The current issue of Acta Psychiatrica Scandinavica features an article by Barnicot et al. (1) investigating the assertion that patients with Borderline Personality Disorder (BPD) have significant dropout rates, which interfere with obtaining adequate and effective treatment. The authors concluded that there have been high completion rates in a number of BPD treatment studies, although it is still not clear what factors determine dropout. In light of the persistent idea among many that BPD is not treatable, it is important to have data showing that many patients with BPD do remain in treatment and experience benefits from it. However, when reflecting on this article, I was struck by how the meta-analysis mirrors in certain ways an unfortunate bias that has taken over our current approach to studying and treating BPD (and in the mental health field in general): the extensive promotion of so-called evidence-based treatments. Bandicot et al. (1) chose therapies with ‘effectiveness’ defined as a ‘demonstration in at least one randomized controlled trial (RCT) that the treatment was effective in improving one or more of the symptoms of BPD as defined by DSM-IV, compared with treatment as usual or another psychotherapy’. This implies that the only effective treatments are manualized – mostly short-term –‘evidence-based’ approaches that have been studied in RCTs. As Wachtel (2) has recently asserted: ‘In contrast to the appropriate concern with examining the evidence for the efficacy of various approaches to therapy and for the theoretical assumptions that underlie them, the “empirically supported treatments” movement has been characterized more by ideology and faulty assumptions than good science’ (p. 251).

First, let us consider the concept of ‘effectiveness’. The idea persists among those who promote an RCT-driven approach that because something has merely been shown to be efficacious in a well-controlled experiment, or sometimes even because it has been studied at all, we can conclude there is sufficient evidence it is effective for everyone with a particular diagnosis. In the Barnicot et al. article (1), the bar has been set quite low for what constitutes an effect: improvement in as little as one symptom is deemed adequate. A significant problem for many of the manualized approaches, as they are studied and often practiced, is that they are not of sufficient duration to work through the complex problems associated with BPD (some of these treatments make provisions for longer intervals, but that is rarely considered). Second, with the exception of Mentalization-based Treatment, few long-term follow-up studies have been conducted to document whether gains have been maintained (this is also true for short-term manualized treatments for other disorders). In some cases, such as with Dialectical Behavior Therapy (DBT), at the end of follow-up periods of 6 months or 1 year, many of the benefits have been lost (3). Worse yet, the pressure to adopt ‘evidence-based’ treatments for BPD has resulted in a proliferation of programs and practices with insufficient resources or training to conduct properly all of the components of the models tested in RCTs.

The insistence on RCT as the gold standard for studying psychotherapy raises a number of other issues. I will mention just a few. Advocates of using the RCT approach to study psychotherapy would have us believe that in the name of good science, it is necessary to mimic the double-blind methodology employed in drug studies (2). However, it is impossible to be blind in a study of psychotherapy. It is interesting to note that in studies comparing manualized treatments, very often, the one most favored by the authors tends to come out looking better than the others.

Then, there is the issue of the ‘treatment-as-usual’ control group. Most clinicians are not adequately trained to treat BPD. Thus, in many cases, using treatment-as-usual as the comparison group is like a race between someone who carbo-loaded the night before and a person who hasn’t eaten in 3 days.

There are 256 ways to meet DSM-IV criteria for BPD. This implies that therapists should be trained to respond flexibly and creatively to the particular and often myriad needs of a given patient and not be confined by the dictates of a manual (2). Are patients with BPD not entitled to something better than the amelioration of one symptom that may or may not remain improved over time after treatment ends? Those of us who have been trained using long-term intensive approaches work as long as it takes to help the patient find a productive life with minimal suffering. Furthermore, increasing evidence has accumulated that long-term psychodynamic approaches are very effective over time and show robust effect sizes (4, 5).

Manualized treatments have largely been developed as such to pursue the RCT chimera. There are other research methods that are better suited to studying psychotherapy (2). Contrived horse races between competing treatments that arbitrarily end should be replaced with better efforts to study what works for which kind of patient and why, and how long does it really take? Rather than continue to proliferate short-term treatments that promise too much and deliver too little (except to the insurance companies), perhaps we should be truthful about the time and effort required to address adequately the serious problems at the heart of BPD, and demand the resources and training needed to get the job – and more appropriate research – done.


  1. Top of page
  2. References