Journal of Psychiatric Research 47 (2013) 1800e1808

Contents lists available at ScienceDirect

Journal of Psychiatric Research

journal homepage: www.elsevier.com/locate/psychires

Clinical, psychological and environmental predictors of prospective

suicide events in patients with Bipolar Disorder
Niki Antypa, Marco Antonioli, Alessandro Serretti*
Department of Biomedical and NeuroMotor Sciences, University of Bologna, Viale Carlo Pepoli 5, 40123 Bologna, Italy

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 14 May 2013
Received in revised form
31 July 2013
Accepted 7 August 2013

Patients with Bipolar Disorder (BD) have high rates of suicide compared to the general population. The
present study investigates the predictive power of baseline clinical, psychological and environmental
characteristics as risk factors of prospective suicide events (attempts and completions). Data was
collected from the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD) study.
3083 bipolar patients were included in this report, among these 140 (4.6%) had a suicide event (8 died by
suicide and 132 attempted suicide). Evaluation and assessment forms were used to collect clinical,
psychological and socio-demographic information. Chi-square and independent t-tests were used to
evaluate baseline characteristics. Potential prospective predictors were selected on the basis of prior
literature and using a screening analysis of all risk factors that were associated with a history of suicide
attempt at baseline and were tested using a Cox regression analysis. The strongest predictor of a suicide
event was a history of suicide attempt (hazard ratio 2.60, p-value < 0.001) in line with prior literature.
Additional predictors were: younger age, a high total score on the personality disorder questionnaire and
a high percentage of days spent depressed in the year prior to study entry. In conclusion, the present
ndings may help clinicians to identify patients at high risk for suicidal behavior upon presentation for
treatment.
2013 Elsevier Ltd. All rights reserved.

Keywords:
Bipolar
Suicide
Prospective
Predictors
STEP-BD

1. Introduction
Bipolar Disorder (BD) is characterized by a high risk of suicide
attempts and completions, resulting in a 15-fold higher risk
compared to that in the general population (Harris and
Barraclough, 1997). Approximately 10% of BD patients die from
completed suicide (Harris and Barraclough, 1997). Studies show
rates of attempted suicide up to 30% among BD patients (Novick
et al., 2010), and rates of suicidal ideation up to 56% for those
with an adult rst episode, and w74% among patients with a pediatric rst episode (Carter et al., 2003). To date, retrospective and
cross sectional studies have proposed many risk factors for suicidal
behavior in BD patients. These include demographic factors (female
gender (Nivoli et al., 2011), age of rst depressive episode (Song
et al., 2012)), comorbidities (such as anxiety disorders
(Baldassano, 2006), borderline personality (Neves et al., 2009),
substance abuse (Finseth et al., 2012) or eating disorder (McElroy
et al., 2011)), clinical features (feelings of hopelessness (Acosta
et al., 2012; Johnson et al., 2005), number of mood episodes,

* Corresponding author. Tel.: 39 051 6584233; fax: 39 051 521030.

E-mail address: alessandro.serretti@unibo.it (A. Serretti).
0022-3956/$ e see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jpsychires.2013.08.005

history of suicide attempt and mixed-state or rapid cycling BD

(Baldassano, 2006)), heritability, such as family history of suicide
attempt (Goldstein et al., 2012), social withdrawal, stressful life
events (Papolos et al., 2005). Prospective studies have shown that
suicide attempts in BD patients can be predicted by: long duration
of untreated illness (Altamura et al., 2010), cigarette smoking
(Ostacher et al., 2009), rapid cycling (Garcia-Amador et al., 2009),
early onset of Bipolar Disorder (Coryell et al., 2012), previous suicide attempt(s) (Nordstrom et al., 1995), index of depression
severity (Angst et al., 2002), alcohol or other substance abuse
(Preuss et al., 2003) and feelings of hopelessness (Maser et al.,
2002). The majority of studies to date has examined small sample
sizes, often concentrating on single, or few, predictors. Such
models, although informative, remain insufcient in predicting
individual course of suicidality.
The Systematic Treatment Enhancement Program for Bipolar
Disorder (STEP-BD) is the largest cohort study to date on BD, and
followed 4361 BD patients (Bowden et al., 2012; Sachs et al., 2003).
To date, published studies have explored risk factors of suicide in
the rst sub-samples only. In the rst 1000 patients, higher number
of psychiatric medications prior to study entry predicted future
suicide attempts (Martinez et al., 2005), while in the rst 1500
participants, previous suicide attempts and time spent depressed

N. Antypa et al. / Journal of Psychiatric Research 47 (2013) 1800e1808

were predictors of suicide attempts or completion, in a 2-year

period (Marangell et al., 2006). One study examined the effect of
lithium on suicide events in the complete sample, but found no
protective effect (Marangell et al., 2008). The present study aims to
investigate the predictive power of baseline clinical, psychological
and environmental characteristics as risk factors of prospective
suicide events (attempts or completions) in the STEP-BD cohort.
The selection of variables was based on prior research evidence of
association with suicide and using a screening analysis of all risk
factors that were associated with a history of suicide attempt at
baseline. We employed multivariate models allowing for the evaluation of unique predictors of suicide. This method may bring
forward distinct characteristics that can be used to identify patients
at high risk for suicidal behavior upon presentation for treatment.
2. Methods
2.1. Participants and study overview
STEP-BD is a prospective study, aiming to develop knowledge on
risk factors and treatment effectiveness for the management of
Bipolar Disorder and evaluate the longitudinal outcome of patients
with this disorder. Patients gave both oral and written consent for
participation, in accordance with declaration of Helsinki. The study
design was approved by the Human Subjects Panel from each site.
Exclusion criteria were being unable or unwilling to follow the
assessments, refuse to give consent or be below 15 years of age.
Patients were followed by trained psychiatrists and treated according to the evidenced based current therapy guidelines (Sachs
et al., 2003). Overall, 4360 patients were included in the STEPBD; we used data from the STEP-BD Standard Care Pathway (SCP)
for this report (n 4107). Patients in the SCP could belong to any
spectrum of BD presented for clinical care, and received pharmacological interventions as clinically indicated by the principles of
evidence-based treatment and published guidelines, updated
annually in the STEP-BD Clinicians Handbook. The protocol promoted the application of evidence-based treatments at regular
clinical sessions during treatment, according to the needs of the
patient and did not follow compliance to a specic treatment algorithm. This naturalistic study used ongoing assessments of
treatment and outcome information.
2.2. Assessments
Several measures were used to investigate patients clinical
status. The diagnosis of Bipolar Disorder, and its characteristics
were assessed with the Mini International Neuropsychiatric Interview (MINI) (Sheehan et al., 1998) and Affective Disorder Evaluation (ADE) form (Sachs, 2004), which was assessed upon study
entry together with a socio-demographic form (Sachs, 2004). The
ADE was an assessment tool that used versions of the mood and
psychosis modules from the Structured Clinical Interview for DSMIV Axis I Disorders, Patient Edition, for use by practicing clinicians.
The ADE also included systematic assessment of lifetime and recent
course of illness, including history of suicide attempt. The ADE form
was also used to assess the presence of suicide ideation at baseline;
suicide ideation was dened as the following: patients had to express symptoms, for the past 7 days, answering to the question
rate associated symptoms for the past week in a range from 2
(less likely) to 2 (very likely) in points increment depending on
how much they had thought about suicide; based on previous
suggestions (Marangell et al., 2006) we dened values from 2 to
0 as no suicide ideation and values from 0 to 2 as positive suicide
ideation. The Young mania rating scale (YMRS) (Young et al., 1978)
was used to assess severity of mania symptoms while depression

1801

severity was measured using the Montgomery and Asberg (1979)

Depression Rating Scale (MADRS). Exposure to lithium or SSRI
treatment prior the time point of the suicide event was recorded
using the Clinical Monitoring Form, a standardized form designed
as a routine progress note during the STEP-BD trial. The 60item Neuroticism Extroversion Openness Five Factor Inventory
(NEO-FFI) was used to evaluate the personality traits dened in ve
basic dimensions (Costa et al., 1992; McCrae and John, 1992).
Hopelessness was measured using Becks hopelessness scale, a 20item true/false questionnaire assessing the patients feelings of
hope toward different aspects of life, such as future, work and
accomplishment (Beck et al., 1974). The 15-item Family History
questionnaire assessed the presence of psychiatric illnesses in the
family members of the patient (Sachs, 2004). The Life Experiences
Survey (LES), a 50-item survey, investigated the impact of single life
events on the persons life (Sarason et al., 1978); negative values
from 0 (no impact) to 3 (extremely negative impact) corresponded to negative life events (values were recoded to nonnegative integers for analyses); values from 0 to 3 (extremely
positive) represented the positive life events. The Personality
Diagnostic Questionnaire (PDQ) was used to assess personality
disorders according to DSM-IV; it consists of a set of 99 true/false
statements and the total score reects a general index of personality disorder (Fossati et al., 1998). The Quality of Life Enjoyment
and Satisfaction short Form (QLES-SF) evaluated the amount of
satisfaction felt by the patients in the past week prior administration; higher scores indicate greater satisfaction (Endicott et al.,
1993).
2.3. Suicide outcome
The main suicide outcome of this report was a prospective
suicide event (dened as attempt or completion and referred to
as such from this point). Suicide events were assessed through the
Serious Adverse Events (SAE) form and/or Care Utilization form
(CU) as suggested previously (Marangell et al., 2006, 2008). The SAE
form was assessed by the clinician during all visits. Questions used
to assess suicidality in the present report were did the patient die
by suicide? and was there any suicide attempt?. An independent
safety ofcer and SAE committee of the STEP-BD reviewed all potential suicide events to ensure accurate classication. The Care
Utilization (CU) was a semi-structured interview (clinician-rated
form), administered every 3 months during the rst year and every
6 months after that, in which the patient was asked to answer
questions on the use of services, as well as on suicidal acts, such as
did you attempt suicide in the last 3 months?. Suicide attempts
were dened as self-injurious behaviors with intent for serious
harm and/or lethality. A suicide event was coded as present if a
suicide attempt or completion was documented in the SAE and/or
CU forms, and absent if no evidence of suicidal behavior was present in the two forms.
2.4. Statistical analysis
Predictors were selected on the basis of prior literature and
using a screening analysis of all risk factors that were associated
with a history of suicide attempt at baseline in the present sample.
Chi-square and independent-sample t-tests were used to identify
differences between baseline characteristic in patients without and
with a history of suicide attempt (Table 1).
For the prospective analysis, risk factors were subsequently
divided in three clusters; (i) clinical factors: age of onset of the rst
manic or depressive episode, number of mania/depressive episodes, bipolar subtype, YMRS total score, MADRS total score, SSRI
and lithium use, percentage of days depressed or anxious in the

1802

N. Antypa et al. / Journal of Psychiatric Research 47 (2013) 1800e1808

Table 1
Demographic and clinical characteristic of patients at baseline, stratied by history of suicide attempt(s).
Characteristics

Patients without history of suicide

attempts
N

Demographics
Age
Male
Race
White or Caucasian
African Americans
Native Americans, Eskimo or Aleut
Asian
No primary race
Martial status
Single
Married/living together
Divorced/separated/widowed
Baseline clinical characteristic
Age onset BD
Age >45 years
Age 30e45
Age 20e30 or <15
Age 15e19
Age Mania 1st episode
Age Depression 1st episode
Longest period of remission
YMRS total
MADRS total
Negative life events (total)
Positive life events (total)
PDQ total
Quality of life total
Cigarettes packet/day
Smoking onset-age
Bipolar subtype
BP1
BP2
BPNOS
Family history of suicide (Positive)
N Manic Episodes
0e4
5e9
10
N Depressive Episodes
0e4
5e9
10
NEO-FFI scores
Neuroticism
Openness
Conscientiousness
Extroversion
Agreeableness
a
Days Depressed Last year
a
Days Anxious Last year
Suicide ideation

1861
1860
1859

Mean/N

SD/%

40.9
862

13
46.3%

1023
1022
1022

40
348

12.1
34.1%

924
60
13
13
6

90.4%
5.9%
1.3%
1.3%
0.6%

346
345
321

34.2%
34.1%
31.7%

1702
77
6
44
14

91.6%
4.1%
0.3%
2.4%
0.8%

667
745
428

36.3%
40.5%
23.3%

1012

1074
45
237
936
667
21
17.4
6.0
6.3
14.6
13
4.7
32.4
43.1
0.3
17.3

2.4%
12.6%
49.6%
35.3%
12.2
12.0
8.23
6.34
10.5
11.3
5.5
16.1
11.4
0.8
6.8

1193
563
136
260

62.2%
29.4%
7.1%
21.1%

631
264
791

37.4%
15.7%
46.9%

539
283
888

31.5%
16.5%
51.9%

40.9
41.6
39.3
37.7
42.1
39.6%
31.0%
128

8.8
7.0
8.0
4.8
7.2
30.4
32.8
6.7%

1710

1089
1089
1089
993
989
760
760
648
702
1036
346
1091

667
965

15
74
545
440
18.6
14.8
4.4
7.9
18.7
16
5
40.8
39.5
0.5
16.6

1.4%
6.9%
50.7%
41.0%
11.01
9.4
6.3
6.9
11.0
11.7
5.8
15.4
11.1
1.2
6.7

772
241
58
209

70.9%
22.1%
5.3%
31.3%

220
143
602

22.8%
14.8%
62.4%

132
137
693

13.7%
14.2%
72.0%

45.2
41.2
37.6
37.0
40.1
48.7%
39.4%
153

8.0
7.3
8.1
5.1
7.1
29.0
33.8
14.1%

972

1137

1868
1850
1918

t/c2

SD/%

1887

1230
1686

Analysis

Mean/N

1840

1918
1918
1918
1805
1816
1304
1304
1135
1223
1837
425
1919

Patients with a history of suicide

attempts

1.8
40.9
18.1

df

2227.1
1
5

0.071
<0.001
0.03

25.6

<0.001

31.3

<0.001

5.6
6.5
6.0
6.0
9.7
5.6
1.1
10.9
6.8
4.9
1.4
27.7

2448.6
2719.5
2758.7
1891.2
2803
2062
2064
1398
1923
1496.8
769
4

<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
0.263
<0.001
<0.001
<0.001
0.138
<0.001

24.1
69.1

1
2

<0.001
<0.001

121.8

<0.001

665

1056
1043
1088

10.6
1.2
4.2
3.1
5.8
6.4
8.1
44.7

1497.6
1800
1800
1800
1800
2110.7
2279.3
1

<0.001
0.23
<0.001
0.03
<0.001
<0.001
<0.001
<0.001

SD: Standard Deviation; df: Degrees of freedom; BP1: bipolar type 1; BP2: bipolar type 2; BPNOS: bipolar not otherwise specied; MADRS: MontgomeryeAsberg Depression
Rating Scale; YMRS: Young Mania Rating Scale; NEO-FFI: Neuroticism Extroversion Openness Five Factor Inventory; PDQ: Personality Diagnostic Questionnaire.
a
Percentage of days spent depressed/anxious in the year prior study entry.

year prior to study entry, suicide ideation at baseline and psychiatric comorbid disorders (post-traumatic stress disorder (PTSD),
generalized anxiety disorder (GAD), obsessive compulsive disorder
(OCD), social phobia, agoraphobia, panic disorder with or without
agoraphobia, schizoaffective disorder, alcohol dependence/abuse,
drug dependence/abuse, bulimia nervosa purging/non-purging
type, anorexia nervosa restricting type); (ii) psychological factors:
PDQ, NEO-FFI and family history of suicide event (the latter entered
in this model due to a better log-likelihood statistic compared to
other models and on the basis of a diathesis rationale) (iii)
environmental factors: negative life events, cigarette smoking
(packets per day), QOL total score, and marital status.

We tested the validity of the predictors for each cluster separately using a Cox Regression Hazard Model, with the number of
days since study entry as time value, and suicide events
(attempted or completed suicide) as outcome. The last observation
was taken for patients without any suicide event. Since we tested a
large number of predictors that could overlap in terms of variance
explained, we ran multi-collinearity tests among variables within
each cluster. If multi-collinearity was observed, the model was run
separately for the correlated predictors. The best-t model was
maintained (evaluated using the log-likelihood statistic of the
model and the higher number of patients as criteria). The predictors
that resulted as signicant from the models of each of the three

N. Antypa et al. / Journal of Psychiatric Research 47 (2013) 1800e1808

clusters were entered in a nal joint Cox regression hazard model.

Continuous variables were dichotomized by median split, as previously suggested (Oquendo et al., 2004) (PDQ total score: patients
who scored >36 were coded as high PDQ otherwise coded as low
PDQ; age was dichotomized at 40 years). A Kaplan Meier survival
analysis was performed to demonstrate different survival curves of
patients using the strongest predictors.
3. Results
3.1. Participant ow
Of the 4107 patients enrolled in the SCP, 904 patients were
excluded because they had no follow-up data on the presence/
absence of suicide events as assessed in the Serious Adverse Events
(SAE) form and/or Care Utilization form (CU) (Marangell et al.,
2006, 2008). Of the remaining patients, 117 reported a recent suicide attempt at baseline in the CU (last 3 months) and were
excluded since this confounded with the aim of the present study,
namely, to examine the predictive value of risk factors prospectively.
Finally, 3 participants reported transsexual gender and were
excluded since gender would be a predictor of future suicide events
and this group was too small to be included (a owchart of the
selection of participants is available in Supplementary Fig. S1). This
yielded a nal sample of 3083 patients. The maximum length of
prospective follow-up among patients in this analysis was 1885
days.
3.2. Baseline clinical characteristics
Patients characteristics were stratied by history of attempt(s)
as reported by the patients at baseline (Table 1). Patients with a
history of suicide attempt had a shorter follow-up period (Mean
days  Standard Deviation (SD): 476  343) compared to patients
without a history of attempt (Mean days  SD: 513  347)
(t(3005) 2.78, p 0.006). Patients with history of suicide attempt(s), compared to the ones without, were more likely to be
female, separated, divorced or widowed, had a higher prevalence of
a diagnoses of BD I subtype with an earlier age of onset of BD (both
earlier rst manic episode and rst depressive episode), a shorter
period of remission in the past 2 years, a higher number of manic
and depressive episodes, higher total scores in the MADRS and
YMRS scales, higher number of negative life events, higher scores
on the PDQ, lower QOL, smoked more packets of cigarettes per day,
had a higher positive family history of suicide, had higher neuroticism traits, had spent more days depressed or anxious in the year
prior study entry, and they were more likely to report suicidal
ideation at baseline. The two groups did not signicantly differ in
age, race, positive life events, age at smoking onset and in the other
NEO-FFI items (extroversion, openness). Agreeableness and
conscientiousness were only marginally different and were not
considered as prospective predictors. Comorbidities associated
with a history of suicide attempt are shown in Table 3.
3.3. Prospective predictors of suicide events
Of the 3083 subjects, 8 patients died by suicide (0.3%) and 132
attempted suicide (4.3%), with a total of 140 suicide events (4.6%).
53 (37.9%) patients had a suicide event within the rst 6 months
from the baseline assessment, and 96 (68.6%) patients experienced
an event within the rst year.
Patients with a prospective suicide event had a shorter followup period (mean days  SD: 337  315) compared to patients
without a suicide event (mean days  SD: 506  345)
(t(3081) 5.67, p < 0.001). We ran Cox proportional hazards

1803

regression models by merging predictors into three clusters: (i)

clinical, (ii) psychological and (iii) environmental predictors (as
outlined under statistical analysis section). The clinical cluster was
subdivided into two models: (a) clinical predictors including current disorder comorbidities, and (b) clinical predictors including
lifetime comorbidities; this division was carried out for decreasing
multi-collinearity and increasing precision. Age and gender were
entered as covariates in all models. Multi-collinearity was observed
between MADRS total score and percentage of days spent
depressed in the year prior to study entry, between lifetime alcohol
abuse and lifetime alcohol dependence, between lifetime drug
abuse and lifetime drug dependence, and between QOL total score
and negative life events. The percentage of days spent depressed in
the year prior to study entry, lifetime alcohol and drug dependence
and negative life events were the variables that were maintained
for the model according to our entry criteria (statistical analysis).
From the variables that were eliminated, only MADRS was a signicant predictor of future suicide events. Univariate analyses have
also been performed and show similar results to the prediction
models (Supplementary Table S1).
Results from the regression models are found in Table 2. Firstly,
the model with clinical predictors (including current comorbidities) showed that age, history of suicide attempt, OCD, current
alcohol abuse, agoraphobia and percentage of days depressed
during the last year prior study entry were signicant clinical
predictors of prospective suicide events. The model with clinical
predictors (including lifetime comorbidities) showed that history of
drug dependence was an additional signicant predictor, whereas
other lifetime psychiatric comorbidities were not signicant predictors in the model.
Secondly, the model with the psychological predictors showed
that only the PDQ total score was a signicant predictor of prospective suicide events, whereas Neuroticism and family history of
suicide failed to reach signicance. Thirdly, the model with the
environmental factors showed that only negative life events was a
signicant predictor of prospective suicide events, while cigarette
smoking (packets per day) and marital status were not signicant.
A nal joint model with all the signicant predictors from the
previous cluster groups (from Table 2) was run and included 78 patients that had a suicide event, and 1569 patients with no suicide
events. Table 4 shows the results of this nal model. Continuous
variables (age, PDQ) were dichotomized in order to also examine
survival curves. The model showed that a prospective suicide event
could be predicted by younger age, a history of suicide attempt, high
percentage of days depressed in the past year prior admission, and a
high PDQ total score. Marginal effects were observed for current OCD
comorbidity, current alcohol abuse and history of drug dependence.
Survival curves for signicant predictors are shown in Fig. 1.
4. Discussion
Our ndings show that a positive history of suicide attempt(s) is
the strongest predictor of a future suicide event, in line with prior
literature. Young age, high total score on the PDQ and percentage of
days spent depressed in the year prior to the study entry were
additional predictors of prospective suicidal behavior.
With regard to age groups at risk for suicidal behavior, prior
literature suggests that young-aged patients are at higher risk of
suicide compared to older patients, in line with the nding of this
report. One previous study in bipolar patients showed a higher
suicide rate in young patients, below the age of 50 years compared
to the ones over 60 (Oostervink et al., 2009). Studies with depressed
patients have shown that young patients report a higher number of
suicide attempts, as for example, in a group of 15e34 year-olds
(Blair-West et al., 1999) and a group below the age of 44

1804

N. Antypa et al. / Journal of Psychiatric Research 47 (2013) 1800e1808

Table 2
Clinical, psychological and environmental predictorsa of prospective suicide events in the STEP-BD.
B

SE

p value

OR

95.0% CI for OR
Lower

Upper

Model with clinical predictors and comorbidities at baseline

Age
L0.03
Gender
0.11
History of suicide attempt
1.35
% of days spent depressed last year
0.01
% of days spent Anxious last year
0.00
Age Mania 1st episode
0.003
Age Depression 1st episode
0.002
YMRS total
0.03
Number of depressive episodes
0.008
Number of manic episodes
0.175
BP1
0.495
BP2
0.657
Mixed state
0.014
SSRI use
0.06
Lithium use
0.286
Suicide ideation
0.395
Current Comorbidities
Obsessive compulsive disorder
0.718
Bulimia purging type
0.8
Schizoaffective disorder
9.211
Alcohol abuse
0.795
Alcohol dependence
0.036
Drug abuse
0.406
Drug dependence
0.73
PTSD
0.457
Generalized anxiety disorder
0.397
Panic Disorder Agoraphobia
0.611
Panic Disorder  Agoraphobia
0.382
Agoraphobia
0.702
Social anxiety disorder
0.055
d
b
Model including lifetime comorbidities
Lifetime comorbidities
Obsessive compulsive disorder
0.329
Anorexia restriction type
0.093
Bulimia Purging Type
0.431
Bulimia Non-Purging Type
0.3
Schizoaffective disorder
0.507
Alcohol dependence
0.257
Drug dependence
0.567
PTSD
0.069
GAD
0.049
Panic Disorder Agoraphobia
0.372
Panic Disorder  Agoraphobia
0.272
Agoraphobia
0.299
Social Anxiety disorder
0.331
e
Psychological factors
Family history of suicide
0.32
PDQ total score
0.02
Neuroticism
0.01
f
Environmental factors
Negative life events
L0.03
Cigarette smoking (Packet per day)
0.04
Marital status
Single
0.19
Married/living together
0.18

0.01
0.23
0.25
0.004
0.004
0.012
0.014
0.017
0.169
0.151
0.395
0.43
0.373
0.213
0.21
0.319

0.005
0.621
<0.001
0.01
0.954
0.792
0.884
0.082
0.96
0.244
0.21
0.126
0.97
0.779
0.174
0.216

0.97
1.12
3.86
1.01
1.00
1.003
1.002
0.97
0.99
1.19
0.61
0.52
0.99
0.94
1.33
1.48

0.95
0.72
2.38
1.003
0.993
0.979
0.975
0.938
0.712
0.887
0.281
0.223
0.474
0.62
0.882
0.794

0.99
1.75
6.25
1.019
1.007
1.028
1.03
1.004
1.38
1.601
1.323
1.204
2.049
1.431
2.008
2.774

0.292
0.481
182.8
0.334
0.376
0.486
0.439
0.319
0.294
0.314
0.458
0.355
0.273

0.014
0.096
0.96
0.017
0.925
0.404
0.096
0.152
0.177
0.051
0.404
0.048
0.84

2.05
2.23
0.00
2.21
1.04
0.66
2.08
1.58
0.67
1.84
1.47
2.02
0.95

1.157
0.868
0.00
1.15
0.496
0.257
0.878
0.846
0.378
0.996
0.597
1.006
0.554

3.629
5.71
4.476
4.264
2.163
1.728
4.908
2.949
1.196
3.409
3.596
4.049
1.616

0.264
0.456
0.301
0.733
1.102
0.246
0.255
0.253
0.253
0.272
0.338
0.324
0.233

0.212
0.838
0.153
0.682
0.645
0.298
0.026
0.786
0.848
0.172
0.422
0.356
0.155

1.389
1.098
1.539
0.741
0.602
0.774
1.763
0.934
0.953
1.45
1.312
1.349
1.392

0.829
0.45
0.852
0.176
0.069
0.477
1.07
0.568
0.58
0.851
0.676
0.715
0.883

2.33
2.682
2.778
3.114
5.217
1.254
2.907
1.534
1.565
2.472
2.547
2.545
2.197

0.32
0.01
0.02

0.317
0.047
0.682

1.38
1.02
1.01

0.74
1.00
0.96

2.59
1.05
1.06

0.01
0.07

<0.001
0.58

0.97
1.04

0.96
0.90

0.99
1.21

0.32
0.28

0.561
0.518

0.83
1.20

0.44
0.69

1.56
2.08

Abbreviation: PDQ: Personality Diagnostic Questionnaire; PTSD: post-traumatic stress disorder; GAD: general anxiety disorder; YMRS: Young Mania Rating Scale.
a
Signicant predictors are in bold.
b
Model includes same demographic and clinical predictors as the previous model with current comorbidities: in the model with lifetime comorbidities, from the clinical
predictors the only signicant ones were age, history of suicide attempt and % of days spent depressed last year.
c
N 2171 (suicide event: N 99, no suicide event: N 2072).
d
N 2179 (suicide event: N 99, no suicide event: N 2080).
e
N 1074 (suicide event: N 46, no suicide event: N 1028).
f
N 1981 (suicide event: N 93, no suicide event: N 1875).

(Wilkowska-Chmielewska et al., 2013), compared to older age

groups, while one study among general psychiatric inpatients
found that patients aged 15e24 years showed more suicidal
behavior than older patients (Park et al., 2012). In line with other
research on mood disorders, we used an age cut-off of 40 years
(median split) to identify age groups at risk: the younger group

showed a higher number of prospective suicide events compared to

the older patients. Patients were therefore more likely to engage in
suicidal behavior when young, while as age increased this tendency
decreased.
Previous research has highlighted the association between
personality disorders and suicidality in bipolar patients. A high PDQ

N. Antypa et al. / Journal of Psychiatric Research 47 (2013) 1800e1808

1805

Table 3
Comorbid psychiatric disorders in patients with and without a history of suicide attempt at baseline.
Psychiatric disorder

Current comorbidities

Lifetime comorbidities

N (%)

PTSD
GAD
OCD
Social Anxiety Disorder
PD with Agoraphobia
PD without Agoraphobia
Agoraphobia without PD
Schizoaffective disorder
Alcohol dependence
Alcohol abuse
Drug dependence
Drug abuse
Bulimia nervosa PT
Bulimia nervosa NPT
Anorexia nervosa RT
Anorexia nervosa BE-PT
Psychotic disorder
Schizophrenia
Schizophreniform disorder
Psychotic disorder NOS
Delusional disorder
Brief psychotic disorder

Patients SA

Patients SA

91
259
115
207
89
52
60
9
118
139
68
85
12
7
2
1
9
3
0
9
1
0

113(11.2%)
215 (21.3%)
91 (9.0%)
216 (21.3%)
122 (12.0%)
55 (5.4%)
72 (7.1%)
12 (1.2%)
97 (9.6%)
102 (10.1%)
63 (6.2%)
73 (7.2%)
25 (2.5%)
8 (0.8%)
2 (0.2%)
3 (0.3%)
9 (0.9%)
2 (0.2%)
1 (0.1%)
5 (0.5%)
0 (0%)
1 (0.1%)

(5%)
(14.3%)
(6.3%)
(11.4%)
(4.9%)
(2.9%)
(3.3%)
(0.5%)
(6.5%)
(7.7%)
(3.8%)
(4.7%)
(0.7%)
(0.4%)
(0.1%)
(0.1%)
(0.5%)
(0.2%)
(0%)
(0.5%)
(0.1%)
(0%)

c2(1)

p value

36.77
22.57
6.8
49.84
47.83
11.69
21.03
4.18
8.72
4.88
9.06
7.93
16.42
2.01
0.35
2.67
1.6
0.04
1.8
0.00
0.56
1.8

<0.001
<0.001
0.009
<0.001
<0.001
0.001
<0.001
0.04
0.003
0.027
0.003
0.005
<0.001
0.16
0.55
0.102
0.206
0.843
0.180
0.998
0.456
0.180

N (%)
Patients SA

Patients SA

240
356
181
347
189
133
116
17
454
560
276
365
60
12
25
15
40
5
0
25
2
4

297
289
140
303
199
103
103
20
365
406
255
296
104
21
38
15
25
4
1
13
0
1

(13.2%)
(19.7%)
(10.0%)
(19.1%)
(10.4%)
(7.3%)
(6.4%)
(0.9%)
(25.1%)
(30.9%)
(15.2%)
(20.1%)
(3.3%)
(0.7%)
(1.4%)
(0.8%)
(2.2%)
(0.3%)
(0%)
(1.4%)
(0.1%)
(0.2%)

(29.4%)
(28.6%)
(13.8%)
(29.9%)
(19.6%)
(10.2%)
(10.2%)
(2.0%)
(36.1%)
(40.2%)
(25.3%)
(29.3%)
(10.3%)
(2.1%)
(3.8%)
(1.5%)
(2.5%)
(0.4%)
(0.1%)
(1.3%)
(0%)
(0.1%)

c2(1)

p value

109.93
29.49
9.59
42.58
47.57
6.79
12.89
5.47
38.24
24.68
42.58
30.53
57.59
11.22
16.81
2.65
0.210
0.3
1.8
0.04
1.11
0.54

<0.001
<0.001
0.002
<0.001
<0.001
0.009
<0.001
0.019
<0.001
<0.001
<0.001
<0.001
<0.001
0.001
<0.001
0.104
0.647
0.587
0.180
0.842
0.291
0.462

Abbreviations: Patients SA: patients with a history of suicide attempt(s), Patients SA: patients without a history of suicide attempt(s), OCD: obsessive compulsive disorder,
PTSD: post-traumatic stress disorder, GAD: generalized anxiety disorder, PD: panic disorder, PT: purging type, NPT: non-purging type, RT: restriction type, BE-PT: binge-eating
purging type, NOS: not otherwise specied. Note: Signicant differences are indicated in bold.

total score was previously associated with suicide attempts in a

sample of 658 bipolar patients (Leverich et al., 2003), as well as in
BD patients with comorbid borderline personality disorder (Neves
et al., 2009). Similarly, our nal prediction model showed that a
high total score in the PDQ was associated with a 2-fold increase in
suicidal risk.
Moreover, in our study we observed that a high percentage of
days spent depressed in the year prior to study entry was a predictive factor of future suicide events. Prior literature has also
shown that suicide in bipolar patients occurs prevalently during the
depressive and mixed state of the disorder (Isometsa et al., 1994;
Simpson and Jamison, 1999; Valtonen et al., 2007), and a high
prevalence of depressive phases in the course of illness was previously suggested as a risk factor for suicide attempts (Finseth et al.,
2012). Our nding reinforces the importance of treating depression
in BD.

Table 4
Final joint model of predictors of prospective suicide events.a
B

Age >40
Gender
History of suicide attempt
% of days spent depressed
last year
Obsessive compulsive
disorder
Current alcohol abuse
Agoraphobia
Lifetime drug dependence
b
High PDQ
Negative life events
a

SE

p value

OR

95.0% CI for OR
Lower

Upper

L0.56
0.19
0.957
0.011

0.01
0.24
0.252
0.004

0.015
0.416
<0.001
0.012

0.57
1.21
2.603
1.011

0.364
0.762
1.588
1.002

0.897
1.931
4.266
1.019

0.518

0.312

0.096

1.679

0.911

3.095

0.547
0.391
0.466
0.674
0.009

0.313
0.362
0.252
0.286
0.01

0.08
0.28
0.064
0.018
0.336

1.728
1.479
1.594
1.963
1.009

0.936
0.727
0.973
1.121
0.991

3.189
3.009
2.612
3.435
1.028

N 1647 (suicide event: n 78, no suicide event: n 1569). Note: Signicant

differences are indicated in bold.
b
Age has been dichotomized at 40 years; PDQ total score has been dichotomized
at 36 (high PDQ: >36).

Our results also show that substance use disorders may also increase risk for suicidal behavior (although these predictors reached
a trend level of signicance in the nal model). Other studies have
also shown a worse suicide outcome for bipolar patients with substance disorder comorbidities (Cassidy, 2011; Finseth et al., 2012).
Among a group of depressed patients, comorbid substance dependence increased risk for suicide, compared to depressed patients
alone (Dumais et al., 2005). Our ndings support the proposition
that a positive history of drug dependence or current alcohol abuse
may increase the risk of future suicide event(s).
We also observed that the comorbidity of OCD was a suggestive
prospective predictor of future suicide events (predictor reached
trend level of signicance in the nal model). Indeed, bipolar patients with comorbid anxiety disorders had a history of more suicide
attempts compared to BD patients without comorbidity (Lee and
Dunner, 2008). Specically, comorbid OCD in bipolar patients has
been associated with higher incidence of prior suicide attempts
(Kruger et al., 2000) and obsessive compulsive tendencies have been
correlated with suicidal tendencies (Lester and Abdel-Khalek, 1999).
Despite the large amount of studies showing a protective effect
of lithium on suicide risk in bipolar patients (Baldessarini et al.,
2006; Cipriani et al., 2005; Young et al., 2010), this association was
not signicant in our model. Our nding is in line with a previous
study that was designed to examine the impact of pharmacotherapy
on prospective suicide attempts in STEP-BD subjects (Marangell
et al., 2008). One explanation could be that STEP-BD study is an
open study, not a comparison or randomized clinical trial, so clinician choice may have biased the effect of lithium treatment.
4.1. Limitations
A smaller number of suicide events (4.6%) was observed in the
STEP-BD sample in comparison to other samples (32.4% in retrospective and 19.8% in prospective studies assessing suicide attempts) (Novick et al., 2010). One explanation for the low suicide
rate may be that the cohort of patients in the STEP-BD study was

1806

N. Antypa et al. / Journal of Psychiatric Research 47 (2013) 1800e1808

Fig. 1. Survival curves of Bipolar Disorder patients over the course of the STEP-BD study, stratied by baseline presence of major predictors of prospective suicide events. Abbreviations: PDQ: Personality Diagnostic Questionnaire. *Prior to study entry A: no history of suicide attempt (events: 43; censored: 1875; total: 1918), history of suicide attempt(s)
(events: 51; censored: 633; total: 684), history of suicide attempt(s) high PDQ scores (events: 41; censored: 364; total: 405), B: no history of suicide attempt (events: 43;
censored: 1875; total: 1918), history of suicide attempt(s) (events: 51; censored: 545; total: 583), history of suicide attempt(s) age<40 (events: 54; censored: 452; total: 506), C:
no history of suicide attempt (events: 43; censored: 1875; total: 1918), history of suicide attempt(s) (events: 21; censored: 414; total: 435), history of suicide attempt(s) high %
days depressed (events: 71; censored: 583; total: 654).

followed by specialized clinicians, who provided patients with

evidence-based pharmacological and psychosocial treatments.
These interventions and close monitoring of patients may have
been preventive. A limitation of the present analysis is that suicide
attempts and completions were merged into one category of suicide events, but the predictors of the two types of suicidal behavior
may differ substantially. Unfortunately, the small sample of suicide
completers (n 8) did not allow for the separate examination of
these outcomes due to limited statistical power. Furthermore, the
differing sample sizes across the regression models may be a source
of bias. However, simulation studies have shown that differences in
sizes of samples with >800 subjects are not at risk for inated odds
ratios, as is the case for smaller sample sizes (Nemes et al., 2009).
Finally, we want to note that although several factors were not
shown to be signicant in multivariate models, this does not mean
they could not be related (univariately) with suicide risk (see
Table S1 for univariate analyses).

5. Conclusions
Our ndings suggest that prior history of suicide attempts is the
strongest predictor for future suicide events in Bipolar Disorder.
Moreover, bipolar patients of younger age, with personality disorder traits, as well as with long periods spent in depressed mood
may have a higher risk for future suicidal behavior. These characteristics are potentially identiable when the patient presents for
treatment and can be addressed with intensive care by clinicians.
Role of founding source
The STEP-BD project was funded in whole or in part with
federal funds from the National Institute of Mental Health
(NIMH), under contract N01-MH-80001. Details of past and current STEP-BD participants can be found at www.stepbd.org/
research/stepacknowledgmentlist.pdf. The work of Niki Antypa

N. Antypa et al. / Journal of Psychiatric Research 47 (2013) 1800e1808

is supported by the Rubicon Grant (#446-11-004), awarded by

the Netherlands Organization for Scientic Research (NWO) and
the Marie Curie Cofund Action. The funding bodies had no role in
the study design, the collection of data, the writing of this report
and decision to submit it for publication.
Contribution of authors
N. Antypa and A. Serretti were responsible for the conception
and design of the analysis. M. Antonioli drafted the rst version of
the manuscript and performed with N. Antypa the statistical analyses. N. Antypa and A. Serretti revised the manuscript contributing
with important intellectual content. All authors approved the nal
version of the manuscript.
Conict of interest
Dr. Serretti is or has been consultant/speaker for: Abbott,
Angelini, Astra Zeneca, Clinical Data, Boehringer, Bristol Myers
Squibb, Eli Lilly, GlaxoSmithKline, Italfarmaco, Janssen, Lundbeck,
Pzer, Sano, Servier. N. Antypa and M. Antonioli have no conict of
interest.
Acknowledgment
None.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.jpsychires.2013.08.005.
References
Acosta FJ, Vega D, Torralba L, Navarro S, Ramallo-Farina Y, Fiuza D, et al. Hopelessness and suicidal risk in bipolar disorder. A study in clinically nonsyndromal
patients. Comprehensive Psychiatry 2012;53:1103e9.
Altamura AC, DellOsso B, Berlin HA, Buoli M, Bassetti R, Mundo E. Duration of
untreated illness and suicide in bipolar disorder: a naturalistic study. European
Archives of Psychiatry and Clinical Neuroscience 2010;260:385e91.
Angst F, Stassen HH, Clayton PJ, Angst J. Mortality of patients with mood disorders:
follow-up over 34e38 years. Journal of Affective Disorders 2002;68:167e81.
Baldassano CF. Illness course, comorbidity, gender, and suicidality in patients with
bipolar disorder. Journal of Clinical Psychiatry 2006;67(Suppl. 11):8e11.
Baldessarini RJ, Tondo L, Davis P, Pompili M, Goodwin FK, Hennen J. Decreased risk
of suicides and attempts during long-term lithium treatment: a meta-analytic
review. Bipolar Disorders 2006;8:625e39.
Beck AT, Weissman A, Lester D, Trexler L. The measurement of pessimism:
the hopelessness scale. Journal of Consulting and Clinical Psychology 1974;42:
861e5.
Blair-West GW, Cantor CH, Mellsop GW, Eyeson-Annan ML. Lifetime suicide risk in
major depression: sex and age determinants. Journal of Affective Disorders
1999;55:171e8.
Bowden CL, Perlis RH, Thase ME, Ketter TA, Ostacher MM, Calabrese JR, et al. Aims
and results of the NIMH systematic treatment enhancement program for bipolar disorder (STEP-BD). CNS Neuroscience & Therapeutics 2012;18:243e9.
Carter TD, Mundo E, Parikh SV, Kennedy JL. Early age at onset as a risk factor for
poor outcome of bipolar disorder. Journal of Psychiatric Research 2003;37:297e
303.
Cassidy F. Risk factors of attempted suicide in bipolar disorder. Suicide & Lifethreatening Behavior 2011;41:6e11.
Cipriani A, Pretty H, Hawton K, Geddes JR. Lithium in the prevention of suicidal
behavior and all-cause mortality in patients with mood disorders: a systematic
review of randomized trials. American Journal of Psychiatry 2005;162:1805e19.
Coryell W, Fiedorowicz J, Leon AC, Endicott J, Keller MB. Age of onset and the
prospectively observed course of illness in bipolar disorder. Journal of Affective
Disorders 2013;146:34e8.
Costa Jr PT, Fagan PJ, Piedmont RL, Ponticas Y, Wise TN. The ve-factor model of
personality and sexual functioning in outpatient men and women. Psychiatric
Medicine 1992;10:199e215.
Dumais A, Lesage AD, Alda M, Rouleau G, Dumont M, Chawky N, et al. Risk factors
for suicide completion in major depression: a case-control study of impulsive
and aggressive behaviors in men. American Journal of Psychiatry 2005;162:
2116e24.

1807

Endicott J, Nee J, Harrison W, Blumenthal R. Quality of life enjoyment and

satisfaction questionnaire: a new measure. Psychopharmacology Bulletin
1993;29:321e6.
Finseth PI, Morken G, Andreassen OA, Malt UF, Vaaler AE. Risk factors related to
lifetime suicide attempts in acutely admitted bipolar disorder inpatients. Bipolar Disorders 2012;14:727e34.
Fossati A, Maffei C, Bagnato M, Donati D, Donini M, Fiorilli M, et al. Brief communication: criterion validity of the Personality Diagnostic Questionnaire-4
(PDQ-4) in a mixed psychiatric sample. Journal of Personality Disorders
1998;12:172e8.
Garcia-Amador M, Colom F, Valenti M, Horga G, Vieta E. Suicide risk in rapid cycling
bipolar patients. Journal of Affective Disorders 2009;117:74e8.
Goldstein TR, Ha W, Axelson DA, Goldstein BI, Liao F, Gill MK, et al. Predictors of
prospectively examined suicide attempts among youth with bipolar disorder.
Archives of General Psychiatry 2012;69:1113e22.
Harris EC, Barraclough B. Suicide as an outcome for mental disorders. A metaanalysis. British Journal of Psychiatry 1997;170:205e28.
Isometsa ET, Heinriksson MM, Aro HM, Lonnqvist JK. Suicide in bipolar disorder in
Finland. American Journal of Psychiatry 1994;151:1020e4.
Johnson SL, McMurrich SL, Yates M. Suicidality in bipolar I disorder. Suicide & Lifethreatening Behavior 2005;35:681e9.
Kruger S, Braunig P, Cooke RG. Comorbidity of obsessive-compulsive disorder in
recovered inpatients with bipolar disorder. Bipolar Disorders 2000;2:71e4.
Lee JH, Dunner DL. The effect of anxiety disorder comorbidity on treatment resistant bipolar disorders. Depression and Anxiety 2008;25:91e7.
Lester D, Abdel-Khalek AM. Manic-depression, suicidality, and obsessivecompulsive tendencies. Psychological Reports 1999;85:1100.
Leverich GS, Altshuler LL, Frye MA, Suppes T, Keck Jr PE, McElroy SL, et al. Factors
associated with suicide attempts in 648 patients with bipolar disorder in the
Stanley Foundation Bipolar Network. Journal of Clinical Psychiatry 2003;64:
506e15.
Marangell LB, Bauer MS, Dennehy EB, Wisniewski SR, Allen MH, Miklowitz DJ,
et al. Prospective predictors of suicide and suicide attempts in 1,556 patients
with bipolar disorders followed for up to 2 years. Bipolar Disorders 2006;8:
566e75.
Marangell LB, Dennehy EB, Wisniewski SR, Bauer MS, Miyahara S, Allen MH, et al.
Case-control analyses of the impact of pharmacotherapy on prospectively
observed suicide attempts and completed suicides in bipolar disorder: ndings
from STEP-BD. Journal of Clinical Psychiatry 2008;69:916e22.
Martinez JM, Marangell LB, Simon NM, Miyahara S, Wisniewski SR, Harrington J,
et al. Baseline predictors of serious adverse events at one year among patients
with bipolar disorder in STEP-BD. Psychiatric Services 2005;56:1541e8.
Maser JD, Akiskal HS, Schettler P, Scheftner W, Mueller T, Endicott J, et al. Can
temperament identify affectively ill patients who engage in lethal or near-lethal
suicidal behavior? A 14-year prospective study. Suicide & Life-threatening
Behavior 2002;32:10e32.
McCrae RR, John OP. An introduction to the ve-factor model and its applications.
Journal of Personality 1992;60:175e215.
McElroy SL, Frye MA, Hellemann G, Altshuler L, Leverich GS, Suppes T, et al. Prevalence and correlates of eating disorders in 875 patients with bipolar disorder.
Journal of Affective Disorders 2011;128:191e8.
Montgomery SA, Asberg M. A new depression scale designed to be sensitive to
change. British Journal of Psychiatry 1979;134:382e9.
Nemes S, Jonasson JM, Genell A, Steineck G. Bias in odds ratios by logistic regression
modelling and sample size. BMC Medical Research Methodology 2009;9:56.
Neves FS, Malloy-Diniz LF, Correa H. Suicidal behavior in bipolar disorder: what is
the inuence of psychiatric comorbidities? Journal of Clinical Psychiatry
2009;70:13e8.
Nivoli AM, Pacchiarotti I, Rosa AR, Popovic D, Murru A, Valenti M, et al. Gender
differences in a cohort study of 604 bipolar patients: the role of predominant
polarity. Journal of Affective Disorders 2011;133:443e9.
Nordstrom P, Asberg M, Aberg-Wistedt A, Nordin C. Attempted suicide predicts
suicide risk in mood disorders. Acta Psychiatrica Scandinavica 1995;92:345e50.
Novick DM, Swartz HA, Frank E. Suicide attempts in bipolar I and bipolar II disorder:
a review and meta-analysis of the evidence. Bipolar Disorders 2010;12:1e9.
Oostervink F, Boomsma MM, Nolen WA. Bipolar disorder in the elderly;
different effects of age and of age of onset. Journal of Affective Disorders
2009;116:176e83.
Oquendo MA, Galfalvy H, Russo S, Ellis SP, Grunebaum MF, Burke A, et al. Prospective study of clinical predictors of suicidal acts after a major depressive
episode in patients with major depressive disorder or bipolar disorder. American Journal of Psychiatry 2004;161:1433e41.
Ostacher MJ, Lebeau RT, Perlis RH, Nierenberg AA, Lund HG, Moshier SJ, et al.
Cigarette smoking is associated with suicidality in bipolar disorder. Bipolar
Disorders 2009;11:766e71.
Papolos D, Hennen J, Cockerham MS. Factors associated with parent-reported suicide threats by children and adolescents with community-diagnosed bipolar
disorder. Journal of Affective Disorders 2005;86:267e75.
Park S, Choi JW, Kyoung Yi K, Hong JP. Suicide mortality and risk factors in the 12
months after discharge from psychiatric inpatient care in Korea: 1989e2006.
Psychiatry Research 2013;208:145e50.
Preuss UW, Schuckit MA, Smith TL, Danko GP, Bucholz KK, Hesselbrock MN,
et al. Predictors and correlates of suicide attempts over 5 years in 1,237
alcohol-dependent men and women. American Journal of Psychiatry
2003;160:56e63.

1808

N. Antypa et al. / Journal of Psychiatric Research 47 (2013) 1800e1808

Sachs GS. Strategies for improving treatment of bipolar disorder: integration of

measurement and management. Acta Psychiatrica Scandinavica Supplementum
2004:7e17.
Sachs GS, Thase ME, Otto MW, Bauer M, Miklowitz D, Wisniewski SR, et al. Rationale, design, and methods of the systematic treatment enhancement program
for bipolar disorder (STEP-BD). Biological Psychiatry 2003;53:1028e42.
Sarason IG, Johnson JH, Siegel JM. Assessing the impact of life changes:
development of the Life Experiences Survey. Journal of Consulting and
Clinical Psychology 1978;46:932e46.
Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E, et al. The MiniInternational Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10.
Journal of Clinical Psychiatry 1998;59(Suppl. 20):22e33 quiz 34e57.
Simpson SG, Jamison KR. The risk of suicide in patients with bipolar disorders.
Journal of Clinical Psychiatry 1999;60:53e6.

Song JY, Yu HY, Kim SH, Hwang SS, Cho HS, Kim YS, et al. Assessment of risk factors
related to suicide attempts in patients with bipolar disorder. The Journal of
Nervous and Mental Disease 2012;200:978e84.
Valtonen HM, Suominen K, Mantere O, Leppamaki S, Arvilommi P, Isometsa E.
Suicidal behaviour during different phases of bipolar disorder. Journal of Affective Disorders 2007;97:101e7.
Wilkowska-Chmielewska J, Szelenberger W, Wojnar M. Age-dependent symptomatology of depression in hospitalized patients and its implications for DSM-5.
Journal of Affective Disorders 2013;150:142e5.
Young AH, McElroy SL, Bauer M, Philips N, Chang W, Olausson B, et al. A doubleblind, placebo-controlled study of quetiapine and lithium monotherapy in
adults in the acute phase of bipolar depression (EMBOLDEN I). Journal of
Clinical Psychiatry 2010;71:150e62.
Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability, validity and sensitivity. British Journal of Psychiatry 1978;133:429e35.