Search

Background Patients with bipolar disorder (BD) differ in their relative predominance of types of episodes, yielding predominant polarity, which has important treatment implications. However, few prospective studies of predominant polarity exist. Methods In the Jorvi Bipolar Study (JoBS), a regionally representative cohort of 191 BD I and BD II in- and outpatients was followed for five years using life-chart methodology. Differences between depressive (DP), manic (MP), and no predominant polarity (NP) groups were examined regarding time ill, incidence of suicide attempts, and comorbidity. Results At baseline, 16% of patients had MP, 36% DP, and 48% NP. During the follow-up the MP group spent significantly more time euthymic, less time in major depressive episodes, and more time in manic states than the DP and NP groups. The MP group had significantly lower incidence of suicide attempts than the DP and NP group, lower prevalence of comorbid anxiety disorders but more psychotic symptoms lifetime and more often (hypo)manic first phase of the illness than the DP group. Classification of predominant polarity was influenced by the timeframe used. Limitations The retrospective counting of former phases is vulnerable to recall bias. Assignment of dominant polarity may necessitate a sufficient number of illness phases. Conclusions Predominant polarity has predictive value in predicting group differences in course of illness, but individual patients’ classification may change over time. Patients with manic polarity may represent a more distinct subgroup than the two others regarding illness course, suicide attempts, and psychiatric comorbidity.

Objectives The long-term outcome of bipolar disorder (BD) has been extensively investigated. However, previous studies may be biased towards hospitalized patients with bipolar I disorder (BD-I), and generalizability to the current treatment era remains uncertain. In this naturalistic study, we followed a secondary-care cohort of patients with BD. Methods In the Jorvi Bipolar Study, 191 patients with BD-I and bipolar II disorder (BD-II) were followed using a life-chart method. Interviews were conducted at six months, 18 months, and five years. Time to full remission, time to first recurrence, total time ill, their predictors, and BD-I versus BD-II differences were investigated among the 151 patients remaining in follow-up. Results Nearly all subjects recovered from the index episode, but almost all (90%) had a recurrence, and most had multiple recurrences. The patients spent about one-third of their time in illness episodes and 15% of their time with subthreshold symptoms; half of the time they were euthymic. After controlling for confounders, no difference in time spent in depressive states between patients with BD-I and BD-II persisted. Among patients with a depressive index phase, cluster C personality disorders [hazard ratio (HR) = 0.452, p = 0.040] and higher 17-item Hamilton Depression Scale score (HR = 0.951, p = 0.022) predicted longer time to remission, whereas lifetime psychotic symptoms (HR = 2.162, p = 0.016) predicted shorter time to first recurrence. Conclusions Among patients with BD, chronicity as uninterrupted persistence of illness was rare, but multiple recurrences were the norm. Patients with BD spent only half of their time euthymic. Patients with BD-I and BD-II may differ little in proneness to depressive states. Severity of depression, cluster C personality disorders, and psychotic symptoms predicted outcome.

Whether levels of neuroticism or extraversion differ between patients with bipolar disorder (BD), major depressive disorder (MDD) and subjects from the general population, or between BD I and BD II patients, remains unclear.BD patients (n=191) from the Jorvi Bipolar Study, and MDD patients (n=358) from both the Vantaa Depression Study and the Vantaa Primary Care Depression Study cohorts, were interviewed at baseline and at 18 months. A general population comparison group (n=347) was surveyed by mail. BD patients' neuroticism and extraversion scores, measured by Eysenck Personality Inventory, were compared at an index interview, when the levels of depression and mania were lowest, with scores of MDD patients and general population controls. Comparisons were also made between BD I (n=99) and BD II (n=92) patients.In multinomial logistic regression, BD patients had higher neuroticism (OR=1.17, p0.001) and lower extraversion (OR=0.92, p=0.003) than the general population. When entered simultaneously into the model, the effect of extraversion disappeared. In logistic regression, the levels of neuroticism and extraversion did not differ between BD and MDD patients, or between BD I and II patients.Patients' personality scores were not pre-morbid.Levels of neuroticism and extraversion are unlikely to differ between BD and MDD patients, or between BD I and II patients. The overall level of neuroticism is moderately higher and extraversion somewhat lower in BD patients than in the general population. High neuroticism may be an indicator of vulnerability to both bipolar and unipolar mood disorders.

Objectives: Bipolar disorder (BD) is correctly diagnosed in only 40–50% of patients. No previous study has investigated the characteristics of bipolar patients in psychiatric care with or without clinical diagnoses of BD. We investigated the demographic and clinical predictors of the absence of a clinical diagnosis of BD I and II among psychiatric patients. Methods: In the Jorvi Bipolar Study, 1,630 psychiatric in- and outpatients were screened with the Mood Disorder Questionnaire. Suspected cases were diagnosed with the Structured Clinical Interview for DSM-IV Axis I Disorders-Patient version (SCID-I/P) for BD. Patients with no preceding clinical diagnosis of BD, despite previous manic, hypomanic or mixed phases and treatment in psychiatric care, were classified as undiagnosed. The clinical characteristics of unrecognized BD I patients (23 of 90 BD I patients) and BD II patients (47 of 93 BD II patients) were compared to those of patients who had been correctly diagnosed. Results: No previous hospitalizations [odds ratio (OR) = 10.6, p = 0.001] or psychotic symptoms (OR = 4.4, p = 0.045), and the presence of rapid cycling (OR = 11.6, p = 0.001) predicted lack of BD I diagnosis. No psychotic symptoms (OR = 3.3, p = 0.01), female gender (OR = 3.0, p = 0.03), and shorter time in treatment (OR = 1.1, p = 0.03) predicted the lack of a BD II diagnosis. Conclusions: Correct diagnosis of BD I is related to the severe phases of illness leading to hospitalizations. In BD II, the illness factors may not be as important as time elapsed in treatment, a factor that often leads to a delay in diagnosis or none at all. Excessive reliance on typical and cross-sectional presentations of illness likely explain the non-recognition of BD. The challenge for correctly diagnosing bipolar patients is in outpatient settings.

Background There are no previous studies comparing the prevalence and risk factors for suicidal behaviour during different phases of bipolar disorder. Methods In the Jorvi Bipolar Study (JoBS), 1630 psychiatric in- and outpatients were screened for bipolar disorders with the Mood Disorder Questionnaire. Using SCID I and II interviews, 191 patients were diagnosed with bipolar disorders (90 bipolar I, 101 bipolar II). Suicidal ideation was measured using the Scale for Suicidal Ideation (SSI). Prevalence and risk factors for ideation and attempts during different phases (depressive, mixed, depressive mixed and hypomanic/manic phases) were investigated. Results There were marked differences between phases regarding suicide attempts and level of suicidal ideation. Hopelessness predicted suicidal behaviour during the depressive phase, whereas a subjective rating of severity of depression and younger age predicted suicide attempts during mixed phases. Limitations The relatively small sample size in some phases. Conclusions Suicidal behaviour varied markedly between different phases of BD. Suicide attempts and suicidal ideation were related to phases which are associated with depressive aspects of the illness. Hopelessness and severity of depression were key indicators of risk in all phases.

Objectives Bipolar disorder (BD) is one of the leading causes of disability worldwide. However, vocational ability and predictors of long-term work disability have rarely been studied among patients with BD. We investigated clinical predictors of work disability among patients with BD in psychiatric care. Methods The Jorvi Bipolar Study (JoBS) is a naturalistic prospective cohort study (N = 191) representing adult (18–59 years) psychiatric inpatients and outpatients with DSM-IV bipolar I disorder (BD-I) and bipolar II disorder (BD-II) in three Finnish cities. Within the JoBS, we investigated the prevalence of disability pensions at baseline, and predictors for being granted a disability pension during an 18-month follow-up of the 151 patients in the labor force at baseline. Cox models were used to determine predictors for onset of disability pension. Results At baseline, 21% (40/191) of the patients already had a disability pension. During the follow-up, a further 38 patients (25% of the 151 followed) were granted a new disability pension. The predictors included older age, male gender, depressive index episode, higher number of psychiatric hospitalizations, generalized anxiety disorder, avoidant personality disorder, and depressive burden during follow-up. However, the predictors differed depending on bipolar subtype, age, and gender. Conclusions BD-I and BD-II are associated with a major risk of long-term work disability, the proportion of patients with a disability pension rising to 41% in the medium-term follow-up of the Finnish cohort investigated in the present study. Severe clinical course, depression, comorbidities, age, and gender are likely to be the main predictors but predictors may vary depending on the subgroup.

Objective: There are few prospective studies on risk factors for attempted suicide among representative samples of psychiatric patients with bipolar I and II disorders. We conducted a prospective study to investigate risk for suicide attempts among a secondary-level sample of psychiatric in- and outpatients with bipolar disorder (BD). Methods: In the Jorvi Bipolar Study (JoBS), 1,630 psychiatric in- and outpatients from three Finnish cities were screened for BDs with the Mood Disorder Questionnaire (MDQ). Using the Structured Clinical Interview for DSM-IV Disorders (SCID)-I and -II, 191 patients were diagnosed with BDs (90 bipolar I and 101 bipolar II). Information on suicide attempts during the follow-up was obtained for 176 patients (92%) at the 6-month follow-up and for 160 patients (84%) at the 18-month follow-up. Results: During the 18-month follow-up 20% of patients (35/176) attempted suicide. In a Cox regression model, baseline previous suicide attempts (OR 3.8, 95% CI 1.7–8.8; p = 0.001), hopelessness (OR 1.2, 95% CI 1.1–1.3; p

Background Poor treatment adherence among patients with bipolar disorder (BD) is a common clinical problem. However, whether adherence is mostly determined by patient characteristics or attitudes, type of treatment or treatment side-effects remains poorly known. Methods The Jorvi Bipolar Study (JoBS) is a naturalistic prospective 18-month study representing psychiatric in- and outpatients with DSM-IV BD I and II in three Finnish cities. During the 18-month follow-up we investigated the continuity of, attitudes towards and adherence to various types of psychopharmacological and psychosocial treatments among 168 psychiatric in- and outpatients with BD I or II. Results One-quarter of the patients using mood stabilizers or atypical antipsychotics discontinued medication during at least one treatment phase of the follow-up autonomously, mostly during depression. When pharmacotherapy continued, adherence was compromised in one-third. Rates of non-adherence to mood stabilizers or antipsychotics did not differ, but the predictors did. One-quarter of the patients receiving psychosocial treatments were non-adherent to them. Limitations Serum concentrations were not estimated. Conclusions More than one-half of BD patients either discontinue pharmacotherapy or use it irregularly. Autonomous discontinuation takes place mostly in depression. Although rates of non-adherence do not necessarily differ between mood-stabilizing medications, the predictors for nonadherence do. Moreover, adherence to one medication does not guarantee adherence to another, nor does adherence at one time-point ensure later adherence. Attitudes towards treatments affect adherence to medications as well as to psychosocial treatments and should be repeatedly monitored. Non-adherence to psychosocial treatment should be given more attention.

Objectives Whether risk of suicide attempts (SAs) differs between patients with bipolar disorder (BD) and patients with major depressive disorder (MDD) is unclear. We investigated whether cumulative risk differences are due to dissimilarities in time spent in high-risk states, incidence per unit time in high-risk states, or both. Methods Incidence rates for SAs during various illness phases, based on prospective life charts, were compared between patients from the Jorvi Bipolar Study (n = 176; 18 months) and the Vantaa Depression Study (n = 249; five years). Risk factors and their interactions with diagnosis were investigated with Cox proportional hazards models. Results By 18 months, 19.9% of patients with BD versus 9.5% of patients with MDD had attempted suicide. However, patients with BD spent 4.6% of the time in mixed episodes, and more time in major depressive episodes (MDEs) (35% versus 21%, respectively) and in subthreshold depression (39% versus 31%, respectively) than those with MDD. Compared with full remission, the combined incidence rates of SAs were 5-, 25-, and 65-fold in subthreshold depression, MDEs, and BD mixed states, respectively. Between cohorts, incidence of attempts was not different during comparable symptom states. In Cox models, hazard was elevated during MDEs and subthreshold depression, and among patients with preceding SAs, female patients, those with poor social support, and those aged

BACKGROUND: Suicidal ideation indicates risk for suicidal acts. How different definitions and measures for suicidal ideation influence its prevalence, correlates and predictive validity among bipolar disorder (BD) patients is unknown. METHODS: Among the 191 BD patients in the Jorvi Bipolar Study (JoBS), suicidal ideation at baseline 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 patients had seriously considered suicide. The predictive value of different definitions of ideation on suicide 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 definitions ranged from low to moderate (kappa coefficient 0.15 to 0.70). The correlates of suicidal ideation overlapped, but were not identical. Of the measures investigated, a baseline SSI score >/=8 had the 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 and means of measurement among BD patients. Different measures for ideation have the potential to cause inconsistency when correlates of suicidal ideation are investigated. For clinically predicting suicide attempts during the next few months, an SSI score >/=8 may best combine sensitivity and specificity. Language: en

Background We investigated the adequacy of maintenance phase pharmacotherapy received by psychiatric in- and outpatients with bipolar I or II disorder, including patients both with and without a clinical diagnosis of bipolar disorder (BD). Methods In the Jorvi Bipolar Study (JoBS), a naturalistic prospective 18-month study representing psychiatric in- and outpatients with DSM-IV BD I and II in three Finnish cities, we studied the adequacy of pharmacological treatment received by 154 patients during the first maintenance phase after index episode. Information on treatments prescribed during the follow-up was gathered in interviews and from psychiatric records. Results Of the patients with a maintenance phase in follow-up, adequate maintenance treatment was received by 75.3% for some time, but by only 61.0% throughout the maintenance phase and for 69.3% of the time (783/1129 patient months) indicated. Uninterrupted adequate maintenance treatment received was most strongly independently associated with having a clinical diagnosis of BD; other associations included inpatient treatment, rapid cycling and not having a personality disorder. Limitations Adequacy of dosage, duration or serum concentrations were not estimated. Findings represent an upper limit for adequate treatment within the cohort. Conclusions Provision or continuity of maintenance treatment was found to be compromised in more than one-third of BD patients during their first follow-up maintenance phase. As expected, clinical diagnosis of BD has a decisive role in determining adequacy of maintenance treatments. However, also rapid cycling may facilitate provision of adequate maintenance treatment, whereas outpatients and those with comorbid personality disorders may be disadvantaged subgroups.

Background Hopelessness, a key risk factor for suicidal behaviour overall, has been studied little among bipolar disorder (BD) patients. For purposes of prevention, it is important to know whether it is predominantly a patient's permanent trait or merely reflects the highly variable illness states. We investigated the degree to which hopelessness is trait- or state-related during the course of BD. Methods The Jorvi Bipolar Study (JoBS) is a naturalistic prospective study representing psychiatric in- and outpatients with DSM-IV BD I and II. Repeated measurements with the Beck Hopelessness Scale of 188 patients at baseline, 6 months and 18 months were analysed using a linear regression model with general estimation equations. Factors covarying with hopelessness during follow-up were investigated. Results Levels of hopelessness varied markedly between illness phases, being highest in depressive and mixed phases, and lowest in euthymia, hypomania or mania. Hopelessness was independently associated with concurrent severity of depression (estimate 0.231, p

Background Learning to detect prodromal symptoms is a key element of psychosocial treatment of bipolar disorder (BD). However, previous studies have described only prodromes of manic and depressive phases of BD I patients, while information on prodromes in BD II, or other phases is lacking. Methods The Jorvi Bipolar Study included 191 in- and outpatients with DSM-IV BD (90 BD I, 101 BD II) in any acute phase of illness at baseline. The prevalence, type and duration of preceding prodromes were investigated using open-ended questions. The effects of type I or II disorder, index phase, socio-demographic factors, comorbidity, illness history and other correlates on report and duration of prodromes were investigated. Results Prodromes were reported by 45.0% of BD I and 50.0% of BD II patients. The first prodromal symptom was usually mood congruent, but sometimes non-specific for mood or a symptom of anxiety; the median duration was 30.5 days. No differences between BD I and II, or between patients who did and those who did not report prodromes were found. Only Axis I comorbidity associated with longer prodromes, but not independently after adjusting for age, gender and type of phase. Limitations The study was cross-sectional. Reporting prodromes depends on patients' insight which was likely affected by a sub-acute phase at time of interview. Conclusions Only half of ordinary, secondary care bipolar patients are able to report prodromes. The chronic and fluctuating course of illness, and sometimes short time interval to full episode may limit the potentials of prodrome-based interventions.

Objective: To obtain a comprehensive view of the clinical epidemiology of bipolar I and II disorder in secondary-level psychiatric settings. Methods: In the Jorvi Bipolar Study (JoBS), 1630 non-schizophrenic psychiatric in- and outpatients in three Finnish cities were screened for bipolar I and II disorders with the Mood Disorder Questionnaire. Diagnoses were made using semistructured SCID-I and -II interviews. Information collected included clinical history, current episode, symptom status, and other characteristics. Results: A total of 191 patients with bipolar disorder (90 bipolar I and 101 bipolar II) were included in the JoBS. The majority of bipolar II (50.5%) and many bipolar I (25.6%) patients were previously undiagnosed; the remainder had a median 7.8 years delay from first episode to diagnosis. Despite several lifetime episodes, 26 and 58% of bipolar I and II patients, respectively, had never been hospitalized. A polyphasic episode was current in 51.3%, rapid cycling in 32.5%, and psychotic symptoms in 16.2% of patients. Mixed episodes occurred in 16.7% of bipolar I, and depressive mixed states in 25.7% of bipolar II patients. Conclusion: Even in psychiatric settings, bipolar disorders usually go undetected, or recognized only after a long delay. A significant proportion of not only bipolar II, but also bipolar I patients are never hospitalized. Polyphasic episodes and rapid cycling are prevalent in both types. Depressive mixed states are at least as common among bipolar II patients as mixed episodes among bipolar I.

Objective Few studies have investigated the prevalence of and risk factors for suicidal ideation and attempts among representative samples of psychiatric patients with bipolar I and II disorders. Method In the Jorvi Bipolar Study (JoBS), psychiatric inpatients and outpatients were screened for bipolar disorders with the Mood Disorder Questionnaire from January 1, 2002, to February 28, 2003. According to Structured Clinical Interviews for DSM-IV Axis I and II Disorders, 191 patients were diagnosed with bipolar disorders (bipolar I, N = 90; bipolar II, N = 101). Suicidal ideation was measured using the Scale for Suicidal Ideation. Prevalence of and risk factors for ideation and attempts were investigated. Results During the current episode, 39 (20%) of the patients had attempted suicide and 116 (61%) had suicidal ideation; all attempters also reported ideation. During their lifetime, 80% of patients (N = 152) had had suicidal behavior and 51% (N = 98) had attempted suicide. In nominal regression models, severity of depressive episode and hopelessness were independent risk factors for suicidal ideation, and hopelessness, comorbid personality disorder, and previous suicide attempt were independent risk factors for suicide attempts. There were no differences in prevalence of suicidal behavior between bipolar I and II disorder; the risk factors were overlapping but not identical. Conclusion Over their lifetime, the vast majority (80%) of psychiatric patients with bipolar disorders have either suicidal ideation or ideation plus suicide attempts. Depression and hopelessness, comorbidity, and preceding suicidal behavior are key indicators of risk. The prevalence of suicidal behavior in bipolar I and II disorders is similar, but the risk factors for it may differ somewhat between the two.

Objective: Our aim was to obtain a comprehensive view of differences between bipolar disorder (BD) patients with onset at early versus adult age in a representative study cohort. Methods: In the Jorvi Bipolar Study (JoBS), 1,630 psychiatric in- and outpatients were systematically screened for BD using the Mood Disorder Questionnaire (MDQ). A total of 191 bipolar I and II patients with a current DSM-IV episode were interviewed to obtain information about age at onset of mood symptoms, clinical course, treatment, comorbidity, and functional status. The patients were classified as either early onset (

Objective: To obtain a comprehensive view of differences in current comorbidity between bipolar I and II disorders (BD) and (unipolar) major depressive disorder (MDD), and Axis I and II comorbidity in BD in secondary-care psychiatric settings. Method: The psychiatric comorbidity of 90 bipolar I and 101 bipolar II patients from the Jorvi Bipolar Study and 269 MDD patients from the Vantaa Depression Study were compared. We used DSM-IV criteria assessed by semistructured interviews. Patients were inpatients and outpatients from secondary-care psychiatric units. Comparable information was collected on clinical history, index episode, symptom status, and patient characteristics. Results: Bipolar disorder and MDD differed in prevalences of current comorbid disorders, MDD patients having significantly more Axis I comorbidity (69.1% vs. 57.1%), specifically anxiety disorders (56.5% vs. 44.5%) and cluster A (19.0% vs. 9.9%) and C (31.6% vs. 23.0%) personality disorders. In contrast, BD had more single cluster B personality disorders (30.9% vs. 24.6%). Bipolar I and bipolar II were similar in current overall comorbidity, but the prevalence of comorbidity was strongly associated with the current illness phase. Conclusions: Major depressive disorder and BD have somewhat different patterns in the prevalences of comorbid disorders at the time of an illness episode, with differences particularly in the prevalences of anxiety and personality disorders. Current illness phase explains differences in psychiatric comorbidity of BD patients better than type of disorder.

Background We investigated our translation of The Mood Disorder Questionnaire (MDQ) as a screening instrument for bipolar disorder in a psychiatric setting in Finland. Methods In a pilot study for the Jorvi Bipolar Study (JoBS), 109 consecutive non-schizophrenic psychiatric out- and inpatients in Espoo, Finland, were screened for bipolar disorder using the Finnish translation of the MDQ, and 38 of them diagnostically interviewed with the SCID. Results Forty subjects (37%) were positive in the MDQ screen. In the SCID interview, twenty patients were found to suffer from bipolar disorder, of whom seven (70%) of ten patients with bipolar I but only two (20%) of ten with bipolar II disorder had been previously clinically correctly diagnosed. The translated MDQ was found internally consistent (alpha 0.79) and a feasible screening tool. Conclusions Bipolar disorder, particularly type II, remains commonly unrecognized in psychiatric settings. The Mood Disorder Questionnaire is a feasible screen for bipolar disorder, which could well be integrated into psychiatric routine practice.

Objectives: To investigate whether the course of bipolar disorder (BD) type II is more depressive than that of BD I, and, if so, to explore the underlying factors that cause this difference. Methods: In a prospective, naturalistic study of 191 secondary care psychiatric in- and outpatients diagnosed in an acute phase of BD I or II, 160 patients (85.1%) were followed for 18 months. Using a life chart, the exact timing of symptom states in follow-up was examined. Differences between BD I (n = 75) and II (n = 85) in duration of index phase and episode, time to full remission and recurrence, and time in any mood episode were investigated. Results: Patients with BD II spent a higher proportion of time ill (47.5% versus 37.7%; p = 0.02) and in depressive symptom states (58.0% versus 41.7%; p = 0.003) than BD I patients. This was a result of the higher proportion (61.7% versus 48.6%; p = 0.03) and mean number (1.69 versus 1.11; p = 0.006) of depressive illness phases in BD II, rather than of differences in the duration of depressive phases. Type of index phase strongly predicted the outcome. In linear regression models, both BD II and type of index phase predicted more time spent in depressive symptom states. Conclusions: In medium-term follow-up, BD II patients spend about 40% more time in depressive symptom states than BD I patients because a higher proportion of BD II patients have depressive phases and the frequency of these is higher. Differences in type of index phase may markedly confound differences in outcome between BD I and II.

Background: Differences in the incidence of suicide attempts during various phases of bipolar disorder (BD), or the relative importance of static versus time-varying risk factors for overall risk for suicide attempts, are unknown. Methods: We investigated the incidence of suicide attempts in different phases of BD as a part of the Jorvi Bipolar Study (JoBS), a naturalistic, prospective, 18-month study representing psychiatric in- and outpatients with DSM-IV BD in three Finnish cities. Life charts were used to classify time spent in follow-up in the different phases of illness among the 81 BD I and 95 BD II patients. Results: Compared to the other phases of the illness, the incidence of suicide attempts was 37-fold higher [95% confidence interval (CI) for relative risk (RR): 11.8–120.3] during combined mixed and depressive mixed states, and 18-fold higher (95% CI: 6.5–50.8) during major depressive phases. In Cox’s proportional hazards regression models, combined mixed (mixed or depressive mixed) or major depressive phases and prior suicide attempts independently predicted suicide attempts. No other factor significantly modified the risks related to these time-varying risk factors; their population-attributable fraction was 86%. Conclusions: The incidence of suicide attempts varies remarkably between illness phases, with mixed and depressive phases involving the highest risk by time. Time spent in high-risk illness phases is likely the major determinant of overall risk for suicide attempts among BD patients. Studies of suicidal behavior should investigate the role of both static and time-varying risk factors in overall risk; clinically, management of mixed and depressive phases may be crucial in reducing risk.