Risk factors for early recurrence after discontinuing lithium in bipolar disorder.

Abstract

Time to a new episode of bipolar disorder (BD) is shorter after discontinuing lithium rapidly. We now address this and other factors associated with the risk of early illness after discontinuing lithium. We compared factors for association with recurrences of BD within 12months of discontinuing long-term lithium treatment, using bivariate and multivariable analyses, as well as survival analysis to evaluate latency to new episodes versus rate of lithium-discontinuation and prior treatment duration. Among 227 BD subjects who received lithium for 4.47 [CI: 3.89-5.04] years and then discontinued, rapid treatment-discontinuation, and stopping for medical reasons were strongly associated with new illness-episodes within 12months, as were diagnosis (BD-I > BD-II), greater morbidity during lithium-treatment, and less education, but neither longer treatment nor serum lithium concentrations. Discontinuation rate was strongly associated with shorter median latency to a new episode (rapid: 3.50; gradual [≥2 weeks]: 10.6 months), even with very early recurrences excluded to avoid potential contributions of emerging illness to treatment-discontinuation. Early recurrence was not associated with treatment-duration of ≥2 or ≥5 years or less. In multivariable logistic regression, rapid discontinuation, stopping for medical reasons, and BD-I diagnosis remained significantly, independently associated with early illness after lithium-discontinuation, with no effect of treatment duration. Early recurrence risk was again much greater after rapid discontinuation of lithium and discontinuing for medical reasons, somewhat greater with BD-I than BD-II, and following greater morbidity during lithium-treatment, but not related to dose or duration of preceding treatment exposure.