Recurrence risks in schizophrenia: Are they model dependent?