Adverse Effects of the Dependent Type in Crossover Trials

Abstract

Carryover effect, otherwise called treatment by period interaction, is a major adverse effect of crossover studies. If the effect of a treatment carries on into the second period of treatment, then the measured response to the second period of treatment will be changed. This carryover effect can be considered an adverse effect of the dependent type. This is, because it changes the outcome of the study. We report levels of carryover required for a carryover effect to be significant, and give examples of negative crossover studies due to carryover adverse effect. If significant, then an overall analysis of the study is pretty meaningless, but it makes sense to analyze the first period of the study for treatment effect and disregard the second period.

The current chapter sows, that the adverse effect of carryover effect from the first into the second period of treatment in the control group can be easily tested for statistical significance. With a significant carryover effect, an overall treatment assessment makes no sense, but an unpaired analysis of the first periods of treatment is a worthwhile alternative for the purpose.