Integrated Treatment of Borderline Personality Disorder

Sponsored by CME, Inc. for 1 hour of category 1 credit
Original release date 4/96. Approved for CME credit through 4/97.

By, Glen O. Gabbard, M.D.

Educational Objectives for Psychiatrists:
After you finish reading this article, you will be familiar with:

The optimal treatment of borderline personality disorder (BPD) is an integration of psychotherapy and pharmacotherapy. In my view, the therapist should also prescribe the medication, so that all transference issues are contained in one relationship, particularly transference issues related to compliance with medication. The two functions, however, can be handled effectively by two different people, as long as the two clinicians maintain good communication and alertness to splitting,


Two major forms of psychotherapy are typically used in the treatment of borderline patients: dynamic therapy and dialectical behavior therapy (DBT). The latter is the only psychotherapy shown to be efficacious by randomized controlled trials, but nonran-domized "pre-post" studies and clinical reports suggest that dynamic therapy may also be useful. While DBT involves once-weekly group and once-weekly individual therapy focused on coping skills and behavioral outcomes, psychodynamic therapy occurs between one and three times weekly and focuses on the patient's internal experiences and relationships in addition to self-destructive behaviors.

Regardless of which type of therapy one chooses, borderline patients generally do better with structure. The therapist needs to be clear about the scheduling of sessions, the length of the session (including clarity about when the session begins and ends) and how the fee payment will take place. These are the kinds of structural arrangements that provide some stability to the treatment and enhance the effectiveness of the therapy.

From the earliest days of research on borderline personality disorder, a clear and consistent diagnostic finding has been that rage is a central problem. Patients will be angry much of the time, and therapists may feel like retaliating in kind, or withdrawing by becoming more aloof or by mentally leaving the room. Treatment will be smoother if the therapist can hold firm and stay in a mid- position between withdrawing or retaliating. Much of the countertransference acting out I have seen as a consultant has been with therapists who became so intensely angry at a patient that they could no longer function in a professional, competent manner.

Therapists must help extremely impulsive borderline patients to see that a connection exists between an impulsive action and an internal state involving feelings and thoughts. If a patient says to me, "I don't know why I did that," I will typically say, "I don't think we should accept тI don't know.' I think we ought to get beyond it. Let's examine what happened before that." I look for a linkage between the behavior and an event, or a feeling about the event.

An important principle in dynamic psychotherapy with borderline patients is to create a sense of an inner world peopled with representations of others, as well as fantasies and feelings, and where actions are motivated both in the patient and in others. This is what Fonagy and his colleagues have called reflective self function, which they define as the "capacity to think of their own and others' actions in terms of mental states [that is, the] ability to invoke mental state constructs: feelings, beliefs, intentions, conflicts and other psychological states."

Another important issue in all psychotherapy with borderline patients is limit-setting. In dealing with action-prone patients who are impulsive, the therapist could be setting limits all the time, trying to tell the patients what to do and what not to do. Waldinger has suggested a useful way of recognizing what behaviors need to be limited-primarily any acting-out behavior that threatens the safety of the patient, the therapist or the therapy.

When matters of safety are involved, therapists can make serious errors by not setting limits. In fact, a spin-off report (Colson and others) from the Psychotherapy Research Project at The Menninger Clinic looked at 11 patients who had negative outcomes, most of whom would now be diagnosed as having borderline personality disorder. In these cases the therapist had not actively set limits with the patient but had simply offered interpretations of the meaning of a variety of acting-out behaviors that jeopardized the therapy.

Data suggest that childhood trauma may be etiologically significant in well over half of borderline personality disorder patients (Zanarini). Hence the patient's anger and manipulative behaviors, which create so many countertransference problems for the therapist, can be reframed as adaptive and understandable in light of early trauma (Gunderson and Chu). In our Menninger Treatment Interventions Project (Gabbard and others 1994; Horwitz and others), we studied audiotaped transcripts to identify what kind of therapist interventions seemed to facilitate a better therapeutic alliance with borderline patients. We found that in patients with early trauma, transference interpretation of the patient's aggression caused the alliance to deteriorate, because the patients experienced the interpretation as a lack of recognition that real tormentors were involved. On the other hand, our study found that empathic validation of the patient's perception as adaptive in some way improved the alliance. In other words, recognizing that something in the past has caused the patient to be filled with rage can often facilitate the feeling that therapist and patient are working together toward a common goal, instead of being at odds with each other. Transference interpretations appeared most effective when a series of supportive interventions created a climate in which the patient could accept the therapist's observations.

Splitting and Projective Identification

Attention to transference and counter-transference is, of course, a cornerstone of dynamic therapy with the borderline patient. An understanding of splitting and projective identification, the principal defense mechanisms of borderline personality disorder patients, is extremely useful as a conceptual framework (Gabbard and Wilkinson). Splitting involves alternating expressions of contradictory behaviors and attitudes that the patient regards with lack of concern and bland denial (Kernberg). These also may be manifested as the coexistence of contradictory self-representations that alternate with one another. In addition, there may be a corresponding compartmentalization of individuals into "all good" and "all bad" camps.

Splitting works hand in hand with projective identification, which involves an unconscious disavowal of aspects of oneself while simultaneously attributing those disavowed qualities to someone else. For example, when patients have internalized an abusive introject as part of their experience of childhood trauma, they may unconsciously attribute that introject to the therapist. These patients may then behave in such a way that interpersonal pressures are exerted on the therapist to begin to feel and behave like the abusive introject. Hence an "all bad" internal object may be split off and projectively disavowed so that the patient actually recreates a childhood relationship with an abusive parent in the transference-countertransference dimensions of the psychotherapy.

There is nothing mystical about this apparent transfer of intrapsychic contents. Rather, clearly observable behavior in the patient induces a corresponding response in the therapist. If a patient acts in an irritating manner, the therapist will become irritated. The therapist often experiences an obligatory quality in projective identification, as though he or she is being "bullied" into conforming to a prescribed role. This process happens automatically and unconsciously and is ordinarily discovered retrospectively after there has already been a controlled enactment of a particular aspect of the patient's inner world.

It would be misleading, of course, if we said that the patient projects into an empty container. There ordinarily is some sort of "hook" within the treater that receives the patient's projection and responds accordingly. For example, any one of us could become filled with murderous rage given the appropriate stimulus. However, our continuous sense of self operating from day to day does not generally include a homicidal, enraged component. These aspects of ourselves are normally repressed and only become activated when we are provoked by the patient's behavior. Hence one could view projective identification as involving a countertrans-ference reaction in the therapist that is a joint creation (Gabbard 1995)-part of it reflects what is induced by the patient, and part of it reflects what is preexisting in the therapist.

Abusers, Victims and Rescuers

The model of projective identification, which provides a bridge between the intrapsychic and the interpersonal, can be applied to three of the common internal self- and object-representations that one finds in borderline patients. These are the abuser, the victim and the omnipotent rescuer, usefully conceptualized as characters in an internal drama. The roles are played out in introjective and projective processes throughout the treatment (Davies and Frawley; Gabbard 1992). Patients who have not been literally sexually and physically abused may still have abusive internal objects, largely because of the power of fantasy to shape internal representations.

These three roles can appear in any particular order or sequence. A common sequence is for the therapist to start out in the role of rescuer (a role to which therapists are naturally inclined), while the patient begins in the role of victim. The therapist may become involved in an effort to be the perfect parent who will repair all of the damage done by the real parent. This role is doomed to failure because the therapist is not a parent and will never be able to fulfill all of the patient's wishes.

The therapist who tries to be an all-good parent starts running the extra mile by extending hours, not collecting the fee, taking repeated late-night calls and hugging the patient. Demands by the patient may escalate until the therapist begins to feel tormented, as though a victim of the patient, who is now viewed as an abuser. Childhood trauma victims have internalized an abusive introject that will be activated in the treatment, even though they may consciously abhor such an identification

Many therapists cannot set limits because of conflicts over their own aggression. The concept of limit-setting, however, applies as much to the therapist's limits as to the patient's. Patients must be told at some point what the reasonable limits of the therapist are and allowed to express their reactions to those limits. In the worst scenario, the situation escalates until the therapist ends up retraumatizing the patient, who again assumes the victim role as the therapist becomes the abuser.

Many of the cases of sexual relations between therapist and patient have followed this model, where the therapist, under the guise of trying to love the patient back to health, instead becomes an abuser (Gabbard and Lester). In such cases therapists often split off awareness of this kind of quasi-psychotic countertrans-ference reaction, so they are oblivious to the fact that they are repeating the same kind of incestuous relationship that occurred in childhood. The therapist can become the abuser to the patient's victim in other, less dramatic ways. For example, the therapist may start making sarcastic or sadistic comments, may stop paying attention, may not show up for hours, or may use excessive abreaction that retraumatizes the patient, who is forced to relive horrendous experiences.


In examining pharmacotherapy of borderline personality disorder, we first need to recognize that pharmacotherapy is always an adjunct to individual psychotherapy. It will not cure the patient by itself, but it may be extremely useful. The whole field of pharmacotherapy of borderline personality disorderis really in its infancy. A lot of trial and error is involved. We have a little bit of data from empirical studies to guide us but not a great deal.

A target-symptom approach provides guidelines for the clinician. For psychotic symptoms (such as paranoid thinking, mild thought disorder or psychotic rage), low-dose antipsychotics can be very useful. The patient should not be kept on antipsychotics for very long, however, unless absolutely necessary, because of the risk of tardive dyskinesia. Carbamazepine (Tegretol) is underutilized with borderline patients but can be very effective, particularly for behavioral dyscontrol (Cowdry and Gardner). Alprazolam (Xanax) should generally be avoided because it may disinhibit the patient.

MAO inhibitors are used for rejection-sensitive, hysteroid dysphoria and also for so-called paradoxical symptoms of depression like hyperphagia and hypersomnia. If a good therapeutic alliance exists, and the patient can be trusted to stay on the required diet, MAO inhibitors can be effective (Cowdry and Gardner); but without sufficient cooperation, the patient may have a convenient tool for self-destruction.

Recent research suggests that selective serotonin reuptake inhibitors (SSRIs) such as fluoxetine (Prozac) should be the first-line treatment for major depression (tricyclics tend not to work very well). Markovitz and others did a 14-week double-blind, placebo-controlled trial on 31 outpatients. Twenty-five of the 31 had coexisting major affective disorder. They found that the treatment group showed significant improvement in depression, anxiety, paranoia, psychoticism, interpersonal sensitivity, obsessionality and hostility, as well as global functioning. A dose of 80 mg per day was effective, while lower doses were not.

Improvement was not limited to patients with comorbid affective disorders-suggesting that something other than a typical major depression was also being treated. In fact, Gunderson and Phillips have argued that borderline patients may have a particular type of depression that is characterized principally by impulsivity instead of affective dysregulation. Rogers and colleagues examined 50 inpatients, 21 of whom had BPD. The aspects of depression most associated with borderline personality disorder were self-condemnation, emptiness, abandonment fears, self-destructiveness and hopelessness. These traits are sometimes associated with a kind of characterological depression in borderline patients that may be qualitatively distinct from an Axis I major depression. Hence, SSRIs may still be beneficial even in the absence of a clear Axis I major depression.

Salzman and his colleagues did a 13-week double-blind study of patients with mild to moderate borderline personality disorder. Thirteen patients received fluoxetine and nine received placebo. A lower daily dose of 40 mg was used, but the patients on fluoxetine showed a clinically and statistically significant decrease in anger.

The clinician treating a borderline patient must think simultaneously about both brain and mind, about both biology and psychology, just like the physicist thinks about particles and waves simultaneously. Waldinger and Frank studied 40 therapists treating borderline patients and found that the therapists were most likely to prescribe medications when they felt pessimistic about the patient's capacity to use psychotherapy. This finding supports the use of a target-symptom model, so that one is thinking systematically and not just prescribing because of counter-transference responses. Certainly, the countertransference roles of rescuer and abuser may be translated into specific actions involving the prescribing of medication. Borderline patients may also have positive and productive reactions to medication that assist the psychotherapy. They may view the pill as a transitional object or a substitute for the therapist, which may help them with intense separation anxiety over a long weekend or the therapist's vacation. In addition, SSRIs may reduce the intensity of affect sufficiently that patients are better able to reflect on their internal experience. There are preliminary data suggesting that serotonergic abnormalities may play a significant role in the etiology and/or pathogenesis of borderline personality disorder (Silk).

At any rate, medication should not be split off as separate from the psychotherapy process. Even if two persons are involved with each of the separate tasks, the treaters must keep both elements in mind when working with the patient and when communicating with each other.


Most patients with borderline personality disorder are treated in a managed care setting, in which the managed care utilization reviewer may limit the number of sessions to only a handful. Unfortunately, borderline patients rarely get better without more extended psychotherapy (Howard and colleagues). A strong argument can be made, however, that long-term psychotherapy is highly cost-effective with borderline patients. Linehan and others compared a group of patients who received dialectical behavior therapy for one year to a "treatment as usual" control group. Those who regularly received the dialectical behavior therapy for one year had significantly fewer psychiatric hospital days-8.46 per year compared to 38.8 for controls-a substantial difference. They also had fewer instances of self-mutilation-1.5 acts compared to nine in the control group-and less medically severe episodes of self-mutilation. The cost savings was $10,000 per patient per year by having regular weekly therapy (Heard). A strong argument can then be made with managed care companies that they will actually save money in the long run by providing regular weekly psychotherapy over a long period of time because overdoses, emergency room visits, intensive care unit hospitalizations and psychiatric hospitalizations can be avoided.

Stevenson and Meares used dynamic psychotherapy twice weekly for one year. Rather than having a control group, they used a "pre-post" or "before-after" model, comparing the patients the year before they received twice-weekly psychotherapy to the same patients the year after they received a year of twice-weekly psychotherapy. They found that time spent away from work for the year after therapy was one-third the amount of time spent during the year before therapy. The number of hospital admissions decreased by 59 percent compared to pretreatment rates, and overall time spent as an inpatient decreased by half. Substantial cost savings are obviously associated with these improvements. At five-year follow-up, most of these changes appeared to be quite durable.

Dr. Gabbard is Callaway Distinguished Professor at The Menninger Clinic, and clinical professor of psychiatry at the University of Kansas School of Medicine, Wichita, Kan.


  1. Colson D, Lewis L, Horwitz L. Negative outcome in psychotherapy and psychoanalysis. In: Mays DT, Franks CM, eds. Negative Outcome in Psychotherapy and What to Do About It. New York: Springer; 1985.
  2. Cowdry RW, Gardner DL. Pharmacotherapy of borderline personality disorder. Alprazolam, carbamazepine, trifluoperazine and tranyl-cypromine. Arch Gen Psychiatry. 1988;45(2):111-119.
  3. Davies JM, Frawley MG. Dissociative processes and transference-countertransference paradigms in the psychoanalytically oriented treatment of adult survivors of childhood sexual abuse. Psychoanalytic Dialogues. 1992;2(1):5-36.
  4. Fonagy P, Steele M, Steele H, et al. The Emanuel Miller Memorial Lecture, 1992. The theory and practice of resilience. J Child Psychol Psychiatry. 1994;35(2):231-257.
  5. Gabbard GO. Commentary on "Dissociative processes and transference-countertransference paradigms" by Jody Messler Davies and Mary Gail Frawley. Psychoanalytic Dialogues. 1992;2(1):37-47.
  6. Gabbard GO. Countertransference: the emerging common ground. Int J Psychoanal. 1995;76(Pt 3):475-485.
  7. Gabbard GO, Horwitz L, Allen JG, et al. Transference interpretation in the psychotherapy of borderline patients: a high-risk, high-gain phenomenon. Harvard Review of Psychiatry. 1994;2:59-69.
  8. Gabbard GO, Lester EP. Boundaries and Boundary Violations in Psychoanalysis. New York: Basic Books; 1995.
  9. Gabbard GO, Wilkinson SW. Management of Countertransference with Borderline Patients. Washington: American Psychiatric Press; 1994.
  10. Gunderson JG, Chu JA. Treatment implications of past trauma in borderline personality disorder. Harvard Review of Psychiatry. 1993;1:75-81.
  11. Gunderson JG, Phillips KA. A current view of the interface between borderline personality disorder and depression. See comments. Am J Psychiatry. 1991;148(8):967-975.
  12. Heard H. Behavior therapies for borderline patients. Presented at American Psychiatric Association Annual Meeting. May 21-26, 1994; Philadelphia.
  13. Horwitz L, Gabbard GO, Allen JG, et al. Borderline Personality Disorder: Tailoring the Psychotherapy to the Patient. Washington: American Psychiatric Press; 1996.
  14. Howard KI, Kopta SM, Krause MS, Orlinsky DE.The dose-effect relationship in psychotherapy. Am Psychol. 1986;41(2):159-164.
  15. Kernberg OF. Borderline Conditions and Pathological Narcissism. New York: Jason Aronson; 1975.
  16. Linehan MM, Armstrong HE, Suarez A, et al. Cognitive-behavioral treatment of chronically parasuicidal borderline patients. See comments. Arch Gen Psychiatry. 1991;48(12):1060-1064.
  17. Markovitz PJ, Trevidi C, Wagner S, et al. A placebo-controlled trial of fluoxetine in borderline personality disorder. Paper presented at the 31st annual meeting of the NCDEU. May 1991; Key Biscayne, Fla.
  18. Rogers JH, Widiger TA, Krupp A. Aspects of depression associated with borderline personality disorder. Am J Psychiatry. 1995;152(2):268-270.
  19. Salzman C, Wolfson AN, Schatzberg A, et al. Effect of fluoxetine on anger in symptomatic volunteers with borderline personality disorder. J Clin Psychopharmacol. 1995;15(1):23-29.
  20. Silk KR. Biological and Neurobehavioral Studies of Borderline Personality Disorder. Washington: American Psychiatric Press; 1994.
  21. Stevenson J, Meares R. An outcome study of psychotherapy for patients with borderline personality disorder. See comments. Am J Psychiatry. 1992;149(3):358-362.
  22. Waldinger RJ. Intensive psychodynamic therapy with borderline patients: an overview. Am J Psychiatry. 1987;144(3):267-274.
  23. Waldinger RJ, Frank AF. Clinicians' experiences in combining medication and psychotherapy in the treatment of borderline patients. Hosp Community Psychiatry. 1989;40(7):712-718.
  24. Zanarini M. The Role of Sexual Abuse in the Etiology of Borderline Personality Disorder. Washington: American Psychiatric Press; 1996.

Psychiatric Times - Category 1 Credit

To earn Category 1 credit, read the article, "Integrated Treatment of Borderline Personality Disorder." Complete the application for credit and mail with your $10 payment to CME, Inc. You must keep your own records of this activity. Copy this information and include it in your continuing education file for reporting purposes.

CME, Inc. is accredited by the Accreditation Council for Continuing Medical Education (ACCME) to sponsor continuing medical education for physicians. CME, Inc. designates this article for up to one hour of Category 1 credit for the Physician's Recognition Award of the American Psychiatric Association, when used and completed as designed.

CME, Inc. invites physicians to take this posttest for Category 1 credit.

  1. The only psychotherapy shown to be efficacious in borderline personality disorder through the use of random, controlled trials is:
    a. psychodynamic therapy
    b. dialectical behavior therapy
    c. rational emotive therapy
    d. systematic family therapy

  2. The two principal defense mechanisms found in patients with borderline personality disorder are:
    a. displacement and identification
    b. reaction formation and isolation of affect
    c. splitting and projective identification
    d. repression and denial

  3. The medication most useful for borderline patients who are experiencing depression is:
    a. an SSRI such as fluoxetine
    b. a benzodiazepine
    c. a tricyclic antidepressant
    d. lithium carbonate

  4. Long-term stable psychotherapy is cost-effective with borderline patients for the following reasons:
    a. It decreases the use of psychiatric hospitalization
    b. It decreases the number of emergency room visits
    c. It decreases the number of self-mutilation episodes
    d. All of the above