The British Journal of Psychiatry
Progress in the treatment of borderline personality disorder


Summary We outline recent evidence suggesting that the natural course of borderline personality disorder is more benign than formerly believed. We explore possible reasons for the change in findings which include both the iatrogenic effects of earlier treatment models and the recent availability of effective interventions. Clinicians should be optimistic about improvement and long-term outcomes.

Few areas of psychiatric investigation have seen progress as radical as in the field of personality disorder, particularly the treatment of borderline personality disorder.

The advance in the understanding of borderline personality disorder has been influenced by two developments: (a) the increasing recognition that the disorder has a far more benign course than previously thought; and (b) the emergence of a range of relatively effective psychosocial interventions that appear to accelerate the rate of improvement. Taken together and placed in the context of recent neuroscientific work, these observations suggest new opportunities for the treatment of personality disorder, presenting both opportunities and risks.


Two carefully designed prospective studies have highlighted the inappropriateness of the attitudes that historically confined individuals with severe personality disorder to the margins of healthcare systems (Zanarini et al, 2003; Shea et al, 2004). The majority of patients with borderline personality disorder experience a substantial reduction in their symptoms far earlier than previously assumed. After 6 years, 75% of patients diagnosed with borderline personality disorder severe enough to require hospitalisation, achieve remission by standardised diagnostic criteria. A 50% remission rate has occurred by 4 years but the remission is steady (10-15% per year). Recurrences are rare, perhaps no more than 10% over 6 years. This contrasts with the natural course of many Axis I disorders, such as affective disorder, where improvement may be somewhat more rapid but recurrence is common.


Dialectical behaviour therapy was the first treatment to challenge the atmosphere of therapeutic nihilism. Three randomised controlled trials (for a review, see Lieb et al, 2004) reported significant dramatic reductions in attempted suicide when contrasted with usual treatment (relative risk= -1.38, 95% CI 1.13-1.69). When compared with an active control group the benefit of dialectical behaviour therapy is still evident although less clearly marked.

A promising evidence base is also available for psychodynamically oriented interventions. A randomised controlled trial of treatment of borderline personality disorder in a psychotherapeutically orientated day hospital offering modified individual and group psychoanalytical psychotherapy (Bateman & Fonagy, 1999, 2001) has shown significant and enduring changes in mood states and interpersonal functioning associated with an 18 month programme (effect size= -2.36, 95% CI -3.18 to -1.54). The benefits, relative to usual treatment, were considerable and observed to increase during the follow-up period of 18 months, rather than staying level as with dialectical behaviour therapy.

The Cornell Medical College Group recently reported the only head-to-head comparison of psychodynamic and dialectical-behavioural therapy (Clarkin et al, 2004). They found significant improvements in impulsivity-related symptoms, as well as mood and interpersonal functioning measures. The trial contrasted transference-focused psychotherapy, dialectical behaviour therapy and supportive psychotherapy. There was significant and equal benefit from all the interventions, although early drop-out rates were higher for dialectical behaviour therapy than for the other treatments.

Possible important additional findings concerning hospital treatment include the greater efficacy of briefer periods of hospitalisation, the general ineffectiveness of brief hospital admissions motivated by suicide threats, and the value of combining in-patient admissions with structured psychotherapeutic interventions.


If a range of well-organised and coordinated treatments are effective for borderline personality disorder (and in any case, in the vast majority of cases, borderline personality disorder naturally resolves within 6 years), why have clinicians worldwide traditionally agreed on its treatment-resistant nature? Earlier surveys indicated that 97% of patients with borderline personality disorder who presented for treatment in the USA received out-patient care from an average of six therapists. An analysis of outcomes measured 2-3 years later suggests that such a treatment approach is, at best, only marginally effective (see Lieb et al, 2004). How can we reconcile such findings with what we know of the potential effects of treatment and the new data on the natural course of the disorder? Has the nature of the disorder changed? Have treatments become that much more effective? Both seem unlikely explanations. The known efficacy of pharmacological agents, new and old, cannot account for this difference; the evidence-based psychosocial treatments are not widely available. One possible conclusion is that some psychosocial treatments practised currently and perhaps even more commonly in the past, have impeded the borderline patient's capacity to recover following the natural course of the disorder and prevented them harnessing advantageous changes in social circumstances. In Michael Stone's (1990) classic follow-up of patients treated nearly 40 years ago, a 66% recovery rate was only achieved in 20 years (four times longer than reported in more recent studies). Could the apparent improvement in the course of the disorder be accounted for by harmful treatments being less frequently offered? If so, this change is possibly more a consequence of changing patterns of healthcare than recognition by clinicians of the possibility of iatrogenic deterioration. This suggestion is speculative but requires further consideration.


Pharmacological studies routinely explore the potential harm that a well-intentioned treatment may cause. In the case of psychosocial treatments we all too readily assume that at worst such treatments are inert. However, there may be particular disorders where psychotherapy represents a significant risk to the patient. Whatever the mechanisms of therapeutic change might be, traditional psychotherapeutic approaches depend for their effectiveness on the capacity of the individual to consider their experience of their own mental state alongside its re-presentation by the psychotherapist. The appreciation of the difference between one's own experience of one's mind and that presented by another person is key. It is the integration of one's current experience of mind with the alternative view presented by the psychotherapist that must be at the foundation of a change process. The capacity to understand behaviour in terms of the associated mental states in self and other (the capacity to mentalise) is essential for the achievement of this integration.

Most individuals with no major psychological problems are in a relatively strong position to make productive use of an alternative perspective presented by the psychotherapist. However, those who have a very poor appreciation of their own and others' perception of mind are unlikely to be able to benefit from traditional (particularly insight-oriented) psychological therapies. We have argued that persons with borderline personality disorder have an impoverished model of their own and others' mental function (Bateman & Fonagy, 2004). Their schematic, rigid, sometimes extreme ideas about their own and others' states of mind make them vulnerable to powerful emotional storms and apparently impulsive actions, and create profound problems of behavioural and affect regulation. The weaker an individual's sense of their own subjectivity, the harder it is for them to compare the validity of their own perceptions of the way their mind works with that which a ‘mind expert’ presents. When presented with a coherent view of mental function in the context of psychotherapy, they are not able to compare the picture offered to them with a self-generated model and may all too often accept alternative perspectives uncritically or reject them wholesale.

Any psychological therapy can generate these divergent responses. Both cognitively based and dynamically orientated therapies offer causal explanations for underlying mental states. These can give ready-made answers and provide illusory stability by inducing a process of pseudo-mentalisation in which the patient takes on the explanations without question and makes them his/her own. Conversely, both types of perspective can be summarily and angrily dismissed as overly simplistic and patronising, which in turn fuels a sense of abandonment, feelings of isolation and desperation. Even focusing on how the patient feels can have its dangers. A person who has little capacity to discern the subjective state associated with anger cannot benefit from being told both that they are feeling angry and the underlying cause of that anger. Such an assertion addresses nothing that is known or can be integrated. It can only be accepted as true or rejected outright, but in neither case is it helpful. The dissonance between the patient's inner experience and the perspective given by the therapist, in the context of feelings of attachment to the therapist, leads to bewilderment which in turn leads to instability as the patient attempts to integrate the different views and experiences. Unsurprisingly, this results in more rather than less mental and behavioural disturbance.


So what is the psychiatrist or other mental health professional supposed to do if intervention might induce psychological dysfunction? The problem is compounded by the fact that attachment and mentalisation are loosely coupled systems existing in a state of interactive but partial exclusivity. Recent intriguing neuroscientific findings have highlighted how the activation of the attachment system tends temporarily to inhibit or decouple the normal adult's capacity to mentalise (Bartels & Zeki, 2004). Whereas mentalisation has its roots in the sense of being understood by an attachment figure, it is also more challenging to maintain in the context of an attachment relationship (e.g. the relationship with the therapist) for those individuals whose problem is fundamentally one of attachment. Elsewhere, we have proposed, on the basis of research findings as well as clinical observation, that individuals with borderline personality disorder have hyperactive attachment systems as a result of their history and/or biological predisposition (Fonagy & Bateman, 2006). This may account for their compromised capacities of mentalisation. So if the patient with reduced mentalising forms a significant emotional relationship with the psychiatrist, behavioural and psychological disturbance may be the result.

If this is correct, the recovery of the capacity for mentalisation in the context of attachment relationships has to be a primary objective of all psychosocial treatments for borderline personality disorder. However, patients with borderline personality disorder are particularly vulnerable to side-effects of psychotherapeutic treatments that activate the attachment system. Yet, without activation of the attachment system these patients will never develop a capacity to function psychologically in the context of interpersonal relationships, which is at the core of their problems. So, the mental health professional must tread a precarious path between stimulating a patient's attachment and involvement with treatment while helping them to maintain mentalisation. Treatment will only be effective to the extent that it is able to enhance the patient's mentalising capacities without generating too many negative iatrogenic effects as it stimulates the attachment system. This may be done by encouraging exploration and identification of emotions within multiple contexts, particularly interpersonal ones, and by helping the patient establish meaningful internal representations while avoiding premature conscious and unconscious explanations.

In treatment, the psychiatrist must take an inquisitive stance rather than an expert role, be flexible rather than set unachievable goals about attendance and behaviour, structure treatment in collaboration with the patient, and develop clear pathways to care in a crisis. The patient-psychiatrist relationship needs careful attention if a positive therapeutic alliance is to develop without encouraging overdependence or erotic attachments. As a guide, interventions focusing on the relationship, a necessity if the detail and understanding of mind states is to be explored, should be used only when the attachment system is not excessively stimulated. If things start to go wrong, for example, the patient becomes increasingly aroused and disturbed, the psychiatrist should retrace the interaction, interaction, openly asking if he/she has made an error him/herself or whether there is some other cause of the problem. The psychiatrist who feels able to reconsider his/her own perspectives - his mind changed by the patient's mind - will foster mentalisation.

Overall, treatments currently shown to be moderately effective have in common an ability to stimulate attachment to the therapist while asking the patient to evaluate the accuracy of statements concerning their own mind states and those of others. More effective treatment lies in balancing these components in an increasingly optimal manner without inducing serious side-effects. This will require more specific treatment protocols and better focused training if psychotherapy for borderline personality disorder is to be provided free from harm.

  • Received April 8, 2005.
  • Revision received April 27, 2005.
  • Accepted April 29, 2005.


View Abstract