by Michael D. Anestis, M.S.
Yesterday brought a brief reprieve from my current obsession with talking about the evidence supporting the idea that particular treatments have been shown to be more useful than others for specific disorders and that common factors (e.g., therapeutic alliance, rationale provision) are not the underlying mechanisms of change in therapy. For those of you who missed that reprieve, I recommend reading Joye's great article from yesterday on whether or not Night Eating Syndrome should be included in DSM-V. For those of you who are new to this topic, which we have been covering extensively on PBB for the past several days, here is a quick recap:
- There is substantial evidence that particular forms of therapy (both psychosocial approaches and pharmacological approaches) work better than others for specific mental illnesses. Such treatments are referred to as empirically supported treatments (ESTs).
- Despite this evidence, many psychologists do not use the treatments supported by scientific evidence
- Much of the rationale given for ignoring the evidence relies on a misunderstanding of how to interpret data (e.g., "all individuals are different, so group data is not useful in my practice") or an affinity for poorly designed studies that mistakenly conclude all treatments are equal because no single treatment is better at treating everything than any other treatment (e.g., the Dodo Bird hypothesis)
- Some individuals who resist scientific evidence claim that common factors - the aspects of therapy that are there regardless of which form of therapy is used - such as therapeutic alliance are what really matters; however, the evidence for this claim is full of substantial holes
We have covered these topics extensively on PBB and will continue to do so and, if you have not read and shared your thoughts on those articles, I hope you will do so regardless of whether or not you agree with the conclusions.
Today, I would like to further the conversation by discussing two studies. The first, recommended by Dr. Julie Buckner of Louisiana State University (whose work we have sited on PBB numerous times) in the comment section of our recent article on therapeutic alliance, covers the relative importance of therapeutic alliance versus homework completion in treatment outcomes for individuals with cocaine dependence. The second is a meta-analysis conducted by Jedidiah Siev and the highly influential Dianne Chambless of the University of Pennsylvania that demonstrates that, even within cognitive-behavioral therapy (CBT), different forms of the treatment yield different results for different mental illnesses, refuting the claim that common factors drive outcomes.
Study #1: Carrol, Nich, & Ball (2005)
I'm so glad that Dr.Buckner called my attention to this study, as it does a great job of further clarifying some of the statements made in the article from earlier this week. In this paper, Kathleen Carroll, Charla Nich, and Samuel Ball of the Yale University School of Medicine wanted to determine whether particular aspects of cognitive-behavioral therapy (CBT) for cocaine dependence would predict treatment outcome. In other words, do certain aspects of what goes on during therapy help determine whether or not an individual will ultimately respond positively to treatment?
Because CBT generally relies heavily upon homework assignments in which the client practices skills learned in session throughout the week in order to ensure that gains generalize to his or her whole life, Carroll and colleagues (2005) were most interested in determining whether or not actually completing homework assignments was predictive of outcome when they statistically accounted for the influence of other relevant variables (e.g., severity of illness).
This particular study was part of a broader project in which 121 individuals who met criteria for cocaine dependence were randomized to receive either CBT or interpersonal psychotherapy (IPT) as well as either disulfiram or placebo over a 12 week period. Because the IPT protocol did not involve homework assignments, the authors only looked at the 60 individuals assigned to CBT in their analyses. 50 of those participants were assigned homework, 43 of whom attempted or completed homework and 26 of whom completed homework once or more during treatment. Homework was assigned during 72% of sessions, attempted or completed in 48% of sessions, and fully completed in 24% of sessions. The therapists involved in the study did not differ from one another with respect to the number of times they assigned homework, the percentage of clients who completed homework, or in the overall percentage of sessions in which homework was completed. Clients were assessed prior to treatment, weekly throughout treatment, at the 12-week termination point, and once every three months through the year after therapy was completed. All evaluations were conducted by independent evaluators unaware of whether the client had been assigned to CBT (and had thus been assigned homework throughout treatment).
So what did they find? First of all, homework completion was associated with greater increases in the quantity and quality of coping skills as well as significantly less cocaine use during treatment and throughout follow-up. In other words, those who completed their homework improve substantially more than did those who did not.
More relevant to the topics we have been discussing, however, is the fact that homework completion was not correlated with therapeutic alliance as rated by the client. In other words, whether or not a client completed homework, which in and of itself predicted a stronger outcome, was not related to whether or not the client felt that he or she had a strong alliance with the therapist. Homework completion was, however, associated with the therapist's view of alliance. Now, I suppose it is possible that somebody could look at those results and say that, because the therapist felt that he or she had a better alliance with the client, he or she was better able to help the client improve. To me, however, it seems as though therapists simply feel better about alliance when the client is willing to do the work, as though the client's willingness to fully engage in treatment is a reflection upon them. The fact that clients' homework completion rate was completely unrelated to homework completion speaks to the idea that what matters in therapy is that the proper treatment be properly implemented, not that both the client and therapist feel as though they have built a strong working relationship with one another. In other words, skill acquisition and its generalization outside of the therapy room is much more important than how we feel about what is going on during a standard 50-minute session.
Study #2 Siev & Chambless (2007)
Let's take this another step further and look at another study that took a different approach to examining whether specific treatments for specific disorders determine treatment outcome through the specific aspects of the treatment that make it unique from others. For those of you interested, Dianne Chambless has done a ton of great work on this topic, so I highly recommend reading her studies for a more thorough understanding of the issues.
Siev and Chambless (2007) opened their article, which was published in the Journal of Consulting and Clinical Psychology, by pointing out a fact we have touched upon regularly on PBB: many opponents of the empirically supported treatments movement point towards studies that have purportedly found equivalence across diagnoses. One of the many problems with those studies, however, is that they use meta-analyses to compare different treatments in general rather than comparing different treatments for particular problems. In other words, they want to see if any one treatment is better than others at treating all mental illnesses. This, of course, is a ridiculous question. Just as we do not use any single medicine to treat all physical ailments, no psychological treatment is used in this manner. As such, the better question would be something along the lines of "which form of therapy produces the best results in the treatment of bulimia nervosa" or some other specific condition.
In an effort to demonstrate their point rather than simply rant about it, Siev and Chambless (2007) conducted their own meta-analysis. In theirs, however, they wanted to see whether two similar but distinct forms of therapy - cognitive therapy (CT) and relaxation therapy (RT) - differed in their utility for treating two related but distinct mental illnesses - panic disorder (PD) and generalized anxiety disorder (GAD). The authors pointed out that, although an absence of a difference between two such similar treatments for two such related disorders would not confirm the idea that common factors are what truly matters and that treatment type is unimportant, significant differences would, in fact, deliver a severe blow to the idea that treatment outcome is determined by factors that are common to all forms of therapy (thereby making all therapies equivalent).
Before discussing the authors' findings, let me provide a quick summary of their definitions of CT and RT. The cognitive model of anxiety says that our anxiety is caused by distorted thoughts that cause us to overestimate the likelihood of an unwanted outcome and the consequences should that outcome occur. As such, CT focuses on challenging anxious thoughts by examining the evidence for and against the client's beliefs. RT, on the other hand, involves teaching the client to engage in relaxation exercises when he or she experiences anxiety, thereby diminishing the physiological symptoms associated with anxiety and decreasing anxious thoughts without directly challenging them. Within the GAD and PD treatment literature, RT is further broken down into progressive relaxation and applied relaxation and if you would like to learn more about these specific subtypes, I recommend reading the original Siev & Chambless (2007) article.
Okay...on to the findings. For GAD, the authors found no difference between CT and RT in the reduction of generalized anxiety, anxiety-related cognitions, or depression. Neither treatment produced greater clinically significant change or exhibited higher levels of client drop-out. In other words, both treatments produced relatively strong results, but neither stood out above the other on any of the outcome measures.
For PD, on the other hand, the results were quite different. Individuals with PD who received CBT fared significantly better than did individuals with PD who received RT. PD clients who received CBT experienced fewer panic symptoms, had less fear of anxiety, and demonstrated greater improvement with respect to panic-related cognitions at post-treatment. Across all of the studies analyzed by Siev and Chambless (2007), 77% of PD individuals who received CBT were panic free post-treatment whereas only 53% of PD individuals who received RT were panic free. 72% of PD individuals who received CBT exhibited clinically significant change as compared to 50% who received RT. The two treatments did not differ on dropout rates.
So what do all of these findings tell us? If a clinician has a client who meets criteria for GAD, than CBT and RT appear to be equally viable choices for treatment. On the other hand, if a clinician has a client who meets criteria for PD, CBT is a better choice than RT. Does this mean that everyone will respond optimally to CBT and nobody would respond that way to RT? Of course not. It simply means that, on average, CBT produces better results in PD than does RT and, as such, the odds are greater that a PD client will respond to treatment if he or she receives CBT. If treatment response is not in line with these results, than switching to RT would make sense.
But what about common factors? Well, Siev and Chambless (2007) certainly can not dismiss common factors as a prominent issue in the GAD findings. No evidence emerged demonstrating a relationship between common factors and outcome; however no evidence emerged that disputed that possibility (or a limitless number of other possibilities). Overall, our treatment results for GAD lag behind those of other anxiety disorders, so it may very well be that common factors are key there at this point because we have not yet designed a particularly great treatment for GAD. For PD, on the other hand, common factors simply can not be said to dictate the results. Clearly, it would have been better if such variables had been directly measured and controlled for; however, the fact that two fairly similar treatments produced substantially different results for a specific disorder implies that there is something about the particular variation in the treatment protocol that contributes to the greater outcomes in CBT relative to RT.
Overall Conclusion
In both of these studies, authors took a look at various factors that influence whether a client will respond to psychotherapy. Again, as we have been demonstrating repeatedly over the past several days, the evidence, while imperfect, appears much more supportive of the idea that specific treatments are particularly helpful for particular diagnoses than for the notion that all improvement in therapy is accounted for by common factors such as therapeutic alliance.
Also, as I have said before, dismissing common factors as the primary mechanism of change in therapy does not equate to saying they are meaningless. Certainly, everyone benefits if both the client and therapist believe that a strong alliance is in place. At the same time, the data simply do not support the idea that what makes therapy work is simply providing a healing environment and a strong therapist-client relationship. If that were the case, Rogerian therapy would be a panacea for all mental health concerns.
If you would like to learn more about cognitive behavioral therapy, we recommend the following items, all of which are available through our online store of scientifically-based psychological resources:
- Cognitive Therapy of Depression
by Aaron Beck, John Rush, Brian Shaw, and Gary Emery
- Cognitive Therapy of Personality Disorders
by Aaron Beck, Arthur Freeman, and Denise Davis
- Cognitive Therapy of Substance Abuse
by Aaron Beck, Fred Wright, Corey Newman, and Bruce Liese
- Cognitive Therapy of Anxiety Disorders: Science and Practice
by David Clark and Aaron Beck
- Cognitive Therapy Techniques: A Practitioner's Guide
by Robert Leahy
- Treatment Plans and Interventions for Depression and Anxiety Disorders
by Robert Leahy
Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University
