I've had time to think about Coffman, et al's (2007) paper and a number of points seem important:
- The sub-group of individuals who did not respond to CT were characterized by severe depression, greater functional impairment and problems with primary support group. Many had major interpersonal problems or major stressors, or both. Most described themselves as having long-standing depressive propensities.
- With the individuals who did not respond to CT, both therapist and client rated the therapeutic alliance less positively as early as the second session and this continued throughout treatment.
- With the individuals who did not respond to CT, both therapists and clients became demoralized.
- The CT therapists did not have access to on-site expert supervision. They often sought expert consultation regarding the individuals who did not respond to CT but there often was a delay in receiving the consultation.
The authors suggest that the CT therapists may have shifted from behavioral interventions to cognitive interventions too soon and it appears that this was the case. When I teach CT (which is quite often) I teach that when treating severely depressed clients one should start with behavioral interventions, begin some of the more concrete cognitive interventions
once the client's mood has improved somewhat and
wait until the client's depression is substantially improved before addressing dysfunctional beliefs and assumptions. None of the individuals who failed to respond to CT had improved enough for it to make sense to shift from behavioral interventions to cognitive interventions yet therapists in the CT condition spent more than half of their time on cognitive strategies.
Also, the authors suggest that many of the individuals who did not respond to CT would have met DSM criteria for an Axis II disorder. However, it appears that the CT therapists were trained in CT for depression
and were not trained in using CT with Axis II disorders. This it particularly salient in light of the observation that problems with the therapeutic alliance were evident with the individuals who did not respond to CT. In CT with personality disorders there is an increased emphasis on dealing with problems with the therapeutic relationship. It appears that the CT therapists in this study did not know how to deal with some of the complexities they encountered.
The authors of this paper make some valuable recommendations for CT therapists who are faced by clients who appear similar to those who did not respond to CT:
- Do not abandon behavioral interventions too soon. If the client is not responding to treatment, spend extra time on behavioral interventions.
- Don't attempt too much in short-term treatment. If the time is limited, don't get into deep or complex issues that you won't have time to resolve.
To these I would add:
3. If there is a poor therapeutic alliance, address this immediately rather than trying to plow ahead anyway.
4. The "Cognitive Therapy for Depression" protocol is designed for uncomplicated unipolar depression. If there are co-morbid disorders (i.e. anxiety disorders, substance abuse, personality disorders, etc.) they probably will need to be addressed as well. That means knowing how to apply CT with anxiety, substance abuse, personality disorders, etc. rather than only knowing CT for depression
This last point highlights one of the drawbacks of CT. It is complex to learn and apply effectively with the wide range of clients one encounters in clinical practice. One of Behavioral Activation's selling points is that it is supposed to be easier to learn than CT. If it does indeed turn out to be easier to learn while remaining equally effective, that would make me a big proponent of BA. However, I must admit that I remain skeptical.
It appears that CT performed more poorly than BA because there was a sub-group of individuals that the CT therapists did not know how to treat effectively. It also appears that simple improvements in how CT is applied would result in much better outcomes for CT with these clients (the authors cite some unpublished data from DeRubeis to support this point). We'll see if this is the case when subsequent research is published.