Original Article

International Society for Bipolar DisordersTask Force on Suicide: meta-analyses andmeta-regression of correlates of suicideattempts and suicide deaths in bipolar disorder

Schaffer A, Isomets€a ET, Tondo L, Moreno DH, Turecki G, Reis C,Cassidy F, Sinyor M, Azorin J-M, Kessing LV, Ha K, Goldstein T,Weizman A, Beautrais A, Chou Y-H, Diazgranados N, Levitt AJ,Zarate Jr CA, Rihmer Z, Yatham LN. International Society for BipolarDisorders Task Force on Suicide: meta-analyses and meta-regression ofcorrelates of suicide attempts and suicide deaths in bipolar disorder.Bipolar Disord 2014: 00: 000–000. © 2014 John Wiley & Sons A/S.Published by John Wiley & Sons Ltd.

Objectives: Bipolar disorder is associated with a high risk of suicideattempts and suicide death. The main objective of the present study wasto identify and quantify the demographic and clinical correlates ofattempted and completed suicide in people with bipolar disorder.

Methods: Within the framework of the International Society for BipolarDisorders Task Force on Suicide, a systematic review of articles publishedsince 1980, characterized by the key terms bipolar disorder and ‘suicideattempts’ or ‘suicide’, was conducted, and data extracted for analysisfrom all eligible articles. Demographic and clinical variables for which ≥3 studies with usable data were available were meta-analyzed using fixedor random-effects models for association with suicide attempts andsuicide deaths. There was considerable heterogeneity in the methodsemployed by the included studies.

Results: Variables significantly associated with suicide attempts were:female gender, younger age at illness onset, depressive polarity of firstillness episode, depressive polarity of current or most recent episode,comorbid anxiety disorder, any comorbid substance use disorder,alcohol use disorder, any illicit substance use, comorbid clusterB/borderline personality disorder, and first-degree family history ofsuicide. Suicide deaths were significantly associated with male genderand first-degree family history of suicide.

Conclusions: This paper reports on the presence and magnitude of thecorrelates of suicide attempts and suicide deaths in bipolar disorder.These findings do not address causation, and the heterogeneity of datasources should limit the direct clinical ranking of correlates. Our resultsnonetheless support the notion of incorporating diagnosis-specific datain the development of models of understanding suicide in bipolardisorder.

Ayal Schaffera, Erkki T Isomets€ab,Leonardo Tondoc,d, Doris HMorenoe, Gustavo Tureckif,Catherine Reisa, Frederick Cassidyg,Mark Sinyora, Jean-Michel Azorinh,Lars Vedel Kessingi, Kyooseob Haj,Tina Goldsteink, Abraham Weizmanl,Annette Beautraism, Yuan-HwaChoun, Nancy Diazgranadoso,Anthony J Levitta, Carlos A ZarateJrp, Zolt�an Rihmerq and Lakshmi NYathamr

Affiliations for all the authors are listed after the

Acknowledgements.

doi: 10.1111/bdi.12271

Key words: bipolar disorder – meta-analysis –

suicide

Received 1 May 2014, revised and accepted for

publication 5 September 2014

Corresponding author:

Ayal Schaffer, M.D., F.R.C.P.C.

Mood and Anxiety Disorders Program

Department of Psychiatry

Sunnybrook Health Sciences Centre

2075 Bayview Avenue, Room FG 52

Toronto, ON

Canada M4N 3M5

Fax: 416-480-4613

E-mail: ayal.schaffer@sunnybrook.ca

Partial findings were presented during the

symposia at the 16th Annual Conference of the

International Society for Bipolar Disorders,

Seoul, South Korea, 18–21 March 2014.

Mental illness is present in nearly all people whoattempt or die by suicide, and among psychiatricdiagnoses, bipolar disorder (BD) may be associ-

ated with the highest suicide risk (1–14). Amongpeople with BD, the estimated rate of death by sui-cide is 0.2–0.4 per 100 person-years (5, 15–17);

1

Bipolar Disorders 2014 © 2014 John Wiley & Sons A/SPublished by John Wiley & Sons Ltd.

BIPOLAR DISORDERS

however, these rates may reflect data from higher-risk periods in the life of someone with BD, andsimple extrapolation to estimates of lifetime riskare inherently unreliable. The absolute risk of sui-cide among patients with a diagnosis of BD at firsthospital contact has been found to be around 8%for men and 5% for women over a median of18 years’ follow-up (12), and the standardizedmortality ratio of suicide deaths in BD comparedto that in the general population has been reportedto be 10–30-fold (18–21). Rates of suicide attemptsare also very high, with an estimated annual risk of0.9% per year, and a lifetime risk of up to one-halfof sufferers with BD (22–25).

There are a number of well-described correlatesof suicide attempts and suicide deaths identified inthe broader mental illness or public health litera-ture, including gender, depression, anxiety, sub-stance use, family history of suicide, and others(26–28). Suicide is a behavioral endpoint thatresults from a multitude of factors, and confirmingwhether these broadly identified correlates of riskof suicide attempts and suicide are also present inBD populations is imperative. Analogous investi-gation of BD-specific clinical factors such as illnesssubtype and polarity of first and most recent epi-sode is also warranted. These types of data wouldinform the emerging effort to move towards morediagnosis-specific approaches to understanding sui-cide risk, risk assessments, and prevention (29–33).

Recently, published reviews on suicide attemptsand suicide in BD shed light on the scope of thepublic health, clinical assessment, and managementchallenges, and also illuminate glaring gaps in theavailable data (18, 29, 30, 34–37). One such gap isthe absence of up-to-date, quantitative, meta-analytic models of the specific demographic andclinical factors putatively associated with suicideattempts and suicide in BD. While individual stud-ies can identify correlates within a specific studypopulation, these findings require both replicationand an estimate of the magnitude of the associa-tion in order to inform risk assessments and sug-gest avenues for prevention. Prior meta-analysesby Hawton et al. (38) and Novick et al. (39) laidthe foundation for this approach in BD; however,a major expansion of data has since occurred inthe past few years, which is not captured in theseearlier publications. This is especially relevant forsuicide deaths, which have now been more thor-oughly examined in several large epidemiologicalBD samples (4, 8, 12, 15, 40).

The International Society for Bipolar Disorders(ISBD) is a leading organization devoted to pro-moting international collaboration in the study ofBD. Under the auspices of the ISBD, a Task Force

on Suicide comprising 20 international expertswas launched, with the objectives of completing asystematic review of the available literature andconducting meta-analyses and meta-regression onputative correlates of suicide attempts and suicidein BD. The present publication is the product ofthe work done by this task force to achieve the goalof identifying and quantifying the degree of associ-ations between demographic and clinical variableswith the risk of suicide attempts and suicide deathin people with BD.

Methods

Study search and selection

We conducted a systematic review of English-language articles using keywords ‘bipolar disorder’and ‘suicide attempts’ or ‘suicide’, publishedbetween 1 January 1980 and 30 June 2013. Thesearch was then expanded via the ancestryapproach by manual examination of reference listsof included articles and recent published reviewson suicide or suicide attempts in BD (18, 29, 30,34–37). Included articles had to have met all thefollowing criteria: (i) subjects with BD comprisedall or a large majority (> 80%) of the study popula-tion; (ii) study population was exclusively > 13years old; (iii) a binary measure of suicide attemptsor suicide deaths was reported; (iv) a non-suicideattempt or non-suicide group was included; and (v)a binary measure of a demographic or clinical vari-able of interest was reported or could be calculatedfrom the published data. Age at onset was includedas a continuous variable as there was no uniformdefinition of early or later age at onset used acrossstudies. Both prospective and retrospective studieswere included, as were studies from either clinicalor epidemiological samples. There is no uniformdefinition of a suicide attempt in relation to intentor lethality; however, studies were excluded if theyonly reported on non-suicidal self-injury, suicidalideation, or if suicidality ratings were reported as acontinuous measure.

Data extraction

Using the Preferred Reporting Items for System-atic Reviews and Meta-Analyses (PRISMA)framework for systematic reviews (41), the initialsearch yielded 1,700 abstracts. These were screenedfor eligibility criteria and duplication of samples,resulting in 74 full-text articles being assessed bytwo trained investigators (AS and CR), whoreviewed each article for inclusion and exclusioncriteria. Articles for which there was any

2

Schaffer et al.

uncertainty about eligibility were discussed and adecision made through consensus. Studies wereexcluded at this stage if they contained subjectoverlap with another article or did not report suffi-cient details for data analysis (n = 33 studiesexcluded). This resulted in an initial group of 41eligible studies, including one for which the corre-sponding author was contacted for clarification.Whenever possible, task force members, as a geo-graphically diverse group of experts who wereauthors on many of the included studies, providedadditional details of data when insufficient infor-mation was available in the published article (addi-tional three studies). This resulted in a total of 44eligible studies.

In total, 34 papers reported on suicide attempts(total N = 50,004 subjects with BD) across one ormore variables, with 31/34 using clinical samples,29/34 using non-representative samples, and 30/34reporting suicide attempts in a retrospective man-ner. There were 12 papers that reported on sui-cide deaths (total N = 75,137 subjects with BD),with 8/12 using clinical samples, 4/12 using non-representative samples, and 8/12 identifying sui-cide in a retrospective sample. Only two studiesreported both on suicide attempts and deaths (15,40).

For each article, the following data wereextracted and coded: (i) author name(s); (ii) yearof publication; (iii) number of subjects with orwithout a suicide attempt for each demographicand clinical variable of interest; and (iv) numberof suicide deaths or non-deaths for each demo-graphic and clinical variable of interest. No studyincluded all variables of interest, so data wereonly extracted for those variables examined in thepublication. Variables of interest were chosenbased on the general and BD-specific literature onsuicide attempts and deaths, and included: (i) gen-der; (ii) age at onset of BD; (iii) subtype of BD[bipolar I disorder (BD-I) or bipolar II disorder(BD-II)]; (iv) polarity of first mood episode; (v)polarity of current or most recent mood episode(mixed episodes were generally reported as a sub-set of mania, and there were insufficient data tospecifically analyze mixed episodes); (vi) lifetimehistory of past suicide attempts (for analysis ofsuicide deaths only); (vii) current or lifetime anxi-ety disorder; (viii) presence of lifetime history ofpsychotic symptoms; (ix) current or lifetime sub-stance use disorder (most studies reported asabuse, dependence, or both), as well as the follow-ing three subcategories: (xA) current or lifetimealcohol use disorder; (xB) current or lifetime can-nabis use; and (xC) any current or lifetime illicitsubstance use disorder; (xi) current or lifetime

personality disorder; and (xii) first-degree familyhistory of death by suicide.

Data analysis

Cochrane Information Management system -Review Manager(RevMan) version 5.2 (November2012) was utilized to conduct the meta-analyses.For each variable of interest, a meta-analysis wasonly conducted if there was a minimum of threestudies with usable data. A total of 13 variableshad sufficient data for analysis on suicide attemptsand four variables had sufficient data for analysison suicide. The same study could yield data formultiple analyses as many studies reported onmore than one variable of interest. Odds ratios(OR) with 95% confidence intervals (CI) were cal-culated for all binary measures, and weightedmean difference was calculated for age at illnessonset, the only continuous measure. Random-effects models were used when significant heteroge-neity was present (Cochrane test p < 0.1); other-wise, fixed-effects models were used. As asensitivity analysis of the effect of very large stud-ies, we re-ran all meta-analyses, removing any sin-gle study that had a > 40% weighting.

A meta-regression was conducted (using STATAsoftware, StataCorp LP, College Station, TX,USA) using suicide attempts in males versusfemales as the meta-analytic outcome variable, andthe covariates included were based on a sufficientnumber of observations within studies thatreported on gender differences. Covariatesincluded polarity of first mood episode, mean ageat illness onset, BD subtype, psychosis, any sub-stance use disorder, alcohol or illicit substanceuse disorder, and family history of suicide. Testsof meta-bias (effect of small studies) and meta-influence (influence of a single study) were alsoconducted. Other meta-regressions could not becompleted owing to an insufficient number ofobservations.

Results

Findings related to suicide attempts

Thirty-four papers reported suicide attemptsacross one or more variables in a manner suitablefor meta-analysis, with a total of 50,004 subjectswith BD included in the non-overlapping samples.(15, 24, 40, 42–72). We examined 13 variables,including gender, age at illness onset, BD subtype(BD-I or BD-II), polarity of first mood episode,polarity of current or most recent mood episode,lifetime comorbid anxiety disorder, lifetime his-

3

ISBD Task Force on suicide in bipolar disorder

tory of psychotic symptoms, any substance usedisorder, alcohol use disorder, cannabis use, anyillicit substance use disorder, comorbid clusterB/borderline personality disorder, and first-degreefamily history of suicide. Each of Figures 1–9 andSupplementary Figures 1–3 shows the results ofthe meta-analysis for the specific variable(s) ofinterest.

Numerous variables were found to be signifi-cantly correlated with presence of suicide attempts.Women were significantly more likely to attemptsuicide (OR = 1.54, 95% CI: 1.44–1.66, p <0.00001), with 8/20 studies reporting a significanteffect in this direction, and the remaining 12/20studies reporting no significant gender-based dif-ference (Fig. 1). A meta-regression conductedusing gender differences in suicide attempts as themeta-analytic outcome found no significant inde-pendent association for any other tested covariate.There were also no significant small-study or single-study effects.

Age at illness onset was 2.99 years younger (95%CI: 2.20–3.78 years, p < 0.00001) among those witha history of suicide attempt compared to those with-out a history of suicide attempt (Fig. 2), with a stan-dardized mean difference of �0.29 (95% CI: �0.36to �0.21, p < 0.0001).

BD subtypes, BD-I or BD-II, were examined in14 relatively evenly weighted studies, with twostudies finding higher rates of suicide attempt inBD-I, two studies finding a higher rate in BD-II,and the remainder finding no difference, resultingin no overall effect of subtype being identified(OR = 1.07, 95% CI: 0.79–1.45, p = 0.68) (Supple-mentary Fig. 1).

Subjects with a depressive polarity of first moodepisode were nearly twice as likely to attempt sui-cide (OR = 1.92, 95% CI: 1.39–2.65, p < 0.0001)(Fig. 3), with all seven studies reporting a similardirection of effect, and most reaching statistical sig-nificance.

Depressive polarity of the current or most recentmood episode had the strongest association with asuicide attempt (OR = 5.99, 95% CI = 1.75–20.5,p = 0.004), but only three studies reported on thisvariable, resulting in wide confidence intervals(Fig. 4).

The presence of a lifetime comorbid anxiety dis-order was significantly associated with suicideattempts in 8/13 studies, with an OR of 1.81 (95%CI: 1.66–1.97, p < 0.0001) (Fig. 5). Although onestudy had a very large weighting in this analysis,there was a consistent direction of effect with theother studies.

Of the seven studies that examined history ofpsychosis, one reported a higher rate of suicide

attempt among subjects with a history of psy-chosis, one reported a higher rate among sub-jects without psychosis, and the remainder foundno significant difference, resulting in no signifi-cant association being identified (OR = 0.91,95% CI: 0.64–1.30, p = 0.61) (SupplementaryFig. 2).

The presence of a current or lifetime comorbidsubstance use disorder was separated into fournon-mutually exclusive groups, including (i) anysubstance use disorder (OR = 1.81, 95% CI: 1.31–2.50, p < 0.0001) (Fig. 6); (ii) alcohol use disorder(OR = 1.60, 95% CI: 1.31–1.97, p < 0.00001)(Fig. 7A); (iii) any cannabis use (OR = 1.29, 95%CI: 0.85–1.94, p = 0.23) (Supplementary Fig. 3);and (iv) any illicit substance use disorder(OR = 1.72, 95% CI: 1.23–2.39, p = 0.001)(Fig. 7B). Each of these substance use variables,except cannabis use, was significantly associatedwith suicide attempts.

Comorbid cluster B/borderline personality dis-order was strongly associated with suicide attemptsin 5/5 studies, resulting in an OR of 2.51 (95% CI:1.91–3.31, p < 0.00001) (Fig. 8). Data on otherpersonality disorders or traits were not available ina sufficient number of studies to permit meta-analysis.

Finally, a first-degree family history of death bysuicide was found to be associated with suicideattempts, among a total of 7,452 subjects with BD(OR = 1.69, 95% CI: 1.25–2.27, p = 0.0006)(Fig. 9).

Findings related to suicide deaths

Twelve studies on completed suicide were available(2, 4, 8, 12, 15, 21, 40, 73–77). We were able toexamine four variables: gender, lifetime history ofpsychotic symptoms, any substance use disorder,and first-degree family history of suicide. Despitethe smaller number of studies, a total of 75,137subjects with BD were included in these analyses,mostly comprising several large US and Europeancohort studies (8, 12, 15, 40).

The gender-based analysis from 11 studiesincluded a large sample size of 75,055 subjects withBD and a total of 1,149 suicide deaths. Suicidedeaths were significantly associated with male gen-der (OR = 1.83, 95% CI: 1.41–2.39, p < 0.00001)(Fig. 10), with each study reporting an effect in thesame direction, most at a significant level.

Similar to the data for suicide attempts, a historyof psychosis had no significant association withsuicide deaths (OR = 0.93, 95% CI: 0.50–1.74,p = 0.82) (Supplementary Fig. 4). However, in con-trast to the finding for suicide attempts, the

4

Schaffer et al.

presence of any substance use disorder was not sig-nificantly associated with suicide deaths(OR = 1.20, 95% CI: 0.93–1.56, p = 0.17) (Supple-mentary Fig. 5), with 3/4 individual studies findingno significant association.

Only four studies reported on first-degree familyhistory of suicide, but nonetheless a significantassociation was found (OR = 2.91, 95% CI: 1.54–5.48, p = 0.001) (Fig. 11).

Figure 12 provides a list of variables that wereor were not associated with suicide attempts or sui-cide in people with BD.

Sensitivity analyses for effect of very large stud-ies found no switch from variables being significantto non-significant, or vice versa. Furthermore,using random effect analyses, even for variableswith non-significant heterogeneity, resulted in onlyvery minor changes to ORs and modest widening

su

11,160

16,231 28,011

1,269 21,200

1,226

2,4791,881

χ2 = 23.86, (p = 0.20)(p < 0.00001)

I2 = 20%

Odds ratio (non-event)Odds ratio (non-event)

Engström

Fig. 1. Meta-analysis of suicide attempts among males and females with bipolar disorder.CI = confidence interval; M-H = Mantel-Haenszel.

τ2 = 1.19 χ2 = 33.79Z = 7.41 (p < 0.00001)

(p < 0.004)

1,556

4,186

Study or subgroupAge at onset (attempters) Age at onset (Non-attempt) Mean difference Mean difference

Younger age at onset Older age at onset

2,540 9,119

[–3.85 to –1.33][–4.54 to –2.66][–3.27 to 0.69]

[–3.53 to –1.37][–10.92 to –2.68]

[–13.44 to –4.36][–5.68 to –2.32][–5.48 to –2.36][–6.47 to –2.33][–6.19 to –1.81]

–

–––

––

[–3.78 to –2.20]

d.f.

ö

Fig. 2. Meta-analysis of suicide attempts based on age at onset of bipolar disorder.CI = confidence interval; IV = inverse variance; SD = standard deviation.

5

ISBD Task Force on suicide in bipolar disorder

of CIs, with the exception being the associationbetween presence of any substance use disorderand suicide deaths, which increased to OR = 1.49(95% CI = 0.88–2.55).

Discussion

This paper reports on a comprehensive set ofmeta-analyses conducted to identify and quantify

the correlates of suicide attempts and suicidedeaths in BD populations. It was undertaken aspart of the work of the ISBD Task Force on Sui-cide – an international collaborative effort to studysuicide in BD. There have been prior meta-analysespublished on BD subtype and suicide attempts(39), and on a broader examination of demo-graphic and clinical correlates of suicide attemptsand suicide in BD, but this latter work by Hawton

su

τ2 = χ2 = p =p <Z = 3.98

[1.59–2.67][0.94–1.36][1.33–7.54][1.39–4.32][0.91–3.21][1.57–4.54][1.12–2.86]

[1.39–2.65]

1,095 1,121

2,4862,200

Fig. 3. Meta-analysis of suicide attempts based on polarity of first episode of bipolar disorder. BD-I = bipolar I disorder; BD-II = bipolar II disorder; CI = confidence interval; M-H = Mantel-Haenszel.

Study or subgroup

1,075

: τ2 = 0.66; χ2 = 4.80, d.f. = 2 (p = 0.09); I2 = 58%p =Z =

[4.97–646.72][1.32–7.60]

[1.27–15.58]

[1.75–20.50]

Fig. 4. Meta-analysis of suicide attempts based on polarity of current or most recent episode of bipolar disorder. CI = confidenceinterval; M-H = Mantel-Haenszel.

su

7,389

10,348 30,620

1,143

2,1841,096

25,345

: τ2 = 18.53, d.f. = 12 (p = 0.10); I2 = 35%p <

[1.02–1.88][0.83–2.43][1.67–3.15][0.57–4.39][1.13–2.69]

[1.38–23.48][0.47–2.04][0.72–2.35][1.76–4.20][1.47–5.41][1.61–1.99][1.19–6.12][0.79–4.43]

[1.66–1.97]

2,011 2,349

Fig. 5. Meta-analysis of suicide attempts in bipolar disorder based on the presence of comorbid anxiety disorder. CI = confidenceinterval; M-H = Mantel-Haenszel.

6

Schaffer et al.

et al. (38) was published in 2005, and therefore didnot include the large number of studies publishedin the past decade. Our results are noteworthy inreporting on a number of general and BD-specificvariables that are relevant for understanding riskof suicide attempts and suicide specific to a popula-tion with BD. Quantifying the associations permitsthe ranking of correlates and informs risk esti-mates in a more meaningful way.

Of the variables for which sufficient data wereavailable to conduct meta-analyses, ten out of 13were significantly associated with suicide attempts,and two out of four were significantly associatedwith suicide deaths. Factors significantly associ-ated with suicide attempts were (ranked from high-est to lowest ORs): depressive polarity of currentor recent episode (OR = 5.99), comorbid clusterB/borderline personality disorder (OR = 2.51),depressive polarity of first illness episode(OR = 1.92), comorbid anxiety disorder(OR = 1.81), any substance use disorder(OR = 1.81), any illicit substance use (OR = 1.72),first-degree family history of suicide (OR = 1.69),alcohol use disorder (OR = 1.60), and female gen-der (OR = 1.54). Earlier age at illness onset wasalso significantly associated with suicide attempts(mean difference 2.99 years, OR = 1.69).

The evidence for depressive polarity of current/most recent episode and first mood episode eachbeing correlated with suicide attempts was gener-ated from clinical samples. Mixed symptoms havepreviously been associated with elevated risk forsuicide attempts (29, 64), but rates of mixed

episodes were low in the studies we analyzed, andwere most often classified together with manic epi-sodes. Recent evidence suggests that broadlydefined mixed states may be associated with thehighest risk of suicide attempts per time periodspent in a specific phase of illness (78), and withthe new broader definition of mixed states inDSM-5, future studies should be able to examinemore accurately the impact of mixed symptoms onsuicide risk, whether during a manic or depressivephase. Nonetheless, our analyses identified currentor most recent depressive episode as the strongestcorrelate of suicide attempts, likely as a result of acombination of elevated risk per time period, aswell as the predominance of the depressive phaseof illness in the natural course of BD.

The comorbidity between cluster B/borderlinepersonality disorder and BD has long been asource of diagnostic complexity, requiring com-prehensive etiological and management consider-ations (79). Our data identified that elevated riskof suicide attempt (OR = 2.51) is another impor-tant factor to consider, with all five studies dem-onstrating a strong association. It is possiblethat having recurrent suicidal behavior or self-harm without intent to die as part of the diag-nostic criteria for borderline personality disordermay have resulted in an inflation of the associa-tion, and we could not address this issue withthe available data; however, other cluster B per-sonality disorders do not include this criterion,and as such it is unlikely to fully account forthe identified association.

4,935

1,532 2,114

1,116 27,799

7,726 31,413

Study or subgroupAny substance use No substance use Odds ratio Odds ratio

: τ2 = 0.37; χ2 = 159.90, d.f. = 17 (p = 0.00001); I2 = 89%Z = 3.57 (p = 0.0004)

–––––––

––––––––

–––

–

Fig. 6. Meta-analysis of suicide attempts in bipolar disorder based on the presence of any substance use disorder. CI = confidenceinterval; M-H = Mantel-Haenszel.

7

ISBD Task Force on suicide in bipolar disorder

The finding of comorbid anxiety being signifi-cantly associated with suicide attempts was highlyconsistent across studies. A majority of people withBD will experience a comorbid anxiety disorder at

some point in their lives (24, 80), so elevated ratesof suicide attempts are highly clinically relevant.There is some evidence that the associationbetween anxiety and suicide attempts in BD may

1,636

1,719

1,081 1,572

2,641

4,566 7,969

: τ2 = 0.09; χ2 = 43.31, d.f. = 15 (p = 0.0001); I2 = 65%: Z = 4.51 (p < 0.00001)

Odds ratio Odds ratio

1,188

1,646

2,057

3,172

8,3712,202

: τ2 = 0.19; χ2 = 43.01,d.f. = 10 (p = 0.00001); I2 = 77%(p = 0.001)

A

B

[1.60–2.39][0.70–2.69][0.68–1.96][1.18–2.17][0.05–5.90][1.02–2.44]

[0.82–12.52][1.51–2.49][1.52–3.34][0.94–3.67][1.28–3.93][0.93–4.66]

[0.96–11.04][0.90–3.33][0.21–1.17]

[1.31–1.97]

[1.23–2.39]

[0.58–1.10][2.15–3.32][1.03–4.21][1.13–2.12]

[0.48–40.31][0.96–2.51]

[0.38–42.37][0.94–2.42][0.46–2.70][1.03–5.71][1.29–4.58]

[0.70–1.16]

Fig. 7. Meta-analyses of suicide attempts in bipolar disorder based on the presence of alcohol use disorder (A) or any illicit substanceuse disorder (B). CI = confidence interval; M-H = Mantel-Haenszel.

cl

cl Cl

: τ2 = 6.29 , d.f. = 4 (p = 0.18); I2 = 36%Z = 6.55 (p < 0.00001)

Fig. 8. Meta-analysis of suicide attempts in bipolar disorder based on the presence of a comorbid cluster B/borderline personalitydisorder. CI = confidence interval; M-H = Mantel-Haenszel.

8

Schaffer et al.

be partially mediated through increased rumina-tion (24), as well as by comorbid cluster B person-ality disorders (47). Unfortunately, most studiesdo not report on the sequencing of the comorbidsymptoms or disorders and the suicide attempt(s),which may or may not be confluent. It is thereforedifficult to develop a clear attribution model

without sufficient prospective data to elucidate theonset and timing of the comorbid anxiety in rela-tion to the suicide attempt.

The association between suicide attempt andsubstance use disorders was of a similar magnitudeto the association with anxiety (both OR = 1.81),but the results for substance use were more variable

: τ2 = 0.13; χ2 = 23.50, d.f. = 10 (p = 0.009); I2 = 57%Z = 3.44 (p = 0.0006) Favours [No history] Favours [Family history]

4,076

6,473

[0.88–5.63][0.87–2.42][0.87–2.51][0.68–5.68][0.26–2.81][0.36–5.33][1.20–2.35][0.66–1.36][1.70–6.00][2.12–7.06][0.39–3.26]

[1.25–2.27]

Engström

Fig. 9. Meta-analysis of suicide attempts in bipolar disorder based on the presence of a family history of suicide. CI = confidenceinterval; M-H = Mantel-Haenszel.

1,819

4,8452,5716,578

11,160

28,694 46,361

: τ2 = 0.08; χ2 = 22.35, d.f. = 10 (p = 0.01); I2 = 55%(p = 0.00001)

[0.88–61.53][0.79–19.35][1.18–80.37][0.71–1.59][1.22–2.23][1.23–1.68][1.01–2.90][0.29–5.63][1.54–3.88][1.12–6.87]

[2.29–44.96]

[1.41–2.39]

3,3568,8081,032

21,200

8,341

2,359

Fig. 10. Meta-analysis of suicide deaths among males and females with bipolar disorder. CI = confidence interval; M-H = Mantel-Haenszel.

fa

Favours [No fam. history] Favours [Fam. history]

Odds ratio Odds ratio

: τ2 = 3.67, d.f. = 3 ( = 0.30); (p = 0.0010)

p

[0.05–14.61]4,076

2,182

6,601

[0.74–4.80][1.88–13.85][0.96–70.32]

[1.54–5.48]

I2 = 18%

Fig. 11. Meta-analysis of suicide deaths in bipolar disorder based on the presence of a family history of suicide. BD = bipolar disor-der; CI = confidence interval; M-H = Mantel-Haenszel.

9

ISBD Task Force on suicide in bipolar disorder

across studies, with several showing a trend in theopposite direction. This led us to break down thesubstance use category into alcohol use disorder,any illicit drug use, and specific cannabis use. Sig-nificant associations were identified among studiesof alcohol use disorder (OR = 1.60) and illicit druguse (OR = 1.72), but not for cannabis use(OR = 1.29). While the largest study of cannabisuse did report a significant association with suicideattempts, the other three studies found no signifi-cant effect. In a recent review by Watkins andMeyer (81), preliminary evidence identified impul-sivity as mediating the association between alcoholuse and suicide attempts in BD, but whether this istrue across substance use disorders is not known.

Female gender was associated with suicideattempts (OR = 1.54), but this was a weaker asso-ciation than has most commonly been reported innon-BD samples (27, 82, 83), and only 8/20 studiesin our analysis reported any significant gender-based association. Results of the meta-regression

did not identify any significant covariates for thisassociation, but there may be other diagnosis-spe-cific factors, such as a greater female preponder-ance for a more depression-prone course of bipolarillness (84, 85), that are additionally relevant. It isalso worth noting that men with BD were under-represented in the available trials, accounting foronly 36.7% of all subjects in the studies thatreported on suicide attempts. Given the lack ofgender-based differences in the prevalence of BD,this suggests that an ascertainment or samplingbias may also be relevant when interpreting theseresults. Overall, the findings highlight the impor-tance of a diagnostic-specific examination of corre-lates of suicide attempts such as gender, asextrapolation from a broader literature on gender-based differences may be inaccurate.

Earlier age at onset of BD was also significantlyassociated with a history of suicide attempts, witha mean difference of 2.99 years. Early onset of ill-ness has consistently been shown to have negative

Fig. 12. List of variables that were or were not associated with suicide attempts or suicide deaths in people with bipolar disorderbased on meta-analytic results.

10

Schaffer et al.

prognostic implications (86, 87), and our data sup-port this observation. It is worth noting that weonly extracted data from reports of study popula-tions exclusively > 13 years of age, and 27/32 sam-ple groups for this analysis had a mean age atillness onset of ≥ 19 years; therefore, our findingsprimarily relate to adult-onset BD.

There were three variables tested that were notsignificantly associated with suicide attempts. Incontrast to the limited data on cannabis use, therewere more studies on BD subtype (14 studies) andhistory of psychosis (seven studies). Considerablevariability was present in the BD subtype studies,with ORs for BD-I varying from 2.38 to 0.27.There were over twice the number of subjects withBD-I compared to subjects with BD-II, which isnot in line with the relative equal lifetime preva-lence of the two subtypes (88), and again suggeststhat there may be a sample bias in the available lit-erature. Nonetheless, analysis of current data sug-gests that subjects with BD-II are just as likely toattempt suicide as those with BD-I.

History of psychosis was examined in relation toboth suicide attempts and suicide deaths, and nei-ther analysis identified a significant association,although there was a wide confidence interval(0.50–1.74) in the suicide death analysis. It isimportant to note that studies identified a historyof psychotic symptoms at any point in the courseof illness, and did not focus on specific phases ofillness. Given the strong correlation between cur-rent or recent depressive episode and suicideattempts, it is possible that a current or recent psy-chotic depression may be associated with suicideattempts or even suicide deaths, but this has notclearly been shown in other studies (89), and datawere not available to test this hypothesis in BD.

Another challenge is that there are far fewerstudies of correlates of suicide deaths in BD, ascompared to suicide attempts. Suicide deaths arerarer and inherently more difficult to study, as evi-denced by only four variables being sufficientlyexamined to allow for meta-analysis. In additionto psychosis, other variables tested included gen-der, any substance use disorder, and first-degreefamily history of suicide. Men with BD were nearlytwice as likely to die by suicide as compared towomen with BD (OR = 1.83). This is in keepingwith the consistent evidence on gender differencesin suicide rates but, as with suicide attempts in BD,the size of the difference is smaller than in generalsuicide samples (28), which show up to a fourfolddifference, and is more in line with an approximate2:1 ratio identified in a review of suicide in depres-sion (26). Gender-based differences in methods ofsuicide have been reported in the broad literature,

and may also be relevant to understanding gender-based differences in BD, but there were insufficientdata to examine this possibility in detail.

A noteworthy finding is that a lifetime history ofany substance use disorder was not associated withhigher rates of suicide deaths in BD; however, theOR of 1.20 (95% CI: 0.93–1.56) was in the direc-tion of a positive correlation. Nonetheless, thiscontrasts with the elevated risk of suicide attemptsin our analyses and the literature on higher rates ofsuicide deaths in general samples (90) as well asspecific to depression (26) and schizophrenia (91).One possible explanation relates to sample charac-teristics, as the two most heavily weighted studiesin our meta-analysis (Supplementary Fig. 4) werelarge population-based epidemiological samplesthat did not find a significant association, in con-trast to the one pure clinical sample from the Sys-tematic Treatment Enhancement Program forBipolar Disorder (STEP-BD), which reported astrong association (OR = 8.05). This suggests thepossible impact of Berkson’s bias (92), with a moresevere sample of comorbid patients entering into aclinical protocol. This hypothesis, however, is spec-ulative at best, and reinforces the importance ofhaving more data on the specific connectionbetween substance use and suicide deaths in BD,including mediating factors and greater details ofthe substance use in relation to phases of illness.Until these can be examined, the methodologicallimitations of our analysis and the divergence fromthe bulk of the broader literature on substance useand suicide suggest that our result must be inter-preted with appropriate caution. It would also beimportant to further understand the association asit pertains to the timing of substance use, as evi-denced by recent data on self-poisoning deathsamong people with BD which identified 41% ofcases having alcohol in the system at the time ofdeath (93).

Finally, first-degree family history of suicide wasfound to have the strongest association with sui-cide deaths in BD (OR = 2.91). This supports priorliterature on the significance of family history ofsuicide on elevating suicide risk from general psy-chiatric samples (94–96), and reinforces the impor-tance of integrating genetic, epigenetic, and sociallearning effects when building models of psychobi-ological contributors to suicidal behavior in BD(36, 97).

There are a number of important limitationsthat must be considered when interpreting thesedata. First, considerable heterogeneity exists in theavailable literature, but as the number of studies isnot very large, we included all studies with eitherepidemiological or clinical samples. The largest

11

ISBD Task Force on suicide in bipolar disorder

studies included were usually prospective epidemi-ological samples, and the predominant concor-dance of results with smaller clinical samplesshould serve to strengthen the generalizability ofour results. However, it is possible that the magni-tude or even presence of specific associations maybe different within certain BD subpopulations.This degree of granularity could not be determinedbased on the available literature. Second, as notedearlier, the relationship in time between the vari-ables of interest and the outcome of a suicideattempt or suicide death was not always known.For instance, higher lifetime rates of suicideattempts in subjects with a lifetime comorbid anxi-ety disorder does not provide information onwhether these were contemporaneous and there-fore cannot be used to determine the direction ofthe effect or other interactions. This was not thecase with all variables as a number are not timebased but, nonetheless, the usual caution aroundassociation not equaling causation is relevant here.In a related limitation, we included studies withretrospective or prospective designs; therefore, it isimportant to consider the significant findings asbeing evidence of correlation rather than for anyattempt to assign levels of risk of suicide attemptsor suicide for a particular patient. Retrospectivestudies also carry the limitation of recall bias;therefore, while the reported data can serve toinform future studies aimed at determining thistype of risk, prospective stratified designs arerequired. In a similar vein, the findings do notaddress causation, and the heterogeneity of datasources should limit the direct clinical ranking ofcorrelates. The results of one variable should there-fore not be compared to another as different studieswere used in the analyses. An additional limitationis that data on correlates of suicide attempts shouldnot be assumed to be relevant for understandingrisk of suicide deaths, as evidenced by the well-described gender paradox (98) also seen here, aswell as the discordant results on substance use dis-orders. Finally, a number of putatively importantvariables such as past suicide attempts, hopeless-ness, recent hospitalization, childhood abuse, andsmoking could not be meta-analyzed because therewere insufficient studies with the required datareporting. This precludes considering the correlatesreported in the present paper as an exhaustive list ofvariables associated with suicide attempts or suicidein BD, and reinforces the importance of includingall potential variables of interest in future studies ofsuicidal behavior in BD.

The principal objective of the present report wasto identify correlates of suicide attempts and sui-cide deaths in studies specific to BD. The meta-

analytic results identified ten significant correlatesof suicide attempts and two significant correlatesof suicide deaths, some of which are only relevantto BD, and others that have been studied morebroadly. As discussed earlier, there are a few note-worthy differences between the results we obtainedand the broader literature on suicidal behavior anddeaths among general samples (28), or even thosefocused on unipolar depressive disorder (26) orschizophrenia (91). While appropriate caution isrequired when interpreting the findings, our resultssupport the importance of incorporating diagno-sis-specific data on clinical variables relevant toBD, as well on the diagnosis-specific results from abroader set of correlates, into the development ofmodels for understanding suicide risk in BD (29,36, 99). This conclusion has implications for futurework on developing specific preventative programsand therapeutic interventions for people with BD.

Acknowledgements

The authors wish to thank the International Society for BipolarDisorders executive and staff who assisted with the organiza-tion of the task force, and the students who assisted with theliterature review and provided statistical work (Jessika Lenchy-shyn, BSc, Randy Rovinski, MSc). Partial support for this pro-ject was provided by the Brenda Smith Bipolar DisorderResearch Fund, Sunnybrook Health Sciences Centre, Univer-sity of Toronto, Toronto, ON, Canada.

Affiliations

aDepartment of Psychiatry, Sunnybrook Health Sciences Cen-tre and Department of Psychiatry, University of Toronto, Tor-onto ON Canada, bDepartment of Psychiatry, University ofHelsinki, Helsinki, Finland, cLucio Bini Center, Cagliari, Italy,dHarvard Medical School, McLean Hospital, Boston MAUSA, eDepartment and Institute of Psychiatry, University ofS~ao Paulo, S~ao Paulo, Brazil, fDepartments of Psychiatry,Human Genetics, and Neurology & Neurosurgery, McGillUniversity, Montreal QC Canada, gDepartment of Psychiatryand Behavioural Sciences, Duke University, Durham NCUSA, hDepartment of Psychiatry, University of Aix-MarseilleII, Marseille, France, iFaculty of Health Sciences, Universityof Copenhagen, Psychiatric Center Copenhagen Department,Copenhagen, Denmark, jDepartment of Psychiatry, SeoulNational University, Bundang Hospital, Seoul, Korea,kDepartment of Child and Adolescent Psychiatry, Universityof Pittsburgh School of Medicine, Pittsburgh PA USA,lDepartment of Psychiatry, Sackler Faculty of Medicine, TelAviv University, Tel Aviv, Israel, mDepartment of EmergencyMedicine, Yale School of Medicine, Yale University, NewHaven CT USA, nDepartment of Psychiatry, Taipei VeteransGeneral Hospital and National Yang Ming University, Taipei,Taiwan, oNational Institute of Alcohol Abuse and Alcoholism,pNational Institute of Mental Health, Bethesda MD USA,qDepartment of Clinical and Theoretical Mental Health andDepartment of Psychiatry and Psychotherapy, SemmelweisMedical University, Budapest, Hungary, rDepartment ofPsychiatry, University of British Columbia, Vancouver BCCanada

12

Schaffer et al.

Disclosures

AS has received research grants, speakers bureauhonoraria, and/or advisory panel funding from EliLilly Canada, Bristol-Myers Squibb, Pfizer Can-ada, Lundbeck Canada, and Sunovion. ETI hasreceived honoraria from Servier for lecturing ineducational meetings. LT has received fundingfrom private donors at Aretæus Association and atMcLean Hospital. DHM has received grant sup-port or served as a speaker for Abbott, Ach�e,Lundbeck, EMS, and Eurofarma. MS has receivedgrant support from the Physicians’ Services Incor-porated (PSI) Foundation and the Brenda SmithBipolar Disorder Research Fund. J-MA hasreceived research support and has acted as consul-tant and/or served on a speakers bureau for Bris-tol-Myers Squibb, Janssen, Eli Lilly & Co.,Lundbeck, Roche, and Sanofi-Aventis. LVK has,within the preceding three years, been a consultantfor Lundbeck and AstraZeneca. TG has receivedresearch support from NIMH, NIDA, NICHD,and The Pittsburgh Foundation; and Royaltiesfrom Guilford Press. AJL has received researchgrants from Janssen Ortho, AstraZeneca, GreatWest Life Insurance, and Eli Lilly Canada; andhas acted as a consultant for Janssen Ortho. CAZis listed as a co-inventor on a patent applicationfor the use of ketamine and its metabolites inmajor depression, and has assigned his rights in thepatent to the US government but will share a per-centage of any royalties that may be received bythe government. LNY has received research grantsand/or been a member of advisory boards and aspeaker for AstraZeneca, Dainippon Sumitomo,Janssen, Eli Lilly & Co., GlaxoSmithKline, Bris-tol-Myers Squibb, Lundbeck, Novartis, Servier,Sunovion, and Pfizer. GT, CR, FC, KH, AW, AB,Y-HC, ND, and ZR do not have any conflicts ofinterest to report.

References

1. Roy A. Risk factors for suicide in psychiatric patients.Arch Gen Psychiatry 1982; 39: 1089–1095.

2. Tondo L, Lepri B, Baldessarini RJ. Suicidal risks among2826 Sardinian major affective disorder patients. ActaPsychiatr Scand 2007; 116: 419–428.

3. Brown GK, Beck AT, Steer RA, Grisham JR. Riskfactors for suicide in psychiatric outpatients: a 20-yearprospective study. J Consult Clin Psychol 2000; 68:371–377.

4. Sani G, Tondo L, Koukopoulos A et al. Suicide in alarge population of former psychiatric inpatients. Psychi-atry Clin Neurosci 2011; 65: 286–295.

5. Osborn D, Levy G, Nazareth I, King M. Suicide andsevere mental illnesses. Cohort study within the UK gen-eral practice research database. Schizophr Res 2008; 99:134–138.

6. Barner-Rasmussen P. Suicide in psychiatric patients inDenmark, 1971-1981. II. Hospital utilization and riskgroups. Acta Psychiatr Scand 1986; 73: 449–455.

7. Hoyer EH, Mortensen PB, Olesen AV. Mortality andcauses of death in a total national sample of patients withaffective disorders admitted for the first time between1973 and 1993. Br J Psychiatry 2000; 176: 76–82.

8. Osby U, Brandt L, Correia N, Ekbom A, Sparen P.Excess mortality in bipolar and unipolar disorder inSweden. Arch Gen Psychiatry 2001; 58: 844–850.

9. Karch DL, Barker L, Strine TW. Race/ethnicity, sub-stance abuse, and mental illness among suicide victimsin 13 US states: 2004 data from the National ViolentDeath Reporting System. Inj Prev 2006; 12 (Suppl. 2):ii22–ii27.

10. Baillargeon J, Penn JV, Thomas CR, Temple JR,Baillargeon G, Murray OJ. Psychiatric disorders andsuicide in the nation’s largest state prison system. J AmAcad Psychiatry Law 2009; 37: 188–193.

11. Chen YY, Lee MB, Chang CM, Liao SC. Methods ofsuicide in different psychiatric diagnostic groups. J AffectDisord 2009; 118: 196–200.

12. Nordentoft M, Mortensen PB, Pedersen CB. Absoluterisk of suicide after first hospital contact in mental disor-der. Arch Gen Psychiatry 2011; 68: 1058–1064.

13. Takizawa T. Suicide due to mental diseases based on theVital Statistics Survey Death Form. Nihon Koshu EiseiZasshi 2012; 59: 399–406.

14. Ilgen MA, Bohnert AS, Ignacio RV et al. Psychiatricdiagnoses and risk of suicide in veterans. Arch Gen Psy-chiatry 2010; 67: 1152–1158.

15. Simon GE, Hunkeler E, Fireman B, Lee JY, Savarino J.Risk of suicide attempt and suicide death in patientstreated for bipolar disorder. Bipolar Disord 2007a; 9:526–530.

16. Tondo L, Isacsson G, Baldessarini R. Suicidal behaviourin bipolar disorder: risk and prevention. CNS Drugs2003; 17: 491–511.

17. Sønderg�ard L, Lopez AG, Andersen PK, Kessing LV.Mood-stabilizing pharmacological treatment in bipolardisorders and risk of suicide. Bipolar Disord 2008; 10:87–94.

18. Pompili M, Gonda X, Serafini G et al. Epidemiology ofsuicide in bipolar disorders: a systematic review of the lit-erature. Bipolar Disord 2013; 15: 457–490.

19. Crump C, Sundquist K, Winkleby MA, Sundquist J.Comorbidities and mortality in bipolar disorder: a Swed-ish national cohort study. JAMA Psychiatry 2013; 70:931–939.

20. Harris EC, Barraclough B. Suicide as an outcome formental disorders. A meta-analysis. Br J Psychiatry 1997;170: 205–228.

21. Kessing LV, Sondergard L, Kvist K, Andersen PK. Sui-cide risk in patients treated with lithium. Arch Gen Psy-chiatry 2005; 62: 860–866.

22. Gonda X, Pompili M, Serafini G et al. Suicidal behaviorin bipolar disorder: epidemiology, characteristics andmajor risk factors. J Affect Disord 2012; 143: 16–26.

23. Chen YW, Dilsaver SC. Lifetime rates of suicideattempts among subjects with bipolar and unipolar dis-orders relative to subjects with other Axis I disorders.Biol Psychiatry 1996; 39: 896–899.

24. Simon NM, Zalta AK, Otto MW et al. The associationof comorbid anxiety disorders with suicide attempts andsuicidal ideation in outpatients with bipolar disorder.J Psychiatr Res 2007b; 41: 255–264.

13

ISBD Task Force on suicide in bipolar disorder

25. Schaffer A, Cairney J, Veldhuizen S, Kurdyak P, CheungA, Levitt A. A population-based analysis of distinguish-ers of bipolar disorder from major depressive disorder.J Affect Disord 2010; 125: 103–110.

26. Hawton K, Casanas ICC, Haw C, Saunders K. Risk fac-tors for suicide in individuals with depression: a system-atic review. J Affect Disord 2013; 147: 17–28.

27. American Psychiatric Association. Practice Guidelinesfor the Assessment and Treatment of Patients with Sui-cidal Behaviors. American Psychiatric Association Prac-tice Guidelines for the Treatment of PsychiatricDisorders: Compendium 2006. American PsychiatricAssociation: Arlington, TX, 2006: 1315–1496.

28. Mann JJ. A current perspective of suicide and attemptedsuicide. Ann Intern Med 2002; 136: 302–311.

29. Saunders KE, Hawton K. Clinical assessment and crisisintervention for the suicidal bipolar disorder patient.Bipolar Disord 2013; 15: 575–583.

30. Chesin M, Stanley B. Risk assessment and psychosocialinterventions for suicidal patients. Bipolar Disord 2013;15: 584–593.

31. Mann JJ, Apter A, Bertolote J et al. Suicide preventionstrategies: a systematic review. JAMA 2005; 294: 2064–2074.

32. Schwartz-Lifshitz M, Zalsman G, Giner L, OquendoMA. Can we really prevent suicide? Curr Psychiatry Rep2012; 14: 624–633.

33. Pitman A, Caine E. The role of the high-risk approach insuicide prevention. Br J Psychiatry 2012; 201: 175–177.

34. Baethge C, Cassidy F. Fighting on the side of life: a spe-cial issue on suicide in bipolar disorder. Bipolar Disord2013; 15: 453–456.

35. Hauser M, Galling B, Correll CU. Suicidal ideation andsuicide attempts in children and adolescents with bipolardisorder: a systematic review of prevalence and incidencerates, correlates, and targeted interventions. Bipolar Dis-ord 2013; 15: 507–523.

36. Malhi GS, Bargh DM, Kuiper S, Coulston CM, Das P.Modeling bipolar disorder suicidality. Bipolar Disord2013; 15: 559–574.

37. Yerevanian BI, Choi YM. Impact of psychotropic drugson suicide and suicidal behaviors. Bipolar Disord 2013;15: 594–621.

38. Hawton K, Sutton L, Haw C, Sinclair J, Harriss L. Sui-cide and attempted suicide in bipolar disorder: a system-atic review of risk factors. J Clin Psychiatry 2005; 66:693–704.

39. Novick DM, Swartz HA, Frank E. Suicide attempts inbipolar I and bipolar II disorder: a review and meta-analysis of the evidence. Bipolar Disord 2010; 12: 1–9.

40. Dennehy EB, Marangell LB, Allen MH, Chessick C,Wisniewski SR, Thase ME. Suicide and suicide attemptsin the Systematic Treatment Enhancement Program forBipolar Disorder (STEP-BD). J Affect Disord 2011; 133:423–427.

41. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferredreporting items for systematic reviews and meta-analy-ses: the PRISMA statement. J Clin Epidemiol 2009; 62:1006–1012.

42. Romero S, Colom F, Iosif AM et al. Relevance of familyhistory of suicide in the long-term outcome of bipolardisorders. J Clin Psychiatry 2007; 68: 1517–1521.

43. Cassidy F. Risk factors of attempted suicide in bipolardisorder. Suicide Life Threat Behav 2011; 41: 6–11.

44. Sublette ME, Carballo JJ, Moreno C et al. Substanceuse disorders and suicide attempts in bipolar subtypes.J Psychiatr Res 2009; 43: 230–238.

45. Potash JB, Kane HS, Chiu YF et al. Attempted suicideand alcoholism in bipolar disorder: clinical and familialrelationships. Am J Psychiatry 2000; 157: 2048–2050.

46. Oquendo MA, Currier D, Liu SM, Hasin DS, Grant BF,Blanco C. Increased risk for suicidal behavior in comor-bid bipolar disorder and alcohol use disorders: resultsfrom the National Epidemiologic Survey on Alcohol andRelated Conditions (NESARC). J Clin Psychiatry 2010;71: 902–909.

47. Nakagawa A, Grunebaum MF, Sullivan GM et al.Comorbid anxiety in bipolar disorder: does it have anindependent effect on suicidality? Bipolar Disord 2008;10: 530–538.

48. Malloy-Diniz LF, Neves FS, Abrantes SS, Fuentes D,Correa H. Suicide behavior and neuropsychologicalassessment of type I bipolar patients. J Affect Disord2009; 112: 231–236.

49. Valtonen HM, Suominen K, Mantere O, Leppamaki S,Arvilommi P, Isometsa E. Suicidal behaviour during dif-ferent phases of bipolar disorder. J Affect Disord 2007;97: 101–107.

50. Liu CY, Bai YM, Yang YY, Lin CC, Sim CC, Lee CH.Suicide and parasuicide in psychiatric inpatients: tenyears experience at a general hospital in Taiwan. PsycholRep 1996; 79: 683–690.

51. Neves FS, Malloy-Diniz LF, Correa H. Suicidal behav-ior in bipolar disorder: what is the influence of psychiat-ric comorbidities? J Clin Psychiatry 2009; 70: 13–18.

52. Chaudhury SR, Grunebaum MF, Galfalvy HC et al.Does first episode polarity predict risk for suicideattempt in bipolar disorder? J Affect Disord 2007; 104:245–250.

53. Tondo L, Baldessarini RJ, Hennen J et al. Suicideattempts in major affective disorder patients with comor-bid substance use disorders. J Clin Psychiatry 1999; 60(Suppl. 2): 63–69.

54. Joyce PR, Light KJ, Rowe SL, Cloninger CR, KennedyMA. Self-mutilation and suicide attempts: relationshipsto bipolar disorder, borderline personality disorder, tem-perament and character. Aust N Z J Psychiatry 2010; 44:250–257.

55. Bega S, Schaffer A, Goldstein B, Levitt A. Differentiat-ing between bipolar disorder types I and II: results fromthe National Epidemiologic Survey on Alcohol andRelated Conditions (NESARC). J Affect Disord 2012;138: 46–53.

56. Valtonen HM, Suominen K, Mantere O, Lepp€am€aki S,Arvilommi P, Isomets€a ET. Prospective study of risk fac-tors for attempted suicide among patients with bipolardisorder. Bipolar Disord 2006; 8: 576–585.

57. Swann AC, Dougherty DM, Pazzaglia PJ, Pham M,Steinberg JL, Moeller FG. Increased impulsivity associ-ated with severity of suicide attempt history in patientswith bipolar disorder. Am J Psychiatry 2005; 162: 1680–1687.

58. Sanchez-Gistau V, Colom F, Mane A, Romero S,Sugranyes G, Vieta E. Atypical depression is associatedwith suicide attempt in bipolar disorder. Acta PsychiatrScand 2009; 120: 30–36.

59. Ryu V, Jon DI, Cho HS et al. Initial depressive episodesaffect the risk of suicide attempts in Korean patients withbipolar disorder. Yonsei Med J 2010; 51: 641–647.

60. Perroud N, Baud P, Preisig M et al. Social phobia isassociated with suicide attempt history in bipolar inpa-tients. Bipolar Disord 2007; 9: 713–721.

61. Parmentier C, Etain B, Yon L, et al. Clinical and dimen-sional characteristics of euthymic bipolar patients with

14

Schaffer et al.

or without suicidal behavior. Eur Psychiatry 2012; 27:570–576.

62. Moor S, Crowe M, Luty S, Carter J, Joyce PR. Effects ofcomorbidity and early age of onset in young people withbipolar disorder on self harming behaviour and suicideattempts. J Affect Disord 2012; 136: 1212–1215.

63. Grunebaum MF, Ramsay SR, Galfalvy HC et al. Corre-lates of suicide attempt history in bipolar disorder: astress-diathesis perspective. Bipolar Disord 2006; 8: 551–557.

64. Goldberg JF, Garno JL, Leon AC, Kocsis JH, PorteraL. Association of recurrent suicidal ideation with nonre-mission from acute mixed mania. Am J Psychiatry 1998;155: 1753–1755.

65. Gazalle FK, Hallal PC, Tramontina J et al. Polypharma-cy and suicide attempts in bipolar disorder. Rev BrasPsiquiatr 2007; 29: 35–38.

66. Galfalvy H, Oquendo MA, Carballo JJ et al. Clinicalpredictors of suicidal acts after major depression in bipo-lar disorder: a prospective study. Bipolar Disord 2006; 8:586–595.

67. Azorin JM, Kaladjian A, Adida M et al. Risk factorsassociated with lifetime suicide attempts in bipolar Ipatients: findings from a French National Cohort.Compr Psychiatry 2009; 50: 115–120.

68. Bellivier F, Yon L, Luquiens A et al. Suicidal attemptsin bipolar disorder: results from an observational study(EMBLEM). Bipolar Disord 2011; 13: 377–386.

69. Dalton EJ, Cate-Carter TD, Mundo E, Parikh SV,Kennedy JL. Suicide risk in bipolar patients: the role ofco-morbid substance use disorders. Bipolar Disord 2003;5: 58–61.

70. Engstr€om C, Br€andstr€om S, Sigvardsson S, CloningerCR, Nylander PO. Bipolar disorder. III: harm avoidancea risk factor for suicide attempts. Bipolar Disord 2004; 6:130–138.

71. Finseth PI, Morken G, Andreassen OA, Malt UF,Vaaler AE. Risk factors related to lifetime suicideattempts in acutely admitted bipolar disorder inpatients.Bipolar Disord 2012; 14: 727–734.

72. Leverich GS, Altshuler LL, Frye MA et al. Factors asso-ciated with suicide attempts in 648 patients with bipolardisorder in the Stanley Foundation Bipolar Network. JClin Psychiatry 2003; 64: 506–515.

73. Tsai SY, Kuo CJ, Chen CC, Lee HC. Risk factors forcompleted suicide in bipolar disorder. J Clin Psychiatry2002; 63: 469–476.

74. Weeke A, Vaeth M. Excess mortality of bipolar andunipolar manic-depressive patients. J Affect Disord1986; 11: 227–234.

75. Sharma R, Markar HR. Mortality in affective disorder.J Affect Disord 1994; 31: 91–96.

76. Black DW, Winokur G, Nasrallah A. Effect of psychosison suicide risk in 1,593 patients with unipolar andbipolar affective disorders. Am J Psychiatry 1988; 145:849–852.

77. Dutta R, Boydell J, Kennedy N, VAN Os J, Fearon P,Murray RM. Suicide and other causes of mortality inbipolar disorder: a longitudinal study. Psychol Med2007; 37: 839–847.

78. Holma KM, Haukka J, Suominen K et al. Differences inincidence of suicide attempts between bipolar I and IIdisorders and major depressive disorder. Bipolar Disord2014; 16: 652–661.

79. Rosenbluth M, Macqueen G, McIntyre RS, Beaulieu S,Schaffer A. The Canadian Network for Mood and Anxi-ety Treatments (CANMAT) task force recommendations

for the management of patients with mood disorders andcomorbid personality disorders. Ann Clin Psychiatry2012; 24: 56–68.

80. Schaffer A, McIntosh D, Goldstein BI et al. TheCANMAT task force recommendations for the manage-ment of patients with mood disorders and comorbid anx-iety disorders. Ann Clin Psychiatry 2012; 24: 6–22.

81. Watkins HB, Meyer TD. Is there an empirical linkbetween impulsivity and suicidality in bipolar disorders?A review of the current literature and the potential psy-chological implications of the relationship. Bipolar Dis-ord 2013; 15: 542–558.

82. Kuo WH, Gallo JJ, Tien AY. Incidence of suicide idea-tion and attempts in adults: the 13-year follow-up of acommunity sample in Baltimore, Maryland. PsycholMed 2001; 31: 1181–1191.

83. Schmidtke A, Bille-Brahe U, DeLeo D et al. Attemptedsuicide in Europe: rates, trends and sociodemographiccharacteristics of suicide attempters during the period1989-1992. Results of the WHO/EUROMulticentre Studyon Parasuicide. Acta Psychiatr Scand 1996; 93: 327–338.

84. Nivoli AM, Pacchiarotti I, Rosa AR et al. Gender differ-ences in a cohort study of 604 bipolar patients: the roleof predominant polarity. J Affect Disord 2011; 133: 443–449.

85. Baldessarini RJ, Undurraga J, Vazquez GH et al. Pre-dominant recurrence polarity among 928 adult interna-tional bipolar I disorder patients. Acta Psychiatr Scand2012; 125: 293–302.

86. Azorin JM, Bellivier F, Kaladjian A et al. Characteris-tics and profiles of bipolar I patients according to age-at-onset: findings from an admixture analysis. J AffectDisord 2013; 150: 993–1000.

87. Coryell W, Fiedorowicz J, Leon AC, Endicott J, KellerMB. Age of onset and the prospectively observed courseof illness in bipolar disorder. J Affect Disord 2013; 146:34–38.

88. Merikangas KR, Akiskal HS, Angst J et al. Lifetime and12-month prevalence of bipolar spectrum disorder in theNational Comorbidity Survey replication. Arch GenPsychiatry 2007; 64: 543–552.

89. Leadholm AK, Rothschild AJ, Nielsen J, Bech P,Ostergaard SD. Risk factors for suicide among 34,671patients with psychotic and non-psychotic severe depres-sion. J Affect Disord 2014; 156: 119–125.

90. Murphy GE, Wetzel RD. The lifetime risk of suicide inalcoholism. Arch Gen Psychiatry 1990; 47: 383–392.

91. Gomez-Duran EL, Martin-Fumado C, Hurtado-Ruiz G.Clinical and epidemiological aspects of suicide inpatients with schizophrenia. Actas Esp Psiquiatr 2012;40: 333–345.

92. Westreich D. Berkson’s bias, selection bias, and missingdata. Epidemiology 2012; 23: 159–164.

93. Schaffer A, Sinyor M, Reis C, Goldstein BI, Levitt AJ.Suicide in bipolar disorder: characteristics and sub-groups. Bipolar Disord 2014. doi:10.1111/bdi.12219.

94. Large M, Smith G, Sharma S, Nielssen O, Singh SP. Sys-tematic review and meta-analysis of the clinical factorsassociated with the suicide of psychiatric in-patients.Acta Psychiatr Scand 2011; 124: 18–29.

95. Ernst C, Mechawar N, Turecki G. Suicide neurobiology.Prog Neurobiol 2009; 89: 315–333.

96. Baldessarini RJ, Hennen J. Genetics of suicide: an over-view. Harv Rev Psychiatry 2004; 12: 1–13.

97. Manchia M, Hajek T, O’Donovan C et al. Genetic riskof suicidal behavior in bipolar spectrum disorder: analy-sis of 737 pedigrees. Bipolar Disord 2013; 15: 496–506.

15

ISBD Task Force on suicide in bipolar disorder

98. Schrijvers DL, Bollen J, Sabbe BG. The gender paradoxin suicidal behavior and its impact on the suicidal pro-cess. J Affect Disord 2012; 138: 19–26.

99. Sublette ME, Oquendo MA, Mann JJ. Rationalapproaches to the neurobiologic study of youth at riskfor bipolar disorder and suicide. Bipolar Disord 2006; 8:526–542.

100. Suominen K, Mantere O, Valtonen H et al. Gender dif-ferences in bipolar disorder type I and II. Acta PsychiatrScand 2009; 120: 464–473.

Supporting Information

Additional Supporting Information may be found in the onlineversion of this article:

Fig. S1. Meta-analysis of suicide attempts based on subtype ofbipolar disorder (type I or II). CI = confidence interval.Fig. S2. Meta-analysis of suicide attempts in bipolar disorderbased on the presence of a history of psychosis. CI = confi-dence interval.Fig. S3. Meta-analyses of suicide attempts in bipolar disorderbased on the presence of cannabis use. CI = confidence inter-val.Fig. S4. Meta-analysis of suicide deaths in bipolar disorderbased on the presence of a history of psychosis. CI = confi-dence interval.Fig. S5. Meta-analysis of suicide deaths in bipolar disorderbased on the presence of any substance use disorder. CI = con-fidence interval.

16

Schaffer et al.