Two-period linear mixed effects models to analyze clinical trials with run-in data when the primary outcome is continuous: Applications to Alzheimer's disease.

Introduction: Study outcomes can be measured repeatedly based on the clinical trial protocol before randomization during what is known as the "run-in" period. However, it has not been established how best to incorporate run-in data into the primary analysis of the trial.

Methods: We proposed two-period (run-in period and randomization period) linear mixed effects models to simultaneously model the run-in data and the postrandomization data.

Results: Compared with the traditional models, the two-period linear mixed effects models can increase the power up to 15% and yield similar power for both unequal randomization and equal randomization.

Discussion: Given that analysis of run-in data using the two-period linear mixed effects models allows more participants (unequal randomization) to be on the active treatment with similar power to that of the equal-randomization trials, it may reduce the dropout by assigning more participants to the active treatment and thus improve the efficiency of AD clinical trials.