Cognitive and Behavioral Practice 20 (2013) 301313 www.elsevier.

com/locate/cabp

Metacognitive Therapy for Generalized Anxiety Disorder: Nature, Evidence and an Individual Case Illustration
Odin Hjemdal, Roger Hagen, and Hans M. Nordahl, Norwegian University of Science and Technology Adrian Wells, Norwegian University of Science and Technology and University of Manchester
Metacognitive therapy (MCT) is based on over 25 years of research focusing on the processes that contribute to the development and maintenance of psychological disorders. The approach identifies a common set of processes in psychopathology, and MCT shows promising results in effectively treating a range of disorders. This paper presents the central theoretical tenets of MCT and uses a clinical vignette to illustrate the structure and techniques of treatment based on Wells's (2009) manual as they relate to a specific case of generalized anxiety disorder.

What Is Metacognition and Why Is It Important?

This paper provides a general introduction to the theory of metacognitive therapy (MCT) and a more specific outline of how to use MCT for generalized anxiety disorder (GAD), illustrated with the clinical case of William. In the final part of the paper the scientific evidence for MCT in GAD is presented. MCT was developed to address the control of cognition and the strategies and knowledge that govern thinking. It contrasts significantly with the theory and focus of standard CBT. Metacognition refers to cognition applied to cognition and may be defined as any knowledge or cognitive processes involved in the appraisal, control, and monitoring of thinking (Flavell, 1979). In short, metacognition is thinking about thinking. Metacognitive theory has distinguished between metacognitive knowledge, which is information that individuals have about their own thinking and about strategies that affect it, and metacognitive regulation, which are the strategies used to change the nature of processing. In the metacognitive theory of psychological disorder (Wells, 2009; Wells & Matthews, 1994), metacognition is central in determining the maintenance and control of negative and biased thinking styles. According to Wells, most people have negative thoughts and beliefs and in most cases these thoughts and beliefs are transitory mental experiences. The
Keywords: metacognitive therapy; generalized anxiety disorder; anxiety; cognition; case study

1077-7229/12/301-313$1.00/0 2013 Association for Behavioral and Cognitive Therapies. Published by Elsevier Ltd. All rights reserved.

negative thoughts become a problem because of the way an individual responds to them. Thus, an important tenet of metacognitive therapy, and one of the features distinguishing it from traditional CBT, is that neither the content nor the subjective validity of thoughts and beliefs are the central source of disorder. In basic terms, according to metacognitive theory, an individual's metacognitions monitor and control their responses to thoughts, which cause persistence or perseveration of ideas and maintain psychological and interpersonal problems. This supposition can be clearly illustrated in the situation of GAD, where the content of worry shifts around. The content of worry in GAD is not dissimilar from everyday worries experienced by most people. However, people with GAD experience their worry as uncontrollable and excessive, and it is associated with marked distress. The metacognitive model provides an explanation of this in terms of differences in the way individuals relate to, appraise, and control their worry The theoretical grounding of MCT is the Self-Regulatory Executive Function Model (S-REF), which emphasizes the similarities in maladaptive cognitive processing across all psychological disorders (Wells, 2000, 2009; Wells & Matthews, 1994, 1996). The S-REF model postulates a thinking style called the cognitive attentional syndrome (CAS). In MCT the CAS is a universal feature of psychiatric disorders and is responsible for prolonging and intensifying distressing emotions. The CAS is a thinking pattern of inflexible self-focused attention (the focus is on self-observation and monitoring of thought processes), perseverative thinking (in the form of worry and rumination), threat monitoring, and coping behaviors that backfire and

302

Hjemdal et al. formed and/or activated. These beliefs fall into two domains: that worry is uncontrollable and that worry is dangerous for mental or physical well-being. When these beliefs are triggered the individual begins to worry about worry (Type 2 worry or meta-worry), which leads to an increase and prolongation of anxiety symptoms. Anxiety can escalate rapidly due to Type 2 worry and occur as panic attacks because of the more imminent threat considered to be posed by worry itself. In response to meta-worry, the individual engages in thought-control strategies and different behaviors aimed at reducing worry and/or the threat it presents. Many of these responses have paradoxical effects that interfere with effective mental control and the development of more adaptive meta-beliefs. For example, a person with GAD may ask a partner for reassurance, which effectively transfers the control of worry to someone else, thus depriving the individual from learning that he or she has control. In some cases the person will search the Internet for information in an attempt to assuage worry or anxiety, but this can actually increase exposure to ambiguous and threat-related informationa further trigger for worrying (e.g., natural disasters, accidents, crime rates, information on specific diseases and accompanying symptoms). Other unhelpful strategies include trying to suppress thoughts that might trigger worrying and/or having to sustain thinking in order to think oneself out of worry. These strategies, described above, simply extend the person's engagement with negative thoughts. As a consequence, such responses reinforce or maintain meta-beliefs about loss of control and an inability to cope. The MCT model is illustrated in Figure 1. Empirical Support for the Model There is substantial empirical evidence supporting this model. This section provides a brief summary of the evidence (see Wells, 2009, for a more detailed review). The negative effects of worrying for emotional and cognitive self-regulation have been demonstrated. Borkovec, Robinson, Pruzinsky, and DePree (1983) showed that brief periods of worrying led to greater anxiety, more depressive symptoms, and more negative thoughts in high compared with low worriers. It was also demonstrated that despite suffering with the negative consequences of worry, people with GAD had positive beliefs about worry (Borkovec, Hazlett-Stevens, & Diaz, 1999; Borkovec & Roemer, 1995). York, Borkovec, Vasey, and Stern (1987) also demonstrated that participants had more negative thought intrusions after the induction of worry than after a neutral condition. Following exposure to the stress of watching a distressing video, brief periods of induced worrying have been shown to be associated with an increase in intrusive thoughts about the stressor over 3 days (Butler, Wells, & Dewick, 1995; Wells & Papageorgiou, 1995). The use of

interfere with effective mental control and adaptive learning. The CAS is considered to be a problem for psychological well-being because it maintains threat-focused processing and fails to provide information that can modify the individual's maladaptive appraisals and beliefs. In addition to this, the CAS uses attentional resources that might otherwise be directed toward more adaptive responses, and biases perception and automatic processing in a negative way. There is a large evidence base supporting the presence and effects of the CAS in emotional disorders (see Wells, 2009, for a review). The CAS is driven by metacognitive beliefs and metacognitive knowledge stored in long-term memory, and MCT implies that all disorders are linked to this higher level of metacognitive beliefs about thinking. These beliefs fall into either positive or negative domains. Positive meta-beliefs concern the advantages of worrying, ruminating, threat monitoring, and controlling cognition (e.g., Worrying about the future helps me be prepared). Having positive meta-beliefs alone is not in itself pathogenic but increases the tendency to worry as a coping strategy, which does not provide the most effective way of managing negative affect and thoughts. According to MCT, psychopathology develops when negative meta-beliefs about loss of control and danger are activated. These beliefs concern the uncontrollability of worries and rumination and beliefs about the dangerousness or importance of thoughts. An example of a negative metacognition is: Worrying is out of control and will make me lose my mind. The patient with GAD can hold both positive and negative beliefs about worrying, which cause conflicting motivations to sustain or try to avoid negative thoughts. However, the negative beliefs are most important and lead to worry about worry resulting in elevated and persistent distress. The negative beliefs about the uncontrollability of the process contribute to the use of unhelpful forms of control or no control at all. Metacognitive Model of GAD The metacognitive model of GAD (Wells, 1995, 1997) proposes that when experiencing a negative thought (also called trigger thoughts; e.g., What if I can't cope with my work?), patients with GAD use extended negative thinking in the form of worry (Type 1 worry) to anticipate and work out ways of how to cope or avoid problems. Most people have positive beliefs about worry, but this is not the proximal feature of GAD. In MCT, the negative beliefs about worrying are considered to be the main cause of pathology, although the overuse of worry as a means of dealing with triggers may produce longer-term difficulties of impairing emotional processing and be unhelpful in the down-regulation of emotion. The metacognitive model proposes that GAD develops when negative meta-beliefs are

MCT for GAD

303

Figure 1. The metacognitive mode of generalized anxiety disorder. From Wells (1997), p. 204; published by Wiley; reproduced with permission licence number 2927550880379.

worry to cope with thoughts is also associated with a range of emotion disorder symptoms (Wells & Davies, 1994). An important tenet is that negative metacognition is central to the development of pathological worry as seen in GAD, and several studies have examined this assertion. Wells and Carter (1999) measured worry about worry (Type 2 worry) and Type 1 worry and demonstrated that the former was a stronger correlate of pathological worry scores than the latter. In a subsequent study with the sole emphasis on worry about the danger of worry belief, the frequency of Type 2 worry discriminated those individuals meeting threshold for GAD from individuals classified as having somatic anxiety or no anxiety (Wells, 2005a, 2005b; 2006). This effect could not be attributed to differences in the overall frequency of worrying since the effect remained when the frequency of Type 1 worry was statistically controlled. There are some initial indications that the positive association between Type 2 worry and pathological worrying may be stable across different ethnic groups. Nuevo, Montorio, and Borkovec (2004) replicated the study of Wells and Carter (1999) and extended it by examining the relationship between meta-worry and worry severity in older Spanish adults. Meta-worry consistently emerged as a significant positive correlate of pathological worry and the amount of interference from worrying, relationships that persisted even when trait-anxiety and Type 1 worry frequency were statistically controlled. Prospective studies address the causal status of metacognition. However, only one unpublished study has examined this in GAD. Nassif (1999, Study 2)

examined the longitudinal predictors of pathological worry and GAD over a period of 12 to 15 weeks. Meta-worry, but not Type 1 worry (assessed at Time 1), predicted the later development of GAD. Furthermore, negative beliefs about the uncontrollability and danger of worry measured at Time 1 predicted the later development of GAD, and this effect remained when trait-anxiety and Type 1 worry frequency assessed at Time 1 were partialled-out. These findings clearly need to be replicated in order to draw firmer conclusions. In the MCT model of GAD, the disorder results from meta-worry linked to negative metacognitive beliefs. Thus, there should be evidence that negative beliefs about worrying are significantly associated with GAD and should distinguish GAD from other disorders. Several studies confirm this prediction (Cartwright-Hatton & Wells, 1997; Davis & Valentiner, 2000). Wells and Carter (2001) compared patients with GAD against gender-matched patients with social phobia, panic disorder, and nonpatient controls. The GAD group could be distinguished from the other groups in endorsing significantly higher negative metacognitive beliefs in two domains: (a) uncontrollability and danger and (b) need for control focusing on negative consequences of not controlling thoughts. Differences in beliefs about uncontrollability and danger remained when the frequency of Type 1 worry was statistically controlled. A study by Ruscio and Borkovec (2004) is especially relevant in addressing the issue of the importance of worry versus metacognitions about worry in GAD. The study is notable for examining the differences between individuals who were high worriers but did not have GAD and those that met criteria for GAD. The results showed that worry was similar across both groups, but significant differences emerged in two separate negative belief domains: beliefs about the uncontrollability of worry and beliefs about the danger of worrying. Despite the emphasis given to negative metacognitive beliefs, positive beliefs are also viewed as contributing to the inflexible use of worrying as a coping strategy, but this is not necessarily unique to GAD and is a feature of the metacognitive account of general stress vulnerability (Wells & Matthews, 1994). Interview studies and questionnaire data support this assertion (Cartwright-Hatton & Wells, 1997; Davey, Tallis, & Capuzzo, 1996; Wells & Papageorgiou, 1998). Individuals with GAD and those without do not show significant differences in the endorsement of positive meta-beliefs about worrying (e.g., Cartwright-Hatton & Wells, 1997; Wells, 2005a, 2005b), but these beliefs are positively correlated with worry tendencies. Thus, it could be useful to challenge these positive meta-beliefs as a general stress prevention strategy and to increase flexibility in responses to stress and negative thoughts.

304

Hjemdal et al.

MCT for GAD

The first step in MCT is to generate a case conceptualization based on the model described above. This forms the basis of interventions aimed to reduce unhelpful coping strategies, bringing worry under effective control and modifying metacognitive beliefs. The second step is to socialize the patient so that he or she learns that the problem is not worry itself, but the patient's own beliefs about worry. Metacognitive beliefs are modified through experiencing new types of relationships with cognitive processes. In the case of GAD, the focus is on the cognitive process of worry, which typically is perceived as a persistent and an uncontrollable process. MCT applies techniques such as detached mindfulness (Wells & Matthews, 1994; Wells, 2005a, 2005b) and worry postponement experiments. These strategies are used to replace the ineffective mental control strategies used by patients. Take the example of trying to suppress thoughts, which has the paradoxical effect of increasing their frequency, and thus often contributes to a perception of loss of control. Detached mindfulness and worry postponement are strategies that give the client increased awareness of control and ultimately change the client's relationship with the cognitive process of worry and with negative thoughts that trigger the experience. The goal is to increase effective control over worry and to challenge beliefs about its uncontrollability. However, the MCT therapist goes on to ensure that the patient is willing to try and lose control of worry later in treatment to challenge other metacognitions concerning the dangerousness of worrying. The treatment adheres to a particular sequence: case conceptualization and socialization are followed by (1) shaping meta-awareness of the distinction between thoughts that act as triggers and the worry response and learning new meta-level responses; (2) challenging beliefs about the uncontrollability of worry; (3) challenging beliefs about the danger of worry; (4) modification of positive beliefs about worry; and (5) relapse prevention. The GAD treatment usually consists of 10 to 12 treatment sessions administered in weekly 50-minute sessions (see Wells, 2009, for a detailed session-based treatment plan). These are the main elements of MCT. What follows is an illustration of MCT as applied to the case of William.

depressive symptoms prior to and during treatment, and in the case of William this is done by using the Beck Depression Inventory (Beck & Steer, 1993b) and the Beck Anxiety Inventory (Beck & Steer, 1993a), which are completed prior to each therapy session. Additional scales that might be useful, particularly pre- and posttreatment, are the Metacognitions Questionnaire (MCQ-30; Wells & Cartwright-Hatton, 2004) and the Penn State Worry Questionnaire (PSWQ; Meyer, Miller, Metzger, & Borkovec, 1990). It is very important to assess the degree to which the client endorses positive and negative meta-beliefs. When working with GAD using MCT, the Generalized Anxiety Disorder ScaleRevised (GADS-R; Wells, 2009), a self-report instrument, is used. The GADS-R assesses how disturbing worries have been for the last week, and how much effort the patient has used in controlling worries. The same scale also measures levels of positive and negative metacognitions, which is essential for the metacognitive therapist, along with the extent of typical maladaptive behaviors used by the patient to cope with or to avoid worry. Using the GADS-R in every session gives the therapist a good overview of the client's current level of positive and negative meta-beliefs, along with the use of maladaptive coping behaviors. It therefore contributes to targeting the factors that are currently the most important to work with at each stage of treatment.

Case Formulation in MCT for GAD

MCT treatment is described in the treatment manuals by Wells (1997; 2009) and always starts with a collaborative case formulation. The main aim of the case formulation is to capture the metacognitive beliefs that maintain the problem along with the use of maladaptive coping behaviors and avoidance. MCT has a structured and usually sequential way of doing this, and follows a case formulation interview (Wells, 2009), as illustrated in Figure 1. The case formulation starts by identifying the trigger that starts a worry episode (1). A trigger in MCT is always an internal intrusive thought, often in the form of a What if. . . ? question, but also sometimes occurs in the form of a negative image. The therapist typically tries to examine these triggers by asking questions such as, What was the first thought you had when you started to get anxious or started to worry? After the therapist has found the trigger for the worry episode, he then explores the thoughts following this trigger (Type 1 worry) by asking, When you had this initial thought, what kind of thoughts followed? and For how long did you worry? (2). This kind of Type 1 worry leads to an increase or persistence of negative emotions and symptoms. The therapist assesses these symptoms (3) by asking the patient, When you have these thoughts, how do they make you feel (emotionally and physically)? (4). During the worry episode, negative beliefs about worrying are activated. To

The Case of William

Assessment Using the ADIS-IV, William (see Robichaud, 2013-this issue, for the clinical case presentation) receives a primary Axis I diagnosis of GAD with a severity rating of 6 (on a 0 to 8 scale). The diagnostic assessment also indicates that he is experiencing some depression, which is a common comorbid problem for individuals suffering from GAD (see, e.g., Moffitt et al., 2007). It is typical to assess anxiety and

MCT for GAD explore these, the therapist looks for negative interpretation of worry (Type 2 worry). Specific questions are again used: When you had these thoughts and feelings, did you think something bad could happen as a result of worrying and feeling this way? What's the worst thing that could happen? These Type 2 worries reinforce negative beliefs of uncontrollability (5a) and danger (5b). These negative meta-beliefs are further explored with the patient by asking, Since worrying seems to be such a problem for you, why don't you just stop it? Do you think that worrying is uncontrollable in any way? If so, how much do you believe that worrying is uncontrollable on a scale from 0 (not at all) to 100 (completely uncontrollable)? The negative beliefs about danger are further examined by asking: Do you believe that worrying is harmful in any way? Do you think that worry may be harmful to your body or to your mind? If so, how much do you believe this on a scale from 0 (not at all harmful) to 100 (very harmful)?Although patients suffering from GAD always hold negative beliefs about worry, but also hold positive meta-beliefs (6). To assess these positive meta-beliefs the therapist asks the following questions: Are there times that you believe that worry may be useful in some way? Do you think that worrying could help you cope? Does worry help you foresee problems and avoid them? Are there any advantages of worrying? Finally, the therapist elicits the coping strategies (7) and thought control strategies (8a and 8b) used by the patient to cope with worry episodes. This part of the formulation is achieved by using questions such as, When you start worrying, what do you do to try and stop it? Probe questions help the patient to answer this: Do you avoid situations, distract yourself, ask for reassurance, and sleep more or less than normal, use alcohol, medication or other drugs? Thought control strategies are typically examined by asking the patients the following (8a): Do you also use more direct strategies like controlling your thoughts by trying to suppress your thoughts, like trying not to think certain thoughts or trying to control them in other ways? The therapist also asks: Have you tried to interrupt worry by deciding not to engage in these thoughts? MCT Case Formulation for William William describes several worry domains but primarily his trigger thoughts are linked to his health, family, and work. A trigger (1) for William is, What if I develop a serious illness? This trigger is followed by Type 1 worry (2) as a stream of thoughts (What if my wife had trouble handling all the bills if I died? What will happen to my kids, will they cope? What will my family do if they would have no money? Will they have to move? Will they suffer?). This kind of Type 1 worry can last for a long period of time. William worries more or less all the time and every day as a reaction to his different trigger thoughts. These different worries give rise to several emotional and physical symptoms (3). He

305

describes that it is difficult for him to focus and concentrate. In MCT one often finds that these difficulties are related to the ongoing worrying process. William also has difficulties falling asleep, feels tired all the time, and is irritable at work and at home. He experiences muscular tension, stomach problems, feels restless, anxious, and reports a low mood as a reaction to his worries. When William is asked about his Type 2 worryfor example, When you have these thoughts and feelings, did you think something bad could happen as a result of worrying and feeling this way?he may report that he is afraid that he will go crazy related to intense levels of worry and anxiety. In MCT, these negative metacognitive beliefs are important. Here William believes that he could go crazy (a danger belief about worry) and there is also the implicit or explicit belief that his worry is uncontrollable because it could lead to anxiety and losing his mind (4). He may think that because his worries seem to be always present and difficult to stop (because he spends about 75% of his waking day worrying), that this is powerful evidence that they are uncontrollable (5a). In metacognitive therapy it is important to assess how much he believes that the worries are uncontrollable, it would not be unusual for this belief to be around 75%. He worries that he may go crazy, and in metacognitive therapy we often find that patients actually worry that their worries will contribute to making them lose their minds (5b). Despite having these negative meta-beliefs about worrying described above, William continues to believe that there are advantages to managing worry with more thinking time (6). In particular, he worries about something happening to him, and responds by making a plan for the family's finances. His excessive planning has taken on proportions that interfere with the family's other activities. In the metacognitive approach the therapist interprets this as a positive meta-belief, common among GAD patients (i.e., I must make a plan or I won't be able to stop worrying). Unfortunately, the planning process is another form of extended thinking. In effect, William is trying to stop worrying by thinking more, when he needs to reduce the amount of thinking in response to negative ideas. William also uses other coping strategies (7): procrastination, seeking reassurance from others, and checking for information on the Internet, among other things. He also describes using thought control strategies, such as trying to distract himself from worrying in addition to suppressing his thoughts (8a). In particular, the thought, What if I develop a serious illness? triggers worrying. Most clients have never tried not to engage with triggers of this kind, and it is something new to them. It is therefore central to explore if William has ever experienced not engaging with the triggers and thoughts that normally lead to a worry episode (8b). The metacognitive case formulation for William is presented diagrammatically in Figure 2.

306

Hjemdal et al.
(1) What if I develop a serious illness?

(6). Worrying will help me solve problems

in life. Worrying will help me cope better. Worrying will help take the right decisions.

(2). What if my wife had trouble handling all the bills if I died? What will happen to my kids, will they cope? What if they would have no money?

(5a). Worrying is uncontrollable 75% (5b). Worrying will make me go crazy

(4). I will have a nervous breakdown if

I dont stop to worry

(7).
Excessive checking, reassurance, procrastination, make others take the decisions for him, search internet for information, double checking,

(8).
a. Distracting himself from thoughts b) Never tried not to engage with trigger thoughts with more thoughts (e.g. arguing with himself if the worry is "true")

(3).
Anxiety, muscle tension, concentration problems, fatigue sleeplessness, stomach distress, irritable, restlessness low mood

Figure 2. The metacognitive case formulation for William.

Socialization to the Model The socialization process starts with sharing the case conceptualization with the client. William would be expected to view the case formulation a good fit with his view of his problems, and likely had no additional information to add at this point. After the case formulation had been shared, some important facilitating questions were asked in order to develop his metacognitive understanding of the problem. How much of a problem would worrying be for you if you knew that you could control your worries?

William may have answered that if he could be certain that he could control his worries, they would not be a problem. If you were certain that you could not go crazy or have a nervous breakdown due to worrying, how big a problem would worrying be for you? William may have answered that if he was 100% sure that worrying would not give him a nervous breakdown, he would be reassured that his problem was not that dangerous. If there were only advantages to worrying, would worrying be a problem for you? Would you recommend worrying to your spouse and family?

MCT for GAD William said his major problem was that he thought that worrying could be dangerous and that he felt no control over his worries. William would be expected not to recommend worrying as a strategy to anyone. It is useful that the client realizes the contradiction between positive and negative meta-beliefs about worry. Questions such as the following are posed: Are you aware of the fact that you have both negative and positive beliefs about worry? What do you think are the effects of holding both types of beliefs? Questioning how it is possible to simultaneously have both positive and negative meta-beliefs about worrying can begin to build a new perspective for the client. To such a question, William likely answered that he had never considered the contradiction between having both positive and negative beliefs regarding worrying as a strategy. He would be expected to be a bit puzzled by this. The therapist can ask if William felt trapped in a circle of continuous worry and anxiety that was difficult to break out off. Holding both positive and negative beliefs about worry could contribute to keeping him trapped in a negative circle of continuous worry and anxiety. The coping strategies the patient uses when anxious are often maladaptive because they only temporarily reduce anxiety or they increase it. Asking questions that illustrate the effect of other coping strategies is therefore central. For example, How does excessive checking work for you? Do you become less worried? If so, does checking have a lasting or a temporary effect? When William took the time to reflect on this, he discovered that his checking behavior led to an increase in his anxiety levels, which was contradictory to his goal for using this behavior. Experiencing paradoxical effects when using different kinds of coping behaviors can be useful in exploring in the metacognitive model. In addition to his coping strategies, William was asked if he had ever tried to suppress thoughts: Do you ever try not to think particular thoughts? And how well has this worked for you? William had used suppression to deal with his worries. He likely stated that it worked occasionally, but when questioned whether this effect was short lived, he admitted that it had not helped him deal with his problems in the long run. The socialization would be expected to improve William's insight into the effects of his coping strategies and the role of beliefs about worry. It began to open up some new perspectives and alternative ways of thinking about his worries and his thinking about thinking. This created a foundation for proceeding with conducting an experiment to check the effect of using suppression as a way

307

to cope with his worries. Such an experiment would be introduced in the following manner: Let us explore how thought suppression would work. For the next minute, I would like you not to think of a white giraffe. Try your very best not to think of it, and suppress it as hard as you can. This is a task that most clients find difficult to achieve. If the client confirms that it was possible to suppress thoughts, the therapist asks if the client had to allocate a lot of effort and if the client found it strenuous to suppress this particular thought. This way the therapist can illustrate that suppression is a strategy that usually backfires by either increasing the frequency of a particular thought or demanding a lot of mental resources in order to block thoughts. Usually, as in William's case, he found that he could not suppress the thought, and this was used to show how his own strategy could contribute to the belief that his worry was uncontrollable. The therapist offered an alternative interpretation to William in order to explore this further: Perhaps your worry is not uncontrollable, but that you are using strategies that do not work? Would you be interested in exploring some other strategies? Have you ever tried not to engage with your worrying thoughts? Most clients have not tried to disengage from their worry process, and if they state that they have tried, it is important to explore how. For example, there are important differences between mindfulness techniques and the techniques of detached mindfulness used in MCT. Detached Mindfulness Early in therapy it is important that the client understands and starts to differentiate between trigger thoughts and worrying. This is a new way of relating to trigger thoughts, in which the person is helped to become an objective observer of their thoughts. The trigger thoughts typically occur spontaneously, which in most cases lead to worry and rumination as a coping response. One of the main goals of MCT is to increase the client's ability to let trigger thoughts come and go without engaging in them (this experience is called detached mindfulness; Wells, 2009; Wells & Matthews, 1994). Detached mindfulness is not related to meditation techniques or to focusing on the present moment. Detached mindfulness strictly refers to being aware of inner cognitive events and thoughts without responding to them. The term detachment refers to two factors: (a) the disengagement of any active response to the inner event and (b) experiencing the event as separate from the self. That is, the person is helped to become aware of the self as an objective observer of thoughts but separate from them. In order to develop detached mindfulness, it is useful to give the client experiences that approximate and achieve the state. Wells (2005b) described 10 different techniques that facilitate the development of detached mindfulness.

308

Hjemdal et al. they do not need to worry, or the worries that they had at some point during the day are no longer relevant. Using the worry period to worry is something that very few actually do, but postponing the worrying to this period is essential. With increased awareness of the ability to postpone worries, the client's erroneous beliefs that worry is uncontrollable are weakened. The rationale for worry postponement was delivered stressing that following detached mindfulness William should postpone his worries whenever they occurred. Just simply say to yourself, here is a worrying thought, I will not deal with this now, but return to it later if it is still relevant. The therapist emphasized that this was an experiment to test William's belief that worry was uncontrollable. Ratings in his uncontrollability belief were taken in the session and then again at the next session after this homework task. Most patients are surprised to find that they can postpone worrying. William may have found it difficult to postpone worries at first between sessions and started worrying about his inability to postpone them. He reported using postponement of worries 60% of the time when worries occurred. However, on exploration it became clear that he was still trying to suppress his worries and anxious thoughts. He was trying to get rid of them, because he found some of his thoughts were frightening and depressing. The therapist therefore returned to explore the concept of detached mindfulness using exercises in the session, which increased William's ability to simply observe thoughts without engaging in them. Several behavioral experiments of worry postponement can also be undertaken in the sessions, in order to give William the experience of this technique. William was asked to bring on a trigger thought and activate the worrying process. The therapist thereafter asked William to postpone his worries for 10 minutes to explore what would happen. During the 10-minute period William and the therapist focused on a different theme in the therapy. At the end of this period the therapist returned to William's worries, checking if William had worried during the last 10 minutes. In this way William discovered that he had control and his worries faded without him trying actively to get rid of them. Although in-session experiments may be somewhat different from the everyday experiences of the worry process between sessions, the in-session experiences can make it easier for William to understand worry postponement between the treatment sessions. For homework he was asked to postpone his worry until 6:00 P.M. each day, and he was instructed that the worry time was not mandatory. Postponement of worry also facilitates changing from a primarily internal focus to an external focus. So rather than focusing on his thoughts and worries, William gradually changed his focus to what was going on around him. The external focus made it easier to take part in everyday life

William used the free association task, the clouds metaphor, and the passenger train analogy, with the cloud and the train metaphors introduced first in order to socialize William to the concept. In the clouds metaphor, thoughts are likened to clouds floating in the sky. It would be futile to try to push the clouds away and to try to control their movements in order to change the weather. Trigger thoughts should be treated in the same way, left alone to take care of themselves. In the train analogy, the client is asked, When you are waiting for a train, do you climb aboard every one that enters the station? Thoughts are like trains. Where would you end up if you climbed on board all of them? William was then asked to try the free association task (Wells, 2005a, 2005b). In this task, the therapist said aloud a series of neutral words: green, water, sun, walking, bicycle, holiday, chocolate. William's task was to let his mind wander freely without trying to control thoughts, and to just observe his mental experiences in a detached manner. With an increased ability to observe and not engage with thoughts, the free association task was gradually expanded by dropping in trigger words. Here, William was asked to treat trigger words in the same way as neutral words. In this way, William grasped the concept of detached mindfulness and how this was very different from his normal way of reacting to thoughts with worrying and planning. It is important that clients understand the difference between detached mindfulness and suppression, and on a regular basis in treatment the therapist asks the client to describe how he uses detached mindfulness to ensure it is not being confused with thought suppression. Initially, many clients have difficulty distinguishing the two. Challenging Uncontrollability: The Worry Postponement Experiment The experiment of not thinking of a white giraffe makes the patient aware that thought suppression is an inefficient strategy for controlling thoughts. Detached mindfulness is a strategy to disengage from trigger thoughts. However, when the chain of thoughts (worry or rumination) has started, William found it difficult to control this chain of thoughts, reporting that when he started worrying about health issues, he could go on for hours, with suppressing thoughts or distraction not working very well. In order to deal with his worrying and begin to challenge uncontrollability beliefs about worrying, the therapist introduced the worry postponement experiment. As part of the worry postponement experiment a specific worry period was established. If William felt that he needed to engage in worrying or rumination, he would do so between 6:00 and 6:15 P.M. each day. It may be beneficial to restrict worrying to a specific place, but MCT does not regard it as essential. Most GAD patients find that when the pre-set worry period arrives,

MCT for GAD and concentrate on the tasks at hand. Within a few sessions he would be expected to have increased the use of detached mindfulness and worry postponement to 80% of his trigger thoughts and the occurrence of worrying. Furthermore, his belief in the uncontrollability of worry had also decreased. Uncontrollability: Verbal Reattribution We have already described how the use of detached mindfulness and worry postponement is a valuable way to give patients new metacognitive experiences and to challenge their beliefs about the uncontrollability of worry. The goal in MCT is to reduce uncontrollability beliefs to 0% or as close to this as possible. William perceived his worries as ongoing and always present. He stated at the beginning of therapy (when doing the case conceptualization) that his worries were highly uncontrollable (75%). In order to further challenge William's belief about uncontrollability the therapist engaged him in a meta-level Socratic dialogue: What happens to your worries if the doorbell rings, or one of the children urgently needs medical attention or the phone rings? and explored if this was evidence that he could choose to worry and therefore it was under his control. This line of questioning continued with asking if there were times he didn't worry in response to thoughts, and the therapist helped him to see that he had experienced this in treatment already with the worry-postponement experiment. Uncontrollability: The Loss of Control Experiment Further work was needed in order to reduce William's belief that worries were uncontrollable, and a new behavioral experiment called losing control was undertaken. In this experiment, William was asked to specify what losing control would mean for him. He had specific expectations that he would not be able to go to work or take care of the family. The therapist asked William how losing control of his worrying could be observable in the sessions for both of them. William might have expected that he would start talking gibberish. He was therefore encouraged by the therapist to try to lose control of his worry in the session. He was asked to activate a trigger, worry about it as usual, and try his very best to lose control. William was not successful in losing control, and he was quite surprised that one of his big fears did not happen as a consequence of letting worry run freely. His belief in worry being uncontrollable would likely be reduced to 5%, after this experiment. Then the losing control experiment was assigned as homework. The therapist and William planned how to run the experiment at home. The therapist went through William's previous experiences, which indicated that he had not lost control at any point in his life. They collectively summed up the results of the in-session experiment, and also repeated what losing control would look like. William was anxious about doing the experiment, but decided that it would be a good test, and wanted to do it

309

the very same day in the afternoon. In the next sessions he reported that he experienced increased control, and his anxiety was markedly reduced, which was also indicated as a decrease on the scores on the BAI and GAD-S. Working With Negative Metacognitions Related to Danger A significant proportion of GAD patients worry about the danger of worrying. In William's case he worried that he would go crazy from the experience of intense anxiety brought on by worrying. By questioning the evidence of the relation between worry and becoming crazy, William was asked how long he had been worrying and if he had lost his mind at any point during this time. He confirmed that he had never lost his mind, and he also stated that he couldn't think of anyone who had lost their mind as a result to worrying. To test his belief further, a behavioral experiment was set up in therapy. In this experiment William was asked to try to worry in the session to such an extent that he would lose his mind. As in the behavioral experiment described earlier, the therapist and William specified how it would be possible to observe that he was crazy. He might have worried that he would lose his mind, start running around, crying uncontrollably, run out of the therapy room and finally would be admitted to a mental hospital. William agreed to test if worrying could make him go crazy in the session, and to make this even worse he had to worry out loud. His level of anxiety prior to the experiment was relatively high, and he might report that it felt uncomfortable for him to test this out, but he did not go crazy. Through this experiment William could discover that although worrying felt uncomfortable, it was not dangerous for his mental health. His belief of becoming crazy fell to zero. The going crazy experiment was given as homework, with much the same rationale and preparation as the losing control experiment. A reduction of this metacognition further reduced his anxiety levels. Working With Positive Metacognitions Related to Worry When the negative meta-beliefs were reduced to 0% 5%, the treatment turned to challenging William's positive metacognitions. William had expectations that worrying would help him solve problems in his life, that it would help him cope better, and that worrying would help him make the right decisions in how to live. These positive beliefs were challenged using such questions as, What is your evidence that worrying is helpful? Has the worrying reduced your anxiety? Even though William's answer to these questions was likely no, he would be expected to still believe that worry could be helpful in some way. William was therefore challenged to take part in a worry-mismatch experiment (Wells, 2007). The retrospective version of this procedure was tested first. William was asked to remember a recent situation where he had worried about the outcome of

310

Hjemdal et al. strategy for a week, and whenever he got the urge to search the Internet for information, he would postpone it (similar to his postponement of worries earlier in therapy). Having tested this out for a week, William would be expected to return to the session feeling calmer and with less anxiety regarding his health-related worries, and would agree to drop his checking behavior. Toward the end of the treatment the therapist asked William if seeking reassurance from others actually reduced his levels of anxiety in the long run, and if it permanently reduced his anxiety levels. Reassurance seeking is a complex social behavior that is under volitional control. Patients may be reluctant to stop this strategy if they continue to believe it is helpful. By reviewing the usefulness of this strategy with the patient, and illustrating that it has not led to a long-term reduction in worry, the metacognitive therapist increases the motivation to try to change this strategy and try new strategies and behaviors. Furthermore, the therapist helps the patient to discover that it is a source of further processing and therefore another example of extended thinking. William experienced temporary reductions of his anxiety when he received reassurance, yet the anxiety always returned. As an experiment he agreed to drop seeking reassurance for a week. He came back the following week, and had discovered that he had experienced little anxiety and felt that he could control his worry on his own. Based on this, he would be expected to conclude that reassurance seeking was a strategy that increased his anxiety, which was contrary to his initial belief, and therefore decided to drop this strategy in the future. Relapse Prevention In this last phase of treatment the therapist worked with William to construct a therapy blueprint, examined his worries about relapse, and worked on residual negative and positive metacognitive beliefs endorsed on the GAD scale (GAD-S). The blueprint consisted of a summary of William's case formulation and strategies he had found useful in learning to relate to negative thoughts in a new way. It also made use of the old plan/new plan worksheet (Wells, 2009), in which his old response to a negative thought was summarized and a new response written in detail (see Table 1 for a more thorough description). He was asked to refer to this as a summary reminder of the new strategies he should implement in the future. In addition, there was a small residual positive belief that it was necessary to plan ahead in order to avoid worrying. The therapist identified this and used a Socratic approach to help William reinforce the idea that he had alternative and more direct means of reducing worry that did not require overthinking. A booster session can be planned 3 months after the last treatment session in order to check on the client's progress.

the situation. He chose a recent visit to the doctor, where he had worried quite a bit about having a serious illness before the visit. A detailed description of the content of the worries involved in the trip to the doctor was written down on the left side of a sheet of paper. On the right side of the paper William had to write down in detail what actually happened at the doctor's office. There is typically a huge discrepancy between the descriptions on the left and the right. William was asked by the therapist how useful worrying seemed to him if it is was not closely matched to reality. In order to work on the positive meta-beliefs even further, he was asked to do another test before the next session. He was asked to conduct a worry-modulation experiment (Wells, 2009) consisting of increasing his worry for one day and then minimizing it on the second day (using detached mindfulness and worry postponement), in order to compare which of these two days were the most pleasant and productive for him. At first the idea of increasing his worries would likely be a bit puzzling to William. But the therapist reminded him that worrying was what William had been doing for many years, and that by increasing his worry for a day, the therapist simply meant that William was to return to his old habit of worrying about most things, and maybe increase the proportion of worry a bit compared to what he used to do. William would be expected to return to the next session reporting that the day he worried less was the better of the two. He did not find it difficult to increase worry because he had extensive experience with worrying; however, he found it curious to try to worry more as the therapy focuses on reducing the amount of worrying. The therapist can then question if there are any advantages to worrying. William likely concluded that this experience illustrated that there were no positive consequences of worrying, and after that he would be expected to no longer have any remaining positive beliefs that worry had any advantages or could be useful for him. Working With Avoidance Strategies and Unhelpful Coping Strategies William used several unhelpful strategies to cope with his worry. Among these strategies (see description in the case formulation) were excessive checking, reassurance, procrastination, searching the Internet for information, arguing with himself if the worry was "true," and trying to distract himself from worry thoughts. One by one these unhelpful coping strategies were challenged and questioned in therapy. For example, when searching the Internet for information on health issues, did this reduce or increase his focus of attention on possible threats? After thinking about this for a while William became certain that searching the Internet could make him more anxious, and that this strategy just kept the worry process going. He therefore agreed that he would try to drop this unhelpful coping

MCT for GAD

Table 1

311

Therapy Blueprint for William My triggers: What if I will get a serious illness? What if something happens to my family? What if I do something wrong?
Old plan: New plan:

Thinking style:
If I have a negative thought, then worry about it to find out what could happen and how to avoid it If I have a negative thought, cover all the negative possibilities so I am prepared for the catastrophe

Thinking style:
If I have a negative thought, just leave it alone and wait to see what happens If I have a negative thought, just let it fade away

Behavior:
When I am worried search for evidence that is supporting or counteracting my worrying When I am worried, ask others to reassure me

Behavior:
If I am worried, don't search for evidence, just stop the thought process (but do not suppress the thought) If I am worrying, then ban asking for reassurance and just let the worry ebb away

Attention focus:
Look for danger, so I am prepared Focus on my thoughts and body to find out what is going on

Attention focus:
Ban threat monitoring, it just makes me worry Focus on the world outside me and continue what I'm doing to find out what I can achieve

Reframe:
My trigger thoughts are best left alone. If so, they will just fade away. I control my worries, they do not control me. Do more novel things, break my old routine without thinking so much.

Outcome Studies of MCT in GAD Different treatment trials have reported promising results for MCT. The first was an open trial in which Wells and King (2006) reported an 87.5% recovery rate at posttreatment, as well as a 75% rate at 6 and 12 months follow-up using formal criteria on the STAI. In a randomized trial the effects of MCT were compared against the effects of applied relaxation. In this study there was an 80% standardized recovery rate for MCT on a measure of pathological worry (PSWQ) compared to 10% for applied relaxation at posttreatment (Wells et al., 2010). The gains in MCT were maintained at 12-month follow-up. However, a limitation of these earlier studies is their small sample size. A recent large independent randomized controlled therapy trial found that both MCT and a variant of intolerance of uncertainty therapy (IU) were effective (van der Heiden et al., 2012). The within-group posttreatment effect sizes for MCT were 2.39 (PSWQ) and 2.01 (STAI) compared

with 1.43 (PSWQ) and 1.42 (STAI) in IU treatment. At 6-month follow-up the effects were 2.38 (PSWQ) and 2.0 (STAI) following MCT compared with 1.6 (PSWQ) and 1.56 (STAI) in the IU condition. The authors reported a superiority of MCT over IU treatment on several of the measures. Fisher (2006) has computed recovery rates measured by PSWQ scores across different treatment trials for GAD at posttreatment and 1-year follow-up. In this analysis the rates for MCT were 80%, and 70% respectively, compared with 46%, and 53% respectively for CBT, 48% and 64% for IU and 37% and 38% for applied relaxation. The results appear to show that MCT is an effective treatment. Consistent with these data MCT and IU are treatments now recommended in the updated NICE guidelines for (NHS Evidence, 2012).

Conclusions
In this paper we have described Wells's metacognitive model and treatment of GAD, illustrated in the case of

312

Hjemdal et al.
Fisher, P. L. (2006). The efficacy of psychological treatments for generalized anxiety disorder? In G. C. L. Davey & A. Wells (Eds.), Worry and its psychological disorders: Theory, assessment and treatment (pp. 359378) Chichester: Wiley. Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of cognitive-developmental inquiry. American Psychologist, 34(10), 906911. Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Development and validation of the Penn State worry questionnaire. Behaviour Research and Therapy, 28, 487495. Moffitt, T. E., Harrington, H., Caspi, A., Kim-Cohen, J., Goldberg, D., Gregory, A. M., & Poulton, R. (2007). Depression and generalized anxiety disorder: Cumulative and sequential comorbidity in a birth cohort followed prospectively to age 32 years. Archives of General Psychiatry, 64(6), 651660. Nassif, Y. (1999). Predictors of pathological worry. Unpublished M.Phil. thesis. University of Manchester, UK. Nuevo, R., Montorio, I., & Borkovec, T. D. (2004). A test of the role of metaworry in the prediction of worry severity in an elderly sample. Journal of Behavior Therapy and Experimental Psychiatry, 35(3), 209218. Robichaud, M. (2013). Cognitive Behavior Therapy Targeting Intolerance of Uncertainty: Application to a Clinical Case of Generalized Anxiety Disorder.Cognitive and Behavioral Practice, 20, 251263 (in this issue). Ruscio, A. M., & Borkovec, T. D. (2004). Experience and appraisal of worry among high worriers with and without generalized anxiety disorder. Behaviour Research and Therapy , 42 , 14691482. van der Heiden, C., Muris, P., & van der Molen, H. T. (2012). Randomized controlled trial on the effectiveness of metacognitive therapy and intolerance-of-uncertainty therapy for generalized anxiety disorder. Behaviour Research and Therapy, 50(2), 100109. Wells, A. (1995). Metacognition and worry: A cognitive model of generalized anxiety disorder. Behavioural and Cognitive Psychotherapy, 23, 301320. Wells, A. (1997). Cognitive therapy of anxiety disorders: A practice manual and conceptual guide. Chichester, UK: Wiley. Wells, A. (2000). Emotional disorders and metacognition: Innovative cognitive therapy. Chichester, UK: Wiley. Wells, A. (2005a). The Metacognitive Model of GAD: Assessment of meta-worry and relationship with DSM-IV Generalized Anxiety Disorder. Cognitive Therapy and Research, 29, 107121. Wells, A. (2005b). Detached mindfulness in cognitive therapy: A metacognitive analysis and ten techniques. Journal of Rational Emotive & Cognitive Behavior Therapy, 23(4), 337355. Wells, A. (2006). The metacognitive model of worry and generalized anxiety disorder. In G. C. L. Dawey & A. Wells (Eds.), Worry and its psychological disorders. Theory, assessment and treatment (pp. 179199). Chichester, UK: Wiley. Wells, A. (2009). Metacognitive therapy for anxiety and depression. New York: Guilford Press. Wells, A., & Carter, C. (1999). Preliminary tests of a cognitive model of Generalized Anxiety Disorder. Behaviour Research and Therapy, 37, 585594. Wells, A., & Carter, K. (2001). Further tests of a cognitive model of Generalized Anxiety Disorder: Metacognitions and worry in GAD, panic disorder, social phobia, depression, and non-patients. Behavior Therapy, 32, 85102. Wells, A., & Cartwright-Hatton, S. (2004). A short form of the metacognitions questionnaire: Properties of the MCQ-30. Behaviour Research and Therapy, 42(4), 385396. Wells, A., & Davies, M. (1994). The Thought Control Questionnaire: A measure of individual differences in the control of unwanted thought. Behaviour Research and Therapy , 32 , 871878. Wells, A., & King, P. (2006). Metacognitive therapy for Generalized Anxiety Disorder: An open trial. Journal of Behavior Therapy and Experimental Psychiatry, 37, 206212. Wells, A., & Matthews, G. (1994). Attention and emotion: A clinical perspective. Hove, UK: Erlbaum.

William (Robichaud, 2013-this issue). The model is based on the principle that worry is not effectively regulated in GAD, because of the effects of metacognition on mental control and extended thinking. In particular, metacognitive beliefs have a crucial role that intensifies the aversive experience of worry and its threat value. GAD patients use paradoxical or incompatible metacognitive control strategies that reduce their exposure to experiences of self-control and/or contribute to instances of impaired control. The patient does not have a deficit or absence of control; rather, beliefs about control are unhelpful and strategies used to cope with worries and negative thoughts are counterproductive. It should be noted that MCT is not simply a matter of gaining better control, because the model emphasizes the importance of using experiences in therapy to challenge beliefs about the uncontrollability of worry and beliefs that worry is dangerous. The metacognitive therapist is cautious not to convey the idea that worry must be controlled; instead, the aim is to show that thoughts are insignificant for further processing, and the individual has a choice about how to respond to his thinking. This approach contrasts with other approaches that focus on reality testing the content of worry, involve problem solving of concerns, or distinguish between types of worry that should be analyzed and responded to or not. In each of these other approaches the therapist may in some cases be seen to be dealing with the problem of excessive thinking with more thinking. MCT suggests instead that treatment should focus on reacting to negative ideas by reducing the reliance on worrying and thinking and by ultimately doing little or nothing in response to negative thoughts. This contrasts with meditation or relaxation strategies, which countenance responding to thoughts with changes in attention, breathing exercises, and with CBT, which focuses on worry exposure or challenging schemas concerning uncertainty.

References
Beck, A. T., & Steer, R. A. (1993a). Beck Anxiety Inventory manual. San Antonio, TX: Psychological Corporation. Beck, A. T., & Steer, R. A. (1993b). Manual for the Beck Depression Inventory. San Antonio, TX: Psychological Corporation. Borkovec, T. D., Robinson, E., Pruzinsky, T., & DePree, J. A. (1983). Preliminary exploration of worry: Some characteristics and processes. Behaviour Research and Therapy, 21, 916. Butler, G., Wells, A., & Dewick, H. (1995). Differential effects of worry and imagery after exposure to a stressful stimulus: A pilot study. Behavioural and Cognitive Psychotherapy, 23, 4556. Cartwright-Hatton, S., & Wells, A. (1997). Beliefs about worry and intrusions: The Meta-Cognitions Questionnaire and its correlates. Journal of Anxiety Disorders, 11, 279296. Davey, G. C. L., Tallis, F., & Capuzzo, N. (1996). Beliefs about the consequences of worrying. Cognitive Therapy and Research, 20, 499520. Davis, R. N., & Valentiner, D. P. (2000). Does meta-cognitive theory enhance our understanding of pathological worry and anxiety? Personality and Individual Differences, 29, 513526.

MCT for GAD

Wells, A., & Matthews, G. (1996). Modelling cognition in emotional disorder: The S-REF model. Behaviour Research and Therapy, 32, 867870. Wells, A., & Papageorgiou, C. (1995). Worry and the incubation of intrusive images following stress. Behaviour Research and Therapy, 33, 579583. Wells, A., & Papageorgiou, C. (1998). Relationship between worry, obsessive-compulsive symptoms and meta-cognitive beliefs. Behaviour Research and Therapy, 36(9), 899913. Wells, A., Welford, M., King, P., Papageorgiou, C., Wisely, J., & Mendel, E. (2010). A pilot randomized trial of metacognitive therapy vs. applied relaxation in the treatment of adults with generalized anxiety disorder. Behaviour Research and Therapy, 48, 429434.

313

York, D., Borkovec, T. D., Vasey, M., & Stern, R. (1987). Effects of worry and somatic anxiety on thoughts, emotion and physiological activity. Behaviour Research and Therapy, 25(6), 524526. Address correspondence to Odin Hjemdal, Department of Psychology, Norwegian University of Science and Technology, N-7491 Trondheim, Norway; e-mail: odin.hjemdal@svt.ntnu.no.

Received: February 19, 2012 Accepted: January 6, 2013 Available online 19 February 2013