Abstract

To help authors recognize and report bias that could lead to inappropriate decisions from systematic reviews of randomized trials, the Cochrane Collaboration recently implemented a “Risk of Bias” tool. More recently, the Cochrane Collaboration Bias Methods group met and discussed modifications that would be necessary to identify all the additional biases present in observational studies. Although traditional epidemiology lists many different types of biases, the structural approach to bias and causal diagrams suggest that only four different types of bias exist – confounding bias (failure to block bias when a common cause exists), collider bias (conditioning on a common effect), information bias (measurement error) and over-adjustment bias (conditioning on a variable that lies within the causal pathway, or is a marker for a variable within the causal pathway). This presentation will briefly review casual diagrams, and show how they can be used to demonstrate that the current Risk of Bias tool requires only a couple of minor modifications to be applicable to all study designs. Using causal diagrams requires that authors and researchers make more of their assumptions explicit, which should improve the quality of both original research and systematic reviews.

Ian Shrier MD, PhD, Dip Sport Med is an Associate Professor of the Centre for Clinical Epidemiology and Community studies, Lady Davis Institute for Medical Research at the Jewish General Hospital, McGill University, Montreal Canada. He has trained in medicine, physiology and epidemiology, and his current research interests include causal inference, methodology of systematic reviews, and sport injury epidemiology. He has over 150 peer-reviewed publications and sits on the editorial board of four international sport medicine journals.