Crossover Study Designs

Definition & rationale — each participant as own control

• Definition — A crossover study is a randomized interventional design in which each participant receives both (or all) the interventions being compared, in sequence, so that every participant serves as his or her own control — the defining feature is within-subject (intra-individual) comparison.
• Two analyses from one trial — The dual structure permits a between-group analysis (group starting active vs group starting control — preserves randomization, like a parallel trial) and a within-group paired analysis (each participant on intervention vs the same participant on control).
• Eliminates, not merely balances, confounders — Because it is literally the same person, time-invariant characteristics (age, sex, genotype, baseline severity) are held identical across the compared conditions — they are eliminated as confounders rather than balanced on average as in a parallel randomized trial. This removes the large between-subject component of variance, shrinking the standard error and raising statistical power.
• Best-suited niche — A crossover is a good choice when the sample is limited and the outcome responds rapidly and reversibly to the intervention — one sentence that captures both the chief advantage (efficiency) and the chief eligibility constraint.
• Conceptual ancestor — The crossover descends from the before–after (pre–post) design (same participants measured before and after, each as own control) but adds randomization of treatment order and a control period. (Parallel-group and factorial designs are out of scope — contrasted only as comparators.)