A simple analysis of crossover studies with one-group interaction.

The crossover trial has an intuitive appeal to clinicians because each patient is use as his own control. Thus, between-subject variability of symptoms is eliminated. However, this study design suffers frequently from the bias of treatment-by-period interaction. If, for example, the effect of the first treatment period carries on into the next one, then it influences the response to the later period (carryover effect). A second problem is that the standard approach (Hills-Armitage analysis) for interaction has little power. This led the FDA and some statisticians to discourage the use of crossovers. In the present report I take issue with the common clinical situation where there is carryover effect in only one of the groups of a trial and present a simplified analysis for this situation with more power than the standard analysis.