Appropriate maintenance treatment is essential to the well-being of bipolar patients. Although mood stabilization tends to be the immediate goal of psychopharmacologic intervention, clinicians should also consider potential suicidality when treating a bipolar patient. Appropriate maintenance treatment can substantially reduce the risk of suicide (Ba-ethge et al. 2003; Coppen and Farmer 1998; Tondo and Baldessarini 2000, 2001a), but how do we decide what constitutes "appropriate maintenance" when it comes to suicide prevention? Although a variety of mood stabilizers, antidepressants, and antipsychotics have been the focus of research in this realm, the answer is not eminently clear.
Given its frequent and historical use in bipolar illness, it is not surprising that lithium has received a great deal of attention in research concerning psychopharmacology and suicide prevention in patients with mood disorders. A variety of meta-analyses, smaller independent studies, and study of lithium and anticonvulsants in two large health insurance databases generally substantiate a cautiously optimistic view of lithium as a prophylactic agent against suicide in bipolar disorder. While research on lithium's protective effects is promising, it is also essential to keep in mind the complications surrounding the research designs as well as the daily clinical problems of treatment noncompliance in the bipolar population.
Goodwin et al. (2004) reported a strong effect favoring lithium compared with divalproex sodium in 20,638 members of two large health maintenance organizations in the United States. The incidence of emergency department admissions for suicide attempts (31.3 for valproate vs. 10.8 for lithium per 1,000 persons/year), suicide attempts resulting in hospitalization (10.5 vs. 4.2 per 1,000 persons/year), and suicidal death (1.7 vs. 0.7 per 1,000 persons/year) were all lower in lithium-
treated patients. After adjustment for a number of demographic factors, including age and psychiatric and medical comorbidity, the risk of suicide death was reported to be 2.7 times greater for patients prescribed divalproex for a diagnosis of bipolar disorder compared to those prescribed lithium. Although intriguing, the nonrandomized nature of the sample leaves open the concern that the groups were clinically different. For instance, it is not known how many of the patients in the dival-proex group had previously failed to respond to lithium and were thus a treatment-resistant group. In any case, the results are strongly in favor of lithium and are consistent with other examinations of the effect of lithium on suicidality.
A case-control study by Modestin and Schwarzenbach (1992) found that matched controls discharged within 1 year from an inpatient psychiatric hospitalization were more likely to be on lithium than those who committed suicide, yet were equally likely to have dropped out of treatment. Whether this represents a protective effect from lithium is unclear.
A meta-analysis of 33 studies investigating long-term lithium treatment between 1970 and 2000 yields results that overwhelmingly favor lithium as a potential means of suicide prevention (Baldessarini et al. 2001). Of the 19 studies comparing groups with and without lithium treatment, 18 found a lower risk of suicide in the treatment group and one had no suicides in either group. The results clearly present a favorable association of lithium with decreased suicidality. Overall, the meta-analysis demonstrates a thirteenfold reduction in suicidality for patients with an affective illness, leading to a largely reduced risk, which nevertheless remains larger than that estimated for the general population. Specifically, the rates of suicide associated with lithium treatment (0.109%-0.224% annually) are 10 times greater than the international base rate (0.017%) (Baldessarini et al. 2001). Thus, in spite of the prophylaxis afforded by lithium against suicidality, its protective effects appear to be somewhat incomplete.
A second meta-analysis concerning lithium use in bipolar disorder further supports the claim of a demonstrably decreased suicide risk in patients treated with lithium. Tondo and Baldessarini's (2000) review of 22 studies reveals a sevenfold decrease in suicides for patients treated with lithium as compared to those not receiving lithium or having discontinued it. The suicide rate for the group receiving lithium treatment was approximately 0.227% per year, while the group without lithium treatment was 1.778±1.444% per year. As with the meta-analysis discussed above, however, the decreased rates associated with lithium treatment are still significantly greater than those found in the general population; in this case, by 13.7 times. In spite of the high risk (compared with the normal population) of suicide in the group treated with lithium, it appears that the risk of suicide attempts is decreased to a greater extent with lithium treatment. In fact, the rate of suicide attempts associated with lithium treatment may actually be less than the rate of suicide attempts in the general population (0.255% annually vs. 0.315% annually).
A third meta-analysis of 32 randomized, controlled trials was recently reported (Cipriani et al. 2005). In these studies, 1,389 patients were randomized to lithium and 2,069 to other agents, including placebo (n = 779), antidepressants (n=247), and other mood stabilizers and combination treatments. Lithium treatment was associated with a reduced likelihood to die by suicide (2 vs. 11 suicides, odds ratio=0.26, 95% confidence interval [CI] = 0.09-0.77), a reduced likelihood for self-injurious behavior (0 vs. 7 events, odds 0.21, 95% CI=0.08-0.50), and reduced death from any reason (9 vs. 22, odds ratio = 0.42, 95% CI = 0.210.87). It follows that lithium may exert particularly strong protective effects for patients who are less inclined to commit suicidal acts with the intention of suicide completion. In other words, although lithium may not lower suicide rates in patients with mood disorders to the level found in the general population, the fact that it seems to drastically reduce the frequency of attempts demonstrates its protective properties.
Several notes must be made regarding the reliability of these meta-analyses, given several substantial limitations. First, the majority of the studies reviewed in both did not seek to address suicidality as a primary outcome variable (Baldessarini et al. 2001; Tondo and Baldessarini 2000). Second, it is noteworthy that the results of these meta-analyses were likely tainted by the presence of a variety of diagnoses in the study samples. Frequent inclusion of individuals with diagnoses of major depressive disorder and schizoaffective disorder confounds our understanding of the effects of lithium in a strictly bipolar population. Additionally, given the high rates of comorbidity in affective illness (i.e., whether the secondary or tertiary diagnosis is one of substance abuse, an anxiety disorder, an Axis II disorder, etc.), it would be instructive to parse out the effects of lithium for patients with a comorbid illness and those with a "pure" bipolar diagnosis. Furthermore, lithium has historically been the first-line treatment; severely ill patients or lithium nonresponders may have disproportionally higher suicide rates. Fortunately, several individual studies address these limitations and may present us with a fuller illustration of lithium effects in specific segments of the bipolar spectrum.
It is difficult to confidently generalize results of many psychophar-
macologic studies to the general bipolar population involving lithium because of the lack of trials applying a randomized design. Tondo, Hennen, and Baldessarini's (2001b) meta-analysis indicated that among 22 studies comparing a group with lithium treatment with a group without this treatment, only three were double-blind, randomized, controlled trials (RCTs). However, these trials further substantiate and strengthen the claim that lithium exerts a protective effect against suicide. In fact, in 328 patients not receiving lithium therapy and 369 receiving lithium (yielded by an analyses of the three RCTs mentioned above), 1.28% per year committed suicide in the former group compared with 0.00% per year in the latter. An additional analysis of nine RCTs involving lithium found similar rates and further illustrates lithium's protective effects against suicide (Burgess 2002; Tondo et al. 2001b). Notably, no study found a negative effect for lithium, and although not all of the studies found statistically significant effects, all effects were in the same direction (i.e., lithium was associated with a lower suicide rate). Given the prospective and randomized designs of these studies, they provide strong support for the use of lithium in patients with mood disorders judged to be at risk for suicidal behavior.
Reasonable rates of treatment compliance are necessary if we are to expect psychopharmacologic interventions to exert the desired effects, yet treatment compliance is a constant struggle for many bipolar patients. In issues of suicidality, particularly with respect to a medication with promising protective effects, compliance may be especially worthy of attention. In fact, it appears that compliance with lithium treatment may not only decrease the risk of suicide but that it may actually decrease general mortality rates. In a sample of 103 patients treated with lithium and followed over the course of 11 years, the projected number of deaths was 18.31 (based on demographic factors for the population), while the actual number of deaths was 10 (Coppen et al. 1991), with no suicides. The authors concluded that lithium may compensate for the high mortality rate found in mood disorders and actually reverse it. Notably, the study sample was highly treatment compliant, with only 10 patients discontinuing treatment over the course of study duration. Alternately, Brodersen et al. (2000) found that over the course of 16 years of observation, 40 of the 133 study participants who were treated with lithium died, with 11 deaths resulting from suicide. While these results could be taken as further support for the theory that lithium possesses an incomplete protective effect, further examination of the study sample indicates that treatment noncompliance may have been a strong risk factor for death. In fact, individuals who were not compliant with their lithium treatment were four times more likely to die from suicide than those who were compliant, although these findings were marginally insignificant (P=0.06). Direct conclusions as to the direction of causality in such a study are impossible. Perhaps lithium does not have as strong a protective effect as is believed or perhaps its negative side effects lead to higher rates of noncompliance. Regardless of the reason, noncompliance is a serious problem in bipolar patients and must be addressed if we are to expect that lithium will provide prophylaxis against suicidality.
The potential dangers associated with lithium discontinuation may further inform our understanding of lithium's effects on suicidality. It appears that rapid or accelerated lithium discontinuation (as may be practiced by noncompliant individuals who decide to simply stop taking their medications) places a person at serious risk for suicidal behavior. In a sample of 165 patients who decided to discontinue lithium for a variety of reasons (whether electively or for some medical reason), the rate of all suicidal acts rose 14 times following discontinuation (Tondo and Baldessarini 2000). In another sample of 128 individuals who chose to discontinue lithium, the rate of suicide-related deaths rose to 1.27% per year from 0.101% per year with lithium maintenance treatment (Baldessarini et al. 1999). Overall mortality in a group of 273 subjects followed after lithium discontinuation was much higher than in the general population, with a standardized mortality ratio (SMR) of 2.5 for the discontinuation group (the SMR for the general population was 1.0). The authors concluded that this is higher than the SMR for patients continued on lithium in their other data, but because there is no direct comparison, it is not possible to estimate the actual effect of lithium discontinuation on mortality (Muller-Oerlinghausen et al. 1996).
It is unclear whether the risk of suicide following lithium discontinuation exceeds that found in untreated affective illness. However, the striking increase in suicides following lithium discontinuation demonstrates the need for vigilance and frequent communication with patients at risk for suicidal actions. Effective management of adverse side effects or other reasons for which patients may choose to discontinue lithium treatment is also warranted. The effects of lithium discontinuation speak further to the benefits of long-term lithium maintenance therapy. Lithium discontinuation can also contribute to suicidality indirectly; some have suggested that it may lead to higher relapse rates and a refractory response to future lithium treatment (Post et al. 1992).
Several theories have been proposed regarding the mechanisms of suicide prevention afforded by lithium. It is possible that decreased sui-cidality is simply one of the additional benefits accompanying the mood stabilization afforded by lithium. However, it also seems proba ble that lithium has additional protective properties. Serotonin mediation may be a primary means through which lithium exerts these effects. Specifically, lithium's actions in the forebrain may address the potential serotonergic deficiencies that are associated with self-harm and violence (Mann et al. 1999a, 1999b). These effects are not associated with the mood-stabilizing properties of carbamazepine and may explain why lithium has been found to be superior to carbamazepine in suicide prevention (Greil et al. 1997).
Lithium may be the first-line choice for clinical management of suicide risk; however, it is essential that study results be interpreted and clinically applied with caution and careful consideration. Ultimately, although the effects of lithium are promising in the realm of suicide prevention, they have not yet been definitively determined. A greater variety of RCTs specifically designed to measure the anti-suicide effects of lithium are needed for a better understanding of this relationship. Additionally, it is essential that we not neglect the potential protective effects of other psychiatric medications aside from lithium, particularly as an option for lithium nonresponders and patients who cannot tolerate lithium. These individuals may be at a higher risk of suicide attempts and completion (Muller-Oerlinghausen et al. 1992).
There are almost no data available to judge whether any anticonvul-sants typically used in the treatment of bipolar disorder have a prophylactic effect against suicidality. Lamotrigine and divalproex are two anticonvulsants that have been studied in the maintenance treatment of bipolar disorder and that may have some prophylactic efficacy in the prevention of depressive episodes. The evidence that lamotrigine prevents episodes of depressive relapse is particularly robust, but the drug is only recently becoming widely used, and there have been no longitudinal trials examining the question of suicide prevention with it (Goodwin et al. 2004). Divalproex may have some prophylactic effect against the return of depression, but no adequate data exist to support its use for that purpose (Gyulai et al. 2003). Despite its flaws, the Goodwin et al. (2003) study concluded that lithium was superior to divalproex for the prevention of hospitalization for suicidal ideation and suicide attempts, yet it was not designed to determine whether divalproex was superior to other treatments for bipolar disorder in suicide prevention.
In a randomized, maintenance trial of bipolar disorder, 171 patients received either lithium or carbamazepine and were followed for 2.5 years (Kleindienst and Greil 2000). Although there were no suicides during this time, four subjects taking carbamazepine made suicide attempts whereas none of the subjects on lithium did. This difference was not significant, but it was consistent with the data suggesting that lithium prevents suicide.
Atypical antipsychotics are increasingly used as first-line treatment for bipolar disorder, in both acute mania and acute depression as well as in maintenance treatment. Because of the increased use of antipsychotics and their potential replacement of lithium and anticonvulsants (such as valproate), in clinical decision-making with patients, their utility in preventing suicide should be explored. The strongest data suggesting a suicide preventive effect of atypical antipsychotics come from the international suicide prevention trial (interSePT), which found that the risk of suicide in schizophrenia was significantly decreased with clozapine versus olanzapine (Meltzer et al. 2003). This study, however, did not include patients with bipolar disorder, and its generalizability to mood disorders is uncertain.
There are some suggestions that atypical antipsychotics may have some anti-suicide potential in bipolar disorder. In a post hoc analysis of a trial of olanzapine added to lithium or divalproex for mixed or manic episodes, suicidality ratings dropped significantly for patients with dysphoric mania who were given olanzapine. Total 21-item Hamilton Rating Scale for Depression scores were also significantly decreased (Baker et al. 2004). However, because this study was not designed to address the potential of olanzapine to prevent suicidality, it remains unclear how meaningful the results are.
As commonly used as antidepressants are in the treatment of bipolar disorder, their utility in the treatment of bipolar depression remains unclear (Altshuler et al. 2003; Ghaemi et al. 2004; Nemeroff et al. 2001). There are no data to suggest whether they are of utility in preventing suicide. In a study of 78 patients with unipolar or bipolar disorder treated with antidepressants, bipolar patients were more likely to switch to mania, develop rapid cycling, and lose antidepressant response (Ghaemi et al. 2004). Although the study is limited by its retrospective design, it does suggest that, for some patients, risk factors for suicide might increase during antidepressant treatment.
A recent report by Jick et al. (2004) suggests that the risk for suicidal behavior increases in the first month after the initiation of antidepres-
sant treatment—especially in the first nine days. Muller-Oerlinghausen and Berghofer (1999) have suggested that SSRI antidepressants may increase suicidal thoughts through akathesia or by energizing depressed patients, but the risk for this (especially in bipolar disorder) is unknown. In their review, they concluded that while lithium may have anti-suicide potential, this effect has not been shown for antidepres-sants or for nonlithium mood stabilizers.
In a population study in England, Morgan et al. (2004) found that the increase in antidepressant prescriptions between 1993 and 2002 was associated with decreased suicide rates, which fell from 98.2 to 84.3 per million population. Although they could not draw a causal relationship between antidepressant use and suicide reduction, they suggested that the more widespread use of less lethal antidepressants than of tricyclic antidepressants may contribute to this. Whether this association would remain in the subset of the population with bipolar disorder is unknown.
Was this article helpful?