Introduction

The appropriate management and assessment of risk in suicidal patients with borderline personality disorder (BPD) is one of the greatest challenges in modern psychiatry. Whereas 60% to 70% of individuals with BPD attempt suicide, during the course of their illness, 5% to 10% ultimately end their lives [1]. Despite accumulating data on suicidality in BPD, the current state of knowledge offers only partial clues to help identify the BPD patients most at risk, and offers a limited armamentarium of treatment targeted to suicide prevention, leaving clinicians open to medical and legal risk and professional challenges. This article reviews the prediction of suicide risk, management, and legal implications for suicidal individuals with BPD. The paper reviewed articles from 2002 to 2011 from MEDLINE and PsycINFO databases containing the keywords “borderline personality disorder,” “suicide,” and “suicidality.” A total of 77 articles were identified, the most pertinent of which were included, along with seminal papers from earlier years.

In this review, we describe the differences between acute and chronic suicidality and distinguish among suicide completions, suicide attempts, parasuicide events, and suicidal ideation. Suicide completions are attempts that end in death, and suicide attempts are nonlethal behavioral actions with some intent to die. We further distinguish between high- and low-lethality suicide attempts. Parasuicide is defined as behaviors in which there is no intent to die that are often conducted in the service of emotional regulation and communication of distress [2]. Suicidal ideation comprises thoughts, urges, and beliefs about suicide but does not include a behavioral component.

Chronic suicidality describes the pattern of behavior exhibited by patients with BPD that is distinctly different from the acute suicidal crises seen in patients with Axis I disorders. Patients with BPD have an average of 3.3 attempts during their lifetime, and this pattern of recurrent suicide attempts has been referred to as the “behavioral specialty of BPD” [3]. The multiplicity of attempts does not preclude the potential seriousness of outcome, as the number of prior attempts predicts future attempts, attempt at lethality, and in some cases completion [4].

Assessing Suicide Risk

Suicidal behavior is multi-determined, and a myriad of risk factors have been identified, including more than 60 listed in the American Psychiatric Association (APA) Guidelines for the Assessment and Treatment of Patients with Suicidal Behavior (2003) [5], such as demographic variables, Axis I and II comorbidities, personality dimensions, traumatic events, interpersonal and social adjustment, and previous suicidality. However, none are robust predictors. Methodologies to study risk factors include retrospective analysis of suicide completers, cross-sectional studies of suicide attempters, and prospective investigations of suicide attempts. Mounting evidence suggests that suicide completers and attempters are clinically distinct groups.

Borderline Personality Disorder Suicide Completers

An important study conducted by McGirr and colleagues [6••] performed psychological autopsies on suicide completers with and without BPD and compared specific DSM BPD criteria between groups. They discovered an inverse relationship between affective instability and risk of death by suicide and the protective quality of psychotic symptomatology. Surprising was the absence of statistically different levels of anger and impulsivity in the two suicide completer groups, suggesting that the impulsive aggression present in successful suicide is not unique to BPD. This is consistent with previous reports highlighting an increased risk of successful suicide in BPD individuals with heightened levels of impulsivity, hostility, and comorbid cluster B antisocial personality disorder (ASPD) [7].

Upon further examination of the role of impulsive aggression in completed suicide, BPD suicide completers as compared with BPD non-completers were less likely to have had previous suicide attempts and hospitalization [7]. Moreover, completers used more violent methods as compared with BPD attempters, who used nonviolent methods and lacked cluster B comorbidity. Additionally, substance dependence, but not major depressive disorder (MDD), distinguished completers from attempters, prompting the authors to speculate that comorbid substance dependence might select for a more refractory form of BPD that is less amenable to treatment and symptom remission. The authors’ main conclusions centered on the necessity of impulsive aggression for lethal suicide and the conviction that BPD attempters and completers are distinct populations. The strength of the contributions by McGirr and colleagues [6••, 7] is that the data do not rely on assessment of treatment-seeking BPD patients but rather consecutive numbers of deceased suicidal individuals obtained through the coroner’s office in Montreal. One limitation of this data collection is the high proportion of males, which differs markedly from clinical populations of BPD, which are predominantly female [8], though most BPD patients who commit suicide are men [9].

Suicide completion often occurs with the first attempt. In a Finnish autopsy study of suicide attempters, 62% of males and 38% of females died with their first attempt [10]. An additional 19% of males and 39% of females had a nonlethal attempt within the year of their death [10], indicating the important role of previous attempts as a risk factor for completion, particularly for females.

Additional findings on suicide completions in BPD are available from long-term longitudinal studies of individuals with BPD. Paris [11] reported that suicide completions occurred after age 30, later in the course of the illness, after unsuccessful treatment efforts, and when the patient was alone and not involved in active treatment. This contrasts with the earlier phases of the BPD illness, when patients are engaged with the mental health system through suicide threats and attempts and therapeutic efforts. These patterns may not be as apparent in short-term studies [12] but emerge in longer follow-ups. Consistent with longitudinal follow-up over 15 years [9, 11], despite multiple threats, younger females with BPD in treatment are at low risk of death by suicide. However, this may not apply to those who are not in treatment. Lesage et al. [13], studying suicide completions in young males, found many cases of untreated BPD.

Taken together, these suicide completion studies highlight the relevance of impulsive aggression, substance dependence, and comorbid ASPD. Suicide completions in males are often first attempts, while for females, completion more likely follows a path of successive attempts and unsuccessful efforts at treatment.

Suicide Attempters

In cross-sectional studies, suicide attempters with BPD were found to have greater levels of psychopathology and depression [14], hostility and impulsivity [15], increased number of past attempts [14, 15], and first attempt at an early age [15] as compared with non-BPD suicide attempters. Additionally, BPD suicide attempters had triggers that were interpersonal in nature, while non-BPD attempters had more job- and heath-related triggers [15]. Adolescent attempters with BPD reported more lifetime sexual abuse–related stress, while MDD adolescent suicide attempters had more lifetime death-related stress [16]. Childhood sexual abuse has been reported to increase the risk of a suicide attempt 10-fold in BPD [17]. Soloff and colleagues [4] compared BPD high- and low-lethality attempters, presuming that high-lethality attempters may be a close proxy to suicide completers. High-lethality attempters were older; less educated; and had increased frequency of comorbid MDD, ASPD, and a family history of substance abuse disorder. High-lethality attempters had more frequent hospitalizations, greater intent to die, and greater number of attempts. Surprisingly, the severity of BPD and specific BPD criteria of impulsivity and aggression were not significantly different, leading the authors to conclude that suicide lethality is not specific to BPD criteria.

Contrasting findings were reported in a path analysis of high-lethality suicide attempts that derived a model highlighting the role of impulsivity in the seriousness of the current attempt and frequency of past suicidal behavior [18•]. Aggression, use of alcohol, and frequency of past suicide attempts were not related to high-lethality suicide. These results argue for a trait-related theory [19] of high-lethality suicide attempts rather than the crescendo model that posits that previous suicidality facilitates further increases in suicide severity [20].

One prospective study examined predictors of suicide attempt in BPD patients using survival analysis [21••]. Within 1 year, 19% of individuals had an attempt that was predicted by comorbid MDD and poor social adjustment. In years 2 to 5, nonspecific elements of clinical severity surfaced and poor social adjustment continued to be a predictor. These findings on suicide attempt risk factors demonstrate the shifting nature of comorbid Axis I disorders, the stability of poor social adjustment, and the lack of influence of aggression and impulsivity.

In summary, risk factors for suicide attempts in BPD differ from risk factors for suicide completion. Aggression and impulsivity are identified as important risk factors for completion but do not distinguish high- from low-lethality attempts. Although this may be due to limitations in measurement [22], attempt lethality is more likely determined by aspects of interpersonal dysfunction, poor social adjustment, and intent to die. Many first-time attempters have a lethal outcome. Better identification and understanding of these individuals, most of whom are not in clinical care, is a priority, as is further clarifying which 10% of the attempters will progress to completion over the course of their illness.

Treatment

Suicide prevention treatment options have greatly expanded over the past decade, with a multitude of psychotherapeutic interventions and pharmacologic agents targeted to treat risk factors such as impulsivity and depression, all of which are comprehensively described in a recent review [16]. Additionally, the APA Guidelines for the Assessment and Treatment of Patients with Suicidal Behavior [5]; the APA Practice Guidelines for the Treatment of Borderline Personality Disorder [1]; and Great Britain’s National Institute of Heath, Clinical Experience (NICE) Guidelines [23] offer expert opinion on the treatment and management of BPD and suicide. The APA Guidelines on Suicidal Behavior, while not specific to BPD, provide specific indications for psychiatric hospitalization for suicide risk and advocate particular agents to target components of suicide risk. The APA Guidelines for the Treatment of BPD published in 2001, now considered to be out of date, were updated in 2005 with a guideline watch [24]. Although they are comprehensive, the APA recommendations have been criticized for resulting in excessive hospitalization without clear benefit [25] and polypharmacy. The NICE guidelines conclude that medications should not be used specifically for BPD or the behaviors associated with the disorder. They instead advocate for a structured and integrated clinical approach to BPD but comment on the lack of appropriately trained clinicians in these interventions.

Oldham [26] proposes varying combinations of medication and psychotherapy approaches based on subtypes of BPD. Although the BPD guidelines recommend that suicide in BPD is best treated with psychotherapy and medication management of Axis I comorbidities and BPD features of impulsivity, aggression, mood instability, and anger, the empiric evidence for medication management of comorbidities is incomplete. All the efficacious psychotherapies for suicide teach BPD patients how to manage suicidal urges, reduce the reinforcements of self-destructive behavior, and maintain the integrity/structure of the treatment in spite of crisis. Despite the expanding evidence base of psychotherapy approaches, including dialectical behavior therapy [27], mentalization-based therapy [28], and transference-focused psychotherapy [29], a need still exists for the development of briefer and less resource-intense treatment [30], of which the STEPSS program [31], the brief dialectical behavior therapy proposed by Stanley et al. [32], or psychoeducation for BPD [33] may be models.

In terms of psychopharmacology in BPD, Leichsenring and colleagues [34] reviewed 19 randomized controlled trials published to date and found variable and inconsistent results regarding the efficacy of various antidepressants, antipsychotics, and mood stabilizers in treating particular problematic symptoms in BPD. They echo the NICE guidelines [23], highlighting the lack of definitive efficacy and potential for weight gain and medication side effects. Because currently available medications have limited efficacy, future directions will require the development of novel agents (eg, molecules that target novel sites such as μ-opioid receptors) [35].

“No-Harm” Contracts

The APA guidelines advocate for the judicious use of the therapeutic alliance to negotiate safety with skilled clinicians but caution about using the contract as a substitute for a rigorous suicide evaluation of the patient [5, 26]. Similarly, others [3, 36] believe that “no-harm” contracts are useful in that they curtail the patient from using suicidal behaviors as a means of communication with the therapist, clarify expectations, and help prevent treatment-sabotaging behaviors [37]. Other expert opinions caution that no-harm contracting should be done in the context of an established therapeutic connection, with regular follow-up and detailed documentation [26, 38•]. This kind of contracting can be in the form of a verbal agreement or a written contract in which there is a declaration that the patient agrees that he or she will not kill himself or herself or engage in self-harm activities (depending on the individual case) and what actions that individual will take should he or she feel inclined to violate this fundamental promise. Specific consequences may be outlined for particular contract violations. All recommendations regarding no-harm contracts are expert opinions, as minimal empiric evidence exists.

Hospitalization

The role of hospitalization in the management of suicidality in BPD has generated considerable controversy, with strong proponents on each side of the debate. The APA’s Practice Guidelines for Suicidal Behavior promote hospitalization and outline fairly liberal indications for admission to inpatient facilities [5]. Additional supporters advocate that hospitals keep patients safe [39] and are appropriate for treatment of comorbid Axis I exacerbations, diagnostic clarification, and simplifying medication regimens [26, 40]. However, strong dissension exists. Paris [25] points out that minimal empiric evidence exists pertaining to the efficacy of hospitalization for chronic suicidality, and that hospital stays are often “little more than a suicide watch.” In addition to having unproven effect, hospitalization can result in negative consequences for BPD patients, including behavioral regression and reinforcing behaviors that clinicians are trying to limit [41], prompting one group to advocate never to hospitalize individuals with BPD [42].

A review of the topic by Vijay and Links [40] offers an integration of these opinions and emphasizes that the paucity of evidence prevents a definitive answer. However, they suggest prudent use of hospitalization and for certain situations admit that hospitalization may be the best option available. If admission is necessary, it should be kept brief [30]. Managed care’s pressure for shortened hospital stays has inadvertently benefited the management of hospitalized suicidal individuals with BPD [42]. Alternative possibilities include 1) partial hospitalization, for which empiric data exist supporting efficacy in mentalization-based programming [28, 43]; however, accessibility to this type of specialized approach is limited; and 2) crisis intervention unit (CIU) [44]. This alternative to traditional inpatient hospitalization was an eight-bed unit with maximum length of stay of 5 days, with a focus on evaluation and brief interventions. Preliminary evidence from a randomized clinical trial was promising, with 3-month follow-up data demonstrating significantly decreased costs and lower rates of repeated suicidality and psychiatric hospitalization for the CIU treatment group compared with the group that did not have this intervention (CIU, 8% and 8%; treatment as usual, 17% and 56%). Additional research and development of alternatives to traditional hospitalization are needed for optimal management of suicidal crises in BPD.

Medicolegal Concerns

It is not surprising that the death of a patient by suicide is a devastating and unnerving experience for any clinician to face, both professionally and personally. There is a 50% risk of losing a patient to suicide during one’s career [45], and 38% of psychiatrists who lose a patient to suicide experience extreme distress [46].

Fear of litigation can become a force that may inappropriately influence clinical decision making. In fact, Krawitz and Batcheler [47] found that among a group of clinicians working with adult BPD patients, 85% admitted that within the past year, they had practiced in a manner that they believed was not likely to be in a patient’s best interest but would relieve their anxiety over medicolegal risks and other broader consequences of negative patient outcomes, such as media attention and conflict with managers and staff. This rate of defensive practice and decision making is by no means surprising, but it is detrimental to long-term patient outcomes.

Also unsettling for clinicians is that regardless of our endless efforts to identify risk factors, “predicting those who will eventually kill themselves is probably not possible” [39]. As Blinder [48] establishes, “catastrophic outcome correlates poorly with the quality of care preceding it.” Although this may vindicate clinicians of some sense of blame and responsibility, it also renders these cases quite intimidating and precarious. If we cannot accurately identify those patients who are truly more likely to complete suicide, and acting in an overly custodial fashion can be detrimental to long-term patient outcomes, what is a clinician to do?

Gutheil [49] delineates risk management approaches for working with this specific and challenging patient population. He suggests that documentation and consultation remain “the eternal mainstays of protection for the clinician,” a fact that has been described in many settings for quite some time. Beyond that, he also addresses the importance of assessing patient competence, as this “aids in distinguishing those patients who could tell the doctor about their suicidality and choose competently not to do so, from those who are, indeed, too crazy or paranoid to make that decision. To withhold vital information from the physician is the competent patient’s problem, not the physician’s.” With this competence established and documented, the patient is then cast “in the role of agent, rather than victim.” Gutheil [49] also emphasizes the value of outreach to family and making efforts to assist the family members in processing and putting aside any guilty feelings they may possess.

Krawitz et al. [50] also advocate for what they describe as “professionally indicated short-term risk-taking,” a concept championed by Maltsberger [51] 10 years earlier. They outline prerequisites for taking this kind of approach when suicidal patients with BPD present. These include that the patient is well-known to the treating clinicians and that the team has agreed upon the BPD diagnosis. A thorough assessment and risk assessment must be done, and there must be a documented chronic pattern of suicidality and/or self-harm. The team must be involved in a well-documented comprehensive treatment plan. A longitudinal risk assessment and risk–benefit analysis must be carefully developed, individualized, documented, and regularly reviewed along with the treatment plan. In specifically preparing for potential medicolegal action, the authors emphasize the importance of documenting the reasons why the specific plan was selected and considered to be the best means of minimizing risk to the patient. Additionally, obtaining second opinions and consultation is often useful and helps to ensure that patients receive an up-to-date standard of care, but this also must be well-documented. It is also important to include the patient directly in this process and family and peers in establishing the plan, and documentation of patient and family agreement and understanding is similarly essential. Ultimately, the authors acknowledge the significant uncertainty and difficulty that is inherently involved in such situations but declare that “when prerequisites are met and clinical and medicolegal practice is sound and thorough, taking short-term risk, as part of a comprehensive treatment, is a legitimate professional consideration in working with some adults with borderline personality disorder.”

Conclusions

Despite advances in our understanding of suicide risk and mounting evidence of effective therapy options, our ability to identify and treat individuals with BPD most at risk of death by suicide is remarkably limited. Table 1 provides a brief list of helpful resources. Based on the findings to date, our best recommendations regarding managing suicidality in BPD include the following:

Table 1 Helpful resources regarding BPD patients most at risk of death by suicide
Full size table

  • Appreciate the differing profiles of suicide attempters and completers, but recognize that a small subgroup of suicide attempters with particular features (eg, more lethal attempts, higher intent to die, older, failed treatment efforts, less engaged in treatment) will progress to completion.

  • Refer to the APA Guidelines for the Management of Suicidal Behaviors [5] for information and expert consensus.

  • Pay attention to features of impulsivity and Axis I comorbidities, including depression and substance abuse, and manage through appropriate medication and psychotherapeutic approaches.

  • Establish outpatient care with a strong therapeutic alliance that includes frequent assessments of suicidality and an ability to take calculated risks with the support of family and use of outside consultation as needed.

  • Judicious use of hospitalization only when necessary and as brief as possible after the consideration of alternative models, including partial hospitalization or short-term crisis unit, if available.

  • Clear documentation of decision making.

  • Recognition that 50% of psychiatrists will lose a patient to suicide during the course of their career and to seek consultation and support should it occur.