The place of the crossover design in infertility trials: A maximum likelihood approach

Abstract

Background: For some years, there has been a debate as to the place of the crossover trial in assisted reproduction technology (ART). We aimed to investigate whether crossover and parallel designs result in different estimates of treatment effects. Methods: We carried out computer simulation of cohorts of patients undergoing either intra-uterine insemination (IUI) or lVF under both parallel and crossover designs, under scenarios involving censoring and carryover effects. Results of the simulation were analysed using a maximum likelihood approach. Results: No relevant difference was found between the designs. The crossover design resulted in slightly more pregnancies than the parallel design. Carryover effects may slightly distort the estimates of treatment effects. Crossover and parallel designs will produce essentially the same statistical estimates of treatment effect and percentage of pregnancies. The crossover design is an acceptable design in infertility research provided the data are analysed correctly. © European Society of Human Reproduction and Embryology 2004; all rights reserved.