Not all patients with borderline personality disorder are the same. Although DSM-IV-TR provides a broad definition of borderline personality disorder as a "pervasive pattern of instability of interpersonal relationships, self-image, and affects, and marked impulsivity," the diagnosis is determined by the presence of any five of the nine diagnostic criteria for the disorder. By this method, there are numerous combinations of criteria that can constitute an "official" diagnosis of borderline personality disorder +(1). In addition, some patients who have clinically significant symptoms of borderline personality disorder may have fewer than the five diagnostic criteria required for an official diagnosis.

A subtyping system based on theories of the etiology of borderline personality disorder has been suggested +(2, +3) (+Table 1). Although this system has not been tested empirically, the typology suggests that suicidal symptoms (DSM-IV-TR criterion 5: "recurrent suicidal behavior, gestures, or threats, or self-mutilating behavior") would be especially prominent when dysregulation of either affect or impulse control (type 1 or type 2) predominates in the overall symptom pattern.

Personality disorders are estimated to be present in more than 30% of individuals who die by suicide, about 40% of individuals who make suicide attempts, and about 50% of psychiatric outpatients who die by suicide +(18). In clinical populations, the rate of suicide of patients with borderline personality disorder is estimated to be between 8% and 10% +(19, +20, +21), a rate far greater than that in the general population. However, since 60%–70% of patients with borderline personality disorder make suicide attempts +(7), unsuccessful suicide attempts are far more frequent than completed suicides in patients with borderline personality disorder +(22).

Self-injurious behavior is characteristic of patients with borderline personality disorder and is sometimes referred to as the borderline patient’s "behavioral specialty" +(7, +22). This type of behavior includes impulsive behavior that is potentially dangerous (e.g., excessive drinking, high-risk sexual activity), deliberate self-injurious behavior (e.g., superficial cutting or burning), suicide attempts, and completed suicide. In the literature, there is some variability in meaning for terms such as suicide attempt, suicide gesture, "parasuicide," and self-injurious behavior. Stanley and Brodsky +(23) suggested that deliberate self-harm includes two forms of self-destructive behavior. One form, suicide attempt, consists of intentionally self-destructive acts accompanied by at least a partial intent to die. This definition would include suicide "gestures." In contrast, self-injurious behavior is nonsuicidal self-injury or self-mutilation—i.e., intentionally self-destructive behavior with no intent to die. The term "parasuicide" refers to any nonlethal intentional self-injurious behavior. Stanley and Brodsky +(23) estimated that as many as 75% of patients with borderline personality disorder make at least one nonlethal suicide attempt, and even higher percentages of patients, especially of hospitalized patients, engage in self-injurious behavior. It is erroneous, however, to assume that patients with borderline personality disorder who show self-injurious behavior are not at risk for suicide. Both forms of self-destructive behavior may occur in the same patient, and it is estimated that the presence of self-injurious behavior in a given patient doubles the patient’s risk for suicide +(22).

Because of the frequency of self-injurious behavior in patients with borderline personality disorder and because such behavior is often viewed as an effort to elicit a desired response from another person, self-injurious behavior can be mistakenly thought of as willful, deliberate, and under the patient’s control. However, self-injurious behavior in patients with borderline personality disorder is incompletely understood, and it may be associated with different motivations, meanings, or goals from one patient to another or at various times in the same patient. It may produce relief of acute dysphoria +(23) and may be accompanied by analgesia, perhaps suggesting the release of endogenous opiates during acute intensification of dysphoric states +(24).

Numerous studies have identified high rates of comorbidity in patients with borderline personality disorder. Intra-axis-II comorbidity is common +(25, +26), but little is known about whether particular combinations of disorders correlate with predictable patterns of suicidal behavior. Among personality disorders, antisocial personality disorder, like borderline personality disorder, is associated with suicide risk. The estimated lifetime suicide risk for patients with antisocial personality disorder is 5% +(27, +28). However, this estimate may be low, because patients with antisocial personality disorder have a high rate of risk-taking behavior, and it is difficult to differentiate suicide from accidental death. The relative rates of suicide in patients with comorbid borderline personality disorder and antisocial personality disorder, compared to patients with either condition alone, are unclear, although Soloff et al. +(29) found a higher level of lethality of suicide attempts in patients with the comorbidity, compared to patients with borderline personality disorder alone. In clinical populations, borderline personality disorder occurs predominantly in female patients +(7), but it is more evenly distributed among males and females in the general population +(30). One possible explanation for this difference in gender ratio in clinical versus nonclinical populations is that the comorbidity of borderline personality disorder and antisocial personality disorder occurs most often in non-treatment-seeking males.

Instead of approaching the question of comorbidity by using the DSM-IV-TR categorical system, many experts in personality studies prefer a dimensional approach. Using a dimensional framework, Links and Kolla +(28) identified three personality characteristics as relevant for suicidal behavior in psychiatric patients—impulsive aggression, perfectionism, and emotional dysregulation. Of these, impulsive aggression and emotional, or affective, dysregulation are the two most relevant factors for suicidal or self-injurious behavior in patients with borderline personality disorder.

Axis I/axis II comorbidity is also common +(26, +30, +31), and there is particular interest in the comorbidity of borderline personality disorder with major depressive disorder +(28, +30, +32+–34) and with substance abuse +(28, +30, +35, +36). In a study of inpatients with borderline personality disorder, Soloff and colleagues +(37) reported that the comorbidity of borderline personality disorder and major depressive episode increased the number and seriousness of suicide attempts. They also identified impulsivity and hopelessness as independent risk factors for suicidal behavior in patients with comorbid borderline personality disorder and major depressive episode. In a patient with borderline personality disorder only, the symptoms of depression and suicidality are usually reactive to interpersonal or other real or perceived stresses and are usually of brief duration. If such a patient makes a suicide attempt, the attempt is often impulsive in nature. In contrast, in a patient with comorbid borderline personality disorder and axis I major depressive episode, symptoms of depression and suicidality develop and deepen gradually and may persist for weeks (or much longer if not treated). Such patients may show loss of appetite, sleep disturbances, loss of interest in ordinarily pleasurable activities, and other persistent signs and symptoms of depression.

The well-documented comorbidity of substance abuse with borderline personality disorder +(28, +30, +35, +36) increases patients’ risk for impulsive suicidal behavior and for impaired judgment. Because of the impaired judgment, a suicidal act made with a low level of intent to die could have lethal results in such patients. The APA Practice Guideline for the Treatment of Patients With Borderline Personality Disorder +(21) specified that patients with comorbid borderline personality disorder and substance abuse problems have a more guarded prognosis and are at greater risk for suicide or death from injury or accident, a heightened risk also noted by others +(28, +34+–36).

Risk factors for suicidal behavior in patients with borderline personality disorder are summarized in +Table 2. The risk factors most readily recognized by clinicians include a history of multiple suicide attempts, especially those with high potential lethality, and the presence of significant, persistent substance use. Despite the ability to identify meaningful risk factors in patients with borderline personality disorder, we cannot with certainty predict future suicidal behavior in an individual patient +(39, +43)—a problem that confronts clinicians in the treatment of all patient populations with potential suicide risk.

The APA Practice Guideline for the Treatment of Patients With Borderline Personality Disorder +(21) recommended psychotherapy as the primary, or core, evidence-based treatment for the disorder, whether or not suicidality is prominent in a given patient. The guideline did not endorse a specific form of psychotherapy, but two forms of psychotherapy—dialectical behavior therapy +(44) and psychodynamic psychotherapy +(45)—were reported in published randomized, controlled trials to have shown benefit in the treatment of borderline personality disorder. More recently, APA issued a Guideline Watch +(46) that summarized significant developments in the treatment of borderline personality disorder since the 2001 publication of the complete guideline. In addition to a review of new published reports on dialectical behavior therapy, the Guideline Watch described other types of psychotherapy that are being tried for the treatment of borderline personality disorder, including interpersonal therapy, cognitive therapy, cognitive analytic therapy, systems training for emotional predictability and problem solving, and transference-focused psychotherapy. Randomized, controlled trials of these treatments have not yet been reported, although several such studies are in process.

The APA practice guideline recommended symptom-targeted pharmacotherapy to be combined with psychotherapy for adjunctive benefit in the treatment of borderline personality disorder +(21). The results of randomized, controlled trials of specific medications were presented in the guideline, along with decision-tree algorithms based on clinical judgment that were organized around the predominance of affective dysregulation, impulsive-behavioral dyscontrol, or cognitive-perceptual symptoms. Newer randomized, controlled trials, as well as a number of case reports and noncontrolled medication trials, were summarized in the Guideline Watch +(46). Newer-generation atypical neuroleptics are recommended, generally in low doses, to treat cognitive-perceptual symptoms, and the selective serotonin reuptake inhibitors (SSRIs) are recommended to stabilize impulsive aggression or affective dysregulation +(21, +46). If the prescribing physician is not also the psychotherapist, effective communication between the two is essential. The adjunctive medication may be needed only for relatively brief periods for patients with borderline personality disorder alone. If an axis I comorbidity such as major depressive disorder is present, other evidence-based guidelines for the axis I condition should be used to guide pharmacotherapy decisions.

+Figure 1 portrays hypothetical relative proportions of psychotherapy and pharmacotherapy in treatment of the different subtypes of borderline personality disorder that are described in +Table 1. Gunderson +(7, pp. 150–151) emphasized that these subtypes could also reflect different phases of treatment or different levels of severity. Thus, a given patient’s predominant symptoms and corresponding need for medication could change in the course of therapy.

The importance of recognizing the possibility of suicide risk and of actively addressing suicidal behavior in the treatment of patients with borderline personality disorder is emphasized in the APA Practice Guideline for the Treatment of Patients With Borderline Personality Disorder +(21) and in the extensive clinical literature +(7, +39, +44). It is recognized that patients with borderline personality disorder have a generally elevated risk of suicide—often referred to as "chronic" suicide risk—that is periodically intensified by situational precipitants, producing "acute" suicide risk +(27, +28, +47+–49). Although patients who engage in self-injurious behavior can also be seriously suicidal, most episodes of self-injurious behavior are seen as forms of self-regulatory behavior that, though clearly an important focus of treatment, differ from behavior resulting from suicidal intent +(23). Methods for managing suicidal impulses differ from one form of psychotherapy to another, but in all forms, it is important for the therapist and patient to negotiate a plan for dealing with suicidal ideation or behavior.

At the initiation of therapy, some therapists recommend to their patients that a "contract for safety" or "no suicide contract" be agreed upon +(7, +22, +48, +49). Although potentially useful, such contracts could inappropriately erode the therapist’s alertness to ongoing risk. More often, an explicit discussion about ways to deal with crises occurs early in treatment, and the patient and therapist agree on the role that each will take at such a time. In dialectical behavior therapy, the treatment is highly structured, and there is a clear expectation that the patient call the therapist freely between sessions, especially before acting on a self-injurious impulse +(50). In transference-focused psychotherapy, patients are usually asked to name a nearby emergency room or urgent care center where they will go if they become suicidal or to identify an emergency medical number that they will call +(51). A high priority of therapy will be for the patient to learn how to refrain from acting on suicidal or self-injurious impulses and to work on understanding and resolving these issues during the therapy sessions. The therapist explains that the treatment goals include protecting the regular schedule of therapy sessions rather than "putting out fires" between sessions. Regardless of the type of psychotherapy, a hierarchy of goals will be important, as illustrated in +Figure 2.

Common to most approaches to the management of suicidal or self-injurious behavior is the effort to reduce reinforcement of the behavior +(44, +49+–51). Patients often report different conscious goals for nonsuicidal self-injury (e.g., self-regulation, self-punishment, relief, distraction) than for actions with suicidal intent (to decrease the burden or to make things better for others) +(52). These thoughts should be explicitly explored in therapy in order to develop alternative, nondestructive options. It is important to keep in mind that patients may have other layers of motivation beyond their conscious awareness, and these motivations may differ from those that are directly identified. Even if the patient has complex motivations for self-destructive behavior, the goal of developing new skills that do not rely on self-harm remains clear.

The use of hospitalization during the treatment of patients with borderline personality disorder is controversial. Long-term inpatient treatment of patients with borderline personality disorder is now largely unavailable, and there is a reasonably clear clinical consensus that inpatient care should be minimized. Still, decisions about level of care will need to be made, and hospital-based care may be recommended at times +(7, +21). Such recommendations are not evidence-based but are derived from expert clinical opinion. The APA Practice Guideline for the Treatment of Patients With Borderline Personality Disorder +(21) suggested indications for partial hospitalization, brief hospitalization, and extended inpatient hospitalization (in the guideline section on initial assessment and determination of the treatment setting).

Several experts have cautioned against use of hospitalization in the treatment of patients with borderline personality disorder, particularly for management of or in reaction to self-injurious behavior +(22, +28, +47, +50, +53+–55). Nonetheless, there may be periods in a patient’s treatment when suicidal ideation is prominent, and a period of inpatient treatment may be essential. This may be especially true when patients with borderline personality disorder have clear episodes of a comorbid axis I condition such as major depressive disorder or bipolar disorder. It is worth noting that such patients, even when they are in the throes of a major axis I episode, still have borderline personality disorder, and the behavioral issues that are attributable to the patients’ personality pathology may persist, complicating the treatment of the axis I episode.

Suicide risk is a frequent companion in the treatment of patients with borderline personality disorder, and it represents a challenge for the patient and the therapist alike. In the early stages of therapy—whether dialectical behavior therapy, cognitive behavior therapy, psychodynamic therapy, or another form of therapy—suicide risk is often the number-one priority. Giving high priority to this concern does not imply the wish to reinforce and perpetuate the patient’s periodic preoccupation with suicide. Rather, it conveys the need to come at this preoccupation head-on, as a recognized risk, and to engage the patient’s motivation to join forces with the therapist and find a better road to travel. The preferred strategy to accomplish this goal is to develop, with the patient, a mutual plan that protects the patient’s life, the patient’s bodily integrity, and the integrity of the treatment itself. In such a plan, the therapist’s assigned role is not as a member of an emergency medical services team but, rather, as the protector of a valuable sanctuary—the therapy itself—as a place where the patient’s confusion, fear, hopelessness, and distorted perceptions can be sorted out. The patient’s assigned role is to understand and accept the importance of this opportunity and to learn to choose alternative methods to deal with the inevitable crises that arise. The stakes can be high, and risk is certainly involved, but the capacity to tolerate risk is an important asset for both the patient and the therapist +(56, +57).

In the patient described at the beginning of this article, suicidal ideation had not been a prominent ongoing concern. In such a case, it would be important to review potential risk factors for suicide, including the patient’s family history. The new appearance of suicidality in the context of ongoing treatment could represent the first time that the borderline personality disorder symptom of reactive suicidality emerged during the treatment period, or it could herald the onset of comorbid major depression. In the patient described in the case vignette, the gradual development of depression, hopelessness, and suicidal ideation (in contrast to impulsive suicidal behavior) suggests the emergence of comorbid axis I depression. Appropriate antidepressant medication might need to be started for such a patient, as the patient has already been effectively engaged in psychotherapy. Brief hospitalization could be necessary, if signs indicated an extremely high suicide risk and the patient’s hold on the lifeline of therapy seemed to be weakening. However, the therapist should have no expectation that accomplishing such a plan would be smooth sailing, because a depressive episode, superimposed on the patient’s not-yet-resolved borderline intrapsychic world, could challenge the tenuous trust being built between the patient and the therapist.

Received Oct. 4, 2005; accepted Oct. 6, 2005. From the Department of Psychiatry and Behavioral Sciences, Medical University of South Carolina. Address correspondence and reprint requests to Dr. Oldham, Department of Psychiatry and Behavioral Sciences, Medical University of South Carolina, 67 President St., Charleston, SC 29425; oldhamj@musc.edu (e-mail).