Erich Studerus1, Katharina Beck1, Anita Riecher-Rössler1; 1Center for Gender Research and Early Detection, University of Basel Psychiatric Hospital, University of Basel, Basel, Switzerland
Improved prediction of transition to psychosis in those with a clinical high risk (CHR) has become an important goal in psychosis research. However, previously developed prediction models almost exclusively rely on baseline data. Thus, their risk prediction cannot be dynamically updated during the follow-up period when new information becomes available. Furthermore, they do not reveal the specific relationship between symptoms during the follow-up and risk of transition. To solve these problems, the present study made use of so called joint models, a relatively recent statistical innovation that allows studying the association between a longitudinal process (i.e. change in symptoms over time) and a time-to-event outcome (i.e. time to transition to psychosis). We fitted three different joint models in which the hazard for transition at any time t was assumed to be related to 1) the absolute level, 2) the average absolute value since baseline, and 3) the velocity of change of (attenuated) positive psychotic symptoms at the same time point t. Data were obtained as part of the “Basel Früherkennung von Psychosen” (FePsy) study and included 191 CHR patients, of whom 42 transitioned to psychosis during follow-up. Although in all three models the association between longitudinal process and time-to-event outcome was statistically significant, the association was strongest in model 3. Our results therefore suggest that the risk of developing psychosis at any time t during follow-up is most strongly predicted by how fast (attenuated) positive psychotic symptoms are increasing at the same time point.
