Prevention & Treatment, Volume 3, Article 24, posted June 2, 2000
Copyright 2000 by the American Psychological Association

IN MEMORY OF NEIL S. JACOBSON

The Art of Randomized Clinical Trials

Keith S. Dobson
University of Calgary

Correspondence concerning this article should be addressed to Keith S. Dobson, Department of Psychology, University of Calgary, Calgary, Alberta, Canada T2N 1N4.
E-mail: ksdobson@ucalgary.ca

When asked to write this article as an "inside" collaborator with Neil Jacobson in the area of depression, I immediately thought that the component analysis of cognitive therapy was, without a doubt, the work that Neil would have wanted cited. I know that he was keenly interested in the project from the outset, and its results only whetted his appetite for more work in the area. In this article, I offer my version of how the project came to be and some of the processes that are not captured in the formal presentation of the research results.

I first met Neil Jacobson in the fall of 1982. I was a new faculty member at the University of British Columbia and was invited by Neil to the University of Washington to consider a role as consultant to a project that was just beginning, comparing behavioral couples therapy to individual cognitive-behavioral therapy to a combination treatment for depressed, married women. Although offered financial remuneration for my work, I instead asked for a more active role in the project and consideration of publication credit. Our collaboration began then and continued until Neil' s untimely death.

As those who have read the results of the above trial (Jacobson, Dobson, Fruzzetti, Schmaling, & Salusky, 1991; Jacobson, Fruzzetti, Dobson, Whisman, & Hops, 1993) know, cognitive therapy had as much effect on couple satisfaction when the depression lifted as did couples therapy on couple satisfaction and depression. Neil was much impressed with the results of this trial and became increasingly interested in cognitive therapy, as it was generally achieving increasing attention in the literature (Dobson, 1989; Hollon, Shelton, & Loosen, 1991). Even at that point in time, Neil was questioning the process of change in cognitive therapy for depression, and I recall long discussions with him about the oft-cited observation that most of the change in depression scores occurred early in the process of cognitive therapy, at a point in the treatment in which behavioral procedures were typically dominant (Persons, 1989).

The component analysis of cognitive therapy was a collaboration between me and Neil in the best sense of the word. My training was in cognitive therapy, and I was (and continue to be) a committed cognitive therapist. As I have written elsewhere, I believe that cognitive therapy is the "gold standard" in psychotherapy for depression (Dobson, Backs-Dermott, & Dozois, 2000). At the same time, I recognize that we know relatively little about the mechanisms of change in cognitive therapy (cf. DeRubeis & Feeley, 1990) or the relative importance of the various techniques found in treatment manuals (A. T. Beck, Rush, Shaw, & Emery, 1979; J. Beck, 1995; Dobson & Shaw, 1988). Neil, as a behaviorist, questioned the need for all of the training and various techniques that comprised cognitive therapy, and he brought a skeptical approach to the study. Thus, when Neil proposed the component analysis, I was more than ready to engage in those discussions.

Component analyses take various forms, but the two primary designs involve either dismantling or constructive methodologies (Kazdin, 1994). In designing the component analysis of cognitive therapy, Neil argued vociferously for a complete dismantling design, in which behavioral interventions would be compared with automatic thought interventions, which in turn would be compared with interventions focusing only on core belief change. I argued that such a design would not be a test of cognitive therapy, because no one in the field would conduct core belief or assumptive interventions without first doing the other work of cognitive therapy. In the end, Neil was convinced both by my arguments and by consultation with other experts in the field (notably, Steve Hollon), and the constructive design was adopted in the final proposal.

Another important design consideration was whether to include a treatment as usual, waiting list or other control condition in the study. Arguments have been made both for and against the idea of requiring a control condition, notably as a consequence of the NIMH Collaborative Study on the Treatment of Depression (Elkin et al., 1989; Jacobson & Hollon 1996). Here, Neil' s expertise as a successful grantee proved indispensable. He correctly read the state of the research community to be receptive to a study with no control condition, as long as some standardized treatment was used. Further, because three treatment study conditions would permit more participants per condition than would a study with four treatment conditions, he was able to argue that the design used was actually superior from a power consideration.

Other aspects of the study's methodology were relatively straightforward. The criteria that were adopted for measuring change, as well as the process measures, required careful consideration, but the operating rule was that wherever possible, state-of-the-art methodology would be adopted. I took the lead in writing the component treatment manuals for the study, which were subjected to some peer review. Our preparation for the study was further enhanced by the fact that we had a group of qualified and cooperative cognitive therapists from the previous trial to participate in the study.

As has been discussed elsewhere (Jacobson & Gortner, 2000), the results of the component analysis were contrary to the therapists' beliefs about which treatment condition should have superior results and were also contrary to my supposition that providing therapists with increased resources (i.e., extra treatment interventions) would result in superior outcomes. When these results failed to materialize, there was some hope that at least at follow-up, the complete treatment might yield superior results, consistent with some indication in the literature that cognitive therapy is superior at follow-up (Dobson et al., 2000). The failure of the cognitive components to add to long-term results is a challenge to the model and technology of cognitive therapy.

The fact that we found no additive value for either automatic thought or assumptive interventions (Jacobson et al., 1996; Gortner, Gollan, Dobson, & Jacobson, 1998) was very much taken in stride by Neil. This finding was consistent with his predominantly behavioral perspective, and as it was contrary to many expectations, it was provocative. I distinctly recall pointed questioning by Neil about whether my support for cognitive therapy was shaken by these results. We had a number of discussions about the utility of the cognitive interventions in cognitive therapy and whether or not behavioral change also requires some basic cognitive processes (e.g., trust, belief in the therapist, acceptance of the treatment rationale, etc.).

For his part, Neil was convinced that the results of the component analysis revealed the primary importance of behavioral processes in the treatment of depression. Behavioral activation, or BA, as it became known, began to take on a life of its own. Although we were acutely aware of the need to replicate our results, Neil began to expand on BA as a viable treatment. In the treatment of depression trial that is currently in progress at the University of Washington, we have pitted BA against cognitive therapy and three medication conditions (i.e., SSRI, with continuation into the follow-up period; SSRI, without continuation into the follow-up period; and placebo medication).

In this more recent trial, BA has continued to develop, and several innovative treatment methods have now been added to the model. One feature of the revised BA that I believe will be of immense importance is its emphasis on avoidance mechanisms in depression. Indeed, one can almost conceptualize several aspects of the behavioral patterns seen in depression as either active or passive avoidance of problematic situations or people with whom depressed individuals do not believe they have the ability or interpersonal strength to effectively deal. One of the key aspects of BA is in working with patients to identify their avoidance patterns and using directive behavioral strategies to either teach effective coping behaviors or correct avoidance with active coping strategies. In some respects, this aspect of BA can be directly traced to the work of Ferster (1973), which was never fully developed or tested.

The lessons to be learned from component analyses of cognitive therapy are still being discovered. Our current trial (with Drs. Michael Addis, David Dunner, Steve Hollon, and Robert Kohlenberg as collaborators) will test the stability of the previous finding about BA. Neil was writing a trade book on BA, coauthored with Judith Woodburn, when he died. We are also aware that several investigators have taken the BA manual and are examining it for clinical trials for use in several locations.

For my part, the opportunity to work with Neil Jacobson on clinical trials was an opportunity to witness first hand a research maestro at home in his field. Neil had a fascination with controversy, and his primary tool in wrestling with controversial issues was data. Despite his deep convictions about his theoretical and clinical judgments, which could have lead to a certain "blindness," he was a consummate scientist. Frustrated at times about the politics of science, he nonetheless knew the system and used it to maximal advantage. His success in obtaining research funds is a clear testament to his grant-writing ability and his sense of timing.

One of the aspects of a researcher's career that does not appear, except indirectly in the author list, is the extent to which he or she supports and encourages those around them. Having had the privilege to work with Neil over approximately 17 years, I can attest to his concern—even passion—for the strongest possible work from his graduate students and research team. He demanded much from his coworkers and rewarded them accordingly. It is no surprise to me that many of his graduate students have gone on to academic careers themselves, given his strong leadership and role modeling. Part of the art of research is working with other people, inspiring in them confidence and a strong commitment to sound science. Without doubt, the project that is the focus of this series is all the better for Neil's inspiring leadership.

References