It is not universally or perhaps even widely accepted in the community of psychotherapy researchers that cognitive therapy works primarily through the use of methods that are directed at changing depressed persons’ thinking. Several trends within clinical psychology, including the “psychotherapy integration” movement, have led to a greater focus on the interpersonal relationship between patient and therapist as the critical factor in the successful relief of symptoms engendered by the psychotherapeutic interaction. A view that has become prominent is that the “helping alliance” between the patient and the therapist is the most important element in psychotherapy, irrespective of the particular aims of the therapy. Although researchers and theorists have not been as specific as Carl Rogers was when he posed a set of “necessary and sufficient conditions” for therapeutic change, the tone of the discussion in the literature suggests that the helping alliance is a major causal factor in all successful psychotherapeutic change. This has been coupled with the view that the particular theory-specific methods of psychotherapies are of relatively little use. Adherents of these views point to findings that: (a) high levels of helping alliance are associated with therapeutic benefit across different types of psychotherapies; and (b) attempts to demonstrate the superiority of one kind of psychotherapy over another have yielded few successes. They conclude that their views have been supported by these findings.
We (DeRubeis, Brotman, & Gibbons, 2005) recently published a paper in which we question these inferences. We noted that researchers have assumed that correlations between symptom change scores and scores on measures of the helping alliance reflect the causal influence of the alliance on symptom reduction, rather than the other way around. We have argued that correlations between these two sets of variables may reflect the fact that symptom reduction leads to a better alliance. My students and I have conducted several studies to test the importance to outcome of variations in therapist behavior, as well as variations in the quality of the helping alliance. When we eliminated this temporal confound in two studies of the relation between the alliance and outcome (DeRubeis & Feeley, 1990; Feeley, DeRubeis, & Gelfand, 1999), our findings were consistent with the hypothesis that symptom reduction improves the alliance, rather than vice-versa. Moreover, we found that therapist’s adherence to the methods of cognitive therapy predicted symptom reduction, suggesting that greater adherence produces better outcome. Former graduate students Melissa Brotman and Daniel Strunk also have found support for these hypotheses in a research effort that employed more sophisticated methods and larger sample sizes than our earlier efforts.
My former student Tony Tang and I (Tang & DeRubeis, 1999) found that sudden, large reductions in symptoms occurred during the course of cognitive therapy in about 40% of cases, and in about 55% of those who improved overall. These “sudden gains” were frequently followed by substantial improvements in the helping alliance. Once again, support was obtained for a causal relation between symptom reduction and the helping alliance that is the reverse of that currently favored by many psychotherapy researchers. Moreover, we found that these sudden improvements were preceded by particularly productive therapy sessions, in which patients voiced substantial changes in their beliefs about themselves and their world. Findings from this study, as well as a replication of it that was just recently published (Tang, DeRubeis, Beberman, & Pham, 2005), suggest that theory-specific aspects of cognitive therapy are indeed important in the reduction of depressive symptoms.
My students and I continue to conduct studies of the process of change in cognitive therapy. The best context for this kind of work is in a well-controlled and intensively observed outcome study. We are especially interested in the nature of the therapist behaviors – and the psychological changes in patients those behaviors produce – that might be responsible for the short-term (relapse) and long-term (recurrence) prevention effects we are finding in our outcome research. We also continue our attempts to characterize the psychological benefits that accrue to patients who improve in cognitive therapy, especially those benefits that do not accrue to patients who improve with medications. We believe that there are very likely to be such benefits because of the evidence that short-term cognitive therapy produces greater resistance to relapse than does short-term medication therapy. Thus, variables that mediate this relapse resistance effect are of particular interest.