A MULTIPLE IMPUTATION STRATEGY FOR CLINICAL-TRIALS WITH TRUNCATION OFPATIENT DATA

Clinical trials of drug treatments for psychiatric disorders commonly employ the parallel groups, placebo-controlled, repeated measure rando mized comparison. When patients stop adhering to their originally assi gned treatment, investigators often abandon data collection. Thus, non -adherence produces a monotone pattern of unit-level missing data, dis abling the analysis by intent-to-treat. We propose an approach based o n multiple imputation of the missing responses, using the approximate Bayesian bootstrap to draw ignorable repeated imputations from the pos terior predictive distribution of the missing data, stratifying by a b alancing score for the observed responses prior to withdrawal. We appl y the method and some variations to data from a large randomized trial of treatments for panic disorder, and compare the results to those ob tained by the original analysis that used the standard (endpoint) meth od.