Suicide Risk and Treatments for Patients With Bipolar Disorder

Bipolar disorder, one of the most common severe mental illnesses, includes type 1 (with mania and usually recurrent depression) and type 2 (recurrent major depression with hypomania).^{1 - 2} Lifetime prevalence for type 1 bipolar disorder is approximately 1%, but inclusion of more broadly defined conditions increases this rate to 2% to 5%.³ Bipolar disorder can begin in childhood or adolescence,⁴ continues throughout life, and is extraordinarily costly—financially as well as clinically and socially.⁵ The course of bipolar disorder is episodic but highly variable, with potential for high levels of severity and recurrence intensity, disproportionately high depressive morbidity, and comorbidity with substance abuse and anxiety disorders.³ Bipolar depression can be present during 20% to 30% of patients' time, even during prophylactic treatment,^{6 - 8} and is closely associated with disability and mortality.^{3 ,9 - 10} Bipolar disorder proves fatal in a high proportion of patients from complications of risk-taking behavior, comorbid stress-sensitive medical illnesses, and especially suicide.^{3 ,9 - 10} These characteristics mark bipolar disorder as a major unsolved public health challenge.

Treatment for bipolar disorder was revolutionized in the early 1970s by the introduction of US Food and Drug Administration (FDA)–approved long-term prophylactic treatment with lithium carbonate for preventing recurrences of mania and bipolar depression. Lithium remains the international standard of comparison for an increasing number of innovative treatments introduced in recent years, including several anticonvulsants and modern antipsychotics.^{11 - 12} However, evidence for long-term prophylactic effectiveness of agents other than lithium remains limited, and comparisons of specific agents in particular phases of bipolar disorder are needed.¹¹ No currently available treatment provides full protection from recurrences of manic, mixed manic-depressive, major depressive, or highly prevalent milder depressive states.^{11 - 12}

Among patients with type 1or 2 bipolar disorder, long-term lithium monotherapy can reduce mania or hypomania and depression similarly, by about two thirds.^{11 - 12} Most other agents with proposed mood-stabilizing effects appear to be more effective against mania and hypomania than against bipolar depression, with the possible exception of lamotrigine.^{11 ,13} For example, in well-designed comparisons with other agents, divalproex has yielded considerably less long-term protective effects against bipolar depression or mixed dysphoric-agitated states than against mania,^{14 - 15} although divalproex has been shown to limit depression by some measures.¹⁶ Addition of an antidepressant is unlikely to resolve the problem of residual depression, owing to inconsistent responsiveness of bipolar depression to antidepressants and their risk of inducing mania or emotional instability in patients with bipolar disorder.^{17 - 18}

As a cause of premature death in patients with bipolar disorder, suicide represents an extraordinarily high risk, with an estimated rate of 0.40% per year vs the international general population average of 0.017% per year, with a standardized mortality ratio of 22.^{19 - 20} For comparison, the average standardized mortality ratio for suicide is about 20 among persons with unipolar major depression and 8.4 for those with schizophrenia.¹⁹ Moreover, suicide attempts by patients with bipolar disorder have an increased lethal potential. Among patients with bipolar disorder, rates of suicide attempts are only 5 times the rates of completed suicides, whereas in the general population, rates of suicide attempts are 10 to 20 times the rates of completed suicides.²¹ Suicide in patients with bipolar disorder is strongly associated with depressive and dysphoric-agitated (mixed) states.^{21 - 22} Suicidal risk may be greater in type 2 bipolar disorder,^{23 - 24} characterized by less severe episodes of elevated mood than type 1 but with severe recurrent depression, often misdiagnosed as nonbipolar major depression and not consistently treated with mood-stabilizing agents.²⁵

Long-term use of lithium is the only treatment consistently associated with much lower rates of suicide and life-threatening suicide attempts in bipolar disorder. In a recent meta-analysis of 22 studies with data permitting estimates of annual rates of suicide in patients with mainly bipolar disorder, the computed risk of completed suicide was 8.85 times less with long-term lithium treatment than without it.²⁶ Nevertheless, the rate of completed suicides among patients treated with lithium remained 10 times higher than that in the general population.²⁷ Similarly large proportional reductions in the risk of suicide attempts (to levels close to population base rates) also were associated with lithium treatment.^{20 - 21 ,27} These benefits were sustained over many years, but discontinuation of lithium was associated with a sharp temporary increase in suicide risk for several months; this increase was half as large when lithium was discontinued gradually during a span of at least several weeks.²¹ Since some earlier studies included patients with recurrent unipolar major depression along with those with bipolar disorder,^{26 - 27} this beneficial effect of lithium may extend to suicidal risk in patients with nonbipolar depression, but this hypothesis requires study.^{26 - 27}

Differential protective effects of specific mood-altering agents against bipolar depression are likely to be important for reducing suicide risk, but relevant direct comparisons are rare.^{11 - 12 ,18} Such comparisons suggest less overall clinical benefit with long-term use of carbamazepine than with lithium²⁸ and greater suicide risk with this anticonvulsant than with lithium.²⁹ Valproic acid preparations have largely displaced lithium in the United States for treatment of mania and long-term prophylaxis for bipolar disorder despite limited research support for the prophylactic effectiveness of this anticonvulsant, particularly against recurrences of bipolar major depression.^{11 ,13 - 14 ,16} Until now, there has been no specific studies of the effects of valproate on suicide risk among patients with bipolar disorder.

In this issue of THE JOURNAL, Goodwin and colleagues³⁰ report evidence that anticonvulsant treatment (almost entirely with divalproex) among patients with bipolar disorder was associated with higher rates of suicide and suicide attempts than among those treated with lithium for varying periods of time. Goodwin et al analyzed data involving large numbers of adult and adolescent bipolar disorder outpatients (N = 20 638) obtained from computerized databases from 2 large health plans in California and Washington State, with efforts to avoid likely confounding. Such records can vary in accuracy and completeness, and estimates of drug exposure times require assumptions about the relationship of pharmacy records to actual drug taking and about combining periods with and without putative drug exposures fragmented over time. Nevertheless, such limitations should apply similarly among the 3 main treatment conditions analyzed (lithium, anticonvulsants divalproex or carbamazepine, or neither), and should not produce systematic bias.

The adjusted risk ratios for completed suicides and attempts during treatment with divalproex vs lithium were 2.7 and 1.8, respectively.³⁰ The observed suicide rate during lithium treatment was 0.066% per year (9 suicides per 13 597 person-years) and during use of anticonvulsants (mainly divalproex, with some carbamazepine) was 0.155% per year (16 per 10 333 person-years).³⁰ Suicide risk without treatment was 0.116% per year (25 per 21 562 person-years),³⁰ or 9.3 times higher than the California and Washington State base rates of about 0.0120% per year.³¹ The difference in risk with vs without lithium treatment was 8-fold (0.116/0.066)—very similar to previous experience (nearly 9-fold).²⁶ However, the rate was not decreased with anticonvulsant treatment (0.116/0.155 = 0.75). Also of interest, the observed ratio of suicide attempts to suicide deaths for all treatments was 6.4 to 1 (0.565/0.088),³⁰ again indicating high lethality of attempts among bipolar disorder patients, given comparable general population ratios of 10:1 to 20:1.²¹

The study by Goodwin et al not only has merits in its methods and findings but also represents a significant contribution to a newly emerging interest in suicide as the major unsolved medical problem that it is.³² Not until this year has the FDA approved any treatment to prevent suicidal behavior, with the recent approval of clozapine for such purposes among patients with schizophrenia or schizoaffective disorder. This approval was supported by a prospective randomized study showing about 32% lower risk of nonlethal suicidal behaviors with clozapine than with olanzapine,³³ which may also reduce risk of suicide.³⁴ Hopefully, such renewed interest in the potentially treatment-modifiable lethality of major mental disorders will be sustained and increasingly successful.