The pitfalls of using a cross-over design for infertility trials

The cross over trial has advantages in terms of reduced sample size and avoiding between patient variability, but is inappropriate for use in infertility when the outcome of interest is pregnancy. A within-subject comparison, wherein the subject receives both experimental and control interventions and serves as her own control is not possible because conception with one intervention excludes the subject from receiving the alternative intervention. Also, the inability to estimate period effect, carry-over affect and sequence effect further compromises the ability of the study to provide a reliable estimate of the magnitude of the effect of the experimental intervention. Consequently, the cross over design should not be used in infertility trials when pregnancy is the outcome of interest.