O-280 Reading between the lines: Evaluating the clinical benefit of laboratory interventions from published studies

Abstract

Abstract Embryologists rely on both randomised and non-randomised studies evaluating the effectiveness and safety of interventions. However, many of these studies contain subtle but critical methodological limitations, which are not identified by peer reviewers. Embryologists must therefore be critical consumers of the published literature. In this talk, I will highlight some common, serious methodological flaws occurring in studies of embryological interventions, and will give intuitive explanations of both the problems and the impact they may have on study results. This includes methods of analysing randomised controlled trials which nullify the benefits conferred by randomisation, and flawed methods of identifying study groups in observational studies which prevent any meaningful evaluation of an intervention's benefit. I will discuss underappreciated methodological challenges in studies evaluating artificial intelligence algorithms for embryo selection, which limit their usefulness, and will outline superior approaches to evaluate these interventions.