An efficient analysis of covariance model for crossover thorough QT studies with period-specific baseline days

Baseline adjustment is an important consideration in thorough QT studies for nonantiarrhythmic drugs. For crossover studies with period-specific baseline days, we propose an analysis of covariance model with change from time-matched baseline as response, time-matched baseline for the current treatment, day-averaged baseline for the current treatment, time-matched baseline averaged across treatments, and day-averaged baseline averaged across treatments as covariates. This model adjusts for within-subject diurnal effects for each treatment and is more efficient than commonly used models for treatment comparisons. We illustrate the benefit using real clinical trial data.