Comparative Effectiveness of Psychodynamic Psychotherapy and Cognitive-Behavioral Therapy: It’s About Time, and What’s Next?

Since depression is one of the world’s greatest public health problems, conducting research to accurately weigh the benefits and risks of commonly used interventions should be as much a research priority as developing novel treatments or investigating mechanisms of disease pathophysiology. Psychotherapy is one of the most widely used classes of treatment, but unfortunately there is no commercial entity analogous to the pharmaceutical industry to support research and development of the current and next generations of interventions. The impact of this state of affairs is particularly evident with respect to the ability to conduct larger-scale studies of comparative treatment effectiveness, for which there are only a handful of relevant studies. Thus, although psychodynamic psychotherapy has been used to treat depressed outpatients for decades, the utility of this time-honored approach, as measured by the results of randomized controlled trials of treatment efficacy and effectiveness, has not been extensively studied. The study by Driessen et al. (1) in this issue of the Journal is therefore noteworthy because it provides some of the strongest evidence to date that short-term psychodynamic psychotherapy is an effective treatment for major depressive disorder.

The study, in which 341 outpatients seeking treatment for major depressive disorder at three different psychiatric centers in Amsterdam were randomly assigned, is the largest randomized controlled trial of psychodynamic psychotherapy ever conducted. It is a relatively inclusive study conducted under real-world conditions. The study had two phases, a 16-session (22 weeks) acute phase contrasting psychodynamic psychotherapy against a standard comparator with established efficacy, cognitive-behavioral therapy (CBT), and a 1-year naturalistic follow-up. Severely depressed patients, who comprised about 40% of the sample, could receive concomitant antidepressant pharmacotherapy. The 93 therapists were practicing psychologists and psychiatrists with an average of ≥7 years posttraining experience. For the purposes of the study, they completed standardized training in one of the two modalities (presumably based on their interest) and subsequently were supervised by members of the respective national associations for psychodynamic psychotherapy and CBT. Outcomes were assessed by independent (albeit unblinded) evaluators using standard measures, including the Hamilton Depression Rating Scale (HAM-D). The primary outcome of interest was the posttreatment remission rate, defined as a final HAM-D score ≤7. Although the investigators included all randomly assigned patients in analyses of symptom change, only 233 of the 341 randomly assigned patients (68%) were counted in the primary analysis.

The primary finding of this trial was that psychodynamic psychotherapy was noninferior to CBT; posttreatment score remission rates were 21% (26/122) and 24% (27/111) for the psychodynamic psychotherapy and CBT groups, respectively. No significant differences were seen between treatments on any measure at any time point, and the overall pattern of results generally followed the primary outcome, namely that psychodynamic psychotherapy was not inferior to CBT.

The statistical concept of noninferiority is not exactly intuitive, although in a sense it is the converse of using a statistical test to assert that one treatment is superior to another. Importantly, it is a much different result than one that states that the difference between two treatments is “not statistically significantly different.” This is because a noninferiority trial has been planned from the outset to test the hypothesis that, with at least 95% certainty, the effect of one treatment falls within a narrow, predetermined margin of the other, more established treatment’s effect. In the Driessen et al. study, the boundaries, which were determined by expert consensus, were a 10% difference in remission rates and a 2.6-point difference on HAM-D scores. Although one might quibble with the generosity of these boundaries (i.e., noninferiority studies should use the narrowest boundaries practicable), there is no argument that the observed between-group differences at all time points, on all key dependent measures and on most secondary analyses of subsets of patients, were not clinically meaningful. On the basis of these findings, there is no reason to believe that psychodynamic psychotherapy is a less effective treatment of major depressive disorder than CBT.

This large study greatly expands the literature on controlled studies of psychodynamic psychotherapy, which heretofore was relatively meager. Of course, conducting randomized controlled trials was not part of the culture of academic psychiatry when psychodynamic psychotherapy was the dominant model of psychotherapy, and, in contrast to the strategy taken by the developers of “second-wave” interventions such as cognitive therapy, behavior therapy, and interpersonal psychotherapy, for which there has been evidence of efficacy from randomized controlled trials since the 1970s (2–4), the traditions of psychodynamic training and practice placed little emphasis on the grouped data generated by randomized controlled trials, instead emphasizing clinical observations from individual cases. It was indeed a curious circumstance that, for several decades, there was far more evidence that the newer forms of psychotherapy were efficacious than there was for the older and more widely practiced one.

The absence of evidence (from randomized controlled trials) is not necessarily the evidence of absence (of efficacy for treatment of major depressive disorder); however, the Driessen et al. study adds to a slowly growing body of work that indicates that psychodynamic psychotherapy is indeed an effective treatment option for outpatients with major depressive disorder. The results of this randomized controlled trial complement those of a relatively recent meta-analysis (5), which also was led by Driessen. This meta-analysis, which included both controlled and uncontrolled studies, generally supported the efficacy of psychodynamic psychotherapy (i.e., clinically meaningful differences in comparisons versus minimal or no treatment conditions), although the results of a subanalysis examining 13 randomized controlled trials that used “other psychotherapies” as an active comparison group suggested that psychodynamic psychotherapy may be less effective than other forms of therapy. Because all but three of these comparisons involved various cognitive and behavioral therapies, one might conclude that the results of the Driessen et al. prospective study (1) are at variance with those from the meta-analysis (5). As such, one might wonder which result is stronger and/or more believable: the one based on a single large-scale study or the one derived from a meta-analysis of a number of smaller studies?

The likely answer to this question is that findings may accurately describe different aspects of the treatment research landscape. Specifically, in the prospective study, great efforts were taken to ensure that the two therapies were delivered with the same expertise and with comparable expectations of benefit. By contrast, some of the smaller studies included in the meta-analysis used a psychodynamic psychotherapy group as an active comparison group, which at times can be perceived by clinicians and patients alike as sort of a psychological “brand X” (6). The advantage of the other therapies in the meta-analysis was modest (an effect size of 0.3) and in the range of the so-called allegiance effect, in which the expertise (and, in all likelihood, the expectations, enthusiasm, and interpretive biases) of the principal investigator(s) has been shown to predict the outcome of the study (6). Thus, when compared on a level playing field, the two forms of therapy may be comparably useful for treatment of depressed outpatients, both singly and in combination with antidepressants.

From another vantage point, whereas Driessen et al. demonstrated that psychodynamic psychotherapy was not inferior to CBT, they also showed that the outcomes of depressed outpatients were far from ideal, even when receiving good treatments from capable therapists. Indeed, the outcomes of both psychotherapy groups are strikingly comparable to those observed in the CBT arms included in the second level of the Sequenced Treatment Alternatives to Relieve Depression study (7), which likewise was an inclusive, multicenter study aimed at evaluating comparative effectiveness under real-world conditions. Since many clinicians may have already believed that the findings of Driessen et al. were true (i.e., the two therapies are comparably effective), perhaps the more important finding of this study is to underscore the harsh reality that we still need more effective treatments for major depressive disorder, and this need is as true for psychotherapy as it is for pharmacotherapy.