cause no adverse effects in terms of side-effects and withdrawal syndromes, and aim to increase coping skills, and hence increase the sense of control and mastery in patients. In other anxiety disorders, the same strategies have been shown to bring about long-term changes in measures thought to represent vulnerability to neurosis. Our present knowledge suggests that treatment should include, at the very least, education about the nature of the disorder and should address the beliefs, attitudes, and expectations relevant to an individual’s worries and fears.
Despite the existence of efficacious treatments, there is still much to be done. For example, it will be important to replicate many of the reported treatment effects in DSM-IV samples and in the delivery of treatment in routine care. Furthermore, the development of treatments appropriate to the primary care setting will be important for the majority of Generalized anxiety disorder sufferers who will never reach specialist treatment settings. It will also be important for future outcome studies to target the prominent feature of worry in terms of assessment and treatment.