Fig. 1

Flow diagram of data inclusion. Data were selected from the starting OXTEXT cohort 2001–2012 according to inclusion criteria designed to reduce missing data. Cohorts for the main and secondary analyses were then selected based on the number of available individual observations such that every participant included had a period of eight consecutive weeks without either a manic or depressive episode (episode-free run-in period) and at least four follow-up observations up to the first episode