A two-part mixture model for longitudinal adverse event severity data.

摘要: We fit a mixed effects logistic regression model to longitudinal adverse event (AE) severity data (four-point ordered categorical response) describe the dose-AE response for an investigational drug. The distribution of predicted interindividual random (Bayes predictions) was extremely bimodal. This extreme bimodality indicated that biased parameter estimates and poor predictive performance were likely. distribution's primary mode composed patients did not experience AE. Moreover, Bayes predictions these non-AE nearly degenerative; i.e., identical. To resolve this we propose using two-part mixture modeling approach. first part models incidence AE's, second grade given patient had Unconditional probability are calculated by mixing predictions. also report results simulation studies, which assess statistical (bias precision) our