Important questions regarding lithium still remain. Greil and colleagues (1996; 1997a; 1997b), in a series of randomized controlled studies with treatment periods of 2.5 years, found that the prophylactic efficacy of lithium on suicidality varied according to the underlying mental disorder. Carbamazepine was more effective than lithium in reducing suicidal behavior in schizoaffective disorder, especially in subgroups with depressive or schizophrenia-like features; in bipolar types it was not superior (Greil et al., 1997a). For unipolar depressed patients, lithium was found to be superior to amitriptyline (Greil et al., 1996), and in bipolar disorder patients, lithium was judged to be superior to carbamazepine (Greil et al., 1997b). Several studies also suggest that bipolar patients with rapid cycling or mixed states are difficult to treat effectively and do not seem to respond as well to lithium (Bowden et al., 2000; Calabrese et al., 2001b; Montgomery et al., 2000). Comorbidities, especially with substance use disorder, also interfere with treatment outcome (Macqueen and Young, 2001; Vestergaard et al., 1998), though comorbidity appears to moderate outcomes via treatment adherence (Calabrese et al., 2001b).
The mechanism of action of lithium is unknown. It has been hypothesized that it exerts antisuicidal effects on aggressive impulsive traits via the serotonergic system or otherwise. Importantly, lithium appears to have a direct effect on suicidal behavior, not simply by reducing the suicidality caused by depressive relapses (Möller, 2001).
The other class of mood stabilizers found to reduce symptoms of bipolar disorder are the anti-convulsants, such as carbamazepine, divalproex, and valproic acid. These medications are recommended for bipolar patients when lithium is not an option, whether due to lithium intolerance or resistance to lithium treatment (Möller, 2001). Valproate is the most commonly prescribed mood stabilizer in the United States, overtaking lithium. However the data are very limited on the efficacy of anticonvulsants to reduce suicidal behavior; only one randomized controlled study was identified in a recent review (Goodwin and Ghaemi, 2000). Thies-Flechtner et al. (1996) conducted a 2.5 year prospective study on 175 inpatients with bipolar disorder. These patients were treated either with carbamazine or with lithium. Of the 6 patients who committed suicide, 4 were taking carbamazine. None were taking lithium at time of death, but one had discontinued lithium. All of the suicide attempts occurred in patients who were taking carbamazine. These data demonstrated a statistically significant benefit of lithium over carbamazine in the prevention of suicide. Because of the frequency with which anticonvulsants are prescribed for bipolar disorder, it is exceptionally