Lead-Time Bias Confounds Association Between Duration of Untreated Psychosis and Illness Course in Schizophrenia.

Abstract

At first hospitalization, a long duration of untreated psychosis (DUP) predicts illness severity and worse treatment outcomes. The mechanism of this association, however, remains unclear. It has been hypothesized that lengthy untreated psychosis is toxic or that it reflects a more severe form of schizophrenia. Alternatively, the association may be an artifact of lead-time bias. These hypotheses are tested in a longitudinal study of schizophrenia with 2,137 observations spanning from childhood to 20 years after first admission. Data were from the Suffolk County Mental Health Project. The cohort included 287 individuals with schizophrenia or schizoaffective disorder. DUP was defined as days from first psychotic symptom to first psychiatric hospitalization. Psychosocial function was assessed using the Premorbid Adjustment Scale and the Global Assessment of Functioning Scale. Psychosocial function trajectories were estimated using multilevel spline regression models adjusted for gender, occupational status, race, and antipsychotic medication. Both long- and short-DUP patients experienced similar declines in psychosocial function, but declines occurred at different times relative to first admission. Long-DUP patients experienced most of these declines prior to first admission, while short-DUP patients experienced declines after first admission. When psychosocial function was analyzed relative to psychosis onset, DUP did not predict illness course. The association between DUP and psychosocial function may be an artifact of early detection, creating the illusion that early intervention is associated with improved outcomes. In other words, DUP may be better understood as an indicator of illness stage than a predictor of course.