The term ‘borderline’ has undergone a checkered career of multiple name changes since the early 1930s but finally emerged as the name of choice for a group of problematic patients whose original defining characteristics was being too disturbed to be treated by classical psychoanalysis. ‘Borderline’ patients were thought to function psychologically somewhere between neurosis and psychosis (Stern, 1938). This in-between state was later referred to variously as ambulatory schizophrenia (Zilboorg, 1941), the ‘as if personality’ (Deutsch, 1942), and ‘pseudoneurotic schizophrenia’ (Hoch and Polatin, 1949). In the 1960s and 1970s significant progress was made in understanding the disorder and findings from psychiatry and psychoanalysis were distilled over time into the nine descriptive criteria of BPD outlined in the DSM (American Psychiatric Association, 1994). Yet the definition remains problematic and the present DSM-IV does not define any ‘core borderline’ type. To the contrary, it has a prototype categorical design as no one of the nine criteria is necessary or jointly sufficient for the diagnosis. All that is required is five positive criteria. Accordingly there are 256 feasible combinations of criteria for a BPD diagnosis.

DSM-IV groups the PDs into Clusters A, B, and C. BPD is defined as belonging to Cluster B together with antisocial (ASPD), narcissistic (NPD), and histrionic (HPD) PDs (the flamboyant, dramatic, or help-seeking disorders). To some extent this clustering concurs with the psychoanalytic view promoted by Kernberg (see p. 292) who sees BPD as an organization (Kernberg, 1975a), which includes within it narcissistic and antisocial disorders and it has face validity in that comorbidity of PDs is widespread. However, diagnostic co-occurrence studies have provided mixed evidence for these associations. The best Axis II diagnostic co-occurrence study so far (McGlashan et al., 2000) found a significant association between BPD and ASPD and dependent PD as well as with posttraumatic stress disorder on Axis I.

BPD is a common condition with a prevalence of between 0.2 and 1.8% of the general population (Swartz et al., 1990), but most studies originate from North America and the situation may be different elsewhere. The most reliable study of the prevalence of the disorder in a community sample conducted in Oslo (Torgersen et al., 2001) suggested that the prevalence of BPD was not as frequent as commonly assumed with only 0.7% of patients being diagnosed as borderline from a representative community sample. But prevalence rates increase if patients within the mental health system are sampled, with the highest rates being found in those patients requiring the most intensive level of care—outpatient rates range from 8 to 11%, inpatient from 14 to 20%, and forensic services from 60 to 80%. In a Dutch forensic psychiatric hospital 80% of patients fulfilled criteria for at least one PD with paranoid, ASPD, and BPD being the most common (Ruiter and Greeven, 2000). Similar rates have been found in England and Sweden (Blackburn et al., 1990) with the most common being BPD and ASPD (Dolan and Coid, 1993).

The prevalence pattern described above reflects the considerable heterogeneity among patients receiving the diagnosis BPD. The question of severity is of utmost importance for the clinician. Severity is partly dependent on which criteria the patient fulfills. Suicidality and self-mutilation are regarded as severe symptoms, often being targets of specific treatment interventions and measures of treatment effect. Grilo et al. (2001) found that 59% of 240 BPD patients in the American collaborative longitudinal PD study fulfilled this criterion. Severity is also related to number of BPD criteria, partly because total number is positively correlated to number of other PD criteria and other PD diagnoses.

Assessing the level of severity, including risk assessment, is of course of paramount importance for treatment planning. The more severe, the more the patient is in need of comprehensive treatment programs, while the more resourceful BPD patient can benefit from less intensive treatment.

The psychodynamic approach to BPD is essentially a developmental view in which genetic vulnerability is unmasked as a result of disruption of the early mother–infant relationship and later environmental influences. This view has been supported by research on attachment disorders in childhood and empirical evidence demonstrating the association of childhood abuse with BPD (Johnson et al., 1999).

It was Kernberg (1967) who was the first to systematize these features using Kleinian ideas, combining classical instinct theory with object relations to define an underlying BPO, which occurs in many psychopathological situations, including BPD, NPD, HPD, psychotic disorders, some eating disorders, and in normal individuals who are exposed to extreme stress. He placed BPD between personality organizations to be found in psychotic and neurotic conditions. The four intrapsychic features pointed to by Kernberg were: (1) identity diffusion; (2) primitive defenses of projection, projective identification, splitting, and denial; (3) partially intact reality testing that is vulnerable to alterations and failures because of aggression; and (4) characteristic object relations.

To some extent the developmental views described above have gained some credence through research on attachment. Attachment theory, developed by John Bowlby (1969, 1973, 1980) postulates a universal human need to form close affectional bonds. At its core is the reciprocity of early relationships, which is a precondition of normal development probably in all mammals, including humans (Hofer, 1995). The attachment behaviors of the human infant (e.g., proximity seeking, smiling, clinging) are reciprocated by adult attachment behaviors (touching, holding, soothing) and these responses strengthen the attachment behavior of the infant toward that particular adult. The activation of attachment behaviors depends on the infant's evaluation of a range of environmental signals that results in the subjective experience of security or insecurity. The experience of security is the goal of the attachment system, which is thus first and foremost a regulator of emotional experience (Sroufe, 1996).

Although attachment theory is about proximity and the evocation of an experience of safety, it is also about the consequential development of robust, flexible, psychological processes that protect the individual from the stresses of human interaction and everyday life. Borderline patients are conceived of as failing to develop a stable sense of self because of disturbance in early attachment relationships. The experience of safety within the context of a close emotional relationship is essential for the development of an autonomous sense of self and anything that undermines the emergent self leads to anxiety and potentially an angry response as the child attempts to stabilize himself (Sroufe, 1996). The emergent self is only under serious (what might be thought of as existential) threat when it is in close emotional contact with another self—when a mind meets a mind—especially if that mind shows little understanding of the internal state of the child. Under ‘good enough’ conditions an agentive sense of oneself as experiencing thoughts and feelings that can effectively guide action is stabilized by a caregiver who provides an intersubjective milieu in which the self is strengthened through the interaction. Under conditions of chronic neglect and insensitivity instability of the self results first in anger and then aggression, which is evoked so frequently because of repeated parental neglect that it becomes incorporated into the self structure with the result that self-assertion, demand, wishes, and needs have to be accompanied by aggression if the self is to remain intact and stable. Such distortions to the self are not irreversible. The acquisition of the capacity to create a ‘narrative’ of one's thoughts and feelings, to mentalize, can overcome flaws in the organization of the self that can flow from the disorganization of early attachment. Thus the robustness of the self structure is dependent on the capacity to mentalize.

Cognitive-behavioral formulations of BPD are already as diverse as those of psychoanalysis even though it is only over the past decade that cognitive therapists have turned their attention to PDs. A clinically based approach has been proposed by a number of workers who have developed detailed conceptualizations and treatment strategies for each of the PDs. Initially these formulations built on the general view of psychopathology taken by cognitive therapy in which biased thinking patterns are considered as the core of a patient's problem and modification of these is necessary if the patient is to improve. Standard cognitive therapy focuses a great deal of attention on automatic thoughts and assumptions or beliefs. Automatic thoughts are akin to an internal running commentary, which is evoked under particular circumstances, for example when writing a chapter for a book the anxious individual may continually say to himself ‘I am never going to get this done and the editors will think that I am lazy’. Assumptions function at a deeper level of cognition and are tacit rules that give rise to automatic thoughts. But it was soon apparent that this formulation was overly simplistic and inadequate and a reformulated model has been proposed to take into account the complex psychological processes and behaviors found in BPD. In a revised model, Beck and associates (Alford, 1997) define personality in terms of patterns of social, motivational, and cognitive-affective processes thereby moving away from a primary emphasis on cognitions. However, personality is considered to be determined by ‘idiosyncratic structures’ known as schemas whose cognitive content gives meaning to the person. But the term schemas has been used in various ways, on the one hand being considered as a structure of cognition that filters and guides the processing of information and on the other hand being suggested as the building block of latent, core beliefs. The latter is the commonest use of the term and implies basic rules that individuals apply to organize their perceptions of the world, self, and future, and to adapt to the challenges of life.

It is schemas that are the cornerstone of cognitive formulations of BPD. Patients with BPD show characteristic assumptions and dichotomous thinking. Basic assumptions in the borderline commonly include ‘the world is a dangerous place’, ‘people cannot be trusted’, and ‘I am inherently unacceptable’. Dichotomous thinking is the tendency to evaluate experiences in terms of mutually exclusive categories such as good and bad, love and hate. Extreme evaluations such as these require extreme reactions and emotions, leading to abrupt changes in mood and immoderate behavior. The assumptions, dichotomous thinking and weak sense of identity are considered to form a mutually reinforcing and self-perpetuating system that governs relationships. Schemas that were once adaptive during childhood persist even after they have become seriously dysfunctional. They are maintained in the face of contradictory evidence because of distortion, discounting, seeing the evidence as an exception to the rule and extinction of the maladaptive systems does not take place as a result of negative reinforcement. In fact new experiences are filtered by the dysfunctional schemas in such a way that new experiences support existing dysfunctional beliefs and behavior patterns. Young (1990) has argued vociferously for a ‘fourth level of cognition’ to be added to this cognitive model of Beck, namely early maladaptive schemas (EMS). These are stable and enduring patterns of thinking and perception that begin early in life and are continually elaborated. EMS are unconditional beliefs linked together to form a core of an individual's self-image. Challenge threatens the core identity, which is defended with alacrity, guile, and yet desperation, as activation of the schemas may evoke aversive emotions. The EMS gives rise to ‘schema coping behavior’, which is the best adaptation to living that the borderline has found. These schemas are different conceptually from some of those discussed by Beck, which are not unconditional beliefs about the self. Beck refers to core beliefs and conditional beliefs, both of which are labeled schemas (Young, 1990). Core beliefs are more like EMS but conditional beliefs require an additional context to become active—‘if he gets close to me he will find out how awful I am and then reject me’.

Safran and Segal (1990) have integrated schemas within an interpersonal context arguing that the impact of an individual's beliefs and schemas is not purely cognitive but interacts with interpersonal behavior, which in turn has a reciprocal effect on beliefs. Thus the person is seen as being in a state of dynamic balance to the extent of provoking responses from others that perpetuate underlying assumptions. The borderline patient holds poorly integrated views of relationships with early caregivers and has extreme and unrealistic expectations that determine both behavior and emotional response. This is exacerbated by problems of identity and a fragile identity leads to a lack of clear and consistent goals and results in poorly co-ordinated actions, badly controlled impulses, and unsustained achievement. Relationships become an attempt to establish a stable identity through dependency, assertiveness, and control. From this viewpoint cognitive therapy is more than just changing assumptions. It becomes much more complex, lasts longer, and requires new techniques. The therapist cannot rely on modifying beliefs through review of evidence that contradicts maladaptive or negative conclusions. Borderlines cannot be argued out of their beliefs especially when they are dissonant with their affects. This has been recognized in cognitive-behavioral therapy (CBT) and attempts are made not only to challenge maladaptive beliefs, but also to help the patient to identify, support, and develop alternative schemas.

DBT is often considered as a CBT, although its focus is primarily behavioral, but it is distinct enough to be considered in its own right. DBT is a manualized, comprehensive psychosocial treatment developed specifically for suicidal individuals with BPD. The philosophy, biosocial theory, treatment targets, structure, strategies, and protocols of standard DBT are described in two treatment manuals (Linehan, 1993a,b).

Linehan has posited that the BPD (BPD) develops as a result of a transaction between biologic dysfunction in the emotion regulation system and an ‘invalidating environment.’ The biological determined dysregulation of the emotion regulating system is characterized by a high sensitivity for emotional stimuli, an intense reaction to even minimal stimuli, and a slow return to baseline, in combination with the incapability to modulate the emotional condition. But studies have shown that borderline patients do not show electrodermal hyporesponsiveness, which would predispose them to stimulus-seeking and dis-inhibited, impulsive behavior (Herpertz et al., 2001) and self-report data and physiological data suggest that the intensity of affective response in BPD is no different from controls (Herpertz et al., 1999). Nevertheless the theory of dysregulation offers a perspective free from implications of manipulation and destructiveness.

Over time, this transaction between emotion vulnerability and the invalidating environment leads to pervasive emotion dysregulation that is so characteristic of BPD. As a result, individuals with BPD frequently have limited learning opportunities to develop interpersonal, self-regulation, emotion regulation and distress tolerance skills. Furthermore, personal and environmental factors interfere with using the behavioral skills that the individual does possess and often reinforce inappropriate borderline behavior. DBT assumes that people with BPD are not at fault for having these motivational and skills deficits; they are trying their best to cope with life. DBT also assumes that people with BPD must fundamentally give up and replace dysfunctional coping behaviors (e.g., cutting, suicide attempts, abusing drugs, etc.) with functional behaviors. It logically follows that individuals with BPD need help in order to enhance their motivation and skills to develop a life worth living.

Consistent with other behavioral approaches, DBT assumes that all behavior, including dysfunctional behavior, occurs as a result of prior learning or biology. People with BPD learned to react in certain maladaptive ways to stimuli. In order to be able to change maladaptive behaviors, it is necessary to know which factors are controlling the behavior by means of a thorough behavioral assessment of the problem behavior. Following this theoretical view DBT uses a number of core methods in treatment, which are described on p. 298.

A number of integrative approaches have been developed over the past few years of which the most promising has been cognitive analytic therapy (CAT) (Ryle, 1997). The central idea of CAT formulation of BPD is that of the reciprocal role template and the procedures that secure it. It is claimed that a model of reciprocal role templates and their relations, known as the PSORM (Procedural Sequence Object Relations Model) is capable of providing a complete account of the symptoms of BPD. ‘States’ are described that consist of two complementary roles bound by a relationship paradigm. They are composed of attitudes to the self, and the world, which involve constellations of characteristic cognitions drives and emotions. The paired roles: caregiver–care receiver, victimizer–victim, and author–reader are all examples of states. They are learned through experience as blocks of reciprocal role pairs. Thus a child who is chastized by her mother for throwing food on the floor can often be observed to reenact this experience later with a toy and with role's reversed.

Another central CAT principle is that, in social situations, the adoption of one pole of a reciprocal role exerts a pressure on others to reciprocate and adopt a congruent pole. In any situation the role anyone adopts will be conditioned partly by the expectancies created by the situation, partly by their own state but also, to a greater or lesser extent by the roles adopted by other actors in the social setting.

In normal individuals reciprocal roles are numerous and for the most part moderated by three levels of control. Level 1 being the nature and number of the reciprocal roles and their attendant states, level 2 the command and control procedures that govern state transitions, and level 3 being the capacity for conscious self-reflection and conscious accounting for at least some of the other two structures.

In BPD often all three levels are abnormal. At level 1 the reciprocal roles are few in number and stark in nature. So that ‘abusing to abused’, ‘contemptuous to contemptible’, ‘ideally caring to ideally cared for’, and ‘abandoning to abandoned’ are all too frequently the only states in a borderline patient's reciprocal role repertoire. At level 2, states often switch rapidly following apparently minor ‘insults’. This accounts for the very common experience of therapists that patients may suddenly be thrown out by an innocuous comment that the therapist has made. Level 3 disruptions are restrictions of conscious reflection. In Ryle's view these may reflect actual injunctions to secrecy by early caregivers, be the consequence of the jerky progress between states, which combined with state-dependent recall disrupts any hope of sustained reflection or be consequent on trauma-induced dissociation.

For many years long-term psychoanalysis or prolonged inpatient admission was the mainstay of psychodynamic treatment of BPD. The approach, particularly inpatient treatment, has been increasingly questioned because of cost-effectiveness and absence of outcome research using randomized controlled designs even though such studies of inpatient treatment may neither be desirable nor feasible. The limited data available on cost suggests that inpatient admission may yield significant savings after completed treatment (Dolan et al., 1996), particularly in the use of criminal justice services in those with forensic histories.

Caution is suggested in ascribing benefits observed to the inpatient treatment itself by a naturalistic 5-year follow-up of individuals receiving inpatient treatment at the Cassel Hospital in London (Rosser et al., 1987). The study showed that, although patients with neurotic pathology, considerable depression, high intelligence, and lack of chronic outpatient history, did well at the end of treatment and over the follow-up period, patients with BPD had a less favorable outcome. Recent research from the same hospital has also suggested that treatment in the community following a shorter hospital treatment phase than usual is more effective than a prolonged hospital stay. Chiesa and Fonagy (2000) assessed the relative effectiveness of three treatment models for a mixed group of PDs: (1) long-term residential treatment using a therapeutic community approach; (2) briefer inpatient treatment followed by community-based dynamic therapy; and (3) general community psychiatric treatment. The results suggest that the brief inpatient therapeutic community treatment followed by outpatient dynamic therapy is more effective than both long-term residential therapeutic community treatment and general psychiatric treatment in the community on most measures, including self-harm, attempted suicide, and readmission rates to general psychiatric admission wards and is more cost-effective (Chiesa et al., 2004). However, this conclusion needs to be confirmed in a randomized study.

Marziali and Monroe-Blum have concentrated on group therapy alone without the additional milieu and social components of therapy. In a randomized controlled trial (Marziali and Monroe-Blum, 1995) they found equivalent results between group and individual therapy, concluding that on cost-effectiveness grounds group therapy is the treatment of choice. But further studies are needed to confirm their findings especially as the treatment offered was less structured than most other treatments and drop-out rates were high. Noncontrolled studies with day hospital stabilization followed by dynamic group therapy alone indicate the utility of the use of groups in BPD (Wilberg et al., 1998).

An uncontrolled study suggests that psychoanalytic psychotherapy based on ideas taken from self-psychology may be useful in BPD. Stevenson and Meares (1992) reported on 48 borderline patients treated with twice-weekly psychoanalytic psychotherapy that focused on a psychology of the self. Significant improvements were observed in the 30 patients who completed the therapy. Subjects made considerable gains in number of episodes of self-harm and violence, time away from work, number and length of hospital admissions, frequency of use of drugs, and self-report index of symptoms. Thirty percent of patients no longer fulfilled the criteria of BPD at the end of treatment. Improvement was maintained over 1 year. Further follow-up at 5 years confirmed the enduring effect of treatment and demonstrated a substantial saving associated with healthcare costs (Stevenson and Meares, 1999). The therapy concentrated early on the development of a therapeutic alliance and a relative or close friend was seen at the start of treatment. Both these factors may account for the low drop-out rate of 16%.

The most recent support for a psychoanalytically based approach has come from a randomized study examining the effectiveness of an attachment based and psychoanalytically oriented partial hospitalization program with standard psychiatric care for patients with BPD (Bateman and Fonagy, 1999, 2001). Understanding BPD as a disorder of the self resulting from a failure of mentalization (see p. 293), treatment interventions in group and individual therapy were organized to increase the reflective capacity of the patient. Thirty-eight patients with diagnosed BPD were allocated randomly to either a partially hospitalized group or to a standard psychiatric care (control) group. Treatment, which included individual and group psychoanalytic psychotherapy, was for a maximum of 18 months. On all outcome measures, including the frequency of suicide attempts and acts of self-harm, the number and duration of inpatient admissions, the use of psychotropic medication, and self-report measures of depression, anxiety, general symptom distress, interpersonal function, and social adjustment, there was significantly greater improvement in those allocated to psychotherapy. The improvement in symptoms and function were delayed by several months but were greatest by the end of treatment at 18 months. In a follow-up study, which was done on an intention-to-treat basis, gains were maintained after a further 18 months indicating that rehabilitative effects were stimulated during the treatment phase, and treatment has been found to be cost-effective (Bateman and Fonagy, 2003).

Studies of TFP are now becoming available and give promising results, although the outcome of a randomized controlled trial comparing TFP, DBT, and supportive psychotherapy is not yet known. In a cohort study (Clarkin et al., 2001) 23 female borderline patients were assessed at baseline and at the end of 12 months of treatment with diagnostic instruments, measures of suicidality, self-injurious behavior, and measures of medical and psychiatric service utilization. Compared with the year prior to treatment, the number of patients who made suicide attempts significantly decreased, as did the medical risk and severity of medical condition following self-injurious behavior. In addition patients during the treatment year had significantly fewer hospitalizations as well as number and days of psychiatric hospitalization compared with the year before. The drop-out rate was 19%.

The focus of DBT research has been on the initial stage of treatment whose aim is to help the patient to achieve behavioral control. Individual therapy in DBT first stage treatment focuses primarily on motivational issues, including the motivation to stay alive and to stay in treatment. Group therapy teaches self-regulation and change skills, and self and other acceptance skills.

In the original study (Linehan et al., 1991) 22 female patients were assigned to DBT and 22 to treatment as usual (TAU). Assessment was carried out during and at the end of therapy, and again after 1 year follow-up (Linehan et al., 1993). Control patients were significantly more likely to make suicide attempts (mean attempts in control and DBT patients, 33.5 and 6.8, respectively), spent significantly longer as inpatients over the year of treatment (mean 38.8 and 8.5 days, respectively), and were significantly more likely to drop out of those therapies they were assigned to (attrition 50% versus 16.7%, respectively). Follow-up was naturalistic, based on the proposition that the morbidity of this group precluded termination of therapy at the end of the experimental period. At 6-month follow-up DBT patients continued to show less parasuicidal behavior than controls, though at 1 year there were no between-group differences. While at 1 year DBT patients had had fewer days in hospital, at the 6-month assessment there were no between-group differences. Overall treatment with DBT for 1 year compared with TAU led to a reduction in the number and severity of suicide attempts and decreased the frequency and length of inpatient admission. However, there were no between-group differences on measures of depression, hopelessness, or reasons for living. Further, there were no differences in medically risky parasuicidal behavior between patients treated with DBT and those in alternative stable therapy suggesting that the stability of treatment may be an important factor in reducing risk.

A Dutch research project investigated standard DBT (Verheul et al., 2002) in 58 women with BPD who were randomly assigned either to 12 months of DBT or TAU using a randomized controlled design. Participants were clinical referrals from both addiction treatment and psychiatric services. Outcome measures included treatment retention, and course of suicidal, self-mutilating, and self-damaging impulsive behaviors. The results showed that DBT resulted in better retention rates and greater reductions of self-mutilating and self-damaging impulsive behaviors than TAU, especially among those with histories of frequent self-mutilation. This suggests that DBT enhances treatment retention, reduces severe dysfunctional behaviors (e.g., parasuicide, substance abuse, and binge eating), and reduce psychiatric hospitalization for both substance using and nonsubstance using BPD patients.

Across studies the effect on levels of depression, hopelessness, and survival and coping beliefs, and overall life satisfaction is inconclusive. Although originally designed for the outpatient treatment of suicidal individuals with BPD, DBT has been applied to many more populations, including comorbid substance dependence and BPD, and juveniles with antisocial behaviors, and in different contexts such as inpatient wards. The studies are discussed in two reviews (Koerner and Dimeff, 2000; Koerner and Linehan, 2000). Barley et al. (1993) evaluated the effectiveness of DBT for treatment of BPD in an inpatient setting. They found that during and following implementation of a DBT program there was a significant fall in rates of parasuicide when compared with a period before implementation of DBT. There was no significant difference, however, between the reported rates of parasuicide on the specialized DBT unit and another unit offering the hospital's standard treatment (TAU control). The results suggest that DBT may have made a successful contribution to reducing parasuicide but it is not unique in preventing parasuicidal behavior. Confirming this argument is a study reported by (Springer et al., 1996). These workers randomly assigned personality disordered patients either to a modified DBT program or to a wellness and life-styles group during a short inpatient stay. Patients in both groups improved significantly on most measures and there were no between-group differences.

Conclusions about the effectiveness of DBT as a treatment for the personality itself are premature (Levendusky, 2000; Scheel, 2000; Turner, 2000). It does seem effective for self-harm but no comparison with other potentially effective approaches have yet been reported. In addition, it is not clear which elements of DBT (psychotherapy, skills training, phone consultation, therapist consultation team) make this treatment method effective. Two process studies investigated the process of change in DBT by focusing on the possible influence of validation (Shearin and Linehan, 1992; Linehan and Heard, 1993; Linehan et al., 2002), but results are inconclusive. What we know thus far is that adding a DBT skills training group to ongoing outpatient individual psychotherapy does not seem to enhance treatment outcomes. Given that DBT is described as primarily a skills-training approach (Koerner and Linehan, 1992) this finding indicates that the central skills training component of DBT may not be of primary importance.

Davidson and Tyrer (1996), in an open study, used cognitive therapy for the treatment of two Cluster B PDs, namely, ASPD and BPD. They evaluated a brief (10-session) cognitive therapy approach using single-case methodology, which showed improvement in target problems and is now currently being evaluated in a three-center randomized controlled trial. Another small (n = 34), randomized controlled trial has recently been carried out using a mixed cognitive therapy and DBT protocol for treating Cluster B personality difficulties and disorders (Evans et al., 1999). Self-harm repeaters with a parasuicide attempt in the preceding 12 months were randomly allocated to Manual Assisted Cognitive Behavior Therapy (MACT) (n = 18), and the rest (n = 16) to TAU. The rate of suicide acts was lower with MACT (median 0.17/month MACT; 0.37/month TAU; P = 0.11) and self-rated depressive symptoms also improved (P = 0.03). The treatment involved a mean of 2.7 sessions and the observed average cost of care was 46% less with MACT (P = 0.22). This work has now been tested further in a randomized controlled trial involving five centers comparing MACT plus a self-help manual with TAU for patients who self-harm (Tyrer et al., 2003). Results are disappointing. Four hundred and eighty patients were randomized to MACT or TAU. MACT was given for five sessions with an additional two sessions later if appropriate. TAU varied across centers but consisted of psychiatric follow-up and support. Neither self-harm episodes, nor other psychometric assessment outcomes which included measures of depression and anxiety, showed any convincing differences between MACT and TAU, either at 6 or 12 months. It is possible that a longer period of treatment or greater engagement in face-to-face treatment, were this achievable in routine healthcare settings, would show more favorable results. However, a cost-effectiveness analysis suggested that there is a 90% probability that MACT is more cost-effective than TAU (Byford, 2003), although having BPD actually increased costs (Tyrer et al., 2004).

The paucity of robust evidence for the use of CAT in BPD is in contrast to the existence of a large number of single case reports or small uncontrolled series in CAT and to the existence of a large theoretical literature (Margison, 2000). However, Ryle and Golynkina (2000) have reported on 27 patients with BPD treated using CAT and attended a 6-month follow-up, and on 18 who also attended a follow-up at 18 months. All patients were formally diagnosed as suffering from BPD and received CAT according to strictly supervised criteria. At 6-month follow-up 14 patients no longer met formal criteria for BPD but given the instability of the diagnosis this is unlikely to have been a result of treatment itself, and in those who attended at 18 months there was a continuing decline in psychometric scores. These results need confirmation in a randomized trial and a recent comparison of CAT with TAU has suggested little benefit for the addition of CAT (personal communication), although this may be because of the small numbers in the trial.

In keeping with all other therapies used in the treatment of BPD there is general agreement that dynamic therapy should be adapted from its pure, traditional form if it is to effect personality change. It is more structured, requires an active, participating therapist (rather than the archetypal passive, unresponsive therapist), may combine individual and group therapy, and commonly agrees treatment priorities with the patient. This is illustrated by TFP. TFP relies on the techniques of clarification, confrontation, and transference interpretation within the evolving relationship between patient and therapist. The primary focus is on the dominant affect-laden themes that emerge in the therapeutic relationship in the here-and-now of the transference. At the beginning of treatment a hierarchy of issues is established: the containment of suicidal and self-destructive behaviors, the various ways of destroying the treatment, and the identification and recapitulation of dominant object relational patterns as they are experienced and expressed in the here-and-now of the transference relationship.

The initial aims in dynamic therapies are to engage the patient in treatment and to develop a ‘secure enough’ relationship to allow the patients inner representational world to become manifest in the relationship with the therapist. There is a continuum of intervention that moves through affirmation, advice and praise, empathic validation, encouragement to elaborate, and clarification to confrontation and interpretation (Gabbard, 1999). Only interpretation is specific to the psychodynamic approach.

Interpretation involves making conscious something that is unconscious. In order to do so the therapist focuses on the patient's experience of the therapeutic relationship from moment to moment to demonstrate to the patient her repertoire of partial internalized representations of self and other, which are represented in the interaction between patient and therapist. Overall, the goal is the resolution of primitive internalized object relations, that is, the integration of split off parts of the self and significant others into integrated conceptions. In effect the therapist shows the patient that their experience of others is not necessarily how they actually are but is a representation and helps her recognize what is hers and what is not. In doing so the self structure is strengthened and distortions are rectified.

Distortion of representations is challenged or confronted and particular emphasis is placed on defense mechanisms as they operate within the therapeutic relationship. Misrepresentation of self and other representations in BPD arises because of splitting and projective identification in which the therapist himself may be changed emotionally. For dynamic therapists it is important to understand these countertransference feelings as a communication and to analyze the role reversals within the patient and their actualization in the transference between patient and therapist. Constant monitoring of countertransference is a key therapeutic tool in dynamic therapy because of the extensive use of projective identification in BPD and the therapist will be cast as both abuser and abused, rescuer and attacker. He needs to remain equidistant from both if he is to help the patient; to do so he interprets the enforced roles in terms of the patient–therapist relationship, the patients past, and its purpose of stabilizing the self structure.

Containment and confrontation of anger and self-destructiveness is also a core aspect of treatment as aggression and its unmodulated expression is thought to be important as a cause of many of the borderline symptoms. The therapist should not act in the heat of the moment but analyze the underlying causes of outbursts once a crisis is over, linking feelings with actions. The aim is to make unbearable feelings bearable, the bewildering occurrence into an understandable experience, and the undigested trauma of the past into an assimilated event of the present.

The core strategies of DBT include: behavioral analysis, solution analysis and solution strategies, skills training (acquisition and strengthening of new skills), insight strategies, contingency management, exposure, cognitive modification, didactical interventions, orienting strategies, and the acquisition and strengthening of commitment. Because it is important that patients show effective behavior outside the therapy, generalization of new, adaptive behaviors needs to occur. Generalization is aimed at during role-play in the group session, in homework assignments of the skills training and through the phone consultation. DBT differs from pure behavior therapy to the extent that it integrates acceptance-based approaches with cognitive-behavioral change-based procedures. Validation, mindfulness practices, reciprocity, and a focus on the patient–therapist relationship are integrated with basic behavioral procedures of skills training, exposure-based procedures, cognitive modification, contingency management, and problem solving. The concept of dialectics, with its emphasis on synthesis of these polar opposite positions, provides a fresh lens in which to envision treatment possibilities. DBT can be differentiated from other therapies by the systematic use of therapist–patient telephone consultation, and, the emphasis in DBT given to the consultation team, where therapists’ capabilities and motivation to treat patients effectively are the focus. In summary, DBT is a multimodel and rather complex and comprehensive treatment strategy that is highly structured.

DBT is operationalized into five stages of treatment, although there is limited literature on treatment beyond stage 2. The goals of treatment for the first stage of DBT treatment are behavioral control, stability, and connection with treatment and care provider. Consistent with other behavioral treatments (Linehan, 1993b), has specified a pragmatic set of hierarchically arranged behavioral targets for this stage: decrease suicidal and other life-threatening behaviors, decrease therapy-interfering behaviors (e.g., not attending or coming late to therapy sessions, falling asleep during sessions, not completing therapy homework assignments), decreasing quality-of-life interfering behaviors (e.g., substance abuse, homelessness, unemployment, etc.), and increasing behavioral skills. Comprehensive DBT treatment for individuals in stage 1 includes five important functions necessary to decrease dysfunctional behaviors, to increase functionality, and to enhance quality of life. These functions include: (1) enhancing behavioral capabilities; (2) improving motivation to change; (3) assuring new capabilities generalize to the natural environment; (4) structuring the environment in the ways essential to support client and therapist capabilities; and (5) enhancing therapist capabilities and motivation to treat patients effectively.

These five functions are addressed within four different standard treatment modes of DBT lasting for 1 year: (1) weekly individual cognitive-behavioral psychotherapy sessions with the primary therapist; (2) weekly skills training groups lasting 2–2.5 hours per session; (3) weekly supervision and consultation meetings for the therapists; and (4) phone consultation, where patients are encouraged to get coaching in the appliance of new effective skills by phoning their primary therapists either during or outside office hours, for in vivo skills coaching to avert crises, facilitate skills generalization, and to repair between-session conflicts or misunderstandings between therapist and client. Individual therapy focuses primarily on motivational issues, including the motivation to stay alive and to stay in treatment. Group therapy teaches self-regulation and change skills, and self and other acceptance skills. Among its central principles is DBT's simultaneous focus on applying both acceptance and validation strategies and change (behavioral) strategies to achieve a synthetic (dialectical) balance in client functioning.

The individual advances to the second stage of treatment once behavioral control is achieved (e.g., when faced with situations that would historically trigger dysfunctional behavior, the individual is successful in applying skillful behavior to solve or withstand the problem rather instead of dysfunctional behavior). The focus during this second stage is emotional experience and processing of trauma from the past. The third stage emphasizes resolving ordinary problems in living (e.g., ordinary happiness and unhappiness). And then there is a fourth stage (transcendence) for those who desire a more meaningful existence. How change in these subsequent stages is brought about remains unclear.

CBT for BPDs differs from standard CBT (Davidson, 2000). Treatment is longer, there is greater emphasis on the therapeutic relationship, a focus on affects in conjunction with core beliefs, encouragement to develop new ways of behaving and thinking, and past history is explored to understand the context in which the core beliefs have developed.

Core beliefs are identified with the patient and are commonly at an extreme: ‘I am no good’, I am very clever’, ‘I am special’, I am worthless’, ‘Other people cannot be trusted’, ‘People will abandon me’. Work on the core beliefs is combined with identification of more intricate relationships between affects, cognitions, motivation, and actions. These ‘schemas’ are elaborated though direct questioning of the patient. A functional analysis is performed considering the affective element, the cognition, and their effect on self-regulation, motivation, and action.

Information that does not fit in with core beliefs is often avoided, ignored, or distorted. Homework tasks such as using a notebook to strengthen new behaviors are given and patients asked to outline evidence for their beliefs in terms of their history and their current life.

The focus on cognitive processes is combined with exploration of current relationships and examination of interpersonal difficulties, which may impact on therapy and prevent change. To this extent there is a limited use of transference, although it is not viewed as a repetition of past relationships.

CAT is offered for longer when used to treat patients with BPD than is traditional in the treatment of neurotic disorders. Twenty-four rather than 12 sessions are offered and the primary aim is to gain a developmental and social understanding of the patients problem and to share this in a clear, user-friendly way. Understanding is operationalized as defining reciprocal emotional roles that are exacerbated or perpetuated by redundant coping procedures. Many of these may have been effective solutions to childhood distress but are now outdated and inappropriate. These reciprocal roles are viewed as being enacted within the relationship to the therapist and working them through in treatment is at the heart of the therapy. The work is active and shared. Diagrams and outlines of problems are written down and emotional roles are drawn out to become tools for use within and without therapy—the practical manifestation of the PSORM. The axiom is reformulation of problems, recognition, and revision.

The initial sessions involve clarification of difficulties and completion of a psychotherapy file, which asks about typical common problems known as traps, dilemmas, and snags.

The problems are explored and at around the fourth session the therapist presents the patient with a reformulation. This includes a description of the patients’ life, their difficulties, and a formulation of their problems as target problem procedures that become the focus of therapy. This may also be presented as a sequential diagrammatic reformulation (SDR) and the patient is asked to reflect on it and to make modifications until a shared understanding is arrived at. The SDR is a jointly constructed diagram of a patient's interpersonal function that maps the movement of their feelings and resulting behaviors. A key task for the therapist is to avoid colluding with unhelpful aspects of the patients beliefs or being forced into specific interpersonal roles, for example becoming the victim in a victim/victimizer dynamic, an abuser in abuser/abused interaction. The different reciprocal roles and self-states are explored and towards termination the therapist writes a summary letter to the patient giving a realistic estimation of changes, an outline of further work to be done, and positive features on how the patient can be his own therapist. In general, where possible, there is a climate of therapeutic optimism and the patient is asked to write his own letter promoting self-evaluation and outlining his understanding of the achievements and disappointments of therapy.

Supervision is of considerable importance for practitioners treating borderline patients because of the strong emotional reactions that are evoked. A distinction needs to be made between attitudinal responses to features of borderline patients, such as insatiable demand, and the emotional responses that are evoked as countertransference reactions. The more intensive the treatment and the less structured it is the more likely that problematic transference and countertransference problems will arise. All therapists have a limit to the amount of frustration, hatred, or even desire that they can tolerate without giving in to action even if these feelings are understood as arising because of the patient experiencing him as an earlier object relationship. When they are part of unresolved aspects in the therapist or overidentification and empathy with the patient treatment becomes more problematic and boundary violations can occur. It is therefore imperative that therapists obtain adequate support and supervision and are given help in structuring sessions according to the treatment model being used if they are to remain on task. The management of countertransference in borderline patients is extensively reviewed by Gabbard and Wilkinson (1994).

One reason that psychotherapy for BPD tends to be structured and requires supportive supervision is to counter regression in the patient, which when combined with unprocessed countertransference responses can stimulate transgression of patient–therapist boundaries. Regression in BPD remains a topic of debate following the recommendation of early practitioners that it can be therapeutic (Balint, 1968). However, expert opinion now suggests that there is no place for such actions in the treatment of borderline patients and maintenance of physical and therapeutic boundaries is as important for therapists as it is for patients, although if applied too rigorously can become antitherapeutic. A balance needs to be struck that allows some regressive process but not enough to encourage acting-out and destabilization, which in turn can lead to boundary violations.

Boundaries of therapy, for example extending the session by a few minutes, are often crossed initially without the patient or therapist understanding their potential consequences and occur under the guise of having to deal with an immediate problem such as suicidal impulses. In retrospect it can be seen that these apparently innocuous occurrences were the beginning of a slippery slope leading towards catastrophic boundary violations such as a sexual relationship between patient and therapist. In general, therapists need to be alert to any changes in their normal practice or use of techniques that are outside accepted consensus and should always seek a colleague's opinion if in doubt. With the more complex borderline patient it is important that practitioners do not work alone.

Gabbard (2003) describes a miscarriage of psychoanalytic treatment in which a suicidal borderline patient induces a belief in the analyst that only he can save her and as he becomes more frantic about her suicide risk and decides not to admit her to hospital, he agrees to allow her to spend a night at his house. Inevitably this leads to a sexual encounter and yet even though wracked with guilt the analyst continues to believe that ‘at least I saved her from suicide’. Gabbard suggests that boundary transgressions such as these are directly related to the mismanagement of aggression and hatred. The analyst is determined to demonstrate that he is completely unlike abusive parents and that he can compensate the patient for her tragic past. In order to do so his analytic posture disavows any connection to an internalized representation of a bad object that torments the aggressor. He has named this ‘disidentification with the aggressor’ (Gabbard, 1997). While this example is from the literature on dynamic therapy there is no evidence that boundary violations by dynamic therapists are more common than by any other group and it is probable that they write about it more.

In general borderline patients are a greater risk to themselves than they are to others and it is important to remember that they also are help seeking and therefore tend to let others know about their suicidal impulses. Assessing risk in BPD is a difficult art partly because of the fluctuating nature of symptoms but also as a result of the chronicity of suicidal ideas in many patients. It can be difficult to know when chronic ideas have tipped over into an acute suicidal crisis. All therapists have to be able to manage their own anxiety; anxiety in the patient will generate anxiety in the therapist, which, if uncontained will lead to misjudgment and mismanagement. Therapists need to be aware of factors that increase suicide risk such as previous serious attempts, use of drugs and alcohol, hopelessness, high anxiety, and lack of social support. Consultation with a colleague should occur and reasons for the severity of the suicide risk considered in the light of the prevailing relationship with the therapist. All threats of suicide should be taken seriously and explored within therapy before a decision is taken about structural intervention such as inpatient admission. Overall, it is best to allow a patient to retain responsibility for his own life and to have arranged admission pathways at the outset of therapy. Many patients are able to control their own admission and gradually learn ways of reducing risk without admission. It is at these times clinicians need to work together and to avoid the splits that arise, often with ill-considered apportionment of blame when things go wrong.

Chronic self-destructive behavior takes many forms, such as drug binges, excessive promiscuity, shoplifting, self-mutilation, head banging, and should be considered within the therapy itself. It has already been mentioned that many therapists, irrespective of treatment model, set contracts, but the danger is that it then becomes the therapist rather than the patient who wants the patient to stop self-destructive behavior. This may become a countertransference enactment and is unlikely to reduce the behavior. Most clinicians move through a series of interventions that include education about the effects, understanding the triggers both internal and external, and understanding the forces behind the acts. Often chronic compulsive behavior relieves anxiety and distress and so alternative routes for reducing anxiety and emotional turmoil need to be identified.