Yatham 2003.

Methods	Allocation: multicentre, randomised, placebo‐controlled trial Blindness: described as "double blind" Duration: 52 weeks Setting: Canadain and Brazilian academic centres
Participants	Bipolar I disorder in remission from recent manic or mixed episode on treatment with mood stabiliser (valproate or lithium) and either risperidone or olanzapine (N = 159, not clear how many had psychotic features). Total of 39% (62/159) of total sample had co‐occurring alcohol or substance use disorder.
Interventions	Discontinuation of risperidone or olanzapine at either 0 weeks, 24 weeks or 52 weeks and substitution with placebo.
Outcomes	Time to any mood episode, YMRS, HAMD‐21, MADRS, CGI‐BP, CGI‐S, Side‐effects UKU scale, ESRS, weight, metabolic measures (glucose, lipid profile).
Notes	Authors contacted. Responded that no data or analyses available at present for subgroups. No information provided on how many participants had bipolar type I with psychotic features.