When the psychiatrist first meets with a patient who may have borderline personality disorder, a number of important issues related to differential diagnosis, etiology, the formulation, and treatment planning need to be considered. The psychiatrist performs an initial assessment to determine the treatment setting, completes a comprehensive evaluation (including differential diagnosis), and works with the patient to mutually establish the treatment framework. The psychiatrist also attends to a number of principles of psychiatric management that form the foundation of care for patients with borderline personality disorder. The psychiatrist next considers several principles of treatment selection (e.g., type, focus, number of clinicians to involve). Finally, the psychiatrist selects specific treatment strategies for the clinical features of borderline personality disorder.

The psychiatrist first performs an initial assessment of the patient and determines the treatment setting (e.g., inpatient or outpatient). Since patients with borderline personality disorder commonly experience suicidal ideation (and 8%–10% commit suicide), safety issues should be given priority in the initial assessment (see Section II.B.1, "Responding to Crises and Safety Monitoring," for a further discussion of this issue). A thorough safety evaluation should be done before a decision can be reached about whether outpatient, inpatient, or another level of care (e.g., partial hospitalization or residential care) is needed. Presented here are some of the more common indications for particular levels of care. However, this list is not intended to be exhaustive. Since indications for level of care are difficult to empirically investigate and studies are lacking, these recommendations are derived primarily from expert clinical opinion.

Indications for partial hospitalization (or brief inpatient hospitalization if partial hospitalization is not available) include the following:

• Dangerous, impulsive behavior unable to be managed with outpatient treatment
• Nonadherence with outpatient treatment and a deteriorating clinical picture
• Complex comorbidity that requires more intensive clinical assessment of response to treatment
• Symptoms of sufficient severity to interfere with functioning, work, or family life that are unresponsive to outpatient treatment

Indications for brief inpatient hospitalization include the following:

• Imminent danger to others
• Loss of control of suicidal impulses or serious suicide attempt
• Transient psychotic episodes associated with loss of impulse control or impaired judgment
• Symptoms of sufficient severity to interfere with functioning, work, or family life that are unresponsive to outpatient treatment and partial hospitalization

Indications for extended inpatient hospitalization include the following:

• Persistent and severe suicidality, self-destructiveness, or nonadherence to outpatient treatment or partial hospitalization
• Comorbid refractory axis I disorder (e.g., eating disorder, mood disorder) that presents a potential threat to life
• Comorbid substance abuse or dependence that is severe and unresponsive to outpatient treatment or partial hospitalization
• Continued risk of assaultive behavior toward others despite brief hospitalization
• Symptoms of sufficient severity to interfere with functioning, work, or family life that are unresponsive to outpatient treatment, partial hospitalization, and brief hospitalization

Once an initial assessment has been done and the treatment setting determined, a more comprehensive evaluation should be completed as soon as clinically feasible. Such an evaluation includes assessing the presence of comorbid disorders, degree and type of functional impairment, needs and goals, intrapsychic conflicts and defenses, developmental progress and arrests, adaptive and maladaptive coping styles, psychosocial stressors, and strengths in the face of stressors (see Part B, Section V.B, "Assessment"). The psychiatrist should attempt to understand the biological, interpersonal, familial, social, and cultural factors that affect the patient (3).

Special attention should be paid to the differential diagnosis of borderline personality disorder versus axis I conditions (see Part B, Sections V.A.2, "Comorbidity," and V.C, "Differential Diagnosis"). Treatment planning should address comorbid disorders from axis I (e.g., substance use disorders, depressive disorders, PTSD) and axis II as well as borderline personality disorder, with priority established according to risk or predominant symptoms. When priority is given to treating comorbid conditions (e.g., substance abuse, depression, PTSD, or an eating disorder), it may be helpful to caution patients or their families about the expected rate of response or extent of improvement. The prognosis for treatment of these axis I disorders is often poorer when borderline personality disorder is present. It is usually better to anticipate realistic problems than to encourage unrealistically high hopes.

It is important at the outset of treatment to establish a clear and explicit treatment framework. This is sometimes called "contract setting." While this process is generally applicable to the treatment of all patients, regardless of diagnosis, such an agreement is particularly important for patients with borderline personality disorder. The clinician and the patient can then refer to this agreement later in the treatment if the patient challenges it.

Patients and clinicians should establish agreements about goals of treatment sessions (e.g., symptom reduction, personal growth, improvement in functioning) and what role each is expected to perform to achieve these goals. Patients, for example, are expected to report on such issues as conflicts, dysfunction, and impending life changes. Clinicians are expected to offer understanding, explanations for treatment interventions, undistracted attention, and respectful, compassionate attitudes, with judicious feedback to patients that can help them attain their goals. In addition, it is essential for patients and clinicians to work toward establishing agreements about 1) when, where, and with what frequency sessions will be held; 2) a plan for crises management; 3) clarification of the clinician's after-hours availability; and 4) the fee, billing, and payment schedule.

Psychiatric management forms the foundation of psychiatric treatment for patients with borderline personality disorder. It consists of an array of ongoing activities and interventions that should be instituted for all patients. These include providing education about borderline personality disorder, facilitating adherence to a psychotherapeutic or psychopharmacological regimen that is satisfactory to both the patient and psychiatrist, and attempting to help the patient solve practical problems, giving advice and guidance when needed.

Specific components of psychiatric management are discussed here as well as additional important issues—such as the potential for splitting and boundary problems—that may complicate treatment and of which the clinician must be aware and manage.

Psychiatrists should assume that crises, such as interpersonal crises or self-destructive behavior, will occur. Psychiatrists may wish to establish an explicit understanding about what they expect a patient to do during crises and may want to be explicit about what the patient can expect from them. While some clinicians believe that this is of critical importance (4, 5), others believe that this approach is too inflexible and potentially adversarial. From the latter perspective, there is often a tension between the psychiatrist's role in helping patients to understand their behavior and the psychiatrist's role in ensuring patients' safety and in managing problematic behaviors. This tension may be particularly prominent when the psychiatrist is using a psychodynamic approach that relies heavily on interpretation and exploration. Regardless of the psychotherapeutic strategy, however, the psychiatrist has a fundamental responsibility to monitor this tension as part of the treatment process.

Patients with borderline personality disorder commonly experience suicidal ideation and are prone to make suicide attempts or engage in self-injurious behavior (e.g., cutting). Monitoring patients' safety is a critically important task. It is important that psychiatrists always evaluate indicators of self-injurious or suicidal ideas and reformulate the treatment plan as appropriate. Serious self-harm can occur if the potential danger is ignored or minimized. Before intervening to prevent self-endangering behaviors, the psychiatrist should first assess the potential danger, the patient's motivations, and to what extent the patient can manage his or her safety without external interventions (6). When the patient's safety is judged to be at serious risk, hospitalization may be indicated. Even in the context of appropriate treatment, some patients with borderline personality disorder will commit suicide.

Patients with borderline personality disorder have difficulty developing and sustaining trusting relationships. This issue may be a focus of treatment as well as a significant barrier to the development of the treatment alliance necessary to carry out the treatment plan. Therefore, the psychiatrist should pay particular attention to ascertaining that the patient agrees with and accepts the treatment plan; adherence or agreement cannot be assumed. Agreements should be explicit.

The first aspect of alliance building, referred to earlier as "contract setting," is establishing an agreement about respective roles and responsibilities and treatment goals. The next aspect of alliance building is to encourage patients to be actively engaged in the treatment, both in their tasks (e.g., monitoring medication effects or noting and reflecting on their feelings) and in the relationship (e.g., disclosing reactions or wishes to the clinician). This can be accomplished by focusing attention on whether the patient 1) understands and accepts what the psychiatrist says and 2) seems to feel understood and accepted by the psychiatrist. Techniques such as confrontation or interpretation may be appropriate over the long term after a "working alliance" (collaboration over a task) has been established. Psychotherapeutic approaches are often helpful in developing a working alliance for a pharmacotherapy component of the treatment plan. Reciprocally, the experience of being helped by medication that the psychiatrist prescribed can help a patient develop trust in his or her psychotherapeutic interventions.

Psychoeducational methods often are helpful and generally are welcomed by patients and, when appropriate, their families. At an appropriate point in treatment, patients should be familiarized with the diagnosis, including its expected course, responsiveness to treatment, and, when appropriate, pathogenic factors. Many patients with borderline personality disorder profit from ongoing education about self-care (e.g., safe sex, potential legal problems, balanced diet). Formal psychoeducational approaches may include having the patient read the text of DSM-IV-TR or books on borderline personality disorder written for laypersons. Some clinicians prefer to frame psychoeducational discussions in everyday terms and use the patient's own language to negotiate a shared understanding of the major areas of difficulty without turning to a text or manual. More extensive psychoeducational intervention, consisting of workshops, lectures, or seminars, may also be helpful.

Families or others—especially those who are younger—living with individuals with borderline personality disorder will also often benefit from psychoeducation about the disorder, its course, and its treatment. It is wise to introduce information about pathogenic issues that may involve family members with sensitivity to the information's likely effects (e.g., it may evoke undesirable reactions of guilt, anger, or defensiveness). Psychoeducation for families should be distinguished from family therapy, which is sometimes a desirable part of the treatment plan and sometimes not, depending on the patient's history and status of current relationships.

Providing optimal treatment for patients with borderline personality disorder who may be dangerously self-destructive frequently requires a treatment team that involves several clinicians. If the team members work collaboratively, the overall treatment will usually be enhanced by being better able to help patients contain their acting out (via fight or flight) and their projections onto others. It is essential that ongoing coordination of the overall treatment plan is assured by clear role definitions, plans for management of crises, and regular communication among the clinicians.

The team members must also have a clear agreement about which clinician is assuming the primary overall responsibility for the patient's safety and treatment. This individual serves as a gatekeeper for the appropriate level of care (whether it be hospitalization, residential treatment, or day hospitalization), oversees the family involvement, makes decisions regarding which potential treatment modalities are useful or should be discontinued, helps assess the impact of medications, and monitors the patient's safety. Because of the diversity of knowledge and expertise required for this oversight function, a psychiatrist is usually optimal for this role.

With all forms of treatment, it is important to monitor the treatment's effectiveness in an ongoing way. Often the course of treatment is uneven, with periodic setbacks (e.g., at times of stress). Such setbacks do not necessarily indicate that the treatment is ineffective. Nonetheless, ultimate improvement should be a reasonably expected outcome.

Patients with borderline personality disorder sometimes regress early in treatment as they begin to engage in the treatment process, getting somewhat worse before they get better. However, sustained deterioration is a problem that requires attention. Examples of such regressive phenomena include dysfunctional behavior (e.g., cessation of work, increased suicidality, onset of compulsive overeating) or immature behavior. This may occur when patients believe that they no longer need to be as responsible for taking care of themselves, thinking that their needs can and will now be met by those providing treatment.

Clinicians should be prepared to recognize this effect and then explore with patients whether their hope for such care is realistic and, if so, whether it is good for their long-term welfare. When the decline of functioning is sustained, it may mean that the focus of treatment needs to shift from exploration to other strategies (e.g., behavioral modification, vocational counseling, family education, or limit-setting). Of special significance is that such declines in function are likely to occur when patients with borderline personality disorder have reductions in the intensity or amount of support they receive, such as moving to a less intensive level of care. Clinicians need to be alert to the fact that such regressions may reflect the need to add support or structure temporarily to the treatment by way of easing the transition to less intensive treatment. Regressions may also occur when patients perceive particularly sympathetic, nurturant, or protective inclinations in those who are providing their care. Under these circumstances, clinicians need to clarify that these inclinations do not signify a readiness to take on a parenting role.

An issue that frequently requires assessment and response by psychiatrists is the sustained return of symptoms, the previous remission of which had been attributed, at least in part, to medications (although placebo effects may also have been involved). Assessment of such symptom "breakthroughs" requires knowledge of the patient's symptom presentation before the use of medication. Has the full symptom presentation returned? Are the current symptoms sustained over time, or do they reflect transitory and reactive moods in response to an interpersonal crisis? Medications can modulate the intensity of affective, cognitive, and impulsive symptoms, but they should not be expected to extinguish feelings of anger, sadness, and pain in response to separations, rejections, or other life stressors. When situational precipitants are identified, the clinician's primary focus should be to facilitate improved coping. Frequent medication changes in pursuit of improving transient mood states are unnecessary and generally ineffective. The patient should not be given the erroneous message that emotional responses to life events are merely biologic symptoms to be regulated by medications.

Clinicians with overall or primary responsibilities for patients with borderline personality disorder should have a low threshold for seeking consultation because of 1) the high frequency of countertransference reactions and medicolegal liability complications; 2) the high frequency of complicated multitreater, multimodality treatments; and 3) the particularly high level of inference, subjectivity, and life/death significance that clinical judgments involve. The principle that should guide whether a consultation is obtained is that improvement (e.g., less distress, more adaptive behaviors, greater trust) is to be expected during treatment. Thus, failure to show improvement in targeted goals by 6–12 months should raise considerations of introducing changes in the treatment. When a patient continues to do poorly after the treatment has been modified, consultation is indicated as a way of introducing and implementing treatment changes. When a consultant believes that the existing treatment cannot be improved, this offers support for continuing this treatment.

The phenomenon of "splitting" signifies an inability to reconcile alternative or opposing perceptions or feelings within the self or others, which is characteristic of borderline personality disorder. As a result, patients with borderline personality disorder tend to see people or situations in "black or white,""all or nothing,""good or bad" terms. In clinical settings, this phenomenon may be evident in their polarized but alternating views of others as either idealized (i.e., "all good") or devalued (i.e., "all bad"). When they perceive primary clinicians as "all bad" (usually prompted by feeling frustrated), this may precipitate flight from treatment. When splitting threatens continuation of the treatment, clinicians should be prepared to examine the transference and countertransference and consider altering treatment. This can be done by offering increased support, by seeking consultation, or by otherwise suggesting changes in the treatment. Clinicians should always arrange to communicate regularly about their patients to avoid splitting within the treatment team (i.e., one clinician or treatment is idealized while another is devalued). Integration of the clinicians helps patients integrate their internal splits.

Clinicians/therapists vary considerably in their tolerance for patient behaviors (e.g., phone calls, silences) and in their expectations of the patient (e.g., promptness, personal disclosures, homework between sessions). It is important to be explicit about these issues, thereby establishing "boundaries" around the treatment relationship and task. It is also important to be consistent with agreed-upon boundaries. Although patients may agree to such boundaries, some patients with borderline personality disorder will attempt to cross them (e.g., request between-session contacts or seek a personal, nonprofessional relationship). It remains the therapist's responsibility to monitor and sustain the treatment boundaries. Certain situations—e.g., practicing in a small community, rural area, or military setting—may complicate the task of maintaining treatment boundaries (7).

To diminish the problems associated with boundary issues, clinicians should be alert to their occurrence. Clinicians should then be proactive in exploring the meaning of the boundary crossing—whether it originated in their own behavior or that of the patient. After efforts are made to examine the meaning, whether the outcome is satisfactory or not, clinicians should restate their expectations about the treatment boundaries and their rationale. If the patient keeps testing the agreed-upon framework of therapy, clinicians should explicate its rationale. An example of this rationale is, "There are times when I may not answer your personal questions if I think it would be better for us to know why you've inquired." If a patient continues to challenge the framework despite exploration and clarification, a limit will eventually need to be set. An example of setting a limit is, "You recall that we agreed that if you feel suicidal, then you will go to an emergency room. If you cannot do this then your treatment may need to be changed."

When a boundary is crossed by the clinician/therapist, it is called a boundary "violation." The boundary can usually be restored with comments like the following: "If I were to call you every time I'm worried, your safety might come to depend too much on my intuition," or "Whenever I tell you something about my personal life, it limits our opportunity to understand more about what you imagine in the absence of knowing." When therapists find themselves making exceptions to their usual treatment boundaries, it is important to examine their motives (see Section IV, "Risk Management Issues"). It often signals the need for consultation or supervision.

Any consideration of sexual boundary violations by therapists must begin with a caveat: Patients can never be blamed for ethical transgressions by their therapists. It is the therapist's responsibility to act ethically, no matter how the patient may behave. Nevertheless, specific transference-countertransference enactments are at high risk for occurring with patients with borderline personality disorder. If a patient has experienced neglect and abuse in childhood, he or she may wish for the therapist to provide the love the patient missed from parents. Therapists may have rescue fantasies that lead them to collude with the patient's wish for the therapist to offer that love. This collusion in some cases leads to physical contact and even inappropriate physical contact between therapist and patient. Clinicians should be alert to these dynamics and seek consultation or personal psychotherapy or both whenever there is a risk of a boundary violation. Sexual interactions between a therapist and a patient are always unethical. When this type of boundary violation occurs, the therapist should immediately refer the patient to another therapist and seek consultation or personal psychotherapy.

Certain types of psychotherapy (as well as other psychosocial modalities) and certain psychotropic medications are effective for the treatment of borderline personality disorder. Although it has not been empirically established that one approach is more effective than another, clinical experience suggests that most patients with borderline personality disorder will need some form of extended psychotherapy in order to resolve interpersonal problems and attain and maintain lasting improvements in their personality and overall functioning. Pharmacotherapy often has an important adjunctive role, especially for diminution of targeted symptoms such as affective instability, impulsivity, psychotic-like symptoms, and self-destructive behavior. However, pharmacotherapy is unlikely to have substantial effects on some interpersonal problems and some of the other primary features of the disorder. Although no studies have compared a combination of psychotherapy and pharmacotherapy with either treatment alone, clinical experience indicates that many patients will benefit most from a combination of psychotherapy and pharmacotherapy.

Patients with borderline personality disorder frequently have comorbid axis I and other axis II conditions. The nature of certain borderline characteristics often complicates the treatment provided, even when treatment is focused on a comorbid axis I condition. For example, chronic self-destructive behaviors in response to perceived abandonment, marked impulsivity, or difficulties in establishing a therapeutic alliance have been referred to as "therapy-interfering behaviors." Treatment planning should address comorbid axis I and axis II disorders as well as borderline personality disorder, with priority established according to risk or predominant symptoms. The coexisting presence of borderline personality disorder with axis I disorders is associated with a poorer outcome of a number of axis I conditions. Treatment should usually be focused on both axis I and axis II disorders to facilitate the treatment of axis I conditions as well as address problematic, treatment-interfering personality features of borderline personality disorder itself. For patients with axis I conditions and coexisting borderline traits who do not meet full criteria for borderline personality disorder, it may be sufficient to focus treatment on the axis I conditions alone, although the therapy should be monitored and the focus changed to include the borderline traits if necessary to ensure the success of the treatment.

Features of borderline personality disorder are of a heterogeneous nature. Some patients, for example, display prominent affective instability, whereas others exhibit marked impulsivity or antisocial traits. The many possible combinations of comorbid axis I and axis II disorders further contribute to the heterogeneity of the clinical picture. Because of this heterogeneity, and because of each patient's unique history, the treatment plan needs to be flexible, adapted to the needs of the individual patient. Flexibility is also needed to respond to the changing characteristics of patients over time (e.g., at one point, the treatment focus may be on safety, whereas at another, it may be on improving relationships and functioning at work). Similarly, the psychiatrist may need to use different treatment modalities or refer the patient for adjunctive treatments (e.g., behavioral, supportive, or psychodynamic psychotherapy) at different times during the treatment.

Successful treatment is a collaborative process between the patient and the clinician. Patient preference is an important factor to consider when developing an individual treatment plan. The psychiatrist should explain and discuss the range of treatments available for the patient's condition, the modalities he or she recommends, and the rationale for having selected them. He or she should take time to elicit the patient's views about this provisional treatment plan and modify it to the extent feasible to take into account the patient's views and preferences. The hazard of nonadherence makes it worthwhile to spend whatever time may be required to gain the patient's assent to a viable treatment plan and his or her agreement to collaborate with the clinician(s) before any therapy is instituted.

Treatment can be provided by more than one clinician, each performing separate treatment tasks, or by a single clinician performing multiple tasks; both are viable approaches to treating borderline personality disorder. When there are multiple clinicians on the treatment team, they may be involved in a number of tasks, including individual psychotherapy, pharmacotherapy, group therapy, family therapy, or couples therapy or be involved as administrators on an inpatient unit, partial hospital setting, halfway house, or other living situation. Such treatment has a number of potential advantages. For example, it brings more types of expertise to the patient's treatment, and multiple clinicians may better contain the patient's self-destructive tendencies. However, because of patients' propensity for engaging in "splitting" (i.e., seeing one clinician as "good" and another as "bad") as well as the real-world difficulties of maintaining good collaboration with all other clinicians, the treatment has the potential to become fragmented. For this type of treatment to be successful, good collaboration of the entire treatment team and clarity of roles are essential (7). Regardless of whether treatment involves multiple clinicians or a single therapist, its effectiveness should be monitored over time, and it should be changed if the patient is not improving.

Although there is a long clinical tradition of treating borderline personality disorder, there are no well-designed studies comparing pharmacotherapy with psychotherapy. Nor are there any systematic investigations of the effects of combined medication and psychotherapy to either modality alone. Hence, in this section we will consider psychotherapy and pharmacotherapy separately, knowing that in clinical practice the two treatments are frequently combined. Indeed, many of the pharmacotherapy studies included patients with borderline personality disorder who were also in psychotherapy, and many patients in psychotherapy studies were also taking medication. A good deal of clinical wisdom supports the notion that carefully focused pharmacotherapy may enhance the patient's capacity to engage in psychotherapy.

Two psychotherapeutic approaches have been shown to have efficacy in randomized controlled trials: psychoanalytic/psychodynamic therapy and dialectical behavior therapy. We emphasize that these are psychotherapeutic approaches because the trials that have demonstrated efficacy (8–10) have involved sophisticated therapeutic programs rather than simply the provision of individual psychotherapy. Both approaches have three key features: 1) weekly meetings with an individual therapist, 2) one or more weekly group sessions, and 3) meetings between therapists for consultation/supervision. No results are available from direct comparisons of the two approaches to suggest which patients may respond better to which modality.

Psychoanalytic/psychodynamic therapy and dialectical behavior therapy are described in more detail in Part B of this guideline (see Section VI.B, "Review of Psychotherapy and Other Psychosocial Treatments"). One characteristic of both dialectical behavior therapy and psychoanalytic/psychodynamic therapy involves the length of treatment. Although brief therapy has not been systematically tested for patients with borderline personality disorder, the studies of extended treatment suggest that substantial improvement may not occur until after approximately 1 year of psychotherapeutic intervention has been provided and that many patients require even longer treatment.

In addition, clinical experience suggests that there are a number of "common features" that help guide the psychotherapist who is treating a patient with borderline personality disorder, regardless of the specific type of therapy used. The psychotherapist must emphasize the building of a strong therapeutic alliance with the patient to withstand the frequent affective storms within the treatment (11, 12). This process of building a positive working relationship is greatly enhanced by careful attention to specific goals for the treatment that both patient and therapist view as reasonable and attainable. Consolidation of a therapeutic alliance is facilitated as well by the establishment of clear boundaries within and around the treatment. Clinicians may find it useful to keep in mind that often patients will attempt to redefine, cross, or even violate boundaries as a test to see whether the treatment situation is safe enough for them to reveal their feelings to the therapist. Regular meeting times with firm expectation of attendance and participation are important as well as an understanding of the relative contributions of patient and therapist to the treatment process (12).

Therapists need to be active, interactive, and responsive to the patient. Self-destructive and suicidal behaviors need to be actively monitored. As seen in Figure 1, some therapists create a hierarchy of priorities to be considered in the treatment. For example, practitioners of dialectical behavior therapy (5) might consider suicidal behaviors first, followed by behaviors that interfere with therapy and then behaviors that interfere with quality of life. Practitioners of psychoanalytic or psychodynamic therapy (4, 13) might construct a similar hierarchy.

Figure 1. Treatment Priorities of Two Psychotherapeutic Approaches for Patients With Borderline Personality Disorder.a aSpecific behaviors that practitioners of each approach may encounter in patients with borderline personality disorder are presented, with those of highest priority sitting atop the "ladder"; treatment priority lessens as one goes down the ladder. bAs described by Linehan et al. (5). cAs described by Kernberg et al. (4) and Clarkin et al. (13).

Add to 'My Saved Images'

Many patients with borderline personality disorder have experienced considerable childhood neglect and abuse, so an empathic validation of the reality of that mistreatment and the suffering it has caused is a valuable intervention (12, 14–17). This process of empathizing with the patient's experience is also valuable in building a stronger therapeutic alliance (11) and paving the way for interpretive comments.

While validating patients' suffering, therapists must also help them take appropriate responsibility for their actions. Many patients with borderline personality disorder who have experienced trauma in the past blame themselves. Effective therapy helps patients realize that while they were not responsible for the neglect and abuse they experienced in childhood, they are currently responsible for controlling and preventing self-destructive patterns in the present. Psychotherapy can become derailed if there is too much focus on past trauma instead of attention to current functioning and problems in relating to others. Most therapists believe that interventions like interpretation, confrontation, and clarification should focus more on here-and-now situations than on the distant past (18). Interpretations of the here and now as it links to events in the past is a particularly useful form of interpretation for helping patients learn about the tendency toward repetition of maladaptive behavior patterns throughout their lives. Moreover, therapists must have a clear expectation of change as they help patients understand the origins of their suffering.

Because patients with borderline personality disorder possess a broad array of strengths and weaknesses, flexibility is a crucial aspect of effective therapy. At times therapists may be able to offer interpretations of unconscious patterns that help the patient develop insight. At other times, support and empathy may be more therapeutic. Supportive strategies should not be misconstrued as simply offering a friendly relationship. Validation or affirmation of the patient's experience, strengthening of adaptive defenses, and specific advice are examples of useful supportive approaches. Interpretive or exploratory comments often work synergistically with supportive interventions. Much of the action of the therapy is focused in the therapeutic relationship, and therapists must directly address unrealistic negative and, at times, unrealistic positive perceptions that patients have about the therapist to keep these perceptions from disrupting the treatment.

Appropriate management of intense feelings in both patient and therapist is a cornerstone of good psychotherapy (15). Consulting with other therapists, enlisting the help of a supervisor, and engaging in personal psychotherapy are useful methods of increasing one's capacity to contain these powerful feelings.

Clinical experience suggests that effective therapy for patients with borderline personality disorder also involves promoting reflection rather than impulsive action. Therapists should encourage the patient to engage in a process of self-observation to generate a greater understanding of how behaviors originate from internal motivations and affect states rather than coming from "out of the blue." Similarly, psychotherapy involves helping patients think through the consequences of their actions so that their judgment improves.

As previously noted, splitting is a major defense mechanism of patients with borderline personality disorder. The self and others are often regarded as "all good" or "all bad." This phenomenon is closely related to what Beck and Freeman (19) call "dichotomous thinking" and what Linehan (17) refers to as "all or none thinking." Psychotherapy must be geared to helping the patient begin to experience the shades of gray between the extremes and integrate the positive and negative aspects of the self and others. A major thrust of psychotherapy is to help patients recognize that their perception of others, including the therapist, is a representation rather than how they really are.

Because of the potential for impulsive behavior, therapists must be comfortable with setting limits on self-destructive behaviors. Similarly, at times therapists may need to convey to patients the limits of the therapist's own capacities. For example, therapists may need to lay out what they see as the necessary conditions to make therapy viable, with the understanding that the particular therapy may not be able to continue if the patient cannot adhere to minimal conditions that make psychotherapy possible.

Individual psychodynamic therapy without concomitant group therapy or other partial hospital modalities has some empirical support (20, 21). These studies, which used nonrandomized waiting list control conditions and "pre-post" comparisons, suggested that twice-weekly psychodynamic therapy for 1 year may be helpful for many patients with borderline personality disorder. In these studies, as in the randomized controlled trials, the therapists met regularly for group consultation.

There is a large clinical literature describing psychoanalytic/psychodynamic individual therapy for patients with borderline personality disorder (12, 14, 15, 18, 22–38). Most of these clinical reports document the difficult transference and countertransference aspects of the treatment, but they also provide considerable encouragement regarding the ultimate treatability of borderline personality disorder. Therapists who persevere describe substantial improvement in well-suited patients. Some of these skilled clinicians have reported success with the use of psychoanalysis four or five times weekly (22, 24, 34, 39). These cases may have involved "higher level" patients with borderline personality disorder who more likely fit into the Kernberg category of borderline personality organization (a broader theoretical rubric that describes a specific intrapsychic structural organization [27]). Some exceptional patients who do meet criteria for borderline personality disorder may be analyzable in the hands of gifted and well-trained clinicians, but most psychotherapists and psychoanalysts agree that psychoanalytic psychotherapy, at a frequency of one to three times a week face-to-face with the patient, is a more suitable treatment than psychoanalysis.

The limited literature on group therapy for patients with borderline personality disorder indicates that group treatment is not harmful and may be helpful, but it does not provide evidence of any clear advantage over individual psychotherapy. In general, group therapy is usually used in combination with individual therapy and other types of treatment, reflecting clinical wisdom that the combination is more effective than group therapy alone. Studies of combined individual dynamic therapy plus group therapy suggest that nonspecified components of combined interventions may have the greatest therapeutic power (40). Clinical experience suggests that a relatively homogeneous group of patients with borderline personality disorder is generally recommended for group therapy, although patients with dependent, schizoid, and narcissistic personality disorders or chronic depression also mix well with patients with borderline personality disorder (12). It is generally recommended that patients with antisocial personality disorder, untreated substance abuse, or psychosis not be included in groups designed for patients with borderline personality disorder.

The published literature on couples therapy with patients with borderline personality disorder consists only of reported clinical experience and case reports. This clinical literature suggests that couples therapy may be a useful and at times essential adjunctive treatment modality, since inherent in the very nature of the illness is the potential for chaotic interpersonal relationships. However, couples therapy is not recommended as the only form of treatment for patients with borderline personality disorder. Clinical experience suggests that it is relatively contraindicated when either partner is unable to listen to the other's criticisms or complaints without becoming too enraged, terrified, or despairing (41).

There is only one published study of family therapy for patients with borderline personality disorder (12), which found that a psychoeducational approach could greatly enhance communication and diminish conflict about independence. Published clinical reports differ in their recommendations about the appropriateness of family therapy and family involvement in the treatment. Whereas some clinicians recommend removing the patient's treatment from the family setting and not attempting family therapy (12), others recommend working with the patient and family together (42).

Clinical experience suggests that family work is most apt to be helpful and can be of critical importance when patients with borderline personality disorder have significant involvement with, or are financially dependent on, the family. Failure to enlist family support is a common reason for treatment dropout. The decision about whether to work with the family should depend on the degree of pathology within the family and strengths and weaknesses of the family members. Clinical experience suggests that a psychoeducational approach may lay the groundwork for the small subset of families for whom subsequent dynamic family therapy may be effective. Family therapy is not recommended as the only form of treatment for patients with borderline personality disorder.

A pharmacological approach to the treatment of borderline personality disorder is based upon evidence that some personality dimensions of patients appear to be mediated by dysregulation of neurotransmitter physiology and are responsive to medication (43). Pharmacotherapy is used to treat state symptoms during periods of acute decompensation as well as trait vulnerabilities. Although medications are widely used to treat patients who have borderline personality disorder, the Food and Drug Administration has not approved any medications specifically for the treatment of this disorder.

Pharmacotherapy may be guided by a set of basic assumptions that provide the theoretical rationale and empirical basis for choosing specific treatments. First, borderline personality disorder is a chronic disorder. Pharmacotherapy has demonstrated significant efficacy in many studies in diminishing symptom severity and optimizing functioning. However, cure is not a realistic goal—medications do not cure character. Second, borderline personality disorder is characterized by a number of dimensions; treatment is symptom-specific, directed at particular behavioral dimensions, rather than the disorder as a whole. Third, affective dysregulation and impulsive aggression are dimensions that require particular attention because they are risk factors for suicidal behavior, self-injury, and assaultiveness and are thus given high priority in selecting pharmacological agents. Fourth, pharmacotherapy targets the neurotransmitter basis of behavioral dimensions, affecting both acute symptomatic expression (e.g., anger treated with dopamine-blocking agents) and chronic vulnerability (e.g., temperamental impulsivity treated with serotonergic agents). Last, symptoms common to both axis I and II disorders may respond similarly to the same medication.

Symptoms exhibited within three behavioral dimensions seen in patients with borderline personality disorder are targeted for pharmacotherapy: affective dysregulation, impulsive-behavioral dyscontrol, and cognitive-perceptual difficulties.

Affective dysregulation in patients with borderline personality disorder is manifested by symptoms such as mood lability, rejection sensitivity, inappropriate intense anger, depressive "mood crashes," and temper outbursts. As seen in Table 2, patients displaying these features should be treated initially with one of the SSRIs, since this recommendation has strong empirical support (44–49). SSRIs have a broad spectrum of therapeutic effects, are relatively safe in overdose (compared with the tricyclic antidepressants or MAOIs), and have a favorable side effect profile, which supports treatment adherence. For example, fluoxetine has been found to improve depressed mood, mood lability, rejection sensitivity, impulsive behavior, self-mutilation, hostility, and even psychotic features. Research trials of SSRIs for treatment of borderline personality disorder have ranged in duration from 6 to 14 weeks for acute treatment studies, with continuation studies lasting up to 12 months. Some patients have retained improvement with maintenance treatment of 1–3 years. Studies have been reported with fluoxetine (in doses of 20–80 mg/day), sertraline (in doses of 100–200 mg/day), and the mixed norepinephrine/serotonin reuptake blocker venlafaxine (in doses of up to 400 mg/day) (45). A reasonable trial of an SSRI for treatment of patients with borderline personality disorder is at least 12 weeks.

Empirical trials of tricyclic antidepressants have produced inconsistent results (50, 51). Patients with comorbid major depression and borderline personality disorder have shown improvement following treatment with tricyclic antidepressants. However, in one placebo-controlled study, amitriptyline had a paradoxical effect in patients with borderline personality disorder, increasing suicidal ideation, paranoid thinking, and assaultiveness (50).

Since affective dysregulation is a dimension of temperament in patients with borderline personality disorder and not an acute illness, the duration of continuation and maintenance phases of pharmacotherapy cannot presently be defined. Significant improvement in the quality of the patient's coping skills and interpersonal relationships may be required before medication can be discontinued. Clinical experience suggests caution in discontinuing a successful antidepressant trial, especially if prior medication trials have failed. In the event of a suboptimal response to an SSRI, consideration should be given to switching to a second SSRI or related antidepressant. In one study of patients with borderline personality disorder (45), one-half of the patients who failed to respond to fluoxetine subsequently responded to sertraline.

When affective dysregulation appears as anxiety, an SSRI may be insufficient. At this point, the use of a benzodiazepine should be considered, although there is little systematic research on the use of these medications in patients with borderline personality disorder. Use of benzodiazepines may be problematic, given the risk of abuse, tolerance, and even behavioral toxicity. Despite clinical use of benzodiazepines (52), the short-acting benzodiazepine alprazolam was associated in one study with serious behavioral dyscontrol (53). Case reports demonstrate some utility for the long half-life benzodiazepine clonazepam (54). Clinical experience suggests that this medication, if used over the longer term, is best used adjunctively with an SSRI.

In theory, buspirone may treat anxiety or impulsive aggression without the risk of abuse or tolerance. However, the absence of an immediate effect generally makes this drug less acceptable to patients with borderline personality disorder. Currently, there are no published data on the use of buspirone for the treatment of affective dysregulation symptoms in patients with borderline personality disorder.

When affective dysregulation appears as disinhibited anger that coexists with other affective symptoms, SSRIs are the treatment of first choice. Fluoxetine has been shown to be effective for anger in patients with borderline personality disorder independent of its effects on depressed mood (44). Effects of fluoxetine on anger and impulsivity may appear within days, much earlier than antidepressant effects. Clinical experience suggests that in patients with severe behavioral dyscontrol, low-dose neuroleptics can be added to the regimen for a rapid response; they may also improve affective symptoms (50). Augmentation with neuroleptics should be considered before trying an MAOI, which requires more patient cooperation and adherence.

The efficacy of MAOIs for affective dysregulation symptoms in patients with borderline personality disorder has strong empirical support (55, 56). However, they are not a first-line treatment because of concerns about adherence to required dietary restrictions and because of their more problematic side effects. The effectiveness of MAOIs is supported by randomized controlled studies in patients with a primary diagnosis of borderline personality disorder as well as syndromes (e.g., atypical depression) in which the diagnosis of borderline personality disorder is considered secondary (57). MAOI antidepressants have demonstrated efficacy for impulsivity, mood reactivity, rejection sensitivity, anger, and hostility. They may also be effective for atypical depression and "hysteroid dysphoria." If a psychiatrist wishes to use an MAOI as a second-line treatment for symptoms of affective dysregulation, care should be taken to allow an adequate washout period after discontinuing SSRIs, particularly those with a long half-life.

Mood stabilizers are another second-line (or adjunctive) treatment for affective dysregulation symptoms in patients with borderline personality disorder. Lithium carbonate, carbamazepine, and valproate have been used for treatment of mood instability in patients with an axis II disorder, but there is a surprising paucity of empirical support for their use in borderline personality disorder, although studies are currently under way. Lithium carbonate has the most research support in randomized controlled trials studying patients with personality disorders (although not specifically borderline personality disorder). However, these studies focused primarily on impulsivity and aggression rather than mood regulation (58–60). Nonetheless, lithium may be helpful for mood lability as a primary presentation in patients with a personality disorder (61). Lithium has the disadvantage of a narrow margin of safety in overdose and the risk of hypothyroidism with long-term use.

Carbamazepine has demonstrated efficacy for impulsivity, anger, suicidality, and anxiety in patients with borderline personality disorder and hysteroid dysphoria (62). However, a small, controlled study of patients with borderline personality disorder with no axis I affective disorder found no significant benefit for carbamazepine (63). Carbamazepine has been reported to precipitate melancholic depression in patients with borderline personality disorder who have a history of this disorder (64), and it has the potential to cause bone marrow suppression.

Valproate demonstrated modest efficacy for depressed mood in patients with borderline personality disorder in one small, randomized, controlled trial (65). Open-label case reports suggest that this medication may also decrease agitation, aggression, anxiety, impulsivity, rejection sensitivity, anger, and irritability in patients with borderline personality disorder (66). Although the use of carbamazepine and valproate is widespread, psychiatrists should be aware of the lack of solid research support for their use in patients with borderline personality disorder.

Although there is a paucity of data on the efficacy of ECT for patients with borderline personality disorder, much of the available data suggest that depressed patients with a personality disorder generally have a poorer outcome with ECT than depressed patients without a personality disorder. Clinical experience suggests that while ECT may sometimes be indicated for patients with borderline personality disorder and severe axis I depression that has been resistant to pharmacotherapy, affective features of the borderline diagnosis are unlikely to respond to ECT.

As seen in Table 3, SSRIs are the treatment of choice for impulsive, disinhibited behavior in patients with borderline personality disorder. Randomized controlled trials and open-label studies with fluoxetine and sertraline have shown that their effect on impulsive behavior is independent of their effect on depression and anxiety (67). The effect of SSRIs on impulsivity may appear earlier than the effect on depression, with onset of action within days in some reports. Similarly, discontinuation of an SSRI following successful treatment may result in the reemergence of impulsive aggression within days. Clinical experience suggests that the duration of treatment following improvement of impulsive aggression should be determined by the clinical state of the patient, including his or her risk of exposure to life stressors and progress in learning coping skills. When the target for treatment is a trait vulnerability, a predetermined limit on treatment duration cannot be set.

When behavioral dyscontrol poses a serious threat to the patient's safety, it may be necessary to add a low-dose neuroleptic to the SSRI. Although this combination has not been studied, randomized controlled trials of neuroleptics alone have demonstrated their efficacy for impulsivity in patients with borderline personality disorder. The effect is rapid in onset, often within hours with oral use (and more rapidly when given intramuscularly), providing immediate control of escalating impulsive-aggressive behavior.

If an SSRI is ineffective, a trial of another SSRI or related antidepressant may be considered, although there are no published studies of this approach with impulsivity as a target symptom.

Clinical experience suggests that partial efficacy of an SSRI may be enhanced by adding lithium carbonate, although this combination has not been studied in patients with borderline personality disorder. Nonetheless, studies in impulsive adults and adolescents with criminal behavior (who were not selected for having borderline personality disorder) demonstrate that lithium alone is effective for impulsive-aggressive symptoms (58–60). If an SSRI is ineffective, switching to an MAOI antidepressant may be considered, although it is critical to have an adequate washout period. In a placebo-controlled crossover study of women with borderline personality disorder and hysteroid dysphoria, tranylcypromine was effective for the treatment of impulsive behavior (55). In another randomized controlled trial, phenelzine was effective for the treatment of anger and irritability (56, 68). On the basis of these findings, MAOIs are recommended for treatment of impulsivity, anger, and irritability in patients with borderline personality disorder. Combining MAOIs with valproate would also appear to be rational for selected patients, although there are no studies of these combinations.

Although the use of MAOIs in patients with borderline personality disorder is supported by randomized controlled trials, because of safety considerations many clinicians prefer to use mood stabilizers for treatment of impulsive behavior. The use of carbamazepine or valproate for impulse control in patients with borderline personality disorder appears to be widespread in clinical practice, although empirical evidence for their efficacy for impulsive aggression is limited and inconclusive. Carbamazepine has been shown to decrease behavioral impulsivity in patients with borderline personality disorder and hysteroid dysphoria. However, in a small controlled study that excluded patients with an affective disorder (63), carbamazepine proved no better than placebo for impulsivity in borderline personality disorder. Support for the use of valproate for impulsivity in borderline personality disorder is derived only from case reports, one small randomized control study, and one open-label trial in which impulsivity significantly improved (65, 66, 69, 70). Preliminary evidence suggests that the atypical neuroleptics may have some efficacy for impulsivity in patients with borderline personality disorder, especially severe self-mutilation and other impulsive behaviors arising from psychotic thinking. One open-label trial (71) and one case report (72) support the use of clozapine for this indication. The difficulties and risks involved in using clozapine (e.g., neutropenia) generally warrant its use only after other treatments have failed. The newer atypical neuroleptics have fewer risks, but there are few published data on their efficacy. Further investigation is warranted for their use as a treatment for refractory impulsive aggression in patients with borderline personality disorder.

Opioid antagonists (e.g., naltrexone) are sometimes used in an attempt to decrease self-injurious behavior in patients with borderline personality disorder. However, empirical support for this approach is very preliminary, since their efficacy has been demonstrated only in case reports and small case series.

As seen in Table 4, low-dose neuroleptics are the treatment of choice for these symptoms. This recommendation is strongly supported by randomized, double-blind controlled studies and open-label trials involving a variety of neuroleptics in both inpatient and outpatient settings and in adult and adolescent populations (50, 51, 55, 73–78).

Low-dose neuroleptics appear to have a broad spectrum of efficacy in acute use, improving not only psychotic-like symptoms but also depressed mood, impulsivity, and anger/hostility. Treatment effects appear within days to several weeks. Patients with cognitive symptoms as a primary complaint respond best to the use of low-dose neuroleptics. Patients with borderline personality disorder with prominent affective dysregulation and labile, depressive moods, in whom cognitive-perceptual distortions are secondary mood-congruent features, may do less well with neuroleptics alone. In this case, treatments more effective for affective dysregulation should be considered. Duration of treatment may be guided by the length of treatment trials in the literature, which are generally up to 12 weeks. Prolonged use of neuroleptic medication alone in patients with borderline personality disorder (i.e., up to 22 weeks in one study) has been associated with progressive nonadherence and dropout from treatment (68, 79). There is currently a paucity of research on the use of neuroleptic medication as long-term maintenance therapy for patients with borderline personality disorder, although many clinicians regularly use low-dose neuroleptics to help patients manage their vulnerability to disruptive anger. One longer-term study (80) found that a depot neuroleptic was effective for recurrent parasuicidal behaviors in patients with borderline personality disorder. The risk of tardive dyskinesia must be weighed carefully against perceived prophylactic benefit if maintenance strategies are considered (although this risk may be lessened by the use of atypical neuroleptics).

If response to treatment with low-dose neuroleptics is suboptimal after 4 to 6 weeks, the dose should be increased into a range suitable for treating axis I disorders and continued for a second trial period of 4–6 weeks. A suboptimal response at this point should prompt rereview of the etiology of the cognitive-perceptual symptoms. If the symptom presentation is truly part of a nonaffective presentation, atypical neuroleptics may be considered. Although there are no published randomized controlled trials of atypical neuroleptics in patients with borderline personality disorder, open-label trials and case studies support the use of clozapine for patients with severe, refractory psychotic symptoms "of an atypical nature" or for severe self-mutilation (71, 72, 81). However, clozapine is best used in patients with refractory borderline personality disorder, given the risk of agranulocytosis. Studies are currently under way with olanzapine and risperidone (82, 83). The generally favorable side effect profiles of risperidone and olanzapine, compared with those of traditional neuroleptics, indicate that these medications warrant careful empirical trials. As yet, there are no published data on the efficacy of quetiapine for borderline personality disorder.

Neuroleptics are often effective for anger and hostility regardless of whether these symptoms occur in the context of cognitive-perceptual symptoms or other types of symptoms. It is important to note that both MAOI and SSRI antidepressants have also been shown in randomized controlled trials to be effective for irritability and anger in some patients with borderline personality disorder with cognitive-perceptual symptoms.