Psychiatry Research 226 (2015) 361–367

Contents lists available at ScienceDirect

Psychiatry Research journal homepage: www.elsevier.com/locate/psychres

Assessing the contribution of borderline personality disorder and features to suicide risk in psychiatric inpatients with bipolar disorder, major depression and schizoaffective disorder Ruifan Zeng a,n, Lisa J. Cohen b, Thachell Tanis b, Azra Qizilbash b, Yana Lopatyuk b, Zimri S. Yaseen b, Igor Galynker b a b

Department of Psychology, Long Island University-Brooklyn, 1 University Plaza, Brooklyn, NY 11201, USA Department of Psychiatry and Behavioral Sciences, Mount Sinai Beth Israel Medical Center, New York, NY 10003, USA

art ic l e i nf o

a b s t r a c t

Article history: Received 26 July 2014 Received in revised form 13 January 2015 Accepted 19 January 2015 Available online 28 January 2015

Suicidal behavior often accompanies both borderline personality disorder (BPD) and severe mood disorders, and comorbidity between the two appears to further increase suicide risk. The current study aims to quantify the risk of suicidality conferred by comorbid BPD diagnosis or features in three affective disorders: major depressive disorder (MDD), bipolar disorder (BP) and schizoaffective disorder. One hundred forty-nine (149) psychiatric inpatients were assessed by SCID I and II, and the Columbia Suicide Severity Rating Scale. Logistic regression analyses investigated the associations between previous suicide attempt and BPD diagnosis or features in patients with MDD, BP, and schizoaffective disorder, as well as a history of manic or major depressive episodes, and psychotic symptoms. Comorbid BPD diagnosis significantly increased suicide risk in the whole sample, and in those with MDD, BP, and history of depressive episode or psychotic symptoms. Each additional borderline feature also increased risk of past suicide attempt in these same groups (excepting BP) and in those with a previous manic episode. Of the BPD criteria, only unstable relationships and impulsivity independently predicted past suicide attempt. Overall, among patients with severe mood disorders, the presence of comorbid BPD features or disorder appears to substantially increase the risk of suicide attempts. & 2015 Elsevier Ireland Ltd. All rights reserved.

Keywords: Borderline personality disorder Suicide Bipolar Depression Schizoaffective Psychosis

1. Introduction Suicidal behavior, including attempted and completed suicide, remains a prominent public health concern in the United States. According to the National Institute of Mental Health (NIMH), it was the tenth leading cause of death, at approximately 11.3 suicide deaths per 100,000 people in a single year in 2007, with 11 attempted suicides per death (NIMH, 2007). In the first few months following hospital discharge, and even within the first week, inpatients are drastically more likely to re-attempt suicide (Pirkola et al., 2005; Yaseen et al., 2014). In fact, the adjusted risk ratio for completed suicide within the first week following inpatient discharge is staggering: a 102-fold and 246-fold increase for men and women, respectively, compared to those who have never been hospitalized (Qin and Nordentoft, 2005). This underscores the importance of adequately identifying those patients in an inpatient population who may be at greatest risk for suicidal behavior post-discharge.

n

Corresponding author. Tel.: þ 1 718 488 1068. E-mail address: [email protected] (R. Zeng).

http://dx.doi.org/10.1016/j.psychres.2015.01.020 0165-1781/& 2015 Elsevier Ireland Ltd. All rights reserved.

Risk factors for suicide attempts include prior attempts, substance use, and mood and personality disorders (Angst et al., 2005; Bolton et al., 2010; Rihmer, 2007). In particular, comorbidity of borderline personality disorder (BPD) with mood disorders in adults and adolescents seems to confer greater risk for increased suicidal behavior, including self-harm and attempted suicide (Bolton et al., 2010; Moor et al., 2012). The combined influence of BPD and a major depressive episode (MDE) has also been associated with a greater number of suicide attempts and serious objective planning (Soloff et al., 2000). Further, adult outpatients with MDD and comorbid BPD had a higher number of previous suicide attempts compared to patients with MDD alone (Galione and Zimmerman, 2010). In addition, depressed bipolar patients with comorbid BPD demonstrated more mood lability and irritability, as well as increased history of suicide attempts (Perugi et al., 2013). In general, BPD appears to increase suicidal risk in those with MDD and bipolar disorder, though none of the above mentioned studies consider the impact of BPD on other severe mental illnesses, such as schizoaffective disorder or schizophrenia. There is some evidence that comorbid BPD in inpatients with schizophrenia has a negative impact on course and outcome of illness compared to either diagnosis alone (Bahorik and Eack, 2010). Moreover, while the combined effect

362

R. Zeng et al. / Psychiatry Research 226 (2015) 361–367

of BPD and a depressive episode (across various psychiatric diagnoses) may increase suicidality (Soloff et al., 2000), there is considerably less research on how a BPD diagnosis affects those who are not predominantly depressed, but who are also manic or psychotic. Therefore, the current study aims to determine the differential influences of BPD on suicide risk across multiple mood disorders, including those characterized by mania or psychosis. By assessing BPD's effect on suicide risk across several mood and psychotic disorder diagnoses within the same sample, the current study can provide a more thorough and nuanced understanding of suicide risk in patients with comorbid mood disorder and personality pathology. We also examine patients based on transdiagnostic syndromes (i.e., those with a history of depressive episodes, manic episodes, or psychotic symptoms), in order to gain a clearer picture of how borderline pathology affects suicidal behavior differentially depending on mood and psychotic symptoms that may cut across discrete DSM-IV diagnoses. Further, it is important to consider the effects of BPD dimensionally rather than categorically (i.e., whether or not one meets criteria for the full diagnosis), given that severity of comorbid Cluster B features might be predictive of suicidal behavior even in subthreshold cases (Corbitt et al., 1996). Thus, in the current study we consider not only the effect on suicide risk of comorbid BPD diagnosis but also of borderline features (i.e., each borderline diagnostic criterion endorsed, independent of BPD diagnosis) in patients with diagnoses of MDD, bipolar disorder, and schizoaffective disorder as well as current or previous syndromes of major depressive episodes, manic episodes, and psychosis. Though schizoaffective disorder is classified as a psychotic disorder, it is characterized by major mood episodes and for the sake of simplicity, will be heretofore referred to as a mood disorder along with MDD and bipolar disorder. We hypothesized that the presence of comorbid BPD diagnosis or features would confer a significant additional risk of past suicide attempts in psychiatric inpatients with severe mood disorders, as well as a lifetime history of depressive, manic, or psychotic symptoms.

2. Methods 2.1. Subjects One hundred forty-nine (149) participants presenting with mood or psychotic disorders were recruited from two psychiatric inpatient units in a large, urban hospital. The study was approved by Mount Sinai Beth Israel's Institutional Review Board (IRB) for Human Subjects Research. All participants were between the ages of 18 and 65, English speaking, able to understand and sign an IRB-approved informed consent, and compensated for completing a 5 hour test battery. Patients met criteria for schizophrenia, schizoaffective disorder, bipolar disorder (I, II, NOS), major depressive disorder, or non-psychotic and non-bipolar mood or anxiety disorder. Patients were excluded if they were experiencing acute psychosis, mania, and/or agitation severe enough to preclude informed consent or task performance, and if they had severe cognitive deficits (DSM-5 intellectual disabilities or neurocognitive disorders). 2.2. Measures 2.2.1. Structured Clinical Interview for DSM-IV-TR Axis I Disorders, Patient Version With Psychotic Screen (SCID-I/P) The SCID-I/P is a widely used, semi-structured interview for determining the major Axis I DSM-IV-TR diagnoses (First et al., 2002). A trained clinician administered modules assessing for manic episodes (past or present), major depressive episodes (past or present), and psychotic symptoms (i.e., delusions, hallucinations, disorganized speech and behavior, negative symptoms), in order to make a diagnosis of schizophrenia, schizoaffective disorder, bipolar I and II disorders, major depressive disorder, or substance induced mood disorder. Interviewers were doctoral students in clinical psychology or psychiatric residents, and were trained by the study Principal Investigator (LJC) in groups of two. Raters administered two practice interviews before conducting assessments with subjects, and final ratings were determined by consensus in the research team using interview responses and any available current and past chart data. For the sake of contrast with the SCID II, this instrument will herein be referred to as the SCID-I.

2.2.2. Structured Clinical Interview for DSM-IV Axis II Disorders (SCID-II) The SCID-II is a well-established, semi-structured interview used for the assessment of DSM-IV Axis II personality disorder diagnoses (First et al., 1997). The SCID-II includes an initial self-report consisting of 119 yes or no questions. Using standardized follow-up probes, trained interviewers determined whether the patient met the personality disorder criterion associated with the items to which they answered yes. Raters kept thorough notes detailing subjects' responses to interview probes. The number of criteria met for the disorder was documented as a dimensional measure of pathology: for the current study, each criterion for a BPD diagnosis that was endorsed represented a borderline feature. A diagnosis of a disorder was made if subjects met or exceeded the required number of criteria for each disorder. SCID II interviewers were doctoral level trainees in psychology and psychiatry, or bachelors and masters level research assistants. Assessment of interrater reliability across all 10 SCID II diagnoses with the PI (LJC) on 20 SCID II's with 6 different raters yielded a Cohen's kappa of 0.856 for items and 0.825 for diagnoses. SCID I and SCID II interviews were administered by different research staff for each subject in order to reduce rater bias.

2.2.3. Columbia Suicide-Severity Rating Scale (C-SSRS) Interviewers administered the C-SSRS to assess for past and present suicidal ideation and behavior. The C-SSRS was developed to quantify the severity of suicidal ideation and behavior. It has demonstrated good convergent and divergent validity with other multi-informant scales measuring suicidal ideation and behavior in several multisite studies of adolescent suicide attempters and adults presenting at a psychiatric emergency department (Posner et al., 2011). The C-SSRS also demonstrates good sensitivity to change over time (Posner et al., 2011). The present study used the history of suicide attempt as a key variable. An actual suicide attempt according to the C-SSRS is a potentially self-injurious act committed with at least some wish to die as a direct result of that act. If any intent or desire to die is associated with the act, then it meets criteria for a suicide attempt. It is not necessary for any injury or harm to have occurred so long as there was the potential for harm. Although a patient may deny their intent to die, intent can be inferred clinically based on behavior or circumstances, such as when someone engages in behavior knowing that the action could be lethal (e.g., jumping from a window of a tall building) (Posner et al., 2011).

2.3. Statistical analysis As an initial step, demographic and clinical variables were analyzed to characterize the sample. The proportion of patients within each major mood disorder (diagnosis based on SCID-I) who also met criteria for a BPD diagnosis (based on SCIDII) was determined as well as the proportion of patients with each major syndrome (mania, depression, and psychosis) who also met criteria for a BPD diagnosis. Next, logistic regression analyses were conducted on the whole sample and then within each of the major mood disorders (MDD, bipolar, and schizoaffective). All analyses were set at an alpha level of 0.05. We first calculated the odds ratio for having a history of suicide attempt when a comorbid BPD diagnosis was present, in the sample as a whole and within each of the major mood diagnoses. Then, we calculated odds ratios for having a history of suicide attempt with each additional BPD criterion met, within the whole sample and then within each of the major mood disorders. Furthermore, given the benefits of examining psychopathology based on transdiagnostic syndromes, rooted in symptom presentation rather than discrete diagnostic categories, we conducted logistic regression analyses to determine the odds ratio for having a previous suicide attempt when a BPD diagnosis was present, in patients with a history of a major depressive episode, manic episode, and psychotic symptoms. Further, in order to examine the impact of borderline pathology dimensionally, the odds ratio for a history of suicide attempt within each syndrome when each additional borderline criterion was met was also calculated. Finally, in order to determine which individual borderline criteria were independently predictive of a history of suicide attempt, we entered eight BPD criteria (excluding criterion 5, repeated suicidal behavior) into a multiple logistic regression analysis using the whole sample.

3. Results 3.1. Demographic and clinical characteristics In total, 149 subjects were included in the analyses. The average age of subjects was 36.7 (713.39) years. The sample was 67.1% female and 30.2% identified as of Hispanic ethnicity. With regard to race, 44.6% of the sample identified as Caucasian, 22.8% African American, 12.8% Asian, 4.7% American Indian or Alaskan Native, and 14.8% Other. The average level of education was 14.18 ( 73.39) years, and 16.8% of the sample was employed full time.

R. Zeng et al. / Psychiatry Research 226 (2015) 361–367

3.2. Descriptive statistics Table 1 presents descriptive statistics on diagnostic and suicide variables. All 149 patients received a SCID-I diagnosis: 36 (24.2%) were diagnosed with MDD, 54 (36.2%) with bipolar disorder, 29 (19.5%) with schizoaffective disorder, 23 (15.4%) with schizophrenia, and seven (4.7%) with substance induced mood disorder. Thirty-seven (24.8%) of the patients were also diagnosed with comorbid BPD. Although we did not analyze patients with schizophrenia or substance induced mood disorder separately, we retained them in the whole sample analysis. We did this both to improve power and generalizability to other inpatient samples. We did not analyze the substance induced mood disorder patients separately as the n was too low (n ¼7); likewise, none of our schizophrenic patients were diagnosed with comorbid BPD.

3.3. Logistic regression analyses Logistic regressions are presented in Table 2.

363

3.3.1. Whole sample and comorbid BPD or borderline features Within the whole sample, 24.8% (n¼37) of the subjects met criteria for comorbid BPD, and 45.6% (n¼68) had a history of a past suicide attempt. Logistic regression analyses demonstrated that within the whole sample, a comorbid BPD diagnosis significantly increased the likelihood of having a past suicide attempt with an odds ratio of 6.49 (95% CI: 2.26–12.28, po0.001). The presence of each additional BPD criterion met also significantly increased the risk of past suicide attempt, with an odds ratio of 1.41 (95% CI: 1.19–1.61, po0.001). In other words, a BPD diagnosis as well as each borderline feature significantly increased the risk of previous suicide attempt in the whole sample. 3.3.2. MDD and comorbid BPD or borderline features Thirty-two subjects (25.6%) out of the whole sample were diagnosed with MDD and 38.8% (n ¼ 14) of these had a comorbid diagnosis of BPD. Logistic regression analyses revealed that MDD subjects with a comorbid diagnosis of BPD were significantly more likely to have a history of suicide attempt (OR ¼6.41, 95% CI: 1.37–30.05, p ¼0.018) compared to those with MDD alone.

Table 1 Diagnostic and clinical characteristics. SCID I diagnosis

Frequency (%) of total sample

Patients with comorbid BPD diagnosis

Patients with past suicide attempt

Patients with both BPD and past suicide attempt

Whole sample MDD Schizoaffective Bipolar disorder Schizophrenia Substance-induced mood disorder Syndromes Z 1 MDE Z 1 Manic episode Lifetime psychotic symptoms

149 36 29 54 23 7

37 14 3 16 0 4

68 19 10 27 6 6

27 11 2 11 0 3

(100%) (24.2%) (19.5%) (36.2%) (15.4%) (4.7%)

100 (67.1%) 66 (44.3%) 104 (69.8%)

(24.8%) (38.8%) (10.3%) (29.6%) (0%) (57.1%)

31 (31.0%) 14 (21.1%) 19 (18.2%)

(45.6%) (52.8%) (34.5%) (54.0%) (26.1%) (85.7%)

53 (53.0%) 29 (43.9%) 46 (44.2%)

(18.1%) (30.5%) (0.07%) (20.4%) (0%) (42.8%)

23 (23.0%) 9 (13.6%) 14 (12.8%)

BPD¼ borderline personality disorder, MDD¼ major depressive disorder, and MDE ¼major depressive episode.

Table 2 Logistic regressions – odds ratios of past suicide attempt (SA) across mood diagnoses and syndromes when comorbid BPD diagnosis or criteria are present. SCID I diagnosis

SA (%) No SA (%) Whole sample BPD 27 (75.0%) 9 (25.0%) No BPD 41 (36.6%) 71 (63.4%) MDD BPD 11 (91.6%) 1 (8.3%) No BPD 8 (36.4%) 14 (63.6% Schizoaffective BPD 2 (66.6%) 1 (33.3%) No BPD 8 (30.7%) 18 (69.3%) Bipolar disorder BPD 11 (73.3%) 4 (26.6%) No BPD 16 (42.1%) 22 (57.9%) Syndromes Z1 Major depressive episode (n¼ 100) BPD 23 (76.6%) 7 (23.3%) No BPD 30 (43.5%) 39 (56.5%) Z1 Manic episode (n ¼66) BPD 9 (69.2%) 4 (30.7%) No BPD 20 (38.5%) 32 (61.5%) Lifetime psychotic symptoms (n ¼104) BPD 14 (77.7%) 4 (22.2%) No BPD 32 (37.6%) 53 (62.4%)

OR for previous SA when BPD diagnosis present

95% C.I.

p

OR for previous SA with each additional BPD criterion

95% C.I.

p

5.27 –

2.26–12.28 –

o0.001 –

1.38 –

1.19–1.61 –

o 0.001 –

6.41 –

1.37–30.05 –

0.018 –

1.56 –

1.12–2.19 –

0.010 –

4.50 –

0.35–57.11 –

0.246 –

1.22 –

0.82–1.82 –

0.321 –

3.78 –

1.02–14.06 –

0.047 –

1.25 –

0.98–1.58 –

0.069 –

4.27 –

1.62–11.28 –

0.003 –

1.29 –

1.08–1.56 –

0.005 –

3.60 –

0.98–13.26 –

0.054 –

1.29 –

1.03–1.62 –

0.030 –

5.79 –

1.76–19.14 –

0.004 –

1.44 –

1.17–1.77 –

o 0.001 –

364

R. Zeng et al. / Psychiatry Research 226 (2015) 361–367

The presence of each BPD criterion in those with an MDD diagnosis also significantly increased the odds of past suicide attempt (OR¼1.56, 95% CI: 1.12–2.19, p¼ 0.010). 3.3.3. Bipolar disorder and comorbid BPD or borderline features Fifty-four subjects (36.2%) were diagnosed with bipolar disorder and 29.6% (n¼16) of those with comorbid BPD. Logistic regression analyses indicated that bipolar subjects with comorbid BPD were significantly more likely to have a history of suicide attempt than bipolar subjects without BPD, with an odds ratio of 3.78 (95% CI: 1.02–14.06, p¼0.047). The presence of each BPD criterion in those with a bipolar disorder diagnosis marginally increased the odds of suicide attempt. In order to assess the effect of comorbid BPD on suicide risk in patients with bipolar disorder independent of depression history, logistic regression analyses were repeated covarying for the number of past depressive episodes. Even when controlling for previous number of depressive episodes, patients with bipolar disorder and comorbid BPD were still more likely to have had a previous suicide attempt (AOR¼3.98, 95% CI: 1.04–15.24, p¼0.044). Each additional borderline criterion in bipolar patients marginally increased the odds of past suicide attempt when controlling for previous depression (AOR¼1.22, 95% CI: 0.97–1.55, p¼0.096). Thus, the risk of past suicide attempt in bipolar patients with comorbid BPD appears to persist independent of depressive history. 3.3.4. Schizoaffective disorder and comorbid BPD or borderline features Twenty-nine subjects (19.5% of the sample) were diagnosed with schizoaffective disorder and 10.3% (n¼ 3) of those with comorbid BPD. Logistic regression analyses indicated that neither the presence of BPD nor of each additional BPD criterion significantly increased the risk of past suicide attempt in subjects with schizoaffective disorder. 3.4. Logistic regression analyses by syndrome 3.4.1. Major depressive episode and comorbid BPD or borderline features One-hundred (67.1%) subjects had a history of at least one major depressive episode (MDE), and 14 (31.0%) of those had comorbid BPD. Logistic regression analyses indicated that subjects with a history of at least one MDE as well as comorbid BPD were significantly more likely to have a history of suicide attempt than MDE patients without BPD, with an odds ratio of 4.27 (95% CI: 1.62–11.28, p ¼0.003). Further, the presence of each BPD criterion in those with a history of MDE significantly increased the odds of suicide attempt, with an odds ratio of 1.29 (95% CI: 1.08–1.56, p ¼ 0.005). 3.4.2. Manic episode and comorbid BPD or borderline features Sixty-six (44.4%) subjects had a history of at least one manic episode and 14 (21.1%) of those had comorbid BPD. Logistic regression analyses indicated that subjects with a history of a single manic episode and comorbid BPD were marginally more likely to have a history of suicide attempt than such subjects without comorbid BPD. However, the presence of each BPD criterion in those with a history of mania significantly increased the likelihood of a history of suicide attempt (OR¼1.29, p¼0.030). In order to assess the unique effect of comorbid BPD and mania on suicide risk (independent of depression history, which co-occurs with mania in many patients), logistic regression analyses were repeated covarying for the number of past depressive episodes. When controlling for previous number of depressive episodes, patients with a history of mania and comorbid BPD were found to be marginally more likely to have had a previous suicide attempt

than those without comorbid BPD, with the findings closely approaching significance (AOR¼ 3.75, 95% CI: 0.98–14.36, p¼0.053). Similarly, each BPD criterion was marginally significant in increasing the odds of having a history of suicide attempt after controlling for previous depression, with the findings closely approaching significance (AOR¼ 1.25, 95% CI: 1.00–1.58, p¼0.054). 3.4.3. Psychotic symptoms and comorbid BPD or borderline features One hundred and four (69.8%) subjects had a history of psychotic symptoms and 19 (18.2%) of those had comorbid BPD. Logistic regression analyses indicated that subjects with a history of psychosis plus comorbid BPD were significantly more likely to have a history of suicide attempt than those without comorbid BPD, with an odds ratio of 5.79 (95% CI: 1.76–19.14). Also, the presence of each BPD criterion in patients with a history of psychosis increased the odds of having had a past suicide attempt by about 44% (OR¼ 1.44, 95% CI: 1.17–1.77, po0.001). 3.5. Logistic regressions for subjects not admitted for suicide attempt or ideation In order to account for the possibility of a hospital admission bias due to the fact that suicidality is one of the most common reasons for hospitalization in borderline patients (Hull et al., 1996), which may not be true for patients with severe mood disorders (who may be admitted for disturbances in mood or psychotic symptoms unrelated to suicidality), we repeated the logistic regressions including only subjects who were not admitted for suicide attempts. Among the subjects of the total sample who were not admitted to inpatient care for a suicide attempt (123; 84.8% of the 145 patients for whom we had this information) the presence of a comorbid BPD diagnosis marginally increased the odds of having a previous suicide attempt (OR¼2.50, 95% CI: 0.91–6.89, p ¼0.077). However, each single BPD criterion did increase the risk of past suicide attempt in these patients (OR¼ 1.24, 95% CI: 1.06–1.46, p¼0.009). Of subjects who were not admitted for having current suicidal ideation (91; 62.7% of 145 subjects), a comorbid BPD diagnosis significantly increased the odds of having a previous attempt (OR¼ 3.83, 95% CI: 1.13–12.94, p¼ 0.031) as did each additional BPD criterion (OR¼1.40, 95% CI: 1.11–1.76, p¼0.005). 3.6. Results after removing BPD criterion 5 (suicidal behavior) In addition, because criterion 5 for BPD (recurrent suicidal behavior, gestures or threats or self-mutilating behavior) poses a potential confound for assessment of the impact of comorbid BPD features on suicide risk, we repeated the analyses of BPD features with criterion 5 removed. The results were largely the same: for the whole sample (OR¼1.43, 95% CI: 1.19–1.71, po0.001), for patients with MDD (OR¼1.48, 95% CI: 1.02–2.13, p¼0.034), for patients with bipolar disorder (OR¼ 1.38, 95% CI: 1.03–1.84, p¼0.031), for patients with schizoaffective disorder (OR¼ 1.04, 95% CI: 0.62–1.74, p¼0.880), for patients with at least one major depressive episode (OR¼1.33, 95% CI: 1.07–1.65, p ¼ 0.009), for patients with a manic episode (OR¼1.28, 95% CI: 0.98–1.68, p¼0.069), and for patients with history of psychotic symptoms (OR¼ 1.47, 95% CI: 1.16–1.85, p¼0.001). Thus to the extent each successive BPD criterion confers additional risk of suicide attempt, it appears to do so independently of criterion 5. 3.7. Multiple regression analysis of independent effect of each BPD criterion in predicting previous suicide attempts Finally, we assessed the independent contribution of each BPD criterion in predicting previous suicide attempts. A logistic regression model including 8 BPD criteria (excluding criterion 5) indicated that

R. Zeng et al. / Psychiatry Research 226 (2015) 361–367

criterion 2 (intense and unstable interpersonal relationships) (β ¼2.88, 95% CI: 1.04–7.99, p¼0.042) and criterion 4 (impulsivity) (β ¼2.77, 95% CI: 1.17–6.54, p ¼ 0.020) significantly predicted a lifetime history of suicide attempts. The overall regression model was significant at the p¼0.001 level according to the model chisquare statistic, and accounted for 22.4% of the variance [Nagelkerke R2 ¼0.224, χ2 (8, n¼ 143) ¼26.37, p¼0.001].

4. Discussion The current study assessed the hypothesis that the presence of comorbid BPD diagnosis or features in psychiatric inpatients with several major mood disorders, including MDD, bipolar disorder, and schizoaffective disorder, would significantly increase the risk of suicide attempts as reflected in incidence of past suicide attempts. Our hypothesis was partially supported, such that comorbid BPD diagnosis was associated with significantly increased odds of having a past history of suicide attempts in the sample as a whole, in patients who met criteria for MDD or bipolar disorder, and in those who had a history of a major depressive episode or psychotic symptoms. The effect of comorbid BPD diagnosis on suicide risk closely approached significance in patients with a past manic episode. Likewise, the presence of each additional BPD criterion significantly increased the likelihood of past suicide attempt in the whole sample, in patients with MDD, and those with a history of major depressive episode, manic episode or psychotic symptoms. In bipolar patients, single borderline features were only marginally significant in increasing this risk. Neither comorbid BPD diagnosis or features had a significant effect on suicide risk in patients with a diagnosis of schizoaffective disorder. Within the whole sample, the effect of BPD diagnosis or features on suicide risk persisted even after excluding patients who had been admitted for either suicide attempt or ideation. Moreover, these results cannot be ascribed to BPD criterion 5 (repeated suicidal behavior), as the findings were largely unchanged after criterion 5 was removed from the analysis. Overall, this points to the clinical utility of conducting a comprehensive assessment of personality when considering suicide risk in patients with severe mood disorders. While there has been considerable research on the increased rates of suicide with comorbid BPD and current unipolar and bipolar depression (Angst et al., 2005; Perugi et al., 2013; Soloff et al., 2000), the present study investigated the likelihood of increased risk both across mood diagnoses as well as in the context of previous mood and psychotic symptoms independent of diagnosis. As expected, the current findings revealed significantly greater odds of suicide attempt in patients with comorbid BPD and MDD or bipolar disorder. However, the effect on suicide risk of borderline features in patients with bipolar disorder and of comorbid BPD diagnosis in those with one or more manic episodes closely approached statistical significance. It is likely that a slightly larger sample would have produced significant findings, given our findings of increased suicide risk in patients with comorbid BPD and bipolar disorder, as well as previous research indicating that such comorbidity increased risk of suicidal behavior in adolescents and adults (Moor et al., 2012; Ucok et al., 1998). Moreover, that the effect of comorbid BPD on suicide risk in bipolar patients remained largely unchanged when we controlled for frequency of depressive episodes suggests a unique effect of comorbid BPD on bipolar disorder independent of history of depression. The current study also found a significant relationship between increased suicide attempts and comorbid BPD and psychosis, which has received less examination in the previous literature. Prior research on inpatients with schizophrenia or schizoaffective disorder and comorbid BPD indicated a long-term negative impact of borderline characteristics on outcome (Bahorik and Eack, 2010). Due to a small sample of schizoaffective patients with a history of

365

suicide attempts as well as comorbid BPD, we were unable to adequately assess the effect of borderline features or diagnosis on increased suicide risk in this group. Nevertheless, as expected, the presence of BPD in those with a lifetime history of psychotic symptoms increased the likelihood of previous suicide attempts. Our results also indicated the considerable influence of subthreshold number of borderline features on increasing suicide risk: each BPD criterion significantly increased the odds of a previous suicide attempt in those who had an MDD diagnosis, as well as those with a previous depressive episode or psychosis. Even when BPD criterion 5 (recurrent suicidal behavior) was removed, these patterns of increased suicide risk held true. These results support our hypotheses, and are in line with previous research indicating that number of BPD criteria and overall severity of Cluster B pathology were superior predictors of previous suicidality in inpatients than symptoms of depression (Corbitt et al., 1996; Galione and Zimmerman, 2010; Soloff et al., 2000). Likewise, the past suicide attempt rate for inpatients has previously been found to increase 33% for every unit increase in BPD features (Stringer et al., 2013). Indeed, our own data suggests a 38% increase in past suicide attempt rate for each additional BPD criterion. Furthermore, we found that impulsivity and unstable interpersonal relationships were the only BPD criteria that independently predicted a past history of suicide attempt (excluding repeated suicidal behavior). In prior research, the BPD criteria of impulsivity and affective lability have been found to be significant predictors of suicide risk (Brodsky et al., 1997; Yen et al., 2004). Although these studies did not identify criterion 2 (unstable interpersonal relationships) as a predictor of suicidality, interpersonal triggers for suicide attempts did differentiate depressed attempters with and without BPD in a separate report (Brodsky et al., 2006). Likewise, the study by Yen et al. (2004) that identified affective lability as a suicide risk factor assessed each BPD criterion separately; thus we cannot know if criterion 2 would have been a significant predictor in a multivariate analysis, perhaps even mediating the effect of affective lability. Our own findings are of interest, however, because they suggest that the added suicidal risk conferred by comorbid borderline pathology is due to features that are both shared (impulsivity) and not shared (intense/unstable interpersonal relationships) with severe mood disorders. In fact, dysfunctional interpersonal function has been proposed as a gateway criterion for personality disorder in the alternative proposed model of personality disorders in DSM-5 (APA, 2013). This suggests that, in psychiatric inpatients with BPD features, the disturbance of interpersonal function poses a suicidal risk factor with that is unique to borderline pathology. Several of our hypotheses were not fully supported. We found a statistically significant increased likelihood of past suicide attempts in those with a comorbid BPD diagnosis and bipolar disorder, which is consistent with previous findings (e.g., Garno et al., 2005; Moor et al., 2012; Perugi et al., 2013), but not in those with comorbid BPD and a lifetime history of mania. While borderline features significantly increased the risk of suicide in patients with a history of mania, it did not have a significant effect in bipolar patients. However, our results were in the expected direction and closely approached significance, which may have been achieved with a larger sample. Nevertheless, our findings regarding depression clearly appear more robust than those regarding mania. This is likely because suicidality has been more strongly associated with depressive than manic episodes (Angst et al., 2005). In fact, when we regressed number of both manic and depressive episodes onto past suicide attempts, only number of depressive episodes was significantly associated with history of suicide attempt (data not shown). The current study, however, suggests that manic patients with comorbid BPD features and diagnoses are still at increased suicide risk compared to those without BPD, even after covarying for depressive history. This is of additional interest given the frequency of manic episodes with depressive features (Bottlender et al, 2004).

366

R. Zeng et al. / Psychiatry Research 226 (2015) 361–367

Also, contrary to our hypotheses, we did not find significant effects of comorbid BPD on suicide risk in patients with schizoaffective disorder. We did not have sufficient power in order to test our hypotheses, as we had only three schizoaffective patients with a BPD diagnosis. This lack of significant findings for comorbid BPD and schizoaffective disorder contrasts with our findings that comorbid BPD and borderline features significantly increased the odds of previous suicide attempts in patients with a history of psychotic symptoms. One explanation is a putatively different mechanism underlying psychosis in patients with BPD versus those with pure psychotic disorders. The DSM diagnosis of BPD lists transient psychotic symptoms under stress as a diagnostic criterion. It is therefore possible that psychosis in BPD patients may be primarily related to affective overarousal, which has been linked to the noradrenergic system (Berridge and Waterhouse, 2003), whereas in patients with psychotic disorders, psychosis may reflect dysfunctional activation of the mesolimbic dopaminergic pathways (Goto and Grace, 2007). This overaroused affective state might explicate our finding of an elevated risk of suicide attempts in patients with psychotic symptoms and BPD. This notion is consistent with prior reports of the aggravating effect of panic and anxiety on suicide risk (Katz et al., 2011; Yaseen et al., 2012).

4.1. Why does comorbid BPD increase suicide risk in patients with a mood diagnosis? Taken together, it appears that a comorbid BPD diagnosis confers greater risk for attempting suicide in patients with major mood disorders. This effect may be mediated by the impulsivity, aggression, emotional reactivity, and interpersonal difficulties characteristic of borderline pathology. Indeed, our own data suggests that borderline characteristics of unstable interpersonal relationships and impulsivity are the most predictive of a history of suicide attempts, which is in line with previous research suggesting that suicide attempters may have more features of impulsivity compared to suicide completers (Goodman et al., 2012). Impulsivity and aggression are associated with suicidal behavior in adolescents with BPD, but not in adolescents with MDD (Horesh et al., 2003). Increased hopelessness and impulsive aggression in adult patients with comorbid BPD and MDD or bipolar disorder also appears to confer greater risk for suicidal behavior (Carpiniello et al., 2011; Soloff et al., 2000). In addition, our finding that unstable and intense interpersonal relationships are also significantly predictive of a history of suicide attempts highlights an important delineation between patients with mood/personality disorder comorbidity versus those with mood diagnoses only. Indeed, while the severity of functional impairment in mood disorders such as MDD (which might have a more shortterm effect on suicide risk) may contribute significantly to increased suicide risk, the chronic nature of severe borderline pathology poses particular risk due to the cumulative impact of poor social adjustment (Soloff and Fabio, 2008). Specific components of poor social adjustment that have been linked to suicide risk include social pain (Eisenberger and Lieberman, 2004), low belonging and social alienation, and perceived burdensomeness (Ribeiro and Joiner, 2009). For mood-disordered patients with comorbid personality pathology, treatment that addresses or involves their social support systems might be particularly helpful. Likewise, family-focused treatments involving caregivers of those with bipolar disorder have been shown to be efficacious in reducing symptoms of depression among both caregivers and patients (Perlick et al., 2010). In sum, psychotherapeutic treatment is likely to be particularly important for this population, in order to address the longstanding features of their presentation that contribute to poorer coping ability, impaired social functioning, and significantly increased distress and suicide risk.

5. Limitations and future directions The conclusions of this study should be considered in the context of its limitations. Limitations to generalizability include the bias of the samples taken from inpatients in an urban hospital setting. Particularly because data on suicide attempts were retrospective, it is possible that more severely disturbed patients did not convey as complete or accurate a history of suicidal behavior as less distressed patients. On the other hand, an inpatient population generally has higher base rates of suicidality compared to an outpatient population, and is an appropriate sample in which to examine a variety of suicide risk factors. In addition, the clinical interview portion of the SCID-II consists only of probes to screening questions endorsed by patients, which might result in an underestimation of BPD diagnoses since no further follow-up would have occurred in response to inappropriately denied questions. We also had limited statistical power within Axis I diagnostic groups, as the proportion of patients with BPD was not spread evenly across diagnoses. Future research can include prospective studies as well as retrospective assessments of suicidality. In addition, the particular stimuli and mood states that trigger suicidality in patients with comorbid BPD merit detailed attention. While previous research has studied the relationship between comorbid BPD diagnosis or features and major depression, there has been somewhat less data on bipolar populations and substantially less in psychotic populations. Our study helps clarify the nature of this relationship, and demonstrated that in a single sample of inpatients with multiple mood disorder diagnoses or episodes, borderline pathology indeed plays a significant role in predicting likelihood of past suicide attempts. While many clinicians may focus on the mood diagnosis as conferring the greatest risk for suicide, the current research supports the substantial impact of comorbid BPD on increasing likelihood of suicide in patients with severe mental illnesses, including those with psychotic features. In several of these populations, even a single endorsed BPD criterion increased the odds of having had a past suicide attempt substantially. The current study highlights the importance of conducting a comprehensive assessment of co-occurring personality diagnosis and subthreshold features when considering suicide risk, over and above the severity of severe mood disturbance.

Conflict of interest The authors have no conflicts of interest to disclose.

Author contributions RZ: literature review and writing of the manuscript; LJC: principal investigator, statistical analysis, and co-writing manuscript; TT: literature review, study design and coordination; AZ and YL: subject recruitment, data collection/scoring, and database management; ZSY: statistical analysis and close editing of manuscript; IG: study conceptualization and design, instrument selection, and close editing of manuscript. All authors have approved this final version of the article.

Acknowledgments Mount Sinai Beth Israel departmental funds (PT30004055) were utilized to fund this study. Funding was originally obtained from Bristol-Meyers-Squibb for a previous study, which was completed under budget. BMS had no input into any aspect of this study or manuscript. The authors would like to thank the Beth Israel research

R. Zeng et al. / Psychiatry Research 226 (2015) 361–367

team for helping read the manuscript, as well as research assistants and externs for their help in collecting and entering data. References American Psychiatric Association, 2013. Diagnostic and statistical manual of mental disorders (5th ed.). American Psychiatric Association, Washington, DC. Angst, J., Angst, F., Gerber-Werder, R., Gamma, A., 2005. Suicide in 406 mooddisorder patients with and without long-term medication: a 40–44 years' follow-up. Archives of Suicide Research 9, 279–300. Bahorik, A.L., Eack, S.M., 2010. Examining the course and outcome of individuals diagnosed with schizophrenia and comorbid borderline personality disorder. Schizophrenia Research 124, 29–35. Berridge, C.W., Waterhouse, B.D., 2003. The locus coeruleus-noradrenergic system: modulation of behavioral state and state-dependent cognitive processes. Brain Research: Brain Research Reviews 42, 33–84. Bolton, J.M., Pagura, J., Enns, M.W., Grant, B., Sareen, J., 2010. A population-based longitudinal study of risk factors for suicide attempts in major depressive disorder. Journal of Psychiatric Research 44, 817–826. Bottlender, R., Sato, T., Kleindienst, N., Strauß, A., Möller, H., 2004. Mixed depressive features predict maniform switch during treatment of depression in bipolar I disorder. Journal of Affective Disorders 78, 149–152. Brodsky, B.S., Groves, S.A., Oquendo, M.A., Mann, J.J., Stanley, B., 2006. Interpersonal precipitants and suicide attempts in borderline personality. Suicide and LifeThreatening Behavior 36, 313–322. Brodsky, B.S., Malone, K.M., Ellis, S.P., Dulit, R.A., Mann, J.J., 1997. Characteristics of borderline personality disorder associated with suicial behavior. American Journal of Psychiatry 154, 1715–1719. Carpiniello, B., Lai, L., Pirarba, S., Sardu, C., Pinna, F., 2011. Impulsivity and aggressiveness in bipolar disorder with co-morbid borderline personality disorder. Psychiatry Research, 188; , pp. 40–44. Corbitt, E.M., Malone, K.M., Haas, G.L., Mann, J.J., 1996. Suicidal behavior in patients with major depression and comorbid personality disorders. Journal of Affective Disorders, 39; , pp. 61–72. Eisenberger, N.I., Lieberman, M.D., 2004. Why rejection hurts: a common neural alarm system for physical and social pain. Trends in Cognitive Sciences, 8; pp. 294–300. First, M.B., Gibbon, M., Spitzer, R.L., Williams, J.B.W., Benjamin, L.S., 1997. Structured Clinical Interview for DSM-IV Axis II Personality Disorders, (SCID-II). American Psychiatric Press, Inc., Washington, D.C.. First, M.B., Spitzer, R. L, Gibbon, M., Williams, J.B.W., 2002. Structured Clinical Interview for DSM-IV-TR Axis I Disorders, Research Version, Patient Edition With Psychotic Screen (SCID-I/P W/ PSY SCREEN). Biometrics Research, New York State Psychiatric Institute, New York. Galione, J., Zimmerman, M., 2010. A comparison of depressed patients with and without borderline personality disorder: implications for interpreting studies of the validity of the bipolar spectrum. Journal of Personality Disorders 24, 763–772. Garno, J.L., Goldberg, J.F., Ramirez, P.M., Ritzler, B.A., 2005. Bipolar disorder with comorbid cluster B personality disorder features: impact on suicidality. Journal of Clinical Psychiatry 66, 339–345. Goodman, M., Roiff, T., Oakes, A.H., Paris, J., 2012. Suicidal risk and management in borderline personality disorder. Current Psychiatry Reports 14, 79–85. Goto, Y., Grace, A.A., 2007. The dopamine system and the pathophysiology of schizophrenia: a basic science perspective. International Review of Neurobiology 78, 41–68. Horesh, N., Orbach, I., Gothelf, D., Efrati, M., Apter, A., 2003. Comparison of the suicidal behavior of adolescent inpatients with borderline personality disorder and major depression. Journal of Nervous and Mental Disease 191, 582–588.

367

Hull, J.W., Yeomans, F., Clarkin, J., Li, C., Goodman, G., 1996. Factors associated with multiple hospitalizations of patients with borderline personality disorder. Psychiatric Services 47, 638–641. Katz, C., Yaseen, Z.S., Mojtabai, R., Cohen, L.J., Galynker, I.I., 2011. Panic as an independent risk factor for suicide attempt in depressive illness: findings from the National Epidemiological Survey on Alcohol and Related Conditions (NESARC). Journal of Clinical Psychiatry 72, 1628–1635. Moor, S., Crowe, M., Luty, S., Carter, J., Joyce, P.R., 2012. Effects of comorbidity and early age of onset in young people with bipolar disorder on self harming behaviour and suicide attempts. Journal of Affective Disorders 136, 1212–1215. NIMH, 2007. Suicide Rates 2007. Retrived June 2014 from 〈http://www.nimh.nih. gov/Statistics/4SR07.shtml〉. Perlick, D.A., Miklowitz, D.J., Lopez, N., Chou, J., Kalvin, C., Adzhiashvili, V., Aronson, A., 2010. Family-focused treatment for caregivers of patients with bipolar disorder. Bipolar Disorders 12, 627–637. Perugi, G., Angst, J., Azorin, J.M., Bowden, C., Vieta, E., Young, A.H., Group, B.S., 2013. Is comorbid borderline personality disorder in patients with major depressive episode and bipolarity a developmental subtype? Findings from the international BRIDGE study. Journal of Affective Disorders 144, 72–78. Pirkola, S., Sohlman, B., Wahlbeck, K., 2005. The characteristics of suicides within a week of discharge after psychiatric hospitalisation – a nationwide register study. BMC Psychiatry 5, 32. Posner, K., Brown, G.K., Stanley, B., Brent, D.A., Yershova, K.V., Oquendo, M.A., Currier, G.W., Melvin, G.A., Greenhill, L., Shen, S., Mann, J.J., 2011. The ColumbiaSuicide Severity Rating Scale: initial validity and internal consistency findings from three multisite studies with adolescents and adults. American Journal of Psychiatry 168, 1266–1277. Qin, P., Nordentoft, M., 2005. Suicide risk in relation to psychiatric hospitalization: evidence based on longitudinal registers. Archives of General Psychiatry 62, 427–432. Ribeiro, J.D., Joiner, T.E., 2009. The interpersonal-psychological theory of suicidal behavior: current status and future directions. Journal of Clinical Psychology 65, 1291–1299. Rihmer, Z., 2007. Suicide risk in mood disorders. Current Opinion in Psychiatry 20, 17–22. Soloff, P.H., Fabio, A., 2008. Prospective predictors of suicide attempts in borderline personalitydisorder at one, two, and two-to-five year follow-up. Journal of Personality Disorders 22, 123–134. Soloff, P.H., Lynch, K.G., Kelly, T.M., Malone, K.M., Mann, J.J., 2000. Characteristics of suicide attempts of patients with major depressive episode and borderline personality disorder: a comparative study. American Journal of Psychiatry 157, 601–608. Stringer, B., van Meijel, B., Eikelenboom, M., Koekkoek, B., Licht, C.M., Kerkhof, A.J., Penninx, B.W., Beekman, A.T., 2013. Recurrent suicide attempts in patients with depressive and anxiety disorders: the role of borderline personality traits. Journal of Affective Disorders 151, 23–30. Ucok, A., Karaveli, D., Kundakci, T., Yazici, O., 1998. Comorbidity of personality disorders with bipolar mood disorders. Comprehensive Psychiatry 39, 72–74. Yaseen, Z.S., Gilmer, E., Modi, J., Cohen, L.J., Galynker, I.I., 2012. Emergency room validation of the revised Suicide Trigger Scale (STS-3): a measure of a hypothesized suicide trigger state. PLoS One 7, e45157. Yaseen, Z.S., Kopeykina, I., Gutkovich, Z., Bassirnia, A., Cohen, L.J., Galynker, I.I., 2014. Predictive validity of the Suicide Trigger Scale (STS-3) for post-discharge suicide attempt in high-risk psychiatric inpatients. PLoS One 9, e86768. Yen, S., Shea, M.T., Sanislow, C.A., Grilo, C.M., Skodol, A.E., Gunderson, J.G., McGlashan, T.H., Zanarini, M.C., Morey, L.C., 2004. Borderline personality disorder criteria associated with prospectively observed suicidal behavior. American Journal of Psychiatry 161, 1296–1298.