Psychotherapies for borderline personality disorder: DBT and psychodynamic approaches do best


As the authors of a recent systematic review (Cristea et al., 2017) and an accompanying editorial (Fonagy, Luyten, & Bateman, 2017) comment: Borderline Personality Disorder (BPD) is generally recognised as being a highly disruptive form of personality disorder, characterised by significant mental distress during times of crisis, as well as having a long-standing impact on social functioning that often responds poorly to support and treatment (Zanarini, Frankenburg, Reich, & Fitzmaurice, 2012).

BPD is often taken as being the archetypal representation for ‘personality disorder’ more generally, a group of disorders traditionally associated with feelings of therapeutic nihilism (Lewis & Appleby, 1988); although the development of psychotherapeutic techniques in the past 30 years has to some extent addressed this concern (Bateman & Fonagy, 2010; Linehan, 1987).

Systematic study into the efficacy of psychotherapy for individuals diagnosed with BPD is therefore important for its ability to clarify understanding, and identify areas for future research in this field.


The authors of the current review therefore sought to identify randomised control trials (RCTs) and to consider, through meta-analysis:

“…the efficacy of psychotherapies for BPD-relevant outcomes at post-test and, where possible, at follow-up.”

Borderline Personality Disorder can bring significant mental distress and result in poor social functioning.

Borderline Personality Disorder can bring significant mental distress and result in poor social functioning.


The authors conducted a systematic search of electronic databases, using the terms ‘borderline personality’ and a restriction to limit identified studies to those labelled as ‘randomised trials.’

Inclusion criteria

  • Studies were included if they were identified as RCTs comparing psychotherapy with a control condition for adults diagnosed with BPD
  • Control groups could consist of ‘treatment as usual’ (TAU), or other treatments not specifically for BPD
    • Head-to-head comparisons between two therapies specifically designed for BPD were excluded
  • Medication use in both control and intervention arms was allowed, so long as this was not a structured intervention within the identified trial.

Appraisal for risk of bias

  • Risk of bias was appraised according to four domains from the Cochrane Collaboration Risk of Bias Tool
  • A ‘risk of bias’ score was computed with a study being awarded a point for each domain being deemed at ‘low’ risk of bias.

Study characteristics

  • Studies were identified as ‘stand alone’ (where comparison was made directly between the intervention and control group) and ‘add on’ (both groups received TAU with an additional intervention for the experimental group)
  • Therapy type was classified as:
    • DBT (Dialectical behaviour therapy)
    • Psychodynamic
    • CBT (Cognitive behavioural therapy)
    • Other
  • Control groups were classified as TAU, Supportive Therapy, ad hoc control (designed as part of the trial) and it was also noted whether control groups were ‘manualised’ or not
  • Any involvement of the study team in the control group was noted.


  • Outcome measures were classified as:
    1. Borderline Relevant (BPD Symptoms, self-harm and parasuicide, or suicidal behaviours)
    2. Symptom measures
    3. Health service use
    4. General psychopathology (measures of anxiety and depression in particular)
  • Differences between intervention and control groups were calculated at two points
    • Post-test
    • At later follow up (up to 2 years)
  • Effect sizes were calculated through the use of Hedges’ g
  • Heterogeneity (difference between study findings) was appraised.


  • The authors identified 33 trials meeting their inclusion criteria
    • Only 28 of these trials reported sufficient data to allow the calculation of effect sizes. Authors of the missing 5 trials did not respond to requests for access to data
    • A total of 2,256 participants were included across the trials (1,169 intervention, 1,087 control)
    • 12 trials included only women as participants
    • Treatment duration varied between 2.5 and 24 months with session numbers varying between 6 and 312.
    • Only 11 trials were rated as being at low risk of bias on 3 out of the 4 considered domains
  • Combining all trial types produced a moderate effect for BPD relevant outcomes (g=0.35 95% CI 0.20 to 0.50) at post-test with a moderate degree of heterogeneity to findings (48%)
  • No difference in treatment retention was identified between treatment and control groups
  • At follow up (up to 2 years) 2 deaths by suicide had occurred in the treatment group, versus 5 in the control group
  • Combining design types at follow up again produced a moderate effect in favour of psychotherapy (g=0.45 95% CI 0.15 to 0.75), heterogeneity was high for this finding (70%).

Subgroup analyses

  • DBT and psychodynamic therapy approaches were found to be effective versus control, CBT and other treatment types were not
  • Analysis of trials with a specifically developed (ad hoc) control group, manualised control intervention, low risk of bias (on at least 3 of 4 bias measures), or where the study team were directly involved in the control group treatment, all led to a loss in the statistical significance of the treatment effect.

Meta-regression analysis

  • The relationship between risk of bias and effect size was found to be linear in nature, such that as the risk of bias decreased so too did the effect size
  • Length of treatment was not found to correlate with treatment efficacy.

Risk of publication bias

  • Tests for the presence of publication bias suggested that some studies were missing from publication
  • Statistical measures to take account of these putative ‘missing studies’ decreased the effect size at post-test analysis to a ‘small effect’, although the finding remained statistically significant (g=0.23 95% CI 0.07 to 0.38). However, statistical significance was lost at the follow up point of analysis (g = 0.19 95% CI -0.15 to 0.53) when ‘missing’ studies were included in the analysis.
DBT and psychodynamic therapy approaches were found to be effective versus control, CBT and other treatment types were not.

DBT and psychodynamic therapy approaches were found to be effective versus control, CBT and other treatment types were not.


The authors concluded:

Psychotherapies, most notably dialectical behaviour therapy and psychodynamic approaches, are effective for borderline symptoms and related problems. Nonetheless, effects are small, inflated by risk of bias and publication bias, and particularly unstable at follow-up.

As the authors of an accompanying editorial (Fonagy et al., 2017), addressing the findings of this systematic review, comment: the findings of a small to moderate effect size, that is maintained at follow up is reassuring, and indicates that the long standing myth of BPD being ‘untreatable’ is now difficult to sustain.

As for all systematic reviews, the current study has some inherent limitations, specifically relating to the level of abstraction necessary to handle the amount of included data in a robust manner. For example, the accompanying editorial questions the combination of varying treatment modalities, with different therapeutic processes under the heading of ‘psychodynamic therapies.’

However, the study also raises some important points for consideration in future research. For example, as the authors suggest, the findings that treatment efficacy is not maintained in comparison with control group interventions when study team members are involved in the treatment of the control group, or when a manualised control intervention is used, perhaps indicate that there is some specific benefit from the ‘focus’ or structured nature of these different states that benefits people with experiences of distress associated with BPD. Furthermore, the lack of an association between length of treatment and outcome also indicates a need for exploration regarding cost effectiveness, or the ‘necessary components’ for therapeutic efficacy.

High quality, structured reviews, such as this one, are significant not only for the findings they themselves demonstrate, but also for the avenues of enquiry that they open up.

The long-standing myth that borderline personality disorder is untreatable is now difficult to sustain. 

The long-standing myth that borderline personality disorder is untreatable is now difficult to sustain.


Primary paper

Cristea, I. A., Gentili, C., Cotet, C. D., Palomba, D., Barbui, C., & Cuijpers, P. (2017). Efficacy of Psychotherapies for Borderline Personality Disorder. JAMA Psychiatry, 74(4), 319–10. [PubMed abstract]

Other references

Bateman, A., & Fonagy, P. (2010). Mentalization based treatment for borderline personality disorder. World Psychiatry : Official Journal of the World Psychiatric Association (WPA), 9(1), 11–15. [Abstract]

Fonagy, P., Luyten, P., & Bateman, A. (2017). Treating Borderline Personality Disorder With Psychotherapy. JAMA Psychiatry, 74(4), 316–2. [PubMed abstract]

Lewis, G., & Appleby, L. (1988). Personality disorder: the patients psychiatrists dislike. The British Journal of Psychiatry, 153(1), 44–49. [PubMed abstract]

Linehan, M. M. (1987). Dialectical behavior therapy for borderline personality disorder: Theory and method. Bulletin of the Menninger Clinic, 51(3), 261–276. [PubMed abstract]

Zanarini, M. C., Frankenburg, F. R., Reich, D. B., & Fitzmaurice, G. (2012). Attainment and stability of sustained symptomatic remission and recovery among patients with borderline personality disorder and axis II comparison subjects: a 16-year prospective follow-up study. The American Journal of Psychiatry, 169(5), 476–483. [PubMed – Open Access]

Photo credits

Share on Facebook Tweet this on Twitter Share on LinkedIn Share on Google+