© 2014 John Wiley & Sons A/S Published by John Wiley & Sons Ltd.
Bipolar Disorders 2014
BIPOLAR DISORDERS
Original Article
Differences in incidence of suicide attempts between bipolar I and II disorders and major depressive disorder Holma KM, Haukka J, Suominen K, Valtonen HM, Mantere O, Melartin TK, Sokero TP, Oquendo MA, Isomets€ a ET. Differences in incidence of suicide attempts between bipolar I and II disorders and major depressive disorder. Bipolar Disord 2014: 00: 000–000. © 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd. Objectives: Whether risk of suicide attempts (SAs) differs between patients with bipolar disorder (BD) and patients with major depressive disorder (MDD) is unclear. We investigated whether cumulative risk differences are due to dissimilarities in time spent in high-risk states, incidence per unit time in high-risk states, or both. Methods: Incidence rates for SAs during various illness phases, based on prospective life charts, were compared between patients from the Jorvi Bipolar Study (n = 176; 18 months) and the Vantaa Depression Study (n = 249; five years). Risk factors and their interactions with diagnosis were investigated with Cox proportional hazards models. Results: By 18 months, 19.9% of patients with BD versus 9.5% of patients with MDD had attempted suicide. However, patients with BD spent 4.6% of the time in mixed episodes, and more time in major depressive episodes (MDEs) (35% versus 21%, respectively) and in subthreshold depression (39% versus 31%, respectively) than those with MDD. Compared with full remission, the combined incidence rates of SAs were 5-, 25-, and 65-fold in subthreshold depression, MDEs, and BD mixed states, respectively. Between cohorts, incidence of attempts was not different during comparable symptom states. In Cox models, hazard was elevated during MDEs and subthreshold depression, and among patients with preceding SAs, female patients, those with poor social support, and those aged < 40 years, but was unrelated to BD diagnosis. Conclusions: The observed higher cumulative incidence of SAs among patients with BD than among those with MDD is mostly due to patients with BD spending more time in high-risk illness phases, not to differences in incidence during these phases, or to bipolarity itself. BD mixed phases contribute to differences involving very high incidence, but short duration. Diminishing the time spent in high-risk phases is crucial for prevention.
About one-half of all individuals completing suicide have major depressive disorder (MDD) or bipolar disorder (BD) at the time of their death (1, 2), and a significant proportion of patients with major mood disorders die by suicide (3), making them a key target population for suicide
K Mikael Holmaa,b, Jari Haukkaa,c, Kirsi Suominena,b, Hanna M Valtonena,b, Outi Manterea,d, Tarja K Melartina,d, T Petteri Sokeroa, Maria A Oquendoe and Erkki T Isometsa€a,d a Mood, Depression, and Suicidal Behaviour Unit, National Institute for Health and Welfare, b Psychiatry, City of Helsinki Health Centre, c Department of Public Health, Faculty of Medicine, University of Helsinki, dDepartment of Psychiatry, Helsinki University Central Hospital, District of Helsinki and Uusimaa, Helsinki, Finland, eDepartment of Psychiatry, New York State Psychiatric Institute and Columbia University, New York, NY, USA
doi: 10.1111/bdi.12195 Key words: bipolar disorder – major depressive disorder – suicide – suicide attempts Received 28 June 2013, revised and accepted for publication 26 November 2013 Corresponding author: Erkki T. Isometsa€, M.D., Ph.D. Department of Psychiatry Institute of Clinical Medicine University of Helsinki P.O. Box 22 (Va€lska€rinkatu 12 A) Helsinki 00014 Finland Fax: +358-9-47163735 E-mail: [email protected]
prevention. Moreover, a history of suicide attempts (SAs) is a robust risk factor for completed suicide (3, 4), and approximately one-half of those with MDD (5) or BD (6) completing suicide have made previous attempts. However, while suicide mortality may be higher among patients
1
Holma et al. with BD (3) compared with MDD, the underlying reasons for presumed differences remain poorly understood. Whether risk of SAs differs between patients with BD and MDD is also unclear. Retrospective studies suggest higher rates of SAs among individuals with BD, especially bipolar II disorder (BD-II), compared to MDD (7–9). However, in prospective studies, BD sometimes has (10) and sometimes has not (11–13) been associated with greater risk of suicidal behavior. Putative differences have been attributed to early onset of illness (10), higher depression severity (10), affective temperament (10, 14), and diagnosis of BD-II in those with bipolar illness (10). SAs are most likely to occur during a major depressive episode (MDE) (9, 15, 16) or mixed phases (9, 15, 17) for patients with BD, and during MDEs for patients with MDD (18–20). However, few prospective studies have accounted for variations in risk over time, associating suicidal acts with variations in clinical states and illness course. This would be possible only by using life-chart methodology. We have previously reported incidence rates and risk factors for SAs for bipolar I disorder (BD-I) and BD-II in the Jorvi Bipolar Study (JoBS) (15), and for MDD in the Vantaa Depression Study (VDS) (19, 20). We now compare incidence rates and risk factors for SAs in pooled data from both cohorts. We aimed to investigate whether putative differences in risk are due to differences in time spent in high-risk states, incidence per unit of time in high-risk states, or both. We hypothesized that (i) differences in cumulative risk are due to differences in time spent in risk states rather than to differences in incidence rates of SA in comparable illness states, and (ii) high risk in mixed illness states contributes to higher cumulative risk in BD. Materials and methods
Our patients came from two separate, comparable cohorts from two adjacent cities. Both are collaborative research projects of the Mood Disorder Research Unit of the Department of Mental Health and Substance Use Services of the National Institute of Health and Welfare, Helsinki, Finland, with the last author (ETI) as the principal investigator. The research protocol for the VDS was approved by the Ethics Committee of Peijas Medical Care District and that for the JoBS by the Ethics Committee of Helsinki University Central Hospital. Detailed methodologies have been described elsewhere for the VDS (21, 22) and the JoBS (23, 24).
2
Screening and baseline evaluation
In brief, outpatients and inpatients in an acute mood episode were screened for MDD (VDS: n = 806) or BD (JoBS: n = 1630). After a positive screen or suspicion of an incident episode, the patient was fully informed about the study and written informed consent was obtained. Diagnosis was made using all available information from face-to-face interviews and psychiatric records. Information was also gathered on demographic characteristics, illness history using a retrospective life chart, and current symptomatology. A current episode of MDD was diagnosed using the World Health Organization Schedules for Clinical Assessment in Neuropsychiatry (SCAN), version 2.0 (25) in the VDS, and a current episode of BD was diagnosed using the Structured Clinical Interview for DSMIV Axis I Disorders, research version with psychotic screen (SCID-I/P) (26). To exclude substance-induced mood disorder, patients with MDD currently abusing alcohol or other substances were interviewed after two to three weeks of abstinence. In the JoBS, mixed episodes were diagnosed following the DSM-IV-TR (27). Included in the BD-II group were patients diagnosed as having BD not otherwise specified (NOS) with recurrent hypomania of two to three days, and, deviating from DSM-IV, those with depressive mixed states (three or more simultaneous intra-episode hypomanic symptoms present at least 50% of the time during an MDE) as defined by Benazzi and Akiskal (28). The ‘soft’ bipolar spectrum was excluded. The final baseline cohorts consisted of 269 patients with MDD, and 191 DSM-IV patients with BD-I and BD-II. Inter-rater agreement in diagnostic interviews was excellent (21, 24). DSM-IV Axis I diagnoses (SCAN in VDS; SCID-I/P in JoBS) and Axis II diagnoses (SCID-II for DSM-III-R in VDS; SCID-II for DSM-IV in JoBS) were made. Baseline measurements included the 17-item Hamilton Rating Scale for Depression (HAM-D) (29), 21-item Beck Depression Inventory (30), Beck Anxiety Inventory (BAI) (31), Beck Hopelessness Scale (BHS) (32), Young Mania Rating Scale in JoBS (33), Scale for Suicidal Ideation (34), Social and Occupational Functioning Assessment Scale of DSM-IV (35), Social Adjustment Scale Self-Report (36), Interview for Recent Life Events (37), Interview Measure of Social Relationships (38), Perceived Social Support Scale-Revised (39), and Eysenck Personality Inventory (40). In addition, the number of chronic medical disorders (Axis III) was documented using a checklist.
Incidence of suicide attempts in BD and MDD Follow-up
After baseline assessments, patients were interviewed at six and 18 months, and at five years. Interviews typically lasted two to three hours. Repeated SCID-I/P (JoBS), SCAN 2.0 (VDS), and SCID-II interviews, all observer- and self-reported symptom scales, and patient record review comprised follow-up assessments. SCID-I/P was used in the VDS five-year follow-up. All data were used to generate graphic life charts, based on DSM-IV criteria and definitions. Change points in psychopathological states, using probes related to important life events, were used to improve assessment accuracy. Our life-chart method was similar, but not identical, to the Longitudinal Interval Followup Evaluation methodology used in the National Institute of Mental Health Collaborative Depression Study (41). In the VDS, follow-up time was divided into three periods: (i) full remission (no MDE symptoms), (ii) partial remission (one to four of the nine symptoms), and (iii) MDE (five or more symptoms). In the JoBS, follow-up time was divided into 10 types of time periods (phases): euthymia, manic, hypomanic, major depressive, mixed, depressive mixed, cyclothymic, substanceinduced mood phase, and depressive and hypomanic symptoms (24). However, we deviated from DSM-IV in two ways: by accepting hypomanias lasting two and three days as hypomanic episodes, and by including depressive mixed episodes (29) for both patients with BD-I and those with BD-II. Mixed states were defined as in DSM-IV (27). Information about SAs during follow-up was based on interview and psychiatric and medical records. An SA was defined as self-injurious behavior with a non-fatal outcome accompanied by evidence (either explicit or implicit) that the person had intended to die (42); self-harm with no suicidal intention was excluded. The attempts were timed and placed on the life chart. Attrition rate and characteristics
In the VDS, of the 269 patients with MDD enrolled, 229 patients participated in the six-month interview, 198 unipolar patients participated in the 18-month interview, and 163 patients participated in the five-year interview. The present study included 249/269 patients (92.6%) who participated in at least one follow-up interview, with a mean follow-up time of 5.2 years [standard deviation (SD) = 2.0 years], resulting in 1,009 patientyears. These patients had a median age of 41.8 years. Of the 249 patients: 185 (74.3%) were female, 77 (33%) had attempted suicide before
follow-up, 60 (24%) had comorbid alcohol dependence or abuse, 142 (57%) had any anxiety comorbidity, 106 (43%) had comorbid personality disorder, and 21 (8%) had psychotic features. At baseline, most patients (88%) were receiving antidepressants, and for the majority (78%) the dosage was adequate for the acute phase. More than one-half (57%) were receiving selective serotonin reuptake inhibitors (SSRIs) alone at baseline, about one-fifth (18%) were receiving newer antidepressants (tetracyclics, a selective norepinephrine reuptake inhibitor, or a reversible mono-amine oxidase-A inhibitor), 8% were receiving tricyclic antidepressants (TCAs), and 6% were receiving a combination treatment that usually consisted of SSRIs and TCAs. Nearly all patients (98%) were receiving psychotherapeutic support in the early acute phase, but only a few had weekly psychotherapy (16%) (21, 43). During the 18 months, 9.5% (20/211) of the patients attempted suicide at least once, with a total of 48 attempts. During the fiveyear period, 36/249 patients with MDD attempted suicide at least once, with a total of 106 attempts (19). Eight (3%) patients died during the 18 months after baseline, three (1%) by suicide. In the JoBS, of the 191 subjects with a major depressive, manic, hypomanic, mixed, or depressive mixed phase of BD at intake, 161 participated in the 18-month follow-up interview. This study included 176 patients who participated in at least one follow-up interview; the mean length of follow-up time was 1.6 person-years. These patients had a median age of 37.7 years. Of the 176 patients: 91 (52%) were female, 81 (46%) were diagnosed with BD-I, 95 (54%) were diagnosed with BD-II, 89 (51%) had attempted suicide before follow-up [40/81 patients (49%) with BD-I and 49/ 95 patients (52%) with BD-II], 88 (50%) had lifetime comorbid substance dependence or abuse, 95 (54%) had any lifetime anxiety comorbidity, 79 (45%) had comorbid personality disorder, and 87 (49%) had lifetime psychotic features (44). A total of 75% of patients received mood stabilizers at some point during the maintenance phase. Valproate was most often prescribed, over two times more often than lithium. Patients with BD-I received lithium more often than patients with BD-II, whereas lamotrigine was only prescribed for patients with BD-II. One-fifth of all patients, and one-fourth of the patients with a clinical diagnosis of BD received atypical antipsychotics at some point during the maintenance phase (45). During the 18 months, 19.9% (35/176) of patients with BD attempted suicide, with a total of 53 attempts. Three (2%) patients died by the end of the
3
Holma et al. 18-month period, two (1%) by suicide. We have reported incidence rates of SAs separately in these cohort studies (15, 19). Statistical analysis
For purposes of clinical description, since mixed and depressive episodes are associated with high risk of suicidal behavior, baseline characteristics of patients with BD who were in a depressed or mixed episode (mixed episode and depressive mixed episode combined) were first compared with patients with MDD who were depressed at baseline. Patients with other mood disorder phases at study entry were excluded from these descriptive analyses. Categorical variables were compared using the v2 test with Yates’ correction or Fisher’s exact test. Analysis of variance (ANOVA) was used for continuous variables with a normal distribution and the Kruskal–Wallis test for those with a non-normal distribution. After detailed univariate analyses, we then chose predictors for multivariate models based on their clinical and statistical validity, significance, and relevance in representing a domain of risk factors. Risk factors for suicidal behavior and their relationship to patients’ diagnoses were investigated with a multinomial logistic regression model dividing patients into three groups based on baseline characteristics: (i) patients with no suicidal behavior, (ii) patients with suicidal ideation without SAs, and (iii) patients who attempted suicide. The mood disorder diagnoses (MDD, BD-I, and BD-II) were included in the model adjusted for age and gender. Kaplan–Meier analysis and the log-rank test were used to compare the proportions of suicide attempters during the first 18 months (for comparability, data were censored at 18.0 months). Cox proportional hazards regression models were used to investigate associations between time-varying clinical mood states and fixed (time-invariant) explanatory variables and hazards of SAs. The association between the time-varying level of depression (full remission, subthreshold depression, and MDE) and SAs in the life chart was analyzed. Mixed episodes were not included as, according to DSM-IV definitions, they did not exist among patients with MDD. For each individual, the follow-up assessment was divided into periods with a constant level of depression. Because there were several observations (time periods) for the same individual, a robust sandwich variance estimator was used. In the model, diagnosis, time-varying level of depression, and baseline variables (SAs before baseline assessment, suicide
4
ideation at baseline, duration of illness, HAM-D score, neuroticism, psychotic features, comorbid alcohol use disorder, cluster A–C personality disorders, marital status, BHS score, level of perceived social support, and size of social network) were used as fixed covariates. For continuous variables (age, HAM-D score, BHS score, duration of illness, perceived social support, neuroticism, and size of social network), hazard ratios were calculated for 10-unit increments. Interactions between the time-varying phase and other risk factors were tested using the Cox model with the likelihood ratio test. Results for the Cox model are presented as hazard ratios. The model describes how the hazard of event (SA) is predicted by current values of explanatory variables. The Predictive Analytics Software Statistics 18 (SPSS, Inc., Chicago, IL, USA) and an R-package (R Development Core Team, Vienna, Austria) were used. Results
Before baseline, one-half (51%) of patients with BD compared to one-third (33%) of patients with MDD had attempted suicide (v2 = 13.9, df = 1, p < 0.001). By the 18-month interview, patients with BD reported follow-up SAs twice as often as patients with MDD (19.9% versus 9.5%, respectively; Kaplan–Meier log-rank v2 = 12.4, df = 1, p < 0.001), and there were altogether 53 attempts versus 48 attempts, respectively. Table 1 lists the characteristics of patients with MDD and BD in a major depressive or mixed episode at baseline. Compared to the two BD groups, patients with MDD were significantly more often female, were less likely to have anxiety disorders and panic disorders, but more often had agoraphobia and simple phobia. They also spent significantly more time in subthreshold depression compared to the BD groups. Patients with BD depressed at baseline had significantly more SAs prior to baseline, were more likely to have cluster B personality disorders, had a longer duration of illness, and spent more time in MDEs. Patients with BD in a mixed episode at baseline were significantly younger, had a lower age at onset, had more anxiety symptoms (BAI), had more frequent psychotic symptoms, and had higher neuroticism compared to the other two groups. They also had a higher level of suicidal ideation as a trend, followed by depressive patients with BD, and patients with MDD. The median length of preceding illness was over seven years longer among patients with BD compared to patients with MDD (11.6 versus 4.1 years, respectively), and their age of onset was 10 years younger (median
Incidence of suicide attempts in BD and MDD 21 versus 31 years, respectively). Moreover, patients with BD spent 4.1% of the time in the (combined) mixed episodes; more time in MDEs
than those with MDD (33.5% versus 22.5%, respectively), but less time with subthreshold depression (16.6% versus 30.3%, respectively).
Table 1. Characteristics of psychiatric patients with unipolar major depressive disorder (MDD) and bipolar disorder in major depressive or mixed episode at baseline
Characteristic Total BD-II (baseline) BD-I (baseline) Female Previous suicide attempt Psychotic symptoms (baseline) Comorbid Axis I disorder Anxiety disorder (any) Panic disorder: With agoraphobia Without agoraphobia Agoraphobia without panic Social phobia OCD Simple phobia GAD Alcohol dependence/abuse (baseline) Personality disorder Cluster A Cluster B Cluster C Married or cohabiting Smoking Age Age at onset Duration of illness HAM-D score BDI score BAI score SSI score 10th 25th 50th 75th 90th BHS SOFAS PSSS-R Neuroticism (at lowest HAM-D)b Extroversion (at lowest HAM-D)b Proportion of time in: MDE Subthreshold depression Full remission Mixed episode
Bipolar mixed episodea n (%)
MDD n (%)
Bipolar MDE n (%)
249 (100) – – 185 (74.3) 77 (33.3) 21 (8.4) 162 (65.1) 142 (57.0)
106 (55.5) 59 (55.7) 47 (44.3) 56 (52.8) 56 (52.8) 11 (10.4) 58 (69.9) 64 (60.4)
41 (21.5) 26 (63.4) 15 (36.6) 26 (63.4) 26 (63.4) 10 (24.4) 30 (73.2) 28 (68.3)
16 (6.4) 24 (9.6) 28 (11.2) 46 (18.5) 17 (6.8) 63 (25.3) 35 (14.1) 60 (24.1) 106 (42.6) 45 (18.1) 34 (13.7) 78 (31.3) 131 (52.6) 94 (40.7) Mean (SD) 40.1 (11.0) 31.8 (12.7) 8.3 (9.8) 19.4 (6.2) 27.6 (8.5) 22.4 (10.8)
20 (20.6) 6 (6.2) 1 (0.9) 19 (17.9) 4 (3.8) 9 (8.5) 17 (16.0) 16 (15.1) 49 (46.2) 12 (11.3) 35 (33.0) 22 (20.8) 46 (43.4) 50 (48.5) Mean (SD) 38.3 (12.2) 24.4 (9.6) 13.9 (10.1) 20.9 (6.7) 26.2 (9.5) 22.9 (11.9)
6 (15.4) 9 (23.0) 2 (4.9) 12 (29.3) 0 (0) 4 (9.8) 8 (19.5) 8 (19.5) 19 (46.3) 4 (9.8) 12 (29.3) 12 (29.3) 14 (34.1) 23 (57.5) Mean (SD) 33.6 (11.7) 20.9 (8.2) 12.6 (10.4) 19.1 (6.8) 26.5 (10.4) 29.3 (10.6)
0 0 1 12 20 10.2 (4.9) 52.0 (10.9) 39.5 (12.7) 13.8 (5.4) 12.0 (4.3)
0 0 5 14 21 11.0 (4.3) 48.3 (11.4) 39.9 (12.3) 16.1 (5.0) 11.1 (4.6)
0 0 8 14 18 11.0 (5.0) 48.4 (11.3) 42.8 (12.1) 17.6 (3.9) 11.6 (4.3)
4.1
0.127
1.4 5.2 1.2 13.3 1.7
0.248 0.006 0.305
Bipolar Disorders 2014
BIPOLAR DISORDERS
Original Article
Differences in incidence of suicide attempts between bipolar I and II disorders and major depressive disorder Holma KM, Haukka J, Suominen K, Valtonen HM, Mantere O, Melartin TK, Sokero TP, Oquendo MA, Isomets€ a ET. Differences in incidence of suicide attempts between bipolar I and II disorders and major depressive disorder. Bipolar Disord 2014: 00: 000–000. © 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd. Objectives: Whether risk of suicide attempts (SAs) differs between patients with bipolar disorder (BD) and patients with major depressive disorder (MDD) is unclear. We investigated whether cumulative risk differences are due to dissimilarities in time spent in high-risk states, incidence per unit time in high-risk states, or both. Methods: Incidence rates for SAs during various illness phases, based on prospective life charts, were compared between patients from the Jorvi Bipolar Study (n = 176; 18 months) and the Vantaa Depression Study (n = 249; five years). Risk factors and their interactions with diagnosis were investigated with Cox proportional hazards models. Results: By 18 months, 19.9% of patients with BD versus 9.5% of patients with MDD had attempted suicide. However, patients with BD spent 4.6% of the time in mixed episodes, and more time in major depressive episodes (MDEs) (35% versus 21%, respectively) and in subthreshold depression (39% versus 31%, respectively) than those with MDD. Compared with full remission, the combined incidence rates of SAs were 5-, 25-, and 65-fold in subthreshold depression, MDEs, and BD mixed states, respectively. Between cohorts, incidence of attempts was not different during comparable symptom states. In Cox models, hazard was elevated during MDEs and subthreshold depression, and among patients with preceding SAs, female patients, those with poor social support, and those aged < 40 years, but was unrelated to BD diagnosis. Conclusions: The observed higher cumulative incidence of SAs among patients with BD than among those with MDD is mostly due to patients with BD spending more time in high-risk illness phases, not to differences in incidence during these phases, or to bipolarity itself. BD mixed phases contribute to differences involving very high incidence, but short duration. Diminishing the time spent in high-risk phases is crucial for prevention.
About one-half of all individuals completing suicide have major depressive disorder (MDD) or bipolar disorder (BD) at the time of their death (1, 2), and a significant proportion of patients with major mood disorders die by suicide (3), making them a key target population for suicide
K Mikael Holmaa,b, Jari Haukkaa,c, Kirsi Suominena,b, Hanna M Valtonena,b, Outi Manterea,d, Tarja K Melartina,d, T Petteri Sokeroa, Maria A Oquendoe and Erkki T Isometsa€a,d a Mood, Depression, and Suicidal Behaviour Unit, National Institute for Health and Welfare, b Psychiatry, City of Helsinki Health Centre, c Department of Public Health, Faculty of Medicine, University of Helsinki, dDepartment of Psychiatry, Helsinki University Central Hospital, District of Helsinki and Uusimaa, Helsinki, Finland, eDepartment of Psychiatry, New York State Psychiatric Institute and Columbia University, New York, NY, USA
doi: 10.1111/bdi.12195 Key words: bipolar disorder – major depressive disorder – suicide – suicide attempts Received 28 June 2013, revised and accepted for publication 26 November 2013 Corresponding author: Erkki T. Isometsa€, M.D., Ph.D. Department of Psychiatry Institute of Clinical Medicine University of Helsinki P.O. Box 22 (Va€lska€rinkatu 12 A) Helsinki 00014 Finland Fax: +358-9-47163735 E-mail: [email protected]
prevention. Moreover, a history of suicide attempts (SAs) is a robust risk factor for completed suicide (3, 4), and approximately one-half of those with MDD (5) or BD (6) completing suicide have made previous attempts. However, while suicide mortality may be higher among patients
1
Holma et al. with BD (3) compared with MDD, the underlying reasons for presumed differences remain poorly understood. Whether risk of SAs differs between patients with BD and MDD is also unclear. Retrospective studies suggest higher rates of SAs among individuals with BD, especially bipolar II disorder (BD-II), compared to MDD (7–9). However, in prospective studies, BD sometimes has (10) and sometimes has not (11–13) been associated with greater risk of suicidal behavior. Putative differences have been attributed to early onset of illness (10), higher depression severity (10), affective temperament (10, 14), and diagnosis of BD-II in those with bipolar illness (10). SAs are most likely to occur during a major depressive episode (MDE) (9, 15, 16) or mixed phases (9, 15, 17) for patients with BD, and during MDEs for patients with MDD (18–20). However, few prospective studies have accounted for variations in risk over time, associating suicidal acts with variations in clinical states and illness course. This would be possible only by using life-chart methodology. We have previously reported incidence rates and risk factors for SAs for bipolar I disorder (BD-I) and BD-II in the Jorvi Bipolar Study (JoBS) (15), and for MDD in the Vantaa Depression Study (VDS) (19, 20). We now compare incidence rates and risk factors for SAs in pooled data from both cohorts. We aimed to investigate whether putative differences in risk are due to differences in time spent in high-risk states, incidence per unit of time in high-risk states, or both. We hypothesized that (i) differences in cumulative risk are due to differences in time spent in risk states rather than to differences in incidence rates of SA in comparable illness states, and (ii) high risk in mixed illness states contributes to higher cumulative risk in BD. Materials and methods
Our patients came from two separate, comparable cohorts from two adjacent cities. Both are collaborative research projects of the Mood Disorder Research Unit of the Department of Mental Health and Substance Use Services of the National Institute of Health and Welfare, Helsinki, Finland, with the last author (ETI) as the principal investigator. The research protocol for the VDS was approved by the Ethics Committee of Peijas Medical Care District and that for the JoBS by the Ethics Committee of Helsinki University Central Hospital. Detailed methodologies have been described elsewhere for the VDS (21, 22) and the JoBS (23, 24).
2
Screening and baseline evaluation
In brief, outpatients and inpatients in an acute mood episode were screened for MDD (VDS: n = 806) or BD (JoBS: n = 1630). After a positive screen or suspicion of an incident episode, the patient was fully informed about the study and written informed consent was obtained. Diagnosis was made using all available information from face-to-face interviews and psychiatric records. Information was also gathered on demographic characteristics, illness history using a retrospective life chart, and current symptomatology. A current episode of MDD was diagnosed using the World Health Organization Schedules for Clinical Assessment in Neuropsychiatry (SCAN), version 2.0 (25) in the VDS, and a current episode of BD was diagnosed using the Structured Clinical Interview for DSMIV Axis I Disorders, research version with psychotic screen (SCID-I/P) (26). To exclude substance-induced mood disorder, patients with MDD currently abusing alcohol or other substances were interviewed after two to three weeks of abstinence. In the JoBS, mixed episodes were diagnosed following the DSM-IV-TR (27). Included in the BD-II group were patients diagnosed as having BD not otherwise specified (NOS) with recurrent hypomania of two to three days, and, deviating from DSM-IV, those with depressive mixed states (three or more simultaneous intra-episode hypomanic symptoms present at least 50% of the time during an MDE) as defined by Benazzi and Akiskal (28). The ‘soft’ bipolar spectrum was excluded. The final baseline cohorts consisted of 269 patients with MDD, and 191 DSM-IV patients with BD-I and BD-II. Inter-rater agreement in diagnostic interviews was excellent (21, 24). DSM-IV Axis I diagnoses (SCAN in VDS; SCID-I/P in JoBS) and Axis II diagnoses (SCID-II for DSM-III-R in VDS; SCID-II for DSM-IV in JoBS) were made. Baseline measurements included the 17-item Hamilton Rating Scale for Depression (HAM-D) (29), 21-item Beck Depression Inventory (30), Beck Anxiety Inventory (BAI) (31), Beck Hopelessness Scale (BHS) (32), Young Mania Rating Scale in JoBS (33), Scale for Suicidal Ideation (34), Social and Occupational Functioning Assessment Scale of DSM-IV (35), Social Adjustment Scale Self-Report (36), Interview for Recent Life Events (37), Interview Measure of Social Relationships (38), Perceived Social Support Scale-Revised (39), and Eysenck Personality Inventory (40). In addition, the number of chronic medical disorders (Axis III) was documented using a checklist.
Incidence of suicide attempts in BD and MDD Follow-up
After baseline assessments, patients were interviewed at six and 18 months, and at five years. Interviews typically lasted two to three hours. Repeated SCID-I/P (JoBS), SCAN 2.0 (VDS), and SCID-II interviews, all observer- and self-reported symptom scales, and patient record review comprised follow-up assessments. SCID-I/P was used in the VDS five-year follow-up. All data were used to generate graphic life charts, based on DSM-IV criteria and definitions. Change points in psychopathological states, using probes related to important life events, were used to improve assessment accuracy. Our life-chart method was similar, but not identical, to the Longitudinal Interval Followup Evaluation methodology used in the National Institute of Mental Health Collaborative Depression Study (41). In the VDS, follow-up time was divided into three periods: (i) full remission (no MDE symptoms), (ii) partial remission (one to four of the nine symptoms), and (iii) MDE (five or more symptoms). In the JoBS, follow-up time was divided into 10 types of time periods (phases): euthymia, manic, hypomanic, major depressive, mixed, depressive mixed, cyclothymic, substanceinduced mood phase, and depressive and hypomanic symptoms (24). However, we deviated from DSM-IV in two ways: by accepting hypomanias lasting two and three days as hypomanic episodes, and by including depressive mixed episodes (29) for both patients with BD-I and those with BD-II. Mixed states were defined as in DSM-IV (27). Information about SAs during follow-up was based on interview and psychiatric and medical records. An SA was defined as self-injurious behavior with a non-fatal outcome accompanied by evidence (either explicit or implicit) that the person had intended to die (42); self-harm with no suicidal intention was excluded. The attempts were timed and placed on the life chart. Attrition rate and characteristics
In the VDS, of the 269 patients with MDD enrolled, 229 patients participated in the six-month interview, 198 unipolar patients participated in the 18-month interview, and 163 patients participated in the five-year interview. The present study included 249/269 patients (92.6%) who participated in at least one follow-up interview, with a mean follow-up time of 5.2 years [standard deviation (SD) = 2.0 years], resulting in 1,009 patientyears. These patients had a median age of 41.8 years. Of the 249 patients: 185 (74.3%) were female, 77 (33%) had attempted suicide before
follow-up, 60 (24%) had comorbid alcohol dependence or abuse, 142 (57%) had any anxiety comorbidity, 106 (43%) had comorbid personality disorder, and 21 (8%) had psychotic features. At baseline, most patients (88%) were receiving antidepressants, and for the majority (78%) the dosage was adequate for the acute phase. More than one-half (57%) were receiving selective serotonin reuptake inhibitors (SSRIs) alone at baseline, about one-fifth (18%) were receiving newer antidepressants (tetracyclics, a selective norepinephrine reuptake inhibitor, or a reversible mono-amine oxidase-A inhibitor), 8% were receiving tricyclic antidepressants (TCAs), and 6% were receiving a combination treatment that usually consisted of SSRIs and TCAs. Nearly all patients (98%) were receiving psychotherapeutic support in the early acute phase, but only a few had weekly psychotherapy (16%) (21, 43). During the 18 months, 9.5% (20/211) of the patients attempted suicide at least once, with a total of 48 attempts. During the fiveyear period, 36/249 patients with MDD attempted suicide at least once, with a total of 106 attempts (19). Eight (3%) patients died during the 18 months after baseline, three (1%) by suicide. In the JoBS, of the 191 subjects with a major depressive, manic, hypomanic, mixed, or depressive mixed phase of BD at intake, 161 participated in the 18-month follow-up interview. This study included 176 patients who participated in at least one follow-up interview; the mean length of follow-up time was 1.6 person-years. These patients had a median age of 37.7 years. Of the 176 patients: 91 (52%) were female, 81 (46%) were diagnosed with BD-I, 95 (54%) were diagnosed with BD-II, 89 (51%) had attempted suicide before follow-up [40/81 patients (49%) with BD-I and 49/ 95 patients (52%) with BD-II], 88 (50%) had lifetime comorbid substance dependence or abuse, 95 (54%) had any lifetime anxiety comorbidity, 79 (45%) had comorbid personality disorder, and 87 (49%) had lifetime psychotic features (44). A total of 75% of patients received mood stabilizers at some point during the maintenance phase. Valproate was most often prescribed, over two times more often than lithium. Patients with BD-I received lithium more often than patients with BD-II, whereas lamotrigine was only prescribed for patients with BD-II. One-fifth of all patients, and one-fourth of the patients with a clinical diagnosis of BD received atypical antipsychotics at some point during the maintenance phase (45). During the 18 months, 19.9% (35/176) of patients with BD attempted suicide, with a total of 53 attempts. Three (2%) patients died by the end of the
3
Holma et al. 18-month period, two (1%) by suicide. We have reported incidence rates of SAs separately in these cohort studies (15, 19). Statistical analysis
For purposes of clinical description, since mixed and depressive episodes are associated with high risk of suicidal behavior, baseline characteristics of patients with BD who were in a depressed or mixed episode (mixed episode and depressive mixed episode combined) were first compared with patients with MDD who were depressed at baseline. Patients with other mood disorder phases at study entry were excluded from these descriptive analyses. Categorical variables were compared using the v2 test with Yates’ correction or Fisher’s exact test. Analysis of variance (ANOVA) was used for continuous variables with a normal distribution and the Kruskal–Wallis test for those with a non-normal distribution. After detailed univariate analyses, we then chose predictors for multivariate models based on their clinical and statistical validity, significance, and relevance in representing a domain of risk factors. Risk factors for suicidal behavior and their relationship to patients’ diagnoses were investigated with a multinomial logistic regression model dividing patients into three groups based on baseline characteristics: (i) patients with no suicidal behavior, (ii) patients with suicidal ideation without SAs, and (iii) patients who attempted suicide. The mood disorder diagnoses (MDD, BD-I, and BD-II) were included in the model adjusted for age and gender. Kaplan–Meier analysis and the log-rank test were used to compare the proportions of suicide attempters during the first 18 months (for comparability, data were censored at 18.0 months). Cox proportional hazards regression models were used to investigate associations between time-varying clinical mood states and fixed (time-invariant) explanatory variables and hazards of SAs. The association between the time-varying level of depression (full remission, subthreshold depression, and MDE) and SAs in the life chart was analyzed. Mixed episodes were not included as, according to DSM-IV definitions, they did not exist among patients with MDD. For each individual, the follow-up assessment was divided into periods with a constant level of depression. Because there were several observations (time periods) for the same individual, a robust sandwich variance estimator was used. In the model, diagnosis, time-varying level of depression, and baseline variables (SAs before baseline assessment, suicide
4
ideation at baseline, duration of illness, HAM-D score, neuroticism, psychotic features, comorbid alcohol use disorder, cluster A–C personality disorders, marital status, BHS score, level of perceived social support, and size of social network) were used as fixed covariates. For continuous variables (age, HAM-D score, BHS score, duration of illness, perceived social support, neuroticism, and size of social network), hazard ratios were calculated for 10-unit increments. Interactions between the time-varying phase and other risk factors were tested using the Cox model with the likelihood ratio test. Results for the Cox model are presented as hazard ratios. The model describes how the hazard of event (SA) is predicted by current values of explanatory variables. The Predictive Analytics Software Statistics 18 (SPSS, Inc., Chicago, IL, USA) and an R-package (R Development Core Team, Vienna, Austria) were used. Results
Before baseline, one-half (51%) of patients with BD compared to one-third (33%) of patients with MDD had attempted suicide (v2 = 13.9, df = 1, p < 0.001). By the 18-month interview, patients with BD reported follow-up SAs twice as often as patients with MDD (19.9% versus 9.5%, respectively; Kaplan–Meier log-rank v2 = 12.4, df = 1, p < 0.001), and there were altogether 53 attempts versus 48 attempts, respectively. Table 1 lists the characteristics of patients with MDD and BD in a major depressive or mixed episode at baseline. Compared to the two BD groups, patients with MDD were significantly more often female, were less likely to have anxiety disorders and panic disorders, but more often had agoraphobia and simple phobia. They also spent significantly more time in subthreshold depression compared to the BD groups. Patients with BD depressed at baseline had significantly more SAs prior to baseline, were more likely to have cluster B personality disorders, had a longer duration of illness, and spent more time in MDEs. Patients with BD in a mixed episode at baseline were significantly younger, had a lower age at onset, had more anxiety symptoms (BAI), had more frequent psychotic symptoms, and had higher neuroticism compared to the other two groups. They also had a higher level of suicidal ideation as a trend, followed by depressive patients with BD, and patients with MDD. The median length of preceding illness was over seven years longer among patients with BD compared to patients with MDD (11.6 versus 4.1 years, respectively), and their age of onset was 10 years younger (median
Incidence of suicide attempts in BD and MDD 21 versus 31 years, respectively). Moreover, patients with BD spent 4.1% of the time in the (combined) mixed episodes; more time in MDEs
than those with MDD (33.5% versus 22.5%, respectively), but less time with subthreshold depression (16.6% versus 30.3%, respectively).
Table 1. Characteristics of psychiatric patients with unipolar major depressive disorder (MDD) and bipolar disorder in major depressive or mixed episode at baseline
Characteristic Total BD-II (baseline) BD-I (baseline) Female Previous suicide attempt Psychotic symptoms (baseline) Comorbid Axis I disorder Anxiety disorder (any) Panic disorder: With agoraphobia Without agoraphobia Agoraphobia without panic Social phobia OCD Simple phobia GAD Alcohol dependence/abuse (baseline) Personality disorder Cluster A Cluster B Cluster C Married or cohabiting Smoking Age Age at onset Duration of illness HAM-D score BDI score BAI score SSI score 10th 25th 50th 75th 90th BHS SOFAS PSSS-R Neuroticism (at lowest HAM-D)b Extroversion (at lowest HAM-D)b Proportion of time in: MDE Subthreshold depression Full remission Mixed episode
Bipolar mixed episodea n (%)
MDD n (%)
Bipolar MDE n (%)
249 (100) – – 185 (74.3) 77 (33.3) 21 (8.4) 162 (65.1) 142 (57.0)
106 (55.5) 59 (55.7) 47 (44.3) 56 (52.8) 56 (52.8) 11 (10.4) 58 (69.9) 64 (60.4)
41 (21.5) 26 (63.4) 15 (36.6) 26 (63.4) 26 (63.4) 10 (24.4) 30 (73.2) 28 (68.3)
16 (6.4) 24 (9.6) 28 (11.2) 46 (18.5) 17 (6.8) 63 (25.3) 35 (14.1) 60 (24.1) 106 (42.6) 45 (18.1) 34 (13.7) 78 (31.3) 131 (52.6) 94 (40.7) Mean (SD) 40.1 (11.0) 31.8 (12.7) 8.3 (9.8) 19.4 (6.2) 27.6 (8.5) 22.4 (10.8)
20 (20.6) 6 (6.2) 1 (0.9) 19 (17.9) 4 (3.8) 9 (8.5) 17 (16.0) 16 (15.1) 49 (46.2) 12 (11.3) 35 (33.0) 22 (20.8) 46 (43.4) 50 (48.5) Mean (SD) 38.3 (12.2) 24.4 (9.6) 13.9 (10.1) 20.9 (6.7) 26.2 (9.5) 22.9 (11.9)
6 (15.4) 9 (23.0) 2 (4.9) 12 (29.3) 0 (0) 4 (9.8) 8 (19.5) 8 (19.5) 19 (46.3) 4 (9.8) 12 (29.3) 12 (29.3) 14 (34.1) 23 (57.5) Mean (SD) 33.6 (11.7) 20.9 (8.2) 12.6 (10.4) 19.1 (6.8) 26.5 (10.4) 29.3 (10.6)
0 0 1 12 20 10.2 (4.9) 52.0 (10.9) 39.5 (12.7) 13.8 (5.4) 12.0 (4.3)
0 0 5 14 21 11.0 (4.3) 48.3 (11.4) 39.9 (12.3) 16.1 (5.0) 11.1 (4.6)
0 0 8 14 18 11.0 (5.0) 48.4 (11.3) 42.8 (12.1) 17.6 (3.9) 11.6 (4.3)
4.1
0.127
1.4 5.2 1.2 13.3 1.7
0.248 0.006 0.305
