Catalogue of bias: allocation bias.

Abstract

This article is part of a series of articles featuring the Catalogue of Bias introduced in this volume of <i>BMJ Evidence-Based Medicine</i> that describes allocation bias and outlines its potential impact on research studies and the preventive steps to minimise its risk. Allocation bias is a type of selection bias and is relevant to clinical trials of interventions. Knowledge of interventions prior to group allocation can result in systematic differences in important characteristics that could influence study findings. Allocation bias can overestimate effect size by up to 30%-40%. Sequentially numbered, opaque, sealed envelopes; containers; pharmacy-controlled randomisation and central computer randomisation are methods to minimise allocation bias.