Nearly 40 years ago, Grinker and his colleagues (1) published a follow-up study on a series of borderline patients who had been hospitalized in Chicago. They concluded that such patients remain stably unstable rather than traversing the deteriorating course more typical of those suffering from schizophrenia. Their findings were appropriated by some to reinforce the clinical impression that these patients were essentially untreatable or, at the very least, carried a guarded prognosis. Indeed, early in my career I was asked to present a workshop on the treatment of borderline personality disorder. The workshop organizer wished to publicize my workshop with the following title: “Borderline Personality Disorder: The New Chronic Patient.” I objected to the title, asserting that such patients, while difficult, were nevertheless treatable.
In the decade following the publication of Grinker’s work, Kernberg and colleagues (2) analyzed the data from the Menninger Foundation Psychotherapy Research Project and concluded that borderline patients treated by skilled therapists who focused their interventions on the transference showed a significantly better outcome than those treated with a more supportive approach. The data on which this conclusion was based did not stem from a randomized controlled trial, and skeptics remained unconvinced. Even those who were favorably disposed to psychoanalytic psychotherapy shook their heads as they contemplated the challenges of designing a randomized controlled trial that would demonstrate the efficacy of a psychoanalytic approach. In essence, they were saying it couldn’t be done. The dropout rate would be high, the therapy could not be manualized, suitable control treatments could not be found, and funding sources were unlikely to emerge.
In this issue, the report by Clarkin et al. defies the naysayers and establishes Kernberg’s brand of psychoanalytic object relations therapy as a treatment that produces substantial change in a 12-month period when applied to persons with borderline personality disorder. In this study, 90 patients were randomly assigned to one of three treatment groups: 1) transference-focused psychotherapy, 2) dialectical behavior therapy, and 3) supportive psychotherapy. Six domains of outcome measures were assessed at 4-month intervals over a 12-month period by raters blind to treatment group.
When the results were analyzed using individual growth curve analysis, all three treatments appeared to have brought about positive change in multiple domains to a roughly equivalent extent. However, in some areas, transference-focused psychotherapy seemed to do better than the alternative treatments. In fact, transference-focused psychotherapy was associated with significant improvement in 10 of the 12 variables across the six symptom domains, compared with improvement in six variables with supportive psychotherapy and five with dialectical behavior therapy. Only transference-focused psychotherapy brought about significant changes in impulsivity, irritability, verbal assault, and direct assault. Both transference-focused psychotherapy and dialectical behavior therapy—therapies that specifically target suicidal behaviors—did better than supportive psychotherapy in reducing suicidality.
In a recently published report on a different dimension of these findings, Levy et al. (3) demonstrated that transference-focused psychotherapy produced additional improvements that were not found with either dialectical behavior therapy or supportive psychotherapy. Participants who received transference-focused psychotherapy were more likely to move from an insecure attachment classification to a secure one. Moreover, they showed significantly greater changes in mentalizing capacity (measured by reflective functioning) and in narrative coherence compared with those in other groups. Problems in mentalization (a capacity to attribute independent mental states to the self and others in order to explain and predict behavior) have been identified as a specific area of psychopathology in borderline personality disorder, and an empirically validated treatment, mentalization-based therapy, has been designed to address it (4). We now have suggestive evidence that other therapeutic approaches may also have beneficial effects on the capacity to mentalize.
The design of the randomized controlled trial used in the Clarkin et al. study has particular significance to the field. Elsewhere my colleagues and I (5) have discussed the thorny problem of finding suitable control treatments for rigorous study of long-term psychoanalytic psychotherapy; we suggested a hierarchy of controls according to the rigor of the design. Clarkin and colleagues’ reliance on alternative extended treatments as controls instead of treatment as usual in the community or a waiting list places their study design at the very top of the hierarchy. Moreover, the therapists conducting the other therapies in the study were well-trained, experienced professionals convinced that their type of therapy was effective. Psychotherapy research has long been cursed by designs in which a therapy that is expected to work is compared with one that is expected to fail.
Another bane of psychotherapy research has been the problem of generalizability of the findings. Efficacy trials in academic settings often exclude a great many potential patients and provide a service to patients that is generally not available in the community. In this study, Clarkin et al. combined features of both efficacy and effectiveness studies; they approximated community standards for patients with borderline personality disorder by using exclusion criteria that were typical of clinical practice and by arranging for the patients to be seen by community practitioners in private offices.
So what can we conclude from their findings? When the findings of the National Institute of Mental Health’s Treatment of Depression Collaborative Research Program were published in 1989 (6), Daniel X. Freedman wrote an accompanying editorial (7) in which he cautioned that a “horse race” mentality should be avoided when evaluating the differences in outcome among interpersonal therapy, cognitive behavior therapy, imipramine, and placebo plus clinical management. This cautionary note applies here as well. However, we can anticipate that it will be difficult to avoid thinking in such terms, especially in an era when resources are limited and competing treatments are vying for respect. It is possible, of course, that transference-focused psychotherapy simply has broader effects than other therapies used in treating borderline personality disorder. On the other hand, these differences might disappear when the study is replicated with larger Ns in each treatment arm. Moreover, a recently published study comparing transference-focused psychotherapy and schema-focused therapy (8) suggested that the latter had advantages over the former. Yet another possibility is that all roads lead to Rome—or at least a suburb in the vicinity of Rome. Could it be that any thoughtful, systematic approach to borderline personality disorder, based on our knowledge of the disorder, is potentially helpful, whatever its theoretical underpinnings or technical approach? In this view, the findings would be regarded as confirming the notion that most therapies work through nonspecific effects. An alternative hypothesis would be that there are differential effects of these three treatments and that this study foreshadows a time when treatments may be tailored to specific clinical constellations presented by patients with borderline personality disorder. Far more research is needed to test that hypothesis.
Whichever of these hypotheses ultimately holds sway, Clarkin et al. have provided the field with a landmark study. At the very least, psychotherapists and patients alike can take heart in the knowledge that borderline personality disorder is a treatable condition. In addition, clinicians everywhere who practice long-term psychoanalytic psychotherapy can draw reassurance from this work that their favored treatment approach can be tested with rigorous scientific methods.