Clinical correlates of sustained response to individual drugs used in naturalistic treatment of patients with bipolar disorder.

Objective: To report use and treatment success rates of medications for bipolar disorder as a function of patients' clinical characteristics.

Method: Outpatients with bipolar illness diagnosed by SCID were rated by research assistants on the NIMH-LCM and those who had an good response for at least 6months (much or very much improved on the CGI-BP) were considered responders (treatment "success"). Clinical characteristics associated with treatment response in the literature were examined for how often a drug was in a successful regimen when a given characteristic was either present or absent.

Results: Lithium was less successful in those with histories of rapid cycling, substance abuse, or (surprisingly) a positive parental history of mood disorders. Valproate was less successful in those with ≥20 prior episodes. Lamotrigine (LTG) was less successful in those with a parental history of mood disorders or in BP-I compared to BP-II disorder. Antidepressants (ADs) had low success rates, especially in those with a history of anxiety disorders. Benzodiazepines had low success rates in those with child abuse, substance use, or ≥20 episodes. Atypical antipsychotics were less successful in the presence of rapid cycling, ≥20 prior episodes, or a greater number of poor prognosis factors.

Conclusion: Success rates reflect medications used in combination with an average of two other drugs during naturalistic treatment and thus should be considered exploratory. However, the low long-term success rates of drugs (even when used in combination with others) that occurred in the presence of many very common clinical characteristics of bipolar illness speak to the need for the development of alternative treatment strategies.