Risk of Recurrence Following Discontinuation of Lithium Treatment in Bipolar Disorder Trisha Suppes, MD, PhD; Ross J. Baldessarini, MD; Gianni L. Faedda, MD; Mauricio Tohen, MD, DrPH \s=b\ Episode recurrence in bipolar disorder following discon- tinuation of stable maintenance treatment with lithium salts was analyzed from 14 studies involving 257 patients with bipolar I disorder. More than 50% of new episodes of illness occurred within 10 weeks of stopping an average of 30 months of treatment. By survival analysis of 124 cases in which the time to a new episode was known, the computed time to 50% failure of remission was 5.0 months after stop- ping therapy; the time to 25% recurrence of mania was 5.2 times earlier than for depression (2.7 vs 14 months). In 16 patients with a mean cycle length before treatment of 11.6 months, the time to a new episode when off lithium therapy was only 1.7 months. Risk of early recurrence of bipolar ill- ness, especially of mania, evidently is increased following discontinuation of lithium use and may exceed that pre- dicted by the course of the untreated disorder. The basis and management of risks associated with discontinuing effective long-term mood-stabilizing treatment require further study. (Arch Gen Psychiatry. 1991;48:1082-1088) Lithium salts are a well-established treatment for mania and recurrences of manic-depressive illness (bipolar disorder) and may be useful in other forms of recurring or treatment-resistant major depressive disorders as well.1"3 Lithium seems to be preventive and mood stabilizing by reducing the incidence, severity, and duration of episodes or recurring cycles of major mood disorders.2-3 Prolonged use of lithium carries medical risks that include structural or functional renal and thyroid abnormalities, subtle or potentially severe cerebral intoxication, dermatologie dis¬ orders, and probable cardiovascular teratogenicity in the fetus.2 An apparently growing number of patients receiv¬ ing maintenance lithium therapy have treatment inter¬ rupted owing to side effects, pregnancy, medical illness, inadequate response, or a wish to try discontinuing the treatment after doing well for a time. Thus, clarification is needed of risks involved when interrupting effective lith¬ ium treatment, and of the optimal treatment of manic- depressive patients taken off mood-stabilizing mainte¬ nance treatment. This summary and analysis of available information seeks to clarify the risk over time of recurrence of a manic or depressive episode following discontinuation of apparently effective long-term lithium treatment. Specifically, it seeks to test the clinically based hypoth¬ esis that recurrence (a new episode of illness following a symptom-free period of several months)4 may appear earlier after discontinuation of lithium therapy than would be predicted from knowledge of the natural history of the course of bipolar disorder or the history of individual patients. MATERIALS AND METHODS For this analysis, we searched for studies involving discon¬ tinuation of previously successful lithium treatment in pa¬ tients with a major mood disorder who had been clinically stable for at least several months. In particular, we sought studies of bipolar patients that would permit an estimate of the overall risk and time to a new episode, as well as assess¬ ment of the time to recurrence for individual patients. A computerized literature search was supplemented with refer¬ ences cited in studies of the clinical efficacy of lithium salts, and by consulting colleagues for access to data from recent or unpublished studies. To be accepted, studies had to have been carried out since 1960 (to minimize risk of uncertain diagnosis); involved at least four cases of a bipolar type I manic-depressive disorder (with mania) diagnosed by criteria permitting a DSM-III-R- or International Classification of Diseases, ninth revision-compatible diagnosis; documented clinical stability on lithium treatment for at least 3 months; and studied patients for at least a week following dis¬ continuation of lithium therapy under either blinded or open observation. Case reports involving one or two patients were excluded, as were cases of bipolar II disorder, cyclothymia, or nonbipolar mood disorders. The definition of recurrence of a new episode of mood disorder typically required rehospitaliza¬ tion or clinical worsening sufficiently severe as to require phar¬ macologie or electroconvulsive therapy. This selection process yielded a total of 14 pertinent studies5"'6 (E. Klein, MD, P. Lavie, MD, R. Meiraz, MD, A. Sadeh, MD, and R. H. Lenox, MD, unpublished data, November 13, 1990, and January 2, 1991; and L. Tondo, MD, unpublished data, Novem¬ ber 30, 1990; summarized in the Table) with usable data on a to¬ tal of 257 patients with bipolar I disorder, including a few (<2%) Accepted for publication April 18, 1991. From the Department of Psychiatry (Drs Suppes, Baldessarini, Faedda, and Tohen) and Neuroscience Program (Drs Suppes and Baldessarini), Harvard Medical School, and the Department of Ep- idemiology, Harvard School of Public Health (Dr Tohen), Boston, Mass; and the Laboratories for Psychiatric Research and Psychotic Disorders Program (Drs Suppes, Baldessarini, Faedda, and Tohen), Mailman Research Center, McLean Division of Massachusetts General Hospital, Belmont. Reprint requests to Mailman Research Center, McLean Hospital, 115 Mill St, Belmont, MA 02178 (Dr Suppes). Downloaded From: http://archpsyc.jamanetwork.com/ by a Mt Sinai School Of Medicine User on 08/18/2012