Dichotomisation in clinical trials: issues and concerns
*Steven Julious, University of Sheffield 

Keywords: Dichotomisation; clinical trials

If you were to choose a statistical test for an analysis how often would you choose one which massively inflates the sample size to maintain the required power; clouds interpretation and increases by a factor the complexity of the analysis? It would be safe to say not often and yet these are the consequences of categorising a continuous outcome into two categories (dichotomisation). It is wrong on many levels to dichotomise. This presentation will describe consequences and effects of dichotomisation as well as highlight concerns.