Complexities of Choosing Effective Treatment

Starting in 1998, the Personality Disorders Institute at Weill Cornell Medical College's Westchester Division, led by Otto Kernberg and John Clarkin, conducted a randomized controlled trial (RCT) comparing three treatments for borderline personality disorder (BPD). The results of this trial, published in the June 2007 issue of the American Journal of Psychiatry, showed that at the end of one year all three treatments—Kernberg's transference-focused psychotherapy (TFP), Marsha Linehan's dialectical behavior therapy (DBT), and a dynamic supportive psychotherapy (SPT), whose design was led by psychoanalyst Anne Applebaum—fostered similar improvements in mood, global functioning, and social adjustment. However, TFP did better than SPT in improving suicidality and better than DBT in its effect on anger and impulsivity. Furthermore, as reported by Kenneth Levy and colleagues in the December 2006 Journal of Consulting and Clinical Psychology, TFP alone led to a significant increase in the number of patients classified as “secure” with respect to attachment states of mind and significant improvement in Reflective Function (RF), as defined by Peter Fonagy and Mary Target's rating scale.

Even before the start of the Cornell West-chester trial, a group of psychotherapy researchers in Holland, led by Arnoud Arntz, initiated a similar RCT, comparing TFP with Jeffrey Young's schema-focused therapy (SFT). Young, who was trained in cognitive behavioral therapy by founder Aaron Beck, has developed a version specialized for BPD in which therapist and patient systematically identify recurrent and pervasive patterns in a patient's thoughts, feelings, and behaviors and discuss strategies for dealing with them. SFT labels four schemas that distinguish BPD: “detached protector,” “punitive parent,” “abandoned/abused child,” and “angry/impulsive child.” The results of the trial, published in the June 2006 issue of the Archives of General Psychiatry by Arntz, Giesen-Bloo, and colleagues, suggest that patients randomly assigned to one year of SFT are more likely to symptomatically improve than those assigned to TFP. A major factor in this difference appears to have been that patients who underwent TFP in this study were significantly more likely to prematurely drop out of their treatment. Arntz and colleagues reported the results of an “intent-to-treat analysis.” They compared the outcome of all patients assigned to TFP with that of patients assigned to SFT, including patients who completed the assigned treatment as well as those who terminated prematurely.

The story continues. The May 2007 Archives of General Psychiatry carried a cautionary letter to the editor from Frank Yeomans, one of Kernberg's co-investigators and a consultant on the Dutch study who traveled to Holland eight times (over the seven years of the study) for short visits to train the TFP therapists. He points to several problems with the reported results. First, he notes that analyzing only those patients who completed each treatment would likely show TFP and SFT to be more comparable than the authors suggest. Second, he observes that the authors' own data about therapist “adherence” (the extent to which a particular treatment is being done according to a predetermined set of criteria, as rated from therapy videotapes by independent observers) show that TFP therapists in their study were less skilled than SFT therapists. Yeomans points out that this confirmed his own concerns, previously expressed to the study's authors, that their training in TFP was too brief for them to be doing it well and that they had little allegiance to TFP as a treatment. The TFP therapists all had previous dynamic training, but none of them were comprehensively trained in techniques specific to BPD.

Arntz and colleagues' response to Yeomans boils down to one statement. “If experienced, valued, and successful therapists after extensive training, ‘piloting,’ and (peer-) supervision are still regarded as limited-adherent or beginner TFP therapists by the developers of TFP,” they argue, “one can wonder how TFP could become widely available as a treatment for borderline personality disorder.”

The ongoing debate contains elements that seem central to psychoanalytic scholarship, training, and the place of both in the wider communities of psychiatry and psychotherapy. What are the techniques that are best suited for patients with borderline psychopathology? Is Kernberg's group correct in promoting a modified version of psychodynamic psychotherapy for BPD in which negative transference and aggression are confronted and interpreted early in the treatment? How does this method compare to those suggested by other psychoanalytic theories, such as self-psychology or Fonagy and Bateman's mentalization based treatment for BPD, and what is the effect of each on premature termination? Are there patients in whom one set of techniques is preferable over others and are there systematic ways in which analysts might apply their own expertise in assessing patients and recommending different types of treatment? How does the individual analyst or therapist decide which technique is appropriate at a given moment in a treatment? How possible is it to manualize these very techniques?

None of the authors treat RCTs as infallible or consider the evidence from any one of these trials as conclusive in a particular direction. Analytic thinkers, most recently Drew Westen and Sidney Blatt, have long pointed out the crucial shortcomings of the RCT as a method for choosing a defined set of “empirically validated treatments.” Above all, they argue, how can we generalize from manuals and randomly assigned groups, when each treatment is so individual and the product of such a highly subjective and intersubjective negotiation. The debate between Yeomans and Arntz itself makes clear how complicated it can be to interpret any one set of data. However, it also suggests that some of the very issues being debated here—training of therapists, identifying techniques to avoid premature dropouts in borderline patients, and defining the goals of treatment—are those confronting psychoanalysis.