Borderline personality disorder is associated with high rates of comorbidity. Individuals with borderline personality disorder meet criteria for an average of 3.4 to 4.2 lifetime axis I disorders (1, 2), including a high prevalence of posttraumatic stress disorder (PTSD), with estimates as high as 56% (1, 3, 4). However, little research has focused on the impact of PTSD on the clinical presentation, functioning, and behavioral patterns of individuals with borderline personality disorder, and no studies have examined these issues within the subgroup of suicidal borderline personality disorder patients.
Previous research suggests that those with both borderline personality disorder and PTSD tend to be more impaired overall. Several studies have indicated that the addition of PTSD does not significantly alter the severity of borderline personality disorder (5, 6), the number of axis II disorders (6), social adjustment (6—8), general health status (7), the frequency of suicide attempts (5, 8), or hostility (7, 8). However, the addition of PTSD to borderline personality disorder is associated with higher rates of other axis I disorders, particularly anxiety disorders and major depression (7, 8), an increased risk of nonsuicidal self-injury (8), more frequent inpatient psychiatric hospitalization (6), poorer physical health (7), more impaired global functioning (6), and a higher rate of childhood sexual and physical abuse (6). Moreover, the presence of PTSD has been found to decrease the likelihood of attaining remission from borderline personality disorder over a 10-year period (9).
In this study, we sought to replicate and extend previous work by examining the similarities and differences between the clinical presentations of suicidal borderline personality disorder outpatients with and without PTSD. Nine groups of variables were investigated, including borderline personality disorder severity, trauma history characteristics, suicidal and nonsuicidal self-injury, comorbid axis I disorders, comorbid axis II disorders, emotion regulation and expressivity, psychosocial functioning, mental health treatment utilization, and physical health status and medical treatment utilization. Based on previous research, we hypothesized that with the exception of borderline personality disorder severity and comorbid axis II disorders, borderline personality disorder patients with PTSD would have greater impairment across each of these domains compared to those without PTSD.
Participants were 94 women with borderline personality disorder who were enrolled in a randomized controlled outpatient psychotherapy outcome study. The measures included in our analyses represent a portion of a larger baseline assessment. To be included in this study, patients had to meet criteria for borderline personality disorder; be 18—60 years of age; be female; have at least two intentional self-injury acts (i.e., suicide attempts or nonsuicidal self-injury) in the past 5 years, including at least one in the 8-week period prior to entering the study; and have at least one suicide attempt in the past year. (One participant was included in the study who had a suicide attempt in the past 8 weeks but no additional intentional self-injury in the past 5 years.) Participants were excluded if they met criteria for a psychotic disorder, mental retardation, or bipolar disorder; had a seizure disorder requiring medication; were mandated to treatment; or required primary treatment for another debilitating condition. All participants read and signed the informed consent form after the study procedures were explained to them.
After an initial telephone screening, potential participants underwent a series of in-person assessments for study eligibility and to gather more detailed diagnostic and pretreatment information. All assessments were conducted by independent clinical assessors who were trained on interview measures by the instrument developers or an approved trainer and then evaluated for reliability.
The Structured Clinical Interview for DSM-IV Axis II Personality Disorders (10) was used to diagnose borderline personality disorder and to compute the number of criteria met. The International Personality Disorder Examination (11) was used to confirm the borderline personality disorder diagnosis, to generate a borderline personality disorder dimensional score (range of possible scores, 0—18), and to assess all other axis II diagnoses.
The Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I; 12) was used to diagnose mood, anxiety, eating, and substance use disorders.
The self-report Traumatic Life Events Questionnaire (13) was used to assess lifetime history of 22 types of traumatic events. The three-item Childhood Experiences Questionnaire (14) was used to assess self-reported history of three types of childhood sexual abuse (unwanted sexual experiences with a person at least 5 years older with clothes on, with clothes off, and with sexual intercourse). To prevent overlap across instruments, we removed the one item from the Traumatic Life Events Questionnaire that assessed childhood sexual abuse. For both instruments, participants reported the frequency of each type of traumatic event on a 7-point Likert scale ranging from 0 (never) to 6 (more than five times) and, when relevant, the age at onset. Data from both instruments were combined to yield the following: 1) the age at onset of the earliest traumatic event, 2) the frequency of crime events (four items; e.g., robbery, stalking), 3) the frequency of physical abuse or assault events (two items: childhood physical abuse and intimate partner violence), 4) the frequency of unwanted sexual events (seven items; e.g., childhood sexual abuse, adult rape, sexual harassment), 5) the frequency of general disaster events (four items; e.g., natural disaster, motor vehicle accident), 6) the frequency of events involving witnessing trauma (four items; e.g., witnessing domestic violence, seeing another person attacked), 7) the frequency of other types of traumatic events (four items; e.g., life-threatening illness, miscarriage), and 8) the total number of all types of traumatic events.
Suicidal and nonsuicidal self-injury.
The Suicide Attempt Self-Injury Interview (15) was used to assess the frequency, intent, and severity of intentional self-injury (i.e., suicide attempts and nonsuicidal self-injury) that occurred in the past year. Several subscales were also used: 1) the lethality subscale (three items) assessing the severity and potential lethality of intentional self-injury acts; 2) the suicide intent subscale (four items) assessing the degree of suicidal intent associated with intentional self-injury acts; 3) the rescue likelihood subscale (two items) assessing the probability of intervention or discovery; 4) the interpersonal influence subscale (eight items) assessing the use of intentional self-injury as a way to get something from others (e.g., to get help, to change others' behavior); and 5) the emotion relief subscale (six items) assessing the use of intentional self-injury as a way to alleviate negative emotions. Finally, six items assessing the presence or absence of trauma-related triggers of intentional self-injury were examined (e.g., flashbacks or nightmares, talking about sexual abuse or rape).
Emotion regulation and expressivity.
Three self-report measures of emotion regulation and expressivity were used: the Difficulties in Emotion Regulation Scale (16), the Berkeley Expressivity Questionnaire (17), and the State-Trait Anger Expression Inventory (18).
Psychosocial functioning was assessed using the Global Assessment of Functioning score from the SCID-I as well as the global social adjustment score for the best week in the past month (range, 1 [very good] to 5 [very poor]) from the Social History Interview (19). The Social History Interview is an adaptation of both the psychosocial functioning portion of the Social Adjustment Scale and the Longitudinal Interval Follow-Up Evaluation base schedule (20). The Inventory of Interpersonal Problems (21) was used to measure self-reported difficulties in interpersonal relationships.
Mental health treatment utilization.
The Treatment History Interview (22) was used to measure participants' utilization of a variety of mental health services during the year prior to entering the study, including inpatient psychiatric hospitalizations, emergency department visits for psychological reasons, outpatient mental health visits, and prescribed psychotropic medications.
Physical health status and medical treatment utilization.
A medical health history questionnaire measured participants' self-reported history of current and lifetime medical problems and general health status (ranging from 0 [poor] to 2 [good]). The Treatment History Interview assessed medical treatment utilization in the past year, including hospitalizations for medical reasons, emergency department visits for medical reasons, doctor visits for medical reasons, and prescribed nonpsychotropic medications.
Dependent variables were grouped into the nine theoretically related categories listed above. For each category, a multivariate analysis of variance (MANOVA) was run, and if a significant multivariate effect for PTSD status was found, univariate tests of between-subject effects for each of the dependent variables were examined. Effect sizes are presented as partial eta-squared values (η2), for which recommended cut-offs for interpretation are 0.01 (small effect), 0.06 (medium effect), and 0.14 (large effect) (23).
Of the 94 participants, 53 (56.4%) met diagnostic criteria for current PTSD. Patients with and without PTSD did not significantly differ on any demographic characteristics (Table 1).
Demographic Characteristics of 94 Women With Borderline Personality Disorder With and Without PTSDaa
| Add to My POL
|Characteristic||Borderline Personality Disorder Subgroup||Total (N=94)|
|PTSD Present (N=53)||PTSD Absent (N=41)|
| Asian American||2||3.8||3||7.5||5||5.4|
|Single, divorced, separated, or widowed||42||79.2||37||92.5||79||84.9|
| <High school||3||5.7||4||9.8||7||7.4|
| High school graduate or equivalency||4||7.5||4||9.7||8||8.5|
| Some college or technical school||32||60.4||23||56.1||55||58.5|
| College graduate||14||26.4||10||24.4||24||25.5|
Comparisons of Borderline Personality Disorder Patients With and Without PTSD
Figure 1 summarizes the results of the between-group comparisons.
Summary of Major Findings in Comparisons of 94 Women With Borderline Personality Disorder With and Without PTSD
aFinding approached significance (p=0.07).
Borderline personality disorder severity.
There was no significant multivariate effect for PTSD status on the number of borderline personality disorder criteria met and the borderline personality disorder dimensional score.
There was a significant multivariate effect for PTSD status (Wilks's lambda=0.78; F=2.67, df=7, 68, p=0.02) when comparing the two groups on trauma history variables. Borderline personality disorder patients with PTSD reported a greater number of total traumatic events and unwanted sexual events (Table 2).
Comparisons of 94 Women With Borderline Personality Disorder With and Without PTSD on Trauma History, Suicidal and Nonsuicidal Self-Injury, Current Axis I Disorders, and Emotion Regulation and Expressivity
| Add to My POL
|Variable||Borderline Personality Disorder Subgroup||Analysisaa|
|PTSD Present (N=53)||PTSD Absent (N=41)|
|Age at first trauma (years)||7.42||4.19||9.48||8.65||0.17||0.03|
|Physical abuse or assault events||5.29||4.23||3.48||3.72||0.06||0.05|
|Unwanted sexual events||16.58||9.39||9.35||10.28||<0.01||0.12|
|General disaster events||1.84||2.03||2.35||2.29||0.31||0.01|
|Events involving witnessing trauma||6.16||4.59||5.42||3.47||0.45||0.01|
|Other types of traumatic events||2.31||2.58||2.61||3.57||0.67||<0.01|
|Total traumatic events||36.04||15.59||25.71||19.69||0.01||0.08|
|Suicidal and nonsuicidal self-injury: Suicide Attempt Self-Injury Interview|
|Suicide attempts, past year||3.00||2.67||3.17||5.09||0.83||0.00|
|Nonsuicidal self-injury acts, past year||29.73||48.61||12.92||34.52||0.07||0.04|
|Suicide intent subscale||3.28||1.97||4.52||2.51||<0.01||0.07|
|Rescue likelihood subscale||5.78||0.94||5.89||1.09||0.64||<0.01|
|Interpersonal influence subscale||1.28||1.45||0.74||1.05||0.05||0.04|
|Emotion relief subscale||3.52||1.39||3.18||1.48||0.26||0.01|
|Feeling unreal or disconnected||0.67||0.42||0.50||0.48||0.08||0.03|
|Flashbacks or nightmares||0.44||0.47||0.08||0.22||<0.001||0.18|
|Thoughts about sexual abuse or rape||0.30||0.41||0.08||0.25||<0.01||0.10|
|Thoughts about physical abuse or assault||0.15||0.33||0.07||0.24||0.23||0.02|
|Talked to someone about sexual abuse or rape||0.11||0.25||0.03||0.16||0.05||0.04|
|Talked to therapist about sexual abuse or rape||0.04||0.13||0.00||0.00||0.07||0.04|
|Comorbid axis I disordersbb|
|Agoraphobia without panic disorder||0.09||0.29||0.00||0.00||0.04||0.04|
|Generalized anxiety disorder||0.06||0.23||0.15||0.36||0.15||0.02|
|Substance use disorders||0.42||0.50||0.34||0.48||0.47||0.01|
|Emotion regulation and expressivity|
|Difficulties in Emotion Regulation Scale score||131.64||20.29||119.51||21.02||<0.01||0.08|
|Berkeley Expressivity Questionnaire|
| Negative expressivity subscore||3.96||1.09||4.04||1.15||0.75||0.001|
| Positive expressivity subscore||5.19||1.18||5.68||1.02||0.04||0.05|
|State-Trait Anger Expression Inventory|
| Anger-in score||23.06||4.02||21.00||4.90||0.03||0.05|
| Controlled anger score||20.13||5.39||17.78||4.93||0.03||0.05|
| Anger-out score||17.91||5.55||20.10||5.98||0.07||0.04|
| Anger expression score||36.83||11.35||39.32||10.75||0.28||0.01|
Suicidal and nonsuicidal self-injury.
There was a significant multivariate effect for PTSD status on suicide and nonsuicidal self-injury variables (Wilks's lambda=0.67; F=2.98, df=13, 78, p=0.001). The patients with PTSD reported less suicide intent when engaging in intentional self-injury (i.e., suicide attempts and nonsuicidal self-injury), lower lethality associated with intentional self-injury, greater use of intentional self-injury for interpersonal influence reasons, and a higher frequency of intentional self-injury being triggered by flashbacks or nightmares, thoughts about sexual abuse or rape, and talking to someone about sexual abuse or rape (Table 2). There was also a trend indicating that patients with PTSD had engaged in more frequent nonsuicidal self-injury in the past year (29.7 acts compared with 12.9 acts, p=0.07).
To better understand these findings, a post hoc MANOVA was conducted that included these same variables but considered only suicide attempts (that is, nonsuicidal self-injury acts were excluded). There was a significant multivariate effect for PTSD status in this model (Wilks's lambda=0.68; F=3.00, df=12, 76, p=0.002). The results indicated that patients with PTSD were more likely to attempt suicide for interpersonal influence reasons (p=0.01, η2=0.07) and that suicide attempts were more likely to be preceded by flashbacks or nightmares (p<0.001, η2=0.15) and thoughts about sexual abuse or rape (p=0.001, η2=0.12). Notably, borderline personality disorder patients with and without PTSD did not differ in terms of the frequency, intent, or lethality of suicide attempts.
Comorbid axis I disorders.
There was a significant multivariate effect for PTSD status on current axis I disorders, excluding PTSD (Wilks's lambda=0.72; F=3.67, df=9, 84, p=0.001). Patients with PTSD had higher rates of panic disorder, agoraphobia without panic disorder, and obsessive-compulsive disorder (Table 2).
Comorbid axis II disorders.
The multivariate effect of PTSD status on current rates of axis II disorders (excluding borderline personality disorder) was nonsignificant.
Emotion regulation and expressivity.
There was a significant multivariate effect for PTSD status on measures of emotion regulation and expressivity (Wilks's lambda=0.82; F=3.29, df=6, 87, p=0.006). Patients with PTSD reported greater emotion dysregulation and anger suppression and less expression of positive emotions (Table 2).
The multivariate effect of PTSD status on measures of psychosocial functioning was nonsignificant.
Mental health treatment utilization.
The multivariate effect of PTSD status on mental health treatment utilization in the past year was nonsignificant.
Physical health status and medical treatment utilization.
There was a nonsignificant multivariate effect for PTSD status on measures of physical health status and medical treatment utilization.
This study adds to a growing body of evidence indicating that individuals with borderline personality disorder and co-occurring PTSD are likely to have more complex clinical presentations than those without PTSD. In patients with both disorders, scores for suicide intent and lethality were lower when averaged across both suicide attempts and nonsuicidal self-injury episodes. This is likely accounted for by the trend-level finding indicating that patients with both disorders engaged in more frequent nonsuicidal self-injury. When comparing suicide intent and lethality for suicide attempts only, there was no difference between the two groups. This is in contrast to a previous study (14) finding that women with borderline personality disorder who had a history of childhood sexual abuse reported more lethal self-injurious behavior than those without such a history, suggesting that it may be childhood sexual abuse, not PTSD, that is predictive of more lethal self-injurious behavior in this population.
Borderline personality disorder patients with and without PTSD also differed in the function and triggers of intentional self-injury. Those with PTSD were more likely to report engaging in intentional self-injury as a way to influence others. They were also more likely to endorse a variety of trauma-related triggers for their episodes of intentional self-injury, including flashbacks, thoughts about sexual trauma, and talking to someone about sexual trauma. It is possible that women with both disorders engage in higher rates of nonsuicidal self-injury because they have a greater number of potential triggers for these behaviors as well as a heightened reactivity to these trauma-related cues. This may suggest that for some individuals with these co-occurring disorders, addressing PTSD criterion behaviors (e.g., trauma cue reactivity) may be necessary in order for functionally related suicidal and self-injurious behavior to decrease. Short-term solutions, such as distress tolerance skills to tolerate difficult emotions and substituting alternative, nonharmful behaviors to manage triggers in more adaptive ways, may be useful in this area (24). However, the long-term solution for this problem will likely require the resolution of PTSD through more targeted treatments.
Patients with PTSD scored significantly higher on measures of emotion dysregulation and anger suppression and lower on expression of positive emotions. Difficulty expressing positive emotions is consistent with the PTSD diagnostic criterion of restricted range of affect, which typically involves an inability to experience loving and intimate feelings. In addition, theories of PTSD emphasize the extremes in emotional responding that are characteristic of this disorder, including an intrusion phase of intense emotional experiencing (e.g., flashbacks, distressing memories of the trauma) that may trigger emotional numbing (25). This vacillation between overwhelming emotions and emotional numbing that is common in PTSD may exacerbate the emotion dysregulation that is a core feature of borderline personality disorder. Moreover, this increased emotion dysregulation may contribute to the higher rate of nonsuicidal self-injury among borderline personality disorder patients with PTSD given that nonsuicidal self-injury most often functions to alleviate negative affect (26).
Patients with both borderline personality disorder and PTSD were more impaired in terms of axis I comorbidity. They were more likely to meet criteria for panic disorder (53% compared with 22%), obsessive-compulsive disorder (21% compared with 0%), and agoraphobia without panic disorder (9% compared with 0%). This is consistent with previous research demonstrating that PTSD has the highest and most diverse rate of comorbid disorders (27) as well as data from the National Comorbidity Survey indicating that women with PTSD are at particularly high risk for developing co-occurring panic disorder (28). However, in contrast to findings of previous research (29), we did not find that PTSD was associated with a higher prevalence of any mood, substance use, or eating disorder. This discrepancy is likely due to the generally high rate of axis I comorbidity found in individuals with borderline personality disorder (2, 30), as in our sample—in which, for example, high rates of co-occurring major depressive disorder (63%—77%) and substance use disorders (34%—42%) were observed. Taken together, these findings suggest that among women with borderline personality disorder who already exhibit high rates of axis I comorbidity, PTSD further increases the risk of other specific anxiety disorders.
Women with borderline personality disorder who had PTSD reported significantly more total traumatic events as well as nearly twice as many past unwanted sexual experiences (e.g., childhood sexual abuse, adult rape, sexual harassment) as their counterparts without PTSD (16.6 compared with 9.4), a finding consistent with previous research (6). Women with or without PTSD reported a high total incidence of trauma exposure, averaging 26 to 36 lifetime traumatic events. It will be important to determine the factors that protect some individuals with borderline personality disorder from developing PTSD despite such high rates of trauma exposure.
Mary: Borderline Personality Disorder With PTSD
Mary sought treatment because, she said, "I am cutting on myself, having flashbacks, feeling suicidal—if anything goes wrong, [suicide] is the first solution I think of." Mary was severely and repeatedly sexually and physically abused by her parents until age 12, when she was removed from their custody after attempting suicide. She developed severe PTSD during her childhood as a result of this abuse, and it had never remitted despite her spending much of her adolescence and adulthood living in inpatient and residential treatment facilities. As an adult, Mary had also been raped and beaten by a stranger and been the victim of a robbery. She had attempted suicide more than 20 times in her life via cutting and overdosing, and she was engaging in nonsuicidal cutting approximately two times per week. Mary's suicidal and nonsuicidal self-injury was primarily triggered by flashbacks and functioned to decrease her anxiety and provide her with an escape from the intense intrusive memories of her childhood abuse.
Laura: Borderline Personality Disorder Without PTSD
Explaining her reason for seeking treatment, Laura said, "I have tried to kill myself two times [in the past 3 months] and I am ready to make a change." She had attempted suicide three times in her life via overdosing and had hit herself to the point of bruising about six times. Each of these episodes was prompted by intense anger caused by interpersonal conflict, and both recent suicide attempts had occurred after fights with her boyfriend. Although she did not meet criteria for PTSD, Laura reported experiencing several traumatic events, including being stalked and physically threatened as an adult, a serious motor vehicle accident, and several natural disasters. However, she reported the most distress in relation to several subthreshold "traumas," including chronic invalidation by her parents and significant and protracted conflict with members of her extended family. As a result of these experiences, Laura exhibited pervasive distrust and suspiciousness of others, which was the cause of much instability and conflict in her relationships. She also experienced considerable self-doubt, which primarily manifested as generalized anxiety and worry about her ability to succeed in school and work contexts.
As hypothesized, no differences were found between the groups on borderline personality disorder severity or the presence of other axis II disorders. These findings are consistent with previous research that has not found borderline personality disorder patients with and without PTSD to differ in terms of the number of borderline personality disorder criteria met (5, 6) or the number of co-occurring axis II disorders (6). Our findings extend this previous research to indicate also that the two groups exhibit comparable rates of each specific axis II disorder.
Contrary to our hypotheses, borderline personality disorder patients with and without PTSD also did not differ in terms of overall psychosocial functioning, mental health treatment utilization, or physical health and medical treatment utilization. The findings related to psychosocial functioning differ from previous research that has found higher levels of psychosocial impairment in borderline personality disorder patients with PTSD than in those without PTSD (6). This discrepancy is likely due to the use of a suicidal sample, as the presence of a recent suicide attempt significantly lowers indices of functioning and may have led to a floor effect. The lack of between-group differences in treatment utilization is also discrepant with previous research (6, 7) and may be due to a number of factors. First, the sample used for this study was treatment-seeking and thus was self-selecting based on this variable. Second, given previous research that has indicated high rates of treatment utilization in borderline personality disorder compared to other disorders (31), it is also possible that a ceiling effect exists. For example, the majority of participants in our sample had gone to an emergency department for psychological reasons and had at least one psychiatric hospitalization in the past year. Finally, it is possible that the addition of PTSD to borderline personality disorder does not account for an increase in treatment-seeking behavior.
This study has important limitations. Because our sample included only female patients, only treatment-seeking patients, and only patients with recent and chronic suicidal and/or nonsuicidal self-injury and excluded patients with bipolar or psychotic disorders, our results may not be representative of individuals with borderline personality disorder and PTSD more broadly.
Our results have both theoretical and clinical implications. This study provides additional empirical support for the current diagnostic system, which considers borderline personality disorder and PTSD to be separate, although often co-occurring, disorders. Some researchers have suggested that borderline personality disorder is better described as a trauma-related condition known as "complex PTSD" (32, 33). However, conceptualizing borderline personality disorder and PTSD as the same disorder would disregard the important distinctions that exist between these two groups. Clinically, given that patients with both disorders appear to have a particularly severe and complex overall presentation, treatments for this population must be able to address these unique features (34). Results from two previous studies suggest that the addition of a high level of borderline personality disorder characteristics, although related to greater overall impairment, did not prevent individuals from making significant gains with cognitive-behavioral treatments for PTSD (35, 36). However, these studies excluded suicidal and/or self-injuring patients and did not assess for the full borderline personality disorder diagnosis. In addition, no research has yet examined whether the presence of PTSD significantly alters the course or outcome of treatments for borderline personality disorder. Thus, it will be important for future research to determine whether individuals with co-occurring borderline personality disorder and PTSD fare worse in treatments for either disorder, particularly among those who are suicidal or self-injuring.
The authors thank the clients, therapists, assessors, and staff at the Behavioral Research and Therapy Clinics for their help with this research.