To assess the metabolic impact of adding an antipsychotic to a mood stabilizer or switching a mood stabilizer to an antipsychotic in patients with bipolar disorder.

A retrospective fixed cohort study was conducted through the claims database of the French health care program for the self-employed workers. The study population consisted of 3.172 patients age 18 and over who were exposed to mood stabilizers (i.e. lithium, valproate) a 3 month-period in 2004 without dispensing of non-sedative antipsychotic, antidiabetic or lipid-lowering drugs. The outcome was the occurrence of a metabolic incident over the follow-up period, using the dispensing of an antidiabetic drug as a marker of diabetes and the dispensing of a lipid-lowering drug as a marker of hypercholesterolemia or hypertriglyceridemia. A Cox proportional hazard model was used to assess the metabolic impact of the antipsychotics; using mood stabilizers as a reference. Antipsychotic exposition was stratified in «current» and «recent» (discontinued for less than 6 months) at the time of the metabolic incident.

196 patients (6.2%) received a first-generation antipsychotic, 352 (11.1%) a second-generation antipsychotic, 565 (17.8%) a sedative antipsychotic and 367 patients (11.6%) presented with a metabolic incident over the study period. The recent dispensing of a second-generation antipsychotic was associated with the occurrence of a metabolic incident [HR 2.1 (95%CI 1.2-3.7) p=0.006], while current dispensing or dispensing of first-generation antipsychotics were not.

Second-generation antipsychotics have a metabolic impact compared to classic mood stabilizers in patients with bipolar disorder.