Targeting suicide prevention to modifiable risk factors: has bipolar II been overlooked?


There is a new momentum of research on suicide as reflected in an increasing database documenting its robust connection to mood disorders (1). The Tondo et al. report in the present issue of Acta Psychiatrica Scandinavica from the Sardinia–Harvard collaboration (2) is exemplary in this respect, presents extensive data on various indices of suicidality from a well-ascertained and large affectively ill Italian cohort, thereby providing the opportunity for this invited editorial. While the report’s rich database can be profitably examined from a variety of suicidologic angles, as a physician and affectiveologist I would like to focus on several unresolved clinical, public health and professional issues of suicide which continue to challenge the field of psychiatry and medicine at large.

Clinical encounters with suicide

Although suicide is a relatively rare event, it ranks among the first ten causes of death in most western countries. It is customary to state that it is a complex behavior with multiple contributory causes, psychologic, social, cultural, and biological. Very early in their training, all psychiatric, other mental health and medical student trainees are provided with a long list of risk factors which include demographic variables, life events or situations, contributory mental and physical disorders, current mental state variables and personality factors, as well as putative biologic indices. Such a list is often supplemented by elaborate statistics and graphs, broken down by age, sex, ethnic group, profession, country, latitude, and season, with the expectation that our trainees would thereby be armed with the tools they need to prevent suicidal behavior.

How do our trainees integrate the foregoing information into their practice? For nearly four decades, in my capacity as an attending who has listened to accounts of suicidal patients presented by interns, residents and psychologists, social workers or nurses, I keep on hearing several scenarios: ‘22-year-old borderline female who slashed her arms in the context of complex romantic entanglements in a Lesbian relationship’; ‘45-year-old male alcoholic with three failed marriages and multiple failures at sobriety programs, presenting now with a failed attempt to shoot himself’; ‘24-year-old sociopathic male, in a cocaine overdose, to avoid prison for possession.’ I submit that these pejorative descriptions represent counter-transference which stem from the young professionals’ helplessness when faced with suicidal passions. I use the term ‘passions’, because suicidality is not simply an isolated act of desperation, but often an active anguish and emotional battle which intimately engulfs others, including us, the clinicians. That the negative appraisal when confronted with a suicidal person is not so much of a reaction to these patients’ mental illnesses per se, but the type of individuals who ‘chose’ to act in such desperate ways, can be seen in the fact that descriptions of elderly physically ill individuals who engage in suicidal communications are no less judgmental: ‘78-year-old non-compliant male with deteriorating chronic obstructive lung disease who feels entitled to smoke’. And finally, the ultimate shock: ‘Why did this 50-year-old physician who had everything hang himself?’.

All physicians, we psychiatrists included, should examine our professional attitudes and emotional reactions to suicidal acts which defy everything we stand for in our busy professional lives dedicated to preserving life. There is a beginning understanding of the urgent need to do so (3). This topic is beyond the present editorial, but I thought we should start here, as a prelude to my main theme: to consider a converging research literature which could provide new hope for bringing more effective suicide prevention strategies in clinical practice and perhaps even at a population level.

‘Depression’ as the final common pathway in suicide

Based on a review of the then relatively sparse data-based literature on suicide, as well as our decade of experience in two mood clinics (one university, the other community-based), we emphasized ‘the necessity of treating contributory psychiatric disorders’ in suicide prevention (4). That was a quarter of a century ago, but even then it was obvious that two-thirds of suicides arose from mood disorders. Recent research (1) confirms that what is common to many if not most patients described in the foregoing clinical suicide scenarios is a depressive state which is in principle treatable if we are willing to go beyond comorbid diagnoses and other Axis II distractions; even among ‘non-affective’ diseases such as schizophrenia (5), alcoholism (6), epilepsy (7), and physical illness (8), the suicidal act typically occurs during a superimposed depressive state. Hopelessness, the most important mental state correlate of suicidality (9), is the hallmark cognitive disturbance of depression (10). Furthermore, the perception of functional impairment, is often more critical in suicidality than the objective disability produced by disease (11, 12), again testifying to the importance of depressive cognitions. That’s why it is tragic that most individuals who commit suicide in the community had rarely received the benefit of adequate pharmacologic, psychotherapeutic or electroconvulsive treatment for depression (13–16).

Fortunately, a great deal of evidence since the introduction of newer classes of antidepressants into clinical practice (which made acceptance of antidepressant treatment by patients less objectionable), in many but not all European countries (17–19), and more recently in Japan (20), there has been marked diminution of suicide rates. Moreover, a prospective trial involving the training of general medical practitioners in Gotland, Sweden, has shown significant reductions in suicide rates on that Island (21, 22). The highest decline in suicide has been achieved in Denmark and Hungary (18, 23–25). Hungary, well-known for its traditionally high suicide rate, owes its success (25) to increasing availability of mental health services and increased physician work-force which have led to increased utilization of antidepressants; the decline in suicide rate has occurred despite post-Soviet increase in unemployment and divorce rates, again testifying that depression is the critical factor in suicide. A more recent prospective intervention (26), involving general practitioner education in Hungary, has too shown significant reductions in suicide. There has also been steady decline in suicide rates from the pre-electroconvulsive era (1900–1939) throughout much of the tricyclic antidepressant era (27). These data and considerations suggest that suicide prevention is possible in the community, even at the national level. An international panel of experts on suicide concluded that training of physicians in the primary care sector and limiting access to lethal means are the only proven strategies in suicide prevention impacting at the population level (28). It also appears possible for a well-trained psychiatrist in solo practice to achieve such results for his own clientele (29).

I would now like to turn to the Tondo et al. report (2). While agreeing with their reservations on the efficacy of antidepressants as agents in the prevention of suicides in the specialized psychiatric setting of mood clinics – where patient profiles are skewed towards bipolar and related recurrent mood disorders – I do not share their implied skepticism about the potential antisuicidal effectiveness of these agents in general psychiatric practice and in community settings. Clozapine – recently approved by the FDA for the prevention of suicide in schizophrenia – and lithium indicated primarily for manic-depressive illness and emphasized throughout the present (2) and several previous papers by the Sardinia-Harvard group (30–32) are not of proven use for the garden-variety type (non-bipolar) depressive patients encountered in clinical practice. On the other hand, an earlier report by the Sardinia-Harvard group (33), suggesting that lithium might be efficacious for the broader spectrum of recurrent ‘unipolar’ depressions – which in my opinion are on the edge of bipolarity (34) – must receive renewed attention.

Although lithium appears to have made a major impact in reducing suicide rates in bipolar and possibly related recurring unipolar patients in bipolar clinics (35, 36), other mood stabilizers, such as valproate and lamotrigine, need to be further investigated in this regard. Valproate could be relevant (37), in light of the fact that mixed states are increasingly being implicated in suicidality as shown, among others (38, 39), by a previous contribution of Tondo et al. (40). Lamotrigine should also be considered, because state depression and its recurrence in bipolar patients contribute significantly to suicidality (41, 42). In a 40-year follow-up study, Angst et al. (43) have suggested that all efficacious treatments for bipolar and related mood disorders might be operative in reducing the suicide rate in the long-term.

Manic depressive illness in its fully developed ‘classic’ form occurs in 1% of the general population, accounting for a relatively small population share of suicidal acts in the community at large (44–46). The majority of individuals who commit suicide in the community come from a pool of people who suffer primarily from depressive conditions, whether or not it is comorbid with other disorders (1, 4, 47). An increasing aging population with multiple physical illnesses and depression contributes to the suicide pool in the general population (48); it is also imperative to identify depression presenting with physical complaints (49) or bereavement (50) in the elderly.

In nearly every country depression, under different guises, is largely encountered in the primary care sector (51). Unfortunately, many non-psychiatric physicians are neither sufficiently experienced in caring for depressive patients, nor inclined to deal with the clinical challenges posed by suicidal patients. If the purpose of medical education is to learn the science and art that equips us to prevent, treat, attenuate disease and reduce its complications, including death, then the fact that many suicides in the community occur in individuals with recent contacts with their physicians (49, 52, 53) does not augur well for our medical school curricula, nor us psychiatric educators and public health-minded suicidologists who have failed to significantly impact such curricula.

The emerging role of bipolar II

There is increasing awareness that bipolar-II is involved in suicidality. First reported by Arato et al. (44), more recent Hungarian research (45, 46) has replicated the pivotal role of BP-II: at least a third of all suicidal acts are committed by patients with BP-II. The Sardinian-Harvard major affective disorder database (2) provides a critical finding regarding the lethality of suicidal acts in these patients: It shows that the attempt/completion ratio is lowest in bipolar II, which effectively makes it the most lethal of all mood disorders; curiously, the implications of the high lethality of BP-II is not discussed by these authors. Their cautiousness is perhaps due to the relative infrequency of BP-II in the Sardinian database (2). I find this somewhat puzzling given that in the Rome Lucio-Bini Center, the ‘parent’ of the Sardinian Center, BP-II is more often diagnosed or encountered (54). This could reflect the use of the ‘narrow’ BP-II DSM-IV construct utilized in the Sardinian sample (2), which is in line with Baldessarini’s skepticism about the bipolar spectrum (55). There are compelling reasons why BP-II is relevant for suicide prevention at both the clinical and population levels.

  • i)Current data suggest that BP-II accounts for 30–58% of all major depressions in psychiatric practice (54, 56–58), and 10–25% of depressions in primary care or family practice (59, 60).
  • ii)Converging data from different epidemiologic studies have shown a general population prevalence of at least 5% for BP-II spectrum (61–63).
  • iii)Mixed states are very common in the BP-II spectrum (64), recently, shown to be a common substrate to many depressive suicide attempts (65).
  • iv)The cyclothymic temperament is the most prevalent temperament in bipolar II (66), and has been shown to be a predictor of suicide in a recent adolescent study (67).
  • v)Rapid mood shifts are the hallmark of BP-II (68, 69).
  • vi)BP-II, under one rubric, can be said to subsume many if not most risk factors involved with suicidality, i.e. depression, panic attacks, mixed state, cycling, aggressivity, and most relevantly, impulsivity, labile temperament, family history of suicide, as well as alcohol, heroin, and stimulant abuse (68–74). BP-II suicides are the most lethal, also in part because they use the most aggressive methods (75).

It is finally relevant to mention that borderline personality, believed (by its very defining characteristics) to be a substrate of suicidality in the borderline field, more often than not represents a missed bipolar spectrum diagnosis or, for that matter, a mixed labile and/or cycling bipolar diagnosis (76, 77), though vigorously denied by borderline experts (78). The complacent belief that suicidality is the ‘specialty’ of borderline patients should not be used as an excuse for the inevitability of these patients’ suicide – nor to exculpate the clinician.

Are sudden downshifts in mood the final suicidal crisis?

The most important lesson to be drawn from the Tondo et al. report (2) deriving from the Sardinian mood clinic – one of the most renowned clinical research settings in the world – is the tragic reality that even in such a sophisticated clinic suicidal behavior exceeds general population expectations by as much as eight-fold (2). Is Sardinia, or some regions thereof, ‘genetically’ more disposed to suicide, or should one seek other explanations?

In a previous publication from the National Institute of Mental Health Collaborative Study on depression (79), we reported that ‘suicide attempts’ and ‘completed suicide’ in mood disorders are very similar in most parameters, including the fact that both arose from a mixed/cycling recurrent substrate and impulsivity. Such data suggest that clinicians and suicide researchers should avoid the temptation to distinguish them too sharply, primarily on the basis of ‘intentionality’, i.e. ‘gestures’ vs. genuine suicidal acts. ‘Successful’ suicide might, in final analysis, be a matter of sudden dips in mood in someone (i.e. a BP-II with labile mixed states, an epileptic with labile dysphoria, or a depressive patient in the expected zigzag phase during the recovery process on antidepressants), who is knowledgeable about and/or has access to the most lethal means.

Protective factors and treatment-adherence

That’s why protective factors, such as one’s immediate family and broader social network and access to mental and, perhaps even more importantly, to general medical care – emphasized in another contribution by Tondo et al. (80) – are so crucial in suicide prevention. Treatment adherence (81) itself could be conceptualized to belong to this class of protective factors.

Even the effectiveness of lithium in traditional‘lithium clinics’ can in part be accounted for by the vigilant caring provided in these clinics on an ongoing regular basis (82). However, judging from the suicide statistics reported by Tondo et al. (2), it did not seem to have worked very well in Sardinia. Perhaps not so surprising, given that the affectiveologist in the newer academic and/or research bipolar centers no longer have the luxury to ‘select’ the most suitable patients for lithium prophylaxis, i.e. those with the mania-depression-free course pattern (83, 84). At least one European report suggests that lithium applied to the treatment of a spectrum of mood disorders might fare much better in general medical practice (85). It is likely, however, these were mostly bipolar patients who were not severe enough to be followed in a specialized lithium clinic in the psychiatric setting; that treatment-adherence is a crucial factor even among bipolar patients followed-up by general practitioners can be seen in the increase in suicidality when they dropped out from the GP clinic.

The need for a new political will

It is obvious from the foregoing review of the research and disquisition that suicide prevention efforts in the specialized psychiatric sector involving bipolar and/or lithium clinics are unlikely to impact the overall suicidal morbidity and mortality in the community. Apart from the fact that such clinics today attract a ‘refractory’ group of bipolar patients, their main limitation in suicide prevention derives from the fact that most suicides in the community are due to depression. Focusing suicide prevention on modifiable depressive disorders in non-psychiatric medical settings would require the political will to mandate depression treatment in all primary care medical settings.

Veterans, who in the United States receive their psychiatric and all their medical care in the same setting, are ideal for such interventions. As a matter of fact, data just released for publication, from veterans’ setting, have shown that in this high risk ex-military population, among bipolar patients suicidal behavior is best achieved with mood stabilizers (lithium, valproate, and charbamazepine) and avoidance of antidepressants (86–88). By contrast, in the much larger pool of depressive veterans, antidepressants appear strongly protective against suicidality (89). The collaborative care model is optimal, if not the ideal, in the veterans setting (90, 91).

The Gotland (22) and Jämtland (17) studies, as well as the Danish (18) and Hungarian (24) experience need to be better publicized. Likewise, regional suicide prevention efforts which have proven effective in large parts of other countries – e.g. in France (92) and the United States (29, 89) – must be showcased. Moreover, national suicide prevention centers and agencies must go beyond their traditional roles of gathering suicide statistics, reporting and supporting research. Although recommendations from the international body representing them (93) are cogent in emphasizing the need to focus on the special suicide characteristics in each nation, they need to develop the ‘political’ base to influence national care health plans to include mood disorder treatment as a priority, along with such major killers as cardiovascular diseases, cancer, pneumonia, and influenza. In other words, declaring suicide prevention as a national priority is too global and unlikely to succeed, unless the focus is on physician training on the recognition and treatment of the mood disorders in the primary sector. Diminishing access to lethal means such as firearms (94, 95) will require a different type of political will beyond the scope of this editorial. Optimally, the respective strengths of the public health model of suicide and the clinical approach to affective and related disorders must be wed for suicide prevention to work in practice.

Ultimately, I propose to go beyond mood and bipolar clinics, to develop affectiveology as a psychiatric subspecialty of 2 years, no less in status than cardiology. The analogy is appropriate inasmuch as mood disorders are fatal not just in their own right, but contribute as risk factors for heart attacks and cardiac mortality (96–98). Affective disorders prevalent in general medical practice (49), represent the bread and butter of psychiatric practice. It could be said, by analogy, that affectivelogy to psychiatry is what internal medicine is to medicine at large.

Just as physicians have incorporated a discussion of such ‘sensitive’ topics as sexually transmitted diseases, cancer and weight, physicians must learn to be comfortable with depression and suicidality. A brief self-rated depression scale – with items on suicidality had hopelessness – must be obtained at every visit, just as temperature, blood pressure and weight are measured as a matter of routine. Such habits are best learned during medical school and post-graduate training, preferably from physician role models who themselves engage in them routinely.

Such training must be required (i.e. the equivalent of half-day per week for 6 months during the post-graduate years) for all primary care medical specialists, such as general practitioners, family medicine, internal medicine, general surgery, and obstretric-gynecology in their own practice settings. Such a program has been successfully instituted, a family practitioner himself serving as the supervising physician who had developed a mood clinic in the very setting of the family practice residency (99), but did not survive funding cuts.

Developing affectiveology as a subspecialty will assure state-of-the-art diagnosis, treatment and preventive approaches to the more recurrent mood disorders at the severe end of these potentially fatal disorders. Mood clinics and affective disorder centers will also provide the academic back-up to primary care physicians who will be involved with the clinical care of the more prevalent and relatively milder end of the spectrum of depressive disorders in the community.

In view of the limited space of this editorial, it was not possible to address the protective role of spirituality and the clergy, nor triage issues involving psychological counseling in high schools and colleges and the community. Nor could I adequately address the controversy on the FDA advisory on suicidal ideation and antidepressants in juvenile subjects, which has been dealt with elsewhere (19) and in the most recent literature (100).

Given a determined political will to prevent suicidal mortality via effective treatment of mood disorders, such prevention – which to some extent is already taking place, among others, in Scandinavia and Hungary and possibly in parts of France and the United States – can be amplified, extended or otherwise generalized elsewhere, just as it is happening in fields like stroke, coronary mortality, cancer, and AIDS in some parts of the world. I have argued that focusing on depression and better understanding of BP-II is crucial in such an ambitious public health project.

As I conclude this editorial, it is befitting to cite Dunner et al. (101), who in their pioneering report on BP-II, emphasized the high suicidality of these patients. A great deal can be learned about the unpredictable sudden mood reverses of BP-II that might be relevant to other depressives, substance and alcohol-abusive and epileptic patients, and the medically ill with ‘secondary’ depression, as well as in those extremely rare instances where individuals without known psychiatric illness might kill themselves.

I would like to briefly return where we started: suicidal behavior is complex and multidimensional, and on the global world stage (102) manifests in significant variations, such that no single approach can be used in understanding it. Nonetheless, I have argued that affective disorders represent a modifiable spectrum of risk factors, along with reducing access to lethal means particular to a country or an age group. I have further argued that affective disorders, particularly depression (whether ‘primary’ or ‘secondary’), represent the most prevalent substrates for suicidality, yet have not been given sufficient space – in the education of both psychiatrists and primary care physicians – commensurate to their clinical and public health importance. Finally I have argued that persuasive data exist from many countries of the world in favor of the position that successful treatment of affective disorders is key to reducing suicidality in the community.

Will we in the near future be able to prevent the first presentation of suicides at the morgue? Given that the pathologist may have little interest in responding to this question, the present editorial is a call for action for psychiatry and primary care medicine.


I wish to thank Drs Tondo, Baldessarini and their team for their painstaking research on suicidal behavior which provided the opportunity for the present editorial, expanded from what was to be an editorial, as well as the many dedicated and talented researchers – both cited and uncited – whose work inspired me. That this is appearing in Acta Psychiatrica Scandinavica is natural, given that Scandinavian countries have played a major role in suicide research. Finally, I wish to thank Drs Zoltan Rihmer, Sidney Zisook and William Norcross who made several insightful suggestions, as well as Kareen Akiskal for her constructive feedback. However, I am solely responsible for the extrapolations of the existing data in order to say what, after nearly four decades of work as an affectiveologist, I felt needs to be said.

Conflict of interest

Dr Akiskal has been a U.S. Advisory Board Member for bipolar disorder for Glaxo-Smith-Kline, as well as on the Latin American bipolar advisory board for Abbott, and has served in a similar capacity for Sanofi-Aventis in Europe and Lilly for Japan.