Prevention & Treatment, Volume 3, Article 25, posted June 2, 2000
Copyright 2000 by the American Psychological Association

IN MEMORY OF NEIL S. JACOBSON

Do Cognitive Change Strategies Matter
in Cognitive Therapy?

Steven D. Hollon
Vanderbilt University

Correspondence concerning this article should be addressed to Steven D. Hollon, Department of Psychology, Vanderbilt University, 306 Wilson Hall, Nashville, Tennessee 37027.
E-mail: steven.d.hollon@vanderbilt.edu

Following Neil Jacobson's untimely death, Keith Dobson and I were asked to select an article that represented Neil's most influential and important contribution to the treatment of depression. Acting independently, we both nominated the same article, his recent component analysis of cognitive therapy (see Jacobson et al., 1996, reprinted in this issue). In that article, Jacobson and colleagues found that the behavioral component of cognitive therapy was as effective and enduring as the full treatment package. This was an unexpected and provocative finding. Following cognitive theory, Neil had predicted that adding strategies and techniques to identify and test specific negative automatic thoughts would enhance the efficacy of behavioral activation alone (Beck, Rush, Shaw, & Emery, 1979). It did not. Also in keeping with cognitive theory, Neil had further predicted that adding efforts to modify and address underlying cognitive vulnerabilities would reduce the risk of subsequent relapse (Hollon & Garber, 1980). Again, this was not the case (see also Gortner, Gollan, Dobson, & Jacobson, 1998).

If these findings replicate, they have two important implications. Theoretically, they call into question the notion that cognitive therapy works, when it works, by virtue of using cognitive change strategies to produce change in beliefs. My own work and the work of others like me have generally supported the notion that cognitive strategies produce change in beliefs that are associated with symptom change and the reduction of subsequent risk (DeRubeis et al., 1990; DeRubeis & Feeley, 1990; Hollon, Evans, & DeRubeis, 1990). However, we have never explicitly isolated the behavioral components from the rest of cognitive therapy. Because we have always retained those behavioral components in our larger treatment package, our studies have not addressed the issue in a manner as free from confound as Neil and his colleagues were able to do. Clinically, Neil's study points the way to a simpler and more cost-efficient methodology. It may even serve to revitalize more purely behavioral approaches to the treatment of depression. Cognitive therapy was revolutionary in its day because it seemed to provide a quick and relatively simple way to treat depression. It is certainly easier to learn and easier to teach than more traditional interventions. The speed with which it worked made it ideal for a treatment climate in which efficiency was increasingly prized. Moreover, there were repeated indications that clients learned something in the course of treatment that served them well after treatment was over; that is, that it had an enduring effect that reduced subsequent risk (Hollon, Shelton, & Loosen, 1991). However, behavioral activation is even easier to learn and easier to teach than the cognitive components of the full treatment package. If it really is true that behavioral activation is comparable in efficacy and just as long lasting, then it might come to be preferred on the grounds of simple expediency.

I am by no means sure that this will prove to be the case, but to his credit, Neil was committed to testing the notion and pushing the limits of his earlier results. To critics like myself who argued that one could only take null findings so far, he responded by pointing out that his outcomes were comparable to those produced by other studies and conducting careful manipulation checks on his data. Attrition was lower than in most other trials, and rates of response and freedom from subsequent relapse were comparable to what is typically observed in those other studies. He and his colleagues relied on experienced therapists with an allegiance to cognitive therapy to conduct the intervention. Further, he secured the services of Keith Dobson, a recognized expert in the modality, to oversee its implementation and provide additional training and supervision. Neil developed (along with his colleagues) a series of adherence ratings that clearly differentiated among the respective components and was able to demonstrate that each of the successive interventions was implemented in the manner intended with a minimum of unintended contamination (Waltz, Addis, Koerner, & Jacobson, 1993). Finally, he was able to show, using ratings by Keith Dobson and two other independent experts, that cognitive therapy was implemented in a competent manner, as indexed by ratings on the Cognitive Therapy Scale (Jacobson & Gortner, 2000).

Nonetheless, questions can still be raised regarding the interpretation of his earlier findings. At the least, one must always be cautious about interpreting null findings, something of which Neil was fully aware (Jacobson & Hollon, 1996). Null findings could reflect comparable efficacy or comparable inefficacy. Neil was a superb behavior therapist, but he did not consider himself to be an expert in cognitive therapy. Keith Dobson has the necessary expertise, but he was off-site. Keith routinely monitored tapes and provided feedback over the phone, but it was rarely the case that supervision based on a given session could be provided prior to the next session, as is typically the case when on-site supervision is available. Moreover, Neil's therapists, although experienced, were not necessarily that proficient with cognitive therapy. Some were excellent, but others were not likely to be invited back for subsequent studies. Finally, it was not clear that the full treatment package was implemented in as fully integrative a manner as is typically the case. In most studies, cognitive therapists introduce behavior change strategies in the context of testing specific beliefs and encourage clients to use behavioral strategies to examine the accuracy of their underlying beliefs and attitudes. That is, cognitive therapy typically integrates behavioral and cognitive strategies in an ongoing and interactive fashion. In Neil's study, the respective components were provided in a more nearly sequential fashion. Whether this would make any difference with respect to outcome remains unclear, but it does represent a procedural difference between the component analysis trial and earlier studies.

Given these considerations, Neil designed a subsequent study to address these concerns. The best way to deal with the possible null findings is to build in additional conditions that are likely to differ in efficacy. Including a known contrast is ideal. In his current study (still ongoing), Neil incorporated a medication condition (Paroxetine) and a pill-placebo control. In order to ensure that drug treatment was adequately implemented, he called on David Dunner, M.D., an experienced research pharmacotherapist, to oversee that condition. Antidepressant medications represent the current standard of treatment for depression, and drug–placebo differences have been replicated in literally hundreds of trials. If cognitive therapy in this current trial performs at a level comparable to medications (and superior to pill-placebo), it would be difficult to argue that it was implemented in a manner that rendered it ineffective. Keith Dobson remained actively involved in the training and supervision of the therapists and was joined by me in that regard. That allowed us to devote more time to the study therapists and shorten the time elapsed before supervision could be provided. Neil also recruited Sandra Coffman, one of the most experienced and highly respected cognitive therapists in the Seattle area, to serve as a therapist in the study and chair the weekly on-site supervision meetings. For the others, he recruited Steve Sholl and David Kosins, two therapists trained by the Center for Cognitive Therapy in the context of a multisite cardiology trial. Steve had also participated in the component analysis study and received excellent reviews. Finally, he made sure that cognitive therapy was implemented in a fully integrative fashion, much as it is in other trials and just as specified by its creators.

Although the trial is still ongoing, results to date suggest that cognitive therapy will do quite well and that medications will prove superior to pill-placebo. If these findings hold, it would be hard to criticize the implementation of either cognitive therapy or drug treatment. It is too early to have any good sense about the prevention of subsequent relapse. But what had Neil most excited was that behavior therapy was at least as effective as either established intervention. For this trial, Neil did not simply rely on the behavioral components of cognitive therapy. Rather, he developed a more fully realized contextual behavioral approach that emphasized teaching patients to conduct a functional analysis of their own behaviors. As opposed to either cognitive therapy or drugs, patients are told that the locus of the problem lies not in themselves (either their beliefs or their biochemistry) but rather in the situations with which they must deal. That is, it is not that patients have deficits but rather that they confront problems that are difficult to solve.

I am a committed cognitive therapist (based on the experiences that I have had and the evidence that I have seen), and I have difficulty imagining how a purely behavioral approach can prove sufficient to deal with the kinds of clients with whom I have worked over the years. Nonetheless, I share a sense that behavior change is an important mechanism (as do most cognitive therapists), and I am impressed with Neil's data (partial though they are) and with the tapes I have seen. We spent many a long hour watching behavior therapy sessions from his ongoing trial, and they really differ from what I do in therapy or what his cognitive therapists do in the cognitive therapy sessions in that same trial. I think Neil may have been on to something, and I very much want to see his work continue. I still believe that patients fall prey to their own negative expectations and underlying beliefs, as specified by cognitive theory. But I think it just might be possible that an approach that focuses on problem solving and behavior change may be more effective than one that requires that patients deal with the more complex issues of identifying and testing specific beliefs and underlying attitudes. This is ultimately an empirical question and one that deserves to be asked. I think Neil and his colleagues were well on their way to providing an initial answer before his untimely death.

The real importance of Neil's component analysis study was that it started us thinking once again about whether more purely behavioral interventions could produce change in depression. The component analysis study suggested that they could, but it left unanswered questions about the efficacy of the full cognitive therapy treatment package. Neil and his colleagues did a truly remarkable job of addressing these questions within the context of the existing data, but there are limits to just how far null findings can be taken. In response to these lingering concerns, Neil and colleagues designed a subsequent study that addressed many of the problems in his component analysis trial. In the process, he became convinced that it was important not to restrict behavior therapy to only those strategies that were a component of cognitive therapy, but to treat it as a bona fide treatment in its own right. In that regard, he created a behavior-analytic framework that emphasized conducting functional analyses of the environmental events that have impinged on clients to create contextual shifts that deny access to reinforcement. In essence, he changed the rationale within which behavioral strategies are pursued and made it a very different approach than was represented in the component analysis study.

Neil and I first met at a conference in Pittsburgh, Pennsylvania in the late 1970s (Rehm, 1981). That conference had been convened to consider behavioral and cognitive-behavioral interventions for depression. At that time, it was unclear whether behavioral or cognitive interventions would generate greater enthusiasm in the field. In part because its advocates were more likely to conduct comparisons to medications in psychiatric settings with real patient populations, cognitive therapy gained the upper hand. Nonetheless, it is remarkable how few studies there have been comparing behavioral and cognitive interventions, or just how few studies there have been that have adequately implemented any behavioral intervention at all. I think what Neil and his colleagues have done is to revitalize behavior therapy for depression. Given that the prior loss of interest was as much a sociological process as an empirical one, I think that this revitalization is a very welcome development. I am skeptical that behavior therapy will prove to be superior to cognitive therapy (or drugs) in the treatment of depression, and I am particularly skeptical that behavior therapy (or drugs) will prove to even equal cognitive therapy in the prevention of subsequent relapse or recurrence. Nonetheless, the preliminary data from Neil's current study are quite impressive. He has some talented colleagues such as Robert Kohlenberg, Michael Addis, Christopher Martel, and Sona Dimidijian who stand ready to join Keith Dobson in continuing the development of a more purely behavioral approach. I very much hope that they have a chance to complete this work. It could represent a most important legacy.

References