how suicidal bipolar patients are depends on how suicidal ideation is defined

Journal of Affective Disorders 118 (2009) 48–54

Contents lists available at ScienceDirect

Journal of Affective Disorders

j ourna l homepage: www.e lsev ie r.com/ locate / j ad

Research report

How suicidal bipolar patients are depends on how suicidal ideationis defined

Hanna M. Valtonen a,b, Kirsi Suominen a,b, Petteri Sokero a, Outi Mantere a,b, Petri Arvilommi a,b,Sami Leppämäki a,c, Erkki T. Isometsä a,c,⁎a Department of Mental Health and Alcohol Research, National Public Health Institute, Helsinki, Finlandb Department of Psychiatry, Jorvi Hospital, Helsinki University Central Hospital, Espoo, Finlandc Department of Psychiatry, Helsinki University Central Hospital, Helsinki, Finland

a r t i c l e i n f o

⁎ Corresponding author. Institute of Clinical MedPsychiatry, P.O. Box 22, 00014 University of Helsinki,471 63728; fax: +358 9 47163735.

E-mail address: [email protected] (E.T. Isomet

0165-0327/$ – see front matter © 2009 Elsevier B.V.doi:10.1016/j.jad.2009.02.008

a b s t r a c t

Article history:Received 21 January 2009Received in revised form 12 February 2009Accepted 12 February 2009Available online 17 March 2009

Background: Suicidal ideation indicates risk for suicidal acts. How different definitions andmeasures for suicidal ideation influence its prevalence, correlates and predictive validity amongbipolar disorder (BD) patients is unknown.Methods: Among the 191 BD patients in the Jorvi Bipolar Study (JoBS), suicidal ideation atbaseline was measured using the Scale for Suicidal Ideation (SSI), Hamilton Depression Scale(HAM-D) item 3 and Beck Depression Inventory (BDI) item 9 and by asking whether patientshad seriously considered suicide. The predictive value of different definitions of ideation onsuicide attempts during a six-month follow-up was investigated.Results: Altogether 74% of patients had suicidal ideation as defined in at least one of the above-mentioned ways, but only 29% met the criteria for all ways; agreement between definitionsranged from low to moderate (kappa coefficient 0.15 to 0.70). The correlates of suicidal ideationoverlapped, but were not identical. Of the measures investigated, a baseline SSI score ≥8 hadthe best combination of sensitivity (0.81) and specificity (0.69) and a positive predictive value(PPV) of 32% for an attempted suicide during follow-up.Limitations: All plausible measures for suicidal ideation could not be investigated.Conclusions:Who is classified as having suicidal ideation depends strongly on the definition andmeans of measurement among BD patients. Different measures for ideation have the potentialto cause inconsistency when correlates of suicidal ideation are investigated. For clinicallypredicting suicide attempts during the next few months, an SSI score ≥8 may best combinesensitivity and specificity.

© 2009 Elsevier B.V. All rights reserved.

Keywords:Bipolar disorderSuicidal ideationSuicide attemptMixed stateDepressive phase

1. Introduction

Suicidal ideation is highlyprevalent inbipolardisorder (BD).Over three-fourths of patients with BD reported lifetimesuicidal ideation (Valtonen et al., 2005). Suicidal ideationappears to be an important marker for identifying patients atrisk for suicide (Brownet al., 2000) and suicide attempts (Mannet al., 1999). In a 20-year prospective study of risk factors for

icine, Department oFinland. Tel.: +358 9

sä).

All rights reserved.

f

suicide in psychiatric outpatients, Brown et al. (2000) foundthat patients who scored three ormore on the Scale for SuicidalIdeation (SSI) were approximately seven times more likely tocommit suicide than patients who scored less than three. Theprevalence of non-fatal suicidal behaviour and the incidence ofsuicide attempts varymarkedly between different phases of BD(Valtonen et al., 2007, 2008).

Assessmentof risk for suicidal behaviour is a challenge for allclinicians taking care of mood disorder patients, especially BDpatients. Duration, frequency, depth, intensity and persistenceof suicidal thoughts should ideally be taken into account whenassessing risk of suicidal behaviour. The assessment of suicidalbehaviour is also a challenging task for researchers and

mailto:[email protected]

http://dx.doi.org/10.1016/j.jad.2009.02.008

http://www.sciencedirect.com/science/journal/01650327

Table 1Socio-demographic characteristics of 191 bipolar disorder patients in theJorvi Bipolar Study.

Characteristics N (%) Bipolar I Bipolar II Total

N=90(47%)

N=101(53%)

191 (100)

Age at entry, yearsMean±SD 39.5 (12.1) 36.0 (12.1) 37.7 (12.2)

SexMale 50 (55.6) 40 (39.6) 90 (47.1)Female 40 (44.4) 61 (60.4) 101 (52.9)

Marital statusNot cohabiting 25 (27.8) 36 (35.6) 61 (31.9)Married or cohabiting 38 (42.5) 43 (42.5) 81 (42.4)Divorced 25 (27.8) 21 (20.8) 46 (24.1)Widowed 2 (2.2) 1 (1.0) 3 (1.6)

Work statusEmployed 43 (47.8) 61 (60.4) 104 (54.5)Student 7 (7.8) 12 (11.9) 19 (9.9)Unemployed 11 (12.2) 13 (12.9) 24 (12.6)Disability pension 29 (32.2) 15 (14.9) 44 (23.0)

Early age at onset (before 18 years age) 29 (32.2) 29 (28.7) 58 (30.4)Number of past episodesMean±SD 11.3±29.2 7.1±8.4 9.1±21.1

Duration of illnessMean±SD 15.4±10.7 12.7±9.9 14.0±10.3

49H.M. Valtonen et al. / Journal of Affective Disorders 118 (2009) 48–54

regulatory bodies. Inconsistent labelling of potential suicidalevents was identified a significant threat to accurate risk-assessment analyses in the Food and Drug Administration's(FDA) paediatric suicidal risk analysis of antidepressants.Standardizing the evaluation of suicidal behaviour by usingthe Columbia Classification Algorithm of Suicide Assessment(C-CASA) was found to be useful and necessary (Posner et al.,2007). The extent to which different ways of defining andmeasuring suicidal ideation among adult mood disorderpatients influence the prevalence, risk factors and predictivevalidity of suicidal ideation has seldom been investigated.

Previous studies of patients with BD have assessed suicidalideation in several ways; by using SSI (Oquendo et al., 2000,2004; Galfalvy et al., 2006; Grunebaum et al., 2006), by self-rated Beck Scale for Suicidal Ideation (BSI) (Simon et al.,2007), by Brief Psychiatric Rating Scale (Khalsa et al., 2008) orby Affective Disorders Evaluation (Marangell et al., 2006). Toour knowledge, no previous clinical studies of representativepatient samples have investigated how the different assess-ments of suicidal ideation affect the estimated prevalence andrisk factors of suicidal ideation among BD patients. Ourhypothesis was that different ways of defining and measuringsuicidal ideation markedly influence which BD patients areclassified as suicidal.We also explored the extent towhich thecorrelates for suicidal ideation differ as a consequence ofdifferent definitions, and investigated the predictive value ofdifferent measures of suicidal ideation for suicide attemptsduring a six-month follow-up.

2. Methods

The background and methodology of the Jorvi BipolarStudy (JoBS) have been described in detail elsewhere(Mantere et al., 2004). In brief, JoBS is a collaborative researchproject between the Department of Mental Health andAlcohol Research of the National Public Health Institute,Helsinki, and the Department of Psychiatry, Jorvi Hospital,Helsinki University Central Hospital (HUCH), Espoo, Finland.The Department of Psychiatry at Jorvi Hospital providessecondary care psychiatric services to all residents of Espoo,Kauniainen and Kirkkonummi (261116 inhabitants in 2002).The Ethics Committee of HUCH approved the study protocol.

All in- and outpatients who currently had a possible new(DSM-IV) BD episode in the catchment area of Jorvi Hospital(n=1630) were screened by using the Mood DisorderQuestionnaire (MDQ) (Hirschfeld et al., 2000) during thestudy period 1.1.2002–28.2.2003. After a positive MDQ screenor suspicion of BD (n=546), the patient was fully informedabout the study protocol and written informed consent wasrequested. The diagnosis of BD was made using the StructuredClinical Interview for DSM-IV Disorders, researcher versionwith psychotic screen (SCID-I/P) (First et al., 2001) withexcellent interrater reliability (κ=1.0 for both BD I and II).The Structured Clinical Interview for DSM-IV personalitydisorders (SCID-II) (First et al., 1996) was used to assessdiagnoses on Axis II. The cohort baseline measurementsincluded the 17-item Hamilton Depression Scale (HAM-D)(Hamilton, 1960), the Scale for Suicidal Ideation (SSI) (Becket al.,1979), the 21-itemBeckDepression Inventory (BDI) (Becket al., 1961), the Beck Hopelessness Scale (BHS) (Beck et al.,1974), the Beck Anxiety Inventory (BAI) (Beck et al., 1988) and

the Social and Occupational Functioning Assessment Scale ofDSM-IV (SOFAS) (American Psychiatric Association, 2000).Depressive mixed state was defined according to Benazzi andAkiskal (2001) as three or more simultaneous intra-episodehypomanic symptoms present for at least 50% of time during amajor depressive episode. The final study group consisted of191DSM-IVBD I and II patientswitha currentepisode(Mantereet al., 2004). Sociodemographic characteristics of the 191 BDpatients in the Jorvi Bipolar Study are presented in Table 1.

2.1. Measurement and alternative definitions for suicidal ideation

Suicidal ideationwas evaluated in several differentways. First,patients were asked whether they had ever seriously consideredsuicide during the current BD episode (Valtonen et al., 2005).Second, Scale for Suicidal Ideation (SSI) was used. SSI is a 19-itemobserver scale designed to quantify the intensity of currentconscious suicide ideation in various dimensions of self-destruc-tive thoughts orwishes: the extent of thewish todie, thedesire tomake an actual suicide attempt, and details of any plans; alsointernal deterrents to an active attempt, and subjective feelings ofcontrol and/or intent regardingaproposedattempt. Eachof the19items consists of three alternative statements graded in intensityfrom 0 to 2, with the maximum total score being 38. Third, BeckDepression Inventory (BDI) item 9 was used. This item has thefollowing alternative statements: 0= I do not have any thoughtsof killingmyself,1= I have thoughts of killingmyself, but I wouldnot carry themout, 2= Iwould like tokillmyself and3=Iwouldkill myself if I had the chance. Fourth, Hamilton Depression Scale(HAM-D) item 3 (Suicide) was used. This item has the followingalternative statements: 0 = absent, 1 = feels life is not worthliving, 2=wisheshewere dead or any thoughts of possible deathto self, 3 = suicide ideas or gestures and 4= attempts at suicide(any serious attempt rates 4). Internal consistency of themeasurements was SSI, Cronbach's alpha 0.84; BDI, Cronbach'salpha 0.89; and HAM-D, Cronbach's alpha 0.78.

Table 2Estimated prevalences of suicidal ideation during different phases of bipolar disorder at baseline.

Consideredsuicide

BDI a BDI a HAM-D b HAM-D b SSI c

sumscore≥6

Suicidalideation asdefined by oneway d

Suicidalideation asdefined by allways e

Item 9≥1 Item 9≥2 Item 3≥2 Item 3≥3

During all phases (N=191) 105 (55%) 122 (64%) 18 (9%) 77 (40%) 45 (24%) 81 (42%) 142 (74%) 55 (29%)During major depressive phase (N=106) 66 (62%) 77 (73%) 12 (11%) 54 (51%) 33 (31%) 50 (47%) 87 (82%) 36 (34%)During depressive mixed or mixed phases (N=41) 30 (73%) 32 (78%) 6 (15%) 18 (44%) 10 (24%) 25 (61%) 36 (88%) 16 (39%)During manic/hypomanic phase (N=44) 9 (21%) 13 (30%) 0 5 (11%) 2 (5%) 6 (14%) 19 (43%) 3 (7%)

a 1% missing information on the Beck Depression Inventory (BDI).b 0.5% missing information on the Hamilton Depression Rating Scale (HAM-D).c Suicidal Ideation Scale.d Suicidal ideation as defined by one way: considered seriously suicide or BDI item 9≥1 or HAM-D item 3≥2 or SSI≥6.e Suicidal ideation as defined by all ways: considered seriously suicide and BDI item 9≥1 and HAM-D item 3≥2 and SSI≥6.

Table 3Agreement between different measurements of suicidal ideation.

Kappacoefficient

BDI BDI HAM-D HAM-D Consideredsuicide


SSI≥6 0.46 0.15 0.70 0.50 0.53BDI 0.41 0.24 0.46item 9≥1

BDI 0.19 0.25 0.16item 9≥2

HAM-D 0.51item 3≥2

HAM-D 0.41item 3≥3

50 H.M. Valtonen et al. / Journal of Affective Disorders 118 (2009) 48–54

2.2. Suicide attempts during follow-up

Information on suicide attempts was obtained for 176/191patients (92%) at the six-month follow-up. During this follow-up15% of patients (27/176) had attempted suicide.A suicide attemptwas defined as self-injurious behaviourwith a non-fatal outcomeaccompanied by evidence (either explicit or implicit) that theperson intended to die (American Psychiatric Association, 2003);self-harm with no suicidal intention was excluded (Valtonenet al., 2006).

2.3. Statistical methods

Spearman's rank correlation was computed to assess therelationship between the SSI, BDI and HAM-D scores.Sensitivity, specificity and positive and negative predictivevalues for different suicidal ideation at baseline for anattempted suicide at the six-month follow-up were calcu-lated. Kappa coefficient was computed to assess the agree-ment between different definitions of suicidal ideation.Binary logistic regression models were created for investigat-ing cross-sectional correlates of different definitions ofsuicidal ideation. The predetermined independent variablescomprised sex, age, comorbid personality disorder, comorbidsubstance dependence or abuse, hopelessness (measured byBHS), severity of anxiety (measured by BAI), severity ofdepression (measured by HAM-D or BDI) and level of Socialand Occupational Functioning (measured by SOFAS), whereasdefinition of different suicidal ideation was the dependentvariable. To avoid circularity, we omitted the suicidality itemsof the depression rating scales. SPSS software, version 15.0,was used.

3. Results

3.1. Estimated prevalences of suicidal ideation

Of the 191 patients, 74% were suicidal ideators measuredby at least one of the alternative ways, i.e. had consideredseriously suicide or had BDI item 9≥1 or HAM-D item 3≥2 orSSI (sum score) ≥6. However, only 29% of the patients metthe criteria for suicidal ideation of all definitions (Table 2).Similar differences were found during the different phases of

BD. Furthermore, using a different cut-off as a threshold forideation in BDI item 9 markedly affected the prevalence ofpatients identified as having suicidal ideation; 64% of patientswere suicidal ideators when suicidal ideation was definedaccording to BDI item 9 as ≥1, whereas only 9% were suicidalwhen BDI item 9 was set at ≥2.

3.2. Correlations between different measures of suicidal ideation

Level of suicidal ideation asmeasured by SSI correlatedwithsubjective rating of depression (BDI) (rs=0.53, pb0.001) andobjective rating of depression (HAM-D) (rs=0.46, pb0.001).Level of suicidal ideation in the SSI also correlated moderatelywith scores of BDI item 9 (rs=0.58, pb0.001) and HAM-Ditem 3 (rs=0.67, pb0.001). Furthermore, scores of BDI item 9correlated moderately with scores of HAM-D item 3 (rs=0.53,pb0.001).

3.3. Agreement between different measurements of suicidalideation

Kappa coefficient was computed to assess the agreementbetween different definitions of suicidal ideation (Table 3). Thebest agreement was found between SSI≥6 and HAM-D item3≥2 (κ=0.70), whereas the lowest agreement was foundbetween SSI≥6 and BDI item 9≥2 (κ=0.15). Agreement was,in general, low to modest.

Table 4The median distribution of Suicidal Ideation Scale (SSI) for different suicidal ideation during different bipolar disorder phases at baseline.

Consideredsuicide

BDI a BDI a HAM-D b HAM-D b Suicidalideation asdefined byone way

Suicidalideators asdefined byall ways


During all phases (N=191) 12.0 (0–36) 9.0 (0–36) 18.0 (0–36) 14.0 (0–36) 15.0 (0–36) 8.0 (0–36) 15.0 (7–36)During major depressive phase (N=106) 12.5 (0–36) 9.0 (0–36) 20.0 (0–36) 13.5 (0–36) 16.0 (0–36) 8.0 (0–36) 16.0 (7–36)During depressive mixed or mixed phases (N=41) 12.0 (0–28) 12.0 (0–28) 16.0 (0–28) 14.0 (0–28) 15.5 (8–28) 11.5 (0–28) 14.0 (7–28)During manic/hypomanic phase (N=44) 0 (0–28) 5.0 (0–28) 0 9.0 (8–15) 12.0 (9–15) 0 (0–28) 12.0 (9–15)

a 1% missing information on the Beck Depression Inventory (BDI).b 0.5% missing information on the Hamilton Depression Rating Scale (HAM-D).


3.4. Distribution of Suicidal Ideation Scale (SSI)

Marked differences existed regarding the median scores ofSSI based on different definitions of suicidal ideation (Table 4).The median score of SSI was 18.0 when suicidal ideatorswas defined according to BDI item 9 as ≥2, whereas themedian score of SSI was 9.0 when suicidal ideators was definedaccording to BDI item 9 as ≥1. The same phenomenon wasfound during different phases of BD.

3.5. Cross-sectional correlates of suicidal ideation

In all logistic regression models, severity of depression(measured by HAM-D or BDI) was an independent cross-sectional correlates of suicidal ideation (Table 5). Hopelessness,severity of anxiety and age were independent correlates ofsuicidal ideation inmany, but not allmodels. SOFAS level, sex andcomorbid personality disorder were independent correlates ofsuicidal ideation in some of the models (Table 5).

3.6. Predictive values of different measures of suicidal ideationat baseline for an attempted suicide at the six-month follow-up

BDI item 9≥2 had the best positive predictive value (0.53)for a suicide attempt during the next six months (Table 6).

Table 5Binary logistic regression models investigating cross-sectional correlates of suicidal

Dependent variable SSI a≥6 SSI≥6 Consideresuicide

OR (p) OR (p) OR (p)

HAM-D or BDI HAM-D BDI HAM-DSex (male/female)Age 0.95 (0.001) 0.95 (0.004) 0.97 (0.04HAM-D b score 1.09 (0.007) 1.08 (0.01BDI c score 1.08 (0.001)BHS d score 1.23 (b0.001) 1.19 (b0.001) 1.11 (0.00BAI e scoreSOFAS f score 0.96 (0.03) 0.96 (0.04)Comorbid personality disorderComorbid substance dependence/abuse

a Scale for Suicidal Ideation.b 0.5% missing information on the Hamilton Depression Rating Scale (HAM-D); ic 1% missing information on the Beck Depression Inventory (BDI); items 2 and 9d Beck Hopelessness Scale.e Beck Anxiety Inventory.f Social and Occupational Functioning.

However, its sensitivity was quite low (33%). In other words,53% (9/17) of those who scored ≥2 on BDI item 9 at baselineattempted suicide during the six-month follow-up. Of thealternative definitions investigated, an SSI score of ≥8provided an optimal combination of sensitivity (0.82) andspecificity (0.69). It had the second best positive predictivevalue (0.32). In other words, 32% (22/69) of those who scored≥8 on the SSI at baseline attempted suicide during the six-month follow-up.

4. Discussion

4.1. Main findings

Prevalence of suicidal ideation in BD depends markedly onthe means of measuring suicidal ideation. Nearly three-fourthsof the patients were suicidal ideators as defined by one of theways investigated,whereas onlyaboutone-fourthwere suicidalbased on the criteria for all ways. Almost half of the patients(46% (87/191))were classified as suicidal ideatorsmeasured bysomebutnot allways. The correlates of differentmeasurementsof suicidal ideation overlapped, but were not identical; thesingle common independent risk factor of suicidal ideationwasseverity of depression. Current phase of illness does not seem tomarkedly influence these findings.

ideation.

d Consideredsuicide

BDI BDI HAM-D HAM-D


OR (p) OR (p) OR (p) OR (p) OR (p)

BDI HAM-D HAM-D BDI BDI0.17 (0.01)

) 0.96 (0.02) 0.97 (0.02)) 1.07 (0.02) 1.14 (0.01)

1.10 (b0.001) 1.11 (b0.001) 1.05 (0.02)4) 1.26 (b0.001) 1.19 (0.008)

1.08 (b0.001)

3.88 (0.002)

tem 3 omitted.omitted.

Table 6Predictive values of different suicidal ideation at baseline for an attempted suicide (27/176 patients) at the six-month follow-up.

Consideredsuicide

BDI a BDI a HAM-D b HAM-D b SSIsumscore ≥2

SSIsumscore ≥6

SSIsumscore ≥8


Positive predictive value 0.25 0.23 0.53 0.24 0.30 0.25 0.29 0.32Negative predictive value 0.96 0.98 0.89 0.90 0.89 0.95 0.95 0.95Sensitivity 0.89 0.96 0.33 0.63 0.48 0.85 0.81 0.81Specificity 0.52 0.41 0.95 0.64 0.80 0.54 0.64 0.69

a 1% missing information on the Beck Depression Inventory (BDI).b 0.5% missing information on the Hamilton Depression Rating Scale (HAM-D).


4.2. Strengths and limitations

To our knowledge, no previous clinical cohort studiesamong BD patients have investigated how the differentassessments of suicidal ideation affect the estimated pre-valence and correlates of suicidal ideation, or have investi-gated the predictive values of different assessments ofsuicidal ideation for a preceding suicide attempt. Our studyhas some major methodological strengths. It comprises arelatively large (N=191) and unselected sample of both in-and outpatients with BD I and II who have a current acute BDepisode. The patients were carefully diagnosed using struc-tured interviews with excellent reliability for diagnosing BD Iand II (κ=1.0 for both). Because suicidal behaviour was oneof the predetermined main foci of the JoBS, suicidal ideationwas assessed comprehensively both at baseline and at the six-month follow-up. The internal consistency of the instrumentsused was either good or excellent (SSI Cronbach's alpha 0.84,BDI Cronbach's alpha 0.89, HAM-D Cronbach's alpha 0.78).Nevertheless, some methodological limitations should benoted. First, although SSI evaluates various dimensions of self-destructive thoughts, no suicide assessment scale has beendefined as the gold standard. However, in the absence of anestablished gold standard, we used SSI to compare intensity ofsuicidal ideation. Second, we did not evaluate the differentdimensions (e.g. active suicidal desire, preparations andpassive suicidal desire) of SSI. Third, our findings concerningBD patientsmay not be generalized to other diagnostic groups(e.g. major depressive disorder patients). We conducted ourstudy among BD patients, who according to Chen and Dilsaver(1996) may be at higher risk of suicidal behaviour than thosesuffering from other Axis I disorders. Fourth, we measuredcurrent suicidal ideation, thus we did not use SSI-W, whichmeasures suicidal ideation at its worst point (Beck et al.,1999). Fifth, we could not investigate all possible measures ofsuicidal ideation (e.g. Brief Psychiatric Rating Scale, AffectiveDisorders Evaluation) used in previous studies of BD patients.Hintikka et al. (1998, 2001) investigated suicidal ideation inthe Finnish general population. In their study (BDI item 9),the response “I don't have any thoughts of harming myself”was selected to indicate the absence of suicidal ideation.Suicidality has often been assessed by HAM-D item 3 inrandomized, placebo-controlled, antidepressant clinical trials.

4.3. Time interval of suicidal ideation

The different measurements of suicidal ideation measuredifferent time intervals. SSI estimates suicidal thoughts

during the past week, HAM-D item 3 during the past threedays, and BDI item 9 at the moment. Furthermore, patientswere asked whether they had ever seriously consideredsuicide during the current BD episode. In theory, the questionof whether a patient had ever seriously considered suicideduring the current episode should have greater estimatedprevalence of suicidal ideation than suicidal ideation at themoment. However, in our study, the cut-off set for BDI item 9had a greater impact on the estimated prevalence of suicidalideation than different time intervals. The longitudinal courseof BD is chronic and dominated by depressive symptoms(Judd et al., 2002, 2003, 2008; Mantere et al., 2008). Thus,among BD patients the different time interval only partlyexplains the differences between the estimated prevalencesof different definitions for suicidal ideation.

4.4. Measurement of suicidal ideation

Beck et al. (1979) reported three dimensions of suicideideation: active suicidal desire, preparations and passivesuicidal desire, in their original article about SSI. Holi et al.(2005) described the same dimensions among depressedadolescents. The median scores of SSI of different measure-ments of suicidal ideation varied markedly probably becauseHAM-D item 3 and BDI item 9 do not assess suicidal ideationas thoroughly as SSI. The broad coverage of aspects of suicidalideation in multi-item suicidality assessment in the SSIseemed to increase the estimated prevalence of ideationless than low threshold of a single item, especially BDI item 9.The threshold of BDI item 9 was important regarding theestimated prevalence and predictive values of suicidalideation. The BDI item 9 responses of” I would like to killmyself” and “I would kill myself if I had the chance” indicateactive suicidal ideation. The response “I have thoughts ofkilling myself, but I would not carry them out” indicate adeath wish and hopelessness, which also belong to the widespectrum of suicidal ideation (Diekstra and Garnefski, 1995).We found that at least among BD patients the threshold in theHAM-D suicidality itemwas less important than the thresholdin the BDI item. The HAM-D item 3 response of “wishes hewere dead or any thoughts of possible death to self” indicatespassive suicidal ideation, whereas the response “suicide ideasor gestures” indicates active suicidal ideation. The differencebetween the estimated prevalence of suicidal ideation basedon different definition of suicidal ideation may be explainedby different aspects and intensity of suicidal thoughtsexplored by these measurements. We recommend use of SSIfor future studies evaluating suicidal ideation.


4.5. Cross-sectional correlates of suicidal ideation

Different measures of suicidal ideation affect whichpatients are considered suicidal and may also influence theassociated risk factors. In all logistic regression models,severity of depression (measured by HAM-D or BDI) was anindependent cross-sectional correlates of suicidal ideation.Previous studies have found suicidal ideation to be associatedwith any comorbidity (Axis I and II) (Vieta et al., 2000) orwithAxis II comorbidity among BD II patients (Vieta et al., 1999).Likewise, suicidal ideation is related to any Axis I comorbidityamong BD I patients (Vieta et al., 2001). Simon et al. (2007)reported that especially comorbid anxiety disorders wereassociated with both suicidal ideation and suicide attemptsamong BD outpatients. In our cross-sectional study, weestablished that severity of depressive episode and hope-lessness were independent risk factors for suicidal ideation(Valtonen et al., 2005). From a methodological point of view,different ways of measuring suicidal ideation may be a sourceof inconsistency in associated risk factors.

4.6. Predictive values of different measures of suicidal ideationat baseline for an attempted suicide

It is essential for clinicians who treat acutely ill BD patientsto identify which patients are at the greatest risk for a suicideattempt. In a 20-year prospective studyof risk factors for suicidein psychiatric outpatients, Brown et al. (2000) found out thatpatients who scored ≥3 on the SSI were approximately seventimes more likely to commit suicide than patients who scoredb3. By contrast, Beck et al. (1985) reported no significantlyhigher SSI scores among those inpatients who committedsuicide comparedwith thosewhodidnot. To ourknowledge, noprevious clinical cohort studies exist that have investigated thepredictive values of different definitions of suicidal ideation fora suicide attempt among BD patients. Only a minority ofindividuals with suicidal thoughts will actually attempt suicide.In our cohort, 25% (26/105) of patients who had suicidalideation (measured as either SSI≥6orhad seriously consideredsuicide) at baseline attempted suicide during the six-monthfollow-up. However, 53% of thosewho scored≥2 on BDI item 9at baseline attempted suicide during the following six months.Furthermore, 32% of thosewho scored≥8 on the SSI at baselineattempted suicide during the six-month follow-up. In ourearlier study, we also found that a score of eight on the SSIappeared to be an optimal cut-off point for predicting suicideattempts during the follow-up (Valtonen et al., 2006). As wehave already reported (Valtonen et al., 2006), the SSI may beconsidered for routine use among BD patient groups atparticularly high risk for suicidal behaviour.

4.7. Suicidal ideation during different illness phases

Suicidal behaviour varies markedly between differentphases of BD (Valtonen et al., 2007). Thus, it is important toinvestigate whether different measures of suicidal ideationduring different BD phases have an effect on the estimatedprevalence of suicidal ideation. Who was classified as asuicidal ideator depended on the measurement of suicidalideation among BD patients also during different illnessphases. The highest estimated prevalence of suicidal ideation

was related to phases associated with depressive aspects ofthe illness. However, the influence of different alternativedefinitions and measures of suicidal ideations appearedconsistent across the phases.

5. Conclusions

Who is classified as having suicidal ideation dependsstrongly on the definition and means of measurement ofsuicidal ideation applied among BD patients. Different ways ofdefining ideation have the potential to cause inconsistencywhen correlates of suicidal ideation are investigated. Forclinically predicting suicide attempts during the next fewmonths, an SSI score of ≥8 may best combine sensitivity andspecificity.

Role of funding sourceFunding for this study was provided by Helsinki University Central

Hospital and the Research Foundation of Orion Corporation. HelsinkiUniversity Central Hospital and the Research Foundation of Orion Corpora-tion had no role in study design; in the collection, analysis and interpretationof data; in the writing of the report; or in the decision to submit the paper forpublication.

Conflict of interestNone.

References

American Psychiatric Association, 2000. Diagnostic and Statistical Manual ofMental Disorders, Fourth Edition. American Psychiatric Association,Washington, DC, pp. 817–818. Text Revision.

American Psychiatric Association, 2003. Practice guideline for the assess-ment and treatment of patients with suicidal behaviors. Am. J. Psychiatry160 November Suppl.

Beck, A.T., Ward, C.H., Mendelson, M., Mock, J., Erbaugh, J., 1961. An inventoryfor measuring depression. Arch. Gen. Psychiatry 4, 561–571.

Beck, A.T., Weissman, A., Lester, D., Trexler, L., 1974. The measure ofpessimism: the hopelessness scale. J. Consult. Clin. Psychol. 42, 861–865.

Beck, A.T., Kovacs, M., Weissman, A., 1979. Assessment of suicidal intention:the scale for suicide ideation. J. Consult. Clin. Psychol. 47, 343–352.

Beck, A.T., Steer, R.A., Kovacs, M., Garrison, B., 1985. Hopelessness andeventual suicide: a 10-year prospective study of patients hospitalizedwith suicidal ideation. Am. J. Psychiatry 142, 559–563.

Beck, A.T., Epstein, N., Brown, G., Steer, R.A., 1988. An inventory for measuringclinical anxiety: psychometric properties. J. Consult. Clin. Psychol. 56, 893–897.

Beck, A.T., Brown, G.K., Steer, R.A., Dahlsgaar, K.K., Grisham, J.R., 1999. Suicidelife threat. Behav. 29, 1–9.

Benazzi, F., Akiskal, H.S., 2001. Delineating bipolar II mixed states in theRavenna–San Diego collaborative study: the relative prevalence anddiagnostic significance of hypomanic features during major depressiveepisodes. J. Affect. Disord. 67, 115–122.

Brown, G.K., Beck, A.T., Steer, R.A., Grisham, J.R., 2000. Risk factors for suicidein psychiatric year outpatients: a 20-prospective study. J. Consult. Clin.Psychol. 68, 371–377.

Chen, Y.W., Dilsaver, S.C., 1996. Lifetime rates of suicide attempts amongsubjects with bipolar and unipolar disorders relative to subjects withother Axis I disorders. Biol. Psychiatry 39, 896–899.

Diekstra, R.F., Garnefski, N., 1995. On the nature, magnitude, and causality ofsuicidal behaviors: an international perspective. Suicide Life-Threat.Behav. 25, 36–57.

First, M.B., Gibbon, M., Spitzer, R.L., Williams, J.B.W., 1996. Structured ClinicalInterview for DSM-IV Axis II Disorders (SCID-II), Version 2. New YorkPsychiatric Institute, Biometrics Research, New York.

First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W., 2001. Structured ClinicalInterview for DSM-IV-TR Axis I Disorders, Research Version, PatientEdition With Psychotic Screen. Biometrics Research, New York StatePsychiatric Institute. (SCID-I/P W/PSY.SCREEN) New York.

Galfalvy, H., Oquendo, M.A., Carballo, J.J., Sher, L., Grunebaum, M.F., Burke, A.,Mann, J.J., 2006. Clinical predictors of suicidal acts after major depressionin bipolar disorder: a prospective study. Bipolar Disord. 8, 586–595.


Grunebaum, M.F., Ramsay, S.R., Galfalvy, H., Ellis, S.P., Burke, A.K., Sher, L.,Printz, D.J., Khan, D.A., Mann, J.J., 2006. Correlates of suicide attempthistory in bipolar disorder: a stress-diathesis perspective. Bipolar Disord.8, 551–557.

Hamilton, M., 1960. A rating scale for depression. J. Neurol. Neurosurg.Psychiatry 23, 56–62.

Hintikka, J., Viinamäki, H., Tanskanen, A., Kontula, O., Koskela, K., 1998.Suicidal ideation and parasuicide in the Finnish general population. ActaPsychiatr. Scand. 98, 23–27.

Hintikka, J., Pesonen, T., Saarinen, P., Tanskanen, A., Lehtonen, J., Viinamäki, H.,2001. Suicidal ideation in the Finnish general population. A 12-monthfollow-up study. Soc. Psychiatry Psychiatr. Epidemiol. 36, 590–594.

Hirschfeld, R.M.A., Williams, J.B.W., Spitzer, R.L., Calabrese, J.R., Flynn, L., KeckJr, P.E., Lewis, L., McElroy, S.L., Post, R.M., Rapport, D.J., Russell, J.M., Sachs,G.S., Zajecka, J., 2000. Development and validation of a screeninginstrument for bipolar spectrum disorder: the Mood Disorder Ques-tionnaire. Am. J. Psychiatry 157, 1873–1875.

Holi, M.M., Pelkonen, M., Karlsson, L., Kiviruusu, O., Ruuttu, T., Heilä, H.,Tuisku, V., Marttunen, M., 2005. Psychometric properties and clinicalutility of the Scale for Suicidal Ideation (SSI) in adolescents. BMCPsychiatry. 5, 8.

Judd, L.L., Akiskal, H.S., Schettler, P.J., Endicott, J., Maser, J.D., Solomon, D.A., Leon,A.C., Rice, J.A., Keller,M.B., 2002. The long-termnatural historyof theweeklysymptomatic status of bipolar I disorder. Arch. Gen. Psychiatry 59, 530–537.

Judd, L.L., Akiskal, H.S., Schettler, P.J., Coryell, W., Endicott, J., Maser, J.D.,Solomon, D.A., Leon, A.C., Keller, M.B., 2003. A prospective investigationof the natural history of the long-term weekly symptomatic status ofbipolar II disorder. Arch. Gen. Psychiatry 60, 261–269.

Judd, L.L., Schettler, P.J., Akiskal, H.S., Coryell, W., Leon, A.C., Maser, J.D.,Solomon, D.A., 2008. Residual symptom recovery from major affectiveepisodes in bipolar disorders and rapid episode relapse/recurrence. Arch.Gen. Psychiatry 65, 386–394.

Khalsa, H.M.K., Salvatore, P., Hennen, J., Baethge, C., Tohen, M., Baldessarini, R.J.,2008. Suicidal events and accidents in 216 first-episode bipolar I disorderpatients: predictive factors. J. Affect. Disord. 106, 179–184.

Mann, J.J., Waternaux, C., Haas, G.L., Malone, K.M., 1999. Toward a clinicalmodel of suicidal behaviour in psychiatric patients. Am. J. Psychiatry 156,181–189.

Mantere, O., Suominen, K., Leppämäki, S., Valtonen, H., Arvilommi, P.,Isometsä, E., 2004. The clinical characteristics of DSM-IV bipolar I andII disorders baseline findings from the Jorvi Bipolar Study (JoBS). BipolarDisord. 6, 395–405.

Mantere, O., Suominen, K., Valtonen, H., Arvilommi, P., Leppämäki, S.,Isometsä, E., 2008. Differences in outcome of DSM-IV bipolar I and IIdisorders. Bipolar Disord. 10, 413–425.

Marangell, L.B., Bauer, M.S., Dennehy, E.B., Wisniewski, S.R., Allen, M.H.,Miklowitz, D.J., Oquendo, M.A., Frank, E., Perlis, R.H., Martinez, J.M.,Fagiolini, A., Otto, M.W., Chessick, C.A., Zboyan, H.A., Miyahara, S., Sachs,G., Thase,M.E., 2006. Prospective predictors of suicide and suicide attemptsin 1,556 patients with bipolar disorders followed for up to 2 years. BipolarDisord. 8, 566–575.

Oquendo, M.A., Waternaux, C., Brodsky, B., Parsons, B., Haas, G.L., Malone, K.M.,Mann, J.J., 2000. Suicidal behavior in bipolar mood disorder: clinicalcharacteristics of attempters and nonattempters. J. Affect. Disord. 59,107–117.

Oquendo, M.A., Galfalvy, H., Russo, S., Ellis, S.P., Grunebaum, M.F., Burke, A.,Mann, J.J., 2004. Prospective study of clinical predictors of suicidal actsafter a major depressive episode in patients with major depressivedisorder or bipolar disorder. Am. J. Psychiatry 161, 1433–1441.

Posner, K., Oquendo, M.A., Gould, M., Stanley, B., Davies, M., 2007. ColumbiaClassification Algorithmof Suicide Assessment (C-CASA): classification ofsuicidal events in the FDA's pediatric suicidal risk analysis of antidepres-sants. Am. J. Psychiatry 164, 1035–1043.

Simon, N.M., Zalta, A.K., Otto, M.W., Ostacher, M.J., Fischmann, D., Chow, C.W.,Thompson, E.H., Stevens, J.C., Demapulos, C.M., Nierenberg, A.A., Pollack,M.H., 2007. The association of comorbid anxiety disorders with suicideattempts and suicidal ideation in outpatients with bipolar disorder. J.Psychiatr. Res. 41, 255–264.

Valtonen, H., Suominen, K., Mantere, O., Leppämäki, S., Arvilommi, P., Isometsä,E., 2005. Suicidal ideation and attempts in bipolar I and II disorders. J. Clin.Psychiatry 66, 1456–1462.

Valtonen, H.M., Suominen, K., Mantere, O., Leppämäki, S., Arvilommi, P.,Isometsä, E.T., 2006. Prospective study of risk factors for attemptedsuicide among patients with bipolar disorder. Bipolar Disord. 8, 576–585.

Valtonen, H.M., Suominen, K., Mantere, O., Leppämäki, S., Arvilommi, P.,Isometsä, E., 2007. Suicidal behaviour during different phases of bipolardisorder. J. Affect. Disord. 97, 101–107.

Valtonen, H.M., Suominen, K., Haukka, J., Mantere, O., Leppämäki, S.,Arvilommi, P., Isometsä, E., 2008. Differences in incidence of suicideattempts during phases of bipolar I and II disorders. Bipolar Disord. 10,588–596.

Vieta, E., Colom, F., Martinez-Aran, A., Benabarre, A., Gasto, C., 1999. Personalitydisorders in bipolar II patients. J. Nerv. Ment. Dis. 187, 245–248.

Vieta, E., Colom, F.,Martinez-Aran, A., Benabarre, A., Reinares,M., Casto, C., 2000.Bipolar II disorder and comorbidity. Compr. Psychiatry 41, 339–343.

Vieta, E., Colom, F., Corbella, B., Martinez-Aran, A., Reinares, M., Benabarre, A.,Casto, C., 2001. Clinical correlates of psychiatric comorbidity in bipolar Ipatients. Bipolar Disord. 3, 253–258.

how suicidal bipolar patients are depends on how suicidal ideation is defined

Documents