An examination of the process of motivational interviewing in the anxiety disorders

Motivational interviewing (MI) is a collaborative, client-centered therapy style that aims to prepare people for behaviour change by helping them to explore and resolve ambivalence (Miller & Rollnick, 2002, 2013). MI was originally developed to treat problematic substance use but is increasingly used as both a stand-alone and adjunctive treatment for a variety of physical and mental health concerns. Proposed mechanisms of MI’s success have been well specified, however, most research that examines MI mechanisms and particularly MI’s proposed causal model has been conducted in the realm of substance use. Little is known about the generalizability of MI mechanisms from the substance use literature to the other problem areas where MI is being applied. The current thesis aims to address this gap by investigating the process of MI in areas beyond substance use. The thesis combines different approaches to address this central question. The first two papers investigate the current state of MI mechanism research. Paper One is a systematic review of evidence for the causal chain model proposed by Miller and Rose (2009). The review draws together research that tests paths of the causal chain in varying treatment domains. Overall, the results provided support for the link between therapist MI-consistent behaviors (MICO) and client change language, and between client language and treatment outcome. In terms of the relational hypotheses, therapist relational style factors were not consistently related to change language or outcome, however studies that employed experimental designs that isolated the effect of MI relational components did demonstrate positive effects of therapist interpersonal style. Paper Two is a meta-analysis that investigates MI mechanisms of change in populations diagnosed with mood, anxiety, psychotic, and eating disorders, and patients with comorbid mental health conditions. Pooled effect sizes demonstrated that the application of MI was related to specific therapist and client behaviors, and that client behaviors may predict treatment outcome. Moreover, there did appear to be some support for the effect of MI in increasing motivation and treatment engagement (homework compliance and treatment attendance), particularly in the context of anxiety disorders. Nevertheless, a caveat of the effect of MI on motivation and engagement is that studies generally did not employ control conditions. Taken together, the review papers pointed to limited use of control conditions and few investigations of therapist behaviors and change language in the context of anxiety disorders. Therefore, the final three papers of this program of research aim to overcome these limitations and are dedicated to an empirical examination of MI processes in the context of social anxiety disorder (SAD). Each paper employs a sample of adults diagnosed with SAD who were randomised to receive either an MI-style treatment called Treatment Expectations and Engagement (TEE) or a supportive counselling control condition (SC) before all received group Cognitive Behavioural Therapy (CBT) for SAD. Paper Three investigates the capacity of MI to decrease ambivalence for people with social anxiety and whether levels of ambivalence are related to treatment outcome. Overall, Paper Three provided mixed support for the relationship between MI and ambivalence in a socially anxious population. MI was not shown to decrease ambivalence, either general ambivalence, treatment ambivalence, or ambivalence related to specific CBT tasks, however, the condition to which participants were allocated did moderate the effect of some measures of baseline ambivalence on treatment outcome. Also, higher CBT task-related ambivalence was related to higher social anxiety symptoms during the CBT program. Together, the findings suggest that while MI might not decrease ambivalence it may alleviate a negative effect of ambivalence on treatment outcome. Paper Four employs observational coding methods to examine the transition between therapist and client behaviour during MI sessions for social anxiety. Consistent with the proposed causal chain of MI, therapist MICO behaviors were more likely to precede client change talk (CT), while MI-inconsistent behaviors (MIIN) were more likely to precede neutral client language and less likely to precede CT. MICO behavior was also more likely to precede counter-change talk (CCT), suggesting that it may facilitate change exploration in general. Furthermore, the findings highlighted the importance of particular types of MICO behaviors particularly open questions and valenced reflections. Taken together, the findings of Paper Four supported the first step of the MI causal model in the context of social anxiety, though the study did not examine whether therapist and client behaviors predicted treatment outcome. Thus the final empirical paper of this thesis had this as an aim. Paper Five further explores the relationship between therapist behaviours and client language in MI, as well as the relationship between therapist and client variables and outcome. Overall, the results of Paper Five supported some elements of the MI causal model in the context of SAD, but not all. First, the TEE condition was generally distinguishable from the SC condition according to MI relevant behaviors, which suggests that TEE was being conducted in an MI fashion. Second, while the MICO behavior category did not predict change language, MI Spirit and specific therapist behaviors did, with MIIN also predicting a greater frequency of CCT. Lastly, therapist and client behaviors did not predict treatment outcome, thus the role of client language as a mediator of treatment outcome was not investigated. Taken together, the results of the empirical papers suggest that elements of the theorized process of MI are relevant to social anxiety treatment, and that while not all hypotheses were supported, there is reason to suggest that these processes should be further investigated within this context. An overall implication of the current thesis is that while each empirical paper contributes to MI process research, they also highlight the complexity associated with understanding how particular MI mechanisms contribute to outcome.