Methodological issues in infertility research.

Abstract

It is now well established that clinical decision-making can be enhanced by incorporating evidence from valid studies. This evidence-based approach to health-care management requires a critical appraisal of the available evidence, most of which is of relatively poor quality. If the best evidence is to be put forward to support clinical decisions then it must be derived from studies of high methodological rigour. This chapter discusses important methodological issues affecting the evaluation of the efficacy of therapeutic interventions; these are supplemented with illustrative examples from the literature. These issues form the basis for establishing an approach to inform gynaecological practice, with particular reference to infertility management using assisted reproductive technology and other therapies. Ten important issues are discussed: the formulation of an appropriate research question, the use of randomization, the importance of concealment of treatment allocation, the importance of blinding or masking to avoid ascertainment bias, the avoidance of cointervention, the requirement of an adequate sample size, the restriction of evaluation to the first cycle of treatment, the avoidance of the crossover trial when pregnancy is the outcome of interest, the importance of analysing data using an intention-to-treat approach, and the clear identification of the orientation of the study from the perspectives of superiority, equivalence or non-inferiority of the interventions being compared.