Dysfunctional beliefs in the. & treatment of obsessivecompulsive

Transcription

1 Dysfunctional beliefs in the understanding & treatment of obsessivecompulsive disorder Annemiek Polman 2010

2 Publication of this dissertation was supported by the Parnassia Bavo Academie and the University Medical Centre Groningen. Cover & design: Studio Stedum Printed by: Ipskamp Drukkers B.V. A. Polman, 2010 ISBN: Den Haag, Parnassia Bavo Groep, 2010

3 Rijksuniversiteit Groningen Dysfunctional beliefs in the understanding and treatment of obsessive-compulsive disorder Proefschrift ter verkrijging van het doctoraat in de Medische Wetenschappen aan de Rijksuniversiteit Groningen op gezag van de Rector Magnificus, dr. F. Zwarts, in het openbaar te verdedigen op woensdag 30 juni 2010 om uur door Annemieke Polman geboren op 13 mei 1975 te Groningen

4 Promotores: Copromotor: Prof. dr. J.A. den Boer Prof. dr. P.J. de Jong Dr. T.K. Bouman Beoordelingscommissie: Prof. dr. P.M.G. Emmelkamp Prof. dr. E. de Haan Prof. dr. R.B. Minderaa

5 Contents 1 Introduction 7 2 Processes 3 Dysfunctional 4 Obsessive 5 Dysfunctional 6 Obsessive-compulsive 7 general of change in cognitive-behavioural treatment of obsessivecompulsive disorder; current status and some future directions 19 beliefs in the process of change of cognitive treatment in obsessive-compulsive checkers 35 beliefs and their relationship to obsessive-compulsive symptom dimensions 61 belief-based subgroups and inferential confusion in obsessive-compulsive disorder 75 behaviours and beliefs; a family affair? 89 discussion 101 References 117 Appendix 131 Samenvatting 135 Dankwoord 145 Curriculum Vitae 149

6

7 Introduction

8

9 INTRODUCTION 9 Current psychological theories stress the importance of dysfunctional beliefs in the development and persistence of obsessive-compulsive symptoms (e.g., washing, checking), and specific belief domains have been identified that are hypothesized to be especially relevant to obsessive-compulsive disorder (OCD). To further the understanding of OCD within a cognitive framework, this thesis examined dysfunctional beliefs as potential mechanism of OCD from different angles. Obsessive-compulsive disorder Currently, OCD is categorized as an anxiety disorder according to the DSM-IV- TR (American Psychiatric Association, 2000), and is characterized by recurrent thoughts, images or impulses (obsessions) and or repeatedly exercised actions or mental acts (compulsions). Since obsessions are experienced as intrusive and inappropriate, and cause marked distress, patients often attempt to ignore or suppress their obsessions. Furthermore, most patients perform compulsions to neutralize anxiety and/or to prevent threatening situations from happening. These behaviours are clearly excessive and/or not realistically connected to the feared situation, e.g., washing hands for 20 minutes, or dressing in a particular order because that would prevent bad things from happening. The most common obsessions involve thoughts about contamination, repeated doubts, a need to have things in a particular order, aggressive or horrific impulses and sexual imagery. Frequently occurring compulsions are washing and cleaning, checking, counting, requesting reassurance, repeating actions, and ordering (American Psychiatric Association, 2000). The lifetime prevalence of OCD ranges from 1.9 to 2.5%, with a one-year prevalence of 1.1 to 1.8%, which was measured across multiple sites (United States, Canada, Puerto Rico, Germany, Korea & New Zealand) (Weissman, Bland, Canino, Greenwald, Hwu, Lee, et al., 1994). Furthermore, OCD seems to be slightly more common in females than males, and has a substantial comorbidity with major depressive disorder. The mean age at onset across sites was 21.9 to 35.5 years (Weissman et al., 1994). A recent study on clinical features and symptoms at intake, reported that 81% of the patients suffering from OCD mentioned a gradual onset, whereas 17% reported a sudden onset (i.e., fully symptomatic within one month). The overall course of symptoms was rated as continuous with mild variations in intensity of symptoms by 67%, and as waxing and waning by 23%. Two percent rated their OCD as deteriorative; symptoms continued to worsen even during treatment (Pinto, Mancebo, Eisen, Pagano & Rasmussen, 2006). Considering the impact and chronic nature of this impairing disorder, and since efficacy rates of OCD treatment are positive but stagnant (about 50% of treated patients are symptomatic at long-term follow-up) (Van Oppen, Van Balkom, De Haan & Van Dyck, 2005; Whittal, Robichaud, Thordarson & McLean, 2008) it is important to understand the mechanisms involved in the generation and maintenance of OCD symptoms, in order to ultimately improve treatment possibilities.

10 10 CHAPTER 1 Cognitive-behavioural models of OCD Cognitive-behavioural approaches are most prominent in current theoretical models of OCD. These models are predominantly based on Beck s schema theory (1976), which postulates dysfunctional schemata to play an important role in the development and maintenance of emotional disorders. Personal knowledge is considered to be organized within schemata (or associative networks), which influence a person s perceptions, interpretations, and memories. In other words, experiences are perceived in relation to a person s schemata. These schemata are postulated to be reflected in a person s automatic thoughts. An important step towards a cognitive-behavioural conceptualization of OCD was the positioning of obsessional thoughts in regard to negative automatic thoughts (Salkovskis, 1985). Obsessions are distinct from negative automatic thoughts with respect to their perceived intrusiveness, immediate accessibility to consciousness, and the extent to which they are seen as being consistent with the individual s belief systems (Salkovskis, 1985). Whereas automatic thoughts are considered to be ego-syntonic, highly plausible, less accessible, and to involve idiosyncratic themes, intrusive thoughts are ego-dystonic, easily accessible, and irrational. Hence, an intrusive thought is defined as an internal stimulus rather than the basis of the discomfort itself. Intrusions also occur in normal people without disturbing mood or coping (Rachman & De Silva, 1977), and are not anxiety provoking by themselves. According to the cognitive-behavioural model, the appraisal of the intrusion combined with the beliefs that influence the appraisal process are crucial in the development of OCD. It is postulated that when the appraisal of an intrusion is guided by anxiety related schemata, this would lead to negative automatic thoughts, which would influence emotions (i.e., anxiety, distress). Subsequently, these emotions would affect behaviour. It is hypothesized that compulsions are performed to reduce anxiety, and in order to prevent future discomfort people would engage in avoidance behaviour. Usually, neutralizing initially leads to a reduction of discomfort, which reinforces the use of compulsive behaviour. However, such strategies may fail and can even be counter-productive, in that they may trigger the thoughts they are supposed to prevent. Furthermore, neutralizing may increase the salience of an intrusion, leading to increased frequency and discomfort of the intrusive thought (Salkovskis,1985), and shielding the beliefs from disconfirmatory evidence (Rachman, 1998). In his cognitive-behavioural analysis, Salkovskis (1985) stressed the importance of beliefs related to responsibility for possible harm to self or others, which are hypothesized to contribute to the perceived salience of intrusions. In addition, beliefs related to the need to control one s thoughts and the perceived failure in controlling intrusive thoughts (Clark & Purdon, 1993), misinterpretations of the importance or significance of intrusive thoughts (Rachman, 1997, 1998), and meta-cognitions in general (Wells, 2000) have been highlighted in further developments of a cognitive-behavioural model of OCD. An international group of researchers identified six OCD relevant belief domains; over-importance of

11 INTRODUCTION 11 thoughts, importance of controlling one s thoughts, perfectionism, inflated responsibility, overestimation of threat, and intolerance of uncertainty (see Table 1) (Obsessive Compulsive Cognitions Working Group (OCCWG), 1997). Development of the Obsessive Beliefs Questionnaire (OBQ), which indexes the relative importance of each of these domains, enabled more systematic investigation of dysfunctional beliefs and their relationship to OCD (OCCWG, 2001, 2003, 2005). Table 1. Belief domains. Belief domain Definition Example Over-importance of thoughts Importance of controlling thoughts Perfectionism Inflated responsibility Overestimation of threat Intolerance of uncertainty Belief that the occurrence of a thought, image, or impulse implies something very important Overvaluation of the importance of exerting complete control over intrusive thoughts, images, and impulses, and the belief that this is both possible and desirable. Belief that there is a perfect solution to every problem; (2) that doing something perfectly is possible and necessary; and (3) that even minor mistakes have serious consequences. Belief that one is especially powerful in producing and preventing personally important negative outcomes. These outcomes are perceived as essential to prevent. Beliefs indicating an exaggerated estimation of the probability or severity of harm. This domain encompasses three types of beliefs: (1) beliefs about the necessity for being certain; (2) beliefs that one has a poor capacity to cope with unpredictable change; and (3) beliefs about the difficulty of adequate functioning in inherently ambiguous situations. Having violent thoughts means I will lose control and become violent. I would be a better person if I gained control over my thoughts. It is important to keep working at something until it is done just right. If I don t act when I foresee danger, I am to blame for any bad consequences. I believe the world is a dangerous place. I cannot tolerate uncertainty.

12 12 CHAPTER 1 Psychological Treatment Therapeutic implications of the cognitive-behavioural model of OCD would be to modify automatic thoughts consequent to the intrusions, and the beliefs that give rise to them (and not the intrusions themselves). The psychological treatment most frequently applied in OCD is exposure in vivo with response prevention (ERP). The goal of ERP is to reduce anxiety and avoidance behaviour by confronting patients with feared or avoided situations or objects in real life, while preventing fear- reducing behaviour (Van Hout & Emmelkamp, 2002). Originally, the main mechanism of ERP was considered to be habituation to the feared stimulus. However, currently a more cognitive rationale is used as well, since ERP provides an opportunity to challenge automatic thoughts and dysfunctional beliefs. ERP is considered to be an efficacious treatment for OCD (Chambless & Ollendick, 2001). Efficacy studies showed significant decreases in obsessive and compulsive behaviour, and results seemed to be maintained at follow up (Marks, 1997). Within clinical practice (outside research context), ERP is found to be effective as well (Franklin, Abramowitz, Kozak, Levitt & Foa, 2000). A meta-analysis, comparing 16 outcome trials with 9 normative samples, showed that at follow up, 50% of the patients showed reliable change, indicating that they were functioning at a level more similar to the general population than to the OCD population (Abramowitz, 1998). However, ERP does not cure OCD; treated patients did remain more symptomatic than members of the general population. In spite of ERP s efficacy in OCD, several arguments were put forward for the introduction of cognitive elements in the treatment of OCD (Clark, 1999). Firstly, apparently 20-30% of OCD patients refused to start ERP because they think it is too frightening. Secondly, long-term results of ERP showed that results are preserved, but residual symptoms remain. Thirdly, considering the prominent role of dysfunctional beliefs in OCD models, it would make sense to explicitly address these beliefs in treatment as well (Clark, 1999). Various treatments are called cognitivebehavioural therapy (CBT) or cognitive therapy (CT). In general, the purpose of CBT is to teach patients to identify and challenge distressing thoughts and exchange them for more rational or functional ones (Van Oppen & Arntz, 1994). The challenging of beliefs also includes behaviour experiments in which hypotheses about behaviour are tested. The study of Van Oppen, De Haan, Van Balkom, Spinhoven, Hoogduin and Van Dyck (1995) was the first randomized controlled trial comparing CT along the lines of Beck to ERP, and results showed that both treatments were effective for OCD. Multivariate results indicated CT to be superior to ERP, but this was not confirmed by the univariate tests. However, significantly more patients were rated recovered in the CT condition than in the ERP condition, as measured by both a self-report questionnaire and a structured clinical interview of OC symptoms. In another randomized controlled trial, CT and ERP were found to be equally effective (Cottraux, Note, Yao, Lafont, Note, Mollard et al., 2001). Furthermore, some

13 INTRODUCTION 13 evidence was provided for the effectiveness of CBT in patients who only report obsessions without having compulsions, a group that is known to be difficult to treat with ERP (Freeston, Ladouceur, Gagnon, Thibodeau, Rheaume, Letarte et al., 1997). Moreover, CT was found to be effective in private practice as well (Warren & Thomas, 2001). Taken together, these results suggest that CT which explicitly targets dysfunctional beliefs seems to be at least equally effective as ERP. Moreover, CT seems promising for subgroups that are difficult to treat with ERP. Finally, CT can be an effective alternative for patients who refuse to start with ERP. The question remains whether the process and mechanism of change in both treatments is in line with predictions based on the cognitive-behavioural model, i.e., how does change unfold and do changes in appraisals and beliefs play a crucial role in this process? Dysfunctional beliefs in the understanding and treatment of OCD Generally, two main research lines can be distinguished in the investigation of the cognitive paradigm (Kindt & Arntz, 1999; McNally, 2001). One tradition is closely linked to clinical practice and was inspired by Beck s ideas (i.e., dysfunctional schemata are reflected in someone s automatic thoughts, and people can become aware of these thoughts). The other tradition stems from cognitive psychology, and investigates the functioning of cognitive processes such as attention, memory and interpretation. To enhance understanding of psychopathology, these processes are compared between people suffering from psychopathology and healthy controls. Both traditions base their work on the assumption that dysfunctional schemata are the bases for the development and maintenance of psychopathology. However, the clinical tradition assumes that people can become aware of their dysfunctional beliefs and reasoning errors, and rely on self-report instruments, whereas experimental research into cognitive processes focuses on performance on experimental tasks (Kindt & Arntz, 1999). In agreement with current research in the field of dysfunctional beliefs (e.g., Obsessive Compulsive Cognitions Working Group) and process studies in OCD, the studies presented in this thesis follow the clinical tradition and are based on self-report measures. Dysfunctional beliefs hypothesized to be relevant to OCD will be studied in different contexts. The first two studies will focus on processes of change during treatment with specific attention for the involvement of dysfunctional beliefs in this process (chapters 2 & 3). Subsequently, dysfunctional beliefs will be studied related to OCD subtypes (chapters 4 & 5). Finally, intergenerational aspects of obsessive-compulsive beliefs and -behaviours will be investigated within families of patients with OCD (chapter 6). By studying dysfunctional beliefs from different angles, this thesis attempts to explore the contribution of this construct in a cognitive (behavioural) explanation of the mechanisms underlying OCD.

14 14 CHAPTER 1 Cognitive change; a process marker in CBT for OCD? To determine whether therapy can be considered a mature application of science, Marks (2002) postulated four criteria; 1. treatment efficacy; 2. identification of responsible treatment components; 3. knowledge of the mechanisms of action; and 4. elucidation of why treatments are only effective in some sufferers. Most clinical research into anxiety and depression is related to criterion one, and in his review, Marks concluded that psychotherapies have the ability to improve these disorders reliably and enduringly. However, still more needs to be known about which crucial variables produce improvement, how they do so, and why certain interventions are not effective for all sufferers (Marks, 2002). Based on the cognitive-behavioural model of OCD, one would expect changes in negative automatic thoughts and dysfunctional beliefs to be crucial in the process of change. Some evidence has been emerging for the relationship between successful treatment and significant changes in dysfunctional beliefs. Two studies involving exposure with response prevention showed that improvement in OC symptomatology was related to a decrease in credibility of dysfunctional beliefs (Emmelkamp, Van Oppen & Van Balkom, 2002). Moreover, dysfunctional beliefs hardly changed over the course of treatment in the least improved patients. Cognitive behaviour therapy seems to be associated with cognitive change as well: Various research groups demonstrated that reductions in OC symptoms correlated with changes in OC beliefs (for a review, see Bouvard, 2002). However, since these studies focused on the comparison of pre- and post-treatment measures, it cannot be corroborated whether changes in interpretations and long-standing beliefs precede, follow, or co-vary with changes in symptomatic outcome (cf. Emmelkamp et al., 2002). Process research, which focuses on patterns of change of overt and covert client behaviour over the course of treatment, could shed more light on how treatment works, and hence could ultimately provide a more critical test of the cognitive-behavioural model of OCD (i.e., determine whether decreases in dysfunctional beliefs mediate clinical outcome) (Tolin, Woods & Abramowitz, 2003). By generating information about patterns of change, process studies would not only provide a test for theoretical models, but can also have several important clinical implications. It has been argued that there is a gap between researchers and clinicians. Outcome studies provide information on treatment efficacy, but do not provide practical information for a therapist, such as how to apply treatment and when to use certain techniques (Pachankis & Goldfried, 2007). Process studies, on the other hand, focus on individual trajectories over the course of treatment, thereby capturing the clinical reality as experienced by therapists much closer, which can have important theoretical and clinical implications. For instance, the recovery trajectories of depressed patients have shown to be characterized by discontinuities, like sudden gains and sudden worsening, both which were associated with lower post-treatment depression scores (Hayes, Feldman, Beevers, Laurenceau, Cardaciotto & Lewis-Smith, 2007; Tang & DeRubeis, 1999). Sudden gains have been reported to be associated with hope (Hayes et al., 2007) and cogni-

15 INTRODUCTION 15 tive change (Tang & DeRubeis, 1999). Intuitively, one might expect sudden worsening to be related to higher post-treatment scores, but it was found that during this period, patients were processing a lot of new information, initially leading to a worsening, but eventually being part of an important process leading to improvement. This type of information about the pattern of change can have important clinical implications. For instance, a therapist does not have to worry about a sudden worsening since this might indicate a patient is going through a phase of cognitive-emotional processing. Furthermore, this type of information can have implications for planning or timing certain interventions. In case of depressed patients, it is important to consider whether a patient has the resources to go through such a destabilizing period. For instance, during significant negative life events, an intervention which aims at stabilization can be more useful than an intervention which leads to more destabilization (Hayes et al., 2007). Taken together, process research could provide a clearer picture of the processes of recovery and poor response, which could advance theory and can lead to further refinement of existing treatment procedures, and the development of new therapeutic techniques that more specifically activate the mechanisms by which clinical change occurs. Regarding OCD, several process studies were conducted, yielding diverse results. One aspect that hampers comparison between studies is the wide variety in methodological approaches. In contrast to treatment outcome studies, there are no straight-forward guidelines for how to conduct a process study. Chapter 2 provides an overview of process studies in OCD, and discusses methodological considerations involved in process research. Subsequently, in Chapter 3, a process study is presented within seven OCD patients predominantly suffering from checking compulsions, which aims to test predictions based on the cognitive-behavioural model. Moreover, the potential value of a new statistical technique called dynamic systems modelling is explored, which should enable close capturing of discontinuous trajectories. Dysfunctional beliefs and OCD subtypes A complicating factor in the study of OCD and the possible mechanisms involved is that OCD is a very heterogeneous disorder, and questions are raised as to whether certain subtypes should be distinguished. First and foremost, this heterogeneity becomes apparent in the specific manifestation of OC symptoms which can differ tremendously from patient to patient, e.g., washing and cleaning, ordering, sexual obsessions, and hoarding. Subtyping research is dominated by two methodologies; cluster-analysis and factor-analysis. Factor-analysis yields dimensions, whereas clustering provides exclusive categories. Several studies were conducted to investigate differential/independent OCD symptom dimensions or symptom clusters based on the Symptom Checklist of the Yale Brown Obsessive Compulsive Scale (Goodman, Price, Rasmussen, Mazure, Fleischmann, Hill et al, 1989), and in general,

16 16 CHAPTER 1 three recurring symptom subtypes/dimensions can be distinguished; washing, doubting-checking, and obsessional phenomena (McKay, Abramowitz, Calamari, Kyrios, Radomsky, Sookman, et al., 2004). Other criteria for subtyping are age at onset and comorbid disorders. Regarding age at onset, two groups can be distinguished; early age at onset <18 years, M=11.8, and late age at onset >18 years, M=26.6). The early age at onset subgroup reported to have at least one first-degree relative with a probable diagnosis of OCD more often (Pinto et al., 2006). With respect to comorbid disorders, a tic-related subtype has been identified which is associated with symptoms related to exactness and symmetry, and compulsions that show similarity with tics in Tourette s syndrome, like touching, tapping, and blinking. Patients in the non-tic related subgroup report more prominent obsessional worries about harm and responsibility. The function of compulsions seemed different in these two groups: Whereas the subgroup without tics seemed to engage in compulsions to reduce anxiety or distress, the tic related subgroup seemed to conduct their compulsions until it felt just right (Leckman, Grice, Barr, De Vries, Martin, Cohen, et al., 1995). It is conceivable that dysfunctional beliefs are less relevant to the tic related subtype. Investigation of associations between dysfunctional beliefs and OCD symptom subtypes in an OCD patient sample revealed significant relationships when controlling for depression: Responsibility and Threat Estimation was related to rumination scores, Perfectionism and Certainty was associated with checking and precision scores, and Importance and Control of Thoughts was associated with impulse scores (Julien, O Connor, Aardema & Todorov, 2006). Symptom dimensions in this study were based on the subscales of a self-report questionnaire. To further the investigation of OCD subtypes, an attempt was made to replicate this study (Chapter 4). In line with Julien et al. (2006) we used the same questionnaire subscales to define subtype dimensions. However, in order to asses a wider variety of OCD symptoms, we added a clinician rated symptom checklist as well. Since previous studies with the English version of the OBQ showed some inconsistencies regarding its structure, the factor structure of the Dutch version of the OBQ was examined before investigating specific associations between beliefs and subtypes. Finally, the relevance of dysfunctional beliefs could be a criterion for subtyping. Two studies showed that dysfunctional beliefs, as measured by the OBQ, were not relevant for substantial parts of the patient samples; respectively 56% and 49% of the participants had scores comparable to normal controls on dysfunctional beliefs (Calamari, Cohen, Rector, Szacun-Shimizu, Riemann & Norberg, 2006; Taylor, Abramowitz, McKay, Calamari, Sookman, Kyrios, et al., 2006). In addition to subgroups scoring high or low on the OBQ, subgroups were found that scored high on one of the OBQ subscales; Responsibility and Threat Estimation, Perfectionism and Certainty, and Importance and Control of Thoughts (Calamari et al., 2006). One explanation for these findings would be that different mechanisms might underlie different OCD subtypes. Furthermore, investigation of other cognitive

17 INTRODUCTION 17 factors than dysfunctional beliefs as measured by the OBQ might shed light on OCD within a cognitive perspective. In addition to studying content domains, the investigation of cognitive processes in OCD could further our understanding of the cognitive mechanisms behind OCD or certain subtypes of OCD. A reasoning process that is hypothesized to be important in OCD is inferential confusion, which reflects the tendency to negate reality on the basis of subjective possibilities (Aardema & O Connor, 2003). For example, a patient might come to the remote possibility that her hands are dirty, whereas there is no indication of them being dirty or there is even contradictory evidence for this to be the case (e.g., hands have just been washed with soap). Questionnaire research in an OCD patient sample showed that inferential confusion relates to OCD independently of cognitive domains (as measured by the OBQ), and mood states (Aardema, O Connor, Emmelkamp, Marchand & Todorov, 2005). A clear benefit of measuring reasoning tendencies is that the questionnaire items refer to a reasoning process rather than to specific belief content, and therefore can apply to different (symptom) subtypes of OCD. In Chapter 5, a study is presented which attempts to replicate the findings of Calamari et al. (2006) and Taylor et al. (2006), and in addition explores inferential confusion in relation to dysfunctional belief based subgroups. Intergenerational transmission of dysfunctional beliefs Several family studies showed that OCD relatives have a higher risk of developing OC symptomatology than controls (Nestadt, Samuals, Riddle, Bienvenu, Liang, LaBuda, et al. 2000; Pauls, Alsobrook, Goodman, Rasmussen & Leckman, 1995). Potential pathways of transmission of OCD can involve both genetic and environmental factors. So far, family studies mostly focused on the occurrence of obsessions and compulsions in relatives. However, from a cognitive-behavioural perspective, it is highly feasible that psychopathology can also be transmitted through adopting a relative s negative appraisals and dysfunctional beliefs. In support of the idea that dysfunctional beliefs of parents might be associated with children s psychopathology, results from a non-clinical sample showed weak though significant associations between parents beliefs concerning responsibility and threat estimation and importance and control of thoughts on the one hand and their children s OC symptoms on the other hand. In a similar vein, a weak significant relationship was found between parents OC symptom scores and children s beliefs concerning responsibility and threat estimation (Jacobi, Calamari & Woodard, 2006). Knowledge about intergenerational factors could contribute to our understanding of the development of OCD, and might have implications for prevention and/or treatment, for instance prevention programs for high risk families, or involvement of parents in treatment. In order to further the investigation of intergenerational factors related to OCD, chapter 6 presents a study within a family context. More specifically, associations between both OC behaviours and

18 18 CHAPTER 1 dysfunctional beliefs in OCD patients and their parents are studied: Do parents of OCD patients have increased levels of OCD symptoms, and are their dysfunctional beliefs associated with those of their offspring? Finally, chapter 7 discusses the theoretical and clinical implications of the studies presented in this thesis.

19 Processes of change in cognitive-behavioural treatment of obsessivecompulsive disorder; current status and some future directions Published as: Polman, A., Bouman, T.K., Van Hout, W.J.P.J., De Jong, P.J. & Den Boer, J.A. (2010). Processes of change in cognitive-behavioural treatment of obsessive-compulsive disorder; current status and some future directions. Clinical Psychology and Psychotherapy, 17, 1-12.

20 Abstract The present chapter discusses theoretical and methodological issues involved in the processes of change in cognitive-behavioural treatment (CBT) of obsessive-compulsive disorder (OCD). Treatment outcome studies showed that CBT is effective in reducing obsessive-compulsive symptoms. However, why and how CBT works cannot be corroborated by comparing pre- and post-assessment. Recently, there has been a resurgence of interest in theory driven process studies. By showing patterns of change over time, process studies can contribute to our insight into the actual mechanisms of change during treatment. We review process research in the field of OCD, and discuss methodological issues involved in process studies for this particular disorder. It is concluded that studying the processes of change harbours promising possibilities for bridging the gap between theory and clinical practice.

21 PROCESSES of CHANGE, CURRENT STATUS & FUTURE DIRECTIONS 21 Introduction According to Marks (2002, p. 200) therapy is coming of age regarding efficacy for anxiety and depression, but is only a toddler regarding the scientific principles to explain its effects. Cognitive-behavioural models are currently the most prominent theories on obsessive-compulsive disorder (OCD), but whether clinical improvement follows the predictions of these models has not been established, yet. To bridge this gap between theory and practice, process oriented research during the actual treatment phase is an important tool. In this chapter, we will review the current status of process studies in OCD, and discuss several methodological issues that seem relevant for process research in this disorder. Contemporary theories emphasize the role of dysfunctional beliefs in the development and maintenance of OCD (see for example, Clark & Purdon, 1993; Frost & Steketee, 2002; Rachman, 1997, 1998; Salkovskis 1985, 1989). Cognitive-behavioural models postulate intrusive thoughts as cognitive stimuli, and negative automatic thoughts and beliefs (concerning for example responsibility) as cognitive responses to the intrusions. These negative automatic thoughts and beliefs in turn would influence one s emotions, and affect behaviour (i.e., compulsions). An international group of researchers identified six OCD relevant belief domains; over-importance of thoughts, importance of controlling one s thoughts, perfectionism, inflated responsibility, overestimation of threat, and intolerance of uncertainty (Obsessive Compulsive Cognitions Working Group (OCCWG), 1997). A therapeutic implication of the cognitive-behavioural model is that automatic thoughts about the meaning of intrusions and the beliefs that give rise to them should be modified. Changing these beliefs (in whichever way) should lead to a decrease in OCD symptoms. Both exposure with response prevention (ERP), the gold standard in the treatment of OCD, and cognitive treatment (CT) are considered effective treatments for OCD (Cottraux, Note, Yao, Lafont, Note, Mollard et al., 2001; Foa, Franklin & Kozak, 1998; Freeston, Ladouceur, Gagnon, Thibodeau, Rhéaume, Letarte et al., 1997; Van Oppen, De Haan, Van Balkom, Spinhoven, Hoogduin & Van Dyck, 1995a; Whittal, Thordarson & McLean, 2005). Some evidence has been emerging for the relationship between successful treatment and significant changes in obsessional beliefs. Two studies involving exposure with response prevention in which cognitive changes were assessed, showed that improvement in OC symptomatology was related to improvement on cognitive measures (Emmelkamp, Van Oppen & Van Balkom, 2002). Furthermore, one of these studies showed that obsessive beliefs hardly changed over the course of treatment in the least improved patients. Cognitive behaviour therapy seems to be associated with cognitive change as well: Different research teams have demonstrated that reductions in OC symptoms correlate with changes in OC beliefs (Bouvard, 2002). However, since the studies covered in these reviews focused on the comparison of pre- and post-treatment measures, it cannot be corroborated whether changes in interpretations and longstanding beliefs precede, follow, or co-vary with changes in symptomatic outcome

22 22 CHAPTER 2 (cf. Emmelkamp et al., 2002). A more critical test of the cognitive-behavioural model would involve process-oriented research to determine whether decreases in dysfunctional beliefs mediate clinical outcome (Tolin, Woods & Abramowitz, 2003). Defining process research In general, two types of process studies can be distinguished; those that focus on change processes, and those that focus on treatment processes (Orlinsky, Rønnestand & Willutzki, 2004). Change processes occur within the patient and not exclusively or mainly during treatment, whereas treatment processes only take place during treatment sessions and involve the patient, the therapist, and their interaction. When process research concerns the investigation of mechanisms of change of a particular treatment, focus is on changes in overt and covert client behaviour during treatment (which can take place both inside and/or outside treatment sessions) (Orlinsky et al., 2004). Since process research concerns a different research question compared to traditional treatment outcome research, it also requires a different research methodology. Whereas treatment outcome research aims to establish treatment efficacy, process studies aim to elucidate how phenomena change over time. Therefore, many data-points per participant are required to reveal patterns of change over time. Data analysis concerns both intra- and inter-individual changes (instead of changes in group means, as is common in treatment outcome research). The additional value of process research is that the within-person level data provides information on within-person changes over time, as well as individual differences in such changes (Bolger, Davis, & Rafaeli, 2003). Individual differences might be linked to the extent of improvement (outcome). It may be assumed that the least improved patients show different patterns of change compared to the most improved patients. Moreover, different patterns of change might lead to improvement as well. For instance, in the investigation of processes of change in CBT for depression, both sudden gains and temporary worsening of symptoms (which was considered to be a sign of a learning process, in which old patterns were changed and new behaviours and beliefs emerged) predicted better treatment outcome (Hayes, Feldman, Beevers, Laurenceau & Cardaciotto, 2007). Process research could contribute to our understanding of a particular disorder and can help to optimize treatment methods and their effects (Whisman, 1993). For instance, in case of the temporary worsening in depressed patients which occurred during an exposure phase of the treatment, a therapist might consider a patient s resources to deal with this temporary worsening before starting with this part of the treatment (Hayes, et al., 2007).

23 PROCESSES of CHANGE, CURRENT STATUS & FUTURE DIRECTIONS 23 In the following sections we will focus on process studies investigating mechanisms of change in OCD, and the methodological issues involved in such studies. Process studies in OCD To obtain an overview of the status of process research focusing on mechanisms and processes of change in OCD, a literature search was conducted, using PSYCHINFO and MEDLINE-Pubmed from 1985 to 2008 (week 52) with keywords process, treatment, obsessive, compulsive as well as synonyms and related words, and using the reference lists of various publications. This literature search yielded papers related to different types of processes, e.g., neurological and biological processes not related to treatment, and papers related to general treatment processes like the therapeutic relationship. Selection was based on two criteria. Firstly, the paper should concern cognitive-behavioural treatment for patients suffering from OCD. Secondly, the processes under study should involve processes of change, as described in the previous section. Papers were examined by the first author in consultation with the second author, which eventually yielded eight relevant papers. Table 1 provides a summary of these studies, describing the samples, theoretical backgrounds, provided treatments, assessment points, types of process measures, data-analytical techniques and results. In this section, we will briefly describe the studies and discuss theoretical implications, in the next section we will discuss their methodological strengths and weaknesses and methodological aspects of studying processes of change in OCD in general. The studies of Kirkby, Berrios, Daniels, Menzies, Clark, and Romano (2000), Kozak, Foa, and Steketee (1988) and Moergen, Maier, Brown and Pollard (1987) mainly concerned exposure with response prevention. Kirkby et al. (2000) studied the efficacy of a computer aided ERP training consisting of 3 sessions. Participants were asked to learn to do exposure and ritual prevention by directing a person pictured on the screen who is stated to have contamination obsessions and washing rituals. An on-screen thermometer provided feedback on the current anxiety of the figure, which increased with hand-dirtying behaviour. Participants scored points when they got the figure s hands dirty without washing them afterwards. As participants continued vicarious exposure to dirt, the figure s anxiety level declined, simulating habituation. Results showed that more hand dirtying enactments in the first session were related to better treatment effect, and it was concluded that repeated vicarious exposure was a therapeutic factor. Kozak et al. (1988) studied processes related to outcome in an exposure treatment in 14 OCD patients based on the emotional processing theory (Foa & Kozak, 1986). According to this theory it was hypothesized that fear activation should

24 24 CHAPTER 2 occur during exposure and that habituation of fear would take place within and across sessions. Processes were measured during sessions 6 and 14 with electrodermal activity, subjective units of discomfort (SUDS) rated on a scale, and heart rate. Results of both physiological and self-report measures showed that anxiety increased during exposure and that habituation took place during the sessions. Habituation over sessions was established with self-report only. In relation to treatment outcome, results showed that activation of fear and habituation over sessions predicted better treatment results. Moergen et al. (1987) studied generalization of fear habituation in a multiple baseline across fear stimuli design (i.e., visual, auditory, behavioural and cognitive), in an OCD patient with fear of the number 13. Following a baseline phase, visual stimulus exposure was provided. Subsequently, exposures to auditory, behavioural and cognitive stimuli were added on the 6th, 9th, and 12th day of treatment. Results showed decreases in anxiety levels in response to all four stimuli forms before prolonged exposure to a new type of stimulus was introduced during treatment. These results suggest that exposure to a single stimulus form may result in sufficient generalization of treatment effect. However, the single case design precludes firm conclusions, for instance it is unclear whether a different order of stimulus types during the exposure would have lead to the same results. Overall, these three studies provided specific information on processes of change related to treatment success of exposure based treatments. The studies of Anholt, Kempe, De Haan, Van Oppen, Cath, Smit and Van Balkom (2008) and Wilson (2002) investigated the order of changes in obsessions and compulsions during respectively ERP versus CT, and CT by measuring obsessions and compulsions on a weekly basis. Anholt et al. (2008) hypothesized that ERP would primarily affect behaviour, thus reducing compulsions first, and that CT would primarily affect thought and therefore obsessions were expected to reduce first. However, changes in compulsions preceded changes in obsessions in both treatments, which lead to the conclusion that reduction of compulsions is the pro cess through which both ERP and CT give rise to change. However, Wilson s (2002) results were not in line with these findings; sometimes changes in obsessions, compulsions and anxiety co-occurred, however sometimes they also changed in opposite directions. Results gave no indication for decreases in compulsions to precede decreases in obsessions. It should be noted that Wilson s (2002) study concerned individual trajectories of change of six subjects, whereas Anholt et al. (2008) studied groups of subjects thereby averaging out individual differences, which could explain the differences in findings. A limitation of these studies is that the only cognitive aspect that was measured were obsessions. Cognitive treatment explicitly challenges negative appraisals and dysfunctional beliefs. Also, ERP is considered to indirectly- affect appraisals and beliefs by disconfirming and thereby weakening the strength of catastrophic thoughts (e.g., Rachman, 1996). Furthermore, the experience of mastering an OCD relevant situation might induce more positive associations with respect to the OCD stimuli and self efficacy, which

25 PROCESSES of CHANGE, CURRENT STATUS & FUTURE DIRECTIONS 25 further affects the strength and impact of dysfunctional cognitive associations (cf. Bandura, 1977). So, both ERP and CT are expected to affect beliefs and appraisals, and therefore, the investigation of these cognitive aspects (besides changes in obsessions) would shed more light on the processes of change of these two treatments. The studies of Rhéaume and Ladouceur (2000), Storchheim and O Mahony (2006), and Williams, Salkovskis, Forrester and Allsopp (2002) explicitly investigated changes in beliefs both during C(B)T and ERP, based on the cognitive-behavioural model of OCD. All three studies reported decreases in dysfunctional beliefs during successful treatment, irrespective of type of treatment, which is in line with the conclusions based on treatment outcome studies that successful treatment of OCD is related to significant changes in obsessional beliefs (Emmelkamp et al., 2002; Bouvard, 2002). As for the process of change of these dysfunctional beliefs, different aspects were studied: Rhéaume and Ladouceur (2000) investigated whether changes in beliefs were a precursor of changes in compulsive behaviour and vice versa, and found this relationship to be bi-directional. Williams et al. (2002), and Storchheim and O Mahony (2006) assumed that changes in beliefs and compulsive behaviour would occur in tandem, and results were in line with this hypothesis. Rhéaume and Ladouceur (2000) also investigated possible differences between ERP and CT in processes of change, and found that in both treatments cognitive changes sometimes preceded behavioural change, and in some instances followed behavioural change within the same subject. Furthermore, patients in the ERP condition showed a larger number of beliefs involved in the process of symptom changes than patients in the CT condition. In spite of methodological differences and limitations, the above mentioned process studies provided specific information on processes of change in relation to treatment outcome, e.g., hand-dirtying behaviour in session one predicted better treatments results in a computer aided treatment (Kirkby et al., 2000), and activation of fear and habituation over sessions predicted better results for ERP (Kozak et al., 1988). When considering the theoretical background (see Table 1), it is interesting to see that mechanisms of change in ERP were studied based on the habituation/emotional processing theory (Foa & Kozak, 1986), as well as based on the cognitive-behavioural model. Kozak et al. (1988) also proposed that besides activation of fear and habituation over sessions, cognitive variables would probably be involved in an exposure treatment as well. With respect to the cognitive-behavioural model of OCD, it can be concluded that cognitive change seems to be related to treatment effect. Furthermore, the relationship between behavioural- and cognitive change seems to be closely linked, and might be bi-directional.

26 26 CHAPTER 2 Methodological issues in process studies Sample size & -composition Concerning methodological characteristics of the studies in Table 1, first of all, the rather small sample sizes of the studies may catch the eye, ranging from 1 participant to 14 participants. The benefit of closely studying a small number of cases however, is that it facilitates more in-depth analysis of individual patterns of change (Wilson & Chambless, 2005), and therefore can shed light on possible qualitative characteristics of change processes. For instance, Rhéaume and Ladouceur (2000) reported that distinct patterns were observable in all patients in terms of number of beliefs involved as well as the various links between them and the symptoms. This indicates that the process might be highly idiosyncratic, something which would not easily be found in a large patient sample which involves averaging data. However, sample size is also one of the major limitations of most studies presented in Table 1. For example, Wilson (2002) reported that two of the six included patients had co-morbid post traumatic stress disorder (PTSD), another Table 1. Overview of the methodological issues of process studies in obsessive-compulsive disorder Study Sample Theoretical background Anholt et al. (2008) Kirkby et al. (2000) Kozak et al. (1988) Moergen et al. (1987) Rhéaume & Ladouceur (2000) Storchheim & O Mahony (2006) Williams et al. (2002) Wilson (2002) n = 61; 54.8% F Mean age: 33.8 (18-65) Diverse symptom themes n = 13; 9 F Mean age: 41 (28-54) 13 washers n = 14; 8 F Mean age: 34.4 (19-50) 10 washers, 4 checkers n=1, M age: 20 fear of number 13 n = 6; 3F Mean age: 35.3 (21-43) 6 checkers n = 6; 6F Mean age: 30.2 (18-43) Diverse symptom themes n = 6; 3 girls Mean age: 14.3 (12-17) Diverse symptom themes n = 6; 3 F Mean age: 33.2 (20-50) Diverse symptom themes Habituation theory & Cognitive-behavioural model Emotional processing theory Habituation theory Cognitive behavioural model Cognitive behavioural model Cognitive behavioural model Cognitive behavioural model Treatment CT n = 31 ERP n = sessions Computer aided ERP training 3 sessions ERP 15 sessions ERP 23 days (inpatient) ERP (n=3) CT (n=3) 24 sessions 5 sessions ERP + 7 sessions ERP & CT CBT 7-10 sessions CT sessions Assessment points Weekly During the sessions During sessions 6 and 14 Daily, after each presentation of number 13 Daily Daily Once per session Daily

27 PROCESSES of CHANGE, CURRENT STATUS & FUTURE DIRECTIONS 27 patient had a co-morbid depression, and another one might have had some slight cognitive deficits. In a large sample the effects of comorbidity may be averaged out across patients. With only six patients however, two out of six with co-morbid PTSD means a third of the sample, which obviously complicates straightforward conclusions. Furthermore, comparing the process of change in ERP and CT is an interesting research question, but cannot be answered by studying three patients per treatment condition (see Rhéaume & Ladouceur, 2000). So, initially process research might benefit from small scale studies to explore processes of change and individual differences closely, but when general tendencies become clear, these should be studied in large samples as well, in order to facilitate generalization of results. Secondly, the composition of the samples deserves attention. Some samples concern a specific subtype of OCD (e.g., checkers in Rhéaume & Ladouceur, 2000), whereas other studies investigated OCD in general. The importance of studying OCD subtypes is currently supported by various researchers (e.g., McKay, Abramowitz, Calamari, Kyrios, Radomsky, Sookman, Taylor & Wilhelm, 2004), and Process measure Data-analysis of process Results Structured Clinical Interview Generalized estimating equations Reductions in compulsions preceded reductions in obsessions in both ERP and CT Human Computer Interactions (HCI) Physiological measures (electrodermal activity, heart rate) and self-report Specifically written computer programs automatically analysed HCI Univariate repeated measures ANOVA + correlation More hand dirtying enactments in the first session were related to better treatment effect repeated vicarious exposure is therapeutic factor Activation of fear and habituation over sessions predicted better treatment results Self-report Visual inspection Generalization over stimuli took place before a new type of stimulus was introduced during treatment Self-report Multivariate Time Series Analysis (V-ARMA) Bidirectional relations between changes in beliefs and behaviour in both ERP & CT Self monitoring forms Visual inspection + calculation of critical difference Self-report Visual inspection + correlation Self-report Visual inspection + piecewise linear growth modelling Changes in beliefs and behaviour occurred in tandem during both treatment phases Changes in responsibility kept pace with changes in symptom level In some time segments obsessions and compulsions changed together, but sometimes also in opposite directions