Process and change in cognitive behaviour therapy for chronic pain

Cognitive behaviour therapy (CBT) is a complex multicomponent treatment that can be applied to the problem of chronic pain rather than to specific diagnostic categories. Its philosophy of active patient participation, education and guided personal experimentation make it relatively easy to incorporate into multidisciplinary pain programs where contributions from other health care disciplines can be framed within the general principles of CBT. There are about 40 published randomized controlled trials (RCT) of CBT in a variety of implementations. The evidence indicates that CBT is more effective than no-treatment (wait list controls) on a wide range of measures and at least as effective as other treatments on some measures and superior on several of them (Morley et al., 1999). In contrast there is relatively little published research on the process of change in CBT. In this issue Nielson and Jensen (2004) report a study of process in a sample of about 200 patients with fibromyalgia (FM) exposed to a CBT-oriented multidisciplinary program. The study is a welcome addition to the sparse literature and its findings broadly replicate previous studies by Jensen et al. (1994, 2001) using similar measures and methodology. The aim of such research is to identify change processes in treatment for theoretical and practical purposes: they test for the validity of the treatment model and highlight aspects of treatment that might be enhanced by further improvement. The magnitude of this task cannot be underestimated and there are major challenges to be addressed in clinical research. Nielson and Jensen’s article should provide the research community with an opportunity to reflect on several of the issues relevant to this work. First, although the evidence for the effectiveness of CBT for chronic pain is considerable the design of current trials (e.g. devising control groups with equal credibility and expectation of change; establishing equivalence of therapist allegiance and competence across groups) means that it is difficult to exclude several other plausible hypotheses that may account for the data. These problems in part reflect the complexity of developing a comprehensive and well specified psychological model of chronic pain and also reflect significant problems in evaluating psychological treatments per se, where double-blind RCT methodology is not readily implemented (Baskin et al., 2003; Schwartz et al., 1997; Wampold, 2001). Although the solutions to these issues will take time, the investigation of processes which underpin changes in outcomes, irrespective of the treatment model, should proceed in parallel. Second is the development of more precise specifications of CBT. Present descriptions (Turk, 2002), provide excellent general descriptions of the assumptions but do not give specific predictions about which particular processes and treatment components are associated with which specific changes. It is unlikely that all possible cognitive changes can be explicitly mapped a priori onto specific outcomes, and it is equally possible that different treatment components may have a common effect on a given outcome. As a consequence researchers articulate their causal models in a relatively general manner—‘some changes in coping will be associated with and precede changes in some outcomes’. The risk in this is that studies may be critiqued, as Nielsen and Jensen do in their discussion, that ‘not all possible pain-related coping strategies and attributions were assessed’. The empiricist’s approach of investigating the matrix of all possible cognitive processes, coping strategies and attributions against all possible relevant outcomes leads to a prohibitive number of tests. A similar analysis of the psychotherapy literature (Stiles et al., 1986) led to a re-evaluation of the best approach to the matrix problem in psychotherapy research. Researchers in CBT and chronic pain need to be strategic and develop more precise models of change, linking specific cognitive changes to specific outcomes. Specification of such relationships is likely be driven by both theoretical and empirical considerations. Nielson and Jensen’s observations that changes in adaptive beliefs are consistently related to a range of outcomes replicates an earlier study (Jensen et al., 2001) and indicates the importance of beliefs of control and harm; beliefs that have been identified as central in a more restricted and highly specified model of CBT (Vlaeyen and Linton, 2000). The complexity of CBT for chronic pain, viz. its content and delivery (multiple components delivered in a group setting by several therapists), should encourage the explication and clear specification of important causal relationships. Increased clarity will: focus research on particular relationships and negate the need for spurious ‘fishing expeditions’; aid the specification of statistical models; indicate better