Crossover Trial Design

Each enrolled child undergoes two separate sedation sessions separated by a minimum 5-day washout period. With α = 0.05 and 80% power, 72 participants are required; allowing for 10% dropout sets the target enrolment at 80 children. An independent DSMB reviews data after every 20 enrolled cases and can recommend early termination using a Haybittle-Peto stopping boundary of p < 0.001. A hierarchical testing strategy controls family-wise type I error, first testing superiority for desaturation and then non-inferiority for psychotomimetic effects. The crossover design was selected because within-patient variability is smaller than between-patient variability, enabling a smaller required sample than a parallel-group design. Carryover effects are assessed via a treatment × period interaction term; if significant, a sensitivity analysis uses first-period data only, treating the trial as a parallel-group study.