Attrition is typically 25% to 50% in psychopharmacology trials. Mixed-effects models can include clinical trial participants with incomplete data. However, the ignorable attrition assumption is usually required for valid inferences. A novel approach to evaluating its plausibility is considered. At baseline, participants rate their Intent to Complete the trial. Throughout the trial, participants rate their Intent to Attend the next assessment session.

Experience with Intent to Attend in recent trials is described. In an effort to reduce attrition bias, blinded raters queried those with low intent and attempted to accommodate their needs. In addition, the ratings were incorporated in sensitivity analyses. A simulation study is discussed that compares bias and coverage of mixed-effects analyses that do and do not include Intent to Attend as a covariate.